{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 34706,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.8813461649282546e-05,
      "grad_norm": 393.82843017578125,
      "learning_rate": 9.596928982725528e-09,
      "loss": 6.6975,
      "step": 1
    },
    {
      "epoch": 5.762692329856509e-05,
      "grad_norm": 396.66815185546875,
      "learning_rate": 1.9193857965451055e-08,
      "loss": 6.6976,
      "step": 2
    },
    {
      "epoch": 8.644038494784764e-05,
      "grad_norm": 389.69732666015625,
      "learning_rate": 2.8790786948176586e-08,
      "loss": 6.5623,
      "step": 3
    },
    {
      "epoch": 0.00011525384659713018,
      "grad_norm": 391.723388671875,
      "learning_rate": 3.838771593090211e-08,
      "loss": 6.6447,
      "step": 4
    },
    {
      "epoch": 0.00014406730824641272,
      "grad_norm": 394.2310485839844,
      "learning_rate": 4.798464491362764e-08,
      "loss": 6.7376,
      "step": 5
    },
    {
      "epoch": 0.00017288076989569528,
      "grad_norm": 394.6132507324219,
      "learning_rate": 5.758157389635317e-08,
      "loss": 6.6357,
      "step": 6
    },
    {
      "epoch": 0.0002016942315449778,
      "grad_norm": 390.2063293457031,
      "learning_rate": 6.71785028790787e-08,
      "loss": 6.6673,
      "step": 7
    },
    {
      "epoch": 0.00023050769319426037,
      "grad_norm": 397.5264587402344,
      "learning_rate": 7.677543186180422e-08,
      "loss": 6.6255,
      "step": 8
    },
    {
      "epoch": 0.0002593211548435429,
      "grad_norm": 383.6949768066406,
      "learning_rate": 8.637236084452976e-08,
      "loss": 6.5264,
      "step": 9
    },
    {
      "epoch": 0.00028813461649282543,
      "grad_norm": 397.8243713378906,
      "learning_rate": 9.596928982725528e-08,
      "loss": 6.644,
      "step": 10
    },
    {
      "epoch": 0.00031694807814210797,
      "grad_norm": 389.0898132324219,
      "learning_rate": 1.0556621880998082e-07,
      "loss": 6.6411,
      "step": 11
    },
    {
      "epoch": 0.00034576153979139055,
      "grad_norm": 362.9072570800781,
      "learning_rate": 1.1516314779270635e-07,
      "loss": 6.4819,
      "step": 12
    },
    {
      "epoch": 0.0003745750014406731,
      "grad_norm": 386.5528869628906,
      "learning_rate": 1.2476007677543187e-07,
      "loss": 6.5776,
      "step": 13
    },
    {
      "epoch": 0.0004033884630899556,
      "grad_norm": 361.44610595703125,
      "learning_rate": 1.343570057581574e-07,
      "loss": 6.4576,
      "step": 14
    },
    {
      "epoch": 0.00043220192473923815,
      "grad_norm": 370.8934326171875,
      "learning_rate": 1.4395393474088293e-07,
      "loss": 6.5188,
      "step": 15
    },
    {
      "epoch": 0.00046101538638852074,
      "grad_norm": 331.69268798828125,
      "learning_rate": 1.5355086372360844e-07,
      "loss": 6.2767,
      "step": 16
    },
    {
      "epoch": 0.0004898288480378033,
      "grad_norm": 334.1194152832031,
      "learning_rate": 1.6314779270633396e-07,
      "loss": 6.2854,
      "step": 17
    },
    {
      "epoch": 0.0005186423096870858,
      "grad_norm": 316.3736877441406,
      "learning_rate": 1.7274472168905953e-07,
      "loss": 6.215,
      "step": 18
    },
    {
      "epoch": 0.0005474557713363683,
      "grad_norm": 307.3128356933594,
      "learning_rate": 1.8234165067178504e-07,
      "loss": 6.1595,
      "step": 19
    },
    {
      "epoch": 0.0005762692329856509,
      "grad_norm": 299.8911437988281,
      "learning_rate": 1.9193857965451055e-07,
      "loss": 6.0858,
      "step": 20
    },
    {
      "epoch": 0.0006050826946349334,
      "grad_norm": 298.2571716308594,
      "learning_rate": 2.0153550863723612e-07,
      "loss": 6.1094,
      "step": 21
    },
    {
      "epoch": 0.0006338961562842159,
      "grad_norm": 159.06195068359375,
      "learning_rate": 2.1113243761996164e-07,
      "loss": 5.4276,
      "step": 22
    },
    {
      "epoch": 0.0006627096179334986,
      "grad_norm": 150.9569549560547,
      "learning_rate": 2.2072936660268715e-07,
      "loss": 5.3936,
      "step": 23
    },
    {
      "epoch": 0.0006915230795827811,
      "grad_norm": 150.22280883789062,
      "learning_rate": 2.303262955854127e-07,
      "loss": 5.42,
      "step": 24
    },
    {
      "epoch": 0.0007203365412320636,
      "grad_norm": 146.5871124267578,
      "learning_rate": 2.3992322456813823e-07,
      "loss": 5.3987,
      "step": 25
    },
    {
      "epoch": 0.0007491500028813462,
      "grad_norm": 139.92697143554688,
      "learning_rate": 2.4952015355086375e-07,
      "loss": 5.3454,
      "step": 26
    },
    {
      "epoch": 0.0007779634645306287,
      "grad_norm": 130.39315795898438,
      "learning_rate": 2.5911708253358926e-07,
      "loss": 5.3039,
      "step": 27
    },
    {
      "epoch": 0.0008067769261799112,
      "grad_norm": 130.30320739746094,
      "learning_rate": 2.687140115163148e-07,
      "loss": 5.2763,
      "step": 28
    },
    {
      "epoch": 0.0008355903878291938,
      "grad_norm": 116.17740631103516,
      "learning_rate": 2.7831094049904034e-07,
      "loss": 5.2165,
      "step": 29
    },
    {
      "epoch": 0.0008644038494784763,
      "grad_norm": 97.97239685058594,
      "learning_rate": 2.8790786948176586e-07,
      "loss": 5.0322,
      "step": 30
    },
    {
      "epoch": 0.0008932173111277589,
      "grad_norm": 80.5440902709961,
      "learning_rate": 2.9750479846449137e-07,
      "loss": 4.8622,
      "step": 31
    },
    {
      "epoch": 0.0009220307727770415,
      "grad_norm": 81.1929931640625,
      "learning_rate": 3.071017274472169e-07,
      "loss": 4.6967,
      "step": 32
    },
    {
      "epoch": 0.000950844234426324,
      "grad_norm": 84.05754089355469,
      "learning_rate": 3.166986564299424e-07,
      "loss": 4.6769,
      "step": 33
    },
    {
      "epoch": 0.0009796576960756065,
      "grad_norm": 80.52777099609375,
      "learning_rate": 3.262955854126679e-07,
      "loss": 4.6386,
      "step": 34
    },
    {
      "epoch": 0.001008471157724889,
      "grad_norm": 78.04695892333984,
      "learning_rate": 3.3589251439539354e-07,
      "loss": 4.6444,
      "step": 35
    },
    {
      "epoch": 0.0010372846193741716,
      "grad_norm": 73.06047821044922,
      "learning_rate": 3.4548944337811905e-07,
      "loss": 4.5724,
      "step": 36
    },
    {
      "epoch": 0.0010660980810234541,
      "grad_norm": 71.63874816894531,
      "learning_rate": 3.5508637236084457e-07,
      "loss": 4.4764,
      "step": 37
    },
    {
      "epoch": 0.0010949115426727367,
      "grad_norm": 67.9021224975586,
      "learning_rate": 3.646833013435701e-07,
      "loss": 4.4543,
      "step": 38
    },
    {
      "epoch": 0.0011237250043220192,
      "grad_norm": 64.00082397460938,
      "learning_rate": 3.742802303262956e-07,
      "loss": 4.4512,
      "step": 39
    },
    {
      "epoch": 0.0011525384659713017,
      "grad_norm": 58.2032585144043,
      "learning_rate": 3.838771593090211e-07,
      "loss": 4.3564,
      "step": 40
    },
    {
      "epoch": 0.0011813519276205843,
      "grad_norm": 52.32915115356445,
      "learning_rate": 3.934740882917467e-07,
      "loss": 4.2693,
      "step": 41
    },
    {
      "epoch": 0.0012101653892698668,
      "grad_norm": 48.81806564331055,
      "learning_rate": 4.0307101727447224e-07,
      "loss": 4.1808,
      "step": 42
    },
    {
      "epoch": 0.0012389788509191493,
      "grad_norm": 48.04615783691406,
      "learning_rate": 4.1266794625719776e-07,
      "loss": 4.1711,
      "step": 43
    },
    {
      "epoch": 0.0012677923125684319,
      "grad_norm": 42.52837371826172,
      "learning_rate": 4.2226487523992327e-07,
      "loss": 4.0573,
      "step": 44
    },
    {
      "epoch": 0.0012966057742177146,
      "grad_norm": 39.75392532348633,
      "learning_rate": 4.318618042226488e-07,
      "loss": 3.9971,
      "step": 45
    },
    {
      "epoch": 0.0013254192358669971,
      "grad_norm": 41.320159912109375,
      "learning_rate": 4.414587332053743e-07,
      "loss": 3.9961,
      "step": 46
    },
    {
      "epoch": 0.0013542326975162797,
      "grad_norm": 37.7562141418457,
      "learning_rate": 4.5105566218809987e-07,
      "loss": 3.8464,
      "step": 47
    },
    {
      "epoch": 0.0013830461591655622,
      "grad_norm": 37.05450439453125,
      "learning_rate": 4.606525911708254e-07,
      "loss": 3.8441,
      "step": 48
    },
    {
      "epoch": 0.0014118596208148447,
      "grad_norm": 35.514583587646484,
      "learning_rate": 4.702495201535509e-07,
      "loss": 3.7703,
      "step": 49
    },
    {
      "epoch": 0.0014406730824641273,
      "grad_norm": 35.457618713378906,
      "learning_rate": 4.798464491362765e-07,
      "loss": 3.7593,
      "step": 50
    },
    {
      "epoch": 0.0014694865441134098,
      "grad_norm": 34.086917877197266,
      "learning_rate": 4.894433781190019e-07,
      "loss": 3.6907,
      "step": 51
    },
    {
      "epoch": 0.0014983000057626923,
      "grad_norm": 32.369544982910156,
      "learning_rate": 4.990403071017275e-07,
      "loss": 3.6312,
      "step": 52
    },
    {
      "epoch": 0.0015271134674119749,
      "grad_norm": 48.54018783569336,
      "learning_rate": 5.086372360844531e-07,
      "loss": 3.5943,
      "step": 53
    },
    {
      "epoch": 0.0015559269290612574,
      "grad_norm": 30.70709800720215,
      "learning_rate": 5.182341650671785e-07,
      "loss": 3.5712,
      "step": 54
    },
    {
      "epoch": 0.00158474039071054,
      "grad_norm": 27.70322608947754,
      "learning_rate": 5.278310940499041e-07,
      "loss": 3.4548,
      "step": 55
    },
    {
      "epoch": 0.0016135538523598225,
      "grad_norm": 27.545629501342773,
      "learning_rate": 5.374280230326296e-07,
      "loss": 3.4448,
      "step": 56
    },
    {
      "epoch": 0.001642367314009105,
      "grad_norm": 32.57813262939453,
      "learning_rate": 5.470249520153551e-07,
      "loss": 3.4361,
      "step": 57
    },
    {
      "epoch": 0.0016711807756583875,
      "grad_norm": 24.940082550048828,
      "learning_rate": 5.566218809980807e-07,
      "loss": 3.3386,
      "step": 58
    },
    {
      "epoch": 0.00169999423730767,
      "grad_norm": 25.836591720581055,
      "learning_rate": 5.662188099808063e-07,
      "loss": 3.3372,
      "step": 59
    },
    {
      "epoch": 0.0017288076989569526,
      "grad_norm": 23.280744552612305,
      "learning_rate": 5.758157389635317e-07,
      "loss": 3.3053,
      "step": 60
    },
    {
      "epoch": 0.0017576211606062351,
      "grad_norm": 22.836181640625,
      "learning_rate": 5.854126679462573e-07,
      "loss": 3.2204,
      "step": 61
    },
    {
      "epoch": 0.0017864346222555179,
      "grad_norm": 21.60952377319336,
      "learning_rate": 5.950095969289827e-07,
      "loss": 3.1671,
      "step": 62
    },
    {
      "epoch": 0.0018152480839048004,
      "grad_norm": 20.83951759338379,
      "learning_rate": 6.046065259117083e-07,
      "loss": 3.1211,
      "step": 63
    },
    {
      "epoch": 0.001844061545554083,
      "grad_norm": 20.25035285949707,
      "learning_rate": 6.142034548944338e-07,
      "loss": 3.1083,
      "step": 64
    },
    {
      "epoch": 0.0018728750072033655,
      "grad_norm": 19.11989974975586,
      "learning_rate": 6.238003838771593e-07,
      "loss": 3.0652,
      "step": 65
    },
    {
      "epoch": 0.001901688468852648,
      "grad_norm": 18.21695899963379,
      "learning_rate": 6.333973128598848e-07,
      "loss": 3.0465,
      "step": 66
    },
    {
      "epoch": 0.0019305019305019305,
      "grad_norm": 17.195859909057617,
      "learning_rate": 6.429942418426104e-07,
      "loss": 3.0141,
      "step": 67
    },
    {
      "epoch": 0.001959315392151213,
      "grad_norm": 16.13874053955078,
      "learning_rate": 6.525911708253358e-07,
      "loss": 2.9779,
      "step": 68
    },
    {
      "epoch": 0.0019881288538004956,
      "grad_norm": 15.462665557861328,
      "learning_rate": 6.621880998080615e-07,
      "loss": 2.9491,
      "step": 69
    },
    {
      "epoch": 0.002016942315449778,
      "grad_norm": 15.16762638092041,
      "learning_rate": 6.717850287907871e-07,
      "loss": 2.8895,
      "step": 70
    },
    {
      "epoch": 0.0020457557770990607,
      "grad_norm": 15.33432388305664,
      "learning_rate": 6.813819577735125e-07,
      "loss": 2.9187,
      "step": 71
    },
    {
      "epoch": 0.002074569238748343,
      "grad_norm": 14.326980590820312,
      "learning_rate": 6.909788867562381e-07,
      "loss": 2.8527,
      "step": 72
    },
    {
      "epoch": 0.0021033827003976257,
      "grad_norm": 16.74235725402832,
      "learning_rate": 7.005758157389636e-07,
      "loss": 2.8746,
      "step": 73
    },
    {
      "epoch": 0.0021321961620469083,
      "grad_norm": 14.358026504516602,
      "learning_rate": 7.101727447216891e-07,
      "loss": 2.8469,
      "step": 74
    },
    {
      "epoch": 0.002161009623696191,
      "grad_norm": 16.454999923706055,
      "learning_rate": 7.197696737044146e-07,
      "loss": 2.75,
      "step": 75
    },
    {
      "epoch": 0.0021898230853454733,
      "grad_norm": 13.891707420349121,
      "learning_rate": 7.293666026871402e-07,
      "loss": 2.7633,
      "step": 76
    },
    {
      "epoch": 0.002218636546994756,
      "grad_norm": 13.558817863464355,
      "learning_rate": 7.389635316698656e-07,
      "loss": 2.6867,
      "step": 77
    },
    {
      "epoch": 0.0022474500086440384,
      "grad_norm": 12.594690322875977,
      "learning_rate": 7.485604606525912e-07,
      "loss": 2.7404,
      "step": 78
    },
    {
      "epoch": 0.002276263470293321,
      "grad_norm": 12.360105514526367,
      "learning_rate": 7.581573896353168e-07,
      "loss": 2.6835,
      "step": 79
    },
    {
      "epoch": 0.0023050769319426035,
      "grad_norm": 12.262704849243164,
      "learning_rate": 7.677543186180422e-07,
      "loss": 2.6666,
      "step": 80
    },
    {
      "epoch": 0.002333890393591886,
      "grad_norm": 11.168907165527344,
      "learning_rate": 7.773512476007679e-07,
      "loss": 2.6549,
      "step": 81
    },
    {
      "epoch": 0.0023627038552411685,
      "grad_norm": 15.071958541870117,
      "learning_rate": 7.869481765834934e-07,
      "loss": 2.6088,
      "step": 82
    },
    {
      "epoch": 0.002391517316890451,
      "grad_norm": 10.61023235321045,
      "learning_rate": 7.965451055662189e-07,
      "loss": 2.6086,
      "step": 83
    },
    {
      "epoch": 0.0024203307785397336,
      "grad_norm": 10.382811546325684,
      "learning_rate": 8.061420345489445e-07,
      "loss": 2.6082,
      "step": 84
    },
    {
      "epoch": 0.002449144240189016,
      "grad_norm": 9.86429214477539,
      "learning_rate": 8.1573896353167e-07,
      "loss": 2.5762,
      "step": 85
    },
    {
      "epoch": 0.0024779577018382987,
      "grad_norm": 9.639535903930664,
      "learning_rate": 8.253358925143955e-07,
      "loss": 2.5037,
      "step": 86
    },
    {
      "epoch": 0.002506771163487581,
      "grad_norm": 9.68486499786377,
      "learning_rate": 8.34932821497121e-07,
      "loss": 2.4909,
      "step": 87
    },
    {
      "epoch": 0.0025355846251368637,
      "grad_norm": 9.800992965698242,
      "learning_rate": 8.445297504798465e-07,
      "loss": 2.52,
      "step": 88
    },
    {
      "epoch": 0.0025643980867861467,
      "grad_norm": 8.985952377319336,
      "learning_rate": 8.54126679462572e-07,
      "loss": 2.48,
      "step": 89
    },
    {
      "epoch": 0.0025932115484354292,
      "grad_norm": 9.001872062683105,
      "learning_rate": 8.637236084452976e-07,
      "loss": 2.496,
      "step": 90
    },
    {
      "epoch": 0.0026220250100847118,
      "grad_norm": 8.709939956665039,
      "learning_rate": 8.73320537428023e-07,
      "loss": 2.3926,
      "step": 91
    },
    {
      "epoch": 0.0026508384717339943,
      "grad_norm": 8.566285133361816,
      "learning_rate": 8.829174664107486e-07,
      "loss": 2.4361,
      "step": 92
    },
    {
      "epoch": 0.002679651933383277,
      "grad_norm": 8.798495292663574,
      "learning_rate": 8.925143953934741e-07,
      "loss": 2.3855,
      "step": 93
    },
    {
      "epoch": 0.0027084653950325594,
      "grad_norm": 8.181546211242676,
      "learning_rate": 9.021113243761997e-07,
      "loss": 2.3603,
      "step": 94
    },
    {
      "epoch": 0.002737278856681842,
      "grad_norm": 8.580350875854492,
      "learning_rate": 9.117082533589253e-07,
      "loss": 2.3815,
      "step": 95
    },
    {
      "epoch": 0.0027660923183311244,
      "grad_norm": 7.5677080154418945,
      "learning_rate": 9.213051823416508e-07,
      "loss": 2.3163,
      "step": 96
    },
    {
      "epoch": 0.002794905779980407,
      "grad_norm": 7.718907833099365,
      "learning_rate": 9.309021113243763e-07,
      "loss": 2.3376,
      "step": 97
    },
    {
      "epoch": 0.0028237192416296895,
      "grad_norm": 7.533500671386719,
      "learning_rate": 9.404990403071018e-07,
      "loss": 2.3625,
      "step": 98
    },
    {
      "epoch": 0.002852532703278972,
      "grad_norm": 7.070401191711426,
      "learning_rate": 9.500959692898274e-07,
      "loss": 2.3245,
      "step": 99
    },
    {
      "epoch": 0.0028813461649282546,
      "grad_norm": 7.3187689781188965,
      "learning_rate": 9.59692898272553e-07,
      "loss": 2.2952,
      "step": 100
    },
    {
      "epoch": 0.002910159626577537,
      "grad_norm": 6.968564033508301,
      "learning_rate": 9.692898272552784e-07,
      "loss": 2.2621,
      "step": 101
    },
    {
      "epoch": 0.0029389730882268196,
      "grad_norm": 6.989243984222412,
      "learning_rate": 9.788867562380039e-07,
      "loss": 2.2948,
      "step": 102
    },
    {
      "epoch": 0.002967786549876102,
      "grad_norm": 6.789176940917969,
      "learning_rate": 9.884836852207293e-07,
      "loss": 2.2524,
      "step": 103
    },
    {
      "epoch": 0.0029966000115253847,
      "grad_norm": 6.620153903961182,
      "learning_rate": 9.98080614203455e-07,
      "loss": 2.2408,
      "step": 104
    },
    {
      "epoch": 0.0030254134731746672,
      "grad_norm": 6.495826721191406,
      "learning_rate": 1.0076775431861805e-06,
      "loss": 2.2359,
      "step": 105
    },
    {
      "epoch": 0.0030542269348239497,
      "grad_norm": 6.528627872467041,
      "learning_rate": 1.0172744721689061e-06,
      "loss": 2.2111,
      "step": 106
    },
    {
      "epoch": 0.0030830403964732323,
      "grad_norm": 5.8867011070251465,
      "learning_rate": 1.0268714011516316e-06,
      "loss": 2.1887,
      "step": 107
    },
    {
      "epoch": 0.003111853858122515,
      "grad_norm": 6.31183385848999,
      "learning_rate": 1.036468330134357e-06,
      "loss": 2.1745,
      "step": 108
    },
    {
      "epoch": 0.0031406673197717973,
      "grad_norm": 6.060816764831543,
      "learning_rate": 1.0460652591170827e-06,
      "loss": 2.1904,
      "step": 109
    },
    {
      "epoch": 0.00316948078142108,
      "grad_norm": 6.278237342834473,
      "learning_rate": 1.0556621880998082e-06,
      "loss": 2.1627,
      "step": 110
    },
    {
      "epoch": 0.0031982942430703624,
      "grad_norm": 6.252457618713379,
      "learning_rate": 1.0652591170825336e-06,
      "loss": 2.1719,
      "step": 111
    },
    {
      "epoch": 0.003227107704719645,
      "grad_norm": 5.955358505249023,
      "learning_rate": 1.074856046065259e-06,
      "loss": 2.0864,
      "step": 112
    },
    {
      "epoch": 0.0032559211663689275,
      "grad_norm": 5.584120273590088,
      "learning_rate": 1.0844529750479848e-06,
      "loss": 2.1112,
      "step": 113
    },
    {
      "epoch": 0.00328473462801821,
      "grad_norm": 8.360549926757812,
      "learning_rate": 1.0940499040307102e-06,
      "loss": 2.1214,
      "step": 114
    },
    {
      "epoch": 0.0033135480896674925,
      "grad_norm": 6.740393161773682,
      "learning_rate": 1.1036468330134357e-06,
      "loss": 2.1094,
      "step": 115
    },
    {
      "epoch": 0.003342361551316775,
      "grad_norm": 5.46950101852417,
      "learning_rate": 1.1132437619961614e-06,
      "loss": 2.0672,
      "step": 116
    },
    {
      "epoch": 0.0033711750129660576,
      "grad_norm": 7.390514850616455,
      "learning_rate": 1.1228406909788868e-06,
      "loss": 2.0772,
      "step": 117
    },
    {
      "epoch": 0.00339998847461534,
      "grad_norm": 5.692412853240967,
      "learning_rate": 1.1324376199616125e-06,
      "loss": 2.0881,
      "step": 118
    },
    {
      "epoch": 0.0034288019362646227,
      "grad_norm": 5.542796611785889,
      "learning_rate": 1.142034548944338e-06,
      "loss": 2.0763,
      "step": 119
    },
    {
      "epoch": 0.003457615397913905,
      "grad_norm": 5.339426517486572,
      "learning_rate": 1.1516314779270634e-06,
      "loss": 2.081,
      "step": 120
    },
    {
      "epoch": 0.0034864288595631877,
      "grad_norm": 5.549520015716553,
      "learning_rate": 1.161228406909789e-06,
      "loss": 2.0691,
      "step": 121
    },
    {
      "epoch": 0.0035152423212124703,
      "grad_norm": 5.2649827003479,
      "learning_rate": 1.1708253358925146e-06,
      "loss": 2.0386,
      "step": 122
    },
    {
      "epoch": 0.003544055782861753,
      "grad_norm": 5.475005626678467,
      "learning_rate": 1.18042226487524e-06,
      "loss": 2.032,
      "step": 123
    },
    {
      "epoch": 0.0035728692445110358,
      "grad_norm": 5.597681522369385,
      "learning_rate": 1.1900191938579655e-06,
      "loss": 2.0528,
      "step": 124
    },
    {
      "epoch": 0.0036016827061603183,
      "grad_norm": 5.048548221588135,
      "learning_rate": 1.1996161228406912e-06,
      "loss": 1.9681,
      "step": 125
    },
    {
      "epoch": 0.003630496167809601,
      "grad_norm": 5.421810626983643,
      "learning_rate": 1.2092130518234166e-06,
      "loss": 2.0099,
      "step": 126
    },
    {
      "epoch": 0.0036593096294588834,
      "grad_norm": 4.958249568939209,
      "learning_rate": 1.218809980806142e-06,
      "loss": 2.0255,
      "step": 127
    },
    {
      "epoch": 0.003688123091108166,
      "grad_norm": 5.283166885375977,
      "learning_rate": 1.2284069097888675e-06,
      "loss": 1.9763,
      "step": 128
    },
    {
      "epoch": 0.0037169365527574484,
      "grad_norm": 4.698947906494141,
      "learning_rate": 1.2380038387715932e-06,
      "loss": 1.9903,
      "step": 129
    },
    {
      "epoch": 0.003745750014406731,
      "grad_norm": 4.986602306365967,
      "learning_rate": 1.2476007677543187e-06,
      "loss": 1.9702,
      "step": 130
    },
    {
      "epoch": 0.0037745634760560135,
      "grad_norm": 5.034524917602539,
      "learning_rate": 1.2571976967370441e-06,
      "loss": 1.9744,
      "step": 131
    },
    {
      "epoch": 0.003803376937705296,
      "grad_norm": 4.701071262359619,
      "learning_rate": 1.2667946257197696e-06,
      "loss": 1.9781,
      "step": 132
    },
    {
      "epoch": 0.0038321903993545786,
      "grad_norm": 4.587594509124756,
      "learning_rate": 1.2763915547024953e-06,
      "loss": 1.9573,
      "step": 133
    },
    {
      "epoch": 0.003861003861003861,
      "grad_norm": 5.363451957702637,
      "learning_rate": 1.2859884836852207e-06,
      "loss": 1.968,
      "step": 134
    },
    {
      "epoch": 0.0038898173226531436,
      "grad_norm": 4.7314677238464355,
      "learning_rate": 1.2955854126679462e-06,
      "loss": 1.9428,
      "step": 135
    },
    {
      "epoch": 0.003918630784302426,
      "grad_norm": 5.1231889724731445,
      "learning_rate": 1.3051823416506717e-06,
      "loss": 1.9619,
      "step": 136
    },
    {
      "epoch": 0.003947444245951708,
      "grad_norm": 4.786021709442139,
      "learning_rate": 1.3147792706333976e-06,
      "loss": 1.9382,
      "step": 137
    },
    {
      "epoch": 0.003976257707600991,
      "grad_norm": 4.723607540130615,
      "learning_rate": 1.324376199616123e-06,
      "loss": 1.9468,
      "step": 138
    },
    {
      "epoch": 0.004005071169250273,
      "grad_norm": 4.389206886291504,
      "learning_rate": 1.3339731285988487e-06,
      "loss": 1.9118,
      "step": 139
    },
    {
      "epoch": 0.004033884630899556,
      "grad_norm": 4.359829902648926,
      "learning_rate": 1.3435700575815741e-06,
      "loss": 1.9614,
      "step": 140
    },
    {
      "epoch": 0.004062698092548838,
      "grad_norm": 4.7029242515563965,
      "learning_rate": 1.3531669865642996e-06,
      "loss": 1.8764,
      "step": 141
    },
    {
      "epoch": 0.004091511554198121,
      "grad_norm": 4.839771270751953,
      "learning_rate": 1.362763915547025e-06,
      "loss": 1.874,
      "step": 142
    },
    {
      "epoch": 0.004120325015847404,
      "grad_norm": 4.690546035766602,
      "learning_rate": 1.3723608445297507e-06,
      "loss": 1.8908,
      "step": 143
    },
    {
      "epoch": 0.004149138477496686,
      "grad_norm": 4.4204277992248535,
      "learning_rate": 1.3819577735124762e-06,
      "loss": 1.889,
      "step": 144
    },
    {
      "epoch": 0.004177951939145969,
      "grad_norm": 4.098184585571289,
      "learning_rate": 1.3915547024952017e-06,
      "loss": 1.8798,
      "step": 145
    },
    {
      "epoch": 0.0042067654007952515,
      "grad_norm": 4.577674865722656,
      "learning_rate": 1.4011516314779271e-06,
      "loss": 1.841,
      "step": 146
    },
    {
      "epoch": 0.0042355788624445345,
      "grad_norm": 4.703514575958252,
      "learning_rate": 1.4107485604606528e-06,
      "loss": 1.8571,
      "step": 147
    },
    {
      "epoch": 0.0042643923240938165,
      "grad_norm": 4.1855058670043945,
      "learning_rate": 1.4203454894433783e-06,
      "loss": 1.8283,
      "step": 148
    },
    {
      "epoch": 0.0042932057857430995,
      "grad_norm": 4.685680389404297,
      "learning_rate": 1.4299424184261037e-06,
      "loss": 1.8138,
      "step": 149
    },
    {
      "epoch": 0.004322019247392382,
      "grad_norm": 4.713202953338623,
      "learning_rate": 1.4395393474088292e-06,
      "loss": 1.8592,
      "step": 150
    },
    {
      "epoch": 0.004350832709041665,
      "grad_norm": 3.8311493396759033,
      "learning_rate": 1.4491362763915549e-06,
      "loss": 1.8253,
      "step": 151
    },
    {
      "epoch": 0.004379646170690947,
      "grad_norm": 3.9864957332611084,
      "learning_rate": 1.4587332053742803e-06,
      "loss": 1.8082,
      "step": 152
    },
    {
      "epoch": 0.00440845963234023,
      "grad_norm": 5.195226192474365,
      "learning_rate": 1.4683301343570058e-06,
      "loss": 1.8326,
      "step": 153
    },
    {
      "epoch": 0.004437273093989512,
      "grad_norm": 3.8718371391296387,
      "learning_rate": 1.4779270633397312e-06,
      "loss": 1.8154,
      "step": 154
    },
    {
      "epoch": 0.004466086555638795,
      "grad_norm": 4.760035037994385,
      "learning_rate": 1.487523992322457e-06,
      "loss": 1.8162,
      "step": 155
    },
    {
      "epoch": 0.004494900017288077,
      "grad_norm": 4.285490989685059,
      "learning_rate": 1.4971209213051824e-06,
      "loss": 1.8319,
      "step": 156
    },
    {
      "epoch": 0.00452371347893736,
      "grad_norm": 3.9495737552642822,
      "learning_rate": 1.5067178502879078e-06,
      "loss": 1.8236,
      "step": 157
    },
    {
      "epoch": 0.004552526940586642,
      "grad_norm": 4.074175834655762,
      "learning_rate": 1.5163147792706335e-06,
      "loss": 1.8084,
      "step": 158
    },
    {
      "epoch": 0.004581340402235925,
      "grad_norm": 4.066157817840576,
      "learning_rate": 1.525911708253359e-06,
      "loss": 1.778,
      "step": 159
    },
    {
      "epoch": 0.004610153863885207,
      "grad_norm": 3.831474781036377,
      "learning_rate": 1.5355086372360844e-06,
      "loss": 1.7984,
      "step": 160
    },
    {
      "epoch": 0.00463896732553449,
      "grad_norm": 3.504338502883911,
      "learning_rate": 1.54510556621881e-06,
      "loss": 1.7765,
      "step": 161
    },
    {
      "epoch": 0.004667780787183772,
      "grad_norm": 4.14990234375,
      "learning_rate": 1.5547024952015358e-06,
      "loss": 1.789,
      "step": 162
    },
    {
      "epoch": 0.004696594248833055,
      "grad_norm": 3.979544162750244,
      "learning_rate": 1.5642994241842612e-06,
      "loss": 1.7297,
      "step": 163
    },
    {
      "epoch": 0.004725407710482337,
      "grad_norm": 3.6304426193237305,
      "learning_rate": 1.5738963531669867e-06,
      "loss": 1.7394,
      "step": 164
    },
    {
      "epoch": 0.00475422117213162,
      "grad_norm": 3.6876895427703857,
      "learning_rate": 1.5834932821497124e-06,
      "loss": 1.7804,
      "step": 165
    },
    {
      "epoch": 0.004783034633780902,
      "grad_norm": 4.428520679473877,
      "learning_rate": 1.5930902111324378e-06,
      "loss": 1.7567,
      "step": 166
    },
    {
      "epoch": 0.004811848095430185,
      "grad_norm": 4.08388090133667,
      "learning_rate": 1.6026871401151633e-06,
      "loss": 1.7395,
      "step": 167
    },
    {
      "epoch": 0.004840661557079467,
      "grad_norm": 3.896146535873413,
      "learning_rate": 1.612284069097889e-06,
      "loss": 1.7739,
      "step": 168
    },
    {
      "epoch": 0.00486947501872875,
      "grad_norm": 3.5500307083129883,
      "learning_rate": 1.6218809980806144e-06,
      "loss": 1.7225,
      "step": 169
    },
    {
      "epoch": 0.004898288480378032,
      "grad_norm": 3.5644984245300293,
      "learning_rate": 1.63147792706334e-06,
      "loss": 1.7213,
      "step": 170
    },
    {
      "epoch": 0.004927101942027315,
      "grad_norm": 3.6356892585754395,
      "learning_rate": 1.6410748560460654e-06,
      "loss": 1.7378,
      "step": 171
    },
    {
      "epoch": 0.004955915403676597,
      "grad_norm": 3.7830700874328613,
      "learning_rate": 1.650671785028791e-06,
      "loss": 1.7305,
      "step": 172
    },
    {
      "epoch": 0.00498472886532588,
      "grad_norm": 3.3788113594055176,
      "learning_rate": 1.6602687140115165e-06,
      "loss": 1.7172,
      "step": 173
    },
    {
      "epoch": 0.005013542326975162,
      "grad_norm": 4.303073883056641,
      "learning_rate": 1.669865642994242e-06,
      "loss": 1.7044,
      "step": 174
    },
    {
      "epoch": 0.005042355788624445,
      "grad_norm": 3.9875235557556152,
      "learning_rate": 1.6794625719769674e-06,
      "loss": 1.7079,
      "step": 175
    },
    {
      "epoch": 0.0050711692502737275,
      "grad_norm": 3.760216474533081,
      "learning_rate": 1.689059500959693e-06,
      "loss": 1.6989,
      "step": 176
    },
    {
      "epoch": 0.00509998271192301,
      "grad_norm": 3.624882936477661,
      "learning_rate": 1.6986564299424186e-06,
      "loss": 1.7176,
      "step": 177
    },
    {
      "epoch": 0.005128796173572293,
      "grad_norm": 3.6979854106903076,
      "learning_rate": 1.708253358925144e-06,
      "loss": 1.6994,
      "step": 178
    },
    {
      "epoch": 0.0051576096352215755,
      "grad_norm": 3.920368194580078,
      "learning_rate": 1.7178502879078695e-06,
      "loss": 1.704,
      "step": 179
    },
    {
      "epoch": 0.0051864230968708585,
      "grad_norm": 3.3504574298858643,
      "learning_rate": 1.7274472168905951e-06,
      "loss": 1.6733,
      "step": 180
    },
    {
      "epoch": 0.0052152365585201406,
      "grad_norm": 3.994112730026245,
      "learning_rate": 1.7370441458733206e-06,
      "loss": 1.6829,
      "step": 181
    },
    {
      "epoch": 0.0052440500201694235,
      "grad_norm": 4.10190486907959,
      "learning_rate": 1.746641074856046e-06,
      "loss": 1.6925,
      "step": 182
    },
    {
      "epoch": 0.005272863481818706,
      "grad_norm": 3.97344970703125,
      "learning_rate": 1.7562380038387715e-06,
      "loss": 1.7043,
      "step": 183
    },
    {
      "epoch": 0.005301676943467989,
      "grad_norm": 3.8960583209991455,
      "learning_rate": 1.7658349328214972e-06,
      "loss": 1.6959,
      "step": 184
    },
    {
      "epoch": 0.005330490405117271,
      "grad_norm": 3.723318576812744,
      "learning_rate": 1.7754318618042227e-06,
      "loss": 1.6995,
      "step": 185
    },
    {
      "epoch": 0.005359303866766554,
      "grad_norm": 4.884278297424316,
      "learning_rate": 1.7850287907869481e-06,
      "loss": 1.7033,
      "step": 186
    },
    {
      "epoch": 0.005388117328415836,
      "grad_norm": 3.547177791595459,
      "learning_rate": 1.794625719769674e-06,
      "loss": 1.7151,
      "step": 187
    },
    {
      "epoch": 0.005416930790065119,
      "grad_norm": 3.6756393909454346,
      "learning_rate": 1.8042226487523995e-06,
      "loss": 1.6664,
      "step": 188
    },
    {
      "epoch": 0.005445744251714401,
      "grad_norm": 3.5810539722442627,
      "learning_rate": 1.813819577735125e-06,
      "loss": 1.6698,
      "step": 189
    },
    {
      "epoch": 0.005474557713363684,
      "grad_norm": 3.3584296703338623,
      "learning_rate": 1.8234165067178506e-06,
      "loss": 1.62,
      "step": 190
    },
    {
      "epoch": 0.005503371175012966,
      "grad_norm": 3.6173720359802246,
      "learning_rate": 1.833013435700576e-06,
      "loss": 1.6748,
      "step": 191
    },
    {
      "epoch": 0.005532184636662249,
      "grad_norm": 3.2084712982177734,
      "learning_rate": 1.8426103646833015e-06,
      "loss": 1.6639,
      "step": 192
    },
    {
      "epoch": 0.005560998098311531,
      "grad_norm": 3.4262542724609375,
      "learning_rate": 1.852207293666027e-06,
      "loss": 1.6216,
      "step": 193
    },
    {
      "epoch": 0.005589811559960814,
      "grad_norm": 3.331244468688965,
      "learning_rate": 1.8618042226487527e-06,
      "loss": 1.6265,
      "step": 194
    },
    {
      "epoch": 0.005618625021610096,
      "grad_norm": 3.384286642074585,
      "learning_rate": 1.8714011516314781e-06,
      "loss": 1.6325,
      "step": 195
    },
    {
      "epoch": 0.005647438483259379,
      "grad_norm": 3.8790652751922607,
      "learning_rate": 1.8809980806142036e-06,
      "loss": 1.646,
      "step": 196
    },
    {
      "epoch": 0.005676251944908661,
      "grad_norm": 3.5226759910583496,
      "learning_rate": 1.890595009596929e-06,
      "loss": 1.605,
      "step": 197
    },
    {
      "epoch": 0.005705065406557944,
      "grad_norm": 3.5120389461517334,
      "learning_rate": 1.9001919385796547e-06,
      "loss": 1.643,
      "step": 198
    },
    {
      "epoch": 0.005733878868207226,
      "grad_norm": 3.593972682952881,
      "learning_rate": 1.90978886756238e-06,
      "loss": 1.6394,
      "step": 199
    },
    {
      "epoch": 0.005762692329856509,
      "grad_norm": 3.4546139240264893,
      "learning_rate": 1.919385796545106e-06,
      "loss": 1.6232,
      "step": 200
    },
    {
      "epoch": 0.005791505791505791,
      "grad_norm": 3.3177106380462646,
      "learning_rate": 1.928982725527831e-06,
      "loss": 1.6282,
      "step": 201
    },
    {
      "epoch": 0.005820319253155074,
      "grad_norm": 3.2646143436431885,
      "learning_rate": 1.9385796545105568e-06,
      "loss": 1.6195,
      "step": 202
    },
    {
      "epoch": 0.005849132714804356,
      "grad_norm": 3.1593902111053467,
      "learning_rate": 1.9481765834932825e-06,
      "loss": 1.607,
      "step": 203
    },
    {
      "epoch": 0.005877946176453639,
      "grad_norm": 3.3148562908172607,
      "learning_rate": 1.9577735124760077e-06,
      "loss": 1.6085,
      "step": 204
    },
    {
      "epoch": 0.005906759638102921,
      "grad_norm": 3.464528799057007,
      "learning_rate": 1.9673704414587334e-06,
      "loss": 1.581,
      "step": 205
    },
    {
      "epoch": 0.005935573099752204,
      "grad_norm": 3.2076122760772705,
      "learning_rate": 1.9769673704414586e-06,
      "loss": 1.6263,
      "step": 206
    },
    {
      "epoch": 0.005964386561401486,
      "grad_norm": 3.7937114238739014,
      "learning_rate": 1.9865642994241843e-06,
      "loss": 1.6223,
      "step": 207
    },
    {
      "epoch": 0.005993200023050769,
      "grad_norm": 3.2249655723571777,
      "learning_rate": 1.99616122840691e-06,
      "loss": 1.6043,
      "step": 208
    },
    {
      "epoch": 0.0060220134847000515,
      "grad_norm": 3.4997873306274414,
      "learning_rate": 2.0057581573896352e-06,
      "loss": 1.6279,
      "step": 209
    },
    {
      "epoch": 0.0060508269463493344,
      "grad_norm": 3.3583927154541016,
      "learning_rate": 2.015355086372361e-06,
      "loss": 1.5853,
      "step": 210
    },
    {
      "epoch": 0.0060796404079986165,
      "grad_norm": 3.661489486694336,
      "learning_rate": 2.0249520153550866e-06,
      "loss": 1.596,
      "step": 211
    },
    {
      "epoch": 0.0061084538696478995,
      "grad_norm": 3.0119338035583496,
      "learning_rate": 2.0345489443378122e-06,
      "loss": 1.5582,
      "step": 212
    },
    {
      "epoch": 0.0061372673312971825,
      "grad_norm": 3.3772499561309814,
      "learning_rate": 2.044145873320538e-06,
      "loss": 1.5681,
      "step": 213
    },
    {
      "epoch": 0.0061660807929464646,
      "grad_norm": 3.982987880706787,
      "learning_rate": 2.053742802303263e-06,
      "loss": 1.5968,
      "step": 214
    },
    {
      "epoch": 0.0061948942545957475,
      "grad_norm": 3.1331093311309814,
      "learning_rate": 2.063339731285989e-06,
      "loss": 1.5954,
      "step": 215
    },
    {
      "epoch": 0.00622370771624503,
      "grad_norm": 3.174027919769287,
      "learning_rate": 2.072936660268714e-06,
      "loss": 1.5815,
      "step": 216
    },
    {
      "epoch": 0.006252521177894313,
      "grad_norm": 3.2830405235290527,
      "learning_rate": 2.0825335892514398e-06,
      "loss": 1.5748,
      "step": 217
    },
    {
      "epoch": 0.006281334639543595,
      "grad_norm": 3.573338747024536,
      "learning_rate": 2.0921305182341654e-06,
      "loss": 1.5679,
      "step": 218
    },
    {
      "epoch": 0.006310148101192878,
      "grad_norm": 14.097940444946289,
      "learning_rate": 2.1017274472168907e-06,
      "loss": 1.5874,
      "step": 219
    },
    {
      "epoch": 0.00633896156284216,
      "grad_norm": 3.3481223583221436,
      "learning_rate": 2.1113243761996164e-06,
      "loss": 1.5658,
      "step": 220
    },
    {
      "epoch": 0.006367775024491443,
      "grad_norm": 3.993922472000122,
      "learning_rate": 2.120921305182342e-06,
      "loss": 1.5407,
      "step": 221
    },
    {
      "epoch": 0.006396588486140725,
      "grad_norm": 3.367459297180176,
      "learning_rate": 2.1305182341650673e-06,
      "loss": 1.5829,
      "step": 222
    },
    {
      "epoch": 0.006425401947790008,
      "grad_norm": 4.138100624084473,
      "learning_rate": 2.140115163147793e-06,
      "loss": 1.5571,
      "step": 223
    },
    {
      "epoch": 0.00645421540943929,
      "grad_norm": 3.7036855220794678,
      "learning_rate": 2.149712092130518e-06,
      "loss": 1.5208,
      "step": 224
    },
    {
      "epoch": 0.006483028871088573,
      "grad_norm": 3.639472723007202,
      "learning_rate": 2.159309021113244e-06,
      "loss": 1.5637,
      "step": 225
    },
    {
      "epoch": 0.006511842332737855,
      "grad_norm": 3.624945878982544,
      "learning_rate": 2.1689059500959696e-06,
      "loss": 1.5659,
      "step": 226
    },
    {
      "epoch": 0.006540655794387138,
      "grad_norm": 3.4496846199035645,
      "learning_rate": 2.178502879078695e-06,
      "loss": 1.5766,
      "step": 227
    },
    {
      "epoch": 0.00656946925603642,
      "grad_norm": 3.4234840869903564,
      "learning_rate": 2.1880998080614205e-06,
      "loss": 1.5408,
      "step": 228
    },
    {
      "epoch": 0.006598282717685703,
      "grad_norm": 3.2074778079986572,
      "learning_rate": 2.197696737044146e-06,
      "loss": 1.553,
      "step": 229
    },
    {
      "epoch": 0.006627096179334985,
      "grad_norm": 3.424532651901245,
      "learning_rate": 2.2072936660268714e-06,
      "loss": 1.5665,
      "step": 230
    },
    {
      "epoch": 0.006655909640984268,
      "grad_norm": 3.3604588508605957,
      "learning_rate": 2.216890595009597e-06,
      "loss": 1.5538,
      "step": 231
    },
    {
      "epoch": 0.00668472310263355,
      "grad_norm": 3.606464385986328,
      "learning_rate": 2.2264875239923228e-06,
      "loss": 1.5229,
      "step": 232
    },
    {
      "epoch": 0.006713536564282833,
      "grad_norm": 2.9673655033111572,
      "learning_rate": 2.236084452975048e-06,
      "loss": 1.5276,
      "step": 233
    },
    {
      "epoch": 0.006742350025932115,
      "grad_norm": 3.548318386077881,
      "learning_rate": 2.2456813819577737e-06,
      "loss": 1.54,
      "step": 234
    },
    {
      "epoch": 0.006771163487581398,
      "grad_norm": 3.249662160873413,
      "learning_rate": 2.255278310940499e-06,
      "loss": 1.52,
      "step": 235
    },
    {
      "epoch": 0.00679997694923068,
      "grad_norm": 3.1688225269317627,
      "learning_rate": 2.264875239923225e-06,
      "loss": 1.5165,
      "step": 236
    },
    {
      "epoch": 0.006828790410879963,
      "grad_norm": 3.785183906555176,
      "learning_rate": 2.2744721689059503e-06,
      "loss": 1.5144,
      "step": 237
    },
    {
      "epoch": 0.006857603872529245,
      "grad_norm": 3.0626754760742188,
      "learning_rate": 2.284069097888676e-06,
      "loss": 1.5207,
      "step": 238
    },
    {
      "epoch": 0.006886417334178528,
      "grad_norm": 4.2194647789001465,
      "learning_rate": 2.2936660268714016e-06,
      "loss": 1.5868,
      "step": 239
    },
    {
      "epoch": 0.00691523079582781,
      "grad_norm": 3.217144727706909,
      "learning_rate": 2.303262955854127e-06,
      "loss": 1.4901,
      "step": 240
    },
    {
      "epoch": 0.006944044257477093,
      "grad_norm": 3.2707879543304443,
      "learning_rate": 2.3128598848368525e-06,
      "loss": 1.5451,
      "step": 241
    },
    {
      "epoch": 0.0069728577191263755,
      "grad_norm": 3.2444162368774414,
      "learning_rate": 2.322456813819578e-06,
      "loss": 1.5149,
      "step": 242
    },
    {
      "epoch": 0.0070016711807756584,
      "grad_norm": 3.4618735313415527,
      "learning_rate": 2.3320537428023035e-06,
      "loss": 1.5015,
      "step": 243
    },
    {
      "epoch": 0.0070304846424249405,
      "grad_norm": 2.956789255142212,
      "learning_rate": 2.341650671785029e-06,
      "loss": 1.5139,
      "step": 244
    },
    {
      "epoch": 0.0070592981040742235,
      "grad_norm": 3.750650405883789,
      "learning_rate": 2.3512476007677544e-06,
      "loss": 1.518,
      "step": 245
    },
    {
      "epoch": 0.007088111565723506,
      "grad_norm": 5.788551330566406,
      "learning_rate": 2.36084452975048e-06,
      "loss": 1.5086,
      "step": 246
    },
    {
      "epoch": 0.007116925027372789,
      "grad_norm": 4.007157802581787,
      "learning_rate": 2.3704414587332057e-06,
      "loss": 1.4778,
      "step": 247
    },
    {
      "epoch": 0.0071457384890220715,
      "grad_norm": 3.3216845989227295,
      "learning_rate": 2.380038387715931e-06,
      "loss": 1.471,
      "step": 248
    },
    {
      "epoch": 0.007174551950671354,
      "grad_norm": 3.776609182357788,
      "learning_rate": 2.3896353166986567e-06,
      "loss": 1.5187,
      "step": 249
    },
    {
      "epoch": 0.007203365412320637,
      "grad_norm": 3.445645570755005,
      "learning_rate": 2.3992322456813823e-06,
      "loss": 1.5119,
      "step": 250
    },
    {
      "epoch": 0.007232178873969919,
      "grad_norm": 3.4455835819244385,
      "learning_rate": 2.4088291746641076e-06,
      "loss": 1.5009,
      "step": 251
    },
    {
      "epoch": 0.007260992335619202,
      "grad_norm": 3.4503016471862793,
      "learning_rate": 2.4184261036468333e-06,
      "loss": 1.4907,
      "step": 252
    },
    {
      "epoch": 0.007289805797268484,
      "grad_norm": 3.370840311050415,
      "learning_rate": 2.4280230326295585e-06,
      "loss": 1.4874,
      "step": 253
    },
    {
      "epoch": 0.007318619258917767,
      "grad_norm": 3.377690315246582,
      "learning_rate": 2.437619961612284e-06,
      "loss": 1.493,
      "step": 254
    },
    {
      "epoch": 0.007347432720567049,
      "grad_norm": 3.2508933544158936,
      "learning_rate": 2.44721689059501e-06,
      "loss": 1.5028,
      "step": 255
    },
    {
      "epoch": 0.007376246182216332,
      "grad_norm": 3.2561404705047607,
      "learning_rate": 2.456813819577735e-06,
      "loss": 1.5486,
      "step": 256
    },
    {
      "epoch": 0.007405059643865614,
      "grad_norm": 3.3471524715423584,
      "learning_rate": 2.4664107485604608e-06,
      "loss": 1.4912,
      "step": 257
    },
    {
      "epoch": 0.007433873105514897,
      "grad_norm": 3.344884157180786,
      "learning_rate": 2.4760076775431864e-06,
      "loss": 1.4945,
      "step": 258
    },
    {
      "epoch": 0.007462686567164179,
      "grad_norm": 3.53515887260437,
      "learning_rate": 2.4856046065259117e-06,
      "loss": 1.4914,
      "step": 259
    },
    {
      "epoch": 0.007491500028813462,
      "grad_norm": 3.4431896209716797,
      "learning_rate": 2.4952015355086374e-06,
      "loss": 1.4924,
      "step": 260
    },
    {
      "epoch": 0.007520313490462744,
      "grad_norm": 3.803395986557007,
      "learning_rate": 2.504798464491363e-06,
      "loss": 1.4574,
      "step": 261
    },
    {
      "epoch": 0.007549126952112027,
      "grad_norm": 3.372272491455078,
      "learning_rate": 2.5143953934740883e-06,
      "loss": 1.5141,
      "step": 262
    },
    {
      "epoch": 0.007577940413761309,
      "grad_norm": 4.255415439605713,
      "learning_rate": 2.523992322456814e-06,
      "loss": 1.481,
      "step": 263
    },
    {
      "epoch": 0.007606753875410592,
      "grad_norm": 3.076793909072876,
      "learning_rate": 2.5335892514395392e-06,
      "loss": 1.4878,
      "step": 264
    },
    {
      "epoch": 0.007635567337059874,
      "grad_norm": 11.185951232910156,
      "learning_rate": 2.5431861804222653e-06,
      "loss": 1.4634,
      "step": 265
    },
    {
      "epoch": 0.007664380798709157,
      "grad_norm": 3.8334898948669434,
      "learning_rate": 2.5527831094049906e-06,
      "loss": 1.4711,
      "step": 266
    },
    {
      "epoch": 0.007693194260358439,
      "grad_norm": 3.9812605381011963,
      "learning_rate": 2.5623800383877162e-06,
      "loss": 1.4798,
      "step": 267
    },
    {
      "epoch": 0.007722007722007722,
      "grad_norm": 3.6371583938598633,
      "learning_rate": 2.5719769673704415e-06,
      "loss": 1.4821,
      "step": 268
    },
    {
      "epoch": 0.007750821183657004,
      "grad_norm": 5.582980155944824,
      "learning_rate": 2.581573896353167e-06,
      "loss": 1.4488,
      "step": 269
    },
    {
      "epoch": 0.007779634645306287,
      "grad_norm": 3.1744184494018555,
      "learning_rate": 2.5911708253358924e-06,
      "loss": 1.4863,
      "step": 270
    },
    {
      "epoch": 0.007808448106955569,
      "grad_norm": 5.081371784210205,
      "learning_rate": 2.600767754318618e-06,
      "loss": 1.4231,
      "step": 271
    },
    {
      "epoch": 0.007837261568604852,
      "grad_norm": 3.2908504009246826,
      "learning_rate": 2.6103646833013433e-06,
      "loss": 1.476,
      "step": 272
    },
    {
      "epoch": 0.007866075030254134,
      "grad_norm": 5.916402339935303,
      "learning_rate": 2.6199616122840694e-06,
      "loss": 1.4528,
      "step": 273
    },
    {
      "epoch": 0.007894888491903417,
      "grad_norm": 3.246432304382324,
      "learning_rate": 2.629558541266795e-06,
      "loss": 1.4814,
      "step": 274
    },
    {
      "epoch": 0.0079237019535527,
      "grad_norm": 4.279084205627441,
      "learning_rate": 2.6391554702495203e-06,
      "loss": 1.4837,
      "step": 275
    },
    {
      "epoch": 0.007952515415201982,
      "grad_norm": 3.192239761352539,
      "learning_rate": 2.648752399232246e-06,
      "loss": 1.4576,
      "step": 276
    },
    {
      "epoch": 0.007981328876851265,
      "grad_norm": 4.053295612335205,
      "learning_rate": 2.6583493282149713e-06,
      "loss": 1.453,
      "step": 277
    },
    {
      "epoch": 0.008010142338500547,
      "grad_norm": 3.28044056892395,
      "learning_rate": 2.6679462571976974e-06,
      "loss": 1.449,
      "step": 278
    },
    {
      "epoch": 0.00803895580014983,
      "grad_norm": 3.6141326427459717,
      "learning_rate": 2.6775431861804226e-06,
      "loss": 1.4449,
      "step": 279
    },
    {
      "epoch": 0.008067769261799113,
      "grad_norm": 3.524606943130493,
      "learning_rate": 2.6871401151631483e-06,
      "loss": 1.4632,
      "step": 280
    },
    {
      "epoch": 0.008096582723448395,
      "grad_norm": 3.6006391048431396,
      "learning_rate": 2.6967370441458735e-06,
      "loss": 1.4405,
      "step": 281
    },
    {
      "epoch": 0.008125396185097677,
      "grad_norm": 3.474442720413208,
      "learning_rate": 2.7063339731285992e-06,
      "loss": 1.4463,
      "step": 282
    },
    {
      "epoch": 0.00815420964674696,
      "grad_norm": 3.3807079792022705,
      "learning_rate": 2.7159309021113245e-06,
      "loss": 1.4588,
      "step": 283
    },
    {
      "epoch": 0.008183023108396243,
      "grad_norm": 3.11946177482605,
      "learning_rate": 2.72552783109405e-06,
      "loss": 1.4522,
      "step": 284
    },
    {
      "epoch": 0.008211836570045525,
      "grad_norm": 3.3651387691497803,
      "learning_rate": 2.7351247600767754e-06,
      "loss": 1.4571,
      "step": 285
    },
    {
      "epoch": 0.008240650031694809,
      "grad_norm": 3.813322067260742,
      "learning_rate": 2.7447216890595015e-06,
      "loss": 1.4523,
      "step": 286
    },
    {
      "epoch": 0.00826946349334409,
      "grad_norm": 3.3042609691619873,
      "learning_rate": 2.7543186180422267e-06,
      "loss": 1.4551,
      "step": 287
    },
    {
      "epoch": 0.008298276954993373,
      "grad_norm": 4.1896138191223145,
      "learning_rate": 2.7639155470249524e-06,
      "loss": 1.4814,
      "step": 288
    },
    {
      "epoch": 0.008327090416642655,
      "grad_norm": 3.133228063583374,
      "learning_rate": 2.7735124760076777e-06,
      "loss": 1.413,
      "step": 289
    },
    {
      "epoch": 0.008355903878291939,
      "grad_norm": 4.002551078796387,
      "learning_rate": 2.7831094049904033e-06,
      "loss": 1.4396,
      "step": 290
    },
    {
      "epoch": 0.008384717339941221,
      "grad_norm": 3.7539901733398438,
      "learning_rate": 2.7927063339731286e-06,
      "loss": 1.4415,
      "step": 291
    },
    {
      "epoch": 0.008413530801590503,
      "grad_norm": 3.8312056064605713,
      "learning_rate": 2.8023032629558543e-06,
      "loss": 1.4232,
      "step": 292
    },
    {
      "epoch": 0.008442344263239785,
      "grad_norm": 3.3991506099700928,
      "learning_rate": 2.8119001919385795e-06,
      "loss": 1.4128,
      "step": 293
    },
    {
      "epoch": 0.008471157724889069,
      "grad_norm": 4.330965042114258,
      "learning_rate": 2.8214971209213056e-06,
      "loss": 1.4032,
      "step": 294
    },
    {
      "epoch": 0.008499971186538351,
      "grad_norm": 3.4113006591796875,
      "learning_rate": 2.831094049904031e-06,
      "loss": 1.4107,
      "step": 295
    },
    {
      "epoch": 0.008528784648187633,
      "grad_norm": 4.213365077972412,
      "learning_rate": 2.8406909788867565e-06,
      "loss": 1.4474,
      "step": 296
    },
    {
      "epoch": 0.008557598109836915,
      "grad_norm": 4.278400897979736,
      "learning_rate": 2.8502879078694818e-06,
      "loss": 1.4495,
      "step": 297
    },
    {
      "epoch": 0.008586411571486199,
      "grad_norm": 3.1214945316314697,
      "learning_rate": 2.8598848368522074e-06,
      "loss": 1.4125,
      "step": 298
    },
    {
      "epoch": 0.008615225033135481,
      "grad_norm": 4.176494121551514,
      "learning_rate": 2.869481765834933e-06,
      "loss": 1.4082,
      "step": 299
    },
    {
      "epoch": 0.008644038494784763,
      "grad_norm": 3.420680284500122,
      "learning_rate": 2.8790786948176584e-06,
      "loss": 1.4333,
      "step": 300
    },
    {
      "epoch": 0.008672851956434045,
      "grad_norm": 3.6708860397338867,
      "learning_rate": 2.8886756238003845e-06,
      "loss": 1.4135,
      "step": 301
    },
    {
      "epoch": 0.00870166541808333,
      "grad_norm": 3.236766815185547,
      "learning_rate": 2.8982725527831097e-06,
      "loss": 1.3939,
      "step": 302
    },
    {
      "epoch": 0.008730478879732611,
      "grad_norm": 4.002427101135254,
      "learning_rate": 2.9078694817658354e-06,
      "loss": 1.4348,
      "step": 303
    },
    {
      "epoch": 0.008759292341381893,
      "grad_norm": 3.3171024322509766,
      "learning_rate": 2.9174664107485606e-06,
      "loss": 1.4274,
      "step": 304
    },
    {
      "epoch": 0.008788105803031175,
      "grad_norm": 3.64874005317688,
      "learning_rate": 2.9270633397312863e-06,
      "loss": 1.4082,
      "step": 305
    },
    {
      "epoch": 0.00881691926468046,
      "grad_norm": 4.206424236297607,
      "learning_rate": 2.9366602687140116e-06,
      "loss": 1.4102,
      "step": 306
    },
    {
      "epoch": 0.008845732726329741,
      "grad_norm": 3.7235331535339355,
      "learning_rate": 2.9462571976967377e-06,
      "loss": 1.4127,
      "step": 307
    },
    {
      "epoch": 0.008874546187979023,
      "grad_norm": 3.9341118335723877,
      "learning_rate": 2.9558541266794625e-06,
      "loss": 1.4104,
      "step": 308
    },
    {
      "epoch": 0.008903359649628306,
      "grad_norm": 3.0694868564605713,
      "learning_rate": 2.9654510556621886e-06,
      "loss": 1.4217,
      "step": 309
    },
    {
      "epoch": 0.00893217311127759,
      "grad_norm": 4.663366317749023,
      "learning_rate": 2.975047984644914e-06,
      "loss": 1.3953,
      "step": 310
    },
    {
      "epoch": 0.008960986572926872,
      "grad_norm": 3.6300666332244873,
      "learning_rate": 2.9846449136276395e-06,
      "loss": 1.3867,
      "step": 311
    },
    {
      "epoch": 0.008989800034576154,
      "grad_norm": 4.309679985046387,
      "learning_rate": 2.9942418426103648e-06,
      "loss": 1.3701,
      "step": 312
    },
    {
      "epoch": 0.009018613496225436,
      "grad_norm": 3.9563188552856445,
      "learning_rate": 3.0038387715930904e-06,
      "loss": 1.3899,
      "step": 313
    },
    {
      "epoch": 0.00904742695787472,
      "grad_norm": 3.3346545696258545,
      "learning_rate": 3.0134357005758157e-06,
      "loss": 1.4047,
      "step": 314
    },
    {
      "epoch": 0.009076240419524002,
      "grad_norm": 3.8193447589874268,
      "learning_rate": 3.0230326295585418e-06,
      "loss": 1.3991,
      "step": 315
    },
    {
      "epoch": 0.009105053881173284,
      "grad_norm": 4.034125328063965,
      "learning_rate": 3.032629558541267e-06,
      "loss": 1.4128,
      "step": 316
    },
    {
      "epoch": 0.009133867342822566,
      "grad_norm": 4.238102912902832,
      "learning_rate": 3.0422264875239927e-06,
      "loss": 1.4,
      "step": 317
    },
    {
      "epoch": 0.00916268080447185,
      "grad_norm": 3.4034364223480225,
      "learning_rate": 3.051823416506718e-06,
      "loss": 1.3887,
      "step": 318
    },
    {
      "epoch": 0.009191494266121132,
      "grad_norm": 3.4041659832000732,
      "learning_rate": 3.0614203454894436e-06,
      "loss": 1.3957,
      "step": 319
    },
    {
      "epoch": 0.009220307727770414,
      "grad_norm": 4.028063774108887,
      "learning_rate": 3.071017274472169e-06,
      "loss": 1.3974,
      "step": 320
    },
    {
      "epoch": 0.009249121189419698,
      "grad_norm": 3.4820220470428467,
      "learning_rate": 3.0806142034548945e-06,
      "loss": 1.3661,
      "step": 321
    },
    {
      "epoch": 0.00927793465106898,
      "grad_norm": 4.306879997253418,
      "learning_rate": 3.09021113243762e-06,
      "loss": 1.3685,
      "step": 322
    },
    {
      "epoch": 0.009306748112718262,
      "grad_norm": 4.1203179359436035,
      "learning_rate": 3.099808061420346e-06,
      "loss": 1.3863,
      "step": 323
    },
    {
      "epoch": 0.009335561574367544,
      "grad_norm": 3.404514789581299,
      "learning_rate": 3.1094049904030716e-06,
      "loss": 1.3635,
      "step": 324
    },
    {
      "epoch": 0.009364375036016828,
      "grad_norm": 4.158893585205078,
      "learning_rate": 3.119001919385797e-06,
      "loss": 1.3984,
      "step": 325
    },
    {
      "epoch": 0.00939318849766611,
      "grad_norm": 3.2679450511932373,
      "learning_rate": 3.1285988483685225e-06,
      "loss": 1.3789,
      "step": 326
    },
    {
      "epoch": 0.009422001959315392,
      "grad_norm": 5.007554054260254,
      "learning_rate": 3.1381957773512477e-06,
      "loss": 1.3392,
      "step": 327
    },
    {
      "epoch": 0.009450815420964674,
      "grad_norm": 3.376283884048462,
      "learning_rate": 3.1477927063339734e-06,
      "loss": 1.377,
      "step": 328
    },
    {
      "epoch": 0.009479628882613958,
      "grad_norm": 16.59280776977539,
      "learning_rate": 3.1573896353166987e-06,
      "loss": 1.3873,
      "step": 329
    },
    {
      "epoch": 0.00950844234426324,
      "grad_norm": 3.582876682281494,
      "learning_rate": 3.1669865642994248e-06,
      "loss": 1.3964,
      "step": 330
    },
    {
      "epoch": 0.009537255805912522,
      "grad_norm": 5.270073413848877,
      "learning_rate": 3.17658349328215e-06,
      "loss": 1.4009,
      "step": 331
    },
    {
      "epoch": 0.009566069267561804,
      "grad_norm": 4.065816879272461,
      "learning_rate": 3.1861804222648757e-06,
      "loss": 1.3719,
      "step": 332
    },
    {
      "epoch": 0.009594882729211088,
      "grad_norm": 4.420246601104736,
      "learning_rate": 3.195777351247601e-06,
      "loss": 1.3495,
      "step": 333
    },
    {
      "epoch": 0.00962369619086037,
      "grad_norm": 4.093062877655029,
      "learning_rate": 3.2053742802303266e-06,
      "loss": 1.3766,
      "step": 334
    },
    {
      "epoch": 0.009652509652509652,
      "grad_norm": 4.439526081085205,
      "learning_rate": 3.214971209213052e-06,
      "loss": 1.3573,
      "step": 335
    },
    {
      "epoch": 0.009681323114158934,
      "grad_norm": 3.122976064682007,
      "learning_rate": 3.224568138195778e-06,
      "loss": 1.3746,
      "step": 336
    },
    {
      "epoch": 0.009710136575808218,
      "grad_norm": 3.666792869567871,
      "learning_rate": 3.2341650671785028e-06,
      "loss": 1.3926,
      "step": 337
    },
    {
      "epoch": 0.0097389500374575,
      "grad_norm": 4.189226150512695,
      "learning_rate": 3.243761996161229e-06,
      "loss": 1.3881,
      "step": 338
    },
    {
      "epoch": 0.009767763499106782,
      "grad_norm": 3.2715718746185303,
      "learning_rate": 3.253358925143954e-06,
      "loss": 1.3501,
      "step": 339
    },
    {
      "epoch": 0.009796576960756065,
      "grad_norm": 3.7821686267852783,
      "learning_rate": 3.26295585412668e-06,
      "loss": 1.3282,
      "step": 340
    },
    {
      "epoch": 0.009825390422405348,
      "grad_norm": 3.3289895057678223,
      "learning_rate": 3.272552783109405e-06,
      "loss": 1.3412,
      "step": 341
    },
    {
      "epoch": 0.00985420388405463,
      "grad_norm": 3.096008777618408,
      "learning_rate": 3.2821497120921307e-06,
      "loss": 1.3353,
      "step": 342
    },
    {
      "epoch": 0.009883017345703913,
      "grad_norm": 4.11373233795166,
      "learning_rate": 3.291746641074856e-06,
      "loss": 1.381,
      "step": 343
    },
    {
      "epoch": 0.009911830807353195,
      "grad_norm": 3.9812657833099365,
      "learning_rate": 3.301343570057582e-06,
      "loss": 1.3364,
      "step": 344
    },
    {
      "epoch": 0.009940644269002478,
      "grad_norm": 3.673691511154175,
      "learning_rate": 3.3109404990403073e-06,
      "loss": 1.3786,
      "step": 345
    },
    {
      "epoch": 0.00996945773065176,
      "grad_norm": 4.511928081512451,
      "learning_rate": 3.320537428023033e-06,
      "loss": 1.3773,
      "step": 346
    },
    {
      "epoch": 0.009998271192301043,
      "grad_norm": 3.8463685512542725,
      "learning_rate": 3.3301343570057582e-06,
      "loss": 1.3455,
      "step": 347
    },
    {
      "epoch": 0.010027084653950325,
      "grad_norm": 3.623436212539673,
      "learning_rate": 3.339731285988484e-06,
      "loss": 1.3838,
      "step": 348
    },
    {
      "epoch": 0.010055898115599609,
      "grad_norm": 4.223214149475098,
      "learning_rate": 3.3493282149712096e-06,
      "loss": 1.3847,
      "step": 349
    },
    {
      "epoch": 0.01008471157724889,
      "grad_norm": 3.6209192276000977,
      "learning_rate": 3.358925143953935e-06,
      "loss": 1.3547,
      "step": 350
    },
    {
      "epoch": 0.010113525038898173,
      "grad_norm": 4.649455547332764,
      "learning_rate": 3.368522072936661e-06,
      "loss": 1.3479,
      "step": 351
    },
    {
      "epoch": 0.010142338500547455,
      "grad_norm": 3.9059903621673584,
      "learning_rate": 3.378119001919386e-06,
      "loss": 1.3273,
      "step": 352
    },
    {
      "epoch": 0.010171151962196739,
      "grad_norm": 4.842840671539307,
      "learning_rate": 3.387715930902112e-06,
      "loss": 1.3771,
      "step": 353
    },
    {
      "epoch": 0.01019996542384602,
      "grad_norm": 3.7005083560943604,
      "learning_rate": 3.397312859884837e-06,
      "loss": 1.3505,
      "step": 354
    },
    {
      "epoch": 0.010228778885495303,
      "grad_norm": 5.891433238983154,
      "learning_rate": 3.4069097888675628e-06,
      "loss": 1.3687,
      "step": 355
    },
    {
      "epoch": 0.010257592347144587,
      "grad_norm": 3.605262279510498,
      "learning_rate": 3.416506717850288e-06,
      "loss": 1.3233,
      "step": 356
    },
    {
      "epoch": 0.010286405808793869,
      "grad_norm": 5.254359722137451,
      "learning_rate": 3.4261036468330137e-06,
      "loss": 1.3213,
      "step": 357
    },
    {
      "epoch": 0.010315219270443151,
      "grad_norm": 5.14846658706665,
      "learning_rate": 3.435700575815739e-06,
      "loss": 1.3562,
      "step": 358
    },
    {
      "epoch": 0.010344032732092433,
      "grad_norm": 3.9988021850585938,
      "learning_rate": 3.445297504798465e-06,
      "loss": 1.3242,
      "step": 359
    },
    {
      "epoch": 0.010372846193741717,
      "grad_norm": 5.781030178070068,
      "learning_rate": 3.4548944337811903e-06,
      "loss": 1.3548,
      "step": 360
    },
    {
      "epoch": 0.010401659655390999,
      "grad_norm": 3.3133456707000732,
      "learning_rate": 3.464491362763916e-06,
      "loss": 1.3522,
      "step": 361
    },
    {
      "epoch": 0.010430473117040281,
      "grad_norm": 4.164439678192139,
      "learning_rate": 3.4740882917466412e-06,
      "loss": 1.338,
      "step": 362
    },
    {
      "epoch": 0.010459286578689563,
      "grad_norm": 3.914935350418091,
      "learning_rate": 3.483685220729367e-06,
      "loss": 1.3356,
      "step": 363
    },
    {
      "epoch": 0.010488100040338847,
      "grad_norm": 3.2479591369628906,
      "learning_rate": 3.493282149712092e-06,
      "loss": 1.3387,
      "step": 364
    },
    {
      "epoch": 0.01051691350198813,
      "grad_norm": 4.598458290100098,
      "learning_rate": 3.502879078694818e-06,
      "loss": 1.3257,
      "step": 365
    },
    {
      "epoch": 0.010545726963637411,
      "grad_norm": 3.4657609462738037,
      "learning_rate": 3.512476007677543e-06,
      "loss": 1.3636,
      "step": 366
    },
    {
      "epoch": 0.010574540425286693,
      "grad_norm": 3.5615031719207764,
      "learning_rate": 3.522072936660269e-06,
      "loss": 1.3211,
      "step": 367
    },
    {
      "epoch": 0.010603353886935977,
      "grad_norm": 3.9387736320495605,
      "learning_rate": 3.5316698656429944e-06,
      "loss": 1.3492,
      "step": 368
    },
    {
      "epoch": 0.01063216734858526,
      "grad_norm": 3.5625531673431396,
      "learning_rate": 3.54126679462572e-06,
      "loss": 1.3326,
      "step": 369
    },
    {
      "epoch": 0.010660980810234541,
      "grad_norm": 4.333132266998291,
      "learning_rate": 3.5508637236084453e-06,
      "loss": 1.3176,
      "step": 370
    },
    {
      "epoch": 0.010689794271883823,
      "grad_norm": 3.5296123027801514,
      "learning_rate": 3.560460652591171e-06,
      "loss": 1.3106,
      "step": 371
    },
    {
      "epoch": 0.010718607733533107,
      "grad_norm": 4.257118225097656,
      "learning_rate": 3.5700575815738963e-06,
      "loss": 1.3539,
      "step": 372
    },
    {
      "epoch": 0.01074742119518239,
      "grad_norm": 3.876345157623291,
      "learning_rate": 3.5796545105566224e-06,
      "loss": 1.344,
      "step": 373
    },
    {
      "epoch": 0.010776234656831672,
      "grad_norm": 4.40437650680542,
      "learning_rate": 3.589251439539348e-06,
      "loss": 1.3337,
      "step": 374
    },
    {
      "epoch": 0.010805048118480954,
      "grad_norm": 3.381840944290161,
      "learning_rate": 3.5988483685220733e-06,
      "loss": 1.2922,
      "step": 375
    },
    {
      "epoch": 0.010833861580130237,
      "grad_norm": 3.6433093547821045,
      "learning_rate": 3.608445297504799e-06,
      "loss": 1.2938,
      "step": 376
    },
    {
      "epoch": 0.01086267504177952,
      "grad_norm": 3.6625397205352783,
      "learning_rate": 3.618042226487524e-06,
      "loss": 1.3465,
      "step": 377
    },
    {
      "epoch": 0.010891488503428802,
      "grad_norm": 3.1260716915130615,
      "learning_rate": 3.62763915547025e-06,
      "loss": 1.2815,
      "step": 378
    },
    {
      "epoch": 0.010920301965078084,
      "grad_norm": 3.585423707962036,
      "learning_rate": 3.637236084452975e-06,
      "loss": 1.3487,
      "step": 379
    },
    {
      "epoch": 0.010949115426727368,
      "grad_norm": 4.078736305236816,
      "learning_rate": 3.6468330134357012e-06,
      "loss": 1.2953,
      "step": 380
    },
    {
      "epoch": 0.01097792888837665,
      "grad_norm": 3.4087395668029785,
      "learning_rate": 3.6564299424184265e-06,
      "loss": 1.2553,
      "step": 381
    },
    {
      "epoch": 0.011006742350025932,
      "grad_norm": 5.096975803375244,
      "learning_rate": 3.666026871401152e-06,
      "loss": 1.3269,
      "step": 382
    },
    {
      "epoch": 0.011035555811675214,
      "grad_norm": 4.335935115814209,
      "learning_rate": 3.6756238003838774e-06,
      "loss": 1.3341,
      "step": 383
    },
    {
      "epoch": 0.011064369273324498,
      "grad_norm": 3.938096761703491,
      "learning_rate": 3.685220729366603e-06,
      "loss": 1.3476,
      "step": 384
    },
    {
      "epoch": 0.01109318273497378,
      "grad_norm": 4.000796318054199,
      "learning_rate": 3.6948176583493283e-06,
      "loss": 1.3211,
      "step": 385
    },
    {
      "epoch": 0.011121996196623062,
      "grad_norm": 4.390284061431885,
      "learning_rate": 3.704414587332054e-06,
      "loss": 1.3371,
      "step": 386
    },
    {
      "epoch": 0.011150809658272344,
      "grad_norm": 3.9382712841033936,
      "learning_rate": 3.7140115163147792e-06,
      "loss": 1.308,
      "step": 387
    },
    {
      "epoch": 0.011179623119921628,
      "grad_norm": 4.244975566864014,
      "learning_rate": 3.7236084452975053e-06,
      "loss": 1.3182,
      "step": 388
    },
    {
      "epoch": 0.01120843658157091,
      "grad_norm": 4.361526012420654,
      "learning_rate": 3.7332053742802306e-06,
      "loss": 1.3341,
      "step": 389
    },
    {
      "epoch": 0.011237250043220192,
      "grad_norm": 3.7903027534484863,
      "learning_rate": 3.7428023032629563e-06,
      "loss": 1.3014,
      "step": 390
    },
    {
      "epoch": 0.011266063504869476,
      "grad_norm": 4.752796173095703,
      "learning_rate": 3.7523992322456815e-06,
      "loss": 1.3249,
      "step": 391
    },
    {
      "epoch": 0.011294876966518758,
      "grad_norm": 4.11468505859375,
      "learning_rate": 3.761996161228407e-06,
      "loss": 1.3095,
      "step": 392
    },
    {
      "epoch": 0.01132369042816804,
      "grad_norm": 4.984833240509033,
      "learning_rate": 3.7715930902111324e-06,
      "loss": 1.3176,
      "step": 393
    },
    {
      "epoch": 0.011352503889817322,
      "grad_norm": 3.8340702056884766,
      "learning_rate": 3.781190019193858e-06,
      "loss": 1.2998,
      "step": 394
    },
    {
      "epoch": 0.011381317351466606,
      "grad_norm": 4.645700931549072,
      "learning_rate": 3.7907869481765834e-06,
      "loss": 1.3217,
      "step": 395
    },
    {
      "epoch": 0.011410130813115888,
      "grad_norm": 4.584367275238037,
      "learning_rate": 3.8003838771593095e-06,
      "loss": 1.3309,
      "step": 396
    },
    {
      "epoch": 0.01143894427476517,
      "grad_norm": 4.080788612365723,
      "learning_rate": 3.809980806142035e-06,
      "loss": 1.314,
      "step": 397
    },
    {
      "epoch": 0.011467757736414452,
      "grad_norm": 3.949188470840454,
      "learning_rate": 3.81957773512476e-06,
      "loss": 1.3339,
      "step": 398
    },
    {
      "epoch": 0.011496571198063736,
      "grad_norm": 3.7570700645446777,
      "learning_rate": 3.829174664107486e-06,
      "loss": 1.3229,
      "step": 399
    },
    {
      "epoch": 0.011525384659713018,
      "grad_norm": 4.891308784484863,
      "learning_rate": 3.838771593090212e-06,
      "loss": 1.2919,
      "step": 400
    },
    {
      "epoch": 0.0115541981213623,
      "grad_norm": 4.01744270324707,
      "learning_rate": 3.848368522072937e-06,
      "loss": 1.321,
      "step": 401
    },
    {
      "epoch": 0.011583011583011582,
      "grad_norm": 4.678388595581055,
      "learning_rate": 3.857965451055662e-06,
      "loss": 1.3116,
      "step": 402
    },
    {
      "epoch": 0.011611825044660866,
      "grad_norm": 3.6857187747955322,
      "learning_rate": 3.867562380038388e-06,
      "loss": 1.295,
      "step": 403
    },
    {
      "epoch": 0.011640638506310148,
      "grad_norm": 5.016856670379639,
      "learning_rate": 3.8771593090211136e-06,
      "loss": 1.3062,
      "step": 404
    },
    {
      "epoch": 0.01166945196795943,
      "grad_norm": 4.675918102264404,
      "learning_rate": 3.886756238003839e-06,
      "loss": 1.3143,
      "step": 405
    },
    {
      "epoch": 0.011698265429608713,
      "grad_norm": 3.9940013885498047,
      "learning_rate": 3.896353166986565e-06,
      "loss": 1.284,
      "step": 406
    },
    {
      "epoch": 0.011727078891257996,
      "grad_norm": 4.781159400939941,
      "learning_rate": 3.905950095969291e-06,
      "loss": 1.3378,
      "step": 407
    },
    {
      "epoch": 0.011755892352907278,
      "grad_norm": 4.842870235443115,
      "learning_rate": 3.915547024952015e-06,
      "loss": 1.3275,
      "step": 408
    },
    {
      "epoch": 0.01178470581455656,
      "grad_norm": 5.0227437019348145,
      "learning_rate": 3.925143953934741e-06,
      "loss": 1.2943,
      "step": 409
    },
    {
      "epoch": 0.011813519276205843,
      "grad_norm": 4.600337028503418,
      "learning_rate": 3.934740882917467e-06,
      "loss": 1.3,
      "step": 410
    },
    {
      "epoch": 0.011842332737855127,
      "grad_norm": 4.778714656829834,
      "learning_rate": 3.9443378119001924e-06,
      "loss": 1.2786,
      "step": 411
    },
    {
      "epoch": 0.011871146199504409,
      "grad_norm": 3.853773593902588,
      "learning_rate": 3.953934740882917e-06,
      "loss": 1.2966,
      "step": 412
    },
    {
      "epoch": 0.01189995966115369,
      "grad_norm": 5.534796714782715,
      "learning_rate": 3.963531669865644e-06,
      "loss": 1.2957,
      "step": 413
    },
    {
      "epoch": 0.011928773122802973,
      "grad_norm": 3.768503189086914,
      "learning_rate": 3.973128598848369e-06,
      "loss": 1.3066,
      "step": 414
    },
    {
      "epoch": 0.011957586584452257,
      "grad_norm": 4.858711242675781,
      "learning_rate": 3.982725527831094e-06,
      "loss": 1.3147,
      "step": 415
    },
    {
      "epoch": 0.011986400046101539,
      "grad_norm": 4.340757369995117,
      "learning_rate": 3.99232245681382e-06,
      "loss": 1.2992,
      "step": 416
    },
    {
      "epoch": 0.01201521350775082,
      "grad_norm": 4.572978973388672,
      "learning_rate": 4.001919385796546e-06,
      "loss": 1.292,
      "step": 417
    },
    {
      "epoch": 0.012044026969400103,
      "grad_norm": 4.249795436859131,
      "learning_rate": 4.0115163147792705e-06,
      "loss": 1.3032,
      "step": 418
    },
    {
      "epoch": 0.012072840431049387,
      "grad_norm": 4.5735883712768555,
      "learning_rate": 4.021113243761996e-06,
      "loss": 1.3173,
      "step": 419
    },
    {
      "epoch": 0.012101653892698669,
      "grad_norm": 3.8514244556427,
      "learning_rate": 4.030710172744722e-06,
      "loss": 1.3123,
      "step": 420
    },
    {
      "epoch": 0.012130467354347951,
      "grad_norm": 4.700437068939209,
      "learning_rate": 4.0403071017274475e-06,
      "loss": 1.3323,
      "step": 421
    },
    {
      "epoch": 0.012159280815997233,
      "grad_norm": 3.7765309810638428,
      "learning_rate": 4.049904030710173e-06,
      "loss": 1.3102,
      "step": 422
    },
    {
      "epoch": 0.012188094277646517,
      "grad_norm": 4.028764247894287,
      "learning_rate": 4.059500959692899e-06,
      "loss": 1.3015,
      "step": 423
    },
    {
      "epoch": 0.012216907739295799,
      "grad_norm": 3.577998638153076,
      "learning_rate": 4.0690978886756245e-06,
      "loss": 1.2625,
      "step": 424
    },
    {
      "epoch": 0.012245721200945081,
      "grad_norm": 4.202712059020996,
      "learning_rate": 4.078694817658349e-06,
      "loss": 1.2715,
      "step": 425
    },
    {
      "epoch": 0.012274534662594365,
      "grad_norm": 3.57448673248291,
      "learning_rate": 4.088291746641076e-06,
      "loss": 1.273,
      "step": 426
    },
    {
      "epoch": 0.012303348124243647,
      "grad_norm": 4.0360107421875,
      "learning_rate": 4.097888675623801e-06,
      "loss": 1.2739,
      "step": 427
    },
    {
      "epoch": 0.012332161585892929,
      "grad_norm": 4.176298141479492,
      "learning_rate": 4.107485604606526e-06,
      "loss": 1.2997,
      "step": 428
    },
    {
      "epoch": 0.012360975047542211,
      "grad_norm": 3.9365711212158203,
      "learning_rate": 4.117082533589252e-06,
      "loss": 1.2841,
      "step": 429
    },
    {
      "epoch": 0.012389788509191495,
      "grad_norm": 4.678027153015137,
      "learning_rate": 4.126679462571978e-06,
      "loss": 1.3016,
      "step": 430
    },
    {
      "epoch": 0.012418601970840777,
      "grad_norm": 4.019702434539795,
      "learning_rate": 4.1362763915547025e-06,
      "loss": 1.3063,
      "step": 431
    },
    {
      "epoch": 0.01244741543249006,
      "grad_norm": 4.830901145935059,
      "learning_rate": 4.145873320537428e-06,
      "loss": 1.2917,
      "step": 432
    },
    {
      "epoch": 0.012476228894139341,
      "grad_norm": 4.380056381225586,
      "learning_rate": 4.155470249520154e-06,
      "loss": 1.2663,
      "step": 433
    },
    {
      "epoch": 0.012505042355788625,
      "grad_norm": 4.664355754852295,
      "learning_rate": 4.1650671785028795e-06,
      "loss": 1.3031,
      "step": 434
    },
    {
      "epoch": 0.012533855817437907,
      "grad_norm": 4.4214982986450195,
      "learning_rate": 4.174664107485605e-06,
      "loss": 1.2976,
      "step": 435
    },
    {
      "epoch": 0.01256266927908719,
      "grad_norm": 4.042685031890869,
      "learning_rate": 4.184261036468331e-06,
      "loss": 1.2867,
      "step": 436
    },
    {
      "epoch": 0.012591482740736471,
      "grad_norm": 4.068833827972412,
      "learning_rate": 4.193857965451056e-06,
      "loss": 1.2737,
      "step": 437
    },
    {
      "epoch": 0.012620296202385755,
      "grad_norm": 4.8223042488098145,
      "learning_rate": 4.203454894433781e-06,
      "loss": 1.3005,
      "step": 438
    },
    {
      "epoch": 0.012649109664035037,
      "grad_norm": 4.331967353820801,
      "learning_rate": 4.213051823416507e-06,
      "loss": 1.3105,
      "step": 439
    },
    {
      "epoch": 0.01267792312568432,
      "grad_norm": 3.904505968093872,
      "learning_rate": 4.222648752399233e-06,
      "loss": 1.2623,
      "step": 440
    },
    {
      "epoch": 0.012706736587333602,
      "grad_norm": 3.804517984390259,
      "learning_rate": 4.2322456813819576e-06,
      "loss": 1.277,
      "step": 441
    },
    {
      "epoch": 0.012735550048982885,
      "grad_norm": 5.258599281311035,
      "learning_rate": 4.241842610364684e-06,
      "loss": 1.2962,
      "step": 442
    },
    {
      "epoch": 0.012764363510632168,
      "grad_norm": 3.947319507598877,
      "learning_rate": 4.251439539347409e-06,
      "loss": 1.2682,
      "step": 443
    },
    {
      "epoch": 0.01279317697228145,
      "grad_norm": 4.99240779876709,
      "learning_rate": 4.2610364683301346e-06,
      "loss": 1.2662,
      "step": 444
    },
    {
      "epoch": 0.012821990433930732,
      "grad_norm": 4.211867809295654,
      "learning_rate": 4.27063339731286e-06,
      "loss": 1.2674,
      "step": 445
    },
    {
      "epoch": 0.012850803895580016,
      "grad_norm": 6.796544075012207,
      "learning_rate": 4.280230326295586e-06,
      "loss": 1.2673,
      "step": 446
    },
    {
      "epoch": 0.012879617357229298,
      "grad_norm": 3.9619479179382324,
      "learning_rate": 4.289827255278312e-06,
      "loss": 1.2958,
      "step": 447
    },
    {
      "epoch": 0.01290843081887858,
      "grad_norm": 5.036623477935791,
      "learning_rate": 4.299424184261036e-06,
      "loss": 1.2586,
      "step": 448
    },
    {
      "epoch": 0.012937244280527862,
      "grad_norm": 3.9598751068115234,
      "learning_rate": 4.309021113243763e-06,
      "loss": 1.286,
      "step": 449
    },
    {
      "epoch": 0.012966057742177146,
      "grad_norm": 4.980327606201172,
      "learning_rate": 4.318618042226488e-06,
      "loss": 1.313,
      "step": 450
    },
    {
      "epoch": 0.012994871203826428,
      "grad_norm": 4.223171710968018,
      "learning_rate": 4.3282149712092134e-06,
      "loss": 1.2708,
      "step": 451
    },
    {
      "epoch": 0.01302368466547571,
      "grad_norm": 3.993025302886963,
      "learning_rate": 4.337811900191939e-06,
      "loss": 1.2886,
      "step": 452
    },
    {
      "epoch": 0.013052498127124992,
      "grad_norm": 4.625014781951904,
      "learning_rate": 4.347408829174665e-06,
      "loss": 1.2566,
      "step": 453
    },
    {
      "epoch": 0.013081311588774276,
      "grad_norm": 3.445340633392334,
      "learning_rate": 4.35700575815739e-06,
      "loss": 1.2979,
      "step": 454
    },
    {
      "epoch": 0.013110125050423558,
      "grad_norm": 4.650182723999023,
      "learning_rate": 4.366602687140115e-06,
      "loss": 1.2765,
      "step": 455
    },
    {
      "epoch": 0.01313893851207284,
      "grad_norm": 4.386460304260254,
      "learning_rate": 4.376199616122841e-06,
      "loss": 1.2741,
      "step": 456
    },
    {
      "epoch": 0.013167751973722122,
      "grad_norm": 4.7678351402282715,
      "learning_rate": 4.385796545105567e-06,
      "loss": 1.2656,
      "step": 457
    },
    {
      "epoch": 0.013196565435371406,
      "grad_norm": 5.154484748840332,
      "learning_rate": 4.395393474088292e-06,
      "loss": 1.2706,
      "step": 458
    },
    {
      "epoch": 0.013225378897020688,
      "grad_norm": 4.001853942871094,
      "learning_rate": 4.404990403071018e-06,
      "loss": 1.2753,
      "step": 459
    },
    {
      "epoch": 0.01325419235866997,
      "grad_norm": 4.647881984710693,
      "learning_rate": 4.414587332053743e-06,
      "loss": 1.26,
      "step": 460
    },
    {
      "epoch": 0.013283005820319254,
      "grad_norm": 3.9682528972625732,
      "learning_rate": 4.4241842610364685e-06,
      "loss": 1.2802,
      "step": 461
    },
    {
      "epoch": 0.013311819281968536,
      "grad_norm": 5.291789531707764,
      "learning_rate": 4.433781190019194e-06,
      "loss": 1.2763,
      "step": 462
    },
    {
      "epoch": 0.013340632743617818,
      "grad_norm": 4.375705718994141,
      "learning_rate": 4.44337811900192e-06,
      "loss": 1.3041,
      "step": 463
    },
    {
      "epoch": 0.0133694462052671,
      "grad_norm": 4.467386722564697,
      "learning_rate": 4.4529750479846455e-06,
      "loss": 1.28,
      "step": 464
    },
    {
      "epoch": 0.013398259666916384,
      "grad_norm": 4.644187927246094,
      "learning_rate": 4.462571976967371e-06,
      "loss": 1.281,
      "step": 465
    },
    {
      "epoch": 0.013427073128565666,
      "grad_norm": 4.157880783081055,
      "learning_rate": 4.472168905950096e-06,
      "loss": 1.2466,
      "step": 466
    },
    {
      "epoch": 0.013455886590214948,
      "grad_norm": 3.9553136825561523,
      "learning_rate": 4.481765834932822e-06,
      "loss": 1.2947,
      "step": 467
    },
    {
      "epoch": 0.01348470005186423,
      "grad_norm": 4.1680827140808105,
      "learning_rate": 4.491362763915547e-06,
      "loss": 1.2846,
      "step": 468
    },
    {
      "epoch": 0.013513513513513514,
      "grad_norm": 3.911255121231079,
      "learning_rate": 4.500959692898273e-06,
      "loss": 1.2615,
      "step": 469
    },
    {
      "epoch": 0.013542326975162796,
      "grad_norm": 4.436634540557861,
      "learning_rate": 4.510556621880998e-06,
      "loss": 1.276,
      "step": 470
    },
    {
      "epoch": 0.013571140436812078,
      "grad_norm": 3.74851393699646,
      "learning_rate": 4.520153550863724e-06,
      "loss": 1.2431,
      "step": 471
    },
    {
      "epoch": 0.01359995389846136,
      "grad_norm": 4.90783166885376,
      "learning_rate": 4.52975047984645e-06,
      "loss": 1.2423,
      "step": 472
    },
    {
      "epoch": 0.013628767360110644,
      "grad_norm": 3.8984756469726562,
      "learning_rate": 4.539347408829175e-06,
      "loss": 1.292,
      "step": 473
    },
    {
      "epoch": 0.013657580821759926,
      "grad_norm": 4.700891017913818,
      "learning_rate": 4.5489443378119005e-06,
      "loss": 1.2514,
      "step": 474
    },
    {
      "epoch": 0.013686394283409209,
      "grad_norm": 3.8666038513183594,
      "learning_rate": 4.558541266794626e-06,
      "loss": 1.255,
      "step": 475
    },
    {
      "epoch": 0.01371520774505849,
      "grad_norm": 4.285643577575684,
      "learning_rate": 4.568138195777352e-06,
      "loss": 1.2633,
      "step": 476
    },
    {
      "epoch": 0.013744021206707775,
      "grad_norm": 4.832088470458984,
      "learning_rate": 4.577735124760077e-06,
      "loss": 1.2879,
      "step": 477
    },
    {
      "epoch": 0.013772834668357057,
      "grad_norm": 4.673577308654785,
      "learning_rate": 4.587332053742803e-06,
      "loss": 1.2455,
      "step": 478
    },
    {
      "epoch": 0.013801648130006339,
      "grad_norm": 4.528984546661377,
      "learning_rate": 4.596928982725528e-06,
      "loss": 1.268,
      "step": 479
    },
    {
      "epoch": 0.01383046159165562,
      "grad_norm": 4.853038311004639,
      "learning_rate": 4.606525911708254e-06,
      "loss": 1.2875,
      "step": 480
    },
    {
      "epoch": 0.013859275053304905,
      "grad_norm": 4.341845989227295,
      "learning_rate": 4.616122840690979e-06,
      "loss": 1.2611,
      "step": 481
    },
    {
      "epoch": 0.013888088514954187,
      "grad_norm": 5.2132391929626465,
      "learning_rate": 4.625719769673705e-06,
      "loss": 1.2654,
      "step": 482
    },
    {
      "epoch": 0.013916901976603469,
      "grad_norm": 4.263676643371582,
      "learning_rate": 4.63531669865643e-06,
      "loss": 1.2466,
      "step": 483
    },
    {
      "epoch": 0.013945715438252751,
      "grad_norm": 3.854499340057373,
      "learning_rate": 4.644913627639156e-06,
      "loss": 1.2418,
      "step": 484
    },
    {
      "epoch": 0.013974528899902035,
      "grad_norm": 4.187376499176025,
      "learning_rate": 4.654510556621881e-06,
      "loss": 1.2584,
      "step": 485
    },
    {
      "epoch": 0.014003342361551317,
      "grad_norm": 4.422837257385254,
      "learning_rate": 4.664107485604607e-06,
      "loss": 1.2685,
      "step": 486
    },
    {
      "epoch": 0.014032155823200599,
      "grad_norm": 3.7106916904449463,
      "learning_rate": 4.673704414587333e-06,
      "loss": 1.2254,
      "step": 487
    },
    {
      "epoch": 0.014060969284849881,
      "grad_norm": 4.536411285400391,
      "learning_rate": 4.683301343570058e-06,
      "loss": 1.278,
      "step": 488
    },
    {
      "epoch": 0.014089782746499165,
      "grad_norm": 3.9828853607177734,
      "learning_rate": 4.692898272552783e-06,
      "loss": 1.2617,
      "step": 489
    },
    {
      "epoch": 0.014118596208148447,
      "grad_norm": 5.361243724822998,
      "learning_rate": 4.702495201535509e-06,
      "loss": 1.2594,
      "step": 490
    },
    {
      "epoch": 0.014147409669797729,
      "grad_norm": 4.82240629196167,
      "learning_rate": 4.7120921305182344e-06,
      "loss": 1.2562,
      "step": 491
    },
    {
      "epoch": 0.014176223131447011,
      "grad_norm": 4.696268558502197,
      "learning_rate": 4.72168905950096e-06,
      "loss": 1.2628,
      "step": 492
    },
    {
      "epoch": 0.014205036593096295,
      "grad_norm": 3.9960832595825195,
      "learning_rate": 4.731285988483685e-06,
      "loss": 1.2515,
      "step": 493
    },
    {
      "epoch": 0.014233850054745577,
      "grad_norm": 5.010351657867432,
      "learning_rate": 4.7408829174664115e-06,
      "loss": 1.2528,
      "step": 494
    },
    {
      "epoch": 0.01426266351639486,
      "grad_norm": 4.079093933105469,
      "learning_rate": 4.750479846449136e-06,
      "loss": 1.2561,
      "step": 495
    },
    {
      "epoch": 0.014291476978044143,
      "grad_norm": 4.793771743774414,
      "learning_rate": 4.760076775431862e-06,
      "loss": 1.2589,
      "step": 496
    },
    {
      "epoch": 0.014320290439693425,
      "grad_norm": 4.2354326248168945,
      "learning_rate": 4.769673704414588e-06,
      "loss": 1.2607,
      "step": 497
    },
    {
      "epoch": 0.014349103901342707,
      "grad_norm": 4.464824199676514,
      "learning_rate": 4.779270633397313e-06,
      "loss": 1.2734,
      "step": 498
    },
    {
      "epoch": 0.01437791736299199,
      "grad_norm": 4.474306106567383,
      "learning_rate": 4.788867562380039e-06,
      "loss": 1.2565,
      "step": 499
    },
    {
      "epoch": 0.014406730824641273,
      "grad_norm": 3.836209774017334,
      "learning_rate": 4.798464491362765e-06,
      "loss": 1.2685,
      "step": 500
    },
    {
      "epoch": 0.014435544286290555,
      "grad_norm": 5.621946334838867,
      "learning_rate": 4.80806142034549e-06,
      "loss": 1.2297,
      "step": 501
    },
    {
      "epoch": 0.014464357747939837,
      "grad_norm": 4.531138896942139,
      "learning_rate": 4.817658349328215e-06,
      "loss": 1.2716,
      "step": 502
    },
    {
      "epoch": 0.01449317120958912,
      "grad_norm": 4.887336730957031,
      "learning_rate": 4.827255278310941e-06,
      "loss": 1.2285,
      "step": 503
    },
    {
      "epoch": 0.014521984671238403,
      "grad_norm": 4.279893398284912,
      "learning_rate": 4.8368522072936665e-06,
      "loss": 1.2539,
      "step": 504
    },
    {
      "epoch": 0.014550798132887685,
      "grad_norm": 4.859304428100586,
      "learning_rate": 4.846449136276392e-06,
      "loss": 1.2493,
      "step": 505
    },
    {
      "epoch": 0.014579611594536968,
      "grad_norm": 4.290686130523682,
      "learning_rate": 4.856046065259117e-06,
      "loss": 1.2625,
      "step": 506
    },
    {
      "epoch": 0.01460842505618625,
      "grad_norm": 4.452437400817871,
      "learning_rate": 4.8656429942418435e-06,
      "loss": 1.2529,
      "step": 507
    },
    {
      "epoch": 0.014637238517835533,
      "grad_norm": 4.366687774658203,
      "learning_rate": 4.875239923224568e-06,
      "loss": 1.2528,
      "step": 508
    },
    {
      "epoch": 0.014666051979484816,
      "grad_norm": 3.938767194747925,
      "learning_rate": 4.884836852207294e-06,
      "loss": 1.228,
      "step": 509
    },
    {
      "epoch": 0.014694865441134098,
      "grad_norm": 5.056628704071045,
      "learning_rate": 4.89443378119002e-06,
      "loss": 1.1958,
      "step": 510
    },
    {
      "epoch": 0.01472367890278338,
      "grad_norm": 4.3993353843688965,
      "learning_rate": 4.904030710172745e-06,
      "loss": 1.2396,
      "step": 511
    },
    {
      "epoch": 0.014752492364432664,
      "grad_norm": 4.175745964050293,
      "learning_rate": 4.91362763915547e-06,
      "loss": 1.2292,
      "step": 512
    },
    {
      "epoch": 0.014781305826081946,
      "grad_norm": 4.077296733856201,
      "learning_rate": 4.923224568138196e-06,
      "loss": 1.2438,
      "step": 513
    },
    {
      "epoch": 0.014810119287731228,
      "grad_norm": 3.619192361831665,
      "learning_rate": 4.9328214971209215e-06,
      "loss": 1.249,
      "step": 514
    },
    {
      "epoch": 0.01483893274938051,
      "grad_norm": 4.884719371795654,
      "learning_rate": 4.942418426103647e-06,
      "loss": 1.2767,
      "step": 515
    },
    {
      "epoch": 0.014867746211029794,
      "grad_norm": 4.484084606170654,
      "learning_rate": 4.952015355086373e-06,
      "loss": 1.2395,
      "step": 516
    },
    {
      "epoch": 0.014896559672679076,
      "grad_norm": 4.112922191619873,
      "learning_rate": 4.9616122840690986e-06,
      "loss": 1.2507,
      "step": 517
    },
    {
      "epoch": 0.014925373134328358,
      "grad_norm": 4.004873752593994,
      "learning_rate": 4.971209213051823e-06,
      "loss": 1.2555,
      "step": 518
    },
    {
      "epoch": 0.01495418659597764,
      "grad_norm": 5.660669326782227,
      "learning_rate": 4.980806142034549e-06,
      "loss": 1.2545,
      "step": 519
    },
    {
      "epoch": 0.014983000057626924,
      "grad_norm": 4.241215229034424,
      "learning_rate": 4.990403071017275e-06,
      "loss": 1.295,
      "step": 520
    },
    {
      "epoch": 0.015011813519276206,
      "grad_norm": 4.800769805908203,
      "learning_rate": 5e-06,
      "loss": 1.2431,
      "step": 521
    },
    {
      "epoch": 0.015040626980925488,
      "grad_norm": 4.248162269592285,
      "learning_rate": 5.009596928982726e-06,
      "loss": 1.243,
      "step": 522
    },
    {
      "epoch": 0.01506944044257477,
      "grad_norm": 4.286783218383789,
      "learning_rate": 5.019193857965452e-06,
      "loss": 1.2541,
      "step": 523
    },
    {
      "epoch": 0.015098253904224054,
      "grad_norm": 4.269665718078613,
      "learning_rate": 5.028790786948177e-06,
      "loss": 1.2495,
      "step": 524
    },
    {
      "epoch": 0.015127067365873336,
      "grad_norm": 4.165854454040527,
      "learning_rate": 5.038387715930902e-06,
      "loss": 1.2186,
      "step": 525
    },
    {
      "epoch": 0.015155880827522618,
      "grad_norm": 5.314565658569336,
      "learning_rate": 5.047984644913628e-06,
      "loss": 1.2589,
      "step": 526
    },
    {
      "epoch": 0.0151846942891719,
      "grad_norm": 4.159290313720703,
      "learning_rate": 5.0575815738963544e-06,
      "loss": 1.2777,
      "step": 527
    },
    {
      "epoch": 0.015213507750821184,
      "grad_norm": 4.444005966186523,
      "learning_rate": 5.0671785028790784e-06,
      "loss": 1.2276,
      "step": 528
    },
    {
      "epoch": 0.015242321212470466,
      "grad_norm": 4.203051567077637,
      "learning_rate": 5.076775431861805e-06,
      "loss": 1.2642,
      "step": 529
    },
    {
      "epoch": 0.015271134674119748,
      "grad_norm": 4.694973945617676,
      "learning_rate": 5.086372360844531e-06,
      "loss": 1.2141,
      "step": 530
    },
    {
      "epoch": 0.015299948135769032,
      "grad_norm": 4.391695022583008,
      "learning_rate": 5.095969289827256e-06,
      "loss": 1.2516,
      "step": 531
    },
    {
      "epoch": 0.015328761597418314,
      "grad_norm": 4.080984115600586,
      "learning_rate": 5.105566218809981e-06,
      "loss": 1.2295,
      "step": 532
    },
    {
      "epoch": 0.015357575059067596,
      "grad_norm": 4.484219074249268,
      "learning_rate": 5.115163147792707e-06,
      "loss": 1.2482,
      "step": 533
    },
    {
      "epoch": 0.015386388520716878,
      "grad_norm": 3.862666130065918,
      "learning_rate": 5.1247600767754325e-06,
      "loss": 1.268,
      "step": 534
    },
    {
      "epoch": 0.015415201982366162,
      "grad_norm": 5.041067600250244,
      "learning_rate": 5.134357005758158e-06,
      "loss": 1.2376,
      "step": 535
    },
    {
      "epoch": 0.015444015444015444,
      "grad_norm": 4.788147449493408,
      "learning_rate": 5.143953934740883e-06,
      "loss": 1.2486,
      "step": 536
    },
    {
      "epoch": 0.015472828905664726,
      "grad_norm": 4.547287940979004,
      "learning_rate": 5.153550863723609e-06,
      "loss": 1.2519,
      "step": 537
    },
    {
      "epoch": 0.015501642367314009,
      "grad_norm": 5.376948356628418,
      "learning_rate": 5.163147792706334e-06,
      "loss": 1.2543,
      "step": 538
    },
    {
      "epoch": 0.015530455828963292,
      "grad_norm": 3.930781602859497,
      "learning_rate": 5.17274472168906e-06,
      "loss": 1.234,
      "step": 539
    },
    {
      "epoch": 0.015559269290612575,
      "grad_norm": 5.917802333831787,
      "learning_rate": 5.182341650671785e-06,
      "loss": 1.214,
      "step": 540
    },
    {
      "epoch": 0.015588082752261857,
      "grad_norm": 3.657301425933838,
      "learning_rate": 5.1919385796545105e-06,
      "loss": 1.2449,
      "step": 541
    },
    {
      "epoch": 0.015616896213911139,
      "grad_norm": 4.450209140777588,
      "learning_rate": 5.201535508637236e-06,
      "loss": 1.2258,
      "step": 542
    },
    {
      "epoch": 0.01564570967556042,
      "grad_norm": 3.8092408180236816,
      "learning_rate": 5.211132437619963e-06,
      "loss": 1.2428,
      "step": 543
    },
    {
      "epoch": 0.015674523137209705,
      "grad_norm": 3.8405072689056396,
      "learning_rate": 5.220729366602687e-06,
      "loss": 1.2289,
      "step": 544
    },
    {
      "epoch": 0.01570333659885899,
      "grad_norm": 3.833569049835205,
      "learning_rate": 5.230326295585413e-06,
      "loss": 1.2209,
      "step": 545
    },
    {
      "epoch": 0.01573215006050827,
      "grad_norm": 4.056374549865723,
      "learning_rate": 5.239923224568139e-06,
      "loss": 1.2721,
      "step": 546
    },
    {
      "epoch": 0.015760963522157553,
      "grad_norm": 3.811256170272827,
      "learning_rate": 5.2495201535508645e-06,
      "loss": 1.2085,
      "step": 547
    },
    {
      "epoch": 0.015789776983806833,
      "grad_norm": 4.3115081787109375,
      "learning_rate": 5.25911708253359e-06,
      "loss": 1.2483,
      "step": 548
    },
    {
      "epoch": 0.015818590445456117,
      "grad_norm": 3.5155439376831055,
      "learning_rate": 5.268714011516315e-06,
      "loss": 1.2487,
      "step": 549
    },
    {
      "epoch": 0.0158474039071054,
      "grad_norm": 4.558698654174805,
      "learning_rate": 5.278310940499041e-06,
      "loss": 1.232,
      "step": 550
    },
    {
      "epoch": 0.01587621736875468,
      "grad_norm": 4.556251525878906,
      "learning_rate": 5.287907869481766e-06,
      "loss": 1.2296,
      "step": 551
    },
    {
      "epoch": 0.015905030830403965,
      "grad_norm": 3.804935932159424,
      "learning_rate": 5.297504798464492e-06,
      "loss": 1.2564,
      "step": 552
    },
    {
      "epoch": 0.01593384429205325,
      "grad_norm": 4.1825175285339355,
      "learning_rate": 5.307101727447217e-06,
      "loss": 1.1915,
      "step": 553
    },
    {
      "epoch": 0.01596265775370253,
      "grad_norm": 3.834832191467285,
      "learning_rate": 5.3166986564299425e-06,
      "loss": 1.2082,
      "step": 554
    },
    {
      "epoch": 0.015991471215351813,
      "grad_norm": 4.375059127807617,
      "learning_rate": 5.326295585412668e-06,
      "loss": 1.2541,
      "step": 555
    },
    {
      "epoch": 0.016020284677001093,
      "grad_norm": 4.369380474090576,
      "learning_rate": 5.335892514395395e-06,
      "loss": 1.2167,
      "step": 556
    },
    {
      "epoch": 0.016049098138650377,
      "grad_norm": 4.315122127532959,
      "learning_rate": 5.345489443378119e-06,
      "loss": 1.235,
      "step": 557
    },
    {
      "epoch": 0.01607791160029966,
      "grad_norm": 4.133538722991943,
      "learning_rate": 5.355086372360845e-06,
      "loss": 1.2358,
      "step": 558
    },
    {
      "epoch": 0.01610672506194894,
      "grad_norm": 3.9727818965911865,
      "learning_rate": 5.364683301343571e-06,
      "loss": 1.225,
      "step": 559
    },
    {
      "epoch": 0.016135538523598225,
      "grad_norm": 4.099331855773926,
      "learning_rate": 5.374280230326297e-06,
      "loss": 1.2161,
      "step": 560
    },
    {
      "epoch": 0.01616435198524751,
      "grad_norm": 4.885702610015869,
      "learning_rate": 5.383877159309021e-06,
      "loss": 1.2388,
      "step": 561
    },
    {
      "epoch": 0.01619316544689679,
      "grad_norm": 4.517354965209961,
      "learning_rate": 5.393474088291747e-06,
      "loss": 1.1933,
      "step": 562
    },
    {
      "epoch": 0.016221978908546073,
      "grad_norm": 4.219283103942871,
      "learning_rate": 5.403071017274473e-06,
      "loss": 1.2243,
      "step": 563
    },
    {
      "epoch": 0.016250792370195354,
      "grad_norm": 5.067386627197266,
      "learning_rate": 5.4126679462571984e-06,
      "loss": 1.2308,
      "step": 564
    },
    {
      "epoch": 0.016279605831844637,
      "grad_norm": 4.732729911804199,
      "learning_rate": 5.422264875239923e-06,
      "loss": 1.2218,
      "step": 565
    },
    {
      "epoch": 0.01630841929349392,
      "grad_norm": 5.54449462890625,
      "learning_rate": 5.431861804222649e-06,
      "loss": 1.1993,
      "step": 566
    },
    {
      "epoch": 0.0163372327551432,
      "grad_norm": 4.889398574829102,
      "learning_rate": 5.441458733205375e-06,
      "loss": 1.2036,
      "step": 567
    },
    {
      "epoch": 0.016366046216792485,
      "grad_norm": 4.280142307281494,
      "learning_rate": 5.4510556621881e-06,
      "loss": 1.1925,
      "step": 568
    },
    {
      "epoch": 0.01639485967844177,
      "grad_norm": 6.498767852783203,
      "learning_rate": 5.460652591170825e-06,
      "loss": 1.2408,
      "step": 569
    },
    {
      "epoch": 0.01642367314009105,
      "grad_norm": 3.7652440071105957,
      "learning_rate": 5.470249520153551e-06,
      "loss": 1.2384,
      "step": 570
    },
    {
      "epoch": 0.016452486601740333,
      "grad_norm": 5.3889641761779785,
      "learning_rate": 5.4798464491362765e-06,
      "loss": 1.2474,
      "step": 571
    },
    {
      "epoch": 0.016481300063389617,
      "grad_norm": 3.9189913272857666,
      "learning_rate": 5.489443378119003e-06,
      "loss": 1.2236,
      "step": 572
    },
    {
      "epoch": 0.016510113525038898,
      "grad_norm": 5.483098030090332,
      "learning_rate": 5.499040307101729e-06,
      "loss": 1.2126,
      "step": 573
    },
    {
      "epoch": 0.01653892698668818,
      "grad_norm": 3.655731201171875,
      "learning_rate": 5.5086372360844535e-06,
      "loss": 1.2161,
      "step": 574
    },
    {
      "epoch": 0.016567740448337462,
      "grad_norm": 5.071920394897461,
      "learning_rate": 5.518234165067179e-06,
      "loss": 1.2326,
      "step": 575
    },
    {
      "epoch": 0.016596553909986746,
      "grad_norm": 3.7652218341827393,
      "learning_rate": 5.527831094049905e-06,
      "loss": 1.2264,
      "step": 576
    },
    {
      "epoch": 0.01662536737163603,
      "grad_norm": 5.205135345458984,
      "learning_rate": 5.5374280230326305e-06,
      "loss": 1.2107,
      "step": 577
    },
    {
      "epoch": 0.01665418083328531,
      "grad_norm": 3.6525561809539795,
      "learning_rate": 5.547024952015355e-06,
      "loss": 1.2034,
      "step": 578
    },
    {
      "epoch": 0.016682994294934594,
      "grad_norm": 4.023789882659912,
      "learning_rate": 5.556621880998081e-06,
      "loss": 1.2481,
      "step": 579
    },
    {
      "epoch": 0.016711807756583878,
      "grad_norm": 4.392200946807861,
      "learning_rate": 5.566218809980807e-06,
      "loss": 1.2312,
      "step": 580
    },
    {
      "epoch": 0.016740621218233158,
      "grad_norm": 4.628062725067139,
      "learning_rate": 5.575815738963532e-06,
      "loss": 1.2173,
      "step": 581
    },
    {
      "epoch": 0.016769434679882442,
      "grad_norm": 4.248229503631592,
      "learning_rate": 5.585412667946257e-06,
      "loss": 1.2171,
      "step": 582
    },
    {
      "epoch": 0.016798248141531722,
      "grad_norm": 4.4733991622924805,
      "learning_rate": 5.595009596928983e-06,
      "loss": 1.2696,
      "step": 583
    },
    {
      "epoch": 0.016827061603181006,
      "grad_norm": 4.544429779052734,
      "learning_rate": 5.6046065259117085e-06,
      "loss": 1.2221,
      "step": 584
    },
    {
      "epoch": 0.01685587506483029,
      "grad_norm": 5.192183971405029,
      "learning_rate": 5.614203454894435e-06,
      "loss": 1.1837,
      "step": 585
    },
    {
      "epoch": 0.01688468852647957,
      "grad_norm": 5.11816930770874,
      "learning_rate": 5.623800383877159e-06,
      "loss": 1.2112,
      "step": 586
    },
    {
      "epoch": 0.016913501988128854,
      "grad_norm": 5.42815637588501,
      "learning_rate": 5.6333973128598855e-06,
      "loss": 1.2264,
      "step": 587
    },
    {
      "epoch": 0.016942315449778138,
      "grad_norm": 4.753808498382568,
      "learning_rate": 5.642994241842611e-06,
      "loss": 1.2062,
      "step": 588
    },
    {
      "epoch": 0.016971128911427418,
      "grad_norm": 5.924714088439941,
      "learning_rate": 5.652591170825337e-06,
      "loss": 1.1996,
      "step": 589
    },
    {
      "epoch": 0.016999942373076702,
      "grad_norm": 4.776706218719482,
      "learning_rate": 5.662188099808062e-06,
      "loss": 1.2577,
      "step": 590
    },
    {
      "epoch": 0.017028755834725982,
      "grad_norm": 4.756631851196289,
      "learning_rate": 5.671785028790787e-06,
      "loss": 1.1933,
      "step": 591
    },
    {
      "epoch": 0.017057569296375266,
      "grad_norm": 4.391541957855225,
      "learning_rate": 5.681381957773513e-06,
      "loss": 1.2149,
      "step": 592
    },
    {
      "epoch": 0.01708638275802455,
      "grad_norm": 5.030577659606934,
      "learning_rate": 5.690978886756239e-06,
      "loss": 1.2195,
      "step": 593
    },
    {
      "epoch": 0.01711519621967383,
      "grad_norm": 3.923156976699829,
      "learning_rate": 5.7005758157389635e-06,
      "loss": 1.2085,
      "step": 594
    },
    {
      "epoch": 0.017144009681323114,
      "grad_norm": 4.645239353179932,
      "learning_rate": 5.710172744721689e-06,
      "loss": 1.1989,
      "step": 595
    },
    {
      "epoch": 0.017172823142972398,
      "grad_norm": 3.839379072189331,
      "learning_rate": 5.719769673704415e-06,
      "loss": 1.2219,
      "step": 596
    },
    {
      "epoch": 0.01720163660462168,
      "grad_norm": 5.64005708694458,
      "learning_rate": 5.7293666026871406e-06,
      "loss": 1.2236,
      "step": 597
    },
    {
      "epoch": 0.017230450066270962,
      "grad_norm": 3.685337543487549,
      "learning_rate": 5.738963531669866e-06,
      "loss": 1.2128,
      "step": 598
    },
    {
      "epoch": 0.017259263527920243,
      "grad_norm": 4.925141334533691,
      "learning_rate": 5.748560460652591e-06,
      "loss": 1.2102,
      "step": 599
    },
    {
      "epoch": 0.017288076989569526,
      "grad_norm": 4.429798126220703,
      "learning_rate": 5.758157389635317e-06,
      "loss": 1.2238,
      "step": 600
    },
    {
      "epoch": 0.01731689045121881,
      "grad_norm": 4.239463806152344,
      "learning_rate": 5.767754318618043e-06,
      "loss": 1.2175,
      "step": 601
    },
    {
      "epoch": 0.01734570391286809,
      "grad_norm": 5.566977500915527,
      "learning_rate": 5.777351247600769e-06,
      "loss": 1.2135,
      "step": 602
    },
    {
      "epoch": 0.017374517374517374,
      "grad_norm": 3.905305862426758,
      "learning_rate": 5.786948176583494e-06,
      "loss": 1.2253,
      "step": 603
    },
    {
      "epoch": 0.01740333083616666,
      "grad_norm": 4.395923614501953,
      "learning_rate": 5.7965451055662194e-06,
      "loss": 1.2125,
      "step": 604
    },
    {
      "epoch": 0.01743214429781594,
      "grad_norm": 4.27286958694458,
      "learning_rate": 5.806142034548945e-06,
      "loss": 1.2128,
      "step": 605
    },
    {
      "epoch": 0.017460957759465223,
      "grad_norm": 4.463138580322266,
      "learning_rate": 5.815738963531671e-06,
      "loss": 1.2106,
      "step": 606
    },
    {
      "epoch": 0.017489771221114506,
      "grad_norm": 4.05118989944458,
      "learning_rate": 5.825335892514396e-06,
      "loss": 1.203,
      "step": 607
    },
    {
      "epoch": 0.017518584682763787,
      "grad_norm": 3.4085190296173096,
      "learning_rate": 5.834932821497121e-06,
      "loss": 1.1688,
      "step": 608
    },
    {
      "epoch": 0.01754739814441307,
      "grad_norm": 4.725858211517334,
      "learning_rate": 5.844529750479847e-06,
      "loss": 1.2062,
      "step": 609
    },
    {
      "epoch": 0.01757621160606235,
      "grad_norm": 3.2487432956695557,
      "learning_rate": 5.854126679462573e-06,
      "loss": 1.1845,
      "step": 610
    },
    {
      "epoch": 0.017605025067711635,
      "grad_norm": 4.695226192474365,
      "learning_rate": 5.8637236084452975e-06,
      "loss": 1.2073,
      "step": 611
    },
    {
      "epoch": 0.01763383852936092,
      "grad_norm": 3.9534595012664795,
      "learning_rate": 5.873320537428023e-06,
      "loss": 1.2043,
      "step": 612
    },
    {
      "epoch": 0.0176626519910102,
      "grad_norm": 4.659251689910889,
      "learning_rate": 5.882917466410749e-06,
      "loss": 1.1823,
      "step": 613
    },
    {
      "epoch": 0.017691465452659483,
      "grad_norm": 4.5810346603393555,
      "learning_rate": 5.892514395393475e-06,
      "loss": 1.1702,
      "step": 614
    },
    {
      "epoch": 0.017720278914308767,
      "grad_norm": 3.7899608612060547,
      "learning_rate": 5.902111324376199e-06,
      "loss": 1.2258,
      "step": 615
    },
    {
      "epoch": 0.017749092375958047,
      "grad_norm": 3.768744707107544,
      "learning_rate": 5.911708253358925e-06,
      "loss": 1.2108,
      "step": 616
    },
    {
      "epoch": 0.01777790583760733,
      "grad_norm": 4.029522895812988,
      "learning_rate": 5.9213051823416515e-06,
      "loss": 1.2039,
      "step": 617
    },
    {
      "epoch": 0.01780671929925661,
      "grad_norm": 3.7536849975585938,
      "learning_rate": 5.930902111324377e-06,
      "loss": 1.2281,
      "step": 618
    },
    {
      "epoch": 0.017835532760905895,
      "grad_norm": 3.7135939598083496,
      "learning_rate": 5.940499040307102e-06,
      "loss": 1.205,
      "step": 619
    },
    {
      "epoch": 0.01786434622255518,
      "grad_norm": 3.889925718307495,
      "learning_rate": 5.950095969289828e-06,
      "loss": 1.1762,
      "step": 620
    },
    {
      "epoch": 0.01789315968420446,
      "grad_norm": 4.414258003234863,
      "learning_rate": 5.959692898272553e-06,
      "loss": 1.2036,
      "step": 621
    },
    {
      "epoch": 0.017921973145853743,
      "grad_norm": 3.8009414672851562,
      "learning_rate": 5.969289827255279e-06,
      "loss": 1.1813,
      "step": 622
    },
    {
      "epoch": 0.017950786607503027,
      "grad_norm": 4.526575088500977,
      "learning_rate": 5.978886756238005e-06,
      "loss": 1.1976,
      "step": 623
    },
    {
      "epoch": 0.017979600069152307,
      "grad_norm": 3.604912281036377,
      "learning_rate": 5.9884836852207295e-06,
      "loss": 1.1948,
      "step": 624
    },
    {
      "epoch": 0.01800841353080159,
      "grad_norm": 4.292920112609863,
      "learning_rate": 5.998080614203455e-06,
      "loss": 1.22,
      "step": 625
    },
    {
      "epoch": 0.01803722699245087,
      "grad_norm": 3.805804491043091,
      "learning_rate": 6.007677543186181e-06,
      "loss": 1.237,
      "step": 626
    },
    {
      "epoch": 0.018066040454100155,
      "grad_norm": 4.074632167816162,
      "learning_rate": 6.0172744721689065e-06,
      "loss": 1.1849,
      "step": 627
    },
    {
      "epoch": 0.01809485391574944,
      "grad_norm": 3.6576027870178223,
      "learning_rate": 6.026871401151631e-06,
      "loss": 1.1922,
      "step": 628
    },
    {
      "epoch": 0.01812366737739872,
      "grad_norm": 4.059773921966553,
      "learning_rate": 6.036468330134357e-06,
      "loss": 1.1619,
      "step": 629
    },
    {
      "epoch": 0.018152480839048003,
      "grad_norm": 3.6923370361328125,
      "learning_rate": 6.0460652591170836e-06,
      "loss": 1.1864,
      "step": 630
    },
    {
      "epoch": 0.018181294300697287,
      "grad_norm": 5.056499004364014,
      "learning_rate": 6.055662188099809e-06,
      "loss": 1.1878,
      "step": 631
    },
    {
      "epoch": 0.018210107762346568,
      "grad_norm": 4.077653884887695,
      "learning_rate": 6.065259117082534e-06,
      "loss": 1.2106,
      "step": 632
    },
    {
      "epoch": 0.01823892122399585,
      "grad_norm": 4.136611461639404,
      "learning_rate": 6.07485604606526e-06,
      "loss": 1.1754,
      "step": 633
    },
    {
      "epoch": 0.01826773468564513,
      "grad_norm": 4.23780632019043,
      "learning_rate": 6.084452975047985e-06,
      "loss": 1.2021,
      "step": 634
    },
    {
      "epoch": 0.018296548147294416,
      "grad_norm": 4.324599266052246,
      "learning_rate": 6.094049904030711e-06,
      "loss": 1.2047,
      "step": 635
    },
    {
      "epoch": 0.0183253616089437,
      "grad_norm": 3.7927911281585693,
      "learning_rate": 6.103646833013436e-06,
      "loss": 1.1832,
      "step": 636
    },
    {
      "epoch": 0.01835417507059298,
      "grad_norm": 4.29425573348999,
      "learning_rate": 6.1132437619961616e-06,
      "loss": 1.2204,
      "step": 637
    },
    {
      "epoch": 0.018382988532242264,
      "grad_norm": 3.6515369415283203,
      "learning_rate": 6.122840690978887e-06,
      "loss": 1.1935,
      "step": 638
    },
    {
      "epoch": 0.018411801993891547,
      "grad_norm": 3.9063048362731934,
      "learning_rate": 6.132437619961613e-06,
      "loss": 1.1776,
      "step": 639
    },
    {
      "epoch": 0.018440615455540828,
      "grad_norm": 3.8100507259368896,
      "learning_rate": 6.142034548944338e-06,
      "loss": 1.1608,
      "step": 640
    },
    {
      "epoch": 0.01846942891719011,
      "grad_norm": 4.126123428344727,
      "learning_rate": 6.151631477927063e-06,
      "loss": 1.1858,
      "step": 641
    },
    {
      "epoch": 0.018498242378839395,
      "grad_norm": 4.261011600494385,
      "learning_rate": 6.161228406909789e-06,
      "loss": 1.2016,
      "step": 642
    },
    {
      "epoch": 0.018527055840488676,
      "grad_norm": 4.187440395355225,
      "learning_rate": 6.170825335892516e-06,
      "loss": 1.1775,
      "step": 643
    },
    {
      "epoch": 0.01855586930213796,
      "grad_norm": 3.8185455799102783,
      "learning_rate": 6.18042226487524e-06,
      "loss": 1.1953,
      "step": 644
    },
    {
      "epoch": 0.01858468276378724,
      "grad_norm": 4.49042272567749,
      "learning_rate": 6.190019193857965e-06,
      "loss": 1.2179,
      "step": 645
    },
    {
      "epoch": 0.018613496225436524,
      "grad_norm": 3.638590097427368,
      "learning_rate": 6.199616122840692e-06,
      "loss": 1.2355,
      "step": 646
    },
    {
      "epoch": 0.018642309687085808,
      "grad_norm": 3.730144500732422,
      "learning_rate": 6.2092130518234175e-06,
      "loss": 1.2009,
      "step": 647
    },
    {
      "epoch": 0.018671123148735088,
      "grad_norm": 3.882189989089966,
      "learning_rate": 6.218809980806143e-06,
      "loss": 1.1843,
      "step": 648
    },
    {
      "epoch": 0.018699936610384372,
      "grad_norm": 3.9581143856048584,
      "learning_rate": 6.228406909788868e-06,
      "loss": 1.1873,
      "step": 649
    },
    {
      "epoch": 0.018728750072033656,
      "grad_norm": 4.832870006561279,
      "learning_rate": 6.238003838771594e-06,
      "loss": 1.1791,
      "step": 650
    },
    {
      "epoch": 0.018757563533682936,
      "grad_norm": 3.3896291255950928,
      "learning_rate": 6.247600767754319e-06,
      "loss": 1.1769,
      "step": 651
    },
    {
      "epoch": 0.01878637699533222,
      "grad_norm": 4.321454048156738,
      "learning_rate": 6.257197696737045e-06,
      "loss": 1.1924,
      "step": 652
    },
    {
      "epoch": 0.0188151904569815,
      "grad_norm": 3.715669631958008,
      "learning_rate": 6.26679462571977e-06,
      "loss": 1.2164,
      "step": 653
    },
    {
      "epoch": 0.018844003918630784,
      "grad_norm": 4.35163688659668,
      "learning_rate": 6.2763915547024955e-06,
      "loss": 1.1788,
      "step": 654
    },
    {
      "epoch": 0.018872817380280068,
      "grad_norm": 4.400252342224121,
      "learning_rate": 6.285988483685221e-06,
      "loss": 1.2232,
      "step": 655
    },
    {
      "epoch": 0.01890163084192935,
      "grad_norm": 4.06829309463501,
      "learning_rate": 6.295585412667947e-06,
      "loss": 1.1882,
      "step": 656
    },
    {
      "epoch": 0.018930444303578632,
      "grad_norm": 4.282623291015625,
      "learning_rate": 6.305182341650672e-06,
      "loss": 1.2042,
      "step": 657
    },
    {
      "epoch": 0.018959257765227916,
      "grad_norm": 4.172646999359131,
      "learning_rate": 6.314779270633397e-06,
      "loss": 1.1666,
      "step": 658
    },
    {
      "epoch": 0.018988071226877196,
      "grad_norm": 4.907839775085449,
      "learning_rate": 6.324376199616124e-06,
      "loss": 1.1734,
      "step": 659
    },
    {
      "epoch": 0.01901688468852648,
      "grad_norm": 3.864046096801758,
      "learning_rate": 6.3339731285988495e-06,
      "loss": 1.1911,
      "step": 660
    },
    {
      "epoch": 0.01904569815017576,
      "grad_norm": 4.635900974273682,
      "learning_rate": 6.343570057581574e-06,
      "loss": 1.1995,
      "step": 661
    },
    {
      "epoch": 0.019074511611825044,
      "grad_norm": 4.286877155303955,
      "learning_rate": 6.3531669865643e-06,
      "loss": 1.1972,
      "step": 662
    },
    {
      "epoch": 0.019103325073474328,
      "grad_norm": 4.252872467041016,
      "learning_rate": 6.362763915547026e-06,
      "loss": 1.1862,
      "step": 663
    },
    {
      "epoch": 0.01913213853512361,
      "grad_norm": 4.229471683502197,
      "learning_rate": 6.372360844529751e-06,
      "loss": 1.1949,
      "step": 664
    },
    {
      "epoch": 0.019160951996772892,
      "grad_norm": 3.8010828495025635,
      "learning_rate": 6.381957773512476e-06,
      "loss": 1.1805,
      "step": 665
    },
    {
      "epoch": 0.019189765458422176,
      "grad_norm": 3.6911299228668213,
      "learning_rate": 6.391554702495202e-06,
      "loss": 1.2152,
      "step": 666
    },
    {
      "epoch": 0.019218578920071457,
      "grad_norm": 3.9711766242980957,
      "learning_rate": 6.4011516314779275e-06,
      "loss": 1.2176,
      "step": 667
    },
    {
      "epoch": 0.01924739238172074,
      "grad_norm": 4.688870906829834,
      "learning_rate": 6.410748560460653e-06,
      "loss": 1.1547,
      "step": 668
    },
    {
      "epoch": 0.01927620584337002,
      "grad_norm": 4.6160478591918945,
      "learning_rate": 6.420345489443378e-06,
      "loss": 1.1705,
      "step": 669
    },
    {
      "epoch": 0.019305019305019305,
      "grad_norm": 5.1994948387146,
      "learning_rate": 6.429942418426104e-06,
      "loss": 1.2042,
      "step": 670
    },
    {
      "epoch": 0.01933383276666859,
      "grad_norm": 4.221469879150391,
      "learning_rate": 6.439539347408829e-06,
      "loss": 1.1841,
      "step": 671
    },
    {
      "epoch": 0.01936264622831787,
      "grad_norm": 4.125655651092529,
      "learning_rate": 6.449136276391556e-06,
      "loss": 1.1815,
      "step": 672
    },
    {
      "epoch": 0.019391459689967153,
      "grad_norm": 4.2449493408203125,
      "learning_rate": 6.4587332053742816e-06,
      "loss": 1.1862,
      "step": 673
    },
    {
      "epoch": 0.019420273151616436,
      "grad_norm": 4.674434185028076,
      "learning_rate": 6.4683301343570056e-06,
      "loss": 1.1828,
      "step": 674
    },
    {
      "epoch": 0.019449086613265717,
      "grad_norm": 3.8638486862182617,
      "learning_rate": 6.477927063339732e-06,
      "loss": 1.1625,
      "step": 675
    },
    {
      "epoch": 0.019477900074915,
      "grad_norm": 4.895153999328613,
      "learning_rate": 6.487523992322458e-06,
      "loss": 1.1757,
      "step": 676
    },
    {
      "epoch": 0.019506713536564285,
      "grad_norm": 4.0760817527771,
      "learning_rate": 6.497120921305183e-06,
      "loss": 1.1929,
      "step": 677
    },
    {
      "epoch": 0.019535526998213565,
      "grad_norm": 5.418991565704346,
      "learning_rate": 6.506717850287908e-06,
      "loss": 1.1923,
      "step": 678
    },
    {
      "epoch": 0.01956434045986285,
      "grad_norm": 4.522409439086914,
      "learning_rate": 6.516314779270634e-06,
      "loss": 1.1901,
      "step": 679
    },
    {
      "epoch": 0.01959315392151213,
      "grad_norm": 4.322537899017334,
      "learning_rate": 6.52591170825336e-06,
      "loss": 1.1586,
      "step": 680
    },
    {
      "epoch": 0.019621967383161413,
      "grad_norm": 4.007564544677734,
      "learning_rate": 6.535508637236085e-06,
      "loss": 1.1479,
      "step": 681
    },
    {
      "epoch": 0.019650780844810697,
      "grad_norm": 4.198097229003906,
      "learning_rate": 6.54510556621881e-06,
      "loss": 1.1668,
      "step": 682
    },
    {
      "epoch": 0.019679594306459977,
      "grad_norm": 4.492678642272949,
      "learning_rate": 6.554702495201536e-06,
      "loss": 1.185,
      "step": 683
    },
    {
      "epoch": 0.01970840776810926,
      "grad_norm": 3.6013541221618652,
      "learning_rate": 6.5642994241842614e-06,
      "loss": 1.1836,
      "step": 684
    },
    {
      "epoch": 0.019737221229758545,
      "grad_norm": 3.8249294757843018,
      "learning_rate": 6.573896353166987e-06,
      "loss": 1.1679,
      "step": 685
    },
    {
      "epoch": 0.019766034691407825,
      "grad_norm": 3.7012345790863037,
      "learning_rate": 6.583493282149712e-06,
      "loss": 1.1579,
      "step": 686
    },
    {
      "epoch": 0.01979484815305711,
      "grad_norm": 3.835911512374878,
      "learning_rate": 6.593090211132438e-06,
      "loss": 1.1831,
      "step": 687
    },
    {
      "epoch": 0.01982366161470639,
      "grad_norm": 3.932298421859741,
      "learning_rate": 6.602687140115164e-06,
      "loss": 1.1729,
      "step": 688
    },
    {
      "epoch": 0.019852475076355673,
      "grad_norm": 3.844480514526367,
      "learning_rate": 6.61228406909789e-06,
      "loss": 1.1816,
      "step": 689
    },
    {
      "epoch": 0.019881288538004957,
      "grad_norm": 4.3791279792785645,
      "learning_rate": 6.621880998080615e-06,
      "loss": 1.2005,
      "step": 690
    },
    {
      "epoch": 0.019910101999654237,
      "grad_norm": 3.498506546020508,
      "learning_rate": 6.63147792706334e-06,
      "loss": 1.2146,
      "step": 691
    },
    {
      "epoch": 0.01993891546130352,
      "grad_norm": 4.037652015686035,
      "learning_rate": 6.641074856046066e-06,
      "loss": 1.1472,
      "step": 692
    },
    {
      "epoch": 0.019967728922952805,
      "grad_norm": 4.521526336669922,
      "learning_rate": 6.650671785028792e-06,
      "loss": 1.1985,
      "step": 693
    },
    {
      "epoch": 0.019996542384602085,
      "grad_norm": 3.9556384086608887,
      "learning_rate": 6.6602687140115165e-06,
      "loss": 1.1619,
      "step": 694
    },
    {
      "epoch": 0.02002535584625137,
      "grad_norm": 4.279000282287598,
      "learning_rate": 6.669865642994242e-06,
      "loss": 1.1906,
      "step": 695
    },
    {
      "epoch": 0.02005416930790065,
      "grad_norm": 3.856889247894287,
      "learning_rate": 6.679462571976968e-06,
      "loss": 1.1725,
      "step": 696
    },
    {
      "epoch": 0.020082982769549933,
      "grad_norm": 5.050195217132568,
      "learning_rate": 6.6890595009596935e-06,
      "loss": 1.1857,
      "step": 697
    },
    {
      "epoch": 0.020111796231199217,
      "grad_norm": 3.500669240951538,
      "learning_rate": 6.698656429942419e-06,
      "loss": 1.1672,
      "step": 698
    },
    {
      "epoch": 0.020140609692848498,
      "grad_norm": 4.532929420471191,
      "learning_rate": 6.708253358925144e-06,
      "loss": 1.1902,
      "step": 699
    },
    {
      "epoch": 0.02016942315449778,
      "grad_norm": 3.6432316303253174,
      "learning_rate": 6.71785028790787e-06,
      "loss": 1.1551,
      "step": 700
    },
    {
      "epoch": 0.020198236616147065,
      "grad_norm": 4.295480728149414,
      "learning_rate": 6.727447216890595e-06,
      "loss": 1.1846,
      "step": 701
    },
    {
      "epoch": 0.020227050077796346,
      "grad_norm": 3.8760733604431152,
      "learning_rate": 6.737044145873322e-06,
      "loss": 1.18,
      "step": 702
    },
    {
      "epoch": 0.02025586353944563,
      "grad_norm": 3.886930227279663,
      "learning_rate": 6.746641074856046e-06,
      "loss": 1.1933,
      "step": 703
    },
    {
      "epoch": 0.02028467700109491,
      "grad_norm": 4.171339988708496,
      "learning_rate": 6.756238003838772e-06,
      "loss": 1.2114,
      "step": 704
    },
    {
      "epoch": 0.020313490462744194,
      "grad_norm": 3.5505270957946777,
      "learning_rate": 6.765834932821498e-06,
      "loss": 1.1829,
      "step": 705
    },
    {
      "epoch": 0.020342303924393478,
      "grad_norm": 3.478480339050293,
      "learning_rate": 6.775431861804224e-06,
      "loss": 1.1521,
      "step": 706
    },
    {
      "epoch": 0.020371117386042758,
      "grad_norm": 3.4826717376708984,
      "learning_rate": 6.7850287907869485e-06,
      "loss": 1.1697,
      "step": 707
    },
    {
      "epoch": 0.02039993084769204,
      "grad_norm": 3.5106420516967773,
      "learning_rate": 6.794625719769674e-06,
      "loss": 1.1443,
      "step": 708
    },
    {
      "epoch": 0.020428744309341326,
      "grad_norm": 3.759477138519287,
      "learning_rate": 6.8042226487524e-06,
      "loss": 1.1575,
      "step": 709
    },
    {
      "epoch": 0.020457557770990606,
      "grad_norm": 3.3493847846984863,
      "learning_rate": 6.8138195777351256e-06,
      "loss": 1.1506,
      "step": 710
    },
    {
      "epoch": 0.02048637123263989,
      "grad_norm": 3.948737859725952,
      "learning_rate": 6.82341650671785e-06,
      "loss": 1.1724,
      "step": 711
    },
    {
      "epoch": 0.020515184694289174,
      "grad_norm": 3.7825000286102295,
      "learning_rate": 6.833013435700576e-06,
      "loss": 1.1497,
      "step": 712
    },
    {
      "epoch": 0.020543998155938454,
      "grad_norm": 4.282146453857422,
      "learning_rate": 6.842610364683302e-06,
      "loss": 1.1679,
      "step": 713
    },
    {
      "epoch": 0.020572811617587738,
      "grad_norm": 3.8032331466674805,
      "learning_rate": 6.852207293666027e-06,
      "loss": 1.187,
      "step": 714
    },
    {
      "epoch": 0.020601625079237018,
      "grad_norm": 4.147481441497803,
      "learning_rate": 6.861804222648752e-06,
      "loss": 1.1683,
      "step": 715
    },
    {
      "epoch": 0.020630438540886302,
      "grad_norm": 4.218388080596924,
      "learning_rate": 6.871401151631478e-06,
      "loss": 1.176,
      "step": 716
    },
    {
      "epoch": 0.020659252002535586,
      "grad_norm": 3.75809907913208,
      "learning_rate": 6.8809980806142044e-06,
      "loss": 1.1565,
      "step": 717
    },
    {
      "epoch": 0.020688065464184866,
      "grad_norm": 3.5651087760925293,
      "learning_rate": 6.89059500959693e-06,
      "loss": 1.167,
      "step": 718
    },
    {
      "epoch": 0.02071687892583415,
      "grad_norm": 4.161599159240723,
      "learning_rate": 6.900191938579655e-06,
      "loss": 1.1571,
      "step": 719
    },
    {
      "epoch": 0.020745692387483434,
      "grad_norm": 3.8092212677001953,
      "learning_rate": 6.909788867562381e-06,
      "loss": 1.1832,
      "step": 720
    },
    {
      "epoch": 0.020774505849132714,
      "grad_norm": 3.933877468109131,
      "learning_rate": 6.919385796545106e-06,
      "loss": 1.1498,
      "step": 721
    },
    {
      "epoch": 0.020803319310781998,
      "grad_norm": 3.762122392654419,
      "learning_rate": 6.928982725527832e-06,
      "loss": 1.197,
      "step": 722
    },
    {
      "epoch": 0.02083213277243128,
      "grad_norm": 4.584457874298096,
      "learning_rate": 6.938579654510558e-06,
      "loss": 1.1747,
      "step": 723
    },
    {
      "epoch": 0.020860946234080562,
      "grad_norm": 3.846095323562622,
      "learning_rate": 6.9481765834932824e-06,
      "loss": 1.1677,
      "step": 724
    },
    {
      "epoch": 0.020889759695729846,
      "grad_norm": 3.9857492446899414,
      "learning_rate": 6.957773512476008e-06,
      "loss": 1.1337,
      "step": 725
    },
    {
      "epoch": 0.020918573157379126,
      "grad_norm": 4.2688069343566895,
      "learning_rate": 6.967370441458734e-06,
      "loss": 1.1741,
      "step": 726
    },
    {
      "epoch": 0.02094738661902841,
      "grad_norm": 3.4339890480041504,
      "learning_rate": 6.9769673704414595e-06,
      "loss": 1.1466,
      "step": 727
    },
    {
      "epoch": 0.020976200080677694,
      "grad_norm": 4.415665626525879,
      "learning_rate": 6.986564299424184e-06,
      "loss": 1.1547,
      "step": 728
    },
    {
      "epoch": 0.021005013542326974,
      "grad_norm": 3.8109378814697266,
      "learning_rate": 6.99616122840691e-06,
      "loss": 1.1887,
      "step": 729
    },
    {
      "epoch": 0.02103382700397626,
      "grad_norm": 4.392058849334717,
      "learning_rate": 7.005758157389636e-06,
      "loss": 1.1789,
      "step": 730
    },
    {
      "epoch": 0.02106264046562554,
      "grad_norm": 3.5967841148376465,
      "learning_rate": 7.015355086372362e-06,
      "loss": 1.1815,
      "step": 731
    },
    {
      "epoch": 0.021091453927274822,
      "grad_norm": 4.503957271575928,
      "learning_rate": 7.024952015355086e-06,
      "loss": 1.2002,
      "step": 732
    },
    {
      "epoch": 0.021120267388924106,
      "grad_norm": 3.6622233390808105,
      "learning_rate": 7.034548944337813e-06,
      "loss": 1.1788,
      "step": 733
    },
    {
      "epoch": 0.021149080850573387,
      "grad_norm": 4.528642654418945,
      "learning_rate": 7.044145873320538e-06,
      "loss": 1.1741,
      "step": 734
    },
    {
      "epoch": 0.02117789431222267,
      "grad_norm": 3.9447076320648193,
      "learning_rate": 7.053742802303264e-06,
      "loss": 1.1804,
      "step": 735
    },
    {
      "epoch": 0.021206707773871954,
      "grad_norm": 4.368509769439697,
      "learning_rate": 7.063339731285989e-06,
      "loss": 1.1535,
      "step": 736
    },
    {
      "epoch": 0.021235521235521235,
      "grad_norm": 3.749584913253784,
      "learning_rate": 7.0729366602687145e-06,
      "loss": 1.1442,
      "step": 737
    },
    {
      "epoch": 0.02126433469717052,
      "grad_norm": 4.344030380249023,
      "learning_rate": 7.08253358925144e-06,
      "loss": 1.1502,
      "step": 738
    },
    {
      "epoch": 0.0212931481588198,
      "grad_norm": 3.6262528896331787,
      "learning_rate": 7.092130518234166e-06,
      "loss": 1.1461,
      "step": 739
    },
    {
      "epoch": 0.021321961620469083,
      "grad_norm": 4.230298042297363,
      "learning_rate": 7.101727447216891e-06,
      "loss": 1.1616,
      "step": 740
    },
    {
      "epoch": 0.021350775082118367,
      "grad_norm": 3.7347044944763184,
      "learning_rate": 7.111324376199616e-06,
      "loss": 1.195,
      "step": 741
    },
    {
      "epoch": 0.021379588543767647,
      "grad_norm": 4.064319133758545,
      "learning_rate": 7.120921305182342e-06,
      "loss": 1.1524,
      "step": 742
    },
    {
      "epoch": 0.02140840200541693,
      "grad_norm": 3.7721335887908936,
      "learning_rate": 7.130518234165068e-06,
      "loss": 1.1815,
      "step": 743
    },
    {
      "epoch": 0.021437215467066215,
      "grad_norm": 3.9380650520324707,
      "learning_rate": 7.1401151631477925e-06,
      "loss": 1.1886,
      "step": 744
    },
    {
      "epoch": 0.021466028928715495,
      "grad_norm": 3.6871273517608643,
      "learning_rate": 7.149712092130518e-06,
      "loss": 1.1359,
      "step": 745
    },
    {
      "epoch": 0.02149484239036478,
      "grad_norm": 3.6457481384277344,
      "learning_rate": 7.159309021113245e-06,
      "loss": 1.1237,
      "step": 746
    },
    {
      "epoch": 0.021523655852014063,
      "grad_norm": 3.716752767562866,
      "learning_rate": 7.16890595009597e-06,
      "loss": 1.1498,
      "step": 747
    },
    {
      "epoch": 0.021552469313663343,
      "grad_norm": 3.942239999771118,
      "learning_rate": 7.178502879078696e-06,
      "loss": 1.1814,
      "step": 748
    },
    {
      "epoch": 0.021581282775312627,
      "grad_norm": 4.095983505249023,
      "learning_rate": 7.188099808061421e-06,
      "loss": 1.1643,
      "step": 749
    },
    {
      "epoch": 0.021610096236961907,
      "grad_norm": 3.9858973026275635,
      "learning_rate": 7.1976967370441466e-06,
      "loss": 1.1565,
      "step": 750
    },
    {
      "epoch": 0.02163890969861119,
      "grad_norm": 3.707629680633545,
      "learning_rate": 7.207293666026872e-06,
      "loss": 1.1625,
      "step": 751
    },
    {
      "epoch": 0.021667723160260475,
      "grad_norm": 4.3112616539001465,
      "learning_rate": 7.216890595009598e-06,
      "loss": 1.1427,
      "step": 752
    },
    {
      "epoch": 0.021696536621909755,
      "grad_norm": 3.4693758487701416,
      "learning_rate": 7.226487523992323e-06,
      "loss": 1.1503,
      "step": 753
    },
    {
      "epoch": 0.02172535008355904,
      "grad_norm": 4.121772766113281,
      "learning_rate": 7.236084452975048e-06,
      "loss": 1.1509,
      "step": 754
    },
    {
      "epoch": 0.021754163545208323,
      "grad_norm": 3.512044906616211,
      "learning_rate": 7.245681381957774e-06,
      "loss": 1.1774,
      "step": 755
    },
    {
      "epoch": 0.021782977006857603,
      "grad_norm": 4.181389331817627,
      "learning_rate": 7.2552783109405e-06,
      "loss": 1.1395,
      "step": 756
    },
    {
      "epoch": 0.021811790468506887,
      "grad_norm": 4.25566291809082,
      "learning_rate": 7.264875239923225e-06,
      "loss": 1.1616,
      "step": 757
    },
    {
      "epoch": 0.021840603930156167,
      "grad_norm": 4.110269546508789,
      "learning_rate": 7.27447216890595e-06,
      "loss": 1.1469,
      "step": 758
    },
    {
      "epoch": 0.02186941739180545,
      "grad_norm": 3.785538911819458,
      "learning_rate": 7.284069097888676e-06,
      "loss": 1.1764,
      "step": 759
    },
    {
      "epoch": 0.021898230853454735,
      "grad_norm": 4.089937686920166,
      "learning_rate": 7.2936660268714024e-06,
      "loss": 1.171,
      "step": 760
    },
    {
      "epoch": 0.021927044315104015,
      "grad_norm": 3.9534730911254883,
      "learning_rate": 7.3032629558541264e-06,
      "loss": 1.1595,
      "step": 761
    },
    {
      "epoch": 0.0219558577767533,
      "grad_norm": 4.0580620765686035,
      "learning_rate": 7.312859884836853e-06,
      "loss": 1.1608,
      "step": 762
    },
    {
      "epoch": 0.021984671238402583,
      "grad_norm": 3.6690328121185303,
      "learning_rate": 7.322456813819579e-06,
      "loss": 1.1476,
      "step": 763
    },
    {
      "epoch": 0.022013484700051864,
      "grad_norm": 3.8926241397857666,
      "learning_rate": 7.332053742802304e-06,
      "loss": 1.1511,
      "step": 764
    },
    {
      "epoch": 0.022042298161701147,
      "grad_norm": 4.072379112243652,
      "learning_rate": 7.341650671785029e-06,
      "loss": 1.1443,
      "step": 765
    },
    {
      "epoch": 0.022071111623350428,
      "grad_norm": 3.6080007553100586,
      "learning_rate": 7.351247600767755e-06,
      "loss": 1.1529,
      "step": 766
    },
    {
      "epoch": 0.02209992508499971,
      "grad_norm": 3.8723642826080322,
      "learning_rate": 7.3608445297504805e-06,
      "loss": 1.1711,
      "step": 767
    },
    {
      "epoch": 0.022128738546648995,
      "grad_norm": 4.160393714904785,
      "learning_rate": 7.370441458733206e-06,
      "loss": 1.1395,
      "step": 768
    },
    {
      "epoch": 0.022157552008298276,
      "grad_norm": 3.8190195560455322,
      "learning_rate": 7.380038387715931e-06,
      "loss": 1.1479,
      "step": 769
    },
    {
      "epoch": 0.02218636546994756,
      "grad_norm": 3.5815212726593018,
      "learning_rate": 7.389635316698657e-06,
      "loss": 1.1461,
      "step": 770
    },
    {
      "epoch": 0.022215178931596843,
      "grad_norm": 4.135644435882568,
      "learning_rate": 7.399232245681382e-06,
      "loss": 1.1536,
      "step": 771
    },
    {
      "epoch": 0.022243992393246124,
      "grad_norm": 3.648899555206299,
      "learning_rate": 7.408829174664108e-06,
      "loss": 1.1453,
      "step": 772
    },
    {
      "epoch": 0.022272805854895408,
      "grad_norm": 4.687570571899414,
      "learning_rate": 7.4184261036468345e-06,
      "loss": 1.169,
      "step": 773
    },
    {
      "epoch": 0.022301619316544688,
      "grad_norm": 3.9212141036987305,
      "learning_rate": 7.4280230326295585e-06,
      "loss": 1.1743,
      "step": 774
    },
    {
      "epoch": 0.022330432778193972,
      "grad_norm": 4.160579204559326,
      "learning_rate": 7.437619961612285e-06,
      "loss": 1.1526,
      "step": 775
    },
    {
      "epoch": 0.022359246239843256,
      "grad_norm": 4.779922008514404,
      "learning_rate": 7.447216890595011e-06,
      "loss": 1.1554,
      "step": 776
    },
    {
      "epoch": 0.022388059701492536,
      "grad_norm": 3.5535025596618652,
      "learning_rate": 7.456813819577736e-06,
      "loss": 1.1641,
      "step": 777
    },
    {
      "epoch": 0.02241687316314182,
      "grad_norm": 4.101378917694092,
      "learning_rate": 7.466410748560461e-06,
      "loss": 1.1872,
      "step": 778
    },
    {
      "epoch": 0.022445686624791104,
      "grad_norm": 3.97537899017334,
      "learning_rate": 7.476007677543187e-06,
      "loss": 1.1422,
      "step": 779
    },
    {
      "epoch": 0.022474500086440384,
      "grad_norm": 3.876757860183716,
      "learning_rate": 7.4856046065259125e-06,
      "loss": 1.1765,
      "step": 780
    },
    {
      "epoch": 0.022503313548089668,
      "grad_norm": 3.5395495891571045,
      "learning_rate": 7.495201535508638e-06,
      "loss": 1.1502,
      "step": 781
    },
    {
      "epoch": 0.02253212700973895,
      "grad_norm": 3.5626094341278076,
      "learning_rate": 7.504798464491363e-06,
      "loss": 1.1521,
      "step": 782
    },
    {
      "epoch": 0.022560940471388232,
      "grad_norm": 4.016055107116699,
      "learning_rate": 7.514395393474089e-06,
      "loss": 1.153,
      "step": 783
    },
    {
      "epoch": 0.022589753933037516,
      "grad_norm": 3.656510353088379,
      "learning_rate": 7.523992322456814e-06,
      "loss": 1.1542,
      "step": 784
    },
    {
      "epoch": 0.022618567394686796,
      "grad_norm": 3.581759214401245,
      "learning_rate": 7.53358925143954e-06,
      "loss": 1.1589,
      "step": 785
    },
    {
      "epoch": 0.02264738085633608,
      "grad_norm": 3.4663209915161133,
      "learning_rate": 7.543186180422265e-06,
      "loss": 1.1478,
      "step": 786
    },
    {
      "epoch": 0.022676194317985364,
      "grad_norm": 4.073664665222168,
      "learning_rate": 7.5527831094049905e-06,
      "loss": 1.1647,
      "step": 787
    },
    {
      "epoch": 0.022705007779634644,
      "grad_norm": 3.5184385776519775,
      "learning_rate": 7.562380038387716e-06,
      "loss": 1.1555,
      "step": 788
    },
    {
      "epoch": 0.022733821241283928,
      "grad_norm": 3.9422996044158936,
      "learning_rate": 7.571976967370443e-06,
      "loss": 1.1392,
      "step": 789
    },
    {
      "epoch": 0.022762634702933212,
      "grad_norm": 3.547504425048828,
      "learning_rate": 7.581573896353167e-06,
      "loss": 1.1753,
      "step": 790
    },
    {
      "epoch": 0.022791448164582492,
      "grad_norm": 5.305600643157959,
      "learning_rate": 7.591170825335893e-06,
      "loss": 1.1835,
      "step": 791
    },
    {
      "epoch": 0.022820261626231776,
      "grad_norm": 3.7376410961151123,
      "learning_rate": 7.600767754318619e-06,
      "loss": 1.1742,
      "step": 792
    },
    {
      "epoch": 0.022849075087881057,
      "grad_norm": 4.5630011558532715,
      "learning_rate": 7.610364683301345e-06,
      "loss": 1.1618,
      "step": 793
    },
    {
      "epoch": 0.02287788854953034,
      "grad_norm": 4.287456512451172,
      "learning_rate": 7.61996161228407e-06,
      "loss": 1.136,
      "step": 794
    },
    {
      "epoch": 0.022906702011179624,
      "grad_norm": 4.114102840423584,
      "learning_rate": 7.629558541266795e-06,
      "loss": 1.1565,
      "step": 795
    },
    {
      "epoch": 0.022935515472828905,
      "grad_norm": 3.99790096282959,
      "learning_rate": 7.63915547024952e-06,
      "loss": 1.1203,
      "step": 796
    },
    {
      "epoch": 0.02296432893447819,
      "grad_norm": 4.1398186683654785,
      "learning_rate": 7.648752399232246e-06,
      "loss": 1.149,
      "step": 797
    },
    {
      "epoch": 0.022993142396127472,
      "grad_norm": 3.929283618927002,
      "learning_rate": 7.658349328214972e-06,
      "loss": 1.1649,
      "step": 798
    },
    {
      "epoch": 0.023021955857776753,
      "grad_norm": 4.05009651184082,
      "learning_rate": 7.667946257197698e-06,
      "loss": 1.168,
      "step": 799
    },
    {
      "epoch": 0.023050769319426036,
      "grad_norm": 3.597883462905884,
      "learning_rate": 7.677543186180423e-06,
      "loss": 1.1119,
      "step": 800
    },
    {
      "epoch": 0.023079582781075317,
      "grad_norm": 4.26542329788208,
      "learning_rate": 7.687140115163149e-06,
      "loss": 1.1274,
      "step": 801
    },
    {
      "epoch": 0.0231083962427246,
      "grad_norm": 3.4430577754974365,
      "learning_rate": 7.696737044145875e-06,
      "loss": 1.1155,
      "step": 802
    },
    {
      "epoch": 0.023137209704373884,
      "grad_norm": 4.452023029327393,
      "learning_rate": 7.706333973128599e-06,
      "loss": 1.1503,
      "step": 803
    },
    {
      "epoch": 0.023166023166023165,
      "grad_norm": 3.6511340141296387,
      "learning_rate": 7.715930902111324e-06,
      "loss": 1.1218,
      "step": 804
    },
    {
      "epoch": 0.02319483662767245,
      "grad_norm": 4.336791515350342,
      "learning_rate": 7.72552783109405e-06,
      "loss": 1.1637,
      "step": 805
    },
    {
      "epoch": 0.023223650089321732,
      "grad_norm": 4.188377857208252,
      "learning_rate": 7.735124760076776e-06,
      "loss": 1.1625,
      "step": 806
    },
    {
      "epoch": 0.023252463550971013,
      "grad_norm": 3.915614604949951,
      "learning_rate": 7.744721689059501e-06,
      "loss": 1.1322,
      "step": 807
    },
    {
      "epoch": 0.023281277012620297,
      "grad_norm": 4.460464954376221,
      "learning_rate": 7.754318618042227e-06,
      "loss": 1.1612,
      "step": 808
    },
    {
      "epoch": 0.023310090474269577,
      "grad_norm": 4.134449005126953,
      "learning_rate": 7.763915547024953e-06,
      "loss": 1.0946,
      "step": 809
    },
    {
      "epoch": 0.02333890393591886,
      "grad_norm": 4.762842178344727,
      "learning_rate": 7.773512476007678e-06,
      "loss": 1.1466,
      "step": 810
    },
    {
      "epoch": 0.023367717397568145,
      "grad_norm": 4.230005264282227,
      "learning_rate": 7.783109404990402e-06,
      "loss": 1.1264,
      "step": 811
    },
    {
      "epoch": 0.023396530859217425,
      "grad_norm": 4.220333099365234,
      "learning_rate": 7.79270633397313e-06,
      "loss": 1.1421,
      "step": 812
    },
    {
      "epoch": 0.02342534432086671,
      "grad_norm": 3.6539177894592285,
      "learning_rate": 7.802303262955856e-06,
      "loss": 1.1425,
      "step": 813
    },
    {
      "epoch": 0.023454157782515993,
      "grad_norm": 4.5919342041015625,
      "learning_rate": 7.811900191938581e-06,
      "loss": 1.1193,
      "step": 814
    },
    {
      "epoch": 0.023482971244165273,
      "grad_norm": 3.8865792751312256,
      "learning_rate": 7.821497120921305e-06,
      "loss": 1.1383,
      "step": 815
    },
    {
      "epoch": 0.023511784705814557,
      "grad_norm": 3.580781936645508,
      "learning_rate": 7.83109404990403e-06,
      "loss": 1.1545,
      "step": 816
    },
    {
      "epoch": 0.02354059816746384,
      "grad_norm": 4.027431964874268,
      "learning_rate": 7.840690978886757e-06,
      "loss": 1.1446,
      "step": 817
    },
    {
      "epoch": 0.02356941162911312,
      "grad_norm": 3.979562759399414,
      "learning_rate": 7.850287907869482e-06,
      "loss": 1.1534,
      "step": 818
    },
    {
      "epoch": 0.023598225090762405,
      "grad_norm": 3.595223903656006,
      "learning_rate": 7.859884836852208e-06,
      "loss": 1.1516,
      "step": 819
    },
    {
      "epoch": 0.023627038552411685,
      "grad_norm": 4.259181499481201,
      "learning_rate": 7.869481765834934e-06,
      "loss": 1.1357,
      "step": 820
    },
    {
      "epoch": 0.02365585201406097,
      "grad_norm": 3.6260271072387695,
      "learning_rate": 7.87907869481766e-06,
      "loss": 1.1425,
      "step": 821
    },
    {
      "epoch": 0.023684665475710253,
      "grad_norm": 4.261026382446289,
      "learning_rate": 7.888675623800385e-06,
      "loss": 1.1329,
      "step": 822
    },
    {
      "epoch": 0.023713478937359533,
      "grad_norm": 3.5031278133392334,
      "learning_rate": 7.89827255278311e-06,
      "loss": 1.1435,
      "step": 823
    },
    {
      "epoch": 0.023742292399008817,
      "grad_norm": 4.11406135559082,
      "learning_rate": 7.907869481765835e-06,
      "loss": 1.144,
      "step": 824
    },
    {
      "epoch": 0.0237711058606581,
      "grad_norm": 3.8027455806732178,
      "learning_rate": 7.91746641074856e-06,
      "loss": 1.1529,
      "step": 825
    },
    {
      "epoch": 0.02379991932230738,
      "grad_norm": 3.503267765045166,
      "learning_rate": 7.927063339731288e-06,
      "loss": 1.1289,
      "step": 826
    },
    {
      "epoch": 0.023828732783956665,
      "grad_norm": 3.501833200454712,
      "learning_rate": 7.936660268714013e-06,
      "loss": 1.125,
      "step": 827
    },
    {
      "epoch": 0.023857546245605946,
      "grad_norm": 3.6713240146636963,
      "learning_rate": 7.946257197696737e-06,
      "loss": 1.1512,
      "step": 828
    },
    {
      "epoch": 0.02388635970725523,
      "grad_norm": 3.96419095993042,
      "learning_rate": 7.955854126679463e-06,
      "loss": 1.157,
      "step": 829
    },
    {
      "epoch": 0.023915173168904513,
      "grad_norm": 3.599118709564209,
      "learning_rate": 7.965451055662189e-06,
      "loss": 1.1354,
      "step": 830
    },
    {
      "epoch": 0.023943986630553794,
      "grad_norm": 4.064572811126709,
      "learning_rate": 7.975047984644914e-06,
      "loss": 1.1434,
      "step": 831
    },
    {
      "epoch": 0.023972800092203077,
      "grad_norm": 3.359172821044922,
      "learning_rate": 7.98464491362764e-06,
      "loss": 1.1491,
      "step": 832
    },
    {
      "epoch": 0.02400161355385236,
      "grad_norm": 3.889259099960327,
      "learning_rate": 7.994241842610366e-06,
      "loss": 1.1549,
      "step": 833
    },
    {
      "epoch": 0.02403042701550164,
      "grad_norm": 3.3280441761016846,
      "learning_rate": 8.003838771593091e-06,
      "loss": 1.1387,
      "step": 834
    },
    {
      "epoch": 0.024059240477150926,
      "grad_norm": 4.185556888580322,
      "learning_rate": 8.013435700575817e-06,
      "loss": 1.1395,
      "step": 835
    },
    {
      "epoch": 0.024088053938800206,
      "grad_norm": 3.3834033012390137,
      "learning_rate": 8.023032629558541e-06,
      "loss": 1.1309,
      "step": 836
    },
    {
      "epoch": 0.02411686740044949,
      "grad_norm": 4.238918781280518,
      "learning_rate": 8.032629558541267e-06,
      "loss": 1.1165,
      "step": 837
    },
    {
      "epoch": 0.024145680862098774,
      "grad_norm": 3.2799878120422363,
      "learning_rate": 8.042226487523992e-06,
      "loss": 1.1025,
      "step": 838
    },
    {
      "epoch": 0.024174494323748054,
      "grad_norm": 3.535830020904541,
      "learning_rate": 8.05182341650672e-06,
      "loss": 1.1316,
      "step": 839
    },
    {
      "epoch": 0.024203307785397338,
      "grad_norm": 3.4083573818206787,
      "learning_rate": 8.061420345489444e-06,
      "loss": 1.1657,
      "step": 840
    },
    {
      "epoch": 0.02423212124704662,
      "grad_norm": 3.964585304260254,
      "learning_rate": 8.07101727447217e-06,
      "loss": 1.1549,
      "step": 841
    },
    {
      "epoch": 0.024260934708695902,
      "grad_norm": 3.8687498569488525,
      "learning_rate": 8.080614203454895e-06,
      "loss": 1.1072,
      "step": 842
    },
    {
      "epoch": 0.024289748170345186,
      "grad_norm": 3.721142053604126,
      "learning_rate": 8.09021113243762e-06,
      "loss": 1.1313,
      "step": 843
    },
    {
      "epoch": 0.024318561631994466,
      "grad_norm": 3.9654173851013184,
      "learning_rate": 8.099808061420346e-06,
      "loss": 1.1089,
      "step": 844
    },
    {
      "epoch": 0.02434737509364375,
      "grad_norm": 3.715157985687256,
      "learning_rate": 8.109404990403072e-06,
      "loss": 1.1192,
      "step": 845
    },
    {
      "epoch": 0.024376188555293034,
      "grad_norm": 4.0100860595703125,
      "learning_rate": 8.119001919385798e-06,
      "loss": 1.1485,
      "step": 846
    },
    {
      "epoch": 0.024405002016942314,
      "grad_norm": 4.1174492835998535,
      "learning_rate": 8.128598848368523e-06,
      "loss": 1.1693,
      "step": 847
    },
    {
      "epoch": 0.024433815478591598,
      "grad_norm": 3.9722213745117188,
      "learning_rate": 8.138195777351249e-06,
      "loss": 1.1061,
      "step": 848
    },
    {
      "epoch": 0.024462628940240882,
      "grad_norm": 3.8607895374298096,
      "learning_rate": 8.147792706333973e-06,
      "loss": 1.156,
      "step": 849
    },
    {
      "epoch": 0.024491442401890162,
      "grad_norm": 4.032652378082275,
      "learning_rate": 8.157389635316699e-06,
      "loss": 1.1258,
      "step": 850
    },
    {
      "epoch": 0.024520255863539446,
      "grad_norm": 3.4204723834991455,
      "learning_rate": 8.166986564299424e-06,
      "loss": 1.1196,
      "step": 851
    },
    {
      "epoch": 0.02454906932518873,
      "grad_norm": 4.43638277053833,
      "learning_rate": 8.176583493282152e-06,
      "loss": 1.1316,
      "step": 852
    },
    {
      "epoch": 0.02457788278683801,
      "grad_norm": 4.046248435974121,
      "learning_rate": 8.186180422264876e-06,
      "loss": 1.1548,
      "step": 853
    },
    {
      "epoch": 0.024606696248487294,
      "grad_norm": 4.050534248352051,
      "learning_rate": 8.195777351247601e-06,
      "loss": 1.1104,
      "step": 854
    },
    {
      "epoch": 0.024635509710136574,
      "grad_norm": 3.815157651901245,
      "learning_rate": 8.205374280230327e-06,
      "loss": 1.1667,
      "step": 855
    },
    {
      "epoch": 0.024664323171785858,
      "grad_norm": 3.777517080307007,
      "learning_rate": 8.214971209213053e-06,
      "loss": 1.1515,
      "step": 856
    },
    {
      "epoch": 0.024693136633435142,
      "grad_norm": 3.2882726192474365,
      "learning_rate": 8.224568138195778e-06,
      "loss": 1.1162,
      "step": 857
    },
    {
      "epoch": 0.024721950095084422,
      "grad_norm": 3.700364112854004,
      "learning_rate": 8.234165067178504e-06,
      "loss": 1.146,
      "step": 858
    },
    {
      "epoch": 0.024750763556733706,
      "grad_norm": 3.7294118404388428,
      "learning_rate": 8.24376199616123e-06,
      "loss": 1.1396,
      "step": 859
    },
    {
      "epoch": 0.02477957701838299,
      "grad_norm": 3.8575987815856934,
      "learning_rate": 8.253358925143955e-06,
      "loss": 1.1156,
      "step": 860
    },
    {
      "epoch": 0.02480839048003227,
      "grad_norm": 4.0562005043029785,
      "learning_rate": 8.26295585412668e-06,
      "loss": 1.1322,
      "step": 861
    },
    {
      "epoch": 0.024837203941681554,
      "grad_norm": 3.581068754196167,
      "learning_rate": 8.272552783109405e-06,
      "loss": 1.1413,
      "step": 862
    },
    {
      "epoch": 0.024866017403330835,
      "grad_norm": 3.716655731201172,
      "learning_rate": 8.28214971209213e-06,
      "loss": 1.1272,
      "step": 863
    },
    {
      "epoch": 0.02489483086498012,
      "grad_norm": 4.080185890197754,
      "learning_rate": 8.291746641074856e-06,
      "loss": 1.1133,
      "step": 864
    },
    {
      "epoch": 0.024923644326629402,
      "grad_norm": 3.5880560874938965,
      "learning_rate": 8.301343570057582e-06,
      "loss": 1.1303,
      "step": 865
    },
    {
      "epoch": 0.024952457788278683,
      "grad_norm": 3.6868550777435303,
      "learning_rate": 8.310940499040308e-06,
      "loss": 1.1161,
      "step": 866
    },
    {
      "epoch": 0.024981271249927967,
      "grad_norm": 3.9171061515808105,
      "learning_rate": 8.320537428023033e-06,
      "loss": 1.1349,
      "step": 867
    },
    {
      "epoch": 0.02501008471157725,
      "grad_norm": 3.789750099182129,
      "learning_rate": 8.330134357005759e-06,
      "loss": 1.1248,
      "step": 868
    },
    {
      "epoch": 0.02503889817322653,
      "grad_norm": 3.82517147064209,
      "learning_rate": 8.339731285988485e-06,
      "loss": 1.1479,
      "step": 869
    },
    {
      "epoch": 0.025067711634875815,
      "grad_norm": 3.973393201828003,
      "learning_rate": 8.34932821497121e-06,
      "loss": 1.0956,
      "step": 870
    },
    {
      "epoch": 0.025096525096525095,
      "grad_norm": 3.9322946071624756,
      "learning_rate": 8.358925143953936e-06,
      "loss": 1.1203,
      "step": 871
    },
    {
      "epoch": 0.02512533855817438,
      "grad_norm": 4.11823844909668,
      "learning_rate": 8.368522072936662e-06,
      "loss": 1.1552,
      "step": 872
    },
    {
      "epoch": 0.025154152019823663,
      "grad_norm": 4.08441162109375,
      "learning_rate": 8.378119001919387e-06,
      "loss": 1.1356,
      "step": 873
    },
    {
      "epoch": 0.025182965481472943,
      "grad_norm": 4.555100440979004,
      "learning_rate": 8.387715930902111e-06,
      "loss": 1.1615,
      "step": 874
    },
    {
      "epoch": 0.025211778943122227,
      "grad_norm": 4.3878631591796875,
      "learning_rate": 8.397312859884837e-06,
      "loss": 1.1306,
      "step": 875
    },
    {
      "epoch": 0.02524059240477151,
      "grad_norm": 4.014100074768066,
      "learning_rate": 8.406909788867563e-06,
      "loss": 1.0862,
      "step": 876
    },
    {
      "epoch": 0.02526940586642079,
      "grad_norm": 4.53034782409668,
      "learning_rate": 8.416506717850288e-06,
      "loss": 1.1196,
      "step": 877
    },
    {
      "epoch": 0.025298219328070075,
      "grad_norm": 3.7845611572265625,
      "learning_rate": 8.426103646833014e-06,
      "loss": 1.1488,
      "step": 878
    },
    {
      "epoch": 0.025327032789719355,
      "grad_norm": 4.334610462188721,
      "learning_rate": 8.43570057581574e-06,
      "loss": 1.1255,
      "step": 879
    },
    {
      "epoch": 0.02535584625136864,
      "grad_norm": 3.179733991622925,
      "learning_rate": 8.445297504798465e-06,
      "loss": 1.1299,
      "step": 880
    },
    {
      "epoch": 0.025384659713017923,
      "grad_norm": 4.251236438751221,
      "learning_rate": 8.454894433781191e-06,
      "loss": 1.1193,
      "step": 881
    },
    {
      "epoch": 0.025413473174667203,
      "grad_norm": 3.6483898162841797,
      "learning_rate": 8.464491362763915e-06,
      "loss": 1.1372,
      "step": 882
    },
    {
      "epoch": 0.025442286636316487,
      "grad_norm": 4.151557922363281,
      "learning_rate": 8.47408829174664e-06,
      "loss": 1.1387,
      "step": 883
    },
    {
      "epoch": 0.02547110009796577,
      "grad_norm": 3.881044626235962,
      "learning_rate": 8.483685220729368e-06,
      "loss": 1.1247,
      "step": 884
    },
    {
      "epoch": 0.02549991355961505,
      "grad_norm": 3.561208486557007,
      "learning_rate": 8.493282149712094e-06,
      "loss": 1.1062,
      "step": 885
    },
    {
      "epoch": 0.025528727021264335,
      "grad_norm": 4.498389720916748,
      "learning_rate": 8.502879078694818e-06,
      "loss": 1.1161,
      "step": 886
    },
    {
      "epoch": 0.02555754048291362,
      "grad_norm": 3.4005863666534424,
      "learning_rate": 8.512476007677543e-06,
      "loss": 1.136,
      "step": 887
    },
    {
      "epoch": 0.0255863539445629,
      "grad_norm": 3.8027422428131104,
      "learning_rate": 8.522072936660269e-06,
      "loss": 1.1264,
      "step": 888
    },
    {
      "epoch": 0.025615167406212183,
      "grad_norm": 3.7187273502349854,
      "learning_rate": 8.531669865642995e-06,
      "loss": 1.1476,
      "step": 889
    },
    {
      "epoch": 0.025643980867861463,
      "grad_norm": 3.614673614501953,
      "learning_rate": 8.54126679462572e-06,
      "loss": 1.1248,
      "step": 890
    },
    {
      "epoch": 0.025672794329510747,
      "grad_norm": 3.7979624271392822,
      "learning_rate": 8.550863723608446e-06,
      "loss": 1.1391,
      "step": 891
    },
    {
      "epoch": 0.02570160779116003,
      "grad_norm": 3.4442977905273438,
      "learning_rate": 8.560460652591172e-06,
      "loss": 1.1246,
      "step": 892
    },
    {
      "epoch": 0.02573042125280931,
      "grad_norm": 3.9618544578552246,
      "learning_rate": 8.570057581573898e-06,
      "loss": 1.1301,
      "step": 893
    },
    {
      "epoch": 0.025759234714458595,
      "grad_norm": 3.35534930229187,
      "learning_rate": 8.579654510556623e-06,
      "loss": 1.1243,
      "step": 894
    },
    {
      "epoch": 0.02578804817610788,
      "grad_norm": 3.4116005897521973,
      "learning_rate": 8.589251439539347e-06,
      "loss": 1.1143,
      "step": 895
    },
    {
      "epoch": 0.02581686163775716,
      "grad_norm": 3.468977689743042,
      "learning_rate": 8.598848368522073e-06,
      "loss": 1.1327,
      "step": 896
    },
    {
      "epoch": 0.025845675099406443,
      "grad_norm": 3.4935736656188965,
      "learning_rate": 8.6084452975048e-06,
      "loss": 1.133,
      "step": 897
    },
    {
      "epoch": 0.025874488561055724,
      "grad_norm": 3.6882457733154297,
      "learning_rate": 8.618042226487526e-06,
      "loss": 1.1475,
      "step": 898
    },
    {
      "epoch": 0.025903302022705008,
      "grad_norm": 3.4292171001434326,
      "learning_rate": 8.62763915547025e-06,
      "loss": 1.1531,
      "step": 899
    },
    {
      "epoch": 0.02593211548435429,
      "grad_norm": 3.2636475563049316,
      "learning_rate": 8.637236084452976e-06,
      "loss": 1.1093,
      "step": 900
    },
    {
      "epoch": 0.025960928946003572,
      "grad_norm": 3.639920234680176,
      "learning_rate": 8.646833013435701e-06,
      "loss": 1.1243,
      "step": 901
    },
    {
      "epoch": 0.025989742407652856,
      "grad_norm": 3.747945547103882,
      "learning_rate": 8.656429942418427e-06,
      "loss": 1.1389,
      "step": 902
    },
    {
      "epoch": 0.02601855586930214,
      "grad_norm": 3.566969633102417,
      "learning_rate": 8.666026871401153e-06,
      "loss": 1.1288,
      "step": 903
    },
    {
      "epoch": 0.02604736933095142,
      "grad_norm": 3.7548422813415527,
      "learning_rate": 8.675623800383878e-06,
      "loss": 1.1341,
      "step": 904
    },
    {
      "epoch": 0.026076182792600704,
      "grad_norm": 3.429175853729248,
      "learning_rate": 8.685220729366604e-06,
      "loss": 1.1543,
      "step": 905
    },
    {
      "epoch": 0.026104996254249984,
      "grad_norm": 3.486872911453247,
      "learning_rate": 8.69481765834933e-06,
      "loss": 1.1004,
      "step": 906
    },
    {
      "epoch": 0.026133809715899268,
      "grad_norm": 3.545693874359131,
      "learning_rate": 8.704414587332054e-06,
      "loss": 1.1293,
      "step": 907
    },
    {
      "epoch": 0.02616262317754855,
      "grad_norm": 3.4742209911346436,
      "learning_rate": 8.71401151631478e-06,
      "loss": 1.1649,
      "step": 908
    },
    {
      "epoch": 0.026191436639197832,
      "grad_norm": 3.5925490856170654,
      "learning_rate": 8.723608445297505e-06,
      "loss": 1.1548,
      "step": 909
    },
    {
      "epoch": 0.026220250100847116,
      "grad_norm": 3.4911487102508545,
      "learning_rate": 8.73320537428023e-06,
      "loss": 1.1212,
      "step": 910
    },
    {
      "epoch": 0.0262490635624964,
      "grad_norm": 3.9086415767669678,
      "learning_rate": 8.742802303262956e-06,
      "loss": 1.125,
      "step": 911
    },
    {
      "epoch": 0.02627787702414568,
      "grad_norm": 4.030948162078857,
      "learning_rate": 8.752399232245682e-06,
      "loss": 1.0982,
      "step": 912
    },
    {
      "epoch": 0.026306690485794964,
      "grad_norm": 3.60432505607605,
      "learning_rate": 8.761996161228408e-06,
      "loss": 1.124,
      "step": 913
    },
    {
      "epoch": 0.026335503947444244,
      "grad_norm": 3.520704746246338,
      "learning_rate": 8.771593090211133e-06,
      "loss": 1.0897,
      "step": 914
    },
    {
      "epoch": 0.026364317409093528,
      "grad_norm": 3.9046900272369385,
      "learning_rate": 8.781190019193859e-06,
      "loss": 1.1142,
      "step": 915
    },
    {
      "epoch": 0.026393130870742812,
      "grad_norm": 3.326414108276367,
      "learning_rate": 8.790786948176585e-06,
      "loss": 1.1073,
      "step": 916
    },
    {
      "epoch": 0.026421944332392092,
      "grad_norm": 3.6688995361328125,
      "learning_rate": 8.80038387715931e-06,
      "loss": 1.1156,
      "step": 917
    },
    {
      "epoch": 0.026450757794041376,
      "grad_norm": 3.916806221008301,
      "learning_rate": 8.809980806142036e-06,
      "loss": 1.144,
      "step": 918
    },
    {
      "epoch": 0.02647957125569066,
      "grad_norm": 3.345334053039551,
      "learning_rate": 8.819577735124762e-06,
      "loss": 1.0866,
      "step": 919
    },
    {
      "epoch": 0.02650838471733994,
      "grad_norm": 3.8898332118988037,
      "learning_rate": 8.829174664107486e-06,
      "loss": 1.1365,
      "step": 920
    },
    {
      "epoch": 0.026537198178989224,
      "grad_norm": 3.635438919067383,
      "learning_rate": 8.838771593090211e-06,
      "loss": 1.1066,
      "step": 921
    },
    {
      "epoch": 0.026566011640638508,
      "grad_norm": 4.200286388397217,
      "learning_rate": 8.848368522072937e-06,
      "loss": 1.1523,
      "step": 922
    },
    {
      "epoch": 0.02659482510228779,
      "grad_norm": 3.407059669494629,
      "learning_rate": 8.857965451055663e-06,
      "loss": 1.1022,
      "step": 923
    },
    {
      "epoch": 0.026623638563937072,
      "grad_norm": 4.225884437561035,
      "learning_rate": 8.867562380038388e-06,
      "loss": 1.1311,
      "step": 924
    },
    {
      "epoch": 0.026652452025586353,
      "grad_norm": 3.580275058746338,
      "learning_rate": 8.877159309021114e-06,
      "loss": 1.1486,
      "step": 925
    },
    {
      "epoch": 0.026681265487235636,
      "grad_norm": 3.721985101699829,
      "learning_rate": 8.88675623800384e-06,
      "loss": 1.136,
      "step": 926
    },
    {
      "epoch": 0.02671007894888492,
      "grad_norm": 3.8082644939422607,
      "learning_rate": 8.896353166986565e-06,
      "loss": 1.1222,
      "step": 927
    },
    {
      "epoch": 0.0267388924105342,
      "grad_norm": 3.53515887260437,
      "learning_rate": 8.905950095969291e-06,
      "loss": 1.1387,
      "step": 928
    },
    {
      "epoch": 0.026767705872183484,
      "grad_norm": 3.515903949737549,
      "learning_rate": 8.915547024952017e-06,
      "loss": 1.1101,
      "step": 929
    },
    {
      "epoch": 0.02679651933383277,
      "grad_norm": 3.4120898246765137,
      "learning_rate": 8.925143953934742e-06,
      "loss": 1.127,
      "step": 930
    },
    {
      "epoch": 0.02682533279548205,
      "grad_norm": 3.6402719020843506,
      "learning_rate": 8.934740882917468e-06,
      "loss": 1.1379,
      "step": 931
    },
    {
      "epoch": 0.026854146257131332,
      "grad_norm": 3.3321821689605713,
      "learning_rate": 8.944337811900192e-06,
      "loss": 1.1188,
      "step": 932
    },
    {
      "epoch": 0.026882959718780613,
      "grad_norm": 3.6361520290374756,
      "learning_rate": 8.953934740882918e-06,
      "loss": 1.1184,
      "step": 933
    },
    {
      "epoch": 0.026911773180429897,
      "grad_norm": 3.3544652462005615,
      "learning_rate": 8.963531669865643e-06,
      "loss": 1.1555,
      "step": 934
    },
    {
      "epoch": 0.02694058664207918,
      "grad_norm": 4.085747718811035,
      "learning_rate": 8.973128598848369e-06,
      "loss": 1.1266,
      "step": 935
    },
    {
      "epoch": 0.02696940010372846,
      "grad_norm": 3.6661884784698486,
      "learning_rate": 8.982725527831095e-06,
      "loss": 1.1383,
      "step": 936
    },
    {
      "epoch": 0.026998213565377745,
      "grad_norm": 3.3774831295013428,
      "learning_rate": 8.99232245681382e-06,
      "loss": 1.1009,
      "step": 937
    },
    {
      "epoch": 0.02702702702702703,
      "grad_norm": 3.3285014629364014,
      "learning_rate": 9.001919385796546e-06,
      "loss": 1.1282,
      "step": 938
    },
    {
      "epoch": 0.02705584048867631,
      "grad_norm": 3.6275596618652344,
      "learning_rate": 9.011516314779272e-06,
      "loss": 1.1044,
      "step": 939
    },
    {
      "epoch": 0.027084653950325593,
      "grad_norm": 3.303631544113159,
      "learning_rate": 9.021113243761996e-06,
      "loss": 1.1548,
      "step": 940
    },
    {
      "epoch": 0.027113467411974873,
      "grad_norm": 3.8610806465148926,
      "learning_rate": 9.030710172744721e-06,
      "loss": 1.1516,
      "step": 941
    },
    {
      "epoch": 0.027142280873624157,
      "grad_norm": 3.733711004257202,
      "learning_rate": 9.040307101727449e-06,
      "loss": 1.1019,
      "step": 942
    },
    {
      "epoch": 0.02717109433527344,
      "grad_norm": 3.346266269683838,
      "learning_rate": 9.049904030710174e-06,
      "loss": 1.0887,
      "step": 943
    },
    {
      "epoch": 0.02719990779692272,
      "grad_norm": 3.5379767417907715,
      "learning_rate": 9.0595009596929e-06,
      "loss": 1.1505,
      "step": 944
    },
    {
      "epoch": 0.027228721258572005,
      "grad_norm": 3.6364848613739014,
      "learning_rate": 9.069097888675624e-06,
      "loss": 1.1246,
      "step": 945
    },
    {
      "epoch": 0.02725753472022129,
      "grad_norm": 3.620232582092285,
      "learning_rate": 9.07869481765835e-06,
      "loss": 1.1596,
      "step": 946
    },
    {
      "epoch": 0.02728634818187057,
      "grad_norm": 3.4510655403137207,
      "learning_rate": 9.088291746641075e-06,
      "loss": 1.1024,
      "step": 947
    },
    {
      "epoch": 0.027315161643519853,
      "grad_norm": 3.862393856048584,
      "learning_rate": 9.097888675623801e-06,
      "loss": 1.1443,
      "step": 948
    },
    {
      "epoch": 0.027343975105169133,
      "grad_norm": 3.5251283645629883,
      "learning_rate": 9.107485604606527e-06,
      "loss": 1.1253,
      "step": 949
    },
    {
      "epoch": 0.027372788566818417,
      "grad_norm": 3.4761600494384766,
      "learning_rate": 9.117082533589252e-06,
      "loss": 1.0925,
      "step": 950
    },
    {
      "epoch": 0.0274016020284677,
      "grad_norm": 3.8886778354644775,
      "learning_rate": 9.126679462571978e-06,
      "loss": 1.1184,
      "step": 951
    },
    {
      "epoch": 0.02743041549011698,
      "grad_norm": 3.570359945297241,
      "learning_rate": 9.136276391554704e-06,
      "loss": 1.1,
      "step": 952
    },
    {
      "epoch": 0.027459228951766265,
      "grad_norm": 3.8500733375549316,
      "learning_rate": 9.145873320537428e-06,
      "loss": 1.1279,
      "step": 953
    },
    {
      "epoch": 0.02748804241341555,
      "grad_norm": 4.3769378662109375,
      "learning_rate": 9.155470249520153e-06,
      "loss": 1.1023,
      "step": 954
    },
    {
      "epoch": 0.02751685587506483,
      "grad_norm": 3.7192792892456055,
      "learning_rate": 9.16506717850288e-06,
      "loss": 1.118,
      "step": 955
    },
    {
      "epoch": 0.027545669336714113,
      "grad_norm": 3.8953371047973633,
      "learning_rate": 9.174664107485606e-06,
      "loss": 1.1395,
      "step": 956
    },
    {
      "epoch": 0.027574482798363397,
      "grad_norm": 3.593451499938965,
      "learning_rate": 9.18426103646833e-06,
      "loss": 1.1256,
      "step": 957
    },
    {
      "epoch": 0.027603296260012677,
      "grad_norm": 4.114181995391846,
      "learning_rate": 9.193857965451056e-06,
      "loss": 1.1047,
      "step": 958
    },
    {
      "epoch": 0.02763210972166196,
      "grad_norm": 4.020487308502197,
      "learning_rate": 9.203454894433782e-06,
      "loss": 1.159,
      "step": 959
    },
    {
      "epoch": 0.02766092318331124,
      "grad_norm": 3.524401903152466,
      "learning_rate": 9.213051823416507e-06,
      "loss": 1.1386,
      "step": 960
    },
    {
      "epoch": 0.027689736644960525,
      "grad_norm": 3.8225221633911133,
      "learning_rate": 9.222648752399233e-06,
      "loss": 1.1158,
      "step": 961
    },
    {
      "epoch": 0.02771855010660981,
      "grad_norm": 3.590688467025757,
      "learning_rate": 9.232245681381959e-06,
      "loss": 1.1358,
      "step": 962
    },
    {
      "epoch": 0.02774736356825909,
      "grad_norm": 4.175741195678711,
      "learning_rate": 9.241842610364684e-06,
      "loss": 1.1079,
      "step": 963
    },
    {
      "epoch": 0.027776177029908374,
      "grad_norm": 3.634157657623291,
      "learning_rate": 9.25143953934741e-06,
      "loss": 1.1315,
      "step": 964
    },
    {
      "epoch": 0.027804990491557657,
      "grad_norm": 3.8993875980377197,
      "learning_rate": 9.261036468330134e-06,
      "loss": 1.1302,
      "step": 965
    },
    {
      "epoch": 0.027833803953206938,
      "grad_norm": 3.1458990573883057,
      "learning_rate": 9.27063339731286e-06,
      "loss": 1.0993,
      "step": 966
    },
    {
      "epoch": 0.02786261741485622,
      "grad_norm": 4.01068639755249,
      "learning_rate": 9.280230326295585e-06,
      "loss": 1.106,
      "step": 967
    },
    {
      "epoch": 0.027891430876505502,
      "grad_norm": 3.488621711730957,
      "learning_rate": 9.289827255278311e-06,
      "loss": 1.102,
      "step": 968
    },
    {
      "epoch": 0.027920244338154786,
      "grad_norm": 3.7665939331054688,
      "learning_rate": 9.299424184261039e-06,
      "loss": 1.1445,
      "step": 969
    },
    {
      "epoch": 0.02794905779980407,
      "grad_norm": 3.424177646636963,
      "learning_rate": 9.309021113243763e-06,
      "loss": 1.1067,
      "step": 970
    },
    {
      "epoch": 0.02797787126145335,
      "grad_norm": 4.111751556396484,
      "learning_rate": 9.318618042226488e-06,
      "loss": 1.1008,
      "step": 971
    },
    {
      "epoch": 0.028006684723102634,
      "grad_norm": 3.6089649200439453,
      "learning_rate": 9.328214971209214e-06,
      "loss": 1.142,
      "step": 972
    },
    {
      "epoch": 0.028035498184751918,
      "grad_norm": 4.8920722007751465,
      "learning_rate": 9.33781190019194e-06,
      "loss": 1.1277,
      "step": 973
    },
    {
      "epoch": 0.028064311646401198,
      "grad_norm": 4.13226318359375,
      "learning_rate": 9.347408829174665e-06,
      "loss": 1.1348,
      "step": 974
    },
    {
      "epoch": 0.028093125108050482,
      "grad_norm": 3.1769611835479736,
      "learning_rate": 9.357005758157391e-06,
      "loss": 1.1165,
      "step": 975
    },
    {
      "epoch": 0.028121938569699762,
      "grad_norm": 3.9498543739318848,
      "learning_rate": 9.366602687140117e-06,
      "loss": 1.1623,
      "step": 976
    },
    {
      "epoch": 0.028150752031349046,
      "grad_norm": 3.8442442417144775,
      "learning_rate": 9.376199616122842e-06,
      "loss": 1.0972,
      "step": 977
    },
    {
      "epoch": 0.02817956549299833,
      "grad_norm": 3.887212038040161,
      "learning_rate": 9.385796545105566e-06,
      "loss": 1.1159,
      "step": 978
    },
    {
      "epoch": 0.02820837895464761,
      "grad_norm": 3.902120590209961,
      "learning_rate": 9.395393474088292e-06,
      "loss": 1.115,
      "step": 979
    },
    {
      "epoch": 0.028237192416296894,
      "grad_norm": 3.5360090732574463,
      "learning_rate": 9.404990403071018e-06,
      "loss": 1.1417,
      "step": 980
    },
    {
      "epoch": 0.028266005877946178,
      "grad_norm": 3.806800365447998,
      "learning_rate": 9.414587332053743e-06,
      "loss": 1.146,
      "step": 981
    },
    {
      "epoch": 0.028294819339595458,
      "grad_norm": 4.111157417297363,
      "learning_rate": 9.424184261036469e-06,
      "loss": 1.1143,
      "step": 982
    },
    {
      "epoch": 0.028323632801244742,
      "grad_norm": 3.380190372467041,
      "learning_rate": 9.433781190019195e-06,
      "loss": 1.0871,
      "step": 983
    },
    {
      "epoch": 0.028352446262894022,
      "grad_norm": 3.811392307281494,
      "learning_rate": 9.44337811900192e-06,
      "loss": 1.1122,
      "step": 984
    },
    {
      "epoch": 0.028381259724543306,
      "grad_norm": 3.5284457206726074,
      "learning_rate": 9.452975047984646e-06,
      "loss": 1.1207,
      "step": 985
    },
    {
      "epoch": 0.02841007318619259,
      "grad_norm": 4.067584991455078,
      "learning_rate": 9.46257197696737e-06,
      "loss": 1.1008,
      "step": 986
    },
    {
      "epoch": 0.02843888664784187,
      "grad_norm": 3.564065933227539,
      "learning_rate": 9.472168905950097e-06,
      "loss": 1.1267,
      "step": 987
    },
    {
      "epoch": 0.028467700109491154,
      "grad_norm": 3.357872724533081,
      "learning_rate": 9.481765834932823e-06,
      "loss": 1.1117,
      "step": 988
    },
    {
      "epoch": 0.028496513571140438,
      "grad_norm": 3.6159183979034424,
      "learning_rate": 9.491362763915549e-06,
      "loss": 1.0639,
      "step": 989
    },
    {
      "epoch": 0.02852532703278972,
      "grad_norm": 3.6005702018737793,
      "learning_rate": 9.500959692898273e-06,
      "loss": 1.1166,
      "step": 990
    },
    {
      "epoch": 0.028554140494439002,
      "grad_norm": 3.55873441696167,
      "learning_rate": 9.510556621880998e-06,
      "loss": 1.0883,
      "step": 991
    },
    {
      "epoch": 0.028582953956088286,
      "grad_norm": 3.438173770904541,
      "learning_rate": 9.520153550863724e-06,
      "loss": 1.1168,
      "step": 992
    },
    {
      "epoch": 0.028611767417737567,
      "grad_norm": 3.3643343448638916,
      "learning_rate": 9.52975047984645e-06,
      "loss": 1.1044,
      "step": 993
    },
    {
      "epoch": 0.02864058087938685,
      "grad_norm": 3.694304943084717,
      "learning_rate": 9.539347408829175e-06,
      "loss": 1.1121,
      "step": 994
    },
    {
      "epoch": 0.02866939434103613,
      "grad_norm": 3.8476574420928955,
      "learning_rate": 9.548944337811901e-06,
      "loss": 1.1439,
      "step": 995
    },
    {
      "epoch": 0.028698207802685415,
      "grad_norm": 3.4998040199279785,
      "learning_rate": 9.558541266794627e-06,
      "loss": 1.1141,
      "step": 996
    },
    {
      "epoch": 0.0287270212643347,
      "grad_norm": 3.065563201904297,
      "learning_rate": 9.568138195777352e-06,
      "loss": 1.0942,
      "step": 997
    },
    {
      "epoch": 0.02875583472598398,
      "grad_norm": 3.46659517288208,
      "learning_rate": 9.577735124760078e-06,
      "loss": 1.1587,
      "step": 998
    },
    {
      "epoch": 0.028784648187633263,
      "grad_norm": 4.073842525482178,
      "learning_rate": 9.587332053742802e-06,
      "loss": 1.1181,
      "step": 999
    },
    {
      "epoch": 0.028813461649282546,
      "grad_norm": 3.4769318103790283,
      "learning_rate": 9.59692898272553e-06,
      "loss": 1.108,
      "step": 1000
    },
    {
      "epoch": 0.028842275110931827,
      "grad_norm": 3.465334177017212,
      "learning_rate": 9.606525911708255e-06,
      "loss": 1.1244,
      "step": 1001
    },
    {
      "epoch": 0.02887108857258111,
      "grad_norm": 3.5895333290100098,
      "learning_rate": 9.61612284069098e-06,
      "loss": 1.1189,
      "step": 1002
    },
    {
      "epoch": 0.02889990203423039,
      "grad_norm": 3.959467887878418,
      "learning_rate": 9.625719769673705e-06,
      "loss": 1.1395,
      "step": 1003
    },
    {
      "epoch": 0.028928715495879675,
      "grad_norm": 3.297013759613037,
      "learning_rate": 9.63531669865643e-06,
      "loss": 1.124,
      "step": 1004
    },
    {
      "epoch": 0.02895752895752896,
      "grad_norm": 3.888859510421753,
      "learning_rate": 9.644913627639156e-06,
      "loss": 1.1043,
      "step": 1005
    },
    {
      "epoch": 0.02898634241917824,
      "grad_norm": 3.454167127609253,
      "learning_rate": 9.654510556621882e-06,
      "loss": 1.1029,
      "step": 1006
    },
    {
      "epoch": 0.029015155880827523,
      "grad_norm": 3.607602834701538,
      "learning_rate": 9.664107485604607e-06,
      "loss": 1.1071,
      "step": 1007
    },
    {
      "epoch": 0.029043969342476807,
      "grad_norm": 3.5585615634918213,
      "learning_rate": 9.673704414587333e-06,
      "loss": 1.1183,
      "step": 1008
    },
    {
      "epoch": 0.029072782804126087,
      "grad_norm": 3.6066243648529053,
      "learning_rate": 9.683301343570059e-06,
      "loss": 1.1005,
      "step": 1009
    },
    {
      "epoch": 0.02910159626577537,
      "grad_norm": 3.4821057319641113,
      "learning_rate": 9.692898272552784e-06,
      "loss": 1.1215,
      "step": 1010
    },
    {
      "epoch": 0.02913040972742465,
      "grad_norm": 3.45961594581604,
      "learning_rate": 9.702495201535508e-06,
      "loss": 1.1164,
      "step": 1011
    },
    {
      "epoch": 0.029159223189073935,
      "grad_norm": 3.0937142372131348,
      "learning_rate": 9.712092130518234e-06,
      "loss": 1.1093,
      "step": 1012
    },
    {
      "epoch": 0.02918803665072322,
      "grad_norm": 3.727264881134033,
      "learning_rate": 9.721689059500961e-06,
      "loss": 1.1106,
      "step": 1013
    },
    {
      "epoch": 0.0292168501123725,
      "grad_norm": 3.4310195446014404,
      "learning_rate": 9.731285988483687e-06,
      "loss": 1.1141,
      "step": 1014
    },
    {
      "epoch": 0.029245663574021783,
      "grad_norm": 3.560608148574829,
      "learning_rate": 9.740882917466411e-06,
      "loss": 1.085,
      "step": 1015
    },
    {
      "epoch": 0.029274477035671067,
      "grad_norm": 3.181030035018921,
      "learning_rate": 9.750479846449137e-06,
      "loss": 1.1113,
      "step": 1016
    },
    {
      "epoch": 0.029303290497320347,
      "grad_norm": 3.350203275680542,
      "learning_rate": 9.760076775431862e-06,
      "loss": 1.1256,
      "step": 1017
    },
    {
      "epoch": 0.02933210395896963,
      "grad_norm": 4.041247367858887,
      "learning_rate": 9.769673704414588e-06,
      "loss": 1.1234,
      "step": 1018
    },
    {
      "epoch": 0.02936091742061891,
      "grad_norm": 3.167494058609009,
      "learning_rate": 9.779270633397314e-06,
      "loss": 1.1067,
      "step": 1019
    },
    {
      "epoch": 0.029389730882268195,
      "grad_norm": 3.649782657623291,
      "learning_rate": 9.78886756238004e-06,
      "loss": 1.1084,
      "step": 1020
    },
    {
      "epoch": 0.02941854434391748,
      "grad_norm": 3.280428171157837,
      "learning_rate": 9.798464491362765e-06,
      "loss": 1.0597,
      "step": 1021
    },
    {
      "epoch": 0.02944735780556676,
      "grad_norm": 3.535348892211914,
      "learning_rate": 9.80806142034549e-06,
      "loss": 1.1141,
      "step": 1022
    },
    {
      "epoch": 0.029476171267216043,
      "grad_norm": 3.2974886894226074,
      "learning_rate": 9.817658349328216e-06,
      "loss": 1.0953,
      "step": 1023
    },
    {
      "epoch": 0.029504984728865327,
      "grad_norm": 3.667489767074585,
      "learning_rate": 9.82725527831094e-06,
      "loss": 1.0929,
      "step": 1024
    },
    {
      "epoch": 0.029533798190514608,
      "grad_norm": 13.384965896606445,
      "learning_rate": 9.836852207293666e-06,
      "loss": 1.1318,
      "step": 1025
    },
    {
      "epoch": 0.02956261165216389,
      "grad_norm": 3.5465915203094482,
      "learning_rate": 9.846449136276392e-06,
      "loss": 1.1013,
      "step": 1026
    },
    {
      "epoch": 0.029591425113813175,
      "grad_norm": 3.6518232822418213,
      "learning_rate": 9.856046065259119e-06,
      "loss": 1.0901,
      "step": 1027
    },
    {
      "epoch": 0.029620238575462456,
      "grad_norm": 3.576354503631592,
      "learning_rate": 9.865642994241843e-06,
      "loss": 1.1277,
      "step": 1028
    },
    {
      "epoch": 0.02964905203711174,
      "grad_norm": 3.7753243446350098,
      "learning_rate": 9.875239923224569e-06,
      "loss": 1.1315,
      "step": 1029
    },
    {
      "epoch": 0.02967786549876102,
      "grad_norm": 3.1885297298431396,
      "learning_rate": 9.884836852207294e-06,
      "loss": 1.1274,
      "step": 1030
    },
    {
      "epoch": 0.029706678960410304,
      "grad_norm": 3.5630791187286377,
      "learning_rate": 9.89443378119002e-06,
      "loss": 1.1229,
      "step": 1031
    },
    {
      "epoch": 0.029735492422059587,
      "grad_norm": 3.0700974464416504,
      "learning_rate": 9.904030710172746e-06,
      "loss": 1.1115,
      "step": 1032
    },
    {
      "epoch": 0.029764305883708868,
      "grad_norm": 7.162629127502441,
      "learning_rate": 9.913627639155471e-06,
      "loss": 1.1393,
      "step": 1033
    },
    {
      "epoch": 0.02979311934535815,
      "grad_norm": 3.3096232414245605,
      "learning_rate": 9.923224568138197e-06,
      "loss": 1.0983,
      "step": 1034
    },
    {
      "epoch": 0.029821932807007435,
      "grad_norm": 3.546894073486328,
      "learning_rate": 9.932821497120923e-06,
      "loss": 1.0909,
      "step": 1035
    },
    {
      "epoch": 0.029850746268656716,
      "grad_norm": 3.9209654331207275,
      "learning_rate": 9.942418426103647e-06,
      "loss": 1.1149,
      "step": 1036
    },
    {
      "epoch": 0.029879559730306,
      "grad_norm": 3.2585508823394775,
      "learning_rate": 9.952015355086372e-06,
      "loss": 1.0974,
      "step": 1037
    },
    {
      "epoch": 0.02990837319195528,
      "grad_norm": 5.055637359619141,
      "learning_rate": 9.961612284069098e-06,
      "loss": 1.1156,
      "step": 1038
    },
    {
      "epoch": 0.029937186653604564,
      "grad_norm": 3.6369500160217285,
      "learning_rate": 9.971209213051824e-06,
      "loss": 1.119,
      "step": 1039
    },
    {
      "epoch": 0.029966000115253848,
      "grad_norm": 4.196709156036377,
      "learning_rate": 9.98080614203455e-06,
      "loss": 1.1174,
      "step": 1040
    },
    {
      "epoch": 0.029994813576903128,
      "grad_norm": 3.7355072498321533,
      "learning_rate": 9.990403071017275e-06,
      "loss": 1.1224,
      "step": 1041
    },
    {
      "epoch": 0.030023627038552412,
      "grad_norm": 4.206112384796143,
      "learning_rate": 1e-05,
      "loss": 1.1107,
      "step": 1042
    },
    {
      "epoch": 0.030052440500201696,
      "grad_norm": 3.745232582092285,
      "learning_rate": 9.9999999782275e-06,
      "loss": 1.0927,
      "step": 1043
    },
    {
      "epoch": 0.030081253961850976,
      "grad_norm": 3.393821954727173,
      "learning_rate": 9.999999912909996e-06,
      "loss": 1.077,
      "step": 1044
    },
    {
      "epoch": 0.03011006742350026,
      "grad_norm": 3.2777559757232666,
      "learning_rate": 9.999999804047494e-06,
      "loss": 1.1128,
      "step": 1045
    },
    {
      "epoch": 0.03013888088514954,
      "grad_norm": 3.629894256591797,
      "learning_rate": 9.99999965163999e-06,
      "loss": 1.1067,
      "step": 1046
    },
    {
      "epoch": 0.030167694346798824,
      "grad_norm": 9.907320022583008,
      "learning_rate": 9.999999455687488e-06,
      "loss": 1.0958,
      "step": 1047
    },
    {
      "epoch": 0.030196507808448108,
      "grad_norm": 3.6301512718200684,
      "learning_rate": 9.99999921618999e-06,
      "loss": 1.1034,
      "step": 1048
    },
    {
      "epoch": 0.03022532127009739,
      "grad_norm": 3.4117822647094727,
      "learning_rate": 9.999998933147494e-06,
      "loss": 1.1279,
      "step": 1049
    },
    {
      "epoch": 0.030254134731746672,
      "grad_norm": 3.6371490955352783,
      "learning_rate": 9.999998606560006e-06,
      "loss": 1.1638,
      "step": 1050
    },
    {
      "epoch": 0.030282948193395956,
      "grad_norm": 3.7794809341430664,
      "learning_rate": 9.999998236427528e-06,
      "loss": 1.1083,
      "step": 1051
    },
    {
      "epoch": 0.030311761655045236,
      "grad_norm": 3.013923168182373,
      "learning_rate": 9.999997822750065e-06,
      "loss": 1.1164,
      "step": 1052
    },
    {
      "epoch": 0.03034057511669452,
      "grad_norm": 4.35982608795166,
      "learning_rate": 9.99999736552762e-06,
      "loss": 1.0922,
      "step": 1053
    },
    {
      "epoch": 0.0303693885783438,
      "grad_norm": 3.393630266189575,
      "learning_rate": 9.999996864760195e-06,
      "loss": 1.0964,
      "step": 1054
    },
    {
      "epoch": 0.030398202039993084,
      "grad_norm": 3.7991855144500732,
      "learning_rate": 9.999996320447795e-06,
      "loss": 1.0733,
      "step": 1055
    },
    {
      "epoch": 0.030427015501642368,
      "grad_norm": 3.8119022846221924,
      "learning_rate": 9.999995732590426e-06,
      "loss": 1.1146,
      "step": 1056
    },
    {
      "epoch": 0.03045582896329165,
      "grad_norm": 3.2988317012786865,
      "learning_rate": 9.999995101188091e-06,
      "loss": 1.1283,
      "step": 1057
    },
    {
      "epoch": 0.030484642424940932,
      "grad_norm": 3.330155849456787,
      "learning_rate": 9.999994426240797e-06,
      "loss": 1.0994,
      "step": 1058
    },
    {
      "epoch": 0.030513455886590216,
      "grad_norm": 3.3248708248138428,
      "learning_rate": 9.999993707748552e-06,
      "loss": 1.0996,
      "step": 1059
    },
    {
      "epoch": 0.030542269348239497,
      "grad_norm": 3.253984212875366,
      "learning_rate": 9.999992945711358e-06,
      "loss": 1.0881,
      "step": 1060
    },
    {
      "epoch": 0.03057108280988878,
      "grad_norm": 3.554265260696411,
      "learning_rate": 9.999992140129225e-06,
      "loss": 1.0596,
      "step": 1061
    },
    {
      "epoch": 0.030599896271538064,
      "grad_norm": 5.174740791320801,
      "learning_rate": 9.999991291002156e-06,
      "loss": 1.1407,
      "step": 1062
    },
    {
      "epoch": 0.030628709733187345,
      "grad_norm": 4.299599647521973,
      "learning_rate": 9.999990398330164e-06,
      "loss": 1.0724,
      "step": 1063
    },
    {
      "epoch": 0.03065752319483663,
      "grad_norm": 3.1480581760406494,
      "learning_rate": 9.999989462113252e-06,
      "loss": 1.0941,
      "step": 1064
    },
    {
      "epoch": 0.03068633665648591,
      "grad_norm": 3.690117835998535,
      "learning_rate": 9.999988482351431e-06,
      "loss": 1.1165,
      "step": 1065
    },
    {
      "epoch": 0.030715150118135193,
      "grad_norm": 3.1593384742736816,
      "learning_rate": 9.999987459044707e-06,
      "loss": 1.0788,
      "step": 1066
    },
    {
      "epoch": 0.030743963579784477,
      "grad_norm": 3.8305749893188477,
      "learning_rate": 9.999986392193091e-06,
      "loss": 1.1072,
      "step": 1067
    },
    {
      "epoch": 0.030772777041433757,
      "grad_norm": 3.512932538986206,
      "learning_rate": 9.999985281796592e-06,
      "loss": 1.0965,
      "step": 1068
    },
    {
      "epoch": 0.03080159050308304,
      "grad_norm": 3.74165678024292,
      "learning_rate": 9.999984127855221e-06,
      "loss": 1.1049,
      "step": 1069
    },
    {
      "epoch": 0.030830403964732325,
      "grad_norm": 3.818477153778076,
      "learning_rate": 9.999982930368983e-06,
      "loss": 1.1152,
      "step": 1070
    },
    {
      "epoch": 0.030859217426381605,
      "grad_norm": 3.5428452491760254,
      "learning_rate": 9.999981689337895e-06,
      "loss": 1.0943,
      "step": 1071
    },
    {
      "epoch": 0.03088803088803089,
      "grad_norm": 3.382140874862671,
      "learning_rate": 9.999980404761963e-06,
      "loss": 1.1085,
      "step": 1072
    },
    {
      "epoch": 0.03091684434968017,
      "grad_norm": 3.8648133277893066,
      "learning_rate": 9.9999790766412e-06,
      "loss": 1.1078,
      "step": 1073
    },
    {
      "epoch": 0.030945657811329453,
      "grad_norm": 3.3852686882019043,
      "learning_rate": 9.999977704975616e-06,
      "loss": 1.0734,
      "step": 1074
    },
    {
      "epoch": 0.030974471272978737,
      "grad_norm": 3.604769229888916,
      "learning_rate": 9.999976289765226e-06,
      "loss": 1.0921,
      "step": 1075
    },
    {
      "epoch": 0.031003284734628017,
      "grad_norm": 3.553745746612549,
      "learning_rate": 9.999974831010041e-06,
      "loss": 1.1096,
      "step": 1076
    },
    {
      "epoch": 0.0310320981962773,
      "grad_norm": 3.2927794456481934,
      "learning_rate": 9.999973328710073e-06,
      "loss": 1.1108,
      "step": 1077
    },
    {
      "epoch": 0.031060911657926585,
      "grad_norm": 3.922945499420166,
      "learning_rate": 9.999971782865336e-06,
      "loss": 1.1191,
      "step": 1078
    },
    {
      "epoch": 0.031089725119575865,
      "grad_norm": 3.406325101852417,
      "learning_rate": 9.99997019347584e-06,
      "loss": 1.1184,
      "step": 1079
    },
    {
      "epoch": 0.03111853858122515,
      "grad_norm": 3.378469944000244,
      "learning_rate": 9.999968560541605e-06,
      "loss": 1.116,
      "step": 1080
    },
    {
      "epoch": 0.03114735204287443,
      "grad_norm": 3.1351099014282227,
      "learning_rate": 9.999966884062642e-06,
      "loss": 1.1243,
      "step": 1081
    },
    {
      "epoch": 0.031176165504523713,
      "grad_norm": 3.37591814994812,
      "learning_rate": 9.999965164038962e-06,
      "loss": 1.0811,
      "step": 1082
    },
    {
      "epoch": 0.031204978966172997,
      "grad_norm": 3.057701587677002,
      "learning_rate": 9.999963400470587e-06,
      "loss": 1.1141,
      "step": 1083
    },
    {
      "epoch": 0.031233792427822277,
      "grad_norm": 3.2208917140960693,
      "learning_rate": 9.999961593357527e-06,
      "loss": 1.1076,
      "step": 1084
    },
    {
      "epoch": 0.03126260588947156,
      "grad_norm": 3.2848517894744873,
      "learning_rate": 9.999959742699802e-06,
      "loss": 1.1162,
      "step": 1085
    },
    {
      "epoch": 0.03129141935112084,
      "grad_norm": 3.601616859436035,
      "learning_rate": 9.999957848497423e-06,
      "loss": 1.1004,
      "step": 1086
    },
    {
      "epoch": 0.031320232812770125,
      "grad_norm": 3.3908092975616455,
      "learning_rate": 9.999955910750412e-06,
      "loss": 1.1197,
      "step": 1087
    },
    {
      "epoch": 0.03134904627441941,
      "grad_norm": 3.9170243740081787,
      "learning_rate": 9.999953929458781e-06,
      "loss": 1.0938,
      "step": 1088
    },
    {
      "epoch": 0.03137785973606869,
      "grad_norm": 3.162487030029297,
      "learning_rate": 9.999951904622551e-06,
      "loss": 1.0909,
      "step": 1089
    },
    {
      "epoch": 0.03140667319771798,
      "grad_norm": 3.738954544067383,
      "learning_rate": 9.999949836241737e-06,
      "loss": 1.1018,
      "step": 1090
    },
    {
      "epoch": 0.031435486659367254,
      "grad_norm": 3.3838658332824707,
      "learning_rate": 9.99994772431636e-06,
      "loss": 1.0828,
      "step": 1091
    },
    {
      "epoch": 0.03146430012101654,
      "grad_norm": 3.609133720397949,
      "learning_rate": 9.999945568846433e-06,
      "loss": 1.0928,
      "step": 1092
    },
    {
      "epoch": 0.03149311358266582,
      "grad_norm": 3.386174201965332,
      "learning_rate": 9.99994336983198e-06,
      "loss": 1.1076,
      "step": 1093
    },
    {
      "epoch": 0.031521927044315105,
      "grad_norm": 3.5067949295043945,
      "learning_rate": 9.999941127273019e-06,
      "loss": 1.0866,
      "step": 1094
    },
    {
      "epoch": 0.03155074050596439,
      "grad_norm": 3.6049511432647705,
      "learning_rate": 9.999938841169568e-06,
      "loss": 1.1037,
      "step": 1095
    },
    {
      "epoch": 0.031579553967613666,
      "grad_norm": 3.6674723625183105,
      "learning_rate": 9.999936511521648e-06,
      "loss": 1.0999,
      "step": 1096
    },
    {
      "epoch": 0.03160836742926295,
      "grad_norm": 3.3214476108551025,
      "learning_rate": 9.99993413832928e-06,
      "loss": 1.1056,
      "step": 1097
    },
    {
      "epoch": 0.031637180890912234,
      "grad_norm": 3.5349247455596924,
      "learning_rate": 9.999931721592483e-06,
      "loss": 1.063,
      "step": 1098
    },
    {
      "epoch": 0.03166599435256152,
      "grad_norm": 3.2242724895477295,
      "learning_rate": 9.999929261311277e-06,
      "loss": 1.0721,
      "step": 1099
    },
    {
      "epoch": 0.0316948078142108,
      "grad_norm": 3.1186423301696777,
      "learning_rate": 9.999926757485688e-06,
      "loss": 1.1025,
      "step": 1100
    },
    {
      "epoch": 0.031723621275860085,
      "grad_norm": 3.3309412002563477,
      "learning_rate": 9.999924210115733e-06,
      "loss": 1.0813,
      "step": 1101
    },
    {
      "epoch": 0.03175243473750936,
      "grad_norm": 3.101499080657959,
      "learning_rate": 9.999921619201437e-06,
      "loss": 1.1032,
      "step": 1102
    },
    {
      "epoch": 0.031781248199158646,
      "grad_norm": 3.4202232360839844,
      "learning_rate": 9.999918984742822e-06,
      "loss": 1.0971,
      "step": 1103
    },
    {
      "epoch": 0.03181006166080793,
      "grad_norm": 3.2079968452453613,
      "learning_rate": 9.99991630673991e-06,
      "loss": 1.0972,
      "step": 1104
    },
    {
      "epoch": 0.031838875122457214,
      "grad_norm": 3.1427431106567383,
      "learning_rate": 9.999913585192726e-06,
      "loss": 1.1069,
      "step": 1105
    },
    {
      "epoch": 0.0318676885841065,
      "grad_norm": 2.967474937438965,
      "learning_rate": 9.999910820101292e-06,
      "loss": 1.1,
      "step": 1106
    },
    {
      "epoch": 0.031896502045755774,
      "grad_norm": 3.7714955806732178,
      "learning_rate": 9.999908011465634e-06,
      "loss": 1.0603,
      "step": 1107
    },
    {
      "epoch": 0.03192531550740506,
      "grad_norm": 3.2652549743652344,
      "learning_rate": 9.999905159285775e-06,
      "loss": 1.1035,
      "step": 1108
    },
    {
      "epoch": 0.03195412896905434,
      "grad_norm": 3.341132640838623,
      "learning_rate": 9.999902263561738e-06,
      "loss": 1.0795,
      "step": 1109
    },
    {
      "epoch": 0.031982942430703626,
      "grad_norm": 3.105576276779175,
      "learning_rate": 9.999899324293553e-06,
      "loss": 1.1361,
      "step": 1110
    },
    {
      "epoch": 0.03201175589235291,
      "grad_norm": 2.9023425579071045,
      "learning_rate": 9.999896341481243e-06,
      "loss": 1.075,
      "step": 1111
    },
    {
      "epoch": 0.03204056935400219,
      "grad_norm": 3.309683084487915,
      "learning_rate": 9.999893315124832e-06,
      "loss": 1.1079,
      "step": 1112
    },
    {
      "epoch": 0.03206938281565147,
      "grad_norm": 2.940322160720825,
      "learning_rate": 9.99989024522435e-06,
      "loss": 1.0927,
      "step": 1113
    },
    {
      "epoch": 0.032098196277300754,
      "grad_norm": 2.9042816162109375,
      "learning_rate": 9.99988713177982e-06,
      "loss": 1.0668,
      "step": 1114
    },
    {
      "epoch": 0.03212700973895004,
      "grad_norm": 3.2485640048980713,
      "learning_rate": 9.999883974791273e-06,
      "loss": 1.0682,
      "step": 1115
    },
    {
      "epoch": 0.03215582320059932,
      "grad_norm": 3.4822051525115967,
      "learning_rate": 9.999880774258735e-06,
      "loss": 1.0959,
      "step": 1116
    },
    {
      "epoch": 0.032184636662248606,
      "grad_norm": 3.0599663257598877,
      "learning_rate": 9.999877530182231e-06,
      "loss": 1.1023,
      "step": 1117
    },
    {
      "epoch": 0.03221345012389788,
      "grad_norm": 3.0917978286743164,
      "learning_rate": 9.999874242561794e-06,
      "loss": 1.0934,
      "step": 1118
    },
    {
      "epoch": 0.032242263585547166,
      "grad_norm": 2.787963390350342,
      "learning_rate": 9.99987091139745e-06,
      "loss": 1.0947,
      "step": 1119
    },
    {
      "epoch": 0.03227107704719645,
      "grad_norm": 3.5615553855895996,
      "learning_rate": 9.999867536689227e-06,
      "loss": 1.0948,
      "step": 1120
    },
    {
      "epoch": 0.032299890508845734,
      "grad_norm": 2.918670177459717,
      "learning_rate": 9.999864118437156e-06,
      "loss": 1.1154,
      "step": 1121
    },
    {
      "epoch": 0.03232870397049502,
      "grad_norm": 3.1501007080078125,
      "learning_rate": 9.999860656641268e-06,
      "loss": 1.0933,
      "step": 1122
    },
    {
      "epoch": 0.032357517432144295,
      "grad_norm": 3.3690199851989746,
      "learning_rate": 9.999857151301594e-06,
      "loss": 1.0737,
      "step": 1123
    },
    {
      "epoch": 0.03238633089379358,
      "grad_norm": 3.3859715461730957,
      "learning_rate": 9.999853602418157e-06,
      "loss": 1.0917,
      "step": 1124
    },
    {
      "epoch": 0.03241514435544286,
      "grad_norm": 3.0294671058654785,
      "learning_rate": 9.999850009990996e-06,
      "loss": 1.0807,
      "step": 1125
    },
    {
      "epoch": 0.032443957817092146,
      "grad_norm": 2.967738151550293,
      "learning_rate": 9.999846374020138e-06,
      "loss": 1.0902,
      "step": 1126
    },
    {
      "epoch": 0.03247277127874143,
      "grad_norm": 2.973522186279297,
      "learning_rate": 9.999842694505619e-06,
      "loss": 1.0705,
      "step": 1127
    },
    {
      "epoch": 0.03250158474039071,
      "grad_norm": 3.2079591751098633,
      "learning_rate": 9.999838971447467e-06,
      "loss": 1.0718,
      "step": 1128
    },
    {
      "epoch": 0.03253039820203999,
      "grad_norm": 2.9772861003875732,
      "learning_rate": 9.999835204845714e-06,
      "loss": 1.0806,
      "step": 1129
    },
    {
      "epoch": 0.032559211663689275,
      "grad_norm": 3.04388427734375,
      "learning_rate": 9.999831394700394e-06,
      "loss": 1.0859,
      "step": 1130
    },
    {
      "epoch": 0.03258802512533856,
      "grad_norm": 3.1287307739257812,
      "learning_rate": 9.999827541011542e-06,
      "loss": 1.0767,
      "step": 1131
    },
    {
      "epoch": 0.03261683858698784,
      "grad_norm": 3.0644009113311768,
      "learning_rate": 9.999823643779189e-06,
      "loss": 1.0922,
      "step": 1132
    },
    {
      "epoch": 0.032645652048637126,
      "grad_norm": 3.18796968460083,
      "learning_rate": 9.999819703003372e-06,
      "loss": 1.0572,
      "step": 1133
    },
    {
      "epoch": 0.0326744655102864,
      "grad_norm": 3.4668374061584473,
      "learning_rate": 9.999815718684123e-06,
      "loss": 1.0993,
      "step": 1134
    },
    {
      "epoch": 0.03270327897193569,
      "grad_norm": 2.974036455154419,
      "learning_rate": 9.999811690821476e-06,
      "loss": 1.0894,
      "step": 1135
    },
    {
      "epoch": 0.03273209243358497,
      "grad_norm": 3.391170024871826,
      "learning_rate": 9.999807619415466e-06,
      "loss": 1.0842,
      "step": 1136
    },
    {
      "epoch": 0.032760905895234255,
      "grad_norm": 2.8783953189849854,
      "learning_rate": 9.999803504466132e-06,
      "loss": 1.1082,
      "step": 1137
    },
    {
      "epoch": 0.03278971935688354,
      "grad_norm": 3.205737352371216,
      "learning_rate": 9.999799345973507e-06,
      "loss": 1.0568,
      "step": 1138
    },
    {
      "epoch": 0.032818532818532815,
      "grad_norm": 3.1393003463745117,
      "learning_rate": 9.999795143937627e-06,
      "loss": 1.0712,
      "step": 1139
    },
    {
      "epoch": 0.0328473462801821,
      "grad_norm": 3.1966347694396973,
      "learning_rate": 9.99979089835853e-06,
      "loss": 1.104,
      "step": 1140
    },
    {
      "epoch": 0.03287615974183138,
      "grad_norm": 3.225139617919922,
      "learning_rate": 9.999786609236252e-06,
      "loss": 1.0844,
      "step": 1141
    },
    {
      "epoch": 0.03290497320348067,
      "grad_norm": 2.9870083332061768,
      "learning_rate": 9.99978227657083e-06,
      "loss": 1.0871,
      "step": 1142
    },
    {
      "epoch": 0.03293378666512995,
      "grad_norm": 3.425553321838379,
      "learning_rate": 9.999777900362302e-06,
      "loss": 1.0653,
      "step": 1143
    },
    {
      "epoch": 0.032962600126779235,
      "grad_norm": 2.993133544921875,
      "learning_rate": 9.999773480610709e-06,
      "loss": 1.0767,
      "step": 1144
    },
    {
      "epoch": 0.03299141358842851,
      "grad_norm": 3.3990318775177,
      "learning_rate": 9.999769017316083e-06,
      "loss": 1.0861,
      "step": 1145
    },
    {
      "epoch": 0.033020227050077795,
      "grad_norm": 2.9408023357391357,
      "learning_rate": 9.999764510478468e-06,
      "loss": 1.0903,
      "step": 1146
    },
    {
      "epoch": 0.03304904051172708,
      "grad_norm": 3.016718864440918,
      "learning_rate": 9.999759960097905e-06,
      "loss": 1.0715,
      "step": 1147
    },
    {
      "epoch": 0.03307785397337636,
      "grad_norm": 3.2805421352386475,
      "learning_rate": 9.999755366174429e-06,
      "loss": 1.0723,
      "step": 1148
    },
    {
      "epoch": 0.03310666743502565,
      "grad_norm": 3.3805222511291504,
      "learning_rate": 9.99975072870808e-06,
      "loss": 1.0927,
      "step": 1149
    },
    {
      "epoch": 0.033135480896674924,
      "grad_norm": 3.0121042728424072,
      "learning_rate": 9.999746047698901e-06,
      "loss": 1.0611,
      "step": 1150
    },
    {
      "epoch": 0.03316429435832421,
      "grad_norm": 2.8882040977478027,
      "learning_rate": 9.999741323146933e-06,
      "loss": 1.0732,
      "step": 1151
    },
    {
      "epoch": 0.03319310781997349,
      "grad_norm": 3.611722946166992,
      "learning_rate": 9.999736555052215e-06,
      "loss": 1.0856,
      "step": 1152
    },
    {
      "epoch": 0.033221921281622775,
      "grad_norm": 3.171499490737915,
      "learning_rate": 9.999731743414788e-06,
      "loss": 1.0562,
      "step": 1153
    },
    {
      "epoch": 0.03325073474327206,
      "grad_norm": 3.3113601207733154,
      "learning_rate": 9.9997268882347e-06,
      "loss": 1.0905,
      "step": 1154
    },
    {
      "epoch": 0.033279548204921336,
      "grad_norm": 3.209303140640259,
      "learning_rate": 9.999721989511984e-06,
      "loss": 1.0872,
      "step": 1155
    },
    {
      "epoch": 0.03330836166657062,
      "grad_norm": 3.718320846557617,
      "learning_rate": 9.99971704724669e-06,
      "loss": 1.1018,
      "step": 1156
    },
    {
      "epoch": 0.033337175128219904,
      "grad_norm": 5.118166923522949,
      "learning_rate": 9.999712061438857e-06,
      "loss": 1.0906,
      "step": 1157
    },
    {
      "epoch": 0.03336598858986919,
      "grad_norm": 3.6990203857421875,
      "learning_rate": 9.99970703208853e-06,
      "loss": 1.0435,
      "step": 1158
    },
    {
      "epoch": 0.03339480205151847,
      "grad_norm": 3.321021318435669,
      "learning_rate": 9.999701959195754e-06,
      "loss": 1.0825,
      "step": 1159
    },
    {
      "epoch": 0.033423615513167755,
      "grad_norm": 4.259055137634277,
      "learning_rate": 9.999696842760572e-06,
      "loss": 1.097,
      "step": 1160
    },
    {
      "epoch": 0.03345242897481703,
      "grad_norm": 3.8461999893188477,
      "learning_rate": 9.999691682783025e-06,
      "loss": 1.0895,
      "step": 1161
    },
    {
      "epoch": 0.033481242436466316,
      "grad_norm": 3.52044677734375,
      "learning_rate": 9.999686479263165e-06,
      "loss": 1.0736,
      "step": 1162
    },
    {
      "epoch": 0.0335100558981156,
      "grad_norm": 3.253544807434082,
      "learning_rate": 9.999681232201031e-06,
      "loss": 1.0718,
      "step": 1163
    },
    {
      "epoch": 0.033538869359764883,
      "grad_norm": 3.888157606124878,
      "learning_rate": 9.999675941596673e-06,
      "loss": 1.0976,
      "step": 1164
    },
    {
      "epoch": 0.03356768282141417,
      "grad_norm": 3.6802520751953125,
      "learning_rate": 9.999670607450134e-06,
      "loss": 1.0838,
      "step": 1165
    },
    {
      "epoch": 0.033596496283063444,
      "grad_norm": 3.185431957244873,
      "learning_rate": 9.999665229761464e-06,
      "loss": 1.0725,
      "step": 1166
    },
    {
      "epoch": 0.03362530974471273,
      "grad_norm": 3.253504753112793,
      "learning_rate": 9.999659808530707e-06,
      "loss": 1.0787,
      "step": 1167
    },
    {
      "epoch": 0.03365412320636201,
      "grad_norm": 3.2560970783233643,
      "learning_rate": 9.99965434375791e-06,
      "loss": 1.0614,
      "step": 1168
    },
    {
      "epoch": 0.033682936668011296,
      "grad_norm": 3.3567593097686768,
      "learning_rate": 9.99964883544312e-06,
      "loss": 1.0716,
      "step": 1169
    },
    {
      "epoch": 0.03371175012966058,
      "grad_norm": 3.5176124572753906,
      "learning_rate": 9.99964328358639e-06,
      "loss": 1.0862,
      "step": 1170
    },
    {
      "epoch": 0.03374056359130986,
      "grad_norm": 3.5820908546447754,
      "learning_rate": 9.999637688187762e-06,
      "loss": 1.1175,
      "step": 1171
    },
    {
      "epoch": 0.03376937705295914,
      "grad_norm": 3.5153181552886963,
      "learning_rate": 9.999632049247288e-06,
      "loss": 1.1079,
      "step": 1172
    },
    {
      "epoch": 0.033798190514608424,
      "grad_norm": 3.2722299098968506,
      "learning_rate": 9.999626366765017e-06,
      "loss": 1.0881,
      "step": 1173
    },
    {
      "epoch": 0.03382700397625771,
      "grad_norm": 3.475400686264038,
      "learning_rate": 9.999620640740998e-06,
      "loss": 1.0693,
      "step": 1174
    },
    {
      "epoch": 0.03385581743790699,
      "grad_norm": 2.931745767593384,
      "learning_rate": 9.999614871175284e-06,
      "loss": 1.0867,
      "step": 1175
    },
    {
      "epoch": 0.033884630899556276,
      "grad_norm": 4.097916126251221,
      "learning_rate": 9.999609058067918e-06,
      "loss": 1.0852,
      "step": 1176
    },
    {
      "epoch": 0.03391344436120555,
      "grad_norm": 2.8016293048858643,
      "learning_rate": 9.999603201418958e-06,
      "loss": 1.1253,
      "step": 1177
    },
    {
      "epoch": 0.033942257822854836,
      "grad_norm": 3.823789358139038,
      "learning_rate": 9.999597301228448e-06,
      "loss": 1.0898,
      "step": 1178
    },
    {
      "epoch": 0.03397107128450412,
      "grad_norm": 3.0435712337493896,
      "learning_rate": 9.999591357496446e-06,
      "loss": 1.0942,
      "step": 1179
    },
    {
      "epoch": 0.033999884746153404,
      "grad_norm": 3.6375670433044434,
      "learning_rate": 9.999585370223001e-06,
      "loss": 1.0796,
      "step": 1180
    },
    {
      "epoch": 0.03402869820780269,
      "grad_norm": 3.3659496307373047,
      "learning_rate": 9.999579339408164e-06,
      "loss": 1.0811,
      "step": 1181
    },
    {
      "epoch": 0.034057511669451965,
      "grad_norm": 3.259652614593506,
      "learning_rate": 9.99957326505199e-06,
      "loss": 1.0604,
      "step": 1182
    },
    {
      "epoch": 0.03408632513110125,
      "grad_norm": 3.296297550201416,
      "learning_rate": 9.99956714715453e-06,
      "loss": 1.1017,
      "step": 1183
    },
    {
      "epoch": 0.03411513859275053,
      "grad_norm": 3.0561773777008057,
      "learning_rate": 9.99956098571584e-06,
      "loss": 1.0706,
      "step": 1184
    },
    {
      "epoch": 0.034143952054399816,
      "grad_norm": 3.5797247886657715,
      "learning_rate": 9.999554780735969e-06,
      "loss": 1.0892,
      "step": 1185
    },
    {
      "epoch": 0.0341727655160491,
      "grad_norm": 2.896066904067993,
      "learning_rate": 9.999548532214973e-06,
      "loss": 1.074,
      "step": 1186
    },
    {
      "epoch": 0.034201578977698384,
      "grad_norm": 2.990103244781494,
      "learning_rate": 9.999542240152909e-06,
      "loss": 1.0703,
      "step": 1187
    },
    {
      "epoch": 0.03423039243934766,
      "grad_norm": 3.134758472442627,
      "learning_rate": 9.999535904549829e-06,
      "loss": 1.0602,
      "step": 1188
    },
    {
      "epoch": 0.034259205900996945,
      "grad_norm": 2.7831313610076904,
      "learning_rate": 9.99952952540579e-06,
      "loss": 1.0819,
      "step": 1189
    },
    {
      "epoch": 0.03428801936264623,
      "grad_norm": 3.393254280090332,
      "learning_rate": 9.999523102720845e-06,
      "loss": 1.0868,
      "step": 1190
    },
    {
      "epoch": 0.03431683282429551,
      "grad_norm": 3.2696313858032227,
      "learning_rate": 9.999516636495053e-06,
      "loss": 1.0764,
      "step": 1191
    },
    {
      "epoch": 0.034345646285944796,
      "grad_norm": 3.196992874145508,
      "learning_rate": 9.999510126728467e-06,
      "loss": 1.1016,
      "step": 1192
    },
    {
      "epoch": 0.03437445974759407,
      "grad_norm": 2.7639856338500977,
      "learning_rate": 9.999503573421145e-06,
      "loss": 1.058,
      "step": 1193
    },
    {
      "epoch": 0.03440327320924336,
      "grad_norm": 3.462584972381592,
      "learning_rate": 9.999496976573147e-06,
      "loss": 1.095,
      "step": 1194
    },
    {
      "epoch": 0.03443208667089264,
      "grad_norm": 2.9062533378601074,
      "learning_rate": 9.999490336184526e-06,
      "loss": 1.0544,
      "step": 1195
    },
    {
      "epoch": 0.034460900132541925,
      "grad_norm": 3.27156662940979,
      "learning_rate": 9.999483652255342e-06,
      "loss": 1.108,
      "step": 1196
    },
    {
      "epoch": 0.03448971359419121,
      "grad_norm": 2.8275716304779053,
      "learning_rate": 9.999476924785655e-06,
      "loss": 1.0892,
      "step": 1197
    },
    {
      "epoch": 0.034518527055840485,
      "grad_norm": 3.451476812362671,
      "learning_rate": 9.99947015377552e-06,
      "loss": 1.0702,
      "step": 1198
    },
    {
      "epoch": 0.03454734051748977,
      "grad_norm": 3.0900325775146484,
      "learning_rate": 9.999463339224998e-06,
      "loss": 1.0634,
      "step": 1199
    },
    {
      "epoch": 0.03457615397913905,
      "grad_norm": 3.2989377975463867,
      "learning_rate": 9.999456481134148e-06,
      "loss": 1.1023,
      "step": 1200
    },
    {
      "epoch": 0.03460496744078834,
      "grad_norm": 3.011378526687622,
      "learning_rate": 9.999449579503029e-06,
      "loss": 1.106,
      "step": 1201
    },
    {
      "epoch": 0.03463378090243762,
      "grad_norm": 3.167503833770752,
      "learning_rate": 9.999442634331703e-06,
      "loss": 1.0885,
      "step": 1202
    },
    {
      "epoch": 0.034662594364086904,
      "grad_norm": 3.2121529579162598,
      "learning_rate": 9.999435645620227e-06,
      "loss": 1.0713,
      "step": 1203
    },
    {
      "epoch": 0.03469140782573618,
      "grad_norm": 3.137540340423584,
      "learning_rate": 9.999428613368668e-06,
      "loss": 1.0921,
      "step": 1204
    },
    {
      "epoch": 0.034720221287385465,
      "grad_norm": 3.2553651332855225,
      "learning_rate": 9.99942153757708e-06,
      "loss": 1.0516,
      "step": 1205
    },
    {
      "epoch": 0.03474903474903475,
      "grad_norm": 3.271681785583496,
      "learning_rate": 9.99941441824553e-06,
      "loss": 1.0499,
      "step": 1206
    },
    {
      "epoch": 0.03477784821068403,
      "grad_norm": 3.1321096420288086,
      "learning_rate": 9.999407255374076e-06,
      "loss": 1.0367,
      "step": 1207
    },
    {
      "epoch": 0.03480666167233332,
      "grad_norm": 3.0015997886657715,
      "learning_rate": 9.999400048962785e-06,
      "loss": 1.0712,
      "step": 1208
    },
    {
      "epoch": 0.034835475133982594,
      "grad_norm": 3.1662638187408447,
      "learning_rate": 9.999392799011715e-06,
      "loss": 1.0789,
      "step": 1209
    },
    {
      "epoch": 0.03486428859563188,
      "grad_norm": 3.058663845062256,
      "learning_rate": 9.999385505520932e-06,
      "loss": 1.1135,
      "step": 1210
    },
    {
      "epoch": 0.03489310205728116,
      "grad_norm": 3.2075209617614746,
      "learning_rate": 9.999378168490499e-06,
      "loss": 1.086,
      "step": 1211
    },
    {
      "epoch": 0.034921915518930445,
      "grad_norm": 3.1136600971221924,
      "learning_rate": 9.999370787920479e-06,
      "loss": 1.1092,
      "step": 1212
    },
    {
      "epoch": 0.03495072898057973,
      "grad_norm": 2.8595099449157715,
      "learning_rate": 9.999363363810937e-06,
      "loss": 1.0655,
      "step": 1213
    },
    {
      "epoch": 0.03497954244222901,
      "grad_norm": 3.1016438007354736,
      "learning_rate": 9.999355896161937e-06,
      "loss": 1.0661,
      "step": 1214
    },
    {
      "epoch": 0.03500835590387829,
      "grad_norm": 3.0868594646453857,
      "learning_rate": 9.999348384973546e-06,
      "loss": 1.0564,
      "step": 1215
    },
    {
      "epoch": 0.03503716936552757,
      "grad_norm": 3.076497793197632,
      "learning_rate": 9.999340830245829e-06,
      "loss": 1.0814,
      "step": 1216
    },
    {
      "epoch": 0.03506598282717686,
      "grad_norm": 3.4097483158111572,
      "learning_rate": 9.999333231978849e-06,
      "loss": 1.0457,
      "step": 1217
    },
    {
      "epoch": 0.03509479628882614,
      "grad_norm": 3.0412700176239014,
      "learning_rate": 9.999325590172675e-06,
      "loss": 1.0617,
      "step": 1218
    },
    {
      "epoch": 0.035123609750475425,
      "grad_norm": 3.380889654159546,
      "learning_rate": 9.999317904827373e-06,
      "loss": 1.0804,
      "step": 1219
    },
    {
      "epoch": 0.0351524232121247,
      "grad_norm": 2.969723701477051,
      "learning_rate": 9.999310175943009e-06,
      "loss": 1.0711,
      "step": 1220
    },
    {
      "epoch": 0.035181236673773986,
      "grad_norm": 3.06955623626709,
      "learning_rate": 9.99930240351965e-06,
      "loss": 1.0728,
      "step": 1221
    },
    {
      "epoch": 0.03521005013542327,
      "grad_norm": 3.0499632358551025,
      "learning_rate": 9.999294587557365e-06,
      "loss": 1.0695,
      "step": 1222
    },
    {
      "epoch": 0.03523886359707255,
      "grad_norm": 2.953009843826294,
      "learning_rate": 9.999286728056222e-06,
      "loss": 1.073,
      "step": 1223
    },
    {
      "epoch": 0.03526767705872184,
      "grad_norm": 3.324148654937744,
      "learning_rate": 9.99927882501629e-06,
      "loss": 1.0731,
      "step": 1224
    },
    {
      "epoch": 0.035296490520371114,
      "grad_norm": 3.0663886070251465,
      "learning_rate": 9.999270878437635e-06,
      "loss": 1.0903,
      "step": 1225
    },
    {
      "epoch": 0.0353253039820204,
      "grad_norm": 3.1732959747314453,
      "learning_rate": 9.999262888320329e-06,
      "loss": 1.0821,
      "step": 1226
    },
    {
      "epoch": 0.03535411744366968,
      "grad_norm": 2.9015190601348877,
      "learning_rate": 9.999254854664441e-06,
      "loss": 1.0861,
      "step": 1227
    },
    {
      "epoch": 0.035382930905318966,
      "grad_norm": 3.144285202026367,
      "learning_rate": 9.99924677747004e-06,
      "loss": 1.0733,
      "step": 1228
    },
    {
      "epoch": 0.03541174436696825,
      "grad_norm": 3.119821071624756,
      "learning_rate": 9.999238656737197e-06,
      "loss": 1.0931,
      "step": 1229
    },
    {
      "epoch": 0.03544055782861753,
      "grad_norm": 3.4297308921813965,
      "learning_rate": 9.999230492465983e-06,
      "loss": 1.0957,
      "step": 1230
    },
    {
      "epoch": 0.03546937129026681,
      "grad_norm": 2.9945285320281982,
      "learning_rate": 9.99922228465647e-06,
      "loss": 1.0919,
      "step": 1231
    },
    {
      "epoch": 0.035498184751916094,
      "grad_norm": 2.974607467651367,
      "learning_rate": 9.999214033308726e-06,
      "loss": 1.1176,
      "step": 1232
    },
    {
      "epoch": 0.03552699821356538,
      "grad_norm": 4.421630859375,
      "learning_rate": 9.999205738422826e-06,
      "loss": 1.0589,
      "step": 1233
    },
    {
      "epoch": 0.03555581167521466,
      "grad_norm": 3.364466905593872,
      "learning_rate": 9.999197399998842e-06,
      "loss": 1.0953,
      "step": 1234
    },
    {
      "epoch": 0.035584625136863945,
      "grad_norm": 2.9893534183502197,
      "learning_rate": 9.999189018036846e-06,
      "loss": 1.0667,
      "step": 1235
    },
    {
      "epoch": 0.03561343859851322,
      "grad_norm": 3.084005355834961,
      "learning_rate": 9.99918059253691e-06,
      "loss": 1.0817,
      "step": 1236
    },
    {
      "epoch": 0.035642252060162506,
      "grad_norm": 3.454380512237549,
      "learning_rate": 9.999172123499109e-06,
      "loss": 1.0625,
      "step": 1237
    },
    {
      "epoch": 0.03567106552181179,
      "grad_norm": 2.74481201171875,
      "learning_rate": 9.999163610923516e-06,
      "loss": 1.0764,
      "step": 1238
    },
    {
      "epoch": 0.035699878983461074,
      "grad_norm": 3.135756254196167,
      "learning_rate": 9.999155054810205e-06,
      "loss": 1.0694,
      "step": 1239
    },
    {
      "epoch": 0.03572869244511036,
      "grad_norm": 2.8903961181640625,
      "learning_rate": 9.999146455159253e-06,
      "loss": 1.0798,
      "step": 1240
    },
    {
      "epoch": 0.03575750590675964,
      "grad_norm": 2.950742483139038,
      "learning_rate": 9.99913781197073e-06,
      "loss": 1.0946,
      "step": 1241
    },
    {
      "epoch": 0.03578631936840892,
      "grad_norm": 2.9497838020324707,
      "learning_rate": 9.999129125244714e-06,
      "loss": 1.0806,
      "step": 1242
    },
    {
      "epoch": 0.0358151328300582,
      "grad_norm": 2.826021194458008,
      "learning_rate": 9.99912039498128e-06,
      "loss": 1.0946,
      "step": 1243
    },
    {
      "epoch": 0.035843946291707486,
      "grad_norm": 3.1755144596099854,
      "learning_rate": 9.999111621180508e-06,
      "loss": 1.0632,
      "step": 1244
    },
    {
      "epoch": 0.03587275975335677,
      "grad_norm": 2.758969783782959,
      "learning_rate": 9.999102803842469e-06,
      "loss": 1.0452,
      "step": 1245
    },
    {
      "epoch": 0.035901573215006054,
      "grad_norm": 3.226987600326538,
      "learning_rate": 9.999093942967241e-06,
      "loss": 1.0944,
      "step": 1246
    },
    {
      "epoch": 0.03593038667665533,
      "grad_norm": 3.3541676998138428,
      "learning_rate": 9.999085038554905e-06,
      "loss": 1.0669,
      "step": 1247
    },
    {
      "epoch": 0.035959200138304614,
      "grad_norm": 3.0785746574401855,
      "learning_rate": 9.999076090605534e-06,
      "loss": 1.0753,
      "step": 1248
    },
    {
      "epoch": 0.0359880135999539,
      "grad_norm": 3.200425148010254,
      "learning_rate": 9.999067099119206e-06,
      "loss": 1.0688,
      "step": 1249
    },
    {
      "epoch": 0.03601682706160318,
      "grad_norm": 2.8379459381103516,
      "learning_rate": 9.999058064096002e-06,
      "loss": 1.0775,
      "step": 1250
    },
    {
      "epoch": 0.036045640523252466,
      "grad_norm": 2.927347421646118,
      "learning_rate": 9.999048985536e-06,
      "loss": 1.0483,
      "step": 1251
    },
    {
      "epoch": 0.03607445398490174,
      "grad_norm": 2.9771554470062256,
      "learning_rate": 9.999039863439278e-06,
      "loss": 1.0782,
      "step": 1252
    },
    {
      "epoch": 0.03610326744655103,
      "grad_norm": 2.9547119140625,
      "learning_rate": 9.999030697805916e-06,
      "loss": 1.0611,
      "step": 1253
    },
    {
      "epoch": 0.03613208090820031,
      "grad_norm": 3.302926778793335,
      "learning_rate": 9.999021488635993e-06,
      "loss": 1.0552,
      "step": 1254
    },
    {
      "epoch": 0.036160894369849594,
      "grad_norm": 3.284910202026367,
      "learning_rate": 9.99901223592959e-06,
      "loss": 1.0644,
      "step": 1255
    },
    {
      "epoch": 0.03618970783149888,
      "grad_norm": 2.9025204181671143,
      "learning_rate": 9.99900293968679e-06,
      "loss": 1.0607,
      "step": 1256
    },
    {
      "epoch": 0.03621852129314816,
      "grad_norm": 3.180159091949463,
      "learning_rate": 9.998993599907669e-06,
      "loss": 1.0698,
      "step": 1257
    },
    {
      "epoch": 0.03624733475479744,
      "grad_norm": 2.9033761024475098,
      "learning_rate": 9.998984216592313e-06,
      "loss": 1.0745,
      "step": 1258
    },
    {
      "epoch": 0.03627614821644672,
      "grad_norm": 2.88061785697937,
      "learning_rate": 9.9989747897408e-06,
      "loss": 1.0637,
      "step": 1259
    },
    {
      "epoch": 0.03630496167809601,
      "grad_norm": 3.343932628631592,
      "learning_rate": 9.998965319353215e-06,
      "loss": 1.1002,
      "step": 1260
    },
    {
      "epoch": 0.03633377513974529,
      "grad_norm": 2.992795467376709,
      "learning_rate": 9.998955805429637e-06,
      "loss": 1.0551,
      "step": 1261
    },
    {
      "epoch": 0.036362588601394574,
      "grad_norm": 3.110762357711792,
      "learning_rate": 9.998946247970153e-06,
      "loss": 1.0745,
      "step": 1262
    },
    {
      "epoch": 0.03639140206304385,
      "grad_norm": 3.2491986751556396,
      "learning_rate": 9.998936646974845e-06,
      "loss": 1.0723,
      "step": 1263
    },
    {
      "epoch": 0.036420215524693135,
      "grad_norm": 2.7656264305114746,
      "learning_rate": 9.998927002443794e-06,
      "loss": 1.1067,
      "step": 1264
    },
    {
      "epoch": 0.03644902898634242,
      "grad_norm": 3.509636402130127,
      "learning_rate": 9.998917314377086e-06,
      "loss": 1.0717,
      "step": 1265
    },
    {
      "epoch": 0.0364778424479917,
      "grad_norm": 2.8539347648620605,
      "learning_rate": 9.998907582774808e-06,
      "loss": 1.0415,
      "step": 1266
    },
    {
      "epoch": 0.036506655909640987,
      "grad_norm": 2.8466644287109375,
      "learning_rate": 9.998897807637039e-06,
      "loss": 1.0986,
      "step": 1267
    },
    {
      "epoch": 0.03653546937129026,
      "grad_norm": 3.075815439224243,
      "learning_rate": 9.998887988963868e-06,
      "loss": 1.0791,
      "step": 1268
    },
    {
      "epoch": 0.03656428283293955,
      "grad_norm": 3.078028678894043,
      "learning_rate": 9.99887812675538e-06,
      "loss": 1.0519,
      "step": 1269
    },
    {
      "epoch": 0.03659309629458883,
      "grad_norm": 3.143941879272461,
      "learning_rate": 9.998868221011661e-06,
      "loss": 1.0625,
      "step": 1270
    },
    {
      "epoch": 0.036621909756238115,
      "grad_norm": 3.156630754470825,
      "learning_rate": 9.998858271732794e-06,
      "loss": 1.0513,
      "step": 1271
    },
    {
      "epoch": 0.0366507232178874,
      "grad_norm": 3.044400453567505,
      "learning_rate": 9.998848278918872e-06,
      "loss": 1.0675,
      "step": 1272
    },
    {
      "epoch": 0.03667953667953668,
      "grad_norm": 2.9445953369140625,
      "learning_rate": 9.998838242569977e-06,
      "loss": 1.083,
      "step": 1273
    },
    {
      "epoch": 0.03670835014118596,
      "grad_norm": 3.008234739303589,
      "learning_rate": 9.998828162686198e-06,
      "loss": 1.0823,
      "step": 1274
    },
    {
      "epoch": 0.03673716360283524,
      "grad_norm": 3.0591166019439697,
      "learning_rate": 9.998818039267623e-06,
      "loss": 1.0682,
      "step": 1275
    },
    {
      "epoch": 0.03676597706448453,
      "grad_norm": 3.009857654571533,
      "learning_rate": 9.99880787231434e-06,
      "loss": 1.0834,
      "step": 1276
    },
    {
      "epoch": 0.03679479052613381,
      "grad_norm": 3.0702717304229736,
      "learning_rate": 9.998797661826439e-06,
      "loss": 1.0373,
      "step": 1277
    },
    {
      "epoch": 0.036823603987783095,
      "grad_norm": 2.7559704780578613,
      "learning_rate": 9.998787407804003e-06,
      "loss": 1.0849,
      "step": 1278
    },
    {
      "epoch": 0.03685241744943237,
      "grad_norm": 3.1655094623565674,
      "learning_rate": 9.998777110247129e-06,
      "loss": 1.0795,
      "step": 1279
    },
    {
      "epoch": 0.036881230911081656,
      "grad_norm": 2.908353090286255,
      "learning_rate": 9.998766769155901e-06,
      "loss": 1.0484,
      "step": 1280
    },
    {
      "epoch": 0.03691004437273094,
      "grad_norm": 3.0926475524902344,
      "learning_rate": 9.998756384530415e-06,
      "loss": 1.0843,
      "step": 1281
    },
    {
      "epoch": 0.03693885783438022,
      "grad_norm": 2.9440574645996094,
      "learning_rate": 9.998745956370755e-06,
      "loss": 1.0732,
      "step": 1282
    },
    {
      "epoch": 0.03696767129602951,
      "grad_norm": 3.058488607406616,
      "learning_rate": 9.998735484677017e-06,
      "loss": 1.0444,
      "step": 1283
    },
    {
      "epoch": 0.03699648475767879,
      "grad_norm": 2.9284775257110596,
      "learning_rate": 9.998724969449287e-06,
      "loss": 1.0748,
      "step": 1284
    },
    {
      "epoch": 0.03702529821932807,
      "grad_norm": 2.8379437923431396,
      "learning_rate": 9.99871441068766e-06,
      "loss": 1.0401,
      "step": 1285
    },
    {
      "epoch": 0.03705411168097735,
      "grad_norm": 3.0618133544921875,
      "learning_rate": 9.998703808392227e-06,
      "loss": 1.0586,
      "step": 1286
    },
    {
      "epoch": 0.037082925142626635,
      "grad_norm": 2.90942120552063,
      "learning_rate": 9.998693162563082e-06,
      "loss": 1.1025,
      "step": 1287
    },
    {
      "epoch": 0.03711173860427592,
      "grad_norm": 3.1378531455993652,
      "learning_rate": 9.998682473200316e-06,
      "loss": 1.0721,
      "step": 1288
    },
    {
      "epoch": 0.0371405520659252,
      "grad_norm": 2.6347317695617676,
      "learning_rate": 9.998671740304023e-06,
      "loss": 1.078,
      "step": 1289
    },
    {
      "epoch": 0.03716936552757448,
      "grad_norm": 3.2113802433013916,
      "learning_rate": 9.998660963874295e-06,
      "loss": 1.0325,
      "step": 1290
    },
    {
      "epoch": 0.037198178989223764,
      "grad_norm": 2.861844301223755,
      "learning_rate": 9.998650143911228e-06,
      "loss": 1.1023,
      "step": 1291
    },
    {
      "epoch": 0.03722699245087305,
      "grad_norm": 3.342911720275879,
      "learning_rate": 9.998639280414914e-06,
      "loss": 1.0805,
      "step": 1292
    },
    {
      "epoch": 0.03725580591252233,
      "grad_norm": 2.811257839202881,
      "learning_rate": 9.998628373385448e-06,
      "loss": 1.0891,
      "step": 1293
    },
    {
      "epoch": 0.037284619374171615,
      "grad_norm": 3.243169069290161,
      "learning_rate": 9.998617422822927e-06,
      "loss": 1.0446,
      "step": 1294
    },
    {
      "epoch": 0.03731343283582089,
      "grad_norm": 3.271787166595459,
      "learning_rate": 9.998606428727445e-06,
      "loss": 1.0871,
      "step": 1295
    },
    {
      "epoch": 0.037342246297470176,
      "grad_norm": 2.9508349895477295,
      "learning_rate": 9.998595391099097e-06,
      "loss": 1.0467,
      "step": 1296
    },
    {
      "epoch": 0.03737105975911946,
      "grad_norm": 3.305802583694458,
      "learning_rate": 9.998584309937982e-06,
      "loss": 1.071,
      "step": 1297
    },
    {
      "epoch": 0.037399873220768744,
      "grad_norm": 2.837724208831787,
      "learning_rate": 9.998573185244192e-06,
      "loss": 1.0445,
      "step": 1298
    },
    {
      "epoch": 0.03742868668241803,
      "grad_norm": 2.9218955039978027,
      "learning_rate": 9.998562017017828e-06,
      "loss": 1.0883,
      "step": 1299
    },
    {
      "epoch": 0.03745750014406731,
      "grad_norm": 2.8607888221740723,
      "learning_rate": 9.998550805258984e-06,
      "loss": 1.05,
      "step": 1300
    },
    {
      "epoch": 0.03748631360571659,
      "grad_norm": 3.1103675365448,
      "learning_rate": 9.99853954996776e-06,
      "loss": 1.0391,
      "step": 1301
    },
    {
      "epoch": 0.03751512706736587,
      "grad_norm": 2.8194241523742676,
      "learning_rate": 9.998528251144253e-06,
      "loss": 1.0779,
      "step": 1302
    },
    {
      "epoch": 0.037543940529015156,
      "grad_norm": 2.9263086318969727,
      "learning_rate": 9.998516908788562e-06,
      "loss": 1.0553,
      "step": 1303
    },
    {
      "epoch": 0.03757275399066444,
      "grad_norm": 3.159972906112671,
      "learning_rate": 9.998505522900785e-06,
      "loss": 1.084,
      "step": 1304
    },
    {
      "epoch": 0.037601567452313724,
      "grad_norm": 3.0700323581695557,
      "learning_rate": 9.998494093481022e-06,
      "loss": 1.0713,
      "step": 1305
    },
    {
      "epoch": 0.037630380913963,
      "grad_norm": 2.880121946334839,
      "learning_rate": 9.998482620529371e-06,
      "loss": 1.0512,
      "step": 1306
    },
    {
      "epoch": 0.037659194375612284,
      "grad_norm": 2.900170087814331,
      "learning_rate": 9.998471104045934e-06,
      "loss": 1.0875,
      "step": 1307
    },
    {
      "epoch": 0.03768800783726157,
      "grad_norm": 2.6726059913635254,
      "learning_rate": 9.99845954403081e-06,
      "loss": 1.0713,
      "step": 1308
    },
    {
      "epoch": 0.03771682129891085,
      "grad_norm": 2.9704909324645996,
      "learning_rate": 9.998447940484101e-06,
      "loss": 1.0616,
      "step": 1309
    },
    {
      "epoch": 0.037745634760560136,
      "grad_norm": 2.8885657787323,
      "learning_rate": 9.998436293405907e-06,
      "loss": 1.0627,
      "step": 1310
    },
    {
      "epoch": 0.03777444822220942,
      "grad_norm": 3.0383408069610596,
      "learning_rate": 9.998424602796328e-06,
      "loss": 1.0558,
      "step": 1311
    },
    {
      "epoch": 0.0378032616838587,
      "grad_norm": 2.793837785720825,
      "learning_rate": 9.99841286865547e-06,
      "loss": 1.0858,
      "step": 1312
    },
    {
      "epoch": 0.03783207514550798,
      "grad_norm": 2.903116226196289,
      "learning_rate": 9.99840109098343e-06,
      "loss": 1.055,
      "step": 1313
    },
    {
      "epoch": 0.037860888607157264,
      "grad_norm": 2.8932077884674072,
      "learning_rate": 9.998389269780313e-06,
      "loss": 1.0644,
      "step": 1314
    },
    {
      "epoch": 0.03788970206880655,
      "grad_norm": 3.0982167720794678,
      "learning_rate": 9.998377405046222e-06,
      "loss": 1.095,
      "step": 1315
    },
    {
      "epoch": 0.03791851553045583,
      "grad_norm": 3.1904969215393066,
      "learning_rate": 9.998365496781261e-06,
      "loss": 1.0672,
      "step": 1316
    },
    {
      "epoch": 0.03794732899210511,
      "grad_norm": 2.8208625316619873,
      "learning_rate": 9.998353544985534e-06,
      "loss": 1.0404,
      "step": 1317
    },
    {
      "epoch": 0.03797614245375439,
      "grad_norm": 2.7801291942596436,
      "learning_rate": 9.998341549659144e-06,
      "loss": 1.0715,
      "step": 1318
    },
    {
      "epoch": 0.038004955915403676,
      "grad_norm": 3.2943403720855713,
      "learning_rate": 9.998329510802194e-06,
      "loss": 1.0497,
      "step": 1319
    },
    {
      "epoch": 0.03803376937705296,
      "grad_norm": 2.8486993312835693,
      "learning_rate": 9.998317428414791e-06,
      "loss": 1.0631,
      "step": 1320
    },
    {
      "epoch": 0.038062582838702244,
      "grad_norm": 3.1535227298736572,
      "learning_rate": 9.99830530249704e-06,
      "loss": 1.0928,
      "step": 1321
    },
    {
      "epoch": 0.03809139630035152,
      "grad_norm": 2.8137319087982178,
      "learning_rate": 9.998293133049047e-06,
      "loss": 1.0673,
      "step": 1322
    },
    {
      "epoch": 0.038120209762000805,
      "grad_norm": 2.9357733726501465,
      "learning_rate": 9.998280920070917e-06,
      "loss": 1.0562,
      "step": 1323
    },
    {
      "epoch": 0.03814902322365009,
      "grad_norm": 3.73565673828125,
      "learning_rate": 9.998268663562756e-06,
      "loss": 1.0517,
      "step": 1324
    },
    {
      "epoch": 0.03817783668529937,
      "grad_norm": 3.0683815479278564,
      "learning_rate": 9.99825636352467e-06,
      "loss": 1.0629,
      "step": 1325
    },
    {
      "epoch": 0.038206650146948656,
      "grad_norm": 3.167086362838745,
      "learning_rate": 9.99824401995677e-06,
      "loss": 1.0663,
      "step": 1326
    },
    {
      "epoch": 0.03823546360859794,
      "grad_norm": 2.8199779987335205,
      "learning_rate": 9.99823163285916e-06,
      "loss": 1.0467,
      "step": 1327
    },
    {
      "epoch": 0.03826427707024722,
      "grad_norm": 3.065359592437744,
      "learning_rate": 9.99821920223195e-06,
      "loss": 1.0577,
      "step": 1328
    },
    {
      "epoch": 0.0382930905318965,
      "grad_norm": 3.513115644454956,
      "learning_rate": 9.998206728075245e-06,
      "loss": 1.0677,
      "step": 1329
    },
    {
      "epoch": 0.038321903993545785,
      "grad_norm": 2.9389865398406982,
      "learning_rate": 9.998194210389158e-06,
      "loss": 1.0543,
      "step": 1330
    },
    {
      "epoch": 0.03835071745519507,
      "grad_norm": 3.2859182357788086,
      "learning_rate": 9.998181649173794e-06,
      "loss": 1.0742,
      "step": 1331
    },
    {
      "epoch": 0.03837953091684435,
      "grad_norm": 2.8178224563598633,
      "learning_rate": 9.998169044429264e-06,
      "loss": 1.0858,
      "step": 1332
    },
    {
      "epoch": 0.03840834437849363,
      "grad_norm": 3.0462098121643066,
      "learning_rate": 9.99815639615568e-06,
      "loss": 1.0685,
      "step": 1333
    },
    {
      "epoch": 0.03843715784014291,
      "grad_norm": 2.8231632709503174,
      "learning_rate": 9.998143704353149e-06,
      "loss": 1.0456,
      "step": 1334
    },
    {
      "epoch": 0.0384659713017922,
      "grad_norm": 3.0978007316589355,
      "learning_rate": 9.998130969021782e-06,
      "loss": 1.0748,
      "step": 1335
    },
    {
      "epoch": 0.03849478476344148,
      "grad_norm": 2.657970666885376,
      "learning_rate": 9.998118190161691e-06,
      "loss": 1.0647,
      "step": 1336
    },
    {
      "epoch": 0.038523598225090765,
      "grad_norm": 8.920374870300293,
      "learning_rate": 9.998105367772988e-06,
      "loss": 1.0804,
      "step": 1337
    },
    {
      "epoch": 0.03855241168674004,
      "grad_norm": 2.898362874984741,
      "learning_rate": 9.998092501855783e-06,
      "loss": 1.1037,
      "step": 1338
    },
    {
      "epoch": 0.038581225148389325,
      "grad_norm": 3.2039730548858643,
      "learning_rate": 9.998079592410187e-06,
      "loss": 1.067,
      "step": 1339
    },
    {
      "epoch": 0.03861003861003861,
      "grad_norm": 2.611529588699341,
      "learning_rate": 9.998066639436314e-06,
      "loss": 1.0729,
      "step": 1340
    },
    {
      "epoch": 0.03863885207168789,
      "grad_norm": 3.064666748046875,
      "learning_rate": 9.99805364293428e-06,
      "loss": 1.0769,
      "step": 1341
    },
    {
      "epoch": 0.03866766553333718,
      "grad_norm": 2.7402334213256836,
      "learning_rate": 9.998040602904192e-06,
      "loss": 1.071,
      "step": 1342
    },
    {
      "epoch": 0.03869647899498646,
      "grad_norm": 2.799985408782959,
      "learning_rate": 9.998027519346166e-06,
      "loss": 1.0529,
      "step": 1343
    },
    {
      "epoch": 0.03872529245663574,
      "grad_norm": 2.9682400226593018,
      "learning_rate": 9.99801439226032e-06,
      "loss": 1.0903,
      "step": 1344
    },
    {
      "epoch": 0.03875410591828502,
      "grad_norm": 2.825181007385254,
      "learning_rate": 9.998001221646763e-06,
      "loss": 1.074,
      "step": 1345
    },
    {
      "epoch": 0.038782919379934305,
      "grad_norm": 2.9687178134918213,
      "learning_rate": 9.99798800750561e-06,
      "loss": 1.0512,
      "step": 1346
    },
    {
      "epoch": 0.03881173284158359,
      "grad_norm": 2.7987606525421143,
      "learning_rate": 9.99797474983698e-06,
      "loss": 1.054,
      "step": 1347
    },
    {
      "epoch": 0.03884054630323287,
      "grad_norm": 3.011683940887451,
      "learning_rate": 9.997961448640985e-06,
      "loss": 1.0719,
      "step": 1348
    },
    {
      "epoch": 0.03886935976488215,
      "grad_norm": 2.8474249839782715,
      "learning_rate": 9.997948103917742e-06,
      "loss": 1.0703,
      "step": 1349
    },
    {
      "epoch": 0.038898173226531434,
      "grad_norm": 2.7745044231414795,
      "learning_rate": 9.997934715667367e-06,
      "loss": 1.0418,
      "step": 1350
    },
    {
      "epoch": 0.03892698668818072,
      "grad_norm": 2.9722445011138916,
      "learning_rate": 9.997921283889977e-06,
      "loss": 1.0645,
      "step": 1351
    },
    {
      "epoch": 0.03895580014983,
      "grad_norm": 2.6417348384857178,
      "learning_rate": 9.997907808585689e-06,
      "loss": 1.0306,
      "step": 1352
    },
    {
      "epoch": 0.038984613611479285,
      "grad_norm": 3.038213014602661,
      "learning_rate": 9.99789428975462e-06,
      "loss": 1.0479,
      "step": 1353
    },
    {
      "epoch": 0.03901342707312857,
      "grad_norm": 2.868222236633301,
      "learning_rate": 9.997880727396887e-06,
      "loss": 1.0396,
      "step": 1354
    },
    {
      "epoch": 0.039042240534777846,
      "grad_norm": 2.963616371154785,
      "learning_rate": 9.997867121512608e-06,
      "loss": 1.0688,
      "step": 1355
    },
    {
      "epoch": 0.03907105399642713,
      "grad_norm": 3.049626350402832,
      "learning_rate": 9.997853472101903e-06,
      "loss": 1.0438,
      "step": 1356
    },
    {
      "epoch": 0.039099867458076414,
      "grad_norm": 2.968829393386841,
      "learning_rate": 9.997839779164893e-06,
      "loss": 1.0468,
      "step": 1357
    },
    {
      "epoch": 0.0391286809197257,
      "grad_norm": 2.86156964302063,
      "learning_rate": 9.997826042701693e-06,
      "loss": 1.0484,
      "step": 1358
    },
    {
      "epoch": 0.03915749438137498,
      "grad_norm": 2.959070920944214,
      "learning_rate": 9.997812262712422e-06,
      "loss": 1.0887,
      "step": 1359
    },
    {
      "epoch": 0.03918630784302426,
      "grad_norm": 2.6756839752197266,
      "learning_rate": 9.997798439197204e-06,
      "loss": 1.0565,
      "step": 1360
    },
    {
      "epoch": 0.03921512130467354,
      "grad_norm": 3.1930227279663086,
      "learning_rate": 9.997784572156157e-06,
      "loss": 1.0765,
      "step": 1361
    },
    {
      "epoch": 0.039243934766322826,
      "grad_norm": 2.7989346981048584,
      "learning_rate": 9.997770661589404e-06,
      "loss": 1.0698,
      "step": 1362
    },
    {
      "epoch": 0.03927274822797211,
      "grad_norm": 3.3100717067718506,
      "learning_rate": 9.997756707497063e-06,
      "loss": 1.0762,
      "step": 1363
    },
    {
      "epoch": 0.03930156168962139,
      "grad_norm": 2.9975483417510986,
      "learning_rate": 9.997742709879257e-06,
      "loss": 1.0523,
      "step": 1364
    },
    {
      "epoch": 0.03933037515127067,
      "grad_norm": 8.425135612487793,
      "learning_rate": 9.997728668736109e-06,
      "loss": 1.0607,
      "step": 1365
    },
    {
      "epoch": 0.039359188612919954,
      "grad_norm": 3.3306334018707275,
      "learning_rate": 9.99771458406774e-06,
      "loss": 1.0808,
      "step": 1366
    },
    {
      "epoch": 0.03938800207456924,
      "grad_norm": 3.523641586303711,
      "learning_rate": 9.997700455874271e-06,
      "loss": 1.0645,
      "step": 1367
    },
    {
      "epoch": 0.03941681553621852,
      "grad_norm": 3.236650228500366,
      "learning_rate": 9.997686284155828e-06,
      "loss": 1.077,
      "step": 1368
    },
    {
      "epoch": 0.039445628997867806,
      "grad_norm": 3.4402999877929688,
      "learning_rate": 9.997672068912535e-06,
      "loss": 1.0559,
      "step": 1369
    },
    {
      "epoch": 0.03947444245951709,
      "grad_norm": 3.4563512802124023,
      "learning_rate": 9.997657810144512e-06,
      "loss": 1.0693,
      "step": 1370
    },
    {
      "epoch": 0.039503255921166366,
      "grad_norm": 3.452043294906616,
      "learning_rate": 9.997643507851885e-06,
      "loss": 1.0811,
      "step": 1371
    },
    {
      "epoch": 0.03953206938281565,
      "grad_norm": 3.462606430053711,
      "learning_rate": 9.99762916203478e-06,
      "loss": 1.0642,
      "step": 1372
    },
    {
      "epoch": 0.039560882844464934,
      "grad_norm": 3.563772678375244,
      "learning_rate": 9.99761477269332e-06,
      "loss": 1.0717,
      "step": 1373
    },
    {
      "epoch": 0.03958969630611422,
      "grad_norm": 3.210331916809082,
      "learning_rate": 9.997600339827632e-06,
      "loss": 1.0441,
      "step": 1374
    },
    {
      "epoch": 0.0396185097677635,
      "grad_norm": 3.578948974609375,
      "learning_rate": 9.99758586343784e-06,
      "loss": 1.0638,
      "step": 1375
    },
    {
      "epoch": 0.03964732322941278,
      "grad_norm": 3.287750720977783,
      "learning_rate": 9.997571343524073e-06,
      "loss": 1.0657,
      "step": 1376
    },
    {
      "epoch": 0.03967613669106206,
      "grad_norm": 3.262556791305542,
      "learning_rate": 9.997556780086453e-06,
      "loss": 1.0409,
      "step": 1377
    },
    {
      "epoch": 0.039704950152711346,
      "grad_norm": 3.1301190853118896,
      "learning_rate": 9.99754217312511e-06,
      "loss": 1.0676,
      "step": 1378
    },
    {
      "epoch": 0.03973376361436063,
      "grad_norm": 3.374526262283325,
      "learning_rate": 9.99752752264017e-06,
      "loss": 1.0259,
      "step": 1379
    },
    {
      "epoch": 0.039762577076009914,
      "grad_norm": 3.0842220783233643,
      "learning_rate": 9.997512828631761e-06,
      "loss": 1.0512,
      "step": 1380
    },
    {
      "epoch": 0.0397913905376592,
      "grad_norm": 3.295990228652954,
      "learning_rate": 9.997498091100011e-06,
      "loss": 1.0914,
      "step": 1381
    },
    {
      "epoch": 0.039820203999308475,
      "grad_norm": 3.1175501346588135,
      "learning_rate": 9.997483310045049e-06,
      "loss": 1.0594,
      "step": 1382
    },
    {
      "epoch": 0.03984901746095776,
      "grad_norm": 3.1381795406341553,
      "learning_rate": 9.997468485467003e-06,
      "loss": 1.0511,
      "step": 1383
    },
    {
      "epoch": 0.03987783092260704,
      "grad_norm": 3.0991735458374023,
      "learning_rate": 9.997453617366001e-06,
      "loss": 1.014,
      "step": 1384
    },
    {
      "epoch": 0.039906644384256326,
      "grad_norm": 3.0596001148223877,
      "learning_rate": 9.997438705742176e-06,
      "loss": 1.0671,
      "step": 1385
    },
    {
      "epoch": 0.03993545784590561,
      "grad_norm": 2.882310390472412,
      "learning_rate": 9.997423750595652e-06,
      "loss": 1.037,
      "step": 1386
    },
    {
      "epoch": 0.03996427130755489,
      "grad_norm": 2.7929189205169678,
      "learning_rate": 9.997408751926565e-06,
      "loss": 1.0847,
      "step": 1387
    },
    {
      "epoch": 0.03999308476920417,
      "grad_norm": 2.6509928703308105,
      "learning_rate": 9.997393709735043e-06,
      "loss": 1.0513,
      "step": 1388
    },
    {
      "epoch": 0.040021898230853455,
      "grad_norm": 2.8673388957977295,
      "learning_rate": 9.997378624021217e-06,
      "loss": 1.0553,
      "step": 1389
    },
    {
      "epoch": 0.04005071169250274,
      "grad_norm": 2.868492841720581,
      "learning_rate": 9.997363494785218e-06,
      "loss": 1.0894,
      "step": 1390
    },
    {
      "epoch": 0.04007952515415202,
      "grad_norm": 2.695035457611084,
      "learning_rate": 9.997348322027181e-06,
      "loss": 1.0519,
      "step": 1391
    },
    {
      "epoch": 0.0401083386158013,
      "grad_norm": 2.9517900943756104,
      "learning_rate": 9.997333105747234e-06,
      "loss": 1.0551,
      "step": 1392
    },
    {
      "epoch": 0.04013715207745058,
      "grad_norm": 3.0223968029022217,
      "learning_rate": 9.99731784594551e-06,
      "loss": 1.0655,
      "step": 1393
    },
    {
      "epoch": 0.04016596553909987,
      "grad_norm": 2.8234636783599854,
      "learning_rate": 9.997302542622144e-06,
      "loss": 1.0791,
      "step": 1394
    },
    {
      "epoch": 0.04019477900074915,
      "grad_norm": 2.775575876235962,
      "learning_rate": 9.997287195777269e-06,
      "loss": 1.0817,
      "step": 1395
    },
    {
      "epoch": 0.040223592462398434,
      "grad_norm": 3.051666021347046,
      "learning_rate": 9.997271805411018e-06,
      "loss": 1.0722,
      "step": 1396
    },
    {
      "epoch": 0.04025240592404772,
      "grad_norm": 3.0206305980682373,
      "learning_rate": 9.997256371523524e-06,
      "loss": 1.0603,
      "step": 1397
    },
    {
      "epoch": 0.040281219385696995,
      "grad_norm": 2.6961188316345215,
      "learning_rate": 9.997240894114924e-06,
      "loss": 1.0657,
      "step": 1398
    },
    {
      "epoch": 0.04031003284734628,
      "grad_norm": 2.7211713790893555,
      "learning_rate": 9.997225373185349e-06,
      "loss": 1.0597,
      "step": 1399
    },
    {
      "epoch": 0.04033884630899556,
      "grad_norm": 2.724719285964966,
      "learning_rate": 9.997209808734938e-06,
      "loss": 1.0945,
      "step": 1400
    },
    {
      "epoch": 0.04036765977064485,
      "grad_norm": 2.70961856842041,
      "learning_rate": 9.997194200763824e-06,
      "loss": 1.0878,
      "step": 1401
    },
    {
      "epoch": 0.04039647323229413,
      "grad_norm": 3.023541212081909,
      "learning_rate": 9.997178549272145e-06,
      "loss": 1.0647,
      "step": 1402
    },
    {
      "epoch": 0.04042528669394341,
      "grad_norm": 2.845315933227539,
      "learning_rate": 9.997162854260036e-06,
      "loss": 1.0678,
      "step": 1403
    },
    {
      "epoch": 0.04045410015559269,
      "grad_norm": 3.0915353298187256,
      "learning_rate": 9.997147115727633e-06,
      "loss": 1.0393,
      "step": 1404
    },
    {
      "epoch": 0.040482913617241975,
      "grad_norm": 2.6913039684295654,
      "learning_rate": 9.997131333675075e-06,
      "loss": 1.0301,
      "step": 1405
    },
    {
      "epoch": 0.04051172707889126,
      "grad_norm": 2.791541576385498,
      "learning_rate": 9.997115508102497e-06,
      "loss": 1.0433,
      "step": 1406
    },
    {
      "epoch": 0.04054054054054054,
      "grad_norm": 2.739292621612549,
      "learning_rate": 9.99709963901004e-06,
      "loss": 1.0414,
      "step": 1407
    },
    {
      "epoch": 0.04056935400218982,
      "grad_norm": 2.651334047317505,
      "learning_rate": 9.997083726397839e-06,
      "loss": 1.0341,
      "step": 1408
    },
    {
      "epoch": 0.040598167463839104,
      "grad_norm": 3.0140697956085205,
      "learning_rate": 9.997067770266035e-06,
      "loss": 1.052,
      "step": 1409
    },
    {
      "epoch": 0.04062698092548839,
      "grad_norm": 15.423968315124512,
      "learning_rate": 9.997051770614765e-06,
      "loss": 1.1708,
      "step": 1410
    },
    {
      "epoch": 0.04065579438713767,
      "grad_norm": 3.5526723861694336,
      "learning_rate": 9.997035727444169e-06,
      "loss": 1.0426,
      "step": 1411
    },
    {
      "epoch": 0.040684607848786955,
      "grad_norm": 3.3425631523132324,
      "learning_rate": 9.997019640754388e-06,
      "loss": 1.0627,
      "step": 1412
    },
    {
      "epoch": 0.04071342131043624,
      "grad_norm": 3.3209903240203857,
      "learning_rate": 9.99700351054556e-06,
      "loss": 1.06,
      "step": 1413
    },
    {
      "epoch": 0.040742234772085516,
      "grad_norm": 3.371011734008789,
      "learning_rate": 9.996987336817829e-06,
      "loss": 1.0799,
      "step": 1414
    },
    {
      "epoch": 0.0407710482337348,
      "grad_norm": 3.6147003173828125,
      "learning_rate": 9.996971119571331e-06,
      "loss": 1.0852,
      "step": 1415
    },
    {
      "epoch": 0.04079986169538408,
      "grad_norm": 3.015821933746338,
      "learning_rate": 9.99695485880621e-06,
      "loss": 1.0556,
      "step": 1416
    },
    {
      "epoch": 0.04082867515703337,
      "grad_norm": 3.373199939727783,
      "learning_rate": 9.996938554522609e-06,
      "loss": 1.0569,
      "step": 1417
    },
    {
      "epoch": 0.04085748861868265,
      "grad_norm": 3.0104377269744873,
      "learning_rate": 9.996922206720667e-06,
      "loss": 1.0509,
      "step": 1418
    },
    {
      "epoch": 0.04088630208033193,
      "grad_norm": 3.2036619186401367,
      "learning_rate": 9.996905815400528e-06,
      "loss": 1.107,
      "step": 1419
    },
    {
      "epoch": 0.04091511554198121,
      "grad_norm": 3.10786771774292,
      "learning_rate": 9.996889380562334e-06,
      "loss": 1.0645,
      "step": 1420
    },
    {
      "epoch": 0.040943929003630496,
      "grad_norm": 2.8860509395599365,
      "learning_rate": 9.99687290220623e-06,
      "loss": 1.0976,
      "step": 1421
    },
    {
      "epoch": 0.04097274246527978,
      "grad_norm": 2.892165422439575,
      "learning_rate": 9.996856380332356e-06,
      "loss": 1.0345,
      "step": 1422
    },
    {
      "epoch": 0.04100155592692906,
      "grad_norm": 3.047703504562378,
      "learning_rate": 9.99683981494086e-06,
      "loss": 1.0636,
      "step": 1423
    },
    {
      "epoch": 0.04103036938857835,
      "grad_norm": 3.1705050468444824,
      "learning_rate": 9.996823206031883e-06,
      "loss": 1.0609,
      "step": 1424
    },
    {
      "epoch": 0.041059182850227624,
      "grad_norm": 2.8838794231414795,
      "learning_rate": 9.996806553605571e-06,
      "loss": 1.0646,
      "step": 1425
    },
    {
      "epoch": 0.04108799631187691,
      "grad_norm": 3.531174659729004,
      "learning_rate": 9.996789857662068e-06,
      "loss": 1.0528,
      "step": 1426
    },
    {
      "epoch": 0.04111680977352619,
      "grad_norm": 3.016420364379883,
      "learning_rate": 9.996773118201522e-06,
      "loss": 1.0681,
      "step": 1427
    },
    {
      "epoch": 0.041145623235175476,
      "grad_norm": 3.0491943359375,
      "learning_rate": 9.996756335224075e-06,
      "loss": 1.0704,
      "step": 1428
    },
    {
      "epoch": 0.04117443669682476,
      "grad_norm": 3.0824503898620605,
      "learning_rate": 9.996739508729877e-06,
      "loss": 1.0416,
      "step": 1429
    },
    {
      "epoch": 0.041203250158474036,
      "grad_norm": 3.0216171741485596,
      "learning_rate": 9.996722638719072e-06,
      "loss": 1.0596,
      "step": 1430
    },
    {
      "epoch": 0.04123206362012332,
      "grad_norm": 3.1762754917144775,
      "learning_rate": 9.996705725191809e-06,
      "loss": 1.0411,
      "step": 1431
    },
    {
      "epoch": 0.041260877081772604,
      "grad_norm": 3.1014657020568848,
      "learning_rate": 9.996688768148233e-06,
      "loss": 1.0595,
      "step": 1432
    },
    {
      "epoch": 0.04128969054342189,
      "grad_norm": 2.8875138759613037,
      "learning_rate": 9.996671767588493e-06,
      "loss": 1.0556,
      "step": 1433
    },
    {
      "epoch": 0.04131850400507117,
      "grad_norm": 2.8177075386047363,
      "learning_rate": 9.996654723512736e-06,
      "loss": 1.0571,
      "step": 1434
    },
    {
      "epoch": 0.04134731746672045,
      "grad_norm": 2.962718963623047,
      "learning_rate": 9.996637635921111e-06,
      "loss": 1.0397,
      "step": 1435
    },
    {
      "epoch": 0.04137613092836973,
      "grad_norm": 2.917926549911499,
      "learning_rate": 9.996620504813767e-06,
      "loss": 1.0555,
      "step": 1436
    },
    {
      "epoch": 0.041404944390019016,
      "grad_norm": 2.9081766605377197,
      "learning_rate": 9.996603330190854e-06,
      "loss": 1.0454,
      "step": 1437
    },
    {
      "epoch": 0.0414337578516683,
      "grad_norm": 3.178953170776367,
      "learning_rate": 9.996586112052522e-06,
      "loss": 1.0793,
      "step": 1438
    },
    {
      "epoch": 0.041462571313317584,
      "grad_norm": 3.0437300205230713,
      "learning_rate": 9.996568850398918e-06,
      "loss": 1.021,
      "step": 1439
    },
    {
      "epoch": 0.04149138477496687,
      "grad_norm": 2.6390511989593506,
      "learning_rate": 9.996551545230195e-06,
      "loss": 1.0838,
      "step": 1440
    },
    {
      "epoch": 0.041520198236616145,
      "grad_norm": 2.678601026535034,
      "learning_rate": 9.9965341965465e-06,
      "loss": 1.0628,
      "step": 1441
    },
    {
      "epoch": 0.04154901169826543,
      "grad_norm": 2.9008307456970215,
      "learning_rate": 9.996516804347991e-06,
      "loss": 1.0791,
      "step": 1442
    },
    {
      "epoch": 0.04157782515991471,
      "grad_norm": 2.7384519577026367,
      "learning_rate": 9.996499368634814e-06,
      "loss": 1.0361,
      "step": 1443
    },
    {
      "epoch": 0.041606638621563996,
      "grad_norm": 2.9903154373168945,
      "learning_rate": 9.99648188940712e-06,
      "loss": 1.0658,
      "step": 1444
    },
    {
      "epoch": 0.04163545208321328,
      "grad_norm": 2.8967032432556152,
      "learning_rate": 9.996464366665065e-06,
      "loss": 1.034,
      "step": 1445
    },
    {
      "epoch": 0.04166426554486256,
      "grad_norm": 2.79427170753479,
      "learning_rate": 9.9964468004088e-06,
      "loss": 1.0736,
      "step": 1446
    },
    {
      "epoch": 0.04169307900651184,
      "grad_norm": 2.692723274230957,
      "learning_rate": 9.996429190638477e-06,
      "loss": 1.0678,
      "step": 1447
    },
    {
      "epoch": 0.041721892468161124,
      "grad_norm": 2.710033416748047,
      "learning_rate": 9.99641153735425e-06,
      "loss": 1.07,
      "step": 1448
    },
    {
      "epoch": 0.04175070592981041,
      "grad_norm": 3.1109769344329834,
      "learning_rate": 9.996393840556276e-06,
      "loss": 1.0373,
      "step": 1449
    },
    {
      "epoch": 0.04177951939145969,
      "grad_norm": 2.705620050430298,
      "learning_rate": 9.996376100244705e-06,
      "loss": 1.0448,
      "step": 1450
    },
    {
      "epoch": 0.041808332853108976,
      "grad_norm": 2.8982949256896973,
      "learning_rate": 9.996358316419693e-06,
      "loss": 1.0339,
      "step": 1451
    },
    {
      "epoch": 0.04183714631475825,
      "grad_norm": 2.6670854091644287,
      "learning_rate": 9.996340489081393e-06,
      "loss": 1.0324,
      "step": 1452
    },
    {
      "epoch": 0.04186595977640754,
      "grad_norm": 3.4215428829193115,
      "learning_rate": 9.996322618229963e-06,
      "loss": 1.0813,
      "step": 1453
    },
    {
      "epoch": 0.04189477323805682,
      "grad_norm": 2.5509724617004395,
      "learning_rate": 9.996304703865557e-06,
      "loss": 1.0479,
      "step": 1454
    },
    {
      "epoch": 0.041923586699706104,
      "grad_norm": 3.179236650466919,
      "learning_rate": 9.996286745988332e-06,
      "loss": 1.0708,
      "step": 1455
    },
    {
      "epoch": 0.04195240016135539,
      "grad_norm": 2.8519322872161865,
      "learning_rate": 9.996268744598443e-06,
      "loss": 1.0307,
      "step": 1456
    },
    {
      "epoch": 0.041981213623004665,
      "grad_norm": 2.7361207008361816,
      "learning_rate": 9.996250699696048e-06,
      "loss": 1.0584,
      "step": 1457
    },
    {
      "epoch": 0.04201002708465395,
      "grad_norm": 2.8430638313293457,
      "learning_rate": 9.996232611281305e-06,
      "loss": 1.0268,
      "step": 1458
    },
    {
      "epoch": 0.04203884054630323,
      "grad_norm": 3.17450213432312,
      "learning_rate": 9.996214479354369e-06,
      "loss": 1.0601,
      "step": 1459
    },
    {
      "epoch": 0.04206765400795252,
      "grad_norm": 2.9459822177886963,
      "learning_rate": 9.9961963039154e-06,
      "loss": 1.0516,
      "step": 1460
    },
    {
      "epoch": 0.0420964674696018,
      "grad_norm": 2.8213272094726562,
      "learning_rate": 9.996178084964554e-06,
      "loss": 1.0409,
      "step": 1461
    },
    {
      "epoch": 0.04212528093125108,
      "grad_norm": 2.6436119079589844,
      "learning_rate": 9.996159822501994e-06,
      "loss": 1.0349,
      "step": 1462
    },
    {
      "epoch": 0.04215409439290036,
      "grad_norm": 2.753875494003296,
      "learning_rate": 9.996141516527875e-06,
      "loss": 1.0371,
      "step": 1463
    },
    {
      "epoch": 0.042182907854549645,
      "grad_norm": 3.146434783935547,
      "learning_rate": 9.996123167042358e-06,
      "loss": 1.0633,
      "step": 1464
    },
    {
      "epoch": 0.04221172131619893,
      "grad_norm": 2.696434497833252,
      "learning_rate": 9.9961047740456e-06,
      "loss": 1.0512,
      "step": 1465
    },
    {
      "epoch": 0.04224053477784821,
      "grad_norm": 3.3079309463500977,
      "learning_rate": 9.996086337537767e-06,
      "loss": 1.0332,
      "step": 1466
    },
    {
      "epoch": 0.042269348239497496,
      "grad_norm": 2.790100574493408,
      "learning_rate": 9.996067857519016e-06,
      "loss": 1.0583,
      "step": 1467
    },
    {
      "epoch": 0.04229816170114677,
      "grad_norm": 3.300511598587036,
      "learning_rate": 9.996049333989505e-06,
      "loss": 1.0601,
      "step": 1468
    },
    {
      "epoch": 0.04232697516279606,
      "grad_norm": 3.0361478328704834,
      "learning_rate": 9.996030766949402e-06,
      "loss": 1.0266,
      "step": 1469
    },
    {
      "epoch": 0.04235578862444534,
      "grad_norm": 3.216221332550049,
      "learning_rate": 9.996012156398864e-06,
      "loss": 1.0619,
      "step": 1470
    },
    {
      "epoch": 0.042384602086094625,
      "grad_norm": 2.999999761581421,
      "learning_rate": 9.995993502338054e-06,
      "loss": 1.0553,
      "step": 1471
    },
    {
      "epoch": 0.04241341554774391,
      "grad_norm": 3.340301513671875,
      "learning_rate": 9.995974804767134e-06,
      "loss": 1.0615,
      "step": 1472
    },
    {
      "epoch": 0.042442229009393186,
      "grad_norm": 2.8336374759674072,
      "learning_rate": 9.99595606368627e-06,
      "loss": 1.0461,
      "step": 1473
    },
    {
      "epoch": 0.04247104247104247,
      "grad_norm": 3.2301173210144043,
      "learning_rate": 9.995937279095621e-06,
      "loss": 1.0602,
      "step": 1474
    },
    {
      "epoch": 0.04249985593269175,
      "grad_norm": 2.8270761966705322,
      "learning_rate": 9.995918450995352e-06,
      "loss": 1.0358,
      "step": 1475
    },
    {
      "epoch": 0.04252866939434104,
      "grad_norm": 3.0351481437683105,
      "learning_rate": 9.995899579385628e-06,
      "loss": 1.0436,
      "step": 1476
    },
    {
      "epoch": 0.04255748285599032,
      "grad_norm": 3.008925199508667,
      "learning_rate": 9.995880664266614e-06,
      "loss": 1.0503,
      "step": 1477
    },
    {
      "epoch": 0.0425862963176396,
      "grad_norm": 2.7967634201049805,
      "learning_rate": 9.995861705638473e-06,
      "loss": 1.0644,
      "step": 1478
    },
    {
      "epoch": 0.04261510977928888,
      "grad_norm": 2.78788161277771,
      "learning_rate": 9.99584270350137e-06,
      "loss": 1.0494,
      "step": 1479
    },
    {
      "epoch": 0.042643923240938165,
      "grad_norm": 2.895944118499756,
      "learning_rate": 9.99582365785547e-06,
      "loss": 1.0136,
      "step": 1480
    },
    {
      "epoch": 0.04267273670258745,
      "grad_norm": 2.6956217288970947,
      "learning_rate": 9.995804568700942e-06,
      "loss": 1.0666,
      "step": 1481
    },
    {
      "epoch": 0.04270155016423673,
      "grad_norm": 3.1833372116088867,
      "learning_rate": 9.995785436037947e-06,
      "loss": 1.0605,
      "step": 1482
    },
    {
      "epoch": 0.04273036362588602,
      "grad_norm": 2.824712038040161,
      "learning_rate": 9.995766259866658e-06,
      "loss": 1.0526,
      "step": 1483
    },
    {
      "epoch": 0.042759177087535294,
      "grad_norm": 2.873511791229248,
      "learning_rate": 9.995747040187238e-06,
      "loss": 1.0573,
      "step": 1484
    },
    {
      "epoch": 0.04278799054918458,
      "grad_norm": 2.6457223892211914,
      "learning_rate": 9.995727776999855e-06,
      "loss": 1.0685,
      "step": 1485
    },
    {
      "epoch": 0.04281680401083386,
      "grad_norm": 2.987253427505493,
      "learning_rate": 9.995708470304676e-06,
      "loss": 1.0437,
      "step": 1486
    },
    {
      "epoch": 0.042845617472483145,
      "grad_norm": 3.1081833839416504,
      "learning_rate": 9.995689120101872e-06,
      "loss": 1.0468,
      "step": 1487
    },
    {
      "epoch": 0.04287443093413243,
      "grad_norm": 3.1582510471343994,
      "learning_rate": 9.995669726391608e-06,
      "loss": 1.0499,
      "step": 1488
    },
    {
      "epoch": 0.042903244395781706,
      "grad_norm": 3.0653913021087646,
      "learning_rate": 9.995650289174056e-06,
      "loss": 1.0568,
      "step": 1489
    },
    {
      "epoch": 0.04293205785743099,
      "grad_norm": 3.0734920501708984,
      "learning_rate": 9.995630808449384e-06,
      "loss": 1.0551,
      "step": 1490
    },
    {
      "epoch": 0.042960871319080274,
      "grad_norm": 2.784834623336792,
      "learning_rate": 9.995611284217758e-06,
      "loss": 1.0507,
      "step": 1491
    },
    {
      "epoch": 0.04298968478072956,
      "grad_norm": 2.9298629760742188,
      "learning_rate": 9.995591716479356e-06,
      "loss": 1.0369,
      "step": 1492
    },
    {
      "epoch": 0.04301849824237884,
      "grad_norm": 2.925407648086548,
      "learning_rate": 9.995572105234341e-06,
      "loss": 1.0271,
      "step": 1493
    },
    {
      "epoch": 0.043047311704028125,
      "grad_norm": 3.168325662612915,
      "learning_rate": 9.995552450482889e-06,
      "loss": 1.0711,
      "step": 1494
    },
    {
      "epoch": 0.0430761251656774,
      "grad_norm": 3.121279716491699,
      "learning_rate": 9.995532752225165e-06,
      "loss": 1.0186,
      "step": 1495
    },
    {
      "epoch": 0.043104938627326686,
      "grad_norm": 3.414674758911133,
      "learning_rate": 9.995513010461348e-06,
      "loss": 1.0871,
      "step": 1496
    },
    {
      "epoch": 0.04313375208897597,
      "grad_norm": 2.979029655456543,
      "learning_rate": 9.995493225191605e-06,
      "loss": 1.0469,
      "step": 1497
    },
    {
      "epoch": 0.043162565550625254,
      "grad_norm": 3.0120818614959717,
      "learning_rate": 9.99547339641611e-06,
      "loss": 1.0578,
      "step": 1498
    },
    {
      "epoch": 0.04319137901227454,
      "grad_norm": 2.8609976768493652,
      "learning_rate": 9.995453524135035e-06,
      "loss": 1.0635,
      "step": 1499
    },
    {
      "epoch": 0.043220192473923814,
      "grad_norm": 2.8117032051086426,
      "learning_rate": 9.995433608348554e-06,
      "loss": 1.0673,
      "step": 1500
    },
    {
      "epoch": 0.0432490059355731,
      "grad_norm": 3.9076595306396484,
      "learning_rate": 9.99541364905684e-06,
      "loss": 1.0679,
      "step": 1501
    },
    {
      "epoch": 0.04327781939722238,
      "grad_norm": 3.162328004837036,
      "learning_rate": 9.995393646260067e-06,
      "loss": 1.0663,
      "step": 1502
    },
    {
      "epoch": 0.043306632858871666,
      "grad_norm": 2.7413382530212402,
      "learning_rate": 9.995373599958407e-06,
      "loss": 1.0236,
      "step": 1503
    },
    {
      "epoch": 0.04333544632052095,
      "grad_norm": 2.905406951904297,
      "learning_rate": 9.995353510152038e-06,
      "loss": 1.0557,
      "step": 1504
    },
    {
      "epoch": 0.04336425978217023,
      "grad_norm": 2.8208134174346924,
      "learning_rate": 9.995333376841134e-06,
      "loss": 1.0363,
      "step": 1505
    },
    {
      "epoch": 0.04339307324381951,
      "grad_norm": 2.6396543979644775,
      "learning_rate": 9.99531320002587e-06,
      "loss": 1.0667,
      "step": 1506
    },
    {
      "epoch": 0.043421886705468794,
      "grad_norm": 2.760120391845703,
      "learning_rate": 9.995292979706419e-06,
      "loss": 1.0385,
      "step": 1507
    },
    {
      "epoch": 0.04345070016711808,
      "grad_norm": 2.583303689956665,
      "learning_rate": 9.99527271588296e-06,
      "loss": 1.0718,
      "step": 1508
    },
    {
      "epoch": 0.04347951362876736,
      "grad_norm": 2.7696011066436768,
      "learning_rate": 9.99525240855567e-06,
      "loss": 1.0604,
      "step": 1509
    },
    {
      "epoch": 0.043508327090416646,
      "grad_norm": 2.8322365283966064,
      "learning_rate": 9.995232057724726e-06,
      "loss": 1.0332,
      "step": 1510
    },
    {
      "epoch": 0.04353714055206592,
      "grad_norm": 2.7473747730255127,
      "learning_rate": 9.995211663390303e-06,
      "loss": 1.0163,
      "step": 1511
    },
    {
      "epoch": 0.04356595401371521,
      "grad_norm": 2.673323154449463,
      "learning_rate": 9.995191225552581e-06,
      "loss": 1.0474,
      "step": 1512
    },
    {
      "epoch": 0.04359476747536449,
      "grad_norm": 2.702691078186035,
      "learning_rate": 9.995170744211735e-06,
      "loss": 1.0425,
      "step": 1513
    },
    {
      "epoch": 0.043623580937013774,
      "grad_norm": 3.2567203044891357,
      "learning_rate": 9.995150219367946e-06,
      "loss": 1.0489,
      "step": 1514
    },
    {
      "epoch": 0.04365239439866306,
      "grad_norm": 3.2553951740264893,
      "learning_rate": 9.995129651021393e-06,
      "loss": 1.0442,
      "step": 1515
    },
    {
      "epoch": 0.043681207860312335,
      "grad_norm": 2.9476687908172607,
      "learning_rate": 9.995109039172253e-06,
      "loss": 1.0409,
      "step": 1516
    },
    {
      "epoch": 0.04371002132196162,
      "grad_norm": 2.509268283843994,
      "learning_rate": 9.995088383820706e-06,
      "loss": 1.0335,
      "step": 1517
    },
    {
      "epoch": 0.0437388347836109,
      "grad_norm": 3.3612821102142334,
      "learning_rate": 9.995067684966933e-06,
      "loss": 1.0476,
      "step": 1518
    },
    {
      "epoch": 0.043767648245260186,
      "grad_norm": 2.615337610244751,
      "learning_rate": 9.995046942611113e-06,
      "loss": 1.0534,
      "step": 1519
    },
    {
      "epoch": 0.04379646170690947,
      "grad_norm": 3.053025245666504,
      "learning_rate": 9.99502615675343e-06,
      "loss": 1.0075,
      "step": 1520
    },
    {
      "epoch": 0.043825275168558754,
      "grad_norm": 2.842210292816162,
      "learning_rate": 9.995005327394059e-06,
      "loss": 1.0324,
      "step": 1521
    },
    {
      "epoch": 0.04385408863020803,
      "grad_norm": 2.6389997005462646,
      "learning_rate": 9.994984454533185e-06,
      "loss": 1.049,
      "step": 1522
    },
    {
      "epoch": 0.043882902091857315,
      "grad_norm": 2.8405704498291016,
      "learning_rate": 9.994963538170993e-06,
      "loss": 1.0465,
      "step": 1523
    },
    {
      "epoch": 0.0439117155535066,
      "grad_norm": 2.8969476222991943,
      "learning_rate": 9.994942578307658e-06,
      "loss": 1.0293,
      "step": 1524
    },
    {
      "epoch": 0.04394052901515588,
      "grad_norm": 3.050722599029541,
      "learning_rate": 9.994921574943368e-06,
      "loss": 1.058,
      "step": 1525
    },
    {
      "epoch": 0.043969342476805166,
      "grad_norm": 2.7692205905914307,
      "learning_rate": 9.994900528078305e-06,
      "loss": 1.0414,
      "step": 1526
    },
    {
      "epoch": 0.04399815593845444,
      "grad_norm": 2.9579336643218994,
      "learning_rate": 9.994879437712648e-06,
      "loss": 1.0503,
      "step": 1527
    },
    {
      "epoch": 0.04402696940010373,
      "grad_norm": 2.8961148262023926,
      "learning_rate": 9.994858303846588e-06,
      "loss": 1.0473,
      "step": 1528
    },
    {
      "epoch": 0.04405578286175301,
      "grad_norm": 2.802523374557495,
      "learning_rate": 9.994837126480304e-06,
      "loss": 1.0536,
      "step": 1529
    },
    {
      "epoch": 0.044084596323402295,
      "grad_norm": 2.9186465740203857,
      "learning_rate": 9.994815905613982e-06,
      "loss": 1.0355,
      "step": 1530
    },
    {
      "epoch": 0.04411340978505158,
      "grad_norm": 2.8992791175842285,
      "learning_rate": 9.994794641247805e-06,
      "loss": 1.0593,
      "step": 1531
    },
    {
      "epoch": 0.044142223246700855,
      "grad_norm": 2.821106195449829,
      "learning_rate": 9.99477333338196e-06,
      "loss": 1.0789,
      "step": 1532
    },
    {
      "epoch": 0.04417103670835014,
      "grad_norm": 3.143019676208496,
      "learning_rate": 9.994751982016633e-06,
      "loss": 1.0311,
      "step": 1533
    },
    {
      "epoch": 0.04419985016999942,
      "grad_norm": 2.86830997467041,
      "learning_rate": 9.994730587152009e-06,
      "loss": 1.0552,
      "step": 1534
    },
    {
      "epoch": 0.04422866363164871,
      "grad_norm": 3.1544036865234375,
      "learning_rate": 9.994709148788273e-06,
      "loss": 1.0554,
      "step": 1535
    },
    {
      "epoch": 0.04425747709329799,
      "grad_norm": 3.026179313659668,
      "learning_rate": 9.994687666925615e-06,
      "loss": 1.018,
      "step": 1536
    },
    {
      "epoch": 0.044286290554947275,
      "grad_norm": 3.021136522293091,
      "learning_rate": 9.99466614156422e-06,
      "loss": 1.0625,
      "step": 1537
    },
    {
      "epoch": 0.04431510401659655,
      "grad_norm": 2.9415547847747803,
      "learning_rate": 9.994644572704274e-06,
      "loss": 1.0606,
      "step": 1538
    },
    {
      "epoch": 0.044343917478245835,
      "grad_norm": 2.9647114276885986,
      "learning_rate": 9.99462296034597e-06,
      "loss": 1.0518,
      "step": 1539
    },
    {
      "epoch": 0.04437273093989512,
      "grad_norm": 2.8385467529296875,
      "learning_rate": 9.994601304489492e-06,
      "loss": 1.05,
      "step": 1540
    },
    {
      "epoch": 0.0444015444015444,
      "grad_norm": 3.005507230758667,
      "learning_rate": 9.994579605135027e-06,
      "loss": 1.0209,
      "step": 1541
    },
    {
      "epoch": 0.04443035786319369,
      "grad_norm": 2.839846134185791,
      "learning_rate": 9.994557862282767e-06,
      "loss": 1.061,
      "step": 1542
    },
    {
      "epoch": 0.044459171324842964,
      "grad_norm": 2.8778045177459717,
      "learning_rate": 9.994536075932903e-06,
      "loss": 1.1033,
      "step": 1543
    },
    {
      "epoch": 0.04448798478649225,
      "grad_norm": 2.7379255294799805,
      "learning_rate": 9.99451424608562e-06,
      "loss": 1.056,
      "step": 1544
    },
    {
      "epoch": 0.04451679824814153,
      "grad_norm": 3.0111963748931885,
      "learning_rate": 9.994492372741111e-06,
      "loss": 1.0317,
      "step": 1545
    },
    {
      "epoch": 0.044545611709790815,
      "grad_norm": 2.6979315280914307,
      "learning_rate": 9.994470455899568e-06,
      "loss": 1.0315,
      "step": 1546
    },
    {
      "epoch": 0.0445744251714401,
      "grad_norm": 2.7595293521881104,
      "learning_rate": 9.994448495561178e-06,
      "loss": 1.0481,
      "step": 1547
    },
    {
      "epoch": 0.044603238633089376,
      "grad_norm": 2.5340704917907715,
      "learning_rate": 9.994426491726134e-06,
      "loss": 1.0495,
      "step": 1548
    },
    {
      "epoch": 0.04463205209473866,
      "grad_norm": 2.8489608764648438,
      "learning_rate": 9.99440444439463e-06,
      "loss": 1.0198,
      "step": 1549
    },
    {
      "epoch": 0.044660865556387944,
      "grad_norm": 2.670135498046875,
      "learning_rate": 9.994382353566854e-06,
      "loss": 1.0549,
      "step": 1550
    },
    {
      "epoch": 0.04468967901803723,
      "grad_norm": 2.840640068054199,
      "learning_rate": 9.994360219243e-06,
      "loss": 1.0671,
      "step": 1551
    },
    {
      "epoch": 0.04471849247968651,
      "grad_norm": 2.8129279613494873,
      "learning_rate": 9.994338041423263e-06,
      "loss": 1.065,
      "step": 1552
    },
    {
      "epoch": 0.044747305941335795,
      "grad_norm": 2.838289260864258,
      "learning_rate": 9.994315820107833e-06,
      "loss": 1.0681,
      "step": 1553
    },
    {
      "epoch": 0.04477611940298507,
      "grad_norm": 2.7664434909820557,
      "learning_rate": 9.994293555296905e-06,
      "loss": 1.0689,
      "step": 1554
    },
    {
      "epoch": 0.044804932864634356,
      "grad_norm": 2.7333860397338867,
      "learning_rate": 9.994271246990671e-06,
      "loss": 1.0456,
      "step": 1555
    },
    {
      "epoch": 0.04483374632628364,
      "grad_norm": 3.2270278930664062,
      "learning_rate": 9.99424889518933e-06,
      "loss": 1.0642,
      "step": 1556
    },
    {
      "epoch": 0.044862559787932924,
      "grad_norm": 2.5166923999786377,
      "learning_rate": 9.994226499893071e-06,
      "loss": 1.0464,
      "step": 1557
    },
    {
      "epoch": 0.04489137324958221,
      "grad_norm": 3.0092570781707764,
      "learning_rate": 9.994204061102093e-06,
      "loss": 1.1002,
      "step": 1558
    },
    {
      "epoch": 0.044920186711231484,
      "grad_norm": 2.5529491901397705,
      "learning_rate": 9.994181578816588e-06,
      "loss": 1.0498,
      "step": 1559
    },
    {
      "epoch": 0.04494900017288077,
      "grad_norm": 2.7890737056732178,
      "learning_rate": 9.994159053036756e-06,
      "loss": 1.052,
      "step": 1560
    },
    {
      "epoch": 0.04497781363453005,
      "grad_norm": 2.616889476776123,
      "learning_rate": 9.994136483762791e-06,
      "loss": 1.0541,
      "step": 1561
    },
    {
      "epoch": 0.045006627096179336,
      "grad_norm": 2.756232976913452,
      "learning_rate": 9.994113870994888e-06,
      "loss": 1.0496,
      "step": 1562
    },
    {
      "epoch": 0.04503544055782862,
      "grad_norm": 2.61063289642334,
      "learning_rate": 9.994091214733245e-06,
      "loss": 1.0126,
      "step": 1563
    },
    {
      "epoch": 0.0450642540194779,
      "grad_norm": 2.8222930431365967,
      "learning_rate": 9.994068514978062e-06,
      "loss": 1.0207,
      "step": 1564
    },
    {
      "epoch": 0.04509306748112718,
      "grad_norm": 2.8506054878234863,
      "learning_rate": 9.994045771729532e-06,
      "loss": 1.0006,
      "step": 1565
    },
    {
      "epoch": 0.045121880942776464,
      "grad_norm": 2.885759115219116,
      "learning_rate": 9.994022984987858e-06,
      "loss": 1.0474,
      "step": 1566
    },
    {
      "epoch": 0.04515069440442575,
      "grad_norm": 3.1718266010284424,
      "learning_rate": 9.994000154753236e-06,
      "loss": 1.0498,
      "step": 1567
    },
    {
      "epoch": 0.04517950786607503,
      "grad_norm": 2.790503978729248,
      "learning_rate": 9.993977281025862e-06,
      "loss": 1.0414,
      "step": 1568
    },
    {
      "epoch": 0.045208321327724316,
      "grad_norm": 3.094797372817993,
      "learning_rate": 9.99395436380594e-06,
      "loss": 1.0387,
      "step": 1569
    },
    {
      "epoch": 0.04523713478937359,
      "grad_norm": 2.6336541175842285,
      "learning_rate": 9.993931403093668e-06,
      "loss": 1.0599,
      "step": 1570
    },
    {
      "epoch": 0.045265948251022876,
      "grad_norm": 2.6810286045074463,
      "learning_rate": 9.993908398889244e-06,
      "loss": 1.0548,
      "step": 1571
    },
    {
      "epoch": 0.04529476171267216,
      "grad_norm": 2.635596513748169,
      "learning_rate": 9.993885351192872e-06,
      "loss": 1.0245,
      "step": 1572
    },
    {
      "epoch": 0.045323575174321444,
      "grad_norm": 2.705263376235962,
      "learning_rate": 9.99386226000475e-06,
      "loss": 1.06,
      "step": 1573
    },
    {
      "epoch": 0.04535238863597073,
      "grad_norm": 2.6436214447021484,
      "learning_rate": 9.993839125325078e-06,
      "loss": 1.0485,
      "step": 1574
    },
    {
      "epoch": 0.045381202097620005,
      "grad_norm": 2.4909064769744873,
      "learning_rate": 9.993815947154061e-06,
      "loss": 1.0458,
      "step": 1575
    },
    {
      "epoch": 0.04541001555926929,
      "grad_norm": 2.857164144515991,
      "learning_rate": 9.9937927254919e-06,
      "loss": 1.0476,
      "step": 1576
    },
    {
      "epoch": 0.04543882902091857,
      "grad_norm": 2.55193829536438,
      "learning_rate": 9.993769460338794e-06,
      "loss": 1.0524,
      "step": 1577
    },
    {
      "epoch": 0.045467642482567856,
      "grad_norm": 4.535765171051025,
      "learning_rate": 9.993746151694949e-06,
      "loss": 1.0494,
      "step": 1578
    },
    {
      "epoch": 0.04549645594421714,
      "grad_norm": 2.7849082946777344,
      "learning_rate": 9.993722799560566e-06,
      "loss": 1.0217,
      "step": 1579
    },
    {
      "epoch": 0.045525269405866424,
      "grad_norm": 2.819801092147827,
      "learning_rate": 9.99369940393585e-06,
      "loss": 1.046,
      "step": 1580
    },
    {
      "epoch": 0.0455540828675157,
      "grad_norm": 2.871772050857544,
      "learning_rate": 9.993675964821006e-06,
      "loss": 1.0349,
      "step": 1581
    },
    {
      "epoch": 0.045582896329164985,
      "grad_norm": 3.035337448120117,
      "learning_rate": 9.993652482216234e-06,
      "loss": 1.0157,
      "step": 1582
    },
    {
      "epoch": 0.04561170979081427,
      "grad_norm": 3.2969605922698975,
      "learning_rate": 9.993628956121741e-06,
      "loss": 1.0545,
      "step": 1583
    },
    {
      "epoch": 0.04564052325246355,
      "grad_norm": 2.742990732192993,
      "learning_rate": 9.993605386537731e-06,
      "loss": 1.0139,
      "step": 1584
    },
    {
      "epoch": 0.045669336714112836,
      "grad_norm": 2.986760377883911,
      "learning_rate": 9.99358177346441e-06,
      "loss": 1.0271,
      "step": 1585
    },
    {
      "epoch": 0.04569815017576211,
      "grad_norm": 2.477505922317505,
      "learning_rate": 9.993558116901984e-06,
      "loss": 1.0536,
      "step": 1586
    },
    {
      "epoch": 0.0457269636374114,
      "grad_norm": 2.763347625732422,
      "learning_rate": 9.99353441685066e-06,
      "loss": 1.025,
      "step": 1587
    },
    {
      "epoch": 0.04575577709906068,
      "grad_norm": 2.919158697128296,
      "learning_rate": 9.993510673310641e-06,
      "loss": 1.0563,
      "step": 1588
    },
    {
      "epoch": 0.045784590560709965,
      "grad_norm": 2.657439947128296,
      "learning_rate": 9.993486886282137e-06,
      "loss": 1.0367,
      "step": 1589
    },
    {
      "epoch": 0.04581340402235925,
      "grad_norm": 2.651254177093506,
      "learning_rate": 9.993463055765353e-06,
      "loss": 1.0618,
      "step": 1590
    },
    {
      "epoch": 0.04584221748400853,
      "grad_norm": 2.8461644649505615,
      "learning_rate": 9.993439181760497e-06,
      "loss": 1.0451,
      "step": 1591
    },
    {
      "epoch": 0.04587103094565781,
      "grad_norm": 2.8058385848999023,
      "learning_rate": 9.993415264267779e-06,
      "loss": 1.0494,
      "step": 1592
    },
    {
      "epoch": 0.04589984440730709,
      "grad_norm": 2.554931163787842,
      "learning_rate": 9.993391303287403e-06,
      "loss": 1.0244,
      "step": 1593
    },
    {
      "epoch": 0.04592865786895638,
      "grad_norm": 2.7879791259765625,
      "learning_rate": 9.993367298819583e-06,
      "loss": 1.0643,
      "step": 1594
    },
    {
      "epoch": 0.04595747133060566,
      "grad_norm": 2.687439441680908,
      "learning_rate": 9.993343250864524e-06,
      "loss": 1.0312,
      "step": 1595
    },
    {
      "epoch": 0.045986284792254944,
      "grad_norm": 2.695126533508301,
      "learning_rate": 9.993319159422436e-06,
      "loss": 1.0278,
      "step": 1596
    },
    {
      "epoch": 0.04601509825390422,
      "grad_norm": 2.7487356662750244,
      "learning_rate": 9.993295024493532e-06,
      "loss": 1.025,
      "step": 1597
    },
    {
      "epoch": 0.046043911715553505,
      "grad_norm": 2.7597124576568604,
      "learning_rate": 9.993270846078019e-06,
      "loss": 1.024,
      "step": 1598
    },
    {
      "epoch": 0.04607272517720279,
      "grad_norm": 2.5628390312194824,
      "learning_rate": 9.993246624176107e-06,
      "loss": 1.0095,
      "step": 1599
    },
    {
      "epoch": 0.04610153863885207,
      "grad_norm": 2.645772695541382,
      "learning_rate": 9.993222358788008e-06,
      "loss": 1.0205,
      "step": 1600
    },
    {
      "epoch": 0.04613035210050136,
      "grad_norm": 2.691754102706909,
      "learning_rate": 9.993198049913936e-06,
      "loss": 1.0509,
      "step": 1601
    },
    {
      "epoch": 0.046159165562150634,
      "grad_norm": 2.8318467140197754,
      "learning_rate": 9.9931736975541e-06,
      "loss": 1.0743,
      "step": 1602
    },
    {
      "epoch": 0.04618797902379992,
      "grad_norm": 2.7442169189453125,
      "learning_rate": 9.993149301708711e-06,
      "loss": 1.037,
      "step": 1603
    },
    {
      "epoch": 0.0462167924854492,
      "grad_norm": 2.6668756008148193,
      "learning_rate": 9.993124862377982e-06,
      "loss": 1.0398,
      "step": 1604
    },
    {
      "epoch": 0.046245605947098485,
      "grad_norm": 2.768238067626953,
      "learning_rate": 9.993100379562129e-06,
      "loss": 1.0555,
      "step": 1605
    },
    {
      "epoch": 0.04627441940874777,
      "grad_norm": 2.582547903060913,
      "learning_rate": 9.993075853261363e-06,
      "loss": 1.0533,
      "step": 1606
    },
    {
      "epoch": 0.04630323287039705,
      "grad_norm": 2.4760794639587402,
      "learning_rate": 9.993051283475897e-06,
      "loss": 1.0636,
      "step": 1607
    },
    {
      "epoch": 0.04633204633204633,
      "grad_norm": 2.7987453937530518,
      "learning_rate": 9.993026670205944e-06,
      "loss": 1.0243,
      "step": 1608
    },
    {
      "epoch": 0.046360859793695613,
      "grad_norm": 2.4654464721679688,
      "learning_rate": 9.993002013451723e-06,
      "loss": 1.0299,
      "step": 1609
    },
    {
      "epoch": 0.0463896732553449,
      "grad_norm": 2.843885898590088,
      "learning_rate": 9.992977313213444e-06,
      "loss": 1.0436,
      "step": 1610
    },
    {
      "epoch": 0.04641848671699418,
      "grad_norm": 2.6724627017974854,
      "learning_rate": 9.992952569491322e-06,
      "loss": 1.0586,
      "step": 1611
    },
    {
      "epoch": 0.046447300178643465,
      "grad_norm": 2.89041805267334,
      "learning_rate": 9.992927782285577e-06,
      "loss": 1.0597,
      "step": 1612
    },
    {
      "epoch": 0.04647611364029274,
      "grad_norm": 2.9250128269195557,
      "learning_rate": 9.99290295159642e-06,
      "loss": 1.0163,
      "step": 1613
    },
    {
      "epoch": 0.046504927101942026,
      "grad_norm": 2.769423484802246,
      "learning_rate": 9.99287807742407e-06,
      "loss": 1.0478,
      "step": 1614
    },
    {
      "epoch": 0.04653374056359131,
      "grad_norm": 2.978161334991455,
      "learning_rate": 9.992853159768744e-06,
      "loss": 1.0391,
      "step": 1615
    },
    {
      "epoch": 0.04656255402524059,
      "grad_norm": 2.750694751739502,
      "learning_rate": 9.992828198630658e-06,
      "loss": 1.001,
      "step": 1616
    },
    {
      "epoch": 0.04659136748688988,
      "grad_norm": 2.6932249069213867,
      "learning_rate": 9.992803194010029e-06,
      "loss": 1.0196,
      "step": 1617
    },
    {
      "epoch": 0.046620180948539154,
      "grad_norm": 2.7046546936035156,
      "learning_rate": 9.992778145907073e-06,
      "loss": 1.035,
      "step": 1618
    },
    {
      "epoch": 0.04664899441018844,
      "grad_norm": 2.8533928394317627,
      "learning_rate": 9.992753054322014e-06,
      "loss": 1.0715,
      "step": 1619
    },
    {
      "epoch": 0.04667780787183772,
      "grad_norm": 3.0260000228881836,
      "learning_rate": 9.992727919255065e-06,
      "loss": 1.043,
      "step": 1620
    },
    {
      "epoch": 0.046706621333487006,
      "grad_norm": 2.661344528198242,
      "learning_rate": 9.992702740706445e-06,
      "loss": 1.0075,
      "step": 1621
    },
    {
      "epoch": 0.04673543479513629,
      "grad_norm": 2.771129846572876,
      "learning_rate": 9.992677518676377e-06,
      "loss": 1.0273,
      "step": 1622
    },
    {
      "epoch": 0.04676424825678557,
      "grad_norm": 2.7508716583251953,
      "learning_rate": 9.992652253165079e-06,
      "loss": 1.0228,
      "step": 1623
    },
    {
      "epoch": 0.04679306171843485,
      "grad_norm": 2.6863560676574707,
      "learning_rate": 9.99262694417277e-06,
      "loss": 1.0833,
      "step": 1624
    },
    {
      "epoch": 0.046821875180084134,
      "grad_norm": 3.0295162200927734,
      "learning_rate": 9.99260159169967e-06,
      "loss": 1.0671,
      "step": 1625
    },
    {
      "epoch": 0.04685068864173342,
      "grad_norm": 2.3764467239379883,
      "learning_rate": 9.992576195746003e-06,
      "loss": 1.0355,
      "step": 1626
    },
    {
      "epoch": 0.0468795021033827,
      "grad_norm": 2.814473867416382,
      "learning_rate": 9.992550756311988e-06,
      "loss": 1.0362,
      "step": 1627
    },
    {
      "epoch": 0.046908315565031986,
      "grad_norm": 2.8993237018585205,
      "learning_rate": 9.992525273397846e-06,
      "loss": 1.0375,
      "step": 1628
    },
    {
      "epoch": 0.04693712902668126,
      "grad_norm": 2.836378574371338,
      "learning_rate": 9.9924997470038e-06,
      "loss": 1.0751,
      "step": 1629
    },
    {
      "epoch": 0.046965942488330546,
      "grad_norm": 2.7581422328948975,
      "learning_rate": 9.99247417713007e-06,
      "loss": 1.0121,
      "step": 1630
    },
    {
      "epoch": 0.04699475594997983,
      "grad_norm": 2.550731658935547,
      "learning_rate": 9.992448563776882e-06,
      "loss": 1.0646,
      "step": 1631
    },
    {
      "epoch": 0.047023569411629114,
      "grad_norm": 2.7411341667175293,
      "learning_rate": 9.992422906944458e-06,
      "loss": 1.0321,
      "step": 1632
    },
    {
      "epoch": 0.0470523828732784,
      "grad_norm": 2.8403496742248535,
      "learning_rate": 9.99239720663302e-06,
      "loss": 1.071,
      "step": 1633
    },
    {
      "epoch": 0.04708119633492768,
      "grad_norm": 2.744438886642456,
      "learning_rate": 9.992371462842794e-06,
      "loss": 1.0546,
      "step": 1634
    },
    {
      "epoch": 0.04711000979657696,
      "grad_norm": 3.0573337078094482,
      "learning_rate": 9.992345675574003e-06,
      "loss": 1.0157,
      "step": 1635
    },
    {
      "epoch": 0.04713882325822624,
      "grad_norm": 2.4849069118499756,
      "learning_rate": 9.99231984482687e-06,
      "loss": 1.0184,
      "step": 1636
    },
    {
      "epoch": 0.047167636719875526,
      "grad_norm": 2.6489756107330322,
      "learning_rate": 9.992293970601624e-06,
      "loss": 1.0193,
      "step": 1637
    },
    {
      "epoch": 0.04719645018152481,
      "grad_norm": 2.981247901916504,
      "learning_rate": 9.992268052898487e-06,
      "loss": 1.0535,
      "step": 1638
    },
    {
      "epoch": 0.047225263643174094,
      "grad_norm": 2.799539804458618,
      "learning_rate": 9.992242091717685e-06,
      "loss": 1.0531,
      "step": 1639
    },
    {
      "epoch": 0.04725407710482337,
      "grad_norm": 2.7974610328674316,
      "learning_rate": 9.992216087059445e-06,
      "loss": 1.0692,
      "step": 1640
    },
    {
      "epoch": 0.047282890566472655,
      "grad_norm": 2.7153921127319336,
      "learning_rate": 9.992190038923995e-06,
      "loss": 1.0609,
      "step": 1641
    },
    {
      "epoch": 0.04731170402812194,
      "grad_norm": 2.9514458179473877,
      "learning_rate": 9.992163947311558e-06,
      "loss": 1.0349,
      "step": 1642
    },
    {
      "epoch": 0.04734051748977122,
      "grad_norm": 2.767427682876587,
      "learning_rate": 9.992137812222365e-06,
      "loss": 1.0608,
      "step": 1643
    },
    {
      "epoch": 0.047369330951420506,
      "grad_norm": 2.966614246368408,
      "learning_rate": 9.992111633656641e-06,
      "loss": 1.02,
      "step": 1644
    },
    {
      "epoch": 0.04739814441306978,
      "grad_norm": 2.736679792404175,
      "learning_rate": 9.992085411614613e-06,
      "loss": 1.0072,
      "step": 1645
    },
    {
      "epoch": 0.04742695787471907,
      "grad_norm": 2.913037061691284,
      "learning_rate": 9.992059146096514e-06,
      "loss": 1.0375,
      "step": 1646
    },
    {
      "epoch": 0.04745577133636835,
      "grad_norm": 3.2779738903045654,
      "learning_rate": 9.99203283710257e-06,
      "loss": 1.0607,
      "step": 1647
    },
    {
      "epoch": 0.047484584798017634,
      "grad_norm": 2.564455270767212,
      "learning_rate": 9.99200648463301e-06,
      "loss": 1.085,
      "step": 1648
    },
    {
      "epoch": 0.04751339825966692,
      "grad_norm": 2.735349178314209,
      "learning_rate": 9.991980088688062e-06,
      "loss": 1.0422,
      "step": 1649
    },
    {
      "epoch": 0.0475422117213162,
      "grad_norm": 2.838698625564575,
      "learning_rate": 9.99195364926796e-06,
      "loss": 1.0491,
      "step": 1650
    },
    {
      "epoch": 0.04757102518296548,
      "grad_norm": 2.4830455780029297,
      "learning_rate": 9.99192716637293e-06,
      "loss": 1.0199,
      "step": 1651
    },
    {
      "epoch": 0.04759983864461476,
      "grad_norm": 3.0584633350372314,
      "learning_rate": 9.991900640003206e-06,
      "loss": 1.0468,
      "step": 1652
    },
    {
      "epoch": 0.04762865210626405,
      "grad_norm": 2.5890135765075684,
      "learning_rate": 9.991874070159016e-06,
      "loss": 1.0182,
      "step": 1653
    },
    {
      "epoch": 0.04765746556791333,
      "grad_norm": 2.7799055576324463,
      "learning_rate": 9.991847456840595e-06,
      "loss": 1.0587,
      "step": 1654
    },
    {
      "epoch": 0.047686279029562614,
      "grad_norm": 2.777207374572754,
      "learning_rate": 9.991820800048171e-06,
      "loss": 1.0617,
      "step": 1655
    },
    {
      "epoch": 0.04771509249121189,
      "grad_norm": 2.6015868186950684,
      "learning_rate": 9.991794099781977e-06,
      "loss": 0.9879,
      "step": 1656
    },
    {
      "epoch": 0.047743905952861175,
      "grad_norm": 2.7081310749053955,
      "learning_rate": 9.991767356042249e-06,
      "loss": 1.0137,
      "step": 1657
    },
    {
      "epoch": 0.04777271941451046,
      "grad_norm": 2.8662221431732178,
      "learning_rate": 9.991740568829215e-06,
      "loss": 1.0517,
      "step": 1658
    },
    {
      "epoch": 0.04780153287615974,
      "grad_norm": 2.5785627365112305,
      "learning_rate": 9.991713738143111e-06,
      "loss": 1.0199,
      "step": 1659
    },
    {
      "epoch": 0.04783034633780903,
      "grad_norm": 2.8682775497436523,
      "learning_rate": 9.991686863984172e-06,
      "loss": 1.0286,
      "step": 1660
    },
    {
      "epoch": 0.0478591597994583,
      "grad_norm": 2.534388542175293,
      "learning_rate": 9.991659946352628e-06,
      "loss": 1.0184,
      "step": 1661
    },
    {
      "epoch": 0.04788797326110759,
      "grad_norm": 2.7196083068847656,
      "learning_rate": 9.991632985248716e-06,
      "loss": 1.0405,
      "step": 1662
    },
    {
      "epoch": 0.04791678672275687,
      "grad_norm": 2.1976749897003174,
      "learning_rate": 9.991605980672672e-06,
      "loss": 0.9988,
      "step": 1663
    },
    {
      "epoch": 0.047945600184406155,
      "grad_norm": 2.7312963008880615,
      "learning_rate": 9.991578932624728e-06,
      "loss": 1.0075,
      "step": 1664
    },
    {
      "epoch": 0.04797441364605544,
      "grad_norm": 2.4017996788024902,
      "learning_rate": 9.991551841105123e-06,
      "loss": 1.0291,
      "step": 1665
    },
    {
      "epoch": 0.04800322710770472,
      "grad_norm": 2.4682018756866455,
      "learning_rate": 9.99152470611409e-06,
      "loss": 1.0416,
      "step": 1666
    },
    {
      "epoch": 0.048032040569354,
      "grad_norm": 2.5167624950408936,
      "learning_rate": 9.991497527651866e-06,
      "loss": 1.0459,
      "step": 1667
    },
    {
      "epoch": 0.04806085403100328,
      "grad_norm": 2.4293341636657715,
      "learning_rate": 9.991470305718689e-06,
      "loss": 1.0521,
      "step": 1668
    },
    {
      "epoch": 0.04808966749265257,
      "grad_norm": 2.6052374839782715,
      "learning_rate": 9.991443040314794e-06,
      "loss": 1.039,
      "step": 1669
    },
    {
      "epoch": 0.04811848095430185,
      "grad_norm": 2.53840970993042,
      "learning_rate": 9.991415731440421e-06,
      "loss": 1.0293,
      "step": 1670
    },
    {
      "epoch": 0.048147294415951135,
      "grad_norm": 2.531217575073242,
      "learning_rate": 9.991388379095807e-06,
      "loss": 1.0229,
      "step": 1671
    },
    {
      "epoch": 0.04817610787760041,
      "grad_norm": 2.6909236907958984,
      "learning_rate": 9.99136098328119e-06,
      "loss": 1.0147,
      "step": 1672
    },
    {
      "epoch": 0.048204921339249696,
      "grad_norm": 2.4365322589874268,
      "learning_rate": 9.991333543996808e-06,
      "loss": 1.0201,
      "step": 1673
    },
    {
      "epoch": 0.04823373480089898,
      "grad_norm": 2.89910888671875,
      "learning_rate": 9.9913060612429e-06,
      "loss": 1.0354,
      "step": 1674
    },
    {
      "epoch": 0.04826254826254826,
      "grad_norm": 2.5222694873809814,
      "learning_rate": 9.991278535019705e-06,
      "loss": 1.0518,
      "step": 1675
    },
    {
      "epoch": 0.04829136172419755,
      "grad_norm": 2.7801003456115723,
      "learning_rate": 9.991250965327463e-06,
      "loss": 1.0348,
      "step": 1676
    },
    {
      "epoch": 0.04832017518584683,
      "grad_norm": 2.550227642059326,
      "learning_rate": 9.991223352166418e-06,
      "loss": 1.0175,
      "step": 1677
    },
    {
      "epoch": 0.04834898864749611,
      "grad_norm": 2.558173418045044,
      "learning_rate": 9.991195695536803e-06,
      "loss": 1.047,
      "step": 1678
    },
    {
      "epoch": 0.04837780210914539,
      "grad_norm": 2.8815042972564697,
      "learning_rate": 9.991167995438864e-06,
      "loss": 1.0369,
      "step": 1679
    },
    {
      "epoch": 0.048406615570794675,
      "grad_norm": 2.6430060863494873,
      "learning_rate": 9.991140251872842e-06,
      "loss": 1.0405,
      "step": 1680
    },
    {
      "epoch": 0.04843542903244396,
      "grad_norm": 3.032721996307373,
      "learning_rate": 9.991112464838977e-06,
      "loss": 1.0218,
      "step": 1681
    },
    {
      "epoch": 0.04846424249409324,
      "grad_norm": 2.4466400146484375,
      "learning_rate": 9.991084634337512e-06,
      "loss": 1.0364,
      "step": 1682
    },
    {
      "epoch": 0.04849305595574252,
      "grad_norm": 2.854180335998535,
      "learning_rate": 9.99105676036869e-06,
      "loss": 1.0627,
      "step": 1683
    },
    {
      "epoch": 0.048521869417391804,
      "grad_norm": 2.4971392154693604,
      "learning_rate": 9.991028842932752e-06,
      "loss": 0.9931,
      "step": 1684
    },
    {
      "epoch": 0.04855068287904109,
      "grad_norm": 2.599733829498291,
      "learning_rate": 9.991000882029943e-06,
      "loss": 1.0168,
      "step": 1685
    },
    {
      "epoch": 0.04857949634069037,
      "grad_norm": 2.9411633014678955,
      "learning_rate": 9.990972877660503e-06,
      "loss": 1.0417,
      "step": 1686
    },
    {
      "epoch": 0.048608309802339655,
      "grad_norm": 2.8755099773406982,
      "learning_rate": 9.99094482982468e-06,
      "loss": 1.0297,
      "step": 1687
    },
    {
      "epoch": 0.04863712326398893,
      "grad_norm": 2.7993204593658447,
      "learning_rate": 9.990916738522716e-06,
      "loss": 1.0372,
      "step": 1688
    },
    {
      "epoch": 0.048665936725638216,
      "grad_norm": 2.915259838104248,
      "learning_rate": 9.990888603754858e-06,
      "loss": 0.9948,
      "step": 1689
    },
    {
      "epoch": 0.0486947501872875,
      "grad_norm": 2.651257038116455,
      "learning_rate": 9.990860425521346e-06,
      "loss": 1.0572,
      "step": 1690
    },
    {
      "epoch": 0.048723563648936784,
      "grad_norm": 3.0263760089874268,
      "learning_rate": 9.990832203822431e-06,
      "loss": 1.0225,
      "step": 1691
    },
    {
      "epoch": 0.04875237711058607,
      "grad_norm": 2.7327797412872314,
      "learning_rate": 9.990803938658357e-06,
      "loss": 1.0378,
      "step": 1692
    },
    {
      "epoch": 0.04878119057223535,
      "grad_norm": 2.8469738960266113,
      "learning_rate": 9.990775630029368e-06,
      "loss": 1.0397,
      "step": 1693
    },
    {
      "epoch": 0.04881000403388463,
      "grad_norm": 2.8074350357055664,
      "learning_rate": 9.990747277935714e-06,
      "loss": 1.056,
      "step": 1694
    },
    {
      "epoch": 0.04883881749553391,
      "grad_norm": 2.470806360244751,
      "learning_rate": 9.99071888237764e-06,
      "loss": 1.0039,
      "step": 1695
    },
    {
      "epoch": 0.048867630957183196,
      "grad_norm": 2.871338367462158,
      "learning_rate": 9.990690443355391e-06,
      "loss": 1.0686,
      "step": 1696
    },
    {
      "epoch": 0.04889644441883248,
      "grad_norm": 2.3602352142333984,
      "learning_rate": 9.990661960869221e-06,
      "loss": 1.0293,
      "step": 1697
    },
    {
      "epoch": 0.048925257880481764,
      "grad_norm": 2.475019931793213,
      "learning_rate": 9.990633434919371e-06,
      "loss": 1.0463,
      "step": 1698
    },
    {
      "epoch": 0.04895407134213104,
      "grad_norm": 2.595524787902832,
      "learning_rate": 9.990604865506094e-06,
      "loss": 1.0296,
      "step": 1699
    },
    {
      "epoch": 0.048982884803780324,
      "grad_norm": 2.7195701599121094,
      "learning_rate": 9.990576252629636e-06,
      "loss": 1.0219,
      "step": 1700
    },
    {
      "epoch": 0.04901169826542961,
      "grad_norm": 2.5852925777435303,
      "learning_rate": 9.99054759629025e-06,
      "loss": 1.0523,
      "step": 1701
    },
    {
      "epoch": 0.04904051172707889,
      "grad_norm": 2.6711537837982178,
      "learning_rate": 9.99051889648818e-06,
      "loss": 1.0338,
      "step": 1702
    },
    {
      "epoch": 0.049069325188728176,
      "grad_norm": 2.457092523574829,
      "learning_rate": 9.990490153223682e-06,
      "loss": 1.0104,
      "step": 1703
    },
    {
      "epoch": 0.04909813865037746,
      "grad_norm": 2.6353509426116943,
      "learning_rate": 9.990461366497003e-06,
      "loss": 1.0347,
      "step": 1704
    },
    {
      "epoch": 0.04912695211202674,
      "grad_norm": 2.6156997680664062,
      "learning_rate": 9.990432536308392e-06,
      "loss": 1.0365,
      "step": 1705
    },
    {
      "epoch": 0.04915576557367602,
      "grad_norm": 2.7292392253875732,
      "learning_rate": 9.990403662658103e-06,
      "loss": 1.0588,
      "step": 1706
    },
    {
      "epoch": 0.049184579035325304,
      "grad_norm": 2.595937728881836,
      "learning_rate": 9.990374745546387e-06,
      "loss": 1.0412,
      "step": 1707
    },
    {
      "epoch": 0.04921339249697459,
      "grad_norm": 2.7139973640441895,
      "learning_rate": 9.990345784973496e-06,
      "loss": 1.0125,
      "step": 1708
    },
    {
      "epoch": 0.04924220595862387,
      "grad_norm": 2.5526957511901855,
      "learning_rate": 9.990316780939682e-06,
      "loss": 1.0364,
      "step": 1709
    },
    {
      "epoch": 0.04927101942027315,
      "grad_norm": 2.7452211380004883,
      "learning_rate": 9.990287733445198e-06,
      "loss": 1.0527,
      "step": 1710
    },
    {
      "epoch": 0.04929983288192243,
      "grad_norm": 2.483111619949341,
      "learning_rate": 9.990258642490294e-06,
      "loss": 1.0314,
      "step": 1711
    },
    {
      "epoch": 0.049328646343571717,
      "grad_norm": 2.58205509185791,
      "learning_rate": 9.990229508075226e-06,
      "loss": 1.0203,
      "step": 1712
    },
    {
      "epoch": 0.049357459805221,
      "grad_norm": 2.804372787475586,
      "learning_rate": 9.99020033020025e-06,
      "loss": 1.0465,
      "step": 1713
    },
    {
      "epoch": 0.049386273266870284,
      "grad_norm": 2.587747812271118,
      "learning_rate": 9.990171108865615e-06,
      "loss": 1.0204,
      "step": 1714
    },
    {
      "epoch": 0.04941508672851956,
      "grad_norm": 3.013373613357544,
      "learning_rate": 9.99014184407158e-06,
      "loss": 1.0591,
      "step": 1715
    },
    {
      "epoch": 0.049443900190168845,
      "grad_norm": 2.9191648960113525,
      "learning_rate": 9.990112535818394e-06,
      "loss": 1.0126,
      "step": 1716
    },
    {
      "epoch": 0.04947271365181813,
      "grad_norm": 2.8643176555633545,
      "learning_rate": 9.99008318410632e-06,
      "loss": 1.0175,
      "step": 1717
    },
    {
      "epoch": 0.04950152711346741,
      "grad_norm": 2.6977388858795166,
      "learning_rate": 9.990053788935609e-06,
      "loss": 1.0326,
      "step": 1718
    },
    {
      "epoch": 0.049530340575116696,
      "grad_norm": 2.9671339988708496,
      "learning_rate": 9.990024350306516e-06,
      "loss": 1.0328,
      "step": 1719
    },
    {
      "epoch": 0.04955915403676598,
      "grad_norm": 2.4672818183898926,
      "learning_rate": 9.9899948682193e-06,
      "loss": 1.0425,
      "step": 1720
    },
    {
      "epoch": 0.04958796749841526,
      "grad_norm": 2.876433849334717,
      "learning_rate": 9.989965342674218e-06,
      "loss": 1.0041,
      "step": 1721
    },
    {
      "epoch": 0.04961678096006454,
      "grad_norm": 2.8163485527038574,
      "learning_rate": 9.989935773671525e-06,
      "loss": 1.042,
      "step": 1722
    },
    {
      "epoch": 0.049645594421713825,
      "grad_norm": 2.4984896183013916,
      "learning_rate": 9.98990616121148e-06,
      "loss": 1.0562,
      "step": 1723
    },
    {
      "epoch": 0.04967440788336311,
      "grad_norm": 2.6103298664093018,
      "learning_rate": 9.98987650529434e-06,
      "loss": 1.0469,
      "step": 1724
    },
    {
      "epoch": 0.04970322134501239,
      "grad_norm": 2.4902961254119873,
      "learning_rate": 9.989846805920363e-06,
      "loss": 0.9974,
      "step": 1725
    },
    {
      "epoch": 0.04973203480666167,
      "grad_norm": 2.707669973373413,
      "learning_rate": 9.98981706308981e-06,
      "loss": 1.0298,
      "step": 1726
    },
    {
      "epoch": 0.04976084826831095,
      "grad_norm": 2.497253894805908,
      "learning_rate": 9.989787276802937e-06,
      "loss": 1.0311,
      "step": 1727
    },
    {
      "epoch": 0.04978966172996024,
      "grad_norm": 2.47780704498291,
      "learning_rate": 9.989757447060005e-06,
      "loss": 1.0168,
      "step": 1728
    },
    {
      "epoch": 0.04981847519160952,
      "grad_norm": 2.4259190559387207,
      "learning_rate": 9.989727573861273e-06,
      "loss": 1.0153,
      "step": 1729
    },
    {
      "epoch": 0.049847288653258805,
      "grad_norm": 2.706632137298584,
      "learning_rate": 9.989697657207003e-06,
      "loss": 1.014,
      "step": 1730
    },
    {
      "epoch": 0.04987610211490808,
      "grad_norm": 2.5560498237609863,
      "learning_rate": 9.989667697097453e-06,
      "loss": 1.0264,
      "step": 1731
    },
    {
      "epoch": 0.049904915576557365,
      "grad_norm": 2.4827637672424316,
      "learning_rate": 9.989637693532887e-06,
      "loss": 0.9985,
      "step": 1732
    },
    {
      "epoch": 0.04993372903820665,
      "grad_norm": 2.9763269424438477,
      "learning_rate": 9.989607646513565e-06,
      "loss": 1.03,
      "step": 1733
    },
    {
      "epoch": 0.04996254249985593,
      "grad_norm": 2.472557544708252,
      "learning_rate": 9.989577556039744e-06,
      "loss": 1.0244,
      "step": 1734
    },
    {
      "epoch": 0.04999135596150522,
      "grad_norm": 2.9955315589904785,
      "learning_rate": 9.989547422111693e-06,
      "loss": 1.066,
      "step": 1735
    },
    {
      "epoch": 0.0500201694231545,
      "grad_norm": 2.5923867225646973,
      "learning_rate": 9.989517244729672e-06,
      "loss": 1.0038,
      "step": 1736
    },
    {
      "epoch": 0.05004898288480378,
      "grad_norm": 2.9606881141662598,
      "learning_rate": 9.989487023893943e-06,
      "loss": 1.0345,
      "step": 1737
    },
    {
      "epoch": 0.05007779634645306,
      "grad_norm": 2.6850967407226562,
      "learning_rate": 9.989456759604769e-06,
      "loss": 1.0483,
      "step": 1738
    },
    {
      "epoch": 0.050106609808102345,
      "grad_norm": 2.9646151065826416,
      "learning_rate": 9.989426451862417e-06,
      "loss": 1.016,
      "step": 1739
    },
    {
      "epoch": 0.05013542326975163,
      "grad_norm": 2.60215163230896,
      "learning_rate": 9.989396100667145e-06,
      "loss": 1.0292,
      "step": 1740
    },
    {
      "epoch": 0.05016423673140091,
      "grad_norm": 3.0310537815093994,
      "learning_rate": 9.989365706019223e-06,
      "loss": 1.0267,
      "step": 1741
    },
    {
      "epoch": 0.05019305019305019,
      "grad_norm": 3.012211322784424,
      "learning_rate": 9.989335267918912e-06,
      "loss": 1.0617,
      "step": 1742
    },
    {
      "epoch": 0.050221863654699474,
      "grad_norm": 2.774832248687744,
      "learning_rate": 9.989304786366478e-06,
      "loss": 1.0262,
      "step": 1743
    },
    {
      "epoch": 0.05025067711634876,
      "grad_norm": 2.6855809688568115,
      "learning_rate": 9.989274261362187e-06,
      "loss": 1.0577,
      "step": 1744
    },
    {
      "epoch": 0.05027949057799804,
      "grad_norm": 2.585996150970459,
      "learning_rate": 9.989243692906305e-06,
      "loss": 1.036,
      "step": 1745
    },
    {
      "epoch": 0.050308304039647325,
      "grad_norm": 2.7178852558135986,
      "learning_rate": 9.989213080999097e-06,
      "loss": 1.0303,
      "step": 1746
    },
    {
      "epoch": 0.05033711750129661,
      "grad_norm": 2.7635412216186523,
      "learning_rate": 9.98918242564083e-06,
      "loss": 1.0558,
      "step": 1747
    },
    {
      "epoch": 0.050365930962945886,
      "grad_norm": 2.6424195766448975,
      "learning_rate": 9.989151726831774e-06,
      "loss": 1.0412,
      "step": 1748
    },
    {
      "epoch": 0.05039474442459517,
      "grad_norm": 2.743082284927368,
      "learning_rate": 9.98912098457219e-06,
      "loss": 1.0267,
      "step": 1749
    },
    {
      "epoch": 0.050423557886244454,
      "grad_norm": 2.8123769760131836,
      "learning_rate": 9.989090198862353e-06,
      "loss": 1.0691,
      "step": 1750
    },
    {
      "epoch": 0.05045237134789374,
      "grad_norm": 2.6771621704101562,
      "learning_rate": 9.989059369702528e-06,
      "loss": 1.0199,
      "step": 1751
    },
    {
      "epoch": 0.05048118480954302,
      "grad_norm": 2.732647657394409,
      "learning_rate": 9.989028497092981e-06,
      "loss": 1.0347,
      "step": 1752
    },
    {
      "epoch": 0.0505099982711923,
      "grad_norm": 2.375073194503784,
      "learning_rate": 9.988997581033984e-06,
      "loss": 1.0176,
      "step": 1753
    },
    {
      "epoch": 0.05053881173284158,
      "grad_norm": 2.6839847564697266,
      "learning_rate": 9.988966621525805e-06,
      "loss": 1.0196,
      "step": 1754
    },
    {
      "epoch": 0.050567625194490866,
      "grad_norm": 2.5754401683807373,
      "learning_rate": 9.988935618568713e-06,
      "loss": 1.023,
      "step": 1755
    },
    {
      "epoch": 0.05059643865614015,
      "grad_norm": 2.7228729724884033,
      "learning_rate": 9.988904572162978e-06,
      "loss": 1.0242,
      "step": 1756
    },
    {
      "epoch": 0.050625252117789434,
      "grad_norm": 2.5826878547668457,
      "learning_rate": 9.988873482308874e-06,
      "loss": 1.0229,
      "step": 1757
    },
    {
      "epoch": 0.05065406557943871,
      "grad_norm": 2.711484670639038,
      "learning_rate": 9.988842349006666e-06,
      "loss": 1.0334,
      "step": 1758
    },
    {
      "epoch": 0.050682879041087994,
      "grad_norm": 2.7648630142211914,
      "learning_rate": 9.988811172256632e-06,
      "loss": 1.0136,
      "step": 1759
    },
    {
      "epoch": 0.05071169250273728,
      "grad_norm": 2.656822681427002,
      "learning_rate": 9.988779952059037e-06,
      "loss": 0.9924,
      "step": 1760
    },
    {
      "epoch": 0.05074050596438656,
      "grad_norm": 2.5658469200134277,
      "learning_rate": 9.988748688414155e-06,
      "loss": 1.0607,
      "step": 1761
    },
    {
      "epoch": 0.050769319426035846,
      "grad_norm": 2.8696398735046387,
      "learning_rate": 9.988717381322262e-06,
      "loss": 1.0309,
      "step": 1762
    },
    {
      "epoch": 0.05079813288768513,
      "grad_norm": 2.4552838802337646,
      "learning_rate": 9.988686030783625e-06,
      "loss": 0.9807,
      "step": 1763
    },
    {
      "epoch": 0.050826946349334406,
      "grad_norm": 3.1933043003082275,
      "learning_rate": 9.98865463679852e-06,
      "loss": 1.0254,
      "step": 1764
    },
    {
      "epoch": 0.05085575981098369,
      "grad_norm": 2.7720253467559814,
      "learning_rate": 9.98862319936722e-06,
      "loss": 1.0152,
      "step": 1765
    },
    {
      "epoch": 0.050884573272632974,
      "grad_norm": 2.951965093612671,
      "learning_rate": 9.988591718489999e-06,
      "loss": 1.0403,
      "step": 1766
    },
    {
      "epoch": 0.05091338673428226,
      "grad_norm": 2.5501606464385986,
      "learning_rate": 9.98856019416713e-06,
      "loss": 1.0315,
      "step": 1767
    },
    {
      "epoch": 0.05094220019593154,
      "grad_norm": 2.56386661529541,
      "learning_rate": 9.988528626398889e-06,
      "loss": 1.0141,
      "step": 1768
    },
    {
      "epoch": 0.05097101365758082,
      "grad_norm": 2.6208784580230713,
      "learning_rate": 9.98849701518555e-06,
      "loss": 1.0291,
      "step": 1769
    },
    {
      "epoch": 0.0509998271192301,
      "grad_norm": 2.6915714740753174,
      "learning_rate": 9.988465360527389e-06,
      "loss": 1.0326,
      "step": 1770
    },
    {
      "epoch": 0.051028640580879386,
      "grad_norm": 2.7161920070648193,
      "learning_rate": 9.988433662424682e-06,
      "loss": 1.025,
      "step": 1771
    },
    {
      "epoch": 0.05105745404252867,
      "grad_norm": 2.79437518119812,
      "learning_rate": 9.988401920877703e-06,
      "loss": 1.0125,
      "step": 1772
    },
    {
      "epoch": 0.051086267504177954,
      "grad_norm": 2.7883620262145996,
      "learning_rate": 9.988370135886731e-06,
      "loss": 1.0062,
      "step": 1773
    },
    {
      "epoch": 0.05111508096582724,
      "grad_norm": 2.9108312129974365,
      "learning_rate": 9.988338307452043e-06,
      "loss": 1.0227,
      "step": 1774
    },
    {
      "epoch": 0.051143894427476515,
      "grad_norm": 2.6112606525421143,
      "learning_rate": 9.988306435573912e-06,
      "loss": 1.0014,
      "step": 1775
    },
    {
      "epoch": 0.0511727078891258,
      "grad_norm": 2.9659273624420166,
      "learning_rate": 9.98827452025262e-06,
      "loss": 1.036,
      "step": 1776
    },
    {
      "epoch": 0.05120152135077508,
      "grad_norm": 2.802523136138916,
      "learning_rate": 9.988242561488444e-06,
      "loss": 1.0928,
      "step": 1777
    },
    {
      "epoch": 0.051230334812424366,
      "grad_norm": 2.601478338241577,
      "learning_rate": 9.98821055928166e-06,
      "loss": 1.0386,
      "step": 1778
    },
    {
      "epoch": 0.05125914827407365,
      "grad_norm": 2.772334098815918,
      "learning_rate": 9.98817851363255e-06,
      "loss": 1.0394,
      "step": 1779
    },
    {
      "epoch": 0.05128796173572293,
      "grad_norm": 2.6272196769714355,
      "learning_rate": 9.98814642454139e-06,
      "loss": 1.0384,
      "step": 1780
    },
    {
      "epoch": 0.05131677519737221,
      "grad_norm": 2.636232614517212,
      "learning_rate": 9.988114292008461e-06,
      "loss": 1.0242,
      "step": 1781
    },
    {
      "epoch": 0.051345588659021495,
      "grad_norm": 2.569330930709839,
      "learning_rate": 9.988082116034043e-06,
      "loss": 1.0237,
      "step": 1782
    },
    {
      "epoch": 0.05137440212067078,
      "grad_norm": 2.4401297569274902,
      "learning_rate": 9.988049896618418e-06,
      "loss": 1.0345,
      "step": 1783
    },
    {
      "epoch": 0.05140321558232006,
      "grad_norm": 2.5976009368896484,
      "learning_rate": 9.988017633761862e-06,
      "loss": 1.038,
      "step": 1784
    },
    {
      "epoch": 0.05143202904396934,
      "grad_norm": 2.4600656032562256,
      "learning_rate": 9.987985327464659e-06,
      "loss": 1.043,
      "step": 1785
    },
    {
      "epoch": 0.05146084250561862,
      "grad_norm": 2.7738895416259766,
      "learning_rate": 9.98795297772709e-06,
      "loss": 1.0163,
      "step": 1786
    },
    {
      "epoch": 0.05148965596726791,
      "grad_norm": 2.788835287094116,
      "learning_rate": 9.987920584549439e-06,
      "loss": 1.0143,
      "step": 1787
    },
    {
      "epoch": 0.05151846942891719,
      "grad_norm": 2.726916551589966,
      "learning_rate": 9.987888147931984e-06,
      "loss": 1.0216,
      "step": 1788
    },
    {
      "epoch": 0.051547282890566475,
      "grad_norm": 2.637521505355835,
      "learning_rate": 9.98785566787501e-06,
      "loss": 1.0343,
      "step": 1789
    },
    {
      "epoch": 0.05157609635221576,
      "grad_norm": 2.6394975185394287,
      "learning_rate": 9.9878231443788e-06,
      "loss": 0.9736,
      "step": 1790
    },
    {
      "epoch": 0.051604909813865035,
      "grad_norm": 2.782914400100708,
      "learning_rate": 9.987790577443633e-06,
      "loss": 1.0282,
      "step": 1791
    },
    {
      "epoch": 0.05163372327551432,
      "grad_norm": 2.6907076835632324,
      "learning_rate": 9.987757967069798e-06,
      "loss": 1.0008,
      "step": 1792
    },
    {
      "epoch": 0.0516625367371636,
      "grad_norm": 2.8687007427215576,
      "learning_rate": 9.987725313257578e-06,
      "loss": 1.0157,
      "step": 1793
    },
    {
      "epoch": 0.05169135019881289,
      "grad_norm": 2.7669668197631836,
      "learning_rate": 9.987692616007254e-06,
      "loss": 1.0321,
      "step": 1794
    },
    {
      "epoch": 0.05172016366046217,
      "grad_norm": 3.2197396755218506,
      "learning_rate": 9.987659875319115e-06,
      "loss": 1.0174,
      "step": 1795
    },
    {
      "epoch": 0.05174897712211145,
      "grad_norm": 2.7566792964935303,
      "learning_rate": 9.987627091193444e-06,
      "loss": 1.0309,
      "step": 1796
    },
    {
      "epoch": 0.05177779058376073,
      "grad_norm": 2.8133952617645264,
      "learning_rate": 9.987594263630527e-06,
      "loss": 1.0657,
      "step": 1797
    },
    {
      "epoch": 0.051806604045410015,
      "grad_norm": 2.4695985317230225,
      "learning_rate": 9.98756139263065e-06,
      "loss": 1.0348,
      "step": 1798
    },
    {
      "epoch": 0.0518354175070593,
      "grad_norm": 2.8663878440856934,
      "learning_rate": 9.987528478194099e-06,
      "loss": 1.0216,
      "step": 1799
    },
    {
      "epoch": 0.05186423096870858,
      "grad_norm": 2.5828869342803955,
      "learning_rate": 9.98749552032116e-06,
      "loss": 1.0453,
      "step": 1800
    },
    {
      "epoch": 0.05189304443035786,
      "grad_norm": 2.845604658126831,
      "learning_rate": 9.987462519012122e-06,
      "loss": 1.0353,
      "step": 1801
    },
    {
      "epoch": 0.051921857892007144,
      "grad_norm": 2.74312162399292,
      "learning_rate": 9.987429474267269e-06,
      "loss": 1.0728,
      "step": 1802
    },
    {
      "epoch": 0.05195067135365643,
      "grad_norm": 2.939842700958252,
      "learning_rate": 9.987396386086892e-06,
      "loss": 0.993,
      "step": 1803
    },
    {
      "epoch": 0.05197948481530571,
      "grad_norm": 2.6881484985351562,
      "learning_rate": 9.987363254471277e-06,
      "loss": 1.0199,
      "step": 1804
    },
    {
      "epoch": 0.052008298276954995,
      "grad_norm": 2.9867310523986816,
      "learning_rate": 9.987330079420715e-06,
      "loss": 1.048,
      "step": 1805
    },
    {
      "epoch": 0.05203711173860428,
      "grad_norm": 2.6284477710723877,
      "learning_rate": 9.987296860935493e-06,
      "loss": 1.0279,
      "step": 1806
    },
    {
      "epoch": 0.052065925200253556,
      "grad_norm": 2.6140999794006348,
      "learning_rate": 9.987263599015902e-06,
      "loss": 0.9971,
      "step": 1807
    },
    {
      "epoch": 0.05209473866190284,
      "grad_norm": 2.623318672180176,
      "learning_rate": 9.987230293662229e-06,
      "loss": 1.0374,
      "step": 1808
    },
    {
      "epoch": 0.052123552123552123,
      "grad_norm": 2.6777563095092773,
      "learning_rate": 9.987196944874765e-06,
      "loss": 0.9798,
      "step": 1809
    },
    {
      "epoch": 0.05215236558520141,
      "grad_norm": 2.437077760696411,
      "learning_rate": 9.987163552653802e-06,
      "loss": 1.0179,
      "step": 1810
    },
    {
      "epoch": 0.05218117904685069,
      "grad_norm": 2.808772087097168,
      "learning_rate": 9.987130116999631e-06,
      "loss": 1.0513,
      "step": 1811
    },
    {
      "epoch": 0.05220999250849997,
      "grad_norm": 2.620182752609253,
      "learning_rate": 9.987096637912539e-06,
      "loss": 1.0403,
      "step": 1812
    },
    {
      "epoch": 0.05223880597014925,
      "grad_norm": 2.740147829055786,
      "learning_rate": 9.987063115392823e-06,
      "loss": 1.0322,
      "step": 1813
    },
    {
      "epoch": 0.052267619431798536,
      "grad_norm": 2.5296027660369873,
      "learning_rate": 9.987029549440773e-06,
      "loss": 1.0249,
      "step": 1814
    },
    {
      "epoch": 0.05229643289344782,
      "grad_norm": 2.495102882385254,
      "learning_rate": 9.986995940056678e-06,
      "loss": 1.0369,
      "step": 1815
    },
    {
      "epoch": 0.0523252463550971,
      "grad_norm": 2.6822283267974854,
      "learning_rate": 9.986962287240836e-06,
      "loss": 1.0212,
      "step": 1816
    },
    {
      "epoch": 0.05235405981674639,
      "grad_norm": 2.782607316970825,
      "learning_rate": 9.986928590993536e-06,
      "loss": 1.0086,
      "step": 1817
    },
    {
      "epoch": 0.052382873278395664,
      "grad_norm": 2.3587281703948975,
      "learning_rate": 9.986894851315074e-06,
      "loss": 1.0262,
      "step": 1818
    },
    {
      "epoch": 0.05241168674004495,
      "grad_norm": 2.9223554134368896,
      "learning_rate": 9.986861068205742e-06,
      "loss": 1.0362,
      "step": 1819
    },
    {
      "epoch": 0.05244050020169423,
      "grad_norm": 2.492492198944092,
      "learning_rate": 9.986827241665836e-06,
      "loss": 1.0221,
      "step": 1820
    },
    {
      "epoch": 0.052469313663343516,
      "grad_norm": 2.802565574645996,
      "learning_rate": 9.98679337169565e-06,
      "loss": 1.0264,
      "step": 1821
    },
    {
      "epoch": 0.0524981271249928,
      "grad_norm": 2.872837781906128,
      "learning_rate": 9.986759458295479e-06,
      "loss": 1.0097,
      "step": 1822
    },
    {
      "epoch": 0.052526940586642076,
      "grad_norm": 2.571472644805908,
      "learning_rate": 9.986725501465616e-06,
      "loss": 1.0124,
      "step": 1823
    },
    {
      "epoch": 0.05255575404829136,
      "grad_norm": 3.0008609294891357,
      "learning_rate": 9.98669150120636e-06,
      "loss": 1.028,
      "step": 1824
    },
    {
      "epoch": 0.052584567509940644,
      "grad_norm": 2.713791608810425,
      "learning_rate": 9.986657457518007e-06,
      "loss": 1.0029,
      "step": 1825
    },
    {
      "epoch": 0.05261338097158993,
      "grad_norm": 2.7318291664123535,
      "learning_rate": 9.986623370400852e-06,
      "loss": 1.0135,
      "step": 1826
    },
    {
      "epoch": 0.05264219443323921,
      "grad_norm": 2.836693286895752,
      "learning_rate": 9.98658923985519e-06,
      "loss": 1.0286,
      "step": 1827
    },
    {
      "epoch": 0.05267100789488849,
      "grad_norm": 2.5681262016296387,
      "learning_rate": 9.986555065881322e-06,
      "loss": 1.0329,
      "step": 1828
    },
    {
      "epoch": 0.05269982135653777,
      "grad_norm": 2.6142377853393555,
      "learning_rate": 9.986520848479545e-06,
      "loss": 1.0671,
      "step": 1829
    },
    {
      "epoch": 0.052728634818187056,
      "grad_norm": 2.8082635402679443,
      "learning_rate": 9.986486587650154e-06,
      "loss": 1.0092,
      "step": 1830
    },
    {
      "epoch": 0.05275744827983634,
      "grad_norm": 2.600754499435425,
      "learning_rate": 9.986452283393452e-06,
      "loss": 1.0037,
      "step": 1831
    },
    {
      "epoch": 0.052786261741485624,
      "grad_norm": 2.7999677658081055,
      "learning_rate": 9.986417935709734e-06,
      "loss": 1.0193,
      "step": 1832
    },
    {
      "epoch": 0.05281507520313491,
      "grad_norm": 2.8084447383880615,
      "learning_rate": 9.986383544599301e-06,
      "loss": 1.0434,
      "step": 1833
    },
    {
      "epoch": 0.052843888664784185,
      "grad_norm": 2.7134931087493896,
      "learning_rate": 9.98634911006245e-06,
      "loss": 0.9791,
      "step": 1834
    },
    {
      "epoch": 0.05287270212643347,
      "grad_norm": 2.78989315032959,
      "learning_rate": 9.986314632099485e-06,
      "loss": 1.0139,
      "step": 1835
    },
    {
      "epoch": 0.05290151558808275,
      "grad_norm": 2.396843194961548,
      "learning_rate": 9.986280110710703e-06,
      "loss": 1.0146,
      "step": 1836
    },
    {
      "epoch": 0.052930329049732036,
      "grad_norm": 2.5648486614227295,
      "learning_rate": 9.986245545896406e-06,
      "loss": 1.0374,
      "step": 1837
    },
    {
      "epoch": 0.05295914251138132,
      "grad_norm": 2.5936243534088135,
      "learning_rate": 9.986210937656896e-06,
      "loss": 1.0117,
      "step": 1838
    },
    {
      "epoch": 0.0529879559730306,
      "grad_norm": 2.605201482772827,
      "learning_rate": 9.98617628599247e-06,
      "loss": 0.9959,
      "step": 1839
    },
    {
      "epoch": 0.05301676943467988,
      "grad_norm": 2.7282614707946777,
      "learning_rate": 9.986141590903435e-06,
      "loss": 1.0472,
      "step": 1840
    },
    {
      "epoch": 0.053045582896329165,
      "grad_norm": 2.464796543121338,
      "learning_rate": 9.986106852390089e-06,
      "loss": 1.0086,
      "step": 1841
    },
    {
      "epoch": 0.05307439635797845,
      "grad_norm": 2.566462993621826,
      "learning_rate": 9.986072070452739e-06,
      "loss": 1.0183,
      "step": 1842
    },
    {
      "epoch": 0.05310320981962773,
      "grad_norm": 2.5087316036224365,
      "learning_rate": 9.986037245091684e-06,
      "loss": 1.0551,
      "step": 1843
    },
    {
      "epoch": 0.053132023281277016,
      "grad_norm": 2.9656901359558105,
      "learning_rate": 9.98600237630723e-06,
      "loss": 0.9827,
      "step": 1844
    },
    {
      "epoch": 0.05316083674292629,
      "grad_norm": 2.672898054122925,
      "learning_rate": 9.98596746409968e-06,
      "loss": 1.0249,
      "step": 1845
    },
    {
      "epoch": 0.05318965020457558,
      "grad_norm": 2.7441844940185547,
      "learning_rate": 9.985932508469336e-06,
      "loss": 1.0111,
      "step": 1846
    },
    {
      "epoch": 0.05321846366622486,
      "grad_norm": 3.0140860080718994,
      "learning_rate": 9.985897509416504e-06,
      "loss": 1.0537,
      "step": 1847
    },
    {
      "epoch": 0.053247277127874144,
      "grad_norm": 2.4531166553497314,
      "learning_rate": 9.98586246694149e-06,
      "loss": 1.0023,
      "step": 1848
    },
    {
      "epoch": 0.05327609058952343,
      "grad_norm": 2.407742500305176,
      "learning_rate": 9.985827381044597e-06,
      "loss": 1.001,
      "step": 1849
    },
    {
      "epoch": 0.053304904051172705,
      "grad_norm": 2.5772178173065186,
      "learning_rate": 9.985792251726132e-06,
      "loss": 1.0229,
      "step": 1850
    },
    {
      "epoch": 0.05333371751282199,
      "grad_norm": 2.3718421459198,
      "learning_rate": 9.9857570789864e-06,
      "loss": 1.0199,
      "step": 1851
    },
    {
      "epoch": 0.05336253097447127,
      "grad_norm": 2.4462783336639404,
      "learning_rate": 9.985721862825708e-06,
      "loss": 1.0247,
      "step": 1852
    },
    {
      "epoch": 0.05339134443612056,
      "grad_norm": 2.420759439468384,
      "learning_rate": 9.985686603244363e-06,
      "loss": 1.0101,
      "step": 1853
    },
    {
      "epoch": 0.05342015789776984,
      "grad_norm": 2.548231840133667,
      "learning_rate": 9.98565130024267e-06,
      "loss": 1.04,
      "step": 1854
    },
    {
      "epoch": 0.05344897135941912,
      "grad_norm": 2.385467052459717,
      "learning_rate": 9.98561595382094e-06,
      "loss": 1.0147,
      "step": 1855
    },
    {
      "epoch": 0.0534777848210684,
      "grad_norm": 2.5415000915527344,
      "learning_rate": 9.985580563979478e-06,
      "loss": 1.0321,
      "step": 1856
    },
    {
      "epoch": 0.053506598282717685,
      "grad_norm": 2.3478426933288574,
      "learning_rate": 9.985545130718593e-06,
      "loss": 0.9896,
      "step": 1857
    },
    {
      "epoch": 0.05353541174436697,
      "grad_norm": 2.4360780715942383,
      "learning_rate": 9.985509654038592e-06,
      "loss": 1.0268,
      "step": 1858
    },
    {
      "epoch": 0.05356422520601625,
      "grad_norm": 2.5898873805999756,
      "learning_rate": 9.985474133939787e-06,
      "loss": 1.0239,
      "step": 1859
    },
    {
      "epoch": 0.05359303866766554,
      "grad_norm": 2.5975520610809326,
      "learning_rate": 9.985438570422487e-06,
      "loss": 1.0111,
      "step": 1860
    },
    {
      "epoch": 0.05362185212931481,
      "grad_norm": 2.57247257232666,
      "learning_rate": 9.985402963486999e-06,
      "loss": 1.0419,
      "step": 1861
    },
    {
      "epoch": 0.0536506655909641,
      "grad_norm": 2.9753241539001465,
      "learning_rate": 9.985367313133636e-06,
      "loss": 1.0291,
      "step": 1862
    },
    {
      "epoch": 0.05367947905261338,
      "grad_norm": 2.457420825958252,
      "learning_rate": 9.985331619362706e-06,
      "loss": 1.0319,
      "step": 1863
    },
    {
      "epoch": 0.053708292514262665,
      "grad_norm": 2.649797201156616,
      "learning_rate": 9.985295882174522e-06,
      "loss": 1.0408,
      "step": 1864
    },
    {
      "epoch": 0.05373710597591195,
      "grad_norm": 2.8933794498443604,
      "learning_rate": 9.985260101569394e-06,
      "loss": 1.0245,
      "step": 1865
    },
    {
      "epoch": 0.053765919437561226,
      "grad_norm": 2.547960042953491,
      "learning_rate": 9.985224277547634e-06,
      "loss": 0.9843,
      "step": 1866
    },
    {
      "epoch": 0.05379473289921051,
      "grad_norm": 2.9214823246002197,
      "learning_rate": 9.985188410109554e-06,
      "loss": 0.9951,
      "step": 1867
    },
    {
      "epoch": 0.05382354636085979,
      "grad_norm": 2.7500133514404297,
      "learning_rate": 9.985152499255465e-06,
      "loss": 1.0137,
      "step": 1868
    },
    {
      "epoch": 0.05385235982250908,
      "grad_norm": 2.7481300830841064,
      "learning_rate": 9.985116544985685e-06,
      "loss": 0.9984,
      "step": 1869
    },
    {
      "epoch": 0.05388117328415836,
      "grad_norm": 2.7574682235717773,
      "learning_rate": 9.985080547300519e-06,
      "loss": 1.0043,
      "step": 1870
    },
    {
      "epoch": 0.05390998674580764,
      "grad_norm": 2.7820215225219727,
      "learning_rate": 9.985044506200288e-06,
      "loss": 1.0042,
      "step": 1871
    },
    {
      "epoch": 0.05393880020745692,
      "grad_norm": 2.8542637825012207,
      "learning_rate": 9.9850084216853e-06,
      "loss": 1.0029,
      "step": 1872
    },
    {
      "epoch": 0.053967613669106206,
      "grad_norm": 2.896406888961792,
      "learning_rate": 9.984972293755873e-06,
      "loss": 1.0225,
      "step": 1873
    },
    {
      "epoch": 0.05399642713075549,
      "grad_norm": 2.929828643798828,
      "learning_rate": 9.984936122412319e-06,
      "loss": 1.0148,
      "step": 1874
    },
    {
      "epoch": 0.05402524059240477,
      "grad_norm": 2.7081220149993896,
      "learning_rate": 9.984899907654957e-06,
      "loss": 1.0183,
      "step": 1875
    },
    {
      "epoch": 0.05405405405405406,
      "grad_norm": 2.54300856590271,
      "learning_rate": 9.984863649484098e-06,
      "loss": 1.0181,
      "step": 1876
    },
    {
      "epoch": 0.054082867515703334,
      "grad_norm": 2.697946310043335,
      "learning_rate": 9.98482734790006e-06,
      "loss": 1.0213,
      "step": 1877
    },
    {
      "epoch": 0.05411168097735262,
      "grad_norm": 2.7797000408172607,
      "learning_rate": 9.98479100290316e-06,
      "loss": 0.9859,
      "step": 1878
    },
    {
      "epoch": 0.0541404944390019,
      "grad_norm": 2.647155284881592,
      "learning_rate": 9.984754614493713e-06,
      "loss": 1.0186,
      "step": 1879
    },
    {
      "epoch": 0.054169307900651185,
      "grad_norm": 2.7866876125335693,
      "learning_rate": 9.984718182672034e-06,
      "loss": 0.9868,
      "step": 1880
    },
    {
      "epoch": 0.05419812136230047,
      "grad_norm": 2.358351945877075,
      "learning_rate": 9.984681707438444e-06,
      "loss": 1.0344,
      "step": 1881
    },
    {
      "epoch": 0.054226934823949746,
      "grad_norm": 2.5184078216552734,
      "learning_rate": 9.98464518879326e-06,
      "loss": 1.0315,
      "step": 1882
    },
    {
      "epoch": 0.05425574828559903,
      "grad_norm": 2.8066039085388184,
      "learning_rate": 9.984608626736797e-06,
      "loss": 1.0462,
      "step": 1883
    },
    {
      "epoch": 0.054284561747248314,
      "grad_norm": 2.6354050636291504,
      "learning_rate": 9.98457202126938e-06,
      "loss": 1.0052,
      "step": 1884
    },
    {
      "epoch": 0.0543133752088976,
      "grad_norm": 2.618013381958008,
      "learning_rate": 9.984535372391318e-06,
      "loss": 1.0308,
      "step": 1885
    },
    {
      "epoch": 0.05434218867054688,
      "grad_norm": 2.6637580394744873,
      "learning_rate": 9.984498680102938e-06,
      "loss": 1.0039,
      "step": 1886
    },
    {
      "epoch": 0.054371002132196165,
      "grad_norm": 2.7484045028686523,
      "learning_rate": 9.984461944404557e-06,
      "loss": 1.0069,
      "step": 1887
    },
    {
      "epoch": 0.05439981559384544,
      "grad_norm": 2.5412583351135254,
      "learning_rate": 9.984425165296496e-06,
      "loss": 1.0028,
      "step": 1888
    },
    {
      "epoch": 0.054428629055494726,
      "grad_norm": 2.5773167610168457,
      "learning_rate": 9.984388342779075e-06,
      "loss": 1.0223,
      "step": 1889
    },
    {
      "epoch": 0.05445744251714401,
      "grad_norm": 2.6928799152374268,
      "learning_rate": 9.984351476852613e-06,
      "loss": 1.0159,
      "step": 1890
    },
    {
      "epoch": 0.054486255978793294,
      "grad_norm": 2.6817541122436523,
      "learning_rate": 9.984314567517433e-06,
      "loss": 1.0432,
      "step": 1891
    },
    {
      "epoch": 0.05451506944044258,
      "grad_norm": 2.757719039916992,
      "learning_rate": 9.984277614773854e-06,
      "loss": 1.0235,
      "step": 1892
    },
    {
      "epoch": 0.054543882902091854,
      "grad_norm": 2.459590196609497,
      "learning_rate": 9.9842406186222e-06,
      "loss": 1.0109,
      "step": 1893
    },
    {
      "epoch": 0.05457269636374114,
      "grad_norm": 2.6014060974121094,
      "learning_rate": 9.984203579062792e-06,
      "loss": 1.0244,
      "step": 1894
    },
    {
      "epoch": 0.05460150982539042,
      "grad_norm": 2.8437299728393555,
      "learning_rate": 9.984166496095956e-06,
      "loss": 1.0044,
      "step": 1895
    },
    {
      "epoch": 0.054630323287039706,
      "grad_norm": 2.3953051567077637,
      "learning_rate": 9.984129369722008e-06,
      "loss": 1.0075,
      "step": 1896
    },
    {
      "epoch": 0.05465913674868899,
      "grad_norm": 2.516387939453125,
      "learning_rate": 9.984092199941278e-06,
      "loss": 1.0144,
      "step": 1897
    },
    {
      "epoch": 0.05468795021033827,
      "grad_norm": 2.2376155853271484,
      "learning_rate": 9.984054986754089e-06,
      "loss": 1.0073,
      "step": 1898
    },
    {
      "epoch": 0.05471676367198755,
      "grad_norm": 2.436163902282715,
      "learning_rate": 9.98401773016076e-06,
      "loss": 1.0531,
      "step": 1899
    },
    {
      "epoch": 0.054745577133636834,
      "grad_norm": 2.656750202178955,
      "learning_rate": 9.98398043016162e-06,
      "loss": 1.0242,
      "step": 1900
    },
    {
      "epoch": 0.05477439059528612,
      "grad_norm": 2.4208414554595947,
      "learning_rate": 9.983943086756993e-06,
      "loss": 1.0045,
      "step": 1901
    },
    {
      "epoch": 0.0548032040569354,
      "grad_norm": 2.3940165042877197,
      "learning_rate": 9.983905699947202e-06,
      "loss": 1.0308,
      "step": 1902
    },
    {
      "epoch": 0.054832017518584686,
      "grad_norm": 2.487002372741699,
      "learning_rate": 9.983868269732577e-06,
      "loss": 1.0042,
      "step": 1903
    },
    {
      "epoch": 0.05486083098023396,
      "grad_norm": 2.394655704498291,
      "learning_rate": 9.98383079611344e-06,
      "loss": 1.0131,
      "step": 1904
    },
    {
      "epoch": 0.05488964444188325,
      "grad_norm": 2.659029245376587,
      "learning_rate": 9.983793279090119e-06,
      "loss": 1.0114,
      "step": 1905
    },
    {
      "epoch": 0.05491845790353253,
      "grad_norm": 2.379152774810791,
      "learning_rate": 9.98375571866294e-06,
      "loss": 1.0549,
      "step": 1906
    },
    {
      "epoch": 0.054947271365181814,
      "grad_norm": 2.5861077308654785,
      "learning_rate": 9.983718114832231e-06,
      "loss": 1.0192,
      "step": 1907
    },
    {
      "epoch": 0.0549760848268311,
      "grad_norm": 2.2278032302856445,
      "learning_rate": 9.98368046759832e-06,
      "loss": 1.0147,
      "step": 1908
    },
    {
      "epoch": 0.055004898288480375,
      "grad_norm": 2.604095697402954,
      "learning_rate": 9.983642776961534e-06,
      "loss": 1.0226,
      "step": 1909
    },
    {
      "epoch": 0.05503371175012966,
      "grad_norm": 2.85732102394104,
      "learning_rate": 9.9836050429222e-06,
      "loss": 1.021,
      "step": 1910
    },
    {
      "epoch": 0.05506252521177894,
      "grad_norm": 2.7458865642547607,
      "learning_rate": 9.983567265480648e-06,
      "loss": 1.0011,
      "step": 1911
    },
    {
      "epoch": 0.055091338673428226,
      "grad_norm": 2.5213394165039062,
      "learning_rate": 9.983529444637206e-06,
      "loss": 1.0357,
      "step": 1912
    },
    {
      "epoch": 0.05512015213507751,
      "grad_norm": 2.6713740825653076,
      "learning_rate": 9.983491580392206e-06,
      "loss": 0.9948,
      "step": 1913
    },
    {
      "epoch": 0.055148965596726794,
      "grad_norm": 2.689687728881836,
      "learning_rate": 9.983453672745976e-06,
      "loss": 1.0396,
      "step": 1914
    },
    {
      "epoch": 0.05517777905837607,
      "grad_norm": 2.5149478912353516,
      "learning_rate": 9.983415721698845e-06,
      "loss": 1.0332,
      "step": 1915
    },
    {
      "epoch": 0.055206592520025355,
      "grad_norm": 2.501779317855835,
      "learning_rate": 9.983377727251145e-06,
      "loss": 1.0142,
      "step": 1916
    },
    {
      "epoch": 0.05523540598167464,
      "grad_norm": 2.779130220413208,
      "learning_rate": 9.983339689403207e-06,
      "loss": 1.0537,
      "step": 1917
    },
    {
      "epoch": 0.05526421944332392,
      "grad_norm": 2.6378045082092285,
      "learning_rate": 9.983301608155362e-06,
      "loss": 1.0387,
      "step": 1918
    },
    {
      "epoch": 0.055293032904973206,
      "grad_norm": 2.6817495822906494,
      "learning_rate": 9.983263483507941e-06,
      "loss": 1.0238,
      "step": 1919
    },
    {
      "epoch": 0.05532184636662248,
      "grad_norm": 2.504392385482788,
      "learning_rate": 9.983225315461276e-06,
      "loss": 1.0007,
      "step": 1920
    },
    {
      "epoch": 0.05535065982827177,
      "grad_norm": 2.7460713386535645,
      "learning_rate": 9.9831871040157e-06,
      "loss": 1.0369,
      "step": 1921
    },
    {
      "epoch": 0.05537947328992105,
      "grad_norm": 2.29585337638855,
      "learning_rate": 9.983148849171547e-06,
      "loss": 0.9887,
      "step": 1922
    },
    {
      "epoch": 0.055408286751570335,
      "grad_norm": 2.6874003410339355,
      "learning_rate": 9.983110550929148e-06,
      "loss": 0.9975,
      "step": 1923
    },
    {
      "epoch": 0.05543710021321962,
      "grad_norm": 2.636837959289551,
      "learning_rate": 9.983072209288837e-06,
      "loss": 1.0347,
      "step": 1924
    },
    {
      "epoch": 0.055465913674868896,
      "grad_norm": 2.528942823410034,
      "learning_rate": 9.98303382425095e-06,
      "loss": 1.0193,
      "step": 1925
    },
    {
      "epoch": 0.05549472713651818,
      "grad_norm": 2.803952693939209,
      "learning_rate": 9.982995395815818e-06,
      "loss": 1.0532,
      "step": 1926
    },
    {
      "epoch": 0.05552354059816746,
      "grad_norm": 2.5820300579071045,
      "learning_rate": 9.982956923983777e-06,
      "loss": 1.032,
      "step": 1927
    },
    {
      "epoch": 0.05555235405981675,
      "grad_norm": 2.5652506351470947,
      "learning_rate": 9.982918408755164e-06,
      "loss": 1.0401,
      "step": 1928
    },
    {
      "epoch": 0.05558116752146603,
      "grad_norm": 2.598097801208496,
      "learning_rate": 9.982879850130311e-06,
      "loss": 1.0253,
      "step": 1929
    },
    {
      "epoch": 0.055609980983115315,
      "grad_norm": 2.561185359954834,
      "learning_rate": 9.982841248109556e-06,
      "loss": 1.0029,
      "step": 1930
    },
    {
      "epoch": 0.05563879444476459,
      "grad_norm": 2.313646078109741,
      "learning_rate": 9.982802602693235e-06,
      "loss": 1.0272,
      "step": 1931
    },
    {
      "epoch": 0.055667607906413875,
      "grad_norm": 2.6690521240234375,
      "learning_rate": 9.982763913881684e-06,
      "loss": 1.0491,
      "step": 1932
    },
    {
      "epoch": 0.05569642136806316,
      "grad_norm": 2.4023759365081787,
      "learning_rate": 9.982725181675242e-06,
      "loss": 1.0228,
      "step": 1933
    },
    {
      "epoch": 0.05572523482971244,
      "grad_norm": 2.3685526847839355,
      "learning_rate": 9.982686406074243e-06,
      "loss": 1.0201,
      "step": 1934
    },
    {
      "epoch": 0.05575404829136173,
      "grad_norm": 2.5411527156829834,
      "learning_rate": 9.982647587079025e-06,
      "loss": 1.0292,
      "step": 1935
    },
    {
      "epoch": 0.055782861753011004,
      "grad_norm": 2.352614402770996,
      "learning_rate": 9.982608724689928e-06,
      "loss": 0.9709,
      "step": 1936
    },
    {
      "epoch": 0.05581167521466029,
      "grad_norm": 2.429647445678711,
      "learning_rate": 9.982569818907291e-06,
      "loss": 1.026,
      "step": 1937
    },
    {
      "epoch": 0.05584048867630957,
      "grad_norm": 2.2987442016601562,
      "learning_rate": 9.982530869731452e-06,
      "loss": 0.9919,
      "step": 1938
    },
    {
      "epoch": 0.055869302137958855,
      "grad_norm": 2.3833518028259277,
      "learning_rate": 9.98249187716275e-06,
      "loss": 1.0109,
      "step": 1939
    },
    {
      "epoch": 0.05589811559960814,
      "grad_norm": 2.45678973197937,
      "learning_rate": 9.982452841201522e-06,
      "loss": 1.036,
      "step": 1940
    },
    {
      "epoch": 0.055926929061257416,
      "grad_norm": 2.4694955348968506,
      "learning_rate": 9.982413761848113e-06,
      "loss": 1.0418,
      "step": 1941
    },
    {
      "epoch": 0.0559557425229067,
      "grad_norm": 2.3493287563323975,
      "learning_rate": 9.982374639102859e-06,
      "loss": 1.011,
      "step": 1942
    },
    {
      "epoch": 0.055984555984555984,
      "grad_norm": 2.2935214042663574,
      "learning_rate": 9.982335472966103e-06,
      "loss": 1.0301,
      "step": 1943
    },
    {
      "epoch": 0.05601336944620527,
      "grad_norm": 2.57289719581604,
      "learning_rate": 9.982296263438185e-06,
      "loss": 1.0141,
      "step": 1944
    },
    {
      "epoch": 0.05604218290785455,
      "grad_norm": 2.3902533054351807,
      "learning_rate": 9.982257010519448e-06,
      "loss": 0.9943,
      "step": 1945
    },
    {
      "epoch": 0.056070996369503835,
      "grad_norm": 2.5896565914154053,
      "learning_rate": 9.982217714210231e-06,
      "loss": 1.0016,
      "step": 1946
    },
    {
      "epoch": 0.05609980983115311,
      "grad_norm": 2.510671615600586,
      "learning_rate": 9.98217837451088e-06,
      "loss": 1.0299,
      "step": 1947
    },
    {
      "epoch": 0.056128623292802396,
      "grad_norm": 2.963829278945923,
      "learning_rate": 9.982138991421735e-06,
      "loss": 1.0024,
      "step": 1948
    },
    {
      "epoch": 0.05615743675445168,
      "grad_norm": 2.6468348503112793,
      "learning_rate": 9.98209956494314e-06,
      "loss": 1.0399,
      "step": 1949
    },
    {
      "epoch": 0.056186250216100964,
      "grad_norm": 2.9073221683502197,
      "learning_rate": 9.982060095075438e-06,
      "loss": 0.9984,
      "step": 1950
    },
    {
      "epoch": 0.05621506367775025,
      "grad_norm": 2.569655418395996,
      "learning_rate": 9.982020581818973e-06,
      "loss": 1.0335,
      "step": 1951
    },
    {
      "epoch": 0.056243877139399524,
      "grad_norm": 2.7954652309417725,
      "learning_rate": 9.981981025174089e-06,
      "loss": 1.0348,
      "step": 1952
    },
    {
      "epoch": 0.05627269060104881,
      "grad_norm": 2.671771287918091,
      "learning_rate": 9.981941425141129e-06,
      "loss": 1.0209,
      "step": 1953
    },
    {
      "epoch": 0.05630150406269809,
      "grad_norm": 2.438964605331421,
      "learning_rate": 9.981901781720442e-06,
      "loss": 1.0187,
      "step": 1954
    },
    {
      "epoch": 0.056330317524347376,
      "grad_norm": 2.815903425216675,
      "learning_rate": 9.981862094912369e-06,
      "loss": 1.0255,
      "step": 1955
    },
    {
      "epoch": 0.05635913098599666,
      "grad_norm": 2.789168357849121,
      "learning_rate": 9.981822364717256e-06,
      "loss": 1.0151,
      "step": 1956
    },
    {
      "epoch": 0.056387944447645943,
      "grad_norm": 2.544715642929077,
      "learning_rate": 9.981782591135452e-06,
      "loss": 1.0064,
      "step": 1957
    },
    {
      "epoch": 0.05641675790929522,
      "grad_norm": 2.876772403717041,
      "learning_rate": 9.9817427741673e-06,
      "loss": 1.0189,
      "step": 1958
    },
    {
      "epoch": 0.056445571370944504,
      "grad_norm": 2.4981186389923096,
      "learning_rate": 9.981702913813149e-06,
      "loss": 0.9682,
      "step": 1959
    },
    {
      "epoch": 0.05647438483259379,
      "grad_norm": 2.452129364013672,
      "learning_rate": 9.981663010073346e-06,
      "loss": 1.004,
      "step": 1960
    },
    {
      "epoch": 0.05650319829424307,
      "grad_norm": 2.4349470138549805,
      "learning_rate": 9.981623062948238e-06,
      "loss": 1.0042,
      "step": 1961
    },
    {
      "epoch": 0.056532011755892356,
      "grad_norm": 2.397974729537964,
      "learning_rate": 9.981583072438174e-06,
      "loss": 1.0034,
      "step": 1962
    },
    {
      "epoch": 0.05656082521754163,
      "grad_norm": 2.554928779602051,
      "learning_rate": 9.9815430385435e-06,
      "loss": 1.0302,
      "step": 1963
    },
    {
      "epoch": 0.056589638679190916,
      "grad_norm": 2.358576536178589,
      "learning_rate": 9.981502961264566e-06,
      "loss": 0.9982,
      "step": 1964
    },
    {
      "epoch": 0.0566184521408402,
      "grad_norm": 2.698357582092285,
      "learning_rate": 9.98146284060172e-06,
      "loss": 1.0212,
      "step": 1965
    },
    {
      "epoch": 0.056647265602489484,
      "grad_norm": 2.517566680908203,
      "learning_rate": 9.981422676555313e-06,
      "loss": 1.0261,
      "step": 1966
    },
    {
      "epoch": 0.05667607906413877,
      "grad_norm": 2.6519789695739746,
      "learning_rate": 9.981382469125694e-06,
      "loss": 1.0316,
      "step": 1967
    },
    {
      "epoch": 0.056704892525788045,
      "grad_norm": 2.3908371925354004,
      "learning_rate": 9.981342218313213e-06,
      "loss": 1.0101,
      "step": 1968
    },
    {
      "epoch": 0.05673370598743733,
      "grad_norm": 2.640371084213257,
      "learning_rate": 9.981301924118222e-06,
      "loss": 1.0242,
      "step": 1969
    },
    {
      "epoch": 0.05676251944908661,
      "grad_norm": 2.3473312854766846,
      "learning_rate": 9.981261586541068e-06,
      "loss": 1.0421,
      "step": 1970
    },
    {
      "epoch": 0.056791332910735896,
      "grad_norm": 2.415618896484375,
      "learning_rate": 9.981221205582107e-06,
      "loss": 1.0197,
      "step": 1971
    },
    {
      "epoch": 0.05682014637238518,
      "grad_norm": 2.2246878147125244,
      "learning_rate": 9.98118078124169e-06,
      "loss": 1.0054,
      "step": 1972
    },
    {
      "epoch": 0.056848959834034464,
      "grad_norm": 2.369499444961548,
      "learning_rate": 9.981140313520166e-06,
      "loss": 1.0312,
      "step": 1973
    },
    {
      "epoch": 0.05687777329568374,
      "grad_norm": 2.471482276916504,
      "learning_rate": 9.981099802417891e-06,
      "loss": 0.9903,
      "step": 1974
    },
    {
      "epoch": 0.056906586757333025,
      "grad_norm": 2.172832727432251,
      "learning_rate": 9.981059247935213e-06,
      "loss": 0.9916,
      "step": 1975
    },
    {
      "epoch": 0.05693540021898231,
      "grad_norm": 2.392627239227295,
      "learning_rate": 9.98101865007249e-06,
      "loss": 1.0307,
      "step": 1976
    },
    {
      "epoch": 0.05696421368063159,
      "grad_norm": 2.356677293777466,
      "learning_rate": 9.980978008830074e-06,
      "loss": 1.0214,
      "step": 1977
    },
    {
      "epoch": 0.056993027142280876,
      "grad_norm": 2.620966672897339,
      "learning_rate": 9.980937324208317e-06,
      "loss": 1.0302,
      "step": 1978
    },
    {
      "epoch": 0.05702184060393015,
      "grad_norm": 2.4029579162597656,
      "learning_rate": 9.980896596207576e-06,
      "loss": 0.9681,
      "step": 1979
    },
    {
      "epoch": 0.05705065406557944,
      "grad_norm": 2.5946104526519775,
      "learning_rate": 9.980855824828205e-06,
      "loss": 1.0233,
      "step": 1980
    },
    {
      "epoch": 0.05707946752722872,
      "grad_norm": 2.4393527507781982,
      "learning_rate": 9.98081501007056e-06,
      "loss": 1.0441,
      "step": 1981
    },
    {
      "epoch": 0.057108280988878005,
      "grad_norm": 2.311836004257202,
      "learning_rate": 9.980774151934993e-06,
      "loss": 1.0078,
      "step": 1982
    },
    {
      "epoch": 0.05713709445052729,
      "grad_norm": 2.4991118907928467,
      "learning_rate": 9.980733250421863e-06,
      "loss": 1.0026,
      "step": 1983
    },
    {
      "epoch": 0.05716590791217657,
      "grad_norm": 2.529592514038086,
      "learning_rate": 9.980692305531527e-06,
      "loss": 1.0072,
      "step": 1984
    },
    {
      "epoch": 0.05719472137382585,
      "grad_norm": 2.4300456047058105,
      "learning_rate": 9.980651317264339e-06,
      "loss": 0.9736,
      "step": 1985
    },
    {
      "epoch": 0.05722353483547513,
      "grad_norm": 2.3909077644348145,
      "learning_rate": 9.980610285620655e-06,
      "loss": 1.0212,
      "step": 1986
    },
    {
      "epoch": 0.05725234829712442,
      "grad_norm": 2.3479371070861816,
      "learning_rate": 9.980569210600835e-06,
      "loss": 1.0099,
      "step": 1987
    },
    {
      "epoch": 0.0572811617587737,
      "grad_norm": 2.593770980834961,
      "learning_rate": 9.980528092205235e-06,
      "loss": 1.0173,
      "step": 1988
    },
    {
      "epoch": 0.057309975220422985,
      "grad_norm": 2.4607093334198,
      "learning_rate": 9.980486930434217e-06,
      "loss": 1.019,
      "step": 1989
    },
    {
      "epoch": 0.05733878868207226,
      "grad_norm": 2.7413418292999268,
      "learning_rate": 9.980445725288134e-06,
      "loss": 1.0185,
      "step": 1990
    },
    {
      "epoch": 0.057367602143721545,
      "grad_norm": 2.6757261753082275,
      "learning_rate": 9.980404476767347e-06,
      "loss": 1.0138,
      "step": 1991
    },
    {
      "epoch": 0.05739641560537083,
      "grad_norm": 2.56976580619812,
      "learning_rate": 9.980363184872217e-06,
      "loss": 1.0312,
      "step": 1992
    },
    {
      "epoch": 0.05742522906702011,
      "grad_norm": 2.608616352081299,
      "learning_rate": 9.980321849603103e-06,
      "loss": 0.975,
      "step": 1993
    },
    {
      "epoch": 0.0574540425286694,
      "grad_norm": 2.4464259147644043,
      "learning_rate": 9.980280470960363e-06,
      "loss": 0.997,
      "step": 1994
    },
    {
      "epoch": 0.057482855990318674,
      "grad_norm": 2.6077024936676025,
      "learning_rate": 9.980239048944359e-06,
      "loss": 1.0192,
      "step": 1995
    },
    {
      "epoch": 0.05751166945196796,
      "grad_norm": 2.4438037872314453,
      "learning_rate": 9.980197583555451e-06,
      "loss": 0.9849,
      "step": 1996
    },
    {
      "epoch": 0.05754048291361724,
      "grad_norm": 2.5204694271087646,
      "learning_rate": 9.980156074794e-06,
      "loss": 1.0178,
      "step": 1997
    },
    {
      "epoch": 0.057569296375266525,
      "grad_norm": 2.354501724243164,
      "learning_rate": 9.980114522660368e-06,
      "loss": 1.0276,
      "step": 1998
    },
    {
      "epoch": 0.05759810983691581,
      "grad_norm": 2.3501641750335693,
      "learning_rate": 9.980072927154918e-06,
      "loss": 1.0279,
      "step": 1999
    },
    {
      "epoch": 0.05762692329856509,
      "grad_norm": 2.3881845474243164,
      "learning_rate": 9.98003128827801e-06,
      "loss": 0.9946,
      "step": 2000
    },
    {
      "epoch": 0.05765573676021437,
      "grad_norm": 2.575242042541504,
      "learning_rate": 9.979989606030008e-06,
      "loss": 0.9885,
      "step": 2001
    },
    {
      "epoch": 0.057684550221863654,
      "grad_norm": 2.364347457885742,
      "learning_rate": 9.979947880411274e-06,
      "loss": 0.9985,
      "step": 2002
    },
    {
      "epoch": 0.05771336368351294,
      "grad_norm": 2.515585422515869,
      "learning_rate": 9.979906111422174e-06,
      "loss": 1.0134,
      "step": 2003
    },
    {
      "epoch": 0.05774217714516222,
      "grad_norm": 2.3182687759399414,
      "learning_rate": 9.979864299063067e-06,
      "loss": 0.9751,
      "step": 2004
    },
    {
      "epoch": 0.057770990606811505,
      "grad_norm": 2.4059040546417236,
      "learning_rate": 9.979822443334321e-06,
      "loss": 1.0316,
      "step": 2005
    },
    {
      "epoch": 0.05779980406846078,
      "grad_norm": 2.57816481590271,
      "learning_rate": 9.979780544236301e-06,
      "loss": 0.9854,
      "step": 2006
    },
    {
      "epoch": 0.057828617530110066,
      "grad_norm": 2.4047696590423584,
      "learning_rate": 9.97973860176937e-06,
      "loss": 1.0373,
      "step": 2007
    },
    {
      "epoch": 0.05785743099175935,
      "grad_norm": 2.277691125869751,
      "learning_rate": 9.979696615933892e-06,
      "loss": 1.0139,
      "step": 2008
    },
    {
      "epoch": 0.05788624445340863,
      "grad_norm": 2.4326529502868652,
      "learning_rate": 9.979654586730235e-06,
      "loss": 1.0226,
      "step": 2009
    },
    {
      "epoch": 0.05791505791505792,
      "grad_norm": 2.438952922821045,
      "learning_rate": 9.979612514158765e-06,
      "loss": 1.0193,
      "step": 2010
    },
    {
      "epoch": 0.057943871376707194,
      "grad_norm": 2.5671143531799316,
      "learning_rate": 9.979570398219848e-06,
      "loss": 1.022,
      "step": 2011
    },
    {
      "epoch": 0.05797268483835648,
      "grad_norm": 2.268808603286743,
      "learning_rate": 9.979528238913849e-06,
      "loss": 0.9835,
      "step": 2012
    },
    {
      "epoch": 0.05800149830000576,
      "grad_norm": 2.7531490325927734,
      "learning_rate": 9.97948603624114e-06,
      "loss": 1.0016,
      "step": 2013
    },
    {
      "epoch": 0.058030311761655046,
      "grad_norm": 2.397332191467285,
      "learning_rate": 9.979443790202082e-06,
      "loss": 1.0155,
      "step": 2014
    },
    {
      "epoch": 0.05805912522330433,
      "grad_norm": 2.4544484615325928,
      "learning_rate": 9.979401500797048e-06,
      "loss": 1.0165,
      "step": 2015
    },
    {
      "epoch": 0.05808793868495361,
      "grad_norm": 2.374368667602539,
      "learning_rate": 9.979359168026403e-06,
      "loss": 1.0077,
      "step": 2016
    },
    {
      "epoch": 0.05811675214660289,
      "grad_norm": 2.445124626159668,
      "learning_rate": 9.97931679189052e-06,
      "loss": 0.9922,
      "step": 2017
    },
    {
      "epoch": 0.058145565608252174,
      "grad_norm": 2.6091461181640625,
      "learning_rate": 9.979274372389761e-06,
      "loss": 1.0621,
      "step": 2018
    },
    {
      "epoch": 0.05817437906990146,
      "grad_norm": 2.470733404159546,
      "learning_rate": 9.979231909524503e-06,
      "loss": 1.02,
      "step": 2019
    },
    {
      "epoch": 0.05820319253155074,
      "grad_norm": 2.432950973510742,
      "learning_rate": 9.979189403295111e-06,
      "loss": 0.9994,
      "step": 2020
    },
    {
      "epoch": 0.058232005993200026,
      "grad_norm": 2.4106285572052,
      "learning_rate": 9.979146853701958e-06,
      "loss": 1.0241,
      "step": 2021
    },
    {
      "epoch": 0.0582608194548493,
      "grad_norm": 2.4056098461151123,
      "learning_rate": 9.979104260745412e-06,
      "loss": 0.9884,
      "step": 2022
    },
    {
      "epoch": 0.058289632916498586,
      "grad_norm": 2.6832613945007324,
      "learning_rate": 9.979061624425845e-06,
      "loss": 1.0397,
      "step": 2023
    },
    {
      "epoch": 0.05831844637814787,
      "grad_norm": 2.710080862045288,
      "learning_rate": 9.97901894474363e-06,
      "loss": 0.9948,
      "step": 2024
    },
    {
      "epoch": 0.058347259839797154,
      "grad_norm": 2.5391738414764404,
      "learning_rate": 9.978976221699136e-06,
      "loss": 0.9827,
      "step": 2025
    },
    {
      "epoch": 0.05837607330144644,
      "grad_norm": 2.488542318344116,
      "learning_rate": 9.978933455292736e-06,
      "loss": 1.0125,
      "step": 2026
    },
    {
      "epoch": 0.05840488676309572,
      "grad_norm": 2.674013376235962,
      "learning_rate": 9.978890645524803e-06,
      "loss": 1.0025,
      "step": 2027
    },
    {
      "epoch": 0.058433700224745,
      "grad_norm": 2.7202529907226562,
      "learning_rate": 9.97884779239571e-06,
      "loss": 1.0283,
      "step": 2028
    },
    {
      "epoch": 0.05846251368639428,
      "grad_norm": 2.5736281871795654,
      "learning_rate": 9.978804895905831e-06,
      "loss": 0.9903,
      "step": 2029
    },
    {
      "epoch": 0.058491327148043566,
      "grad_norm": 2.5957374572753906,
      "learning_rate": 9.978761956055535e-06,
      "loss": 1.0043,
      "step": 2030
    },
    {
      "epoch": 0.05852014060969285,
      "grad_norm": 2.599252700805664,
      "learning_rate": 9.978718972845203e-06,
      "loss": 1.0346,
      "step": 2031
    },
    {
      "epoch": 0.058548954071342134,
      "grad_norm": 2.324362277984619,
      "learning_rate": 9.978675946275203e-06,
      "loss": 0.9931,
      "step": 2032
    },
    {
      "epoch": 0.05857776753299141,
      "grad_norm": 2.392223834991455,
      "learning_rate": 9.978632876345912e-06,
      "loss": 1.003,
      "step": 2033
    },
    {
      "epoch": 0.058606580994640695,
      "grad_norm": 2.4562065601348877,
      "learning_rate": 9.978589763057707e-06,
      "loss": 1.0195,
      "step": 2034
    },
    {
      "epoch": 0.05863539445628998,
      "grad_norm": 2.371885061264038,
      "learning_rate": 9.978546606410962e-06,
      "loss": 1.0279,
      "step": 2035
    },
    {
      "epoch": 0.05866420791793926,
      "grad_norm": 2.4260597229003906,
      "learning_rate": 9.978503406406053e-06,
      "loss": 1.0205,
      "step": 2036
    },
    {
      "epoch": 0.058693021379588546,
      "grad_norm": 2.2854971885681152,
      "learning_rate": 9.978460163043356e-06,
      "loss": 1.0042,
      "step": 2037
    },
    {
      "epoch": 0.05872183484123782,
      "grad_norm": 2.1472532749176025,
      "learning_rate": 9.978416876323246e-06,
      "loss": 0.9892,
      "step": 2038
    },
    {
      "epoch": 0.05875064830288711,
      "grad_norm": 2.3978383541107178,
      "learning_rate": 9.978373546246103e-06,
      "loss": 0.9953,
      "step": 2039
    },
    {
      "epoch": 0.05877946176453639,
      "grad_norm": 2.3572990894317627,
      "learning_rate": 9.978330172812303e-06,
      "loss": 1.0154,
      "step": 2040
    },
    {
      "epoch": 0.058808275226185674,
      "grad_norm": 2.332167148590088,
      "learning_rate": 9.978286756022222e-06,
      "loss": 1.0108,
      "step": 2041
    },
    {
      "epoch": 0.05883708868783496,
      "grad_norm": 2.459416151046753,
      "learning_rate": 9.978243295876242e-06,
      "loss": 1.0248,
      "step": 2042
    },
    {
      "epoch": 0.05886590214948424,
      "grad_norm": 2.3306658267974854,
      "learning_rate": 9.978199792374739e-06,
      "loss": 1.0038,
      "step": 2043
    },
    {
      "epoch": 0.05889471561113352,
      "grad_norm": 2.503257989883423,
      "learning_rate": 9.978156245518092e-06,
      "loss": 1.0174,
      "step": 2044
    },
    {
      "epoch": 0.0589235290727828,
      "grad_norm": 2.3845443725585938,
      "learning_rate": 9.978112655306679e-06,
      "loss": 1.0082,
      "step": 2045
    },
    {
      "epoch": 0.05895234253443209,
      "grad_norm": 2.4164681434631348,
      "learning_rate": 9.978069021740881e-06,
      "loss": 1.0103,
      "step": 2046
    },
    {
      "epoch": 0.05898115599608137,
      "grad_norm": 2.5298995971679688,
      "learning_rate": 9.97802534482108e-06,
      "loss": 0.9899,
      "step": 2047
    },
    {
      "epoch": 0.059009969457730654,
      "grad_norm": 2.3038268089294434,
      "learning_rate": 9.977981624547653e-06,
      "loss": 1.0142,
      "step": 2048
    },
    {
      "epoch": 0.05903878291937993,
      "grad_norm": 2.518307685852051,
      "learning_rate": 9.977937860920983e-06,
      "loss": 1.0303,
      "step": 2049
    },
    {
      "epoch": 0.059067596381029215,
      "grad_norm": 2.402263879776001,
      "learning_rate": 9.97789405394145e-06,
      "loss": 1.0286,
      "step": 2050
    },
    {
      "epoch": 0.0590964098426785,
      "grad_norm": 2.9483628273010254,
      "learning_rate": 9.977850203609436e-06,
      "loss": 0.9915,
      "step": 2051
    },
    {
      "epoch": 0.05912522330432778,
      "grad_norm": 2.829740524291992,
      "learning_rate": 9.977806309925325e-06,
      "loss": 1.0406,
      "step": 2052
    },
    {
      "epoch": 0.05915403676597707,
      "grad_norm": 2.447704553604126,
      "learning_rate": 9.977762372889493e-06,
      "loss": 0.9768,
      "step": 2053
    },
    {
      "epoch": 0.05918285022762635,
      "grad_norm": 2.4446094036102295,
      "learning_rate": 9.977718392502329e-06,
      "loss": 0.9951,
      "step": 2054
    },
    {
      "epoch": 0.05921166368927563,
      "grad_norm": 2.4215598106384277,
      "learning_rate": 9.977674368764213e-06,
      "loss": 1.027,
      "step": 2055
    },
    {
      "epoch": 0.05924047715092491,
      "grad_norm": 2.612002372741699,
      "learning_rate": 9.977630301675529e-06,
      "loss": 0.9819,
      "step": 2056
    },
    {
      "epoch": 0.059269290612574195,
      "grad_norm": 2.479423999786377,
      "learning_rate": 9.97758619123666e-06,
      "loss": 1.035,
      "step": 2057
    },
    {
      "epoch": 0.05929810407422348,
      "grad_norm": 2.491562604904175,
      "learning_rate": 9.977542037447993e-06,
      "loss": 1.0056,
      "step": 2058
    },
    {
      "epoch": 0.05932691753587276,
      "grad_norm": 2.665959358215332,
      "learning_rate": 9.97749784030991e-06,
      "loss": 1.0106,
      "step": 2059
    },
    {
      "epoch": 0.05935573099752204,
      "grad_norm": 2.3288662433624268,
      "learning_rate": 9.977453599822797e-06,
      "loss": 1.0009,
      "step": 2060
    },
    {
      "epoch": 0.05938454445917132,
      "grad_norm": 2.5119376182556152,
      "learning_rate": 9.977409315987039e-06,
      "loss": 1.0094,
      "step": 2061
    },
    {
      "epoch": 0.05941335792082061,
      "grad_norm": 2.3667171001434326,
      "learning_rate": 9.97736498880302e-06,
      "loss": 1.0131,
      "step": 2062
    },
    {
      "epoch": 0.05944217138246989,
      "grad_norm": 2.197549343109131,
      "learning_rate": 9.977320618271127e-06,
      "loss": 1.0124,
      "step": 2063
    },
    {
      "epoch": 0.059470984844119175,
      "grad_norm": 2.693903684616089,
      "learning_rate": 9.977276204391748e-06,
      "loss": 1.0472,
      "step": 2064
    },
    {
      "epoch": 0.05949979830576845,
      "grad_norm": 2.4963531494140625,
      "learning_rate": 9.977231747165267e-06,
      "loss": 0.9928,
      "step": 2065
    },
    {
      "epoch": 0.059528611767417736,
      "grad_norm": 2.3337044715881348,
      "learning_rate": 9.977187246592077e-06,
      "loss": 1.0369,
      "step": 2066
    },
    {
      "epoch": 0.05955742522906702,
      "grad_norm": 2.6950490474700928,
      "learning_rate": 9.977142702672558e-06,
      "loss": 1.0103,
      "step": 2067
    },
    {
      "epoch": 0.0595862386907163,
      "grad_norm": 2.3680591583251953,
      "learning_rate": 9.977098115407102e-06,
      "loss": 1.0045,
      "step": 2068
    },
    {
      "epoch": 0.05961505215236559,
      "grad_norm": 2.566232681274414,
      "learning_rate": 9.977053484796098e-06,
      "loss": 1.0285,
      "step": 2069
    },
    {
      "epoch": 0.05964386561401487,
      "grad_norm": 2.6373393535614014,
      "learning_rate": 9.977008810839932e-06,
      "loss": 1.0118,
      "step": 2070
    },
    {
      "epoch": 0.05967267907566415,
      "grad_norm": 2.3490712642669678,
      "learning_rate": 9.976964093538995e-06,
      "loss": 1.0351,
      "step": 2071
    },
    {
      "epoch": 0.05970149253731343,
      "grad_norm": 2.653240442276001,
      "learning_rate": 9.976919332893675e-06,
      "loss": 1.0241,
      "step": 2072
    },
    {
      "epoch": 0.059730305998962716,
      "grad_norm": 2.474536418914795,
      "learning_rate": 9.976874528904364e-06,
      "loss": 0.9735,
      "step": 2073
    },
    {
      "epoch": 0.059759119460612,
      "grad_norm": 2.6943235397338867,
      "learning_rate": 9.976829681571451e-06,
      "loss": 1.0085,
      "step": 2074
    },
    {
      "epoch": 0.05978793292226128,
      "grad_norm": 2.76127552986145,
      "learning_rate": 9.976784790895326e-06,
      "loss": 1.0127,
      "step": 2075
    },
    {
      "epoch": 0.05981674638391056,
      "grad_norm": 2.3856282234191895,
      "learning_rate": 9.97673985687638e-06,
      "loss": 1.0153,
      "step": 2076
    },
    {
      "epoch": 0.059845559845559844,
      "grad_norm": 2.7265677452087402,
      "learning_rate": 9.976694879515005e-06,
      "loss": 0.9928,
      "step": 2077
    },
    {
      "epoch": 0.05987437330720913,
      "grad_norm": 2.4844415187835693,
      "learning_rate": 9.976649858811592e-06,
      "loss": 1.0305,
      "step": 2078
    },
    {
      "epoch": 0.05990318676885841,
      "grad_norm": 3.0109424591064453,
      "learning_rate": 9.976604794766533e-06,
      "loss": 1.0106,
      "step": 2079
    },
    {
      "epoch": 0.059932000230507695,
      "grad_norm": 2.563406229019165,
      "learning_rate": 9.976559687380223e-06,
      "loss": 1.0211,
      "step": 2080
    },
    {
      "epoch": 0.05996081369215697,
      "grad_norm": 2.3214454650878906,
      "learning_rate": 9.976514536653052e-06,
      "loss": 0.9811,
      "step": 2081
    },
    {
      "epoch": 0.059989627153806256,
      "grad_norm": 2.5821189880371094,
      "learning_rate": 9.976469342585415e-06,
      "loss": 1.0049,
      "step": 2082
    },
    {
      "epoch": 0.06001844061545554,
      "grad_norm": 2.2628307342529297,
      "learning_rate": 9.976424105177703e-06,
      "loss": 0.9899,
      "step": 2083
    },
    {
      "epoch": 0.060047254077104824,
      "grad_norm": 2.4345779418945312,
      "learning_rate": 9.976378824430313e-06,
      "loss": 1.0241,
      "step": 2084
    },
    {
      "epoch": 0.06007606753875411,
      "grad_norm": 2.435356616973877,
      "learning_rate": 9.976333500343638e-06,
      "loss": 0.9905,
      "step": 2085
    },
    {
      "epoch": 0.06010488100040339,
      "grad_norm": 2.578073024749756,
      "learning_rate": 9.97628813291807e-06,
      "loss": 1.003,
      "step": 2086
    },
    {
      "epoch": 0.06013369446205267,
      "grad_norm": 2.1804306507110596,
      "learning_rate": 9.97624272215401e-06,
      "loss": 1.0033,
      "step": 2087
    },
    {
      "epoch": 0.06016250792370195,
      "grad_norm": 2.7876598834991455,
      "learning_rate": 9.976197268051848e-06,
      "loss": 0.9974,
      "step": 2088
    },
    {
      "epoch": 0.060191321385351236,
      "grad_norm": 2.291868209838867,
      "learning_rate": 9.976151770611983e-06,
      "loss": 1.0027,
      "step": 2089
    },
    {
      "epoch": 0.06022013484700052,
      "grad_norm": 2.3874688148498535,
      "learning_rate": 9.976106229834812e-06,
      "loss": 0.9744,
      "step": 2090
    },
    {
      "epoch": 0.060248948308649804,
      "grad_norm": 2.4208059310913086,
      "learning_rate": 9.976060645720729e-06,
      "loss": 0.9679,
      "step": 2091
    },
    {
      "epoch": 0.06027776177029908,
      "grad_norm": 2.404837131500244,
      "learning_rate": 9.976015018270131e-06,
      "loss": 1.0034,
      "step": 2092
    },
    {
      "epoch": 0.060306575231948364,
      "grad_norm": 2.384868860244751,
      "learning_rate": 9.97596934748342e-06,
      "loss": 1.0113,
      "step": 2093
    },
    {
      "epoch": 0.06033538869359765,
      "grad_norm": 2.6341428756713867,
      "learning_rate": 9.975923633360985e-06,
      "loss": 1.0016,
      "step": 2094
    },
    {
      "epoch": 0.06036420215524693,
      "grad_norm": 2.3807547092437744,
      "learning_rate": 9.975877875903233e-06,
      "loss": 1.0187,
      "step": 2095
    },
    {
      "epoch": 0.060393015616896216,
      "grad_norm": 2.6022255420684814,
      "learning_rate": 9.975832075110556e-06,
      "loss": 0.9599,
      "step": 2096
    },
    {
      "epoch": 0.0604218290785455,
      "grad_norm": 2.483699321746826,
      "learning_rate": 9.975786230983357e-06,
      "loss": 1.0169,
      "step": 2097
    },
    {
      "epoch": 0.06045064254019478,
      "grad_norm": 2.589235305786133,
      "learning_rate": 9.975740343522033e-06,
      "loss": 0.9987,
      "step": 2098
    },
    {
      "epoch": 0.06047945600184406,
      "grad_norm": 2.301091432571411,
      "learning_rate": 9.975694412726987e-06,
      "loss": 0.9977,
      "step": 2099
    },
    {
      "epoch": 0.060508269463493344,
      "grad_norm": 2.56459641456604,
      "learning_rate": 9.975648438598615e-06,
      "loss": 1.006,
      "step": 2100
    },
    {
      "epoch": 0.06053708292514263,
      "grad_norm": 2.378584384918213,
      "learning_rate": 9.975602421137318e-06,
      "loss": 0.9883,
      "step": 2101
    },
    {
      "epoch": 0.06056589638679191,
      "grad_norm": 2.636408805847168,
      "learning_rate": 9.975556360343499e-06,
      "loss": 0.9973,
      "step": 2102
    },
    {
      "epoch": 0.06059470984844119,
      "grad_norm": 2.5797278881073,
      "learning_rate": 9.975510256217558e-06,
      "loss": 1.0151,
      "step": 2103
    },
    {
      "epoch": 0.06062352331009047,
      "grad_norm": 2.5234482288360596,
      "learning_rate": 9.975464108759896e-06,
      "loss": 1.01,
      "step": 2104
    },
    {
      "epoch": 0.06065233677173976,
      "grad_norm": 2.552995204925537,
      "learning_rate": 9.975417917970914e-06,
      "loss": 1.0138,
      "step": 2105
    },
    {
      "epoch": 0.06068115023338904,
      "grad_norm": 2.4974076747894287,
      "learning_rate": 9.975371683851017e-06,
      "loss": 0.9931,
      "step": 2106
    },
    {
      "epoch": 0.060709963695038324,
      "grad_norm": 2.5381760597229004,
      "learning_rate": 9.975325406400605e-06,
      "loss": 0.9967,
      "step": 2107
    },
    {
      "epoch": 0.0607387771566876,
      "grad_norm": 2.6262404918670654,
      "learning_rate": 9.975279085620084e-06,
      "loss": 0.9927,
      "step": 2108
    },
    {
      "epoch": 0.060767590618336885,
      "grad_norm": 2.354743480682373,
      "learning_rate": 9.975232721509854e-06,
      "loss": 1.005,
      "step": 2109
    },
    {
      "epoch": 0.06079640407998617,
      "grad_norm": 2.602369785308838,
      "learning_rate": 9.97518631407032e-06,
      "loss": 0.9754,
      "step": 2110
    },
    {
      "epoch": 0.06082521754163545,
      "grad_norm": 2.639984607696533,
      "learning_rate": 9.975139863301887e-06,
      "loss": 1.0132,
      "step": 2111
    },
    {
      "epoch": 0.060854031003284736,
      "grad_norm": 2.428678035736084,
      "learning_rate": 9.975093369204961e-06,
      "loss": 1.0363,
      "step": 2112
    },
    {
      "epoch": 0.06088284446493402,
      "grad_norm": 2.8014304637908936,
      "learning_rate": 9.975046831779945e-06,
      "loss": 1.0311,
      "step": 2113
    },
    {
      "epoch": 0.0609116579265833,
      "grad_norm": 2.4257278442382812,
      "learning_rate": 9.975000251027242e-06,
      "loss": 1.0264,
      "step": 2114
    },
    {
      "epoch": 0.06094047138823258,
      "grad_norm": 2.49753737449646,
      "learning_rate": 9.974953626947262e-06,
      "loss": 1.0632,
      "step": 2115
    },
    {
      "epoch": 0.060969284849881865,
      "grad_norm": 2.1890084743499756,
      "learning_rate": 9.974906959540407e-06,
      "loss": 1.0199,
      "step": 2116
    },
    {
      "epoch": 0.06099809831153115,
      "grad_norm": 2.494508743286133,
      "learning_rate": 9.974860248807088e-06,
      "loss": 0.99,
      "step": 2117
    },
    {
      "epoch": 0.06102691177318043,
      "grad_norm": 2.3798391819000244,
      "learning_rate": 9.974813494747706e-06,
      "loss": 0.9919,
      "step": 2118
    },
    {
      "epoch": 0.06105572523482971,
      "grad_norm": 2.444899082183838,
      "learning_rate": 9.974766697362675e-06,
      "loss": 1.0037,
      "step": 2119
    },
    {
      "epoch": 0.06108453869647899,
      "grad_norm": 2.16119647026062,
      "learning_rate": 9.974719856652395e-06,
      "loss": 1.0232,
      "step": 2120
    },
    {
      "epoch": 0.06111335215812828,
      "grad_norm": 2.433903932571411,
      "learning_rate": 9.97467297261728e-06,
      "loss": 0.9957,
      "step": 2121
    },
    {
      "epoch": 0.06114216561977756,
      "grad_norm": 2.534923553466797,
      "learning_rate": 9.974626045257738e-06,
      "loss": 1.022,
      "step": 2122
    },
    {
      "epoch": 0.061170979081426845,
      "grad_norm": 2.156123161315918,
      "learning_rate": 9.974579074574175e-06,
      "loss": 1.0142,
      "step": 2123
    },
    {
      "epoch": 0.06119979254307613,
      "grad_norm": 2.377105474472046,
      "learning_rate": 9.974532060567e-06,
      "loss": 0.993,
      "step": 2124
    },
    {
      "epoch": 0.061228606004725405,
      "grad_norm": 2.235823392868042,
      "learning_rate": 9.974485003236623e-06,
      "loss": 0.9768,
      "step": 2125
    },
    {
      "epoch": 0.06125741946637469,
      "grad_norm": 2.27599835395813,
      "learning_rate": 9.974437902583455e-06,
      "loss": 1.0118,
      "step": 2126
    },
    {
      "epoch": 0.06128623292802397,
      "grad_norm": 2.552100896835327,
      "learning_rate": 9.974390758607907e-06,
      "loss": 0.9929,
      "step": 2127
    },
    {
      "epoch": 0.06131504638967326,
      "grad_norm": 2.745971918106079,
      "learning_rate": 9.974343571310387e-06,
      "loss": 1.0138,
      "step": 2128
    },
    {
      "epoch": 0.06134385985132254,
      "grad_norm": 2.5530166625976562,
      "learning_rate": 9.974296340691306e-06,
      "loss": 1.0275,
      "step": 2129
    },
    {
      "epoch": 0.06137267331297182,
      "grad_norm": 2.431527853012085,
      "learning_rate": 9.974249066751078e-06,
      "loss": 0.9729,
      "step": 2130
    },
    {
      "epoch": 0.0614014867746211,
      "grad_norm": 2.3884129524230957,
      "learning_rate": 9.974201749490112e-06,
      "loss": 1.028,
      "step": 2131
    },
    {
      "epoch": 0.061430300236270385,
      "grad_norm": 2.3488802909851074,
      "learning_rate": 9.974154388908821e-06,
      "loss": 1.0035,
      "step": 2132
    },
    {
      "epoch": 0.06145911369791967,
      "grad_norm": 2.310351610183716,
      "learning_rate": 9.974106985007618e-06,
      "loss": 1.0292,
      "step": 2133
    },
    {
      "epoch": 0.06148792715956895,
      "grad_norm": 2.4133729934692383,
      "learning_rate": 9.974059537786915e-06,
      "loss": 1.0458,
      "step": 2134
    },
    {
      "epoch": 0.06151674062121823,
      "grad_norm": 2.4983129501342773,
      "learning_rate": 9.974012047247126e-06,
      "loss": 1.0211,
      "step": 2135
    },
    {
      "epoch": 0.061545554082867514,
      "grad_norm": 2.5764033794403076,
      "learning_rate": 9.973964513388665e-06,
      "loss": 1.0222,
      "step": 2136
    },
    {
      "epoch": 0.0615743675445168,
      "grad_norm": 2.3903653621673584,
      "learning_rate": 9.973916936211944e-06,
      "loss": 1.0055,
      "step": 2137
    },
    {
      "epoch": 0.06160318100616608,
      "grad_norm": 2.792933225631714,
      "learning_rate": 9.973869315717379e-06,
      "loss": 0.9836,
      "step": 2138
    },
    {
      "epoch": 0.061631994467815365,
      "grad_norm": 2.5203535556793213,
      "learning_rate": 9.973821651905384e-06,
      "loss": 1.0177,
      "step": 2139
    },
    {
      "epoch": 0.06166080792946465,
      "grad_norm": 2.678276538848877,
      "learning_rate": 9.973773944776376e-06,
      "loss": 0.9984,
      "step": 2140
    },
    {
      "epoch": 0.061689621391113926,
      "grad_norm": 2.6762917041778564,
      "learning_rate": 9.973726194330768e-06,
      "loss": 1.0016,
      "step": 2141
    },
    {
      "epoch": 0.06171843485276321,
      "grad_norm": 2.59594464302063,
      "learning_rate": 9.973678400568976e-06,
      "loss": 1.0203,
      "step": 2142
    },
    {
      "epoch": 0.061747248314412494,
      "grad_norm": 2.5866706371307373,
      "learning_rate": 9.973630563491417e-06,
      "loss": 0.9903,
      "step": 2143
    },
    {
      "epoch": 0.06177606177606178,
      "grad_norm": 2.6761317253112793,
      "learning_rate": 9.973582683098509e-06,
      "loss": 0.9962,
      "step": 2144
    },
    {
      "epoch": 0.06180487523771106,
      "grad_norm": 2.5507469177246094,
      "learning_rate": 9.973534759390667e-06,
      "loss": 1.0135,
      "step": 2145
    },
    {
      "epoch": 0.06183368869936034,
      "grad_norm": 2.7124927043914795,
      "learning_rate": 9.973486792368308e-06,
      "loss": 1.0131,
      "step": 2146
    },
    {
      "epoch": 0.06186250216100962,
      "grad_norm": 2.398672580718994,
      "learning_rate": 9.97343878203185e-06,
      "loss": 0.9894,
      "step": 2147
    },
    {
      "epoch": 0.061891315622658906,
      "grad_norm": 2.35683274269104,
      "learning_rate": 9.973390728381713e-06,
      "loss": 0.9826,
      "step": 2148
    },
    {
      "epoch": 0.06192012908430819,
      "grad_norm": 2.307539701461792,
      "learning_rate": 9.973342631418315e-06,
      "loss": 0.9911,
      "step": 2149
    },
    {
      "epoch": 0.061948942545957474,
      "grad_norm": 2.422532796859741,
      "learning_rate": 9.973294491142073e-06,
      "loss": 1.0063,
      "step": 2150
    },
    {
      "epoch": 0.06197775600760675,
      "grad_norm": 2.1024551391601562,
      "learning_rate": 9.97324630755341e-06,
      "loss": 1.0076,
      "step": 2151
    },
    {
      "epoch": 0.062006569469256034,
      "grad_norm": 2.42878794670105,
      "learning_rate": 9.97319808065274e-06,
      "loss": 0.9876,
      "step": 2152
    },
    {
      "epoch": 0.06203538293090532,
      "grad_norm": 2.2018020153045654,
      "learning_rate": 9.973149810440487e-06,
      "loss": 1.0057,
      "step": 2153
    },
    {
      "epoch": 0.0620641963925546,
      "grad_norm": 2.486335039138794,
      "learning_rate": 9.973101496917071e-06,
      "loss": 1.0025,
      "step": 2154
    },
    {
      "epoch": 0.062093009854203886,
      "grad_norm": 2.5210278034210205,
      "learning_rate": 9.973053140082913e-06,
      "loss": 1.0207,
      "step": 2155
    },
    {
      "epoch": 0.06212182331585317,
      "grad_norm": 2.707798719406128,
      "learning_rate": 9.973004739938432e-06,
      "loss": 1.0291,
      "step": 2156
    },
    {
      "epoch": 0.062150636777502447,
      "grad_norm": 2.6711933612823486,
      "learning_rate": 9.972956296484051e-06,
      "loss": 0.9827,
      "step": 2157
    },
    {
      "epoch": 0.06217945023915173,
      "grad_norm": 2.6763970851898193,
      "learning_rate": 9.972907809720192e-06,
      "loss": 1.0109,
      "step": 2158
    },
    {
      "epoch": 0.062208263700801014,
      "grad_norm": 2.6968088150024414,
      "learning_rate": 9.972859279647277e-06,
      "loss": 1.0257,
      "step": 2159
    },
    {
      "epoch": 0.0622370771624503,
      "grad_norm": 2.676684617996216,
      "learning_rate": 9.97281070626573e-06,
      "loss": 1.001,
      "step": 2160
    },
    {
      "epoch": 0.06226589062409958,
      "grad_norm": 2.454536199569702,
      "learning_rate": 9.972762089575972e-06,
      "loss": 1.0201,
      "step": 2161
    },
    {
      "epoch": 0.06229470408574886,
      "grad_norm": 2.907231569290161,
      "learning_rate": 9.972713429578427e-06,
      "loss": 0.9931,
      "step": 2162
    },
    {
      "epoch": 0.06232351754739814,
      "grad_norm": 2.503106117248535,
      "learning_rate": 9.972664726273518e-06,
      "loss": 0.9601,
      "step": 2163
    },
    {
      "epoch": 0.062352331009047426,
      "grad_norm": 3.0061399936676025,
      "learning_rate": 9.97261597966167e-06,
      "loss": 0.9983,
      "step": 2164
    },
    {
      "epoch": 0.06238114447069671,
      "grad_norm": 2.251436710357666,
      "learning_rate": 9.97256718974331e-06,
      "loss": 0.9744,
      "step": 2165
    },
    {
      "epoch": 0.062409957932345994,
      "grad_norm": 2.9032437801361084,
      "learning_rate": 9.972518356518858e-06,
      "loss": 0.9867,
      "step": 2166
    },
    {
      "epoch": 0.06243877139399528,
      "grad_norm": 2.5713744163513184,
      "learning_rate": 9.972469479988743e-06,
      "loss": 1.0,
      "step": 2167
    },
    {
      "epoch": 0.062467584855644555,
      "grad_norm": 2.8426175117492676,
      "learning_rate": 9.972420560153389e-06,
      "loss": 1.0059,
      "step": 2168
    },
    {
      "epoch": 0.06249639831729384,
      "grad_norm": 2.570744514465332,
      "learning_rate": 9.972371597013223e-06,
      "loss": 0.9926,
      "step": 2169
    },
    {
      "epoch": 0.06252521177894312,
      "grad_norm": 2.2455992698669434,
      "learning_rate": 9.97232259056867e-06,
      "loss": 1.006,
      "step": 2170
    },
    {
      "epoch": 0.0625540252405924,
      "grad_norm": 2.3646600246429443,
      "learning_rate": 9.97227354082016e-06,
      "loss": 0.9934,
      "step": 2171
    },
    {
      "epoch": 0.06258283870224168,
      "grad_norm": 2.53669810295105,
      "learning_rate": 9.972224447768114e-06,
      "loss": 1.0138,
      "step": 2172
    },
    {
      "epoch": 0.06261165216389097,
      "grad_norm": 2.4084153175354004,
      "learning_rate": 9.972175311412967e-06,
      "loss": 0.9875,
      "step": 2173
    },
    {
      "epoch": 0.06264046562554025,
      "grad_norm": 2.467430830001831,
      "learning_rate": 9.972126131755142e-06,
      "loss": 1.0155,
      "step": 2174
    },
    {
      "epoch": 0.06266927908718954,
      "grad_norm": 2.4257566928863525,
      "learning_rate": 9.97207690879507e-06,
      "loss": 1.0174,
      "step": 2175
    },
    {
      "epoch": 0.06269809254883882,
      "grad_norm": 2.5117666721343994,
      "learning_rate": 9.972027642533177e-06,
      "loss": 0.9806,
      "step": 2176
    },
    {
      "epoch": 0.0627269060104881,
      "grad_norm": 2.268986701965332,
      "learning_rate": 9.971978332969894e-06,
      "loss": 0.9926,
      "step": 2177
    },
    {
      "epoch": 0.06275571947213739,
      "grad_norm": 2.5855352878570557,
      "learning_rate": 9.97192898010565e-06,
      "loss": 1.0128,
      "step": 2178
    },
    {
      "epoch": 0.06278453293378666,
      "grad_norm": 2.3575265407562256,
      "learning_rate": 9.971879583940873e-06,
      "loss": 1.004,
      "step": 2179
    },
    {
      "epoch": 0.06281334639543595,
      "grad_norm": 2.5672690868377686,
      "learning_rate": 9.971830144475996e-06,
      "loss": 0.9932,
      "step": 2180
    },
    {
      "epoch": 0.06284215985708523,
      "grad_norm": 2.5016930103302,
      "learning_rate": 9.97178066171145e-06,
      "loss": 1.0127,
      "step": 2181
    },
    {
      "epoch": 0.06287097331873451,
      "grad_norm": 2.4621713161468506,
      "learning_rate": 9.971731135647662e-06,
      "loss": 1.0104,
      "step": 2182
    },
    {
      "epoch": 0.0628997867803838,
      "grad_norm": 2.6482889652252197,
      "learning_rate": 9.971681566285067e-06,
      "loss": 0.9976,
      "step": 2183
    },
    {
      "epoch": 0.06292860024203308,
      "grad_norm": 2.246960163116455,
      "learning_rate": 9.971631953624094e-06,
      "loss": 1.0217,
      "step": 2184
    },
    {
      "epoch": 0.06295741370368237,
      "grad_norm": 2.364781141281128,
      "learning_rate": 9.97158229766518e-06,
      "loss": 1.0097,
      "step": 2185
    },
    {
      "epoch": 0.06298622716533164,
      "grad_norm": 2.5147109031677246,
      "learning_rate": 9.97153259840875e-06,
      "loss": 1.0479,
      "step": 2186
    },
    {
      "epoch": 0.06301504062698092,
      "grad_norm": 2.2841427326202393,
      "learning_rate": 9.971482855855243e-06,
      "loss": 1.0036,
      "step": 2187
    },
    {
      "epoch": 0.06304385408863021,
      "grad_norm": 2.517595052719116,
      "learning_rate": 9.97143307000509e-06,
      "loss": 0.9942,
      "step": 2188
    },
    {
      "epoch": 0.06307266755027949,
      "grad_norm": 2.3634462356567383,
      "learning_rate": 9.971383240858724e-06,
      "loss": 1.0201,
      "step": 2189
    },
    {
      "epoch": 0.06310148101192878,
      "grad_norm": 2.2443292140960693,
      "learning_rate": 9.971333368416578e-06,
      "loss": 0.9907,
      "step": 2190
    },
    {
      "epoch": 0.06313029447357806,
      "grad_norm": 2.4791083335876465,
      "learning_rate": 9.97128345267909e-06,
      "loss": 1.0264,
      "step": 2191
    },
    {
      "epoch": 0.06315910793522733,
      "grad_norm": 2.2722456455230713,
      "learning_rate": 9.971233493646692e-06,
      "loss": 0.9903,
      "step": 2192
    },
    {
      "epoch": 0.06318792139687662,
      "grad_norm": 2.149482011795044,
      "learning_rate": 9.97118349131982e-06,
      "loss": 1.0014,
      "step": 2193
    },
    {
      "epoch": 0.0632167348585259,
      "grad_norm": 2.244758367538452,
      "learning_rate": 9.971133445698908e-06,
      "loss": 0.953,
      "step": 2194
    },
    {
      "epoch": 0.06324554832017519,
      "grad_norm": 2.366865396499634,
      "learning_rate": 9.971083356784394e-06,
      "loss": 0.9975,
      "step": 2195
    },
    {
      "epoch": 0.06327436178182447,
      "grad_norm": 2.485238552093506,
      "learning_rate": 9.971033224576714e-06,
      "loss": 1.0006,
      "step": 2196
    },
    {
      "epoch": 0.06330317524347374,
      "grad_norm": 2.4203362464904785,
      "learning_rate": 9.970983049076303e-06,
      "loss": 1.0078,
      "step": 2197
    },
    {
      "epoch": 0.06333198870512304,
      "grad_norm": 2.5232858657836914,
      "learning_rate": 9.970932830283598e-06,
      "loss": 0.9862,
      "step": 2198
    },
    {
      "epoch": 0.06336080216677231,
      "grad_norm": 2.317943572998047,
      "learning_rate": 9.97088256819904e-06,
      "loss": 0.9601,
      "step": 2199
    },
    {
      "epoch": 0.0633896156284216,
      "grad_norm": 2.3899691104888916,
      "learning_rate": 9.97083226282306e-06,
      "loss": 0.9901,
      "step": 2200
    },
    {
      "epoch": 0.06341842909007088,
      "grad_norm": 2.2252197265625,
      "learning_rate": 9.970781914156101e-06,
      "loss": 1.0239,
      "step": 2201
    },
    {
      "epoch": 0.06344724255172017,
      "grad_norm": 2.3731799125671387,
      "learning_rate": 9.970731522198603e-06,
      "loss": 0.9978,
      "step": 2202
    },
    {
      "epoch": 0.06347605601336945,
      "grad_norm": 2.313272714614868,
      "learning_rate": 9.970681086950999e-06,
      "loss": 1.0004,
      "step": 2203
    },
    {
      "epoch": 0.06350486947501872,
      "grad_norm": 2.350519895553589,
      "learning_rate": 9.970630608413733e-06,
      "loss": 0.9971,
      "step": 2204
    },
    {
      "epoch": 0.06353368293666802,
      "grad_norm": 2.4248523712158203,
      "learning_rate": 9.970580086587244e-06,
      "loss": 1.0217,
      "step": 2205
    },
    {
      "epoch": 0.06356249639831729,
      "grad_norm": 2.289583683013916,
      "learning_rate": 9.97052952147197e-06,
      "loss": 1.0015,
      "step": 2206
    },
    {
      "epoch": 0.06359130985996658,
      "grad_norm": 2.299117088317871,
      "learning_rate": 9.970478913068353e-06,
      "loss": 0.9961,
      "step": 2207
    },
    {
      "epoch": 0.06362012332161586,
      "grad_norm": 2.55357027053833,
      "learning_rate": 9.970428261376834e-06,
      "loss": 0.9985,
      "step": 2208
    },
    {
      "epoch": 0.06364893678326514,
      "grad_norm": 2.0366322994232178,
      "learning_rate": 9.970377566397853e-06,
      "loss": 1.0017,
      "step": 2209
    },
    {
      "epoch": 0.06367775024491443,
      "grad_norm": 2.4736056327819824,
      "learning_rate": 9.970326828131852e-06,
      "loss": 0.9834,
      "step": 2210
    },
    {
      "epoch": 0.0637065637065637,
      "grad_norm": 2.4435153007507324,
      "learning_rate": 9.970276046579273e-06,
      "loss": 1.0117,
      "step": 2211
    },
    {
      "epoch": 0.063735377168213,
      "grad_norm": 2.14302921295166,
      "learning_rate": 9.970225221740557e-06,
      "loss": 0.9565,
      "step": 2212
    },
    {
      "epoch": 0.06376419062986227,
      "grad_norm": 2.5697522163391113,
      "learning_rate": 9.970174353616149e-06,
      "loss": 1.0194,
      "step": 2213
    },
    {
      "epoch": 0.06379300409151155,
      "grad_norm": 2.2716758251190186,
      "learning_rate": 9.97012344220649e-06,
      "loss": 1.0384,
      "step": 2214
    },
    {
      "epoch": 0.06382181755316084,
      "grad_norm": 2.421668529510498,
      "learning_rate": 9.970072487512025e-06,
      "loss": 0.9849,
      "step": 2215
    },
    {
      "epoch": 0.06385063101481012,
      "grad_norm": 2.2636516094207764,
      "learning_rate": 9.970021489533197e-06,
      "loss": 0.9995,
      "step": 2216
    },
    {
      "epoch": 0.06387944447645941,
      "grad_norm": 2.439093589782715,
      "learning_rate": 9.969970448270448e-06,
      "loss": 1.0175,
      "step": 2217
    },
    {
      "epoch": 0.06390825793810868,
      "grad_norm": 2.335186243057251,
      "learning_rate": 9.969919363724226e-06,
      "loss": 0.9948,
      "step": 2218
    },
    {
      "epoch": 0.06393707139975796,
      "grad_norm": 2.348343849182129,
      "learning_rate": 9.969868235894974e-06,
      "loss": 1.0125,
      "step": 2219
    },
    {
      "epoch": 0.06396588486140725,
      "grad_norm": 2.3001255989074707,
      "learning_rate": 9.969817064783139e-06,
      "loss": 0.9831,
      "step": 2220
    },
    {
      "epoch": 0.06399469832305653,
      "grad_norm": 2.3527133464813232,
      "learning_rate": 9.969765850389164e-06,
      "loss": 0.9853,
      "step": 2221
    },
    {
      "epoch": 0.06402351178470582,
      "grad_norm": 2.5064361095428467,
      "learning_rate": 9.969714592713497e-06,
      "loss": 0.9886,
      "step": 2222
    },
    {
      "epoch": 0.0640523252463551,
      "grad_norm": 2.4801273345947266,
      "learning_rate": 9.969663291756585e-06,
      "loss": 1.0064,
      "step": 2223
    },
    {
      "epoch": 0.06408113870800437,
      "grad_norm": 2.5219790935516357,
      "learning_rate": 9.969611947518872e-06,
      "loss": 1.0088,
      "step": 2224
    },
    {
      "epoch": 0.06410995216965366,
      "grad_norm": 2.463270902633667,
      "learning_rate": 9.969560560000806e-06,
      "loss": 0.9848,
      "step": 2225
    },
    {
      "epoch": 0.06413876563130294,
      "grad_norm": 2.6318769454956055,
      "learning_rate": 9.969509129202836e-06,
      "loss": 1.022,
      "step": 2226
    },
    {
      "epoch": 0.06416757909295223,
      "grad_norm": 2.5788981914520264,
      "learning_rate": 9.96945765512541e-06,
      "loss": 1.0199,
      "step": 2227
    },
    {
      "epoch": 0.06419639255460151,
      "grad_norm": 2.313776969909668,
      "learning_rate": 9.969406137768975e-06,
      "loss": 1.0481,
      "step": 2228
    },
    {
      "epoch": 0.0642252060162508,
      "grad_norm": 2.504612445831299,
      "learning_rate": 9.969354577133978e-06,
      "loss": 0.9703,
      "step": 2229
    },
    {
      "epoch": 0.06425401947790008,
      "grad_norm": 2.278928756713867,
      "learning_rate": 9.969302973220873e-06,
      "loss": 1.0008,
      "step": 2230
    },
    {
      "epoch": 0.06428283293954935,
      "grad_norm": 2.4211173057556152,
      "learning_rate": 9.969251326030105e-06,
      "loss": 1.0122,
      "step": 2231
    },
    {
      "epoch": 0.06431164640119864,
      "grad_norm": 2.261854887008667,
      "learning_rate": 9.969199635562127e-06,
      "loss": 1.0206,
      "step": 2232
    },
    {
      "epoch": 0.06434045986284792,
      "grad_norm": 2.4256670475006104,
      "learning_rate": 9.969147901817386e-06,
      "loss": 1.0339,
      "step": 2233
    },
    {
      "epoch": 0.06436927332449721,
      "grad_norm": 2.365610122680664,
      "learning_rate": 9.969096124796335e-06,
      "loss": 0.9857,
      "step": 2234
    },
    {
      "epoch": 0.06439808678614649,
      "grad_norm": 2.153628349304199,
      "learning_rate": 9.969044304499425e-06,
      "loss": 1.0152,
      "step": 2235
    },
    {
      "epoch": 0.06442690024779577,
      "grad_norm": 2.51027512550354,
      "learning_rate": 9.968992440927104e-06,
      "loss": 1.0005,
      "step": 2236
    },
    {
      "epoch": 0.06445571370944506,
      "grad_norm": 2.2169008255004883,
      "learning_rate": 9.968940534079828e-06,
      "loss": 0.9746,
      "step": 2237
    },
    {
      "epoch": 0.06448452717109433,
      "grad_norm": 2.4697530269622803,
      "learning_rate": 9.968888583958045e-06,
      "loss": 1.0117,
      "step": 2238
    },
    {
      "epoch": 0.06451334063274362,
      "grad_norm": 2.1610100269317627,
      "learning_rate": 9.968836590562211e-06,
      "loss": 0.9781,
      "step": 2239
    },
    {
      "epoch": 0.0645421540943929,
      "grad_norm": 2.2230122089385986,
      "learning_rate": 9.968784553892777e-06,
      "loss": 0.9968,
      "step": 2240
    },
    {
      "epoch": 0.06457096755604218,
      "grad_norm": 2.2491040229797363,
      "learning_rate": 9.968732473950197e-06,
      "loss": 1.0135,
      "step": 2241
    },
    {
      "epoch": 0.06459978101769147,
      "grad_norm": 2.162461042404175,
      "learning_rate": 9.968680350734922e-06,
      "loss": 0.9939,
      "step": 2242
    },
    {
      "epoch": 0.06462859447934075,
      "grad_norm": 2.2360246181488037,
      "learning_rate": 9.968628184247411e-06,
      "loss": 0.9819,
      "step": 2243
    },
    {
      "epoch": 0.06465740794099004,
      "grad_norm": 2.396108865737915,
      "learning_rate": 9.968575974488114e-06,
      "loss": 0.9973,
      "step": 2244
    },
    {
      "epoch": 0.06468622140263931,
      "grad_norm": 2.1854262351989746,
      "learning_rate": 9.968523721457486e-06,
      "loss": 0.9926,
      "step": 2245
    },
    {
      "epoch": 0.06471503486428859,
      "grad_norm": 2.2523980140686035,
      "learning_rate": 9.968471425155986e-06,
      "loss": 1.0202,
      "step": 2246
    },
    {
      "epoch": 0.06474384832593788,
      "grad_norm": 2.4918158054351807,
      "learning_rate": 9.968419085584063e-06,
      "loss": 0.9885,
      "step": 2247
    },
    {
      "epoch": 0.06477266178758716,
      "grad_norm": 2.2799835205078125,
      "learning_rate": 9.968366702742179e-06,
      "loss": 0.9981,
      "step": 2248
    },
    {
      "epoch": 0.06480147524923645,
      "grad_norm": 2.5158395767211914,
      "learning_rate": 9.968314276630787e-06,
      "loss": 0.9736,
      "step": 2249
    },
    {
      "epoch": 0.06483028871088573,
      "grad_norm": 2.4810597896575928,
      "learning_rate": 9.968261807250343e-06,
      "loss": 0.9976,
      "step": 2250
    },
    {
      "epoch": 0.064859102172535,
      "grad_norm": 2.7422144412994385,
      "learning_rate": 9.968209294601305e-06,
      "loss": 0.9951,
      "step": 2251
    },
    {
      "epoch": 0.06488791563418429,
      "grad_norm": 2.4038619995117188,
      "learning_rate": 9.968156738684132e-06,
      "loss": 0.9844,
      "step": 2252
    },
    {
      "epoch": 0.06491672909583357,
      "grad_norm": 2.627668857574463,
      "learning_rate": 9.968104139499279e-06,
      "loss": 1.0155,
      "step": 2253
    },
    {
      "epoch": 0.06494554255748286,
      "grad_norm": 2.498142719268799,
      "learning_rate": 9.968051497047205e-06,
      "loss": 1.0103,
      "step": 2254
    },
    {
      "epoch": 0.06497435601913214,
      "grad_norm": 2.337707996368408,
      "learning_rate": 9.96799881132837e-06,
      "loss": 1.0146,
      "step": 2255
    },
    {
      "epoch": 0.06500316948078141,
      "grad_norm": 2.631868362426758,
      "learning_rate": 9.96794608234323e-06,
      "loss": 1.0076,
      "step": 2256
    },
    {
      "epoch": 0.0650319829424307,
      "grad_norm": 2.4020469188690186,
      "learning_rate": 9.967893310092247e-06,
      "loss": 1.0024,
      "step": 2257
    },
    {
      "epoch": 0.06506079640407998,
      "grad_norm": 2.5079033374786377,
      "learning_rate": 9.96784049457588e-06,
      "loss": 0.9969,
      "step": 2258
    },
    {
      "epoch": 0.06508960986572927,
      "grad_norm": 2.3301703929901123,
      "learning_rate": 9.967787635794586e-06,
      "loss": 0.9903,
      "step": 2259
    },
    {
      "epoch": 0.06511842332737855,
      "grad_norm": 2.3317184448242188,
      "learning_rate": 9.967734733748829e-06,
      "loss": 1.0051,
      "step": 2260
    },
    {
      "epoch": 0.06514723678902784,
      "grad_norm": 2.3600921630859375,
      "learning_rate": 9.967681788439068e-06,
      "loss": 1.0118,
      "step": 2261
    },
    {
      "epoch": 0.06517605025067712,
      "grad_norm": 2.263029098510742,
      "learning_rate": 9.967628799865764e-06,
      "loss": 0.9925,
      "step": 2262
    },
    {
      "epoch": 0.0652048637123264,
      "grad_norm": 2.2756733894348145,
      "learning_rate": 9.96757576802938e-06,
      "loss": 1.0171,
      "step": 2263
    },
    {
      "epoch": 0.06523367717397568,
      "grad_norm": 2.3849780559539795,
      "learning_rate": 9.967522692930377e-06,
      "loss": 1.0061,
      "step": 2264
    },
    {
      "epoch": 0.06526249063562496,
      "grad_norm": 2.4640724658966064,
      "learning_rate": 9.967469574569217e-06,
      "loss": 1.0204,
      "step": 2265
    },
    {
      "epoch": 0.06529130409727425,
      "grad_norm": 2.3183631896972656,
      "learning_rate": 9.967416412946362e-06,
      "loss": 1.0119,
      "step": 2266
    },
    {
      "epoch": 0.06532011755892353,
      "grad_norm": 2.267005681991577,
      "learning_rate": 9.967363208062276e-06,
      "loss": 0.9856,
      "step": 2267
    },
    {
      "epoch": 0.0653489310205728,
      "grad_norm": 2.4344613552093506,
      "learning_rate": 9.967309959917421e-06,
      "loss": 1.0114,
      "step": 2268
    },
    {
      "epoch": 0.0653777444822221,
      "grad_norm": 2.30772066116333,
      "learning_rate": 9.967256668512263e-06,
      "loss": 0.9914,
      "step": 2269
    },
    {
      "epoch": 0.06540655794387137,
      "grad_norm": 2.305413246154785,
      "learning_rate": 9.967203333847264e-06,
      "loss": 1.0099,
      "step": 2270
    },
    {
      "epoch": 0.06543537140552066,
      "grad_norm": 2.39780855178833,
      "learning_rate": 9.96714995592289e-06,
      "loss": 1.0216,
      "step": 2271
    },
    {
      "epoch": 0.06546418486716994,
      "grad_norm": 2.300901174545288,
      "learning_rate": 9.967096534739605e-06,
      "loss": 1.0278,
      "step": 2272
    },
    {
      "epoch": 0.06549299832881922,
      "grad_norm": 2.2505292892456055,
      "learning_rate": 9.967043070297875e-06,
      "loss": 0.9895,
      "step": 2273
    },
    {
      "epoch": 0.06552181179046851,
      "grad_norm": 2.3762784004211426,
      "learning_rate": 9.966989562598163e-06,
      "loss": 0.9966,
      "step": 2274
    },
    {
      "epoch": 0.06555062525211779,
      "grad_norm": 2.151771306991577,
      "learning_rate": 9.96693601164094e-06,
      "loss": 0.9925,
      "step": 2275
    },
    {
      "epoch": 0.06557943871376708,
      "grad_norm": 2.211430788040161,
      "learning_rate": 9.96688241742667e-06,
      "loss": 0.9956,
      "step": 2276
    },
    {
      "epoch": 0.06560825217541635,
      "grad_norm": 2.184471607208252,
      "learning_rate": 9.966828779955814e-06,
      "loss": 1.0089,
      "step": 2277
    },
    {
      "epoch": 0.06563706563706563,
      "grad_norm": 2.393521308898926,
      "learning_rate": 9.966775099228848e-06,
      "loss": 1.0386,
      "step": 2278
    },
    {
      "epoch": 0.06566587909871492,
      "grad_norm": 2.233767032623291,
      "learning_rate": 9.966721375246236e-06,
      "loss": 0.9838,
      "step": 2279
    },
    {
      "epoch": 0.0656946925603642,
      "grad_norm": 2.3610880374908447,
      "learning_rate": 9.966667608008446e-06,
      "loss": 0.9896,
      "step": 2280
    },
    {
      "epoch": 0.06572350602201349,
      "grad_norm": 2.4014058113098145,
      "learning_rate": 9.966613797515944e-06,
      "loss": 1.0161,
      "step": 2281
    },
    {
      "epoch": 0.06575231948366277,
      "grad_norm": 2.422654390335083,
      "learning_rate": 9.966559943769202e-06,
      "loss": 0.996,
      "step": 2282
    },
    {
      "epoch": 0.06578113294531204,
      "grad_norm": 2.338836193084717,
      "learning_rate": 9.966506046768689e-06,
      "loss": 0.9765,
      "step": 2283
    },
    {
      "epoch": 0.06580994640696133,
      "grad_norm": 2.431084156036377,
      "learning_rate": 9.96645210651487e-06,
      "loss": 1.0273,
      "step": 2284
    },
    {
      "epoch": 0.06583875986861061,
      "grad_norm": 2.256260871887207,
      "learning_rate": 9.966398123008221e-06,
      "loss": 0.9775,
      "step": 2285
    },
    {
      "epoch": 0.0658675733302599,
      "grad_norm": 2.2773828506469727,
      "learning_rate": 9.966344096249206e-06,
      "loss": 1.0039,
      "step": 2286
    },
    {
      "epoch": 0.06589638679190918,
      "grad_norm": 2.430891990661621,
      "learning_rate": 9.9662900262383e-06,
      "loss": 0.9925,
      "step": 2287
    },
    {
      "epoch": 0.06592520025355847,
      "grad_norm": 2.594332695007324,
      "learning_rate": 9.966235912975972e-06,
      "loss": 0.9883,
      "step": 2288
    },
    {
      "epoch": 0.06595401371520775,
      "grad_norm": 2.5645346641540527,
      "learning_rate": 9.966181756462693e-06,
      "loss": 1.017,
      "step": 2289
    },
    {
      "epoch": 0.06598282717685702,
      "grad_norm": 2.2081356048583984,
      "learning_rate": 9.966127556698937e-06,
      "loss": 1.006,
      "step": 2290
    },
    {
      "epoch": 0.06601164063850631,
      "grad_norm": 2.4850032329559326,
      "learning_rate": 9.966073313685173e-06,
      "loss": 1.041,
      "step": 2291
    },
    {
      "epoch": 0.06604045410015559,
      "grad_norm": 2.3801214694976807,
      "learning_rate": 9.966019027421875e-06,
      "loss": 1.0188,
      "step": 2292
    },
    {
      "epoch": 0.06606926756180488,
      "grad_norm": 2.293879508972168,
      "learning_rate": 9.965964697909514e-06,
      "loss": 0.9726,
      "step": 2293
    },
    {
      "epoch": 0.06609808102345416,
      "grad_norm": 2.463852643966675,
      "learning_rate": 9.965910325148567e-06,
      "loss": 1.0054,
      "step": 2294
    },
    {
      "epoch": 0.06612689448510344,
      "grad_norm": 2.331894636154175,
      "learning_rate": 9.965855909139503e-06,
      "loss": 0.9963,
      "step": 2295
    },
    {
      "epoch": 0.06615570794675273,
      "grad_norm": 2.429931163787842,
      "learning_rate": 9.9658014498828e-06,
      "loss": 0.9979,
      "step": 2296
    },
    {
      "epoch": 0.066184521408402,
      "grad_norm": 2.2885518074035645,
      "learning_rate": 9.965746947378929e-06,
      "loss": 1.0167,
      "step": 2297
    },
    {
      "epoch": 0.0662133348700513,
      "grad_norm": 2.5745580196380615,
      "learning_rate": 9.965692401628368e-06,
      "loss": 1.0398,
      "step": 2298
    },
    {
      "epoch": 0.06624214833170057,
      "grad_norm": 2.327693223953247,
      "learning_rate": 9.965637812631588e-06,
      "loss": 0.997,
      "step": 2299
    },
    {
      "epoch": 0.06627096179334985,
      "grad_norm": 2.473015069961548,
      "learning_rate": 9.965583180389066e-06,
      "loss": 1.0213,
      "step": 2300
    },
    {
      "epoch": 0.06629977525499914,
      "grad_norm": 2.3111069202423096,
      "learning_rate": 9.965528504901281e-06,
      "loss": 1.0291,
      "step": 2301
    },
    {
      "epoch": 0.06632858871664842,
      "grad_norm": 2.520678758621216,
      "learning_rate": 9.965473786168705e-06,
      "loss": 1.0003,
      "step": 2302
    },
    {
      "epoch": 0.0663574021782977,
      "grad_norm": 2.4146604537963867,
      "learning_rate": 9.965419024191815e-06,
      "loss": 1.0066,
      "step": 2303
    },
    {
      "epoch": 0.06638621563994698,
      "grad_norm": 2.3836426734924316,
      "learning_rate": 9.96536421897109e-06,
      "loss": 0.9918,
      "step": 2304
    },
    {
      "epoch": 0.06641502910159626,
      "grad_norm": 2.345244884490967,
      "learning_rate": 9.965309370507006e-06,
      "loss": 1.0269,
      "step": 2305
    },
    {
      "epoch": 0.06644384256324555,
      "grad_norm": 2.248009443283081,
      "learning_rate": 9.96525447880004e-06,
      "loss": 1.0081,
      "step": 2306
    },
    {
      "epoch": 0.06647265602489483,
      "grad_norm": 2.5021932125091553,
      "learning_rate": 9.965199543850671e-06,
      "loss": 0.9801,
      "step": 2307
    },
    {
      "epoch": 0.06650146948654412,
      "grad_norm": 2.530040979385376,
      "learning_rate": 9.965144565659379e-06,
      "loss": 1.0261,
      "step": 2308
    },
    {
      "epoch": 0.0665302829481934,
      "grad_norm": 2.277308225631714,
      "learning_rate": 9.96508954422664e-06,
      "loss": 1.0085,
      "step": 2309
    },
    {
      "epoch": 0.06655909640984267,
      "grad_norm": 2.5646347999572754,
      "learning_rate": 9.965034479552936e-06,
      "loss": 1.0018,
      "step": 2310
    },
    {
      "epoch": 0.06658790987149196,
      "grad_norm": 2.194056749343872,
      "learning_rate": 9.964979371638743e-06,
      "loss": 0.9874,
      "step": 2311
    },
    {
      "epoch": 0.06661672333314124,
      "grad_norm": 2.563389778137207,
      "learning_rate": 9.964924220484542e-06,
      "loss": 0.9734,
      "step": 2312
    },
    {
      "epoch": 0.06664553679479053,
      "grad_norm": 2.3731930255889893,
      "learning_rate": 9.964869026090816e-06,
      "loss": 1.0106,
      "step": 2313
    },
    {
      "epoch": 0.06667435025643981,
      "grad_norm": 2.3923795223236084,
      "learning_rate": 9.964813788458043e-06,
      "loss": 1.0056,
      "step": 2314
    },
    {
      "epoch": 0.0667031637180891,
      "grad_norm": 2.395941734313965,
      "learning_rate": 9.964758507586704e-06,
      "loss": 1.0143,
      "step": 2315
    },
    {
      "epoch": 0.06673197717973837,
      "grad_norm": 2.338829278945923,
      "learning_rate": 9.964703183477285e-06,
      "loss": 0.9649,
      "step": 2316
    },
    {
      "epoch": 0.06676079064138765,
      "grad_norm": 2.3717777729034424,
      "learning_rate": 9.964647816130261e-06,
      "loss": 0.9825,
      "step": 2317
    },
    {
      "epoch": 0.06678960410303694,
      "grad_norm": 2.460930824279785,
      "learning_rate": 9.964592405546117e-06,
      "loss": 1.0218,
      "step": 2318
    },
    {
      "epoch": 0.06681841756468622,
      "grad_norm": 2.2703909873962402,
      "learning_rate": 9.964536951725338e-06,
      "loss": 0.9875,
      "step": 2319
    },
    {
      "epoch": 0.06684723102633551,
      "grad_norm": 2.2921559810638428,
      "learning_rate": 9.964481454668403e-06,
      "loss": 0.9817,
      "step": 2320
    },
    {
      "epoch": 0.06687604448798479,
      "grad_norm": 2.1506032943725586,
      "learning_rate": 9.964425914375798e-06,
      "loss": 1.0176,
      "step": 2321
    },
    {
      "epoch": 0.06690485794963406,
      "grad_norm": 2.3000268936157227,
      "learning_rate": 9.964370330848005e-06,
      "loss": 0.9985,
      "step": 2322
    },
    {
      "epoch": 0.06693367141128335,
      "grad_norm": 2.1602931022644043,
      "learning_rate": 9.964314704085511e-06,
      "loss": 0.9904,
      "step": 2323
    },
    {
      "epoch": 0.06696248487293263,
      "grad_norm": 2.115161657333374,
      "learning_rate": 9.964259034088797e-06,
      "loss": 0.9849,
      "step": 2324
    },
    {
      "epoch": 0.06699129833458192,
      "grad_norm": 2.1418585777282715,
      "learning_rate": 9.96420332085835e-06,
      "loss": 0.971,
      "step": 2325
    },
    {
      "epoch": 0.0670201117962312,
      "grad_norm": 2.3251237869262695,
      "learning_rate": 9.964147564394653e-06,
      "loss": 0.9928,
      "step": 2326
    },
    {
      "epoch": 0.06704892525788048,
      "grad_norm": 2.1157989501953125,
      "learning_rate": 9.964091764698194e-06,
      "loss": 0.9855,
      "step": 2327
    },
    {
      "epoch": 0.06707773871952977,
      "grad_norm": 2.2800559997558594,
      "learning_rate": 9.96403592176946e-06,
      "loss": 1.0418,
      "step": 2328
    },
    {
      "epoch": 0.06710655218117904,
      "grad_norm": 2.1833837032318115,
      "learning_rate": 9.963980035608931e-06,
      "loss": 0.9855,
      "step": 2329
    },
    {
      "epoch": 0.06713536564282833,
      "grad_norm": 2.138303279876709,
      "learning_rate": 9.963924106217102e-06,
      "loss": 0.995,
      "step": 2330
    },
    {
      "epoch": 0.06716417910447761,
      "grad_norm": 2.1101458072662354,
      "learning_rate": 9.963868133594454e-06,
      "loss": 1.0026,
      "step": 2331
    },
    {
      "epoch": 0.06719299256612689,
      "grad_norm": 2.053072214126587,
      "learning_rate": 9.963812117741478e-06,
      "loss": 0.9763,
      "step": 2332
    },
    {
      "epoch": 0.06722180602777618,
      "grad_norm": 2.336707353591919,
      "learning_rate": 9.96375605865866e-06,
      "loss": 1.0002,
      "step": 2333
    },
    {
      "epoch": 0.06725061948942546,
      "grad_norm": 2.1566178798675537,
      "learning_rate": 9.963699956346488e-06,
      "loss": 1.0351,
      "step": 2334
    },
    {
      "epoch": 0.06727943295107475,
      "grad_norm": 2.1971006393432617,
      "learning_rate": 9.963643810805451e-06,
      "loss": 0.9716,
      "step": 2335
    },
    {
      "epoch": 0.06730824641272402,
      "grad_norm": 2.2108540534973145,
      "learning_rate": 9.96358762203604e-06,
      "loss": 1.0117,
      "step": 2336
    },
    {
      "epoch": 0.0673370598743733,
      "grad_norm": 2.4743494987487793,
      "learning_rate": 9.963531390038742e-06,
      "loss": 1.0122,
      "step": 2337
    },
    {
      "epoch": 0.06736587333602259,
      "grad_norm": 2.2054388523101807,
      "learning_rate": 9.963475114814046e-06,
      "loss": 0.9966,
      "step": 2338
    },
    {
      "epoch": 0.06739468679767187,
      "grad_norm": 2.4240951538085938,
      "learning_rate": 9.963418796362443e-06,
      "loss": 1.0203,
      "step": 2339
    },
    {
      "epoch": 0.06742350025932116,
      "grad_norm": 2.302229881286621,
      "learning_rate": 9.963362434684425e-06,
      "loss": 0.9659,
      "step": 2340
    },
    {
      "epoch": 0.06745231372097044,
      "grad_norm": 2.31769061088562,
      "learning_rate": 9.963306029780482e-06,
      "loss": 0.9965,
      "step": 2341
    },
    {
      "epoch": 0.06748112718261973,
      "grad_norm": 2.198381185531616,
      "learning_rate": 9.963249581651105e-06,
      "loss": 0.9774,
      "step": 2342
    },
    {
      "epoch": 0.067509940644269,
      "grad_norm": 2.3048808574676514,
      "learning_rate": 9.963193090296786e-06,
      "loss": 0.9571,
      "step": 2343
    },
    {
      "epoch": 0.06753875410591828,
      "grad_norm": 2.3469507694244385,
      "learning_rate": 9.963136555718015e-06,
      "loss": 1.0151,
      "step": 2344
    },
    {
      "epoch": 0.06756756756756757,
      "grad_norm": 2.3178048133850098,
      "learning_rate": 9.963079977915286e-06,
      "loss": 0.9796,
      "step": 2345
    },
    {
      "epoch": 0.06759638102921685,
      "grad_norm": 2.258331298828125,
      "learning_rate": 9.963023356889093e-06,
      "loss": 0.9812,
      "step": 2346
    },
    {
      "epoch": 0.06762519449086614,
      "grad_norm": 2.405698299407959,
      "learning_rate": 9.962966692639928e-06,
      "loss": 1.0025,
      "step": 2347
    },
    {
      "epoch": 0.06765400795251542,
      "grad_norm": 2.2928028106689453,
      "learning_rate": 9.962909985168283e-06,
      "loss": 1.018,
      "step": 2348
    },
    {
      "epoch": 0.06768282141416469,
      "grad_norm": 2.464587450027466,
      "learning_rate": 9.962853234474655e-06,
      "loss": 1.0007,
      "step": 2349
    },
    {
      "epoch": 0.06771163487581398,
      "grad_norm": 2.3973207473754883,
      "learning_rate": 9.962796440559533e-06,
      "loss": 0.9861,
      "step": 2350
    },
    {
      "epoch": 0.06774044833746326,
      "grad_norm": 2.48561429977417,
      "learning_rate": 9.962739603423417e-06,
      "loss": 1.0131,
      "step": 2351
    },
    {
      "epoch": 0.06776926179911255,
      "grad_norm": 2.0428848266601562,
      "learning_rate": 9.962682723066802e-06,
      "loss": 0.9961,
      "step": 2352
    },
    {
      "epoch": 0.06779807526076183,
      "grad_norm": 2.544642686843872,
      "learning_rate": 9.962625799490177e-06,
      "loss": 1.0017,
      "step": 2353
    },
    {
      "epoch": 0.0678268887224111,
      "grad_norm": 2.331752300262451,
      "learning_rate": 9.962568832694046e-06,
      "loss": 0.9945,
      "step": 2354
    },
    {
      "epoch": 0.0678557021840604,
      "grad_norm": 2.4162399768829346,
      "learning_rate": 9.962511822678898e-06,
      "loss": 0.9889,
      "step": 2355
    },
    {
      "epoch": 0.06788451564570967,
      "grad_norm": 2.4295403957366943,
      "learning_rate": 9.962454769445235e-06,
      "loss": 1.0085,
      "step": 2356
    },
    {
      "epoch": 0.06791332910735896,
      "grad_norm": 2.4352684020996094,
      "learning_rate": 9.962397672993552e-06,
      "loss": 0.9877,
      "step": 2357
    },
    {
      "epoch": 0.06794214256900824,
      "grad_norm": 2.2295656204223633,
      "learning_rate": 9.962340533324343e-06,
      "loss": 0.9958,
      "step": 2358
    },
    {
      "epoch": 0.06797095603065752,
      "grad_norm": 2.5118632316589355,
      "learning_rate": 9.962283350438111e-06,
      "loss": 1.0253,
      "step": 2359
    },
    {
      "epoch": 0.06799976949230681,
      "grad_norm": 2.2150731086730957,
      "learning_rate": 9.96222612433535e-06,
      "loss": 1.0063,
      "step": 2360
    },
    {
      "epoch": 0.06802858295395608,
      "grad_norm": 2.485504627227783,
      "learning_rate": 9.962168855016561e-06,
      "loss": 0.9929,
      "step": 2361
    },
    {
      "epoch": 0.06805739641560538,
      "grad_norm": 2.4695019721984863,
      "learning_rate": 9.962111542482242e-06,
      "loss": 0.992,
      "step": 2362
    },
    {
      "epoch": 0.06808620987725465,
      "grad_norm": 2.5083396434783936,
      "learning_rate": 9.96205418673289e-06,
      "loss": 1.0207,
      "step": 2363
    },
    {
      "epoch": 0.06811502333890393,
      "grad_norm": 2.5082075595855713,
      "learning_rate": 9.961996787769008e-06,
      "loss": 0.99,
      "step": 2364
    },
    {
      "epoch": 0.06814383680055322,
      "grad_norm": 2.296293020248413,
      "learning_rate": 9.961939345591093e-06,
      "loss": 0.9994,
      "step": 2365
    },
    {
      "epoch": 0.0681726502622025,
      "grad_norm": 2.216867685317993,
      "learning_rate": 9.961881860199647e-06,
      "loss": 1.0124,
      "step": 2366
    },
    {
      "epoch": 0.06820146372385179,
      "grad_norm": 2.368764638900757,
      "learning_rate": 9.96182433159517e-06,
      "loss": 0.9974,
      "step": 2367
    },
    {
      "epoch": 0.06823027718550106,
      "grad_norm": 2.276560068130493,
      "learning_rate": 9.961766759778163e-06,
      "loss": 1.0042,
      "step": 2368
    },
    {
      "epoch": 0.06825909064715036,
      "grad_norm": 2.581339120864868,
      "learning_rate": 9.961709144749128e-06,
      "loss": 0.9975,
      "step": 2369
    },
    {
      "epoch": 0.06828790410879963,
      "grad_norm": 2.2534701824188232,
      "learning_rate": 9.961651486508565e-06,
      "loss": 0.9351,
      "step": 2370
    },
    {
      "epoch": 0.06831671757044891,
      "grad_norm": 2.3229072093963623,
      "learning_rate": 9.961593785056978e-06,
      "loss": 0.9925,
      "step": 2371
    },
    {
      "epoch": 0.0683455310320982,
      "grad_norm": 2.324479579925537,
      "learning_rate": 9.961536040394869e-06,
      "loss": 0.9913,
      "step": 2372
    },
    {
      "epoch": 0.06837434449374748,
      "grad_norm": 2.266282320022583,
      "learning_rate": 9.96147825252274e-06,
      "loss": 1.026,
      "step": 2373
    },
    {
      "epoch": 0.06840315795539677,
      "grad_norm": 2.4819164276123047,
      "learning_rate": 9.961420421441098e-06,
      "loss": 0.9921,
      "step": 2374
    },
    {
      "epoch": 0.06843197141704604,
      "grad_norm": 2.385321855545044,
      "learning_rate": 9.96136254715044e-06,
      "loss": 0.9877,
      "step": 2375
    },
    {
      "epoch": 0.06846078487869532,
      "grad_norm": 2.2907018661499023,
      "learning_rate": 9.961304629651276e-06,
      "loss": 0.9921,
      "step": 2376
    },
    {
      "epoch": 0.06848959834034461,
      "grad_norm": 2.3878695964813232,
      "learning_rate": 9.961246668944107e-06,
      "loss": 0.9988,
      "step": 2377
    },
    {
      "epoch": 0.06851841180199389,
      "grad_norm": 2.3029823303222656,
      "learning_rate": 9.96118866502944e-06,
      "loss": 1.0057,
      "step": 2378
    },
    {
      "epoch": 0.06854722526364318,
      "grad_norm": 2.365260601043701,
      "learning_rate": 9.961130617907778e-06,
      "loss": 0.9672,
      "step": 2379
    },
    {
      "epoch": 0.06857603872529246,
      "grad_norm": 2.429253101348877,
      "learning_rate": 9.961072527579628e-06,
      "loss": 0.991,
      "step": 2380
    },
    {
      "epoch": 0.06860485218694173,
      "grad_norm": 2.318235397338867,
      "learning_rate": 9.961014394045497e-06,
      "loss": 0.984,
      "step": 2381
    },
    {
      "epoch": 0.06863366564859102,
      "grad_norm": 2.2225685119628906,
      "learning_rate": 9.960956217305889e-06,
      "loss": 1.0117,
      "step": 2382
    },
    {
      "epoch": 0.0686624791102403,
      "grad_norm": 2.5437493324279785,
      "learning_rate": 9.96089799736131e-06,
      "loss": 0.9816,
      "step": 2383
    },
    {
      "epoch": 0.06869129257188959,
      "grad_norm": 2.5952203273773193,
      "learning_rate": 9.960839734212268e-06,
      "loss": 0.9944,
      "step": 2384
    },
    {
      "epoch": 0.06872010603353887,
      "grad_norm": 2.2688469886779785,
      "learning_rate": 9.960781427859273e-06,
      "loss": 0.9593,
      "step": 2385
    },
    {
      "epoch": 0.06874891949518815,
      "grad_norm": 2.2725605964660645,
      "learning_rate": 9.960723078302831e-06,
      "loss": 0.9699,
      "step": 2386
    },
    {
      "epoch": 0.06877773295683744,
      "grad_norm": 2.544497489929199,
      "learning_rate": 9.960664685543449e-06,
      "loss": 1.0234,
      "step": 2387
    },
    {
      "epoch": 0.06880654641848671,
      "grad_norm": 2.4167518615722656,
      "learning_rate": 9.960606249581636e-06,
      "loss": 0.9909,
      "step": 2388
    },
    {
      "epoch": 0.068835359880136,
      "grad_norm": 2.3563225269317627,
      "learning_rate": 9.960547770417903e-06,
      "loss": 0.9813,
      "step": 2389
    },
    {
      "epoch": 0.06886417334178528,
      "grad_norm": 2.29868483543396,
      "learning_rate": 9.960489248052755e-06,
      "loss": 1.0122,
      "step": 2390
    },
    {
      "epoch": 0.06889298680343456,
      "grad_norm": 2.2500627040863037,
      "learning_rate": 9.960430682486708e-06,
      "loss": 0.9991,
      "step": 2391
    },
    {
      "epoch": 0.06892180026508385,
      "grad_norm": 2.304147243499756,
      "learning_rate": 9.960372073720266e-06,
      "loss": 1.0014,
      "step": 2392
    },
    {
      "epoch": 0.06895061372673313,
      "grad_norm": 2.394120216369629,
      "learning_rate": 9.960313421753943e-06,
      "loss": 0.997,
      "step": 2393
    },
    {
      "epoch": 0.06897942718838242,
      "grad_norm": 2.582807779312134,
      "learning_rate": 9.960254726588246e-06,
      "loss": 1.0127,
      "step": 2394
    },
    {
      "epoch": 0.0690082406500317,
      "grad_norm": 2.4091944694519043,
      "learning_rate": 9.960195988223691e-06,
      "loss": 1.0159,
      "step": 2395
    },
    {
      "epoch": 0.06903705411168097,
      "grad_norm": 2.4504408836364746,
      "learning_rate": 9.960137206660787e-06,
      "loss": 0.9893,
      "step": 2396
    },
    {
      "epoch": 0.06906586757333026,
      "grad_norm": 2.1124019622802734,
      "learning_rate": 9.960078381900045e-06,
      "loss": 1.0077,
      "step": 2397
    },
    {
      "epoch": 0.06909468103497954,
      "grad_norm": 2.354612350463867,
      "learning_rate": 9.96001951394198e-06,
      "loss": 0.991,
      "step": 2398
    },
    {
      "epoch": 0.06912349449662883,
      "grad_norm": 2.3628695011138916,
      "learning_rate": 9.959960602787102e-06,
      "loss": 1.0014,
      "step": 2399
    },
    {
      "epoch": 0.0691523079582781,
      "grad_norm": 2.4416167736053467,
      "learning_rate": 9.959901648435927e-06,
      "loss": 1.0208,
      "step": 2400
    },
    {
      "epoch": 0.0691811214199274,
      "grad_norm": 2.315272331237793,
      "learning_rate": 9.959842650888967e-06,
      "loss": 0.9779,
      "step": 2401
    },
    {
      "epoch": 0.06920993488157667,
      "grad_norm": 2.241560459136963,
      "learning_rate": 9.959783610146734e-06,
      "loss": 0.9982,
      "step": 2402
    },
    {
      "epoch": 0.06923874834322595,
      "grad_norm": 2.293191432952881,
      "learning_rate": 9.959724526209745e-06,
      "loss": 0.9892,
      "step": 2403
    },
    {
      "epoch": 0.06926756180487524,
      "grad_norm": 2.3855323791503906,
      "learning_rate": 9.95966539907851e-06,
      "loss": 1.0238,
      "step": 2404
    },
    {
      "epoch": 0.06929637526652452,
      "grad_norm": 2.22379207611084,
      "learning_rate": 9.959606228753551e-06,
      "loss": 0.9572,
      "step": 2405
    },
    {
      "epoch": 0.06932518872817381,
      "grad_norm": 2.5047965049743652,
      "learning_rate": 9.959547015235379e-06,
      "loss": 0.9885,
      "step": 2406
    },
    {
      "epoch": 0.06935400218982309,
      "grad_norm": 2.3664791584014893,
      "learning_rate": 9.959487758524509e-06,
      "loss": 0.9908,
      "step": 2407
    },
    {
      "epoch": 0.06938281565147236,
      "grad_norm": 2.3071131706237793,
      "learning_rate": 9.95942845862146e-06,
      "loss": 0.9922,
      "step": 2408
    },
    {
      "epoch": 0.06941162911312165,
      "grad_norm": 2.344881296157837,
      "learning_rate": 9.959369115526744e-06,
      "loss": 1.0185,
      "step": 2409
    },
    {
      "epoch": 0.06944044257477093,
      "grad_norm": 2.410640001296997,
      "learning_rate": 9.959309729240882e-06,
      "loss": 0.9869,
      "step": 2410
    },
    {
      "epoch": 0.06946925603642022,
      "grad_norm": 2.4388043880462646,
      "learning_rate": 9.959250299764389e-06,
      "loss": 0.9715,
      "step": 2411
    },
    {
      "epoch": 0.0694980694980695,
      "grad_norm": 2.440200090408325,
      "learning_rate": 9.959190827097785e-06,
      "loss": 0.9865,
      "step": 2412
    },
    {
      "epoch": 0.06952688295971877,
      "grad_norm": 2.504075765609741,
      "learning_rate": 9.959131311241586e-06,
      "loss": 0.9875,
      "step": 2413
    },
    {
      "epoch": 0.06955569642136807,
      "grad_norm": 2.396613836288452,
      "learning_rate": 9.95907175219631e-06,
      "loss": 0.9951,
      "step": 2414
    },
    {
      "epoch": 0.06958450988301734,
      "grad_norm": 2.211709976196289,
      "learning_rate": 9.959012149962474e-06,
      "loss": 1.0162,
      "step": 2415
    },
    {
      "epoch": 0.06961332334466663,
      "grad_norm": 2.4441161155700684,
      "learning_rate": 9.958952504540602e-06,
      "loss": 0.9756,
      "step": 2416
    },
    {
      "epoch": 0.06964213680631591,
      "grad_norm": 2.298862934112549,
      "learning_rate": 9.958892815931212e-06,
      "loss": 0.9916,
      "step": 2417
    },
    {
      "epoch": 0.06967095026796519,
      "grad_norm": 2.5355846881866455,
      "learning_rate": 9.958833084134821e-06,
      "loss": 0.9933,
      "step": 2418
    },
    {
      "epoch": 0.06969976372961448,
      "grad_norm": 2.291532039642334,
      "learning_rate": 9.95877330915195e-06,
      "loss": 0.9867,
      "step": 2419
    },
    {
      "epoch": 0.06972857719126375,
      "grad_norm": 2.247281074523926,
      "learning_rate": 9.958713490983121e-06,
      "loss": 0.9552,
      "step": 2420
    },
    {
      "epoch": 0.06975739065291305,
      "grad_norm": 2.144707202911377,
      "learning_rate": 9.958653629628856e-06,
      "loss": 0.9967,
      "step": 2421
    },
    {
      "epoch": 0.06978620411456232,
      "grad_norm": 2.219632387161255,
      "learning_rate": 9.958593725089674e-06,
      "loss": 0.9655,
      "step": 2422
    },
    {
      "epoch": 0.0698150175762116,
      "grad_norm": 2.203809976577759,
      "learning_rate": 9.958533777366095e-06,
      "loss": 0.9975,
      "step": 2423
    },
    {
      "epoch": 0.06984383103786089,
      "grad_norm": 2.112185001373291,
      "learning_rate": 9.958473786458647e-06,
      "loss": 1.0365,
      "step": 2424
    },
    {
      "epoch": 0.06987264449951017,
      "grad_norm": 2.195230484008789,
      "learning_rate": 9.958413752367845e-06,
      "loss": 0.9874,
      "step": 2425
    },
    {
      "epoch": 0.06990145796115946,
      "grad_norm": 2.4456489086151123,
      "learning_rate": 9.95835367509422e-06,
      "loss": 0.9978,
      "step": 2426
    },
    {
      "epoch": 0.06993027142280873,
      "grad_norm": 2.27913498878479,
      "learning_rate": 9.95829355463829e-06,
      "loss": 1.0141,
      "step": 2427
    },
    {
      "epoch": 0.06995908488445803,
      "grad_norm": 2.299072027206421,
      "learning_rate": 9.958233391000577e-06,
      "loss": 1.0158,
      "step": 2428
    },
    {
      "epoch": 0.0699878983461073,
      "grad_norm": 2.262037515640259,
      "learning_rate": 9.95817318418161e-06,
      "loss": 0.9839,
      "step": 2429
    },
    {
      "epoch": 0.07001671180775658,
      "grad_norm": 2.162101984024048,
      "learning_rate": 9.958112934181912e-06,
      "loss": 1.0452,
      "step": 2430
    },
    {
      "epoch": 0.07004552526940587,
      "grad_norm": 2.1540703773498535,
      "learning_rate": 9.958052641002004e-06,
      "loss": 0.9755,
      "step": 2431
    },
    {
      "epoch": 0.07007433873105515,
      "grad_norm": 2.233914613723755,
      "learning_rate": 9.957992304642415e-06,
      "loss": 1.0004,
      "step": 2432
    },
    {
      "epoch": 0.07010315219270444,
      "grad_norm": 2.246392250061035,
      "learning_rate": 9.957931925103669e-06,
      "loss": 0.9935,
      "step": 2433
    },
    {
      "epoch": 0.07013196565435371,
      "grad_norm": 2.133199453353882,
      "learning_rate": 9.957871502386292e-06,
      "loss": 0.9508,
      "step": 2434
    },
    {
      "epoch": 0.07016077911600299,
      "grad_norm": 2.213273286819458,
      "learning_rate": 9.95781103649081e-06,
      "loss": 1.0085,
      "step": 2435
    },
    {
      "epoch": 0.07018959257765228,
      "grad_norm": 2.2520828247070312,
      "learning_rate": 9.957750527417751e-06,
      "loss": 0.9795,
      "step": 2436
    },
    {
      "epoch": 0.07021840603930156,
      "grad_norm": 2.2904229164123535,
      "learning_rate": 9.957689975167639e-06,
      "loss": 0.9915,
      "step": 2437
    },
    {
      "epoch": 0.07024721950095085,
      "grad_norm": 2.2994656562805176,
      "learning_rate": 9.957629379741005e-06,
      "loss": 0.987,
      "step": 2438
    },
    {
      "epoch": 0.07027603296260013,
      "grad_norm": 2.1635196208953857,
      "learning_rate": 9.957568741138374e-06,
      "loss": 0.9872,
      "step": 2439
    },
    {
      "epoch": 0.0703048464242494,
      "grad_norm": 2.3030757904052734,
      "learning_rate": 9.957508059360275e-06,
      "loss": 1.0009,
      "step": 2440
    },
    {
      "epoch": 0.0703336598858987,
      "grad_norm": 2.1158671379089355,
      "learning_rate": 9.957447334407236e-06,
      "loss": 1.0016,
      "step": 2441
    },
    {
      "epoch": 0.07036247334754797,
      "grad_norm": 2.5807528495788574,
      "learning_rate": 9.957386566279788e-06,
      "loss": 1.0058,
      "step": 2442
    },
    {
      "epoch": 0.07039128680919726,
      "grad_norm": 2.208220958709717,
      "learning_rate": 9.957325754978457e-06,
      "loss": 1.0164,
      "step": 2443
    },
    {
      "epoch": 0.07042010027084654,
      "grad_norm": 2.296309232711792,
      "learning_rate": 9.957264900503774e-06,
      "loss": 0.9666,
      "step": 2444
    },
    {
      "epoch": 0.07044891373249582,
      "grad_norm": 2.3795766830444336,
      "learning_rate": 9.95720400285627e-06,
      "loss": 1.0012,
      "step": 2445
    },
    {
      "epoch": 0.0704777271941451,
      "grad_norm": 2.218492031097412,
      "learning_rate": 9.957143062036475e-06,
      "loss": 0.9834,
      "step": 2446
    },
    {
      "epoch": 0.07050654065579438,
      "grad_norm": 2.248262643814087,
      "learning_rate": 9.95708207804492e-06,
      "loss": 1.008,
      "step": 2447
    },
    {
      "epoch": 0.07053535411744367,
      "grad_norm": 2.303840160369873,
      "learning_rate": 9.957021050882134e-06,
      "loss": 1.0139,
      "step": 2448
    },
    {
      "epoch": 0.07056416757909295,
      "grad_norm": 2.3251852989196777,
      "learning_rate": 9.956959980548648e-06,
      "loss": 0.9932,
      "step": 2449
    },
    {
      "epoch": 0.07059298104074223,
      "grad_norm": 2.192722797393799,
      "learning_rate": 9.956898867045e-06,
      "loss": 0.991,
      "step": 2450
    },
    {
      "epoch": 0.07062179450239152,
      "grad_norm": 2.3678998947143555,
      "learning_rate": 9.956837710371714e-06,
      "loss": 0.9739,
      "step": 2451
    },
    {
      "epoch": 0.0706506079640408,
      "grad_norm": 2.3885579109191895,
      "learning_rate": 9.95677651052933e-06,
      "loss": 1.0419,
      "step": 2452
    },
    {
      "epoch": 0.07067942142569009,
      "grad_norm": 2.4410266876220703,
      "learning_rate": 9.956715267518373e-06,
      "loss": 0.9881,
      "step": 2453
    },
    {
      "epoch": 0.07070823488733936,
      "grad_norm": 2.369304656982422,
      "learning_rate": 9.956653981339385e-06,
      "loss": 0.9851,
      "step": 2454
    },
    {
      "epoch": 0.07073704834898865,
      "grad_norm": 2.259614944458008,
      "learning_rate": 9.956592651992893e-06,
      "loss": 0.9859,
      "step": 2455
    },
    {
      "epoch": 0.07076586181063793,
      "grad_norm": 2.417222023010254,
      "learning_rate": 9.956531279479436e-06,
      "loss": 0.9576,
      "step": 2456
    },
    {
      "epoch": 0.07079467527228721,
      "grad_norm": 2.0837509632110596,
      "learning_rate": 9.956469863799545e-06,
      "loss": 1.012,
      "step": 2457
    },
    {
      "epoch": 0.0708234887339365,
      "grad_norm": 2.2801942825317383,
      "learning_rate": 9.956408404953757e-06,
      "loss": 0.9889,
      "step": 2458
    },
    {
      "epoch": 0.07085230219558578,
      "grad_norm": 2.145397901535034,
      "learning_rate": 9.956346902942606e-06,
      "loss": 0.9836,
      "step": 2459
    },
    {
      "epoch": 0.07088111565723507,
      "grad_norm": 2.0775017738342285,
      "learning_rate": 9.95628535776663e-06,
      "loss": 0.9494,
      "step": 2460
    },
    {
      "epoch": 0.07090992911888434,
      "grad_norm": 2.2193963527679443,
      "learning_rate": 9.956223769426358e-06,
      "loss": 0.993,
      "step": 2461
    },
    {
      "epoch": 0.07093874258053362,
      "grad_norm": 2.0982065200805664,
      "learning_rate": 9.956162137922335e-06,
      "loss": 0.9815,
      "step": 2462
    },
    {
      "epoch": 0.07096755604218291,
      "grad_norm": 2.2658863067626953,
      "learning_rate": 9.956100463255093e-06,
      "loss": 1.0235,
      "step": 2463
    },
    {
      "epoch": 0.07099636950383219,
      "grad_norm": 2.1921157836914062,
      "learning_rate": 9.956038745425172e-06,
      "loss": 0.9761,
      "step": 2464
    },
    {
      "epoch": 0.07102518296548148,
      "grad_norm": 2.2493598461151123,
      "learning_rate": 9.955976984433106e-06,
      "loss": 1.0084,
      "step": 2465
    },
    {
      "epoch": 0.07105399642713076,
      "grad_norm": 2.4779586791992188,
      "learning_rate": 9.955915180279435e-06,
      "loss": 1.008,
      "step": 2466
    },
    {
      "epoch": 0.07108280988878003,
      "grad_norm": 2.1713526248931885,
      "learning_rate": 9.955853332964696e-06,
      "loss": 0.9931,
      "step": 2467
    },
    {
      "epoch": 0.07111162335042932,
      "grad_norm": 2.2764899730682373,
      "learning_rate": 9.95579144248943e-06,
      "loss": 0.9972,
      "step": 2468
    },
    {
      "epoch": 0.0711404368120786,
      "grad_norm": 2.6591618061065674,
      "learning_rate": 9.955729508854174e-06,
      "loss": 1.0039,
      "step": 2469
    },
    {
      "epoch": 0.07116925027372789,
      "grad_norm": 2.163245439529419,
      "learning_rate": 9.955667532059468e-06,
      "loss": 0.9684,
      "step": 2470
    },
    {
      "epoch": 0.07119806373537717,
      "grad_norm": 2.3412082195281982,
      "learning_rate": 9.955605512105851e-06,
      "loss": 0.9608,
      "step": 2471
    },
    {
      "epoch": 0.07122687719702644,
      "grad_norm": 2.5296106338500977,
      "learning_rate": 9.955543448993865e-06,
      "loss": 0.9707,
      "step": 2472
    },
    {
      "epoch": 0.07125569065867574,
      "grad_norm": 2.2401037216186523,
      "learning_rate": 9.955481342724048e-06,
      "loss": 1.0001,
      "step": 2473
    },
    {
      "epoch": 0.07128450412032501,
      "grad_norm": 2.3638875484466553,
      "learning_rate": 9.955419193296942e-06,
      "loss": 0.9733,
      "step": 2474
    },
    {
      "epoch": 0.0713133175819743,
      "grad_norm": 2.273171901702881,
      "learning_rate": 9.95535700071309e-06,
      "loss": 0.9924,
      "step": 2475
    },
    {
      "epoch": 0.07134213104362358,
      "grad_norm": 2.3573837280273438,
      "learning_rate": 9.955294764973032e-06,
      "loss": 1.0319,
      "step": 2476
    },
    {
      "epoch": 0.07137094450527286,
      "grad_norm": 2.094520092010498,
      "learning_rate": 9.955232486077309e-06,
      "loss": 0.9605,
      "step": 2477
    },
    {
      "epoch": 0.07139975796692215,
      "grad_norm": 2.2499465942382812,
      "learning_rate": 9.955170164026464e-06,
      "loss": 0.9821,
      "step": 2478
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 2.31689190864563,
      "learning_rate": 9.955107798821041e-06,
      "loss": 0.9996,
      "step": 2479
    },
    {
      "epoch": 0.07145738489022072,
      "grad_norm": 2.351228713989258,
      "learning_rate": 9.955045390461582e-06,
      "loss": 0.9957,
      "step": 2480
    },
    {
      "epoch": 0.07148619835186999,
      "grad_norm": 2.079395294189453,
      "learning_rate": 9.954982938948632e-06,
      "loss": 1.001,
      "step": 2481
    },
    {
      "epoch": 0.07151501181351928,
      "grad_norm": 2.233203172683716,
      "learning_rate": 9.954920444282733e-06,
      "loss": 0.9962,
      "step": 2482
    },
    {
      "epoch": 0.07154382527516856,
      "grad_norm": 2.2618396282196045,
      "learning_rate": 9.95485790646443e-06,
      "loss": 0.9797,
      "step": 2483
    },
    {
      "epoch": 0.07157263873681784,
      "grad_norm": 2.091554880142212,
      "learning_rate": 9.954795325494268e-06,
      "loss": 0.9831,
      "step": 2484
    },
    {
      "epoch": 0.07160145219846713,
      "grad_norm": 2.288332462310791,
      "learning_rate": 9.954732701372792e-06,
      "loss": 0.9921,
      "step": 2485
    },
    {
      "epoch": 0.0716302656601164,
      "grad_norm": 2.2691731452941895,
      "learning_rate": 9.954670034100546e-06,
      "loss": 0.9932,
      "step": 2486
    },
    {
      "epoch": 0.0716590791217657,
      "grad_norm": 2.2996792793273926,
      "learning_rate": 9.954607323678078e-06,
      "loss": 0.9959,
      "step": 2487
    },
    {
      "epoch": 0.07168789258341497,
      "grad_norm": 2.39432954788208,
      "learning_rate": 9.954544570105931e-06,
      "loss": 0.9843,
      "step": 2488
    },
    {
      "epoch": 0.07171670604506425,
      "grad_norm": 2.5189366340637207,
      "learning_rate": 9.954481773384656e-06,
      "loss": 0.9573,
      "step": 2489
    },
    {
      "epoch": 0.07174551950671354,
      "grad_norm": 2.3677167892456055,
      "learning_rate": 9.954418933514796e-06,
      "loss": 0.9591,
      "step": 2490
    },
    {
      "epoch": 0.07177433296836282,
      "grad_norm": 2.4776391983032227,
      "learning_rate": 9.954356050496901e-06,
      "loss": 0.9776,
      "step": 2491
    },
    {
      "epoch": 0.07180314643001211,
      "grad_norm": 2.192394733428955,
      "learning_rate": 9.954293124331516e-06,
      "loss": 1.0069,
      "step": 2492
    },
    {
      "epoch": 0.07183195989166138,
      "grad_norm": 2.674888849258423,
      "learning_rate": 9.954230155019192e-06,
      "loss": 0.9681,
      "step": 2493
    },
    {
      "epoch": 0.07186077335331066,
      "grad_norm": 2.301088333129883,
      "learning_rate": 9.954167142560473e-06,
      "loss": 0.9989,
      "step": 2494
    },
    {
      "epoch": 0.07188958681495995,
      "grad_norm": 2.3000099658966064,
      "learning_rate": 9.954104086955912e-06,
      "loss": 0.9977,
      "step": 2495
    },
    {
      "epoch": 0.07191840027660923,
      "grad_norm": 2.5092971324920654,
      "learning_rate": 9.954040988206058e-06,
      "loss": 0.9896,
      "step": 2496
    },
    {
      "epoch": 0.07194721373825852,
      "grad_norm": 2.39044189453125,
      "learning_rate": 9.953977846311457e-06,
      "loss": 0.9773,
      "step": 2497
    },
    {
      "epoch": 0.0719760271999078,
      "grad_norm": 2.270968437194824,
      "learning_rate": 9.953914661272662e-06,
      "loss": 0.9815,
      "step": 2498
    },
    {
      "epoch": 0.07200484066155707,
      "grad_norm": 2.3651819229125977,
      "learning_rate": 9.953851433090222e-06,
      "loss": 0.9704,
      "step": 2499
    },
    {
      "epoch": 0.07203365412320636,
      "grad_norm": 2.198551893234253,
      "learning_rate": 9.953788161764689e-06,
      "loss": 0.9846,
      "step": 2500
    },
    {
      "epoch": 0.07206246758485564,
      "grad_norm": 2.37371563911438,
      "learning_rate": 9.953724847296613e-06,
      "loss": 0.9725,
      "step": 2501
    },
    {
      "epoch": 0.07209128104650493,
      "grad_norm": 2.4232568740844727,
      "learning_rate": 9.953661489686544e-06,
      "loss": 0.987,
      "step": 2502
    },
    {
      "epoch": 0.07212009450815421,
      "grad_norm": 2.073180675506592,
      "learning_rate": 9.953598088935038e-06,
      "loss": 0.9955,
      "step": 2503
    },
    {
      "epoch": 0.07214890796980349,
      "grad_norm": 2.4000210762023926,
      "learning_rate": 9.953534645042642e-06,
      "loss": 0.9699,
      "step": 2504
    },
    {
      "epoch": 0.07217772143145278,
      "grad_norm": 2.28096079826355,
      "learning_rate": 9.953471158009911e-06,
      "loss": 0.989,
      "step": 2505
    },
    {
      "epoch": 0.07220653489310205,
      "grad_norm": 2.3435401916503906,
      "learning_rate": 9.9534076278374e-06,
      "loss": 0.9844,
      "step": 2506
    },
    {
      "epoch": 0.07223534835475134,
      "grad_norm": 2.197009325027466,
      "learning_rate": 9.953344054525658e-06,
      "loss": 0.9831,
      "step": 2507
    },
    {
      "epoch": 0.07226416181640062,
      "grad_norm": 2.2278449535369873,
      "learning_rate": 9.95328043807524e-06,
      "loss": 0.9925,
      "step": 2508
    },
    {
      "epoch": 0.0722929752780499,
      "grad_norm": 2.368023157119751,
      "learning_rate": 9.953216778486705e-06,
      "loss": 0.9691,
      "step": 2509
    },
    {
      "epoch": 0.07232178873969919,
      "grad_norm": 2.2026901245117188,
      "learning_rate": 9.9531530757606e-06,
      "loss": 1.0166,
      "step": 2510
    },
    {
      "epoch": 0.07235060220134847,
      "grad_norm": 2.272066593170166,
      "learning_rate": 9.953089329897484e-06,
      "loss": 0.9766,
      "step": 2511
    },
    {
      "epoch": 0.07237941566299776,
      "grad_norm": 2.2321975231170654,
      "learning_rate": 9.953025540897911e-06,
      "loss": 1.0025,
      "step": 2512
    },
    {
      "epoch": 0.07240822912464703,
      "grad_norm": 2.364866018295288,
      "learning_rate": 9.952961708762436e-06,
      "loss": 0.9962,
      "step": 2513
    },
    {
      "epoch": 0.07243704258629632,
      "grad_norm": 2.359774351119995,
      "learning_rate": 9.952897833491617e-06,
      "loss": 0.9997,
      "step": 2514
    },
    {
      "epoch": 0.0724658560479456,
      "grad_norm": 2.347757577896118,
      "learning_rate": 9.95283391508601e-06,
      "loss": 0.9929,
      "step": 2515
    },
    {
      "epoch": 0.07249466950959488,
      "grad_norm": 2.173973560333252,
      "learning_rate": 9.952769953546169e-06,
      "loss": 0.9894,
      "step": 2516
    },
    {
      "epoch": 0.07252348297124417,
      "grad_norm": 2.628905773162842,
      "learning_rate": 9.952705948872651e-06,
      "loss": 1.0063,
      "step": 2517
    },
    {
      "epoch": 0.07255229643289345,
      "grad_norm": 2.1479227542877197,
      "learning_rate": 9.952641901066018e-06,
      "loss": 0.9657,
      "step": 2518
    },
    {
      "epoch": 0.07258110989454274,
      "grad_norm": 2.2303545475006104,
      "learning_rate": 9.952577810126823e-06,
      "loss": 1.0013,
      "step": 2519
    },
    {
      "epoch": 0.07260992335619201,
      "grad_norm": 2.2524232864379883,
      "learning_rate": 9.952513676055626e-06,
      "loss": 0.9836,
      "step": 2520
    },
    {
      "epoch": 0.07263873681784129,
      "grad_norm": 2.27439546585083,
      "learning_rate": 9.952449498852987e-06,
      "loss": 1.0098,
      "step": 2521
    },
    {
      "epoch": 0.07266755027949058,
      "grad_norm": 2.380934476852417,
      "learning_rate": 9.952385278519463e-06,
      "loss": 0.9802,
      "step": 2522
    },
    {
      "epoch": 0.07269636374113986,
      "grad_norm": 2.3179869651794434,
      "learning_rate": 9.952321015055613e-06,
      "loss": 1.0026,
      "step": 2523
    },
    {
      "epoch": 0.07272517720278915,
      "grad_norm": 2.103610038757324,
      "learning_rate": 9.952256708461999e-06,
      "loss": 0.9743,
      "step": 2524
    },
    {
      "epoch": 0.07275399066443843,
      "grad_norm": 2.1659278869628906,
      "learning_rate": 9.95219235873918e-06,
      "loss": 0.9815,
      "step": 2525
    },
    {
      "epoch": 0.0727828041260877,
      "grad_norm": 2.0898921489715576,
      "learning_rate": 9.952127965887712e-06,
      "loss": 0.9891,
      "step": 2526
    },
    {
      "epoch": 0.072811617587737,
      "grad_norm": 2.253584384918213,
      "learning_rate": 9.952063529908162e-06,
      "loss": 0.9791,
      "step": 2527
    },
    {
      "epoch": 0.07284043104938627,
      "grad_norm": 2.04335355758667,
      "learning_rate": 9.95199905080109e-06,
      "loss": 0.9936,
      "step": 2528
    },
    {
      "epoch": 0.07286924451103556,
      "grad_norm": 2.22394061088562,
      "learning_rate": 9.951934528567055e-06,
      "loss": 1.0084,
      "step": 2529
    },
    {
      "epoch": 0.07289805797268484,
      "grad_norm": 2.168241024017334,
      "learning_rate": 9.951869963206622e-06,
      "loss": 0.9368,
      "step": 2530
    },
    {
      "epoch": 0.07292687143433411,
      "grad_norm": 2.1589324474334717,
      "learning_rate": 9.95180535472035e-06,
      "loss": 0.9727,
      "step": 2531
    },
    {
      "epoch": 0.0729556848959834,
      "grad_norm": 2.252913236618042,
      "learning_rate": 9.951740703108805e-06,
      "loss": 0.9724,
      "step": 2532
    },
    {
      "epoch": 0.07298449835763268,
      "grad_norm": 2.1054978370666504,
      "learning_rate": 9.951676008372546e-06,
      "loss": 0.972,
      "step": 2533
    },
    {
      "epoch": 0.07301331181928197,
      "grad_norm": 2.284806728363037,
      "learning_rate": 9.95161127051214e-06,
      "loss": 1.002,
      "step": 2534
    },
    {
      "epoch": 0.07304212528093125,
      "grad_norm": 2.135355234146118,
      "learning_rate": 9.95154648952815e-06,
      "loss": 0.9795,
      "step": 2535
    },
    {
      "epoch": 0.07307093874258053,
      "grad_norm": 2.0612330436706543,
      "learning_rate": 9.95148166542114e-06,
      "loss": 1.0175,
      "step": 2536
    },
    {
      "epoch": 0.07309975220422982,
      "grad_norm": 2.2455339431762695,
      "learning_rate": 9.951416798191674e-06,
      "loss": 1.0023,
      "step": 2537
    },
    {
      "epoch": 0.0731285656658791,
      "grad_norm": 2.2920711040496826,
      "learning_rate": 9.951351887840317e-06,
      "loss": 0.9487,
      "step": 2538
    },
    {
      "epoch": 0.07315737912752839,
      "grad_norm": 2.234449625015259,
      "learning_rate": 9.951286934367635e-06,
      "loss": 0.9801,
      "step": 2539
    },
    {
      "epoch": 0.07318619258917766,
      "grad_norm": 2.1831774711608887,
      "learning_rate": 9.951221937774194e-06,
      "loss": 0.9992,
      "step": 2540
    },
    {
      "epoch": 0.07321500605082695,
      "grad_norm": 2.0880351066589355,
      "learning_rate": 9.951156898060558e-06,
      "loss": 1.0193,
      "step": 2541
    },
    {
      "epoch": 0.07324381951247623,
      "grad_norm": 2.3643147945404053,
      "learning_rate": 9.951091815227296e-06,
      "loss": 0.9809,
      "step": 2542
    },
    {
      "epoch": 0.0732726329741255,
      "grad_norm": 2.2041471004486084,
      "learning_rate": 9.951026689274971e-06,
      "loss": 0.9856,
      "step": 2543
    },
    {
      "epoch": 0.0733014464357748,
      "grad_norm": 2.271176815032959,
      "learning_rate": 9.950961520204156e-06,
      "loss": 0.9792,
      "step": 2544
    },
    {
      "epoch": 0.07333025989742407,
      "grad_norm": 2.201324701309204,
      "learning_rate": 9.950896308015414e-06,
      "loss": 0.9704,
      "step": 2545
    },
    {
      "epoch": 0.07335907335907337,
      "grad_norm": 2.2524852752685547,
      "learning_rate": 9.950831052709315e-06,
      "loss": 0.9769,
      "step": 2546
    },
    {
      "epoch": 0.07338788682072264,
      "grad_norm": 2.35714054107666,
      "learning_rate": 9.950765754286425e-06,
      "loss": 1.0044,
      "step": 2547
    },
    {
      "epoch": 0.07341670028237192,
      "grad_norm": 2.301506996154785,
      "learning_rate": 9.950700412747315e-06,
      "loss": 0.982,
      "step": 2548
    },
    {
      "epoch": 0.07344551374402121,
      "grad_norm": 2.236926555633545,
      "learning_rate": 9.950635028092554e-06,
      "loss": 0.9597,
      "step": 2549
    },
    {
      "epoch": 0.07347432720567049,
      "grad_norm": 2.378938674926758,
      "learning_rate": 9.950569600322708e-06,
      "loss": 0.9967,
      "step": 2550
    },
    {
      "epoch": 0.07350314066731978,
      "grad_norm": 2.2932589054107666,
      "learning_rate": 9.950504129438352e-06,
      "loss": 0.9792,
      "step": 2551
    },
    {
      "epoch": 0.07353195412896905,
      "grad_norm": 2.198375701904297,
      "learning_rate": 9.950438615440054e-06,
      "loss": 0.9653,
      "step": 2552
    },
    {
      "epoch": 0.07356076759061833,
      "grad_norm": 2.313544511795044,
      "learning_rate": 9.950373058328384e-06,
      "loss": 0.9663,
      "step": 2553
    },
    {
      "epoch": 0.07358958105226762,
      "grad_norm": 2.376627206802368,
      "learning_rate": 9.950307458103911e-06,
      "loss": 0.9911,
      "step": 2554
    },
    {
      "epoch": 0.0736183945139169,
      "grad_norm": 2.39599871635437,
      "learning_rate": 9.950241814767211e-06,
      "loss": 0.9702,
      "step": 2555
    },
    {
      "epoch": 0.07364720797556619,
      "grad_norm": 2.3402953147888184,
      "learning_rate": 9.950176128318854e-06,
      "loss": 0.9796,
      "step": 2556
    },
    {
      "epoch": 0.07367602143721547,
      "grad_norm": 2.243586301803589,
      "learning_rate": 9.950110398759409e-06,
      "loss": 1.0025,
      "step": 2557
    },
    {
      "epoch": 0.07370483489886474,
      "grad_norm": 2.383876323699951,
      "learning_rate": 9.950044626089453e-06,
      "loss": 0.9595,
      "step": 2558
    },
    {
      "epoch": 0.07373364836051403,
      "grad_norm": 2.297055959701538,
      "learning_rate": 9.949978810309554e-06,
      "loss": 1.0221,
      "step": 2559
    },
    {
      "epoch": 0.07376246182216331,
      "grad_norm": 2.0847315788269043,
      "learning_rate": 9.949912951420289e-06,
      "loss": 0.9724,
      "step": 2560
    },
    {
      "epoch": 0.0737912752838126,
      "grad_norm": 2.33788800239563,
      "learning_rate": 9.94984704942223e-06,
      "loss": 0.9899,
      "step": 2561
    },
    {
      "epoch": 0.07382008874546188,
      "grad_norm": 2.271982431411743,
      "learning_rate": 9.94978110431595e-06,
      "loss": 1.0132,
      "step": 2562
    },
    {
      "epoch": 0.07384890220711116,
      "grad_norm": 2.0669872760772705,
      "learning_rate": 9.949715116102027e-06,
      "loss": 0.9948,
      "step": 2563
    },
    {
      "epoch": 0.07387771566876045,
      "grad_norm": 2.305006504058838,
      "learning_rate": 9.949649084781031e-06,
      "loss": 0.9724,
      "step": 2564
    },
    {
      "epoch": 0.07390652913040972,
      "grad_norm": 2.453258752822876,
      "learning_rate": 9.94958301035354e-06,
      "loss": 1.0011,
      "step": 2565
    },
    {
      "epoch": 0.07393534259205901,
      "grad_norm": 2.464043140411377,
      "learning_rate": 9.94951689282013e-06,
      "loss": 0.9772,
      "step": 2566
    },
    {
      "epoch": 0.07396415605370829,
      "grad_norm": 2.2943735122680664,
      "learning_rate": 9.949450732181377e-06,
      "loss": 0.9642,
      "step": 2567
    },
    {
      "epoch": 0.07399296951535758,
      "grad_norm": 2.496837854385376,
      "learning_rate": 9.949384528437851e-06,
      "loss": 1.0103,
      "step": 2568
    },
    {
      "epoch": 0.07402178297700686,
      "grad_norm": 2.3101727962493896,
      "learning_rate": 9.949318281590137e-06,
      "loss": 0.9686,
      "step": 2569
    },
    {
      "epoch": 0.07405059643865614,
      "grad_norm": 2.390782117843628,
      "learning_rate": 9.949251991638806e-06,
      "loss": 0.9803,
      "step": 2570
    },
    {
      "epoch": 0.07407940990030543,
      "grad_norm": 2.2259531021118164,
      "learning_rate": 9.949185658584439e-06,
      "loss": 0.9515,
      "step": 2571
    },
    {
      "epoch": 0.0741082233619547,
      "grad_norm": 2.280712127685547,
      "learning_rate": 9.949119282427613e-06,
      "loss": 0.9726,
      "step": 2572
    },
    {
      "epoch": 0.074137036823604,
      "grad_norm": 2.0860800743103027,
      "learning_rate": 9.949052863168904e-06,
      "loss": 0.9672,
      "step": 2573
    },
    {
      "epoch": 0.07416585028525327,
      "grad_norm": 2.297279119491577,
      "learning_rate": 9.948986400808892e-06,
      "loss": 0.9785,
      "step": 2574
    },
    {
      "epoch": 0.07419466374690255,
      "grad_norm": 2.1182780265808105,
      "learning_rate": 9.948919895348156e-06,
      "loss": 0.981,
      "step": 2575
    },
    {
      "epoch": 0.07422347720855184,
      "grad_norm": 2.0868959426879883,
      "learning_rate": 9.948853346787274e-06,
      "loss": 1.0169,
      "step": 2576
    },
    {
      "epoch": 0.07425229067020112,
      "grad_norm": 2.219008684158325,
      "learning_rate": 9.948786755126827e-06,
      "loss": 0.9687,
      "step": 2577
    },
    {
      "epoch": 0.0742811041318504,
      "grad_norm": 2.197680711746216,
      "learning_rate": 9.948720120367395e-06,
      "loss": 0.9694,
      "step": 2578
    },
    {
      "epoch": 0.07430991759349968,
      "grad_norm": 2.1804027557373047,
      "learning_rate": 9.948653442509558e-06,
      "loss": 0.9862,
      "step": 2579
    },
    {
      "epoch": 0.07433873105514896,
      "grad_norm": 2.2150652408599854,
      "learning_rate": 9.948586721553895e-06,
      "loss": 1.0152,
      "step": 2580
    },
    {
      "epoch": 0.07436754451679825,
      "grad_norm": 2.2076637744903564,
      "learning_rate": 9.948519957500989e-06,
      "loss": 0.9758,
      "step": 2581
    },
    {
      "epoch": 0.07439635797844753,
      "grad_norm": 2.397956132888794,
      "learning_rate": 9.94845315035142e-06,
      "loss": 0.9724,
      "step": 2582
    },
    {
      "epoch": 0.07442517144009682,
      "grad_norm": 2.392975330352783,
      "learning_rate": 9.948386300105772e-06,
      "loss": 1.0037,
      "step": 2583
    },
    {
      "epoch": 0.0744539849017461,
      "grad_norm": 2.1636624336242676,
      "learning_rate": 9.948319406764623e-06,
      "loss": 0.9853,
      "step": 2584
    },
    {
      "epoch": 0.07448279836339537,
      "grad_norm": 2.563668966293335,
      "learning_rate": 9.948252470328562e-06,
      "loss": 0.9741,
      "step": 2585
    },
    {
      "epoch": 0.07451161182504466,
      "grad_norm": 2.178892135620117,
      "learning_rate": 9.948185490798167e-06,
      "loss": 0.9509,
      "step": 2586
    },
    {
      "epoch": 0.07454042528669394,
      "grad_norm": 2.56827974319458,
      "learning_rate": 9.948118468174024e-06,
      "loss": 0.9902,
      "step": 2587
    },
    {
      "epoch": 0.07456923874834323,
      "grad_norm": 2.223055601119995,
      "learning_rate": 9.948051402456714e-06,
      "loss": 0.9786,
      "step": 2588
    },
    {
      "epoch": 0.07459805220999251,
      "grad_norm": 2.1674461364746094,
      "learning_rate": 9.947984293646822e-06,
      "loss": 0.9762,
      "step": 2589
    },
    {
      "epoch": 0.07462686567164178,
      "grad_norm": 2.461519241333008,
      "learning_rate": 9.947917141744934e-06,
      "loss": 0.9917,
      "step": 2590
    },
    {
      "epoch": 0.07465567913329108,
      "grad_norm": 2.1507134437561035,
      "learning_rate": 9.947849946751635e-06,
      "loss": 0.9755,
      "step": 2591
    },
    {
      "epoch": 0.07468449259494035,
      "grad_norm": 2.2081432342529297,
      "learning_rate": 9.947782708667505e-06,
      "loss": 0.98,
      "step": 2592
    },
    {
      "epoch": 0.07471330605658964,
      "grad_norm": 2.4069085121154785,
      "learning_rate": 9.947715427493138e-06,
      "loss": 0.971,
      "step": 2593
    },
    {
      "epoch": 0.07474211951823892,
      "grad_norm": 2.2884433269500732,
      "learning_rate": 9.947648103229114e-06,
      "loss": 0.9816,
      "step": 2594
    },
    {
      "epoch": 0.07477093297988821,
      "grad_norm": 2.2767374515533447,
      "learning_rate": 9.94758073587602e-06,
      "loss": 0.9721,
      "step": 2595
    },
    {
      "epoch": 0.07479974644153749,
      "grad_norm": 2.194028854370117,
      "learning_rate": 9.947513325434444e-06,
      "loss": 0.95,
      "step": 2596
    },
    {
      "epoch": 0.07482855990318676,
      "grad_norm": 2.2622902393341064,
      "learning_rate": 9.947445871904974e-06,
      "loss": 0.9891,
      "step": 2597
    },
    {
      "epoch": 0.07485737336483606,
      "grad_norm": 2.1890382766723633,
      "learning_rate": 9.947378375288195e-06,
      "loss": 0.974,
      "step": 2598
    },
    {
      "epoch": 0.07488618682648533,
      "grad_norm": 2.3805770874023438,
      "learning_rate": 9.947310835584697e-06,
      "loss": 0.9933,
      "step": 2599
    },
    {
      "epoch": 0.07491500028813462,
      "grad_norm": 2.227419853210449,
      "learning_rate": 9.947243252795065e-06,
      "loss": 0.9733,
      "step": 2600
    },
    {
      "epoch": 0.0749438137497839,
      "grad_norm": 2.463669776916504,
      "learning_rate": 9.94717562691989e-06,
      "loss": 0.9858,
      "step": 2601
    },
    {
      "epoch": 0.07497262721143318,
      "grad_norm": 2.3194406032562256,
      "learning_rate": 9.947107957959761e-06,
      "loss": 0.9639,
      "step": 2602
    },
    {
      "epoch": 0.07500144067308247,
      "grad_norm": 2.2350382804870605,
      "learning_rate": 9.947040245915267e-06,
      "loss": 0.9909,
      "step": 2603
    },
    {
      "epoch": 0.07503025413473174,
      "grad_norm": 2.2577285766601562,
      "learning_rate": 9.946972490786998e-06,
      "loss": 0.9808,
      "step": 2604
    },
    {
      "epoch": 0.07505906759638103,
      "grad_norm": 2.2817580699920654,
      "learning_rate": 9.946904692575545e-06,
      "loss": 1.0202,
      "step": 2605
    },
    {
      "epoch": 0.07508788105803031,
      "grad_norm": 2.2083754539489746,
      "learning_rate": 9.946836851281496e-06,
      "loss": 0.9702,
      "step": 2606
    },
    {
      "epoch": 0.07511669451967959,
      "grad_norm": 2.497682809829712,
      "learning_rate": 9.946768966905443e-06,
      "loss": 0.983,
      "step": 2607
    },
    {
      "epoch": 0.07514550798132888,
      "grad_norm": 2.137549638748169,
      "learning_rate": 9.946701039447976e-06,
      "loss": 0.9783,
      "step": 2608
    },
    {
      "epoch": 0.07517432144297816,
      "grad_norm": 2.4482369422912598,
      "learning_rate": 9.94663306890969e-06,
      "loss": 0.9734,
      "step": 2609
    },
    {
      "epoch": 0.07520313490462745,
      "grad_norm": 2.244407892227173,
      "learning_rate": 9.946565055291175e-06,
      "loss": 1.0042,
      "step": 2610
    },
    {
      "epoch": 0.07523194836627672,
      "grad_norm": 2.1560230255126953,
      "learning_rate": 9.946496998593021e-06,
      "loss": 0.9811,
      "step": 2611
    },
    {
      "epoch": 0.075260761827926,
      "grad_norm": 2.266227960586548,
      "learning_rate": 9.946428898815824e-06,
      "loss": 0.9751,
      "step": 2612
    },
    {
      "epoch": 0.07528957528957529,
      "grad_norm": 2.0934641361236572,
      "learning_rate": 9.946360755960176e-06,
      "loss": 0.9736,
      "step": 2613
    },
    {
      "epoch": 0.07531838875122457,
      "grad_norm": 2.2061336040496826,
      "learning_rate": 9.94629257002667e-06,
      "loss": 0.9779,
      "step": 2614
    },
    {
      "epoch": 0.07534720221287386,
      "grad_norm": 2.327948570251465,
      "learning_rate": 9.9462243410159e-06,
      "loss": 0.9712,
      "step": 2615
    },
    {
      "epoch": 0.07537601567452314,
      "grad_norm": 2.21142840385437,
      "learning_rate": 9.94615606892846e-06,
      "loss": 1.0108,
      "step": 2616
    },
    {
      "epoch": 0.07540482913617241,
      "grad_norm": 2.3205697536468506,
      "learning_rate": 9.946087753764947e-06,
      "loss": 0.983,
      "step": 2617
    },
    {
      "epoch": 0.0754336425978217,
      "grad_norm": 2.2649481296539307,
      "learning_rate": 9.946019395525953e-06,
      "loss": 0.9912,
      "step": 2618
    },
    {
      "epoch": 0.07546245605947098,
      "grad_norm": 2.1833298206329346,
      "learning_rate": 9.945950994212073e-06,
      "loss": 0.9872,
      "step": 2619
    },
    {
      "epoch": 0.07549126952112027,
      "grad_norm": 2.3230087757110596,
      "learning_rate": 9.945882549823906e-06,
      "loss": 0.974,
      "step": 2620
    },
    {
      "epoch": 0.07552008298276955,
      "grad_norm": 2.0805585384368896,
      "learning_rate": 9.945814062362044e-06,
      "loss": 1.0118,
      "step": 2621
    },
    {
      "epoch": 0.07554889644441884,
      "grad_norm": 2.117201805114746,
      "learning_rate": 9.945745531827087e-06,
      "loss": 0.975,
      "step": 2622
    },
    {
      "epoch": 0.07557770990606812,
      "grad_norm": 2.194232702255249,
      "learning_rate": 9.94567695821963e-06,
      "loss": 0.9601,
      "step": 2623
    },
    {
      "epoch": 0.0756065233677174,
      "grad_norm": 2.0336244106292725,
      "learning_rate": 9.945608341540271e-06,
      "loss": 0.9903,
      "step": 2624
    },
    {
      "epoch": 0.07563533682936668,
      "grad_norm": 2.2961220741271973,
      "learning_rate": 9.945539681789606e-06,
      "loss": 0.9579,
      "step": 2625
    },
    {
      "epoch": 0.07566415029101596,
      "grad_norm": 2.0809175968170166,
      "learning_rate": 9.945470978968235e-06,
      "loss": 0.9852,
      "step": 2626
    },
    {
      "epoch": 0.07569296375266525,
      "grad_norm": 2.1084682941436768,
      "learning_rate": 9.945402233076755e-06,
      "loss": 0.9996,
      "step": 2627
    },
    {
      "epoch": 0.07572177721431453,
      "grad_norm": 2.3212332725524902,
      "learning_rate": 9.945333444115766e-06,
      "loss": 0.9532,
      "step": 2628
    },
    {
      "epoch": 0.0757505906759638,
      "grad_norm": 2.0800461769104004,
      "learning_rate": 9.945264612085865e-06,
      "loss": 0.9657,
      "step": 2629
    },
    {
      "epoch": 0.0757794041376131,
      "grad_norm": 1.9858871698379517,
      "learning_rate": 9.945195736987654e-06,
      "loss": 0.9867,
      "step": 2630
    },
    {
      "epoch": 0.07580821759926237,
      "grad_norm": 2.3709540367126465,
      "learning_rate": 9.945126818821729e-06,
      "loss": 0.9642,
      "step": 2631
    },
    {
      "epoch": 0.07583703106091166,
      "grad_norm": 2.316051483154297,
      "learning_rate": 9.945057857588695e-06,
      "loss": 0.9914,
      "step": 2632
    },
    {
      "epoch": 0.07586584452256094,
      "grad_norm": 2.0431742668151855,
      "learning_rate": 9.94498885328915e-06,
      "loss": 0.9612,
      "step": 2633
    },
    {
      "epoch": 0.07589465798421022,
      "grad_norm": 2.137714147567749,
      "learning_rate": 9.944919805923695e-06,
      "loss": 0.9851,
      "step": 2634
    },
    {
      "epoch": 0.07592347144585951,
      "grad_norm": 2.120844841003418,
      "learning_rate": 9.944850715492932e-06,
      "loss": 0.9758,
      "step": 2635
    },
    {
      "epoch": 0.07595228490750879,
      "grad_norm": 2.091789722442627,
      "learning_rate": 9.944781581997463e-06,
      "loss": 0.9754,
      "step": 2636
    },
    {
      "epoch": 0.07598109836915808,
      "grad_norm": 2.151965856552124,
      "learning_rate": 9.944712405437886e-06,
      "loss": 0.9835,
      "step": 2637
    },
    {
      "epoch": 0.07600991183080735,
      "grad_norm": 2.167713165283203,
      "learning_rate": 9.944643185814811e-06,
      "loss": 0.9567,
      "step": 2638
    },
    {
      "epoch": 0.07603872529245663,
      "grad_norm": 2.0822489261627197,
      "learning_rate": 9.944573923128835e-06,
      "loss": 0.9616,
      "step": 2639
    },
    {
      "epoch": 0.07606753875410592,
      "grad_norm": 2.2002899646759033,
      "learning_rate": 9.944504617380562e-06,
      "loss": 0.9585,
      "step": 2640
    },
    {
      "epoch": 0.0760963522157552,
      "grad_norm": 2.2020328044891357,
      "learning_rate": 9.944435268570597e-06,
      "loss": 0.9975,
      "step": 2641
    },
    {
      "epoch": 0.07612516567740449,
      "grad_norm": 2.3331029415130615,
      "learning_rate": 9.944365876699545e-06,
      "loss": 0.9605,
      "step": 2642
    },
    {
      "epoch": 0.07615397913905377,
      "grad_norm": 2.075636863708496,
      "learning_rate": 9.944296441768007e-06,
      "loss": 0.995,
      "step": 2643
    },
    {
      "epoch": 0.07618279260070304,
      "grad_norm": 1.9851022958755493,
      "learning_rate": 9.944226963776588e-06,
      "loss": 0.9493,
      "step": 2644
    },
    {
      "epoch": 0.07621160606235233,
      "grad_norm": 2.1680309772491455,
      "learning_rate": 9.944157442725896e-06,
      "loss": 0.9756,
      "step": 2645
    },
    {
      "epoch": 0.07624041952400161,
      "grad_norm": 2.0683224201202393,
      "learning_rate": 9.944087878616537e-06,
      "loss": 0.9608,
      "step": 2646
    },
    {
      "epoch": 0.0762692329856509,
      "grad_norm": 2.016690969467163,
      "learning_rate": 9.944018271449114e-06,
      "loss": 1.0036,
      "step": 2647
    },
    {
      "epoch": 0.07629804644730018,
      "grad_norm": 2.0676252841949463,
      "learning_rate": 9.943948621224232e-06,
      "loss": 0.9949,
      "step": 2648
    },
    {
      "epoch": 0.07632685990894945,
      "grad_norm": 2.1975433826446533,
      "learning_rate": 9.943878927942501e-06,
      "loss": 1.0217,
      "step": 2649
    },
    {
      "epoch": 0.07635567337059875,
      "grad_norm": 2.0273427963256836,
      "learning_rate": 9.943809191604527e-06,
      "loss": 1.0037,
      "step": 2650
    },
    {
      "epoch": 0.07638448683224802,
      "grad_norm": 2.105828285217285,
      "learning_rate": 9.943739412210918e-06,
      "loss": 0.964,
      "step": 2651
    },
    {
      "epoch": 0.07641330029389731,
      "grad_norm": 2.0512349605560303,
      "learning_rate": 9.94366958976228e-06,
      "loss": 0.9819,
      "step": 2652
    },
    {
      "epoch": 0.07644211375554659,
      "grad_norm": 2.5128042697906494,
      "learning_rate": 9.94359972425922e-06,
      "loss": 0.9823,
      "step": 2653
    },
    {
      "epoch": 0.07647092721719588,
      "grad_norm": 2.114727020263672,
      "learning_rate": 9.943529815702349e-06,
      "loss": 0.9882,
      "step": 2654
    },
    {
      "epoch": 0.07649974067884516,
      "grad_norm": 2.2958381175994873,
      "learning_rate": 9.943459864092276e-06,
      "loss": 0.9981,
      "step": 2655
    },
    {
      "epoch": 0.07652855414049443,
      "grad_norm": 1.9359239339828491,
      "learning_rate": 9.943389869429607e-06,
      "loss": 0.9534,
      "step": 2656
    },
    {
      "epoch": 0.07655736760214372,
      "grad_norm": 2.236980676651001,
      "learning_rate": 9.943319831714955e-06,
      "loss": 0.9828,
      "step": 2657
    },
    {
      "epoch": 0.076586181063793,
      "grad_norm": 2.1637532711029053,
      "learning_rate": 9.943249750948929e-06,
      "loss": 0.9684,
      "step": 2658
    },
    {
      "epoch": 0.07661499452544229,
      "grad_norm": 2.166926860809326,
      "learning_rate": 9.94317962713214e-06,
      "loss": 0.9704,
      "step": 2659
    },
    {
      "epoch": 0.07664380798709157,
      "grad_norm": 2.0087170600891113,
      "learning_rate": 9.943109460265197e-06,
      "loss": 0.9675,
      "step": 2660
    },
    {
      "epoch": 0.07667262144874085,
      "grad_norm": 2.0709595680236816,
      "learning_rate": 9.943039250348712e-06,
      "loss": 0.9633,
      "step": 2661
    },
    {
      "epoch": 0.07670143491039014,
      "grad_norm": 2.2457292079925537,
      "learning_rate": 9.942968997383295e-06,
      "loss": 0.9963,
      "step": 2662
    },
    {
      "epoch": 0.07673024837203941,
      "grad_norm": 2.06207537651062,
      "learning_rate": 9.942898701369561e-06,
      "loss": 0.9769,
      "step": 2663
    },
    {
      "epoch": 0.0767590618336887,
      "grad_norm": 2.137471914291382,
      "learning_rate": 9.94282836230812e-06,
      "loss": 0.97,
      "step": 2664
    },
    {
      "epoch": 0.07678787529533798,
      "grad_norm": 2.0478525161743164,
      "learning_rate": 9.942757980199584e-06,
      "loss": 0.9875,
      "step": 2665
    },
    {
      "epoch": 0.07681668875698726,
      "grad_norm": 2.201580762863159,
      "learning_rate": 9.94268755504457e-06,
      "loss": 0.9823,
      "step": 2666
    },
    {
      "epoch": 0.07684550221863655,
      "grad_norm": 2.1403651237487793,
      "learning_rate": 9.942617086843684e-06,
      "loss": 0.9821,
      "step": 2667
    },
    {
      "epoch": 0.07687431568028583,
      "grad_norm": 2.1413538455963135,
      "learning_rate": 9.942546575597547e-06,
      "loss": 0.9571,
      "step": 2668
    },
    {
      "epoch": 0.07690312914193512,
      "grad_norm": 2.2077620029449463,
      "learning_rate": 9.942476021306768e-06,
      "loss": 0.9651,
      "step": 2669
    },
    {
      "epoch": 0.0769319426035844,
      "grad_norm": 2.1583452224731445,
      "learning_rate": 9.942405423971966e-06,
      "loss": 0.9857,
      "step": 2670
    },
    {
      "epoch": 0.07696075606523367,
      "grad_norm": 2.2437915802001953,
      "learning_rate": 9.942334783593752e-06,
      "loss": 0.9989,
      "step": 2671
    },
    {
      "epoch": 0.07698956952688296,
      "grad_norm": 2.2652347087860107,
      "learning_rate": 9.942264100172743e-06,
      "loss": 0.97,
      "step": 2672
    },
    {
      "epoch": 0.07701838298853224,
      "grad_norm": 2.3620622158050537,
      "learning_rate": 9.942193373709553e-06,
      "loss": 0.9774,
      "step": 2673
    },
    {
      "epoch": 0.07704719645018153,
      "grad_norm": 2.1444900035858154,
      "learning_rate": 9.942122604204801e-06,
      "loss": 0.9524,
      "step": 2674
    },
    {
      "epoch": 0.0770760099118308,
      "grad_norm": 2.4348883628845215,
      "learning_rate": 9.942051791659099e-06,
      "loss": 0.9936,
      "step": 2675
    },
    {
      "epoch": 0.07710482337348008,
      "grad_norm": 2.1356260776519775,
      "learning_rate": 9.941980936073069e-06,
      "loss": 1.0115,
      "step": 2676
    },
    {
      "epoch": 0.07713363683512937,
      "grad_norm": 2.240068197250366,
      "learning_rate": 9.941910037447323e-06,
      "loss": 1.0261,
      "step": 2677
    },
    {
      "epoch": 0.07716245029677865,
      "grad_norm": 2.089101791381836,
      "learning_rate": 9.94183909578248e-06,
      "loss": 0.9781,
      "step": 2678
    },
    {
      "epoch": 0.07719126375842794,
      "grad_norm": 2.114003896713257,
      "learning_rate": 9.94176811107916e-06,
      "loss": 0.9735,
      "step": 2679
    },
    {
      "epoch": 0.07722007722007722,
      "grad_norm": 2.1536478996276855,
      "learning_rate": 9.94169708333798e-06,
      "loss": 0.978,
      "step": 2680
    },
    {
      "epoch": 0.07724889068172651,
      "grad_norm": 2.3177905082702637,
      "learning_rate": 9.941626012559558e-06,
      "loss": 0.9409,
      "step": 2681
    },
    {
      "epoch": 0.07727770414337579,
      "grad_norm": 2.1859819889068604,
      "learning_rate": 9.941554898744513e-06,
      "loss": 0.9567,
      "step": 2682
    },
    {
      "epoch": 0.07730651760502506,
      "grad_norm": 2.285491466522217,
      "learning_rate": 9.941483741893463e-06,
      "loss": 0.9743,
      "step": 2683
    },
    {
      "epoch": 0.07733533106667435,
      "grad_norm": 2.22946834564209,
      "learning_rate": 9.941412542007032e-06,
      "loss": 0.9556,
      "step": 2684
    },
    {
      "epoch": 0.07736414452832363,
      "grad_norm": 2.080073833465576,
      "learning_rate": 9.941341299085836e-06,
      "loss": 0.9767,
      "step": 2685
    },
    {
      "epoch": 0.07739295798997292,
      "grad_norm": 2.2591636180877686,
      "learning_rate": 9.941270013130497e-06,
      "loss": 0.9805,
      "step": 2686
    },
    {
      "epoch": 0.0774217714516222,
      "grad_norm": 2.18764591217041,
      "learning_rate": 9.941198684141635e-06,
      "loss": 0.9783,
      "step": 2687
    },
    {
      "epoch": 0.07745058491327148,
      "grad_norm": 2.3660621643066406,
      "learning_rate": 9.941127312119873e-06,
      "loss": 0.9889,
      "step": 2688
    },
    {
      "epoch": 0.07747939837492077,
      "grad_norm": 2.217341899871826,
      "learning_rate": 9.94105589706583e-06,
      "loss": 1.0211,
      "step": 2689
    },
    {
      "epoch": 0.07750821183657004,
      "grad_norm": 2.170452117919922,
      "learning_rate": 9.94098443898013e-06,
      "loss": 0.9792,
      "step": 2690
    },
    {
      "epoch": 0.07753702529821933,
      "grad_norm": 2.3406124114990234,
      "learning_rate": 9.940912937863397e-06,
      "loss": 0.9998,
      "step": 2691
    },
    {
      "epoch": 0.07756583875986861,
      "grad_norm": 2.2193682193756104,
      "learning_rate": 9.940841393716249e-06,
      "loss": 0.9788,
      "step": 2692
    },
    {
      "epoch": 0.07759465222151789,
      "grad_norm": 2.168647289276123,
      "learning_rate": 9.940769806539312e-06,
      "loss": 0.9699,
      "step": 2693
    },
    {
      "epoch": 0.07762346568316718,
      "grad_norm": 2.2808709144592285,
      "learning_rate": 9.940698176333207e-06,
      "loss": 0.9551,
      "step": 2694
    },
    {
      "epoch": 0.07765227914481646,
      "grad_norm": 2.241556167602539,
      "learning_rate": 9.940626503098562e-06,
      "loss": 0.9793,
      "step": 2695
    },
    {
      "epoch": 0.07768109260646575,
      "grad_norm": 2.2028446197509766,
      "learning_rate": 9.940554786836e-06,
      "loss": 0.9576,
      "step": 2696
    },
    {
      "epoch": 0.07770990606811502,
      "grad_norm": 2.291069746017456,
      "learning_rate": 9.940483027546144e-06,
      "loss": 0.9873,
      "step": 2697
    },
    {
      "epoch": 0.0777387195297643,
      "grad_norm": 2.202850341796875,
      "learning_rate": 9.940411225229619e-06,
      "loss": 0.9737,
      "step": 2698
    },
    {
      "epoch": 0.07776753299141359,
      "grad_norm": 2.2190887928009033,
      "learning_rate": 9.94033937988705e-06,
      "loss": 0.996,
      "step": 2699
    },
    {
      "epoch": 0.07779634645306287,
      "grad_norm": 2.149923801422119,
      "learning_rate": 9.940267491519064e-06,
      "loss": 0.982,
      "step": 2700
    },
    {
      "epoch": 0.07782515991471216,
      "grad_norm": 2.226970672607422,
      "learning_rate": 9.940195560126286e-06,
      "loss": 0.9574,
      "step": 2701
    },
    {
      "epoch": 0.07785397337636143,
      "grad_norm": 2.165456771850586,
      "learning_rate": 9.940123585709345e-06,
      "loss": 0.939,
      "step": 2702
    },
    {
      "epoch": 0.07788278683801071,
      "grad_norm": 2.0783801078796387,
      "learning_rate": 9.940051568268864e-06,
      "loss": 0.9459,
      "step": 2703
    },
    {
      "epoch": 0.07791160029966,
      "grad_norm": 2.291491985321045,
      "learning_rate": 9.939979507805473e-06,
      "loss": 0.9781,
      "step": 2704
    },
    {
      "epoch": 0.07794041376130928,
      "grad_norm": 2.098557949066162,
      "learning_rate": 9.939907404319797e-06,
      "loss": 0.9462,
      "step": 2705
    },
    {
      "epoch": 0.07796922722295857,
      "grad_norm": 2.0919201374053955,
      "learning_rate": 9.939835257812468e-06,
      "loss": 0.9634,
      "step": 2706
    },
    {
      "epoch": 0.07799804068460785,
      "grad_norm": 2.208944320678711,
      "learning_rate": 9.939763068284111e-06,
      "loss": 0.9737,
      "step": 2707
    },
    {
      "epoch": 0.07802685414625714,
      "grad_norm": 2.104077100753784,
      "learning_rate": 9.939690835735356e-06,
      "loss": 0.9683,
      "step": 2708
    },
    {
      "epoch": 0.07805566760790641,
      "grad_norm": 2.2597970962524414,
      "learning_rate": 9.93961856016683e-06,
      "loss": 0.9577,
      "step": 2709
    },
    {
      "epoch": 0.07808448106955569,
      "grad_norm": 2.174929618835449,
      "learning_rate": 9.939546241579166e-06,
      "loss": 0.9768,
      "step": 2710
    },
    {
      "epoch": 0.07811329453120498,
      "grad_norm": 2.1450693607330322,
      "learning_rate": 9.939473879972991e-06,
      "loss": 1.0116,
      "step": 2711
    },
    {
      "epoch": 0.07814210799285426,
      "grad_norm": 2.270179510116577,
      "learning_rate": 9.939401475348938e-06,
      "loss": 0.9672,
      "step": 2712
    },
    {
      "epoch": 0.07817092145450355,
      "grad_norm": 2.0644688606262207,
      "learning_rate": 9.939329027707634e-06,
      "loss": 0.9849,
      "step": 2713
    },
    {
      "epoch": 0.07819973491615283,
      "grad_norm": 2.246748447418213,
      "learning_rate": 9.93925653704971e-06,
      "loss": 0.9832,
      "step": 2714
    },
    {
      "epoch": 0.0782285483778021,
      "grad_norm": 2.2338523864746094,
      "learning_rate": 9.939184003375803e-06,
      "loss": 0.9651,
      "step": 2715
    },
    {
      "epoch": 0.0782573618394514,
      "grad_norm": 2.275984287261963,
      "learning_rate": 9.939111426686538e-06,
      "loss": 0.9968,
      "step": 2716
    },
    {
      "epoch": 0.07828617530110067,
      "grad_norm": 2.6048219203948975,
      "learning_rate": 9.939038806982551e-06,
      "loss": 1.0032,
      "step": 2717
    },
    {
      "epoch": 0.07831498876274996,
      "grad_norm": 2.1670169830322266,
      "learning_rate": 9.938966144264473e-06,
      "loss": 0.9776,
      "step": 2718
    },
    {
      "epoch": 0.07834380222439924,
      "grad_norm": 2.4084770679473877,
      "learning_rate": 9.938893438532936e-06,
      "loss": 0.9652,
      "step": 2719
    },
    {
      "epoch": 0.07837261568604852,
      "grad_norm": 2.2705554962158203,
      "learning_rate": 9.938820689788575e-06,
      "loss": 0.9814,
      "step": 2720
    },
    {
      "epoch": 0.07840142914769781,
      "grad_norm": 2.0980546474456787,
      "learning_rate": 9.938747898032022e-06,
      "loss": 0.9694,
      "step": 2721
    },
    {
      "epoch": 0.07843024260934708,
      "grad_norm": 2.5220131874084473,
      "learning_rate": 9.938675063263914e-06,
      "loss": 0.9848,
      "step": 2722
    },
    {
      "epoch": 0.07845905607099637,
      "grad_norm": 2.3763201236724854,
      "learning_rate": 9.938602185484881e-06,
      "loss": 0.9617,
      "step": 2723
    },
    {
      "epoch": 0.07848786953264565,
      "grad_norm": 2.2300076484680176,
      "learning_rate": 9.93852926469556e-06,
      "loss": 0.973,
      "step": 2724
    },
    {
      "epoch": 0.07851668299429493,
      "grad_norm": 2.321465015411377,
      "learning_rate": 9.938456300896587e-06,
      "loss": 0.9932,
      "step": 2725
    },
    {
      "epoch": 0.07854549645594422,
      "grad_norm": 2.3839590549468994,
      "learning_rate": 9.938383294088596e-06,
      "loss": 0.9837,
      "step": 2726
    },
    {
      "epoch": 0.0785743099175935,
      "grad_norm": 2.316938877105713,
      "learning_rate": 9.938310244272222e-06,
      "loss": 1.0062,
      "step": 2727
    },
    {
      "epoch": 0.07860312337924279,
      "grad_norm": 2.390044689178467,
      "learning_rate": 9.938237151448102e-06,
      "loss": 0.9744,
      "step": 2728
    },
    {
      "epoch": 0.07863193684089206,
      "grad_norm": 2.2073583602905273,
      "learning_rate": 9.938164015616874e-06,
      "loss": 0.9372,
      "step": 2729
    },
    {
      "epoch": 0.07866075030254134,
      "grad_norm": 2.127246141433716,
      "learning_rate": 9.938090836779175e-06,
      "loss": 0.9557,
      "step": 2730
    },
    {
      "epoch": 0.07868956376419063,
      "grad_norm": 2.220980167388916,
      "learning_rate": 9.938017614935637e-06,
      "loss": 0.9613,
      "step": 2731
    },
    {
      "epoch": 0.07871837722583991,
      "grad_norm": 1.9075053930282593,
      "learning_rate": 9.937944350086905e-06,
      "loss": 0.9608,
      "step": 2732
    },
    {
      "epoch": 0.0787471906874892,
      "grad_norm": 2.150620460510254,
      "learning_rate": 9.937871042233612e-06,
      "loss": 0.9714,
      "step": 2733
    },
    {
      "epoch": 0.07877600414913848,
      "grad_norm": 2.1144633293151855,
      "learning_rate": 9.937797691376401e-06,
      "loss": 0.9945,
      "step": 2734
    },
    {
      "epoch": 0.07880481761078777,
      "grad_norm": 2.0945706367492676,
      "learning_rate": 9.937724297515907e-06,
      "loss": 0.9808,
      "step": 2735
    },
    {
      "epoch": 0.07883363107243704,
      "grad_norm": 2.0997064113616943,
      "learning_rate": 9.93765086065277e-06,
      "loss": 0.9667,
      "step": 2736
    },
    {
      "epoch": 0.07886244453408632,
      "grad_norm": 2.1839728355407715,
      "learning_rate": 9.937577380787629e-06,
      "loss": 0.9699,
      "step": 2737
    },
    {
      "epoch": 0.07889125799573561,
      "grad_norm": 2.284444808959961,
      "learning_rate": 9.937503857921126e-06,
      "loss": 0.9786,
      "step": 2738
    },
    {
      "epoch": 0.07892007145738489,
      "grad_norm": 2.156270980834961,
      "learning_rate": 9.9374302920539e-06,
      "loss": 1.0096,
      "step": 2739
    },
    {
      "epoch": 0.07894888491903418,
      "grad_norm": 2.149846076965332,
      "learning_rate": 9.93735668318659e-06,
      "loss": 1.0011,
      "step": 2740
    },
    {
      "epoch": 0.07897769838068346,
      "grad_norm": 2.310809850692749,
      "learning_rate": 9.93728303131984e-06,
      "loss": 0.978,
      "step": 2741
    },
    {
      "epoch": 0.07900651184233273,
      "grad_norm": 2.1772682666778564,
      "learning_rate": 9.937209336454292e-06,
      "loss": 0.959,
      "step": 2742
    },
    {
      "epoch": 0.07903532530398202,
      "grad_norm": 2.20652174949646,
      "learning_rate": 9.937135598590583e-06,
      "loss": 1.0334,
      "step": 2743
    },
    {
      "epoch": 0.0790641387656313,
      "grad_norm": 2.1234378814697266,
      "learning_rate": 9.937061817729361e-06,
      "loss": 0.9931,
      "step": 2744
    },
    {
      "epoch": 0.07909295222728059,
      "grad_norm": 2.0705044269561768,
      "learning_rate": 9.936987993871265e-06,
      "loss": 0.9762,
      "step": 2745
    },
    {
      "epoch": 0.07912176568892987,
      "grad_norm": 2.112032175064087,
      "learning_rate": 9.936914127016937e-06,
      "loss": 0.9531,
      "step": 2746
    },
    {
      "epoch": 0.07915057915057915,
      "grad_norm": 2.2752792835235596,
      "learning_rate": 9.936840217167026e-06,
      "loss": 0.9522,
      "step": 2747
    },
    {
      "epoch": 0.07917939261222844,
      "grad_norm": 2.140674114227295,
      "learning_rate": 9.936766264322169e-06,
      "loss": 1.005,
      "step": 2748
    },
    {
      "epoch": 0.07920820607387771,
      "grad_norm": 2.196394681930542,
      "learning_rate": 9.936692268483012e-06,
      "loss": 0.9744,
      "step": 2749
    },
    {
      "epoch": 0.079237019535527,
      "grad_norm": 2.119417428970337,
      "learning_rate": 9.936618229650202e-06,
      "loss": 0.9788,
      "step": 2750
    },
    {
      "epoch": 0.07926583299717628,
      "grad_norm": 2.22542667388916,
      "learning_rate": 9.93654414782438e-06,
      "loss": 0.9684,
      "step": 2751
    },
    {
      "epoch": 0.07929464645882556,
      "grad_norm": 2.1288704872131348,
      "learning_rate": 9.936470023006194e-06,
      "loss": 1.0234,
      "step": 2752
    },
    {
      "epoch": 0.07932345992047485,
      "grad_norm": 2.192265748977661,
      "learning_rate": 9.93639585519629e-06,
      "loss": 0.9649,
      "step": 2753
    },
    {
      "epoch": 0.07935227338212412,
      "grad_norm": 2.124159812927246,
      "learning_rate": 9.936321644395313e-06,
      "loss": 0.9918,
      "step": 2754
    },
    {
      "epoch": 0.07938108684377342,
      "grad_norm": 2.1303508281707764,
      "learning_rate": 9.936247390603907e-06,
      "loss": 0.956,
      "step": 2755
    },
    {
      "epoch": 0.07940990030542269,
      "grad_norm": 2.090057611465454,
      "learning_rate": 9.93617309382272e-06,
      "loss": 0.9911,
      "step": 2756
    },
    {
      "epoch": 0.07943871376707197,
      "grad_norm": 2.117151975631714,
      "learning_rate": 9.936098754052403e-06,
      "loss": 0.9853,
      "step": 2757
    },
    {
      "epoch": 0.07946752722872126,
      "grad_norm": 2.2851924896240234,
      "learning_rate": 9.936024371293598e-06,
      "loss": 0.9879,
      "step": 2758
    },
    {
      "epoch": 0.07949634069037054,
      "grad_norm": 2.096491813659668,
      "learning_rate": 9.935949945546955e-06,
      "loss": 0.9885,
      "step": 2759
    },
    {
      "epoch": 0.07952515415201983,
      "grad_norm": 2.1147282123565674,
      "learning_rate": 9.935875476813124e-06,
      "loss": 1.0056,
      "step": 2760
    },
    {
      "epoch": 0.0795539676136691,
      "grad_norm": 2.151289463043213,
      "learning_rate": 9.93580096509275e-06,
      "loss": 0.9805,
      "step": 2761
    },
    {
      "epoch": 0.0795827810753184,
      "grad_norm": 2.196347713470459,
      "learning_rate": 9.935726410386484e-06,
      "loss": 0.9791,
      "step": 2762
    },
    {
      "epoch": 0.07961159453696767,
      "grad_norm": 2.140389919281006,
      "learning_rate": 9.935651812694973e-06,
      "loss": 0.9651,
      "step": 2763
    },
    {
      "epoch": 0.07964040799861695,
      "grad_norm": 2.136211633682251,
      "learning_rate": 9.93557717201887e-06,
      "loss": 0.9747,
      "step": 2764
    },
    {
      "epoch": 0.07966922146026624,
      "grad_norm": 2.2693324089050293,
      "learning_rate": 9.935502488358825e-06,
      "loss": 0.9854,
      "step": 2765
    },
    {
      "epoch": 0.07969803492191552,
      "grad_norm": 2.212010383605957,
      "learning_rate": 9.935427761715486e-06,
      "loss": 0.998,
      "step": 2766
    },
    {
      "epoch": 0.07972684838356481,
      "grad_norm": 2.0497641563415527,
      "learning_rate": 9.935352992089505e-06,
      "loss": 0.9643,
      "step": 2767
    },
    {
      "epoch": 0.07975566184521408,
      "grad_norm": 2.0942087173461914,
      "learning_rate": 9.93527817948153e-06,
      "loss": 0.9304,
      "step": 2768
    },
    {
      "epoch": 0.07978447530686336,
      "grad_norm": 2.345224380493164,
      "learning_rate": 9.93520332389222e-06,
      "loss": 0.981,
      "step": 2769
    },
    {
      "epoch": 0.07981328876851265,
      "grad_norm": 2.0269484519958496,
      "learning_rate": 9.93512842532222e-06,
      "loss": 0.9829,
      "step": 2770
    },
    {
      "epoch": 0.07984210223016193,
      "grad_norm": 2.3314104080200195,
      "learning_rate": 9.935053483772186e-06,
      "loss": 0.973,
      "step": 2771
    },
    {
      "epoch": 0.07987091569181122,
      "grad_norm": 2.1984477043151855,
      "learning_rate": 9.93497849924277e-06,
      "loss": 0.9765,
      "step": 2772
    },
    {
      "epoch": 0.0798997291534605,
      "grad_norm": 2.046936511993408,
      "learning_rate": 9.934903471734622e-06,
      "loss": 0.9873,
      "step": 2773
    },
    {
      "epoch": 0.07992854261510977,
      "grad_norm": 2.419588565826416,
      "learning_rate": 9.9348284012484e-06,
      "loss": 0.9794,
      "step": 2774
    },
    {
      "epoch": 0.07995735607675906,
      "grad_norm": 2.0603530406951904,
      "learning_rate": 9.934753287784755e-06,
      "loss": 0.984,
      "step": 2775
    },
    {
      "epoch": 0.07998616953840834,
      "grad_norm": 2.2062432765960693,
      "learning_rate": 9.93467813134434e-06,
      "loss": 0.9687,
      "step": 2776
    },
    {
      "epoch": 0.08001498300005763,
      "grad_norm": 2.211176633834839,
      "learning_rate": 9.934602931927812e-06,
      "loss": 1.0242,
      "step": 2777
    },
    {
      "epoch": 0.08004379646170691,
      "grad_norm": 2.066373348236084,
      "learning_rate": 9.934527689535827e-06,
      "loss": 0.9725,
      "step": 2778
    },
    {
      "epoch": 0.08007260992335619,
      "grad_norm": 2.1312289237976074,
      "learning_rate": 9.934452404169038e-06,
      "loss": 0.9811,
      "step": 2779
    },
    {
      "epoch": 0.08010142338500548,
      "grad_norm": 2.287449359893799,
      "learning_rate": 9.934377075828102e-06,
      "loss": 1.0033,
      "step": 2780
    },
    {
      "epoch": 0.08013023684665475,
      "grad_norm": 2.1063430309295654,
      "learning_rate": 9.93430170451367e-06,
      "loss": 0.9955,
      "step": 2781
    },
    {
      "epoch": 0.08015905030830404,
      "grad_norm": 2.3982560634613037,
      "learning_rate": 9.934226290226407e-06,
      "loss": 0.972,
      "step": 2782
    },
    {
      "epoch": 0.08018786376995332,
      "grad_norm": 2.230274200439453,
      "learning_rate": 9.934150832966964e-06,
      "loss": 0.9936,
      "step": 2783
    },
    {
      "epoch": 0.0802166772316026,
      "grad_norm": 2.191537857055664,
      "learning_rate": 9.934075332736002e-06,
      "loss": 0.9824,
      "step": 2784
    },
    {
      "epoch": 0.08024549069325189,
      "grad_norm": 2.149782180786133,
      "learning_rate": 9.933999789534173e-06,
      "loss": 0.9797,
      "step": 2785
    },
    {
      "epoch": 0.08027430415490117,
      "grad_norm": 1.9971328973770142,
      "learning_rate": 9.933924203362138e-06,
      "loss": 0.9458,
      "step": 2786
    },
    {
      "epoch": 0.08030311761655046,
      "grad_norm": 2.2620975971221924,
      "learning_rate": 9.933848574220556e-06,
      "loss": 0.9782,
      "step": 2787
    },
    {
      "epoch": 0.08033193107819973,
      "grad_norm": 2.2289388179779053,
      "learning_rate": 9.933772902110085e-06,
      "loss": 0.9599,
      "step": 2788
    },
    {
      "epoch": 0.08036074453984901,
      "grad_norm": 2.169830560684204,
      "learning_rate": 9.933697187031386e-06,
      "loss": 0.9258,
      "step": 2789
    },
    {
      "epoch": 0.0803895580014983,
      "grad_norm": 2.0567238330841064,
      "learning_rate": 9.933621428985115e-06,
      "loss": 0.9259,
      "step": 2790
    },
    {
      "epoch": 0.08041837146314758,
      "grad_norm": 2.122612476348877,
      "learning_rate": 9.933545627971932e-06,
      "loss": 0.9611,
      "step": 2791
    },
    {
      "epoch": 0.08044718492479687,
      "grad_norm": 2.2567100524902344,
      "learning_rate": 9.933469783992499e-06,
      "loss": 0.9517,
      "step": 2792
    },
    {
      "epoch": 0.08047599838644615,
      "grad_norm": 2.158280849456787,
      "learning_rate": 9.933393897047476e-06,
      "loss": 0.9901,
      "step": 2793
    },
    {
      "epoch": 0.08050481184809544,
      "grad_norm": 2.299952507019043,
      "learning_rate": 9.933317967137523e-06,
      "loss": 0.9559,
      "step": 2794
    },
    {
      "epoch": 0.08053362530974471,
      "grad_norm": 2.2527451515197754,
      "learning_rate": 9.933241994263303e-06,
      "loss": 0.9668,
      "step": 2795
    },
    {
      "epoch": 0.08056243877139399,
      "grad_norm": 2.1724400520324707,
      "learning_rate": 9.933165978425477e-06,
      "loss": 0.9827,
      "step": 2796
    },
    {
      "epoch": 0.08059125223304328,
      "grad_norm": 2.1543948650360107,
      "learning_rate": 9.933089919624706e-06,
      "loss": 0.9756,
      "step": 2797
    },
    {
      "epoch": 0.08062006569469256,
      "grad_norm": 2.2229833602905273,
      "learning_rate": 9.933013817861653e-06,
      "loss": 0.9705,
      "step": 2798
    },
    {
      "epoch": 0.08064887915634185,
      "grad_norm": 2.2237391471862793,
      "learning_rate": 9.93293767313698e-06,
      "loss": 0.9686,
      "step": 2799
    },
    {
      "epoch": 0.08067769261799113,
      "grad_norm": 2.27683424949646,
      "learning_rate": 9.932861485451353e-06,
      "loss": 0.9518,
      "step": 2800
    },
    {
      "epoch": 0.0807065060796404,
      "grad_norm": 2.2133870124816895,
      "learning_rate": 9.932785254805432e-06,
      "loss": 0.9614,
      "step": 2801
    },
    {
      "epoch": 0.0807353195412897,
      "grad_norm": 2.30541729927063,
      "learning_rate": 9.932708981199884e-06,
      "loss": 0.9712,
      "step": 2802
    },
    {
      "epoch": 0.08076413300293897,
      "grad_norm": 2.1537179946899414,
      "learning_rate": 9.93263266463537e-06,
      "loss": 0.981,
      "step": 2803
    },
    {
      "epoch": 0.08079294646458826,
      "grad_norm": 2.031431198120117,
      "learning_rate": 9.932556305112557e-06,
      "loss": 0.9437,
      "step": 2804
    },
    {
      "epoch": 0.08082175992623754,
      "grad_norm": 2.2802836894989014,
      "learning_rate": 9.93247990263211e-06,
      "loss": 0.9673,
      "step": 2805
    },
    {
      "epoch": 0.08085057338788681,
      "grad_norm": 2.2891457080841064,
      "learning_rate": 9.932403457194693e-06,
      "loss": 0.9856,
      "step": 2806
    },
    {
      "epoch": 0.0808793868495361,
      "grad_norm": 2.276081085205078,
      "learning_rate": 9.932326968800971e-06,
      "loss": 0.983,
      "step": 2807
    },
    {
      "epoch": 0.08090820031118538,
      "grad_norm": 2.233408212661743,
      "learning_rate": 9.932250437451614e-06,
      "loss": 0.9508,
      "step": 2808
    },
    {
      "epoch": 0.08093701377283467,
      "grad_norm": 2.159149646759033,
      "learning_rate": 9.932173863147286e-06,
      "loss": 0.9661,
      "step": 2809
    },
    {
      "epoch": 0.08096582723448395,
      "grad_norm": 2.132706880569458,
      "learning_rate": 9.932097245888653e-06,
      "loss": 0.9483,
      "step": 2810
    },
    {
      "epoch": 0.08099464069613323,
      "grad_norm": 2.25028133392334,
      "learning_rate": 9.932020585676383e-06,
      "loss": 0.9865,
      "step": 2811
    },
    {
      "epoch": 0.08102345415778252,
      "grad_norm": 2.1224138736724854,
      "learning_rate": 9.931943882511144e-06,
      "loss": 1.0009,
      "step": 2812
    },
    {
      "epoch": 0.0810522676194318,
      "grad_norm": 2.1097748279571533,
      "learning_rate": 9.931867136393604e-06,
      "loss": 0.943,
      "step": 2813
    },
    {
      "epoch": 0.08108108108108109,
      "grad_norm": 2.2535507678985596,
      "learning_rate": 9.931790347324433e-06,
      "loss": 0.9507,
      "step": 2814
    },
    {
      "epoch": 0.08110989454273036,
      "grad_norm": 2.0148444175720215,
      "learning_rate": 9.931713515304295e-06,
      "loss": 0.9707,
      "step": 2815
    },
    {
      "epoch": 0.08113870800437964,
      "grad_norm": 2.149611234664917,
      "learning_rate": 9.931636640333866e-06,
      "loss": 0.9749,
      "step": 2816
    },
    {
      "epoch": 0.08116752146602893,
      "grad_norm": 2.206902503967285,
      "learning_rate": 9.931559722413809e-06,
      "loss": 0.9682,
      "step": 2817
    },
    {
      "epoch": 0.08119633492767821,
      "grad_norm": 2.134965658187866,
      "learning_rate": 9.931482761544797e-06,
      "loss": 1.0005,
      "step": 2818
    },
    {
      "epoch": 0.0812251483893275,
      "grad_norm": 1.8768082857131958,
      "learning_rate": 9.931405757727499e-06,
      "loss": 0.9661,
      "step": 2819
    },
    {
      "epoch": 0.08125396185097677,
      "grad_norm": 2.1223700046539307,
      "learning_rate": 9.931328710962589e-06,
      "loss": 0.9703,
      "step": 2820
    },
    {
      "epoch": 0.08128277531262607,
      "grad_norm": 2.021552324295044,
      "learning_rate": 9.931251621250733e-06,
      "loss": 0.9767,
      "step": 2821
    },
    {
      "epoch": 0.08131158877427534,
      "grad_norm": 2.219550371170044,
      "learning_rate": 9.931174488592605e-06,
      "loss": 0.9587,
      "step": 2822
    },
    {
      "epoch": 0.08134040223592462,
      "grad_norm": 2.1259841918945312,
      "learning_rate": 9.931097312988878e-06,
      "loss": 0.9999,
      "step": 2823
    },
    {
      "epoch": 0.08136921569757391,
      "grad_norm": 2.170210123062134,
      "learning_rate": 9.931020094440223e-06,
      "loss": 0.9403,
      "step": 2824
    },
    {
      "epoch": 0.08139802915922319,
      "grad_norm": 2.0631191730499268,
      "learning_rate": 9.93094283294731e-06,
      "loss": 0.9541,
      "step": 2825
    },
    {
      "epoch": 0.08142684262087248,
      "grad_norm": 2.24965238571167,
      "learning_rate": 9.930865528510816e-06,
      "loss": 0.9895,
      "step": 2826
    },
    {
      "epoch": 0.08145565608252175,
      "grad_norm": 2.1449594497680664,
      "learning_rate": 9.930788181131413e-06,
      "loss": 0.9739,
      "step": 2827
    },
    {
      "epoch": 0.08148446954417103,
      "grad_norm": 2.095551013946533,
      "learning_rate": 9.93071079080977e-06,
      "loss": 0.9756,
      "step": 2828
    },
    {
      "epoch": 0.08151328300582032,
      "grad_norm": 2.1649200916290283,
      "learning_rate": 9.93063335754657e-06,
      "loss": 0.9552,
      "step": 2829
    },
    {
      "epoch": 0.0815420964674696,
      "grad_norm": 2.012016773223877,
      "learning_rate": 9.930555881342479e-06,
      "loss": 0.9631,
      "step": 2830
    },
    {
      "epoch": 0.08157090992911889,
      "grad_norm": 2.1958632469177246,
      "learning_rate": 9.930478362198175e-06,
      "loss": 0.9591,
      "step": 2831
    },
    {
      "epoch": 0.08159972339076817,
      "grad_norm": 2.2113289833068848,
      "learning_rate": 9.930400800114334e-06,
      "loss": 0.9523,
      "step": 2832
    },
    {
      "epoch": 0.08162853685241744,
      "grad_norm": 2.073514461517334,
      "learning_rate": 9.93032319509163e-06,
      "loss": 0.9685,
      "step": 2833
    },
    {
      "epoch": 0.08165735031406673,
      "grad_norm": 2.2938036918640137,
      "learning_rate": 9.93024554713074e-06,
      "loss": 0.9727,
      "step": 2834
    },
    {
      "epoch": 0.08168616377571601,
      "grad_norm": 2.219327926635742,
      "learning_rate": 9.93016785623234e-06,
      "loss": 0.9773,
      "step": 2835
    },
    {
      "epoch": 0.0817149772373653,
      "grad_norm": 2.0267083644866943,
      "learning_rate": 9.930090122397106e-06,
      "loss": 0.962,
      "step": 2836
    },
    {
      "epoch": 0.08174379069901458,
      "grad_norm": 2.3331425189971924,
      "learning_rate": 9.930012345625715e-06,
      "loss": 0.9938,
      "step": 2837
    },
    {
      "epoch": 0.08177260416066386,
      "grad_norm": 2.056718111038208,
      "learning_rate": 9.929934525918845e-06,
      "loss": 0.9454,
      "step": 2838
    },
    {
      "epoch": 0.08180141762231315,
      "grad_norm": 2.2671544551849365,
      "learning_rate": 9.929856663277173e-06,
      "loss": 0.9703,
      "step": 2839
    },
    {
      "epoch": 0.08183023108396242,
      "grad_norm": 2.276638984680176,
      "learning_rate": 9.929778757701379e-06,
      "loss": 0.9676,
      "step": 2840
    },
    {
      "epoch": 0.08185904454561171,
      "grad_norm": 2.0914573669433594,
      "learning_rate": 9.929700809192139e-06,
      "loss": 0.9798,
      "step": 2841
    },
    {
      "epoch": 0.08188785800726099,
      "grad_norm": 2.204667091369629,
      "learning_rate": 9.929622817750133e-06,
      "loss": 0.9976,
      "step": 2842
    },
    {
      "epoch": 0.08191667146891027,
      "grad_norm": 2.3135669231414795,
      "learning_rate": 9.92954478337604e-06,
      "loss": 0.9833,
      "step": 2843
    },
    {
      "epoch": 0.08194548493055956,
      "grad_norm": 2.1456947326660156,
      "learning_rate": 9.929466706070538e-06,
      "loss": 0.9645,
      "step": 2844
    },
    {
      "epoch": 0.08197429839220884,
      "grad_norm": 2.2303459644317627,
      "learning_rate": 9.929388585834313e-06,
      "loss": 0.9717,
      "step": 2845
    },
    {
      "epoch": 0.08200311185385813,
      "grad_norm": 2.2035465240478516,
      "learning_rate": 9.929310422668037e-06,
      "loss": 0.9691,
      "step": 2846
    },
    {
      "epoch": 0.0820319253155074,
      "grad_norm": 2.212186098098755,
      "learning_rate": 9.929232216572395e-06,
      "loss": 0.9671,
      "step": 2847
    },
    {
      "epoch": 0.0820607387771567,
      "grad_norm": 2.11682391166687,
      "learning_rate": 9.92915396754807e-06,
      "loss": 0.9823,
      "step": 2848
    },
    {
      "epoch": 0.08208955223880597,
      "grad_norm": 2.1729896068573,
      "learning_rate": 9.92907567559574e-06,
      "loss": 0.9695,
      "step": 2849
    },
    {
      "epoch": 0.08211836570045525,
      "grad_norm": 2.330381155014038,
      "learning_rate": 9.928997340716087e-06,
      "loss": 0.964,
      "step": 2850
    },
    {
      "epoch": 0.08214717916210454,
      "grad_norm": 2.139496088027954,
      "learning_rate": 9.928918962909797e-06,
      "loss": 0.9918,
      "step": 2851
    },
    {
      "epoch": 0.08217599262375382,
      "grad_norm": 2.1446263790130615,
      "learning_rate": 9.928840542177548e-06,
      "loss": 0.9548,
      "step": 2852
    },
    {
      "epoch": 0.0822048060854031,
      "grad_norm": 2.1898868083953857,
      "learning_rate": 9.928762078520025e-06,
      "loss": 0.9803,
      "step": 2853
    },
    {
      "epoch": 0.08223361954705238,
      "grad_norm": 2.1276051998138428,
      "learning_rate": 9.928683571937912e-06,
      "loss": 0.9681,
      "step": 2854
    },
    {
      "epoch": 0.08226243300870166,
      "grad_norm": 2.0289270877838135,
      "learning_rate": 9.928605022431893e-06,
      "loss": 0.9796,
      "step": 2855
    },
    {
      "epoch": 0.08229124647035095,
      "grad_norm": 2.174251079559326,
      "learning_rate": 9.928526430002648e-06,
      "loss": 0.9824,
      "step": 2856
    },
    {
      "epoch": 0.08232005993200023,
      "grad_norm": 2.0217995643615723,
      "learning_rate": 9.928447794650866e-06,
      "loss": 0.9513,
      "step": 2857
    },
    {
      "epoch": 0.08234887339364952,
      "grad_norm": 2.014677047729492,
      "learning_rate": 9.928369116377231e-06,
      "loss": 0.965,
      "step": 2858
    },
    {
      "epoch": 0.0823776868552988,
      "grad_norm": 2.251401424407959,
      "learning_rate": 9.928290395182427e-06,
      "loss": 0.9329,
      "step": 2859
    },
    {
      "epoch": 0.08240650031694807,
      "grad_norm": 2.0145764350891113,
      "learning_rate": 9.928211631067141e-06,
      "loss": 0.9657,
      "step": 2860
    },
    {
      "epoch": 0.08243531377859736,
      "grad_norm": 2.089280605316162,
      "learning_rate": 9.928132824032057e-06,
      "loss": 0.9411,
      "step": 2861
    },
    {
      "epoch": 0.08246412724024664,
      "grad_norm": 2.0496859550476074,
      "learning_rate": 9.928053974077863e-06,
      "loss": 0.9801,
      "step": 2862
    },
    {
      "epoch": 0.08249294070189593,
      "grad_norm": 2.0061709880828857,
      "learning_rate": 9.927975081205246e-06,
      "loss": 0.9843,
      "step": 2863
    },
    {
      "epoch": 0.08252175416354521,
      "grad_norm": 1.9370445013046265,
      "learning_rate": 9.92789614541489e-06,
      "loss": 0.9714,
      "step": 2864
    },
    {
      "epoch": 0.08255056762519448,
      "grad_norm": 2.1698527336120605,
      "learning_rate": 9.927817166707485e-06,
      "loss": 0.9727,
      "step": 2865
    },
    {
      "epoch": 0.08257938108684378,
      "grad_norm": 2.027928113937378,
      "learning_rate": 9.92773814508372e-06,
      "loss": 0.9699,
      "step": 2866
    },
    {
      "epoch": 0.08260819454849305,
      "grad_norm": 2.1460328102111816,
      "learning_rate": 9.927659080544281e-06,
      "loss": 0.9721,
      "step": 2867
    },
    {
      "epoch": 0.08263700801014234,
      "grad_norm": 2.06014347076416,
      "learning_rate": 9.927579973089858e-06,
      "loss": 0.9679,
      "step": 2868
    },
    {
      "epoch": 0.08266582147179162,
      "grad_norm": 2.1603474617004395,
      "learning_rate": 9.927500822721138e-06,
      "loss": 0.9556,
      "step": 2869
    },
    {
      "epoch": 0.0826946349334409,
      "grad_norm": 2.0506649017333984,
      "learning_rate": 9.927421629438811e-06,
      "loss": 0.9529,
      "step": 2870
    },
    {
      "epoch": 0.08272344839509019,
      "grad_norm": 2.0487329959869385,
      "learning_rate": 9.927342393243569e-06,
      "loss": 0.9723,
      "step": 2871
    },
    {
      "epoch": 0.08275226185673946,
      "grad_norm": 2.2806758880615234,
      "learning_rate": 9.9272631141361e-06,
      "loss": 0.9416,
      "step": 2872
    },
    {
      "epoch": 0.08278107531838876,
      "grad_norm": 2.104682683944702,
      "learning_rate": 9.927183792117095e-06,
      "loss": 0.9694,
      "step": 2873
    },
    {
      "epoch": 0.08280988878003803,
      "grad_norm": 2.1124894618988037,
      "learning_rate": 9.927104427187243e-06,
      "loss": 0.9678,
      "step": 2874
    },
    {
      "epoch": 0.08283870224168732,
      "grad_norm": 2.1684699058532715,
      "learning_rate": 9.927025019347239e-06,
      "loss": 0.9712,
      "step": 2875
    },
    {
      "epoch": 0.0828675157033366,
      "grad_norm": 2.402925729751587,
      "learning_rate": 9.92694556859777e-06,
      "loss": 0.9764,
      "step": 2876
    },
    {
      "epoch": 0.08289632916498588,
      "grad_norm": 2.161571741104126,
      "learning_rate": 9.92686607493953e-06,
      "loss": 0.9393,
      "step": 2877
    },
    {
      "epoch": 0.08292514262663517,
      "grad_norm": 2.1546695232391357,
      "learning_rate": 9.926786538373213e-06,
      "loss": 0.9486,
      "step": 2878
    },
    {
      "epoch": 0.08295395608828444,
      "grad_norm": 2.076002836227417,
      "learning_rate": 9.92670695889951e-06,
      "loss": 0.9735,
      "step": 2879
    },
    {
      "epoch": 0.08298276954993374,
      "grad_norm": 2.0329763889312744,
      "learning_rate": 9.926627336519115e-06,
      "loss": 0.9628,
      "step": 2880
    },
    {
      "epoch": 0.08301158301158301,
      "grad_norm": 1.9553191661834717,
      "learning_rate": 9.92654767123272e-06,
      "loss": 0.9669,
      "step": 2881
    },
    {
      "epoch": 0.08304039647323229,
      "grad_norm": 1.9608378410339355,
      "learning_rate": 9.926467963041018e-06,
      "loss": 0.9532,
      "step": 2882
    },
    {
      "epoch": 0.08306920993488158,
      "grad_norm": 1.9304393529891968,
      "learning_rate": 9.926388211944707e-06,
      "loss": 0.9716,
      "step": 2883
    },
    {
      "epoch": 0.08309802339653086,
      "grad_norm": 2.0218570232391357,
      "learning_rate": 9.92630841794448e-06,
      "loss": 0.989,
      "step": 2884
    },
    {
      "epoch": 0.08312683685818015,
      "grad_norm": 2.0714213848114014,
      "learning_rate": 9.926228581041029e-06,
      "loss": 0.9871,
      "step": 2885
    },
    {
      "epoch": 0.08315565031982942,
      "grad_norm": 1.9102534055709839,
      "learning_rate": 9.926148701235051e-06,
      "loss": 0.954,
      "step": 2886
    },
    {
      "epoch": 0.0831844637814787,
      "grad_norm": 2.350364923477173,
      "learning_rate": 9.926068778527244e-06,
      "loss": 0.945,
      "step": 2887
    },
    {
      "epoch": 0.08321327724312799,
      "grad_norm": 2.194718360900879,
      "learning_rate": 9.9259888129183e-06,
      "loss": 0.9988,
      "step": 2888
    },
    {
      "epoch": 0.08324209070477727,
      "grad_norm": 1.9649279117584229,
      "learning_rate": 9.92590880440892e-06,
      "loss": 0.9204,
      "step": 2889
    },
    {
      "epoch": 0.08327090416642656,
      "grad_norm": 2.2192635536193848,
      "learning_rate": 9.925828752999797e-06,
      "loss": 0.9583,
      "step": 2890
    },
    {
      "epoch": 0.08329971762807584,
      "grad_norm": 2.129734992980957,
      "learning_rate": 9.925748658691631e-06,
      "loss": 0.9687,
      "step": 2891
    },
    {
      "epoch": 0.08332853108972511,
      "grad_norm": 1.973003625869751,
      "learning_rate": 9.925668521485117e-06,
      "loss": 0.9655,
      "step": 2892
    },
    {
      "epoch": 0.0833573445513744,
      "grad_norm": 2.2881205081939697,
      "learning_rate": 9.925588341380953e-06,
      "loss": 0.959,
      "step": 2893
    },
    {
      "epoch": 0.08338615801302368,
      "grad_norm": 2.085858106613159,
      "learning_rate": 9.925508118379841e-06,
      "loss": 0.964,
      "step": 2894
    },
    {
      "epoch": 0.08341497147467297,
      "grad_norm": 2.1742212772369385,
      "learning_rate": 9.925427852482475e-06,
      "loss": 0.9322,
      "step": 2895
    },
    {
      "epoch": 0.08344378493632225,
      "grad_norm": 2.111589193344116,
      "learning_rate": 9.925347543689556e-06,
      "loss": 0.9609,
      "step": 2896
    },
    {
      "epoch": 0.08347259839797153,
      "grad_norm": 2.16345477104187,
      "learning_rate": 9.925267192001786e-06,
      "loss": 0.9824,
      "step": 2897
    },
    {
      "epoch": 0.08350141185962082,
      "grad_norm": 2.2645626068115234,
      "learning_rate": 9.925186797419859e-06,
      "loss": 0.9885,
      "step": 2898
    },
    {
      "epoch": 0.0835302253212701,
      "grad_norm": 2.0542590618133545,
      "learning_rate": 9.92510635994448e-06,
      "loss": 0.9851,
      "step": 2899
    },
    {
      "epoch": 0.08355903878291938,
      "grad_norm": 2.154439926147461,
      "learning_rate": 9.925025879576347e-06,
      "loss": 0.9462,
      "step": 2900
    },
    {
      "epoch": 0.08358785224456866,
      "grad_norm": 2.1372392177581787,
      "learning_rate": 9.924945356316163e-06,
      "loss": 0.9627,
      "step": 2901
    },
    {
      "epoch": 0.08361666570621795,
      "grad_norm": 2.221234083175659,
      "learning_rate": 9.924864790164628e-06,
      "loss": 0.9462,
      "step": 2902
    },
    {
      "epoch": 0.08364547916786723,
      "grad_norm": 2.1468803882598877,
      "learning_rate": 9.924784181122442e-06,
      "loss": 0.9442,
      "step": 2903
    },
    {
      "epoch": 0.0836742926295165,
      "grad_norm": 2.1868863105773926,
      "learning_rate": 9.92470352919031e-06,
      "loss": 0.9667,
      "step": 2904
    },
    {
      "epoch": 0.0837031060911658,
      "grad_norm": 2.103848457336426,
      "learning_rate": 9.924622834368934e-06,
      "loss": 0.9578,
      "step": 2905
    },
    {
      "epoch": 0.08373191955281507,
      "grad_norm": 1.9236189126968384,
      "learning_rate": 9.924542096659015e-06,
      "loss": 0.9807,
      "step": 2906
    },
    {
      "epoch": 0.08376073301446436,
      "grad_norm": 2.144848346710205,
      "learning_rate": 9.924461316061257e-06,
      "loss": 0.9622,
      "step": 2907
    },
    {
      "epoch": 0.08378954647611364,
      "grad_norm": 1.9632627964019775,
      "learning_rate": 9.924380492576364e-06,
      "loss": 0.9357,
      "step": 2908
    },
    {
      "epoch": 0.08381835993776292,
      "grad_norm": 2.013915777206421,
      "learning_rate": 9.924299626205039e-06,
      "loss": 0.9928,
      "step": 2909
    },
    {
      "epoch": 0.08384717339941221,
      "grad_norm": 2.166172981262207,
      "learning_rate": 9.924218716947987e-06,
      "loss": 0.9981,
      "step": 2910
    },
    {
      "epoch": 0.08387598686106149,
      "grad_norm": 2.0869879722595215,
      "learning_rate": 9.924137764805913e-06,
      "loss": 0.9522,
      "step": 2911
    },
    {
      "epoch": 0.08390480032271078,
      "grad_norm": 2.100126266479492,
      "learning_rate": 9.924056769779519e-06,
      "loss": 0.9923,
      "step": 2912
    },
    {
      "epoch": 0.08393361378436005,
      "grad_norm": 2.033914566040039,
      "learning_rate": 9.923975731869514e-06,
      "loss": 0.9595,
      "step": 2913
    },
    {
      "epoch": 0.08396242724600933,
      "grad_norm": 2.0576603412628174,
      "learning_rate": 9.923894651076605e-06,
      "loss": 0.9885,
      "step": 2914
    },
    {
      "epoch": 0.08399124070765862,
      "grad_norm": 2.157574415206909,
      "learning_rate": 9.923813527401494e-06,
      "loss": 0.9719,
      "step": 2915
    },
    {
      "epoch": 0.0840200541693079,
      "grad_norm": 2.1859474182128906,
      "learning_rate": 9.923732360844889e-06,
      "loss": 0.9908,
      "step": 2916
    },
    {
      "epoch": 0.08404886763095719,
      "grad_norm": 2.032874822616577,
      "learning_rate": 9.923651151407496e-06,
      "loss": 0.9815,
      "step": 2917
    },
    {
      "epoch": 0.08407768109260647,
      "grad_norm": 1.961538553237915,
      "learning_rate": 9.923569899090025e-06,
      "loss": 0.9509,
      "step": 2918
    },
    {
      "epoch": 0.08410649455425574,
      "grad_norm": 2.199566125869751,
      "learning_rate": 9.923488603893181e-06,
      "loss": 0.9578,
      "step": 2919
    },
    {
      "epoch": 0.08413530801590503,
      "grad_norm": 2.134355306625366,
      "learning_rate": 9.923407265817673e-06,
      "loss": 0.9645,
      "step": 2920
    },
    {
      "epoch": 0.08416412147755431,
      "grad_norm": 2.0338664054870605,
      "learning_rate": 9.92332588486421e-06,
      "loss": 0.9372,
      "step": 2921
    },
    {
      "epoch": 0.0841929349392036,
      "grad_norm": 2.0461270809173584,
      "learning_rate": 9.9232444610335e-06,
      "loss": 0.9315,
      "step": 2922
    },
    {
      "epoch": 0.08422174840085288,
      "grad_norm": 2.213078022003174,
      "learning_rate": 9.923162994326252e-06,
      "loss": 0.9705,
      "step": 2923
    },
    {
      "epoch": 0.08425056186250215,
      "grad_norm": 2.2086122035980225,
      "learning_rate": 9.923081484743176e-06,
      "loss": 0.9773,
      "step": 2924
    },
    {
      "epoch": 0.08427937532415145,
      "grad_norm": 1.903257966041565,
      "learning_rate": 9.92299993228498e-06,
      "loss": 0.9621,
      "step": 2925
    },
    {
      "epoch": 0.08430818878580072,
      "grad_norm": 2.091140031814575,
      "learning_rate": 9.922918336952377e-06,
      "loss": 0.9727,
      "step": 2926
    },
    {
      "epoch": 0.08433700224745001,
      "grad_norm": 2.050372362136841,
      "learning_rate": 9.922836698746077e-06,
      "loss": 0.9637,
      "step": 2927
    },
    {
      "epoch": 0.08436581570909929,
      "grad_norm": 2.0120248794555664,
      "learning_rate": 9.92275501766679e-06,
      "loss": 0.9793,
      "step": 2928
    },
    {
      "epoch": 0.08439462917074857,
      "grad_norm": 2.157655954360962,
      "learning_rate": 9.922673293715227e-06,
      "loss": 0.9738,
      "step": 2929
    },
    {
      "epoch": 0.08442344263239786,
      "grad_norm": 2.026685953140259,
      "learning_rate": 9.9225915268921e-06,
      "loss": 0.9668,
      "step": 2930
    },
    {
      "epoch": 0.08445225609404713,
      "grad_norm": 2.077845335006714,
      "learning_rate": 9.922509717198123e-06,
      "loss": 0.9682,
      "step": 2931
    },
    {
      "epoch": 0.08448106955569643,
      "grad_norm": 2.028949737548828,
      "learning_rate": 9.922427864634006e-06,
      "loss": 0.962,
      "step": 2932
    },
    {
      "epoch": 0.0845098830173457,
      "grad_norm": 2.125418186187744,
      "learning_rate": 9.922345969200465e-06,
      "loss": 0.9854,
      "step": 2933
    },
    {
      "epoch": 0.08453869647899499,
      "grad_norm": 2.054126739501953,
      "learning_rate": 9.922264030898208e-06,
      "loss": 0.9476,
      "step": 2934
    },
    {
      "epoch": 0.08456750994064427,
      "grad_norm": 2.0700759887695312,
      "learning_rate": 9.922182049727955e-06,
      "loss": 0.9545,
      "step": 2935
    },
    {
      "epoch": 0.08459632340229355,
      "grad_norm": 2.012803316116333,
      "learning_rate": 9.922100025690414e-06,
      "loss": 0.9732,
      "step": 2936
    },
    {
      "epoch": 0.08462513686394284,
      "grad_norm": 2.1273951530456543,
      "learning_rate": 9.922017958786303e-06,
      "loss": 0.9958,
      "step": 2937
    },
    {
      "epoch": 0.08465395032559211,
      "grad_norm": 2.042963981628418,
      "learning_rate": 9.921935849016337e-06,
      "loss": 0.9904,
      "step": 2938
    },
    {
      "epoch": 0.0846827637872414,
      "grad_norm": 2.3008599281311035,
      "learning_rate": 9.921853696381228e-06,
      "loss": 0.972,
      "step": 2939
    },
    {
      "epoch": 0.08471157724889068,
      "grad_norm": 1.9656789302825928,
      "learning_rate": 9.921771500881694e-06,
      "loss": 0.9489,
      "step": 2940
    },
    {
      "epoch": 0.08474039071053996,
      "grad_norm": 2.2934443950653076,
      "learning_rate": 9.921689262518452e-06,
      "loss": 0.9834,
      "step": 2941
    },
    {
      "epoch": 0.08476920417218925,
      "grad_norm": 1.9828383922576904,
      "learning_rate": 9.921606981292214e-06,
      "loss": 0.9507,
      "step": 2942
    },
    {
      "epoch": 0.08479801763383853,
      "grad_norm": 2.2128705978393555,
      "learning_rate": 9.921524657203701e-06,
      "loss": 0.9371,
      "step": 2943
    },
    {
      "epoch": 0.08482683109548782,
      "grad_norm": 2.222759246826172,
      "learning_rate": 9.921442290253627e-06,
      "loss": 0.9752,
      "step": 2944
    },
    {
      "epoch": 0.0848556445571371,
      "grad_norm": 2.1965625286102295,
      "learning_rate": 9.921359880442709e-06,
      "loss": 0.9988,
      "step": 2945
    },
    {
      "epoch": 0.08488445801878637,
      "grad_norm": 2.133988857269287,
      "learning_rate": 9.921277427771667e-06,
      "loss": 0.9789,
      "step": 2946
    },
    {
      "epoch": 0.08491327148043566,
      "grad_norm": 2.066587448120117,
      "learning_rate": 9.921194932241219e-06,
      "loss": 0.95,
      "step": 2947
    },
    {
      "epoch": 0.08494208494208494,
      "grad_norm": 2.1159908771514893,
      "learning_rate": 9.921112393852082e-06,
      "loss": 0.9561,
      "step": 2948
    },
    {
      "epoch": 0.08497089840373423,
      "grad_norm": 2.173868179321289,
      "learning_rate": 9.921029812604974e-06,
      "loss": 0.9506,
      "step": 2949
    },
    {
      "epoch": 0.0849997118653835,
      "grad_norm": 2.3621134757995605,
      "learning_rate": 9.920947188500617e-06,
      "loss": 0.9759,
      "step": 2950
    },
    {
      "epoch": 0.08502852532703278,
      "grad_norm": 2.024392604827881,
      "learning_rate": 9.920864521539727e-06,
      "loss": 0.9601,
      "step": 2951
    },
    {
      "epoch": 0.08505733878868207,
      "grad_norm": 2.1784043312072754,
      "learning_rate": 9.92078181172303e-06,
      "loss": 0.9835,
      "step": 2952
    },
    {
      "epoch": 0.08508615225033135,
      "grad_norm": 2.1237077713012695,
      "learning_rate": 9.92069905905124e-06,
      "loss": 0.9408,
      "step": 2953
    },
    {
      "epoch": 0.08511496571198064,
      "grad_norm": 1.900551676750183,
      "learning_rate": 9.92061626352508e-06,
      "loss": 0.9795,
      "step": 2954
    },
    {
      "epoch": 0.08514377917362992,
      "grad_norm": 2.0978074073791504,
      "learning_rate": 9.920533425145272e-06,
      "loss": 0.9497,
      "step": 2955
    },
    {
      "epoch": 0.0851725926352792,
      "grad_norm": 2.0569305419921875,
      "learning_rate": 9.920450543912537e-06,
      "loss": 0.9499,
      "step": 2956
    },
    {
      "epoch": 0.08520140609692849,
      "grad_norm": 1.980980396270752,
      "learning_rate": 9.920367619827595e-06,
      "loss": 0.9499,
      "step": 2957
    },
    {
      "epoch": 0.08523021955857776,
      "grad_norm": 2.1221959590911865,
      "learning_rate": 9.92028465289117e-06,
      "loss": 0.9739,
      "step": 2958
    },
    {
      "epoch": 0.08525903302022705,
      "grad_norm": 2.0501792430877686,
      "learning_rate": 9.920201643103983e-06,
      "loss": 0.9673,
      "step": 2959
    },
    {
      "epoch": 0.08528784648187633,
      "grad_norm": 2.1646618843078613,
      "learning_rate": 9.92011859046676e-06,
      "loss": 0.9608,
      "step": 2960
    },
    {
      "epoch": 0.08531665994352562,
      "grad_norm": 2.2240376472473145,
      "learning_rate": 9.920035494980222e-06,
      "loss": 0.968,
      "step": 2961
    },
    {
      "epoch": 0.0853454734051749,
      "grad_norm": 2.1267263889312744,
      "learning_rate": 9.919952356645093e-06,
      "loss": 0.9839,
      "step": 2962
    },
    {
      "epoch": 0.08537428686682418,
      "grad_norm": 2.265833616256714,
      "learning_rate": 9.919869175462097e-06,
      "loss": 0.9636,
      "step": 2963
    },
    {
      "epoch": 0.08540310032847347,
      "grad_norm": 2.0498385429382324,
      "learning_rate": 9.919785951431959e-06,
      "loss": 0.9717,
      "step": 2964
    },
    {
      "epoch": 0.08543191379012274,
      "grad_norm": 2.271906852722168,
      "learning_rate": 9.919702684555402e-06,
      "loss": 0.954,
      "step": 2965
    },
    {
      "epoch": 0.08546072725177203,
      "grad_norm": 2.2793073654174805,
      "learning_rate": 9.919619374833154e-06,
      "loss": 0.9445,
      "step": 2966
    },
    {
      "epoch": 0.08548954071342131,
      "grad_norm": 2.1579508781433105,
      "learning_rate": 9.919536022265939e-06,
      "loss": 0.9915,
      "step": 2967
    },
    {
      "epoch": 0.08551835417507059,
      "grad_norm": 2.0001206398010254,
      "learning_rate": 9.91945262685448e-06,
      "loss": 0.9524,
      "step": 2968
    },
    {
      "epoch": 0.08554716763671988,
      "grad_norm": 2.0511319637298584,
      "learning_rate": 9.91936918859951e-06,
      "loss": 0.9979,
      "step": 2969
    },
    {
      "epoch": 0.08557598109836916,
      "grad_norm": 2.0187599658966064,
      "learning_rate": 9.919285707501751e-06,
      "loss": 0.9796,
      "step": 2970
    },
    {
      "epoch": 0.08560479456001845,
      "grad_norm": 2.065824508666992,
      "learning_rate": 9.91920218356193e-06,
      "loss": 0.97,
      "step": 2971
    },
    {
      "epoch": 0.08563360802166772,
      "grad_norm": 2.0914742946624756,
      "learning_rate": 9.919118616780777e-06,
      "loss": 0.9833,
      "step": 2972
    },
    {
      "epoch": 0.085662421483317,
      "grad_norm": 3.1573803424835205,
      "learning_rate": 9.919035007159018e-06,
      "loss": 0.9785,
      "step": 2973
    },
    {
      "epoch": 0.08569123494496629,
      "grad_norm": 1.956786870956421,
      "learning_rate": 9.91895135469738e-06,
      "loss": 0.9643,
      "step": 2974
    },
    {
      "epoch": 0.08572004840661557,
      "grad_norm": 2.100423812866211,
      "learning_rate": 9.918867659396592e-06,
      "loss": 0.9596,
      "step": 2975
    },
    {
      "epoch": 0.08574886186826486,
      "grad_norm": 2.02746319770813,
      "learning_rate": 9.918783921257385e-06,
      "loss": 0.9441,
      "step": 2976
    },
    {
      "epoch": 0.08577767532991414,
      "grad_norm": 2.0309383869171143,
      "learning_rate": 9.918700140280488e-06,
      "loss": 0.9761,
      "step": 2977
    },
    {
      "epoch": 0.08580648879156341,
      "grad_norm": 2.037168264389038,
      "learning_rate": 9.918616316466628e-06,
      "loss": 0.9632,
      "step": 2978
    },
    {
      "epoch": 0.0858353022532127,
      "grad_norm": 1.9949301481246948,
      "learning_rate": 9.918532449816537e-06,
      "loss": 0.9395,
      "step": 2979
    },
    {
      "epoch": 0.08586411571486198,
      "grad_norm": 2.0459794998168945,
      "learning_rate": 9.918448540330948e-06,
      "loss": 0.9497,
      "step": 2980
    },
    {
      "epoch": 0.08589292917651127,
      "grad_norm": 2.150765895843506,
      "learning_rate": 9.918364588010585e-06,
      "loss": 0.9596,
      "step": 2981
    },
    {
      "epoch": 0.08592174263816055,
      "grad_norm": 2.2445693016052246,
      "learning_rate": 9.918280592856185e-06,
      "loss": 0.9929,
      "step": 2982
    },
    {
      "epoch": 0.08595055609980982,
      "grad_norm": 2.072519302368164,
      "learning_rate": 9.918196554868477e-06,
      "loss": 0.9498,
      "step": 2983
    },
    {
      "epoch": 0.08597936956145912,
      "grad_norm": 2.1978609561920166,
      "learning_rate": 9.918112474048193e-06,
      "loss": 0.9583,
      "step": 2984
    },
    {
      "epoch": 0.08600818302310839,
      "grad_norm": 2.1171927452087402,
      "learning_rate": 9.918028350396067e-06,
      "loss": 0.976,
      "step": 2985
    },
    {
      "epoch": 0.08603699648475768,
      "grad_norm": 2.043869733810425,
      "learning_rate": 9.917944183912829e-06,
      "loss": 0.9558,
      "step": 2986
    },
    {
      "epoch": 0.08606580994640696,
      "grad_norm": 2.1708579063415527,
      "learning_rate": 9.917859974599213e-06,
      "loss": 0.9501,
      "step": 2987
    },
    {
      "epoch": 0.08609462340805625,
      "grad_norm": 2.038865327835083,
      "learning_rate": 9.917775722455953e-06,
      "loss": 0.9526,
      "step": 2988
    },
    {
      "epoch": 0.08612343686970553,
      "grad_norm": 2.451590061187744,
      "learning_rate": 9.917691427483783e-06,
      "loss": 0.956,
      "step": 2989
    },
    {
      "epoch": 0.0861522503313548,
      "grad_norm": 2.0468010902404785,
      "learning_rate": 9.917607089683436e-06,
      "loss": 0.9447,
      "step": 2990
    },
    {
      "epoch": 0.0861810637930041,
      "grad_norm": 1.9205352067947388,
      "learning_rate": 9.917522709055647e-06,
      "loss": 0.924,
      "step": 2991
    },
    {
      "epoch": 0.08620987725465337,
      "grad_norm": 2.187934398651123,
      "learning_rate": 9.917438285601151e-06,
      "loss": 0.9739,
      "step": 2992
    },
    {
      "epoch": 0.08623869071630266,
      "grad_norm": 2.016303300857544,
      "learning_rate": 9.917353819320685e-06,
      "loss": 0.9082,
      "step": 2993
    },
    {
      "epoch": 0.08626750417795194,
      "grad_norm": 2.0274064540863037,
      "learning_rate": 9.917269310214981e-06,
      "loss": 0.9586,
      "step": 2994
    },
    {
      "epoch": 0.08629631763960122,
      "grad_norm": 1.9904043674468994,
      "learning_rate": 9.917184758284777e-06,
      "loss": 0.9722,
      "step": 2995
    },
    {
      "epoch": 0.08632513110125051,
      "grad_norm": 2.140721321105957,
      "learning_rate": 9.917100163530809e-06,
      "loss": 0.9565,
      "step": 2996
    },
    {
      "epoch": 0.08635394456289978,
      "grad_norm": 2.0741989612579346,
      "learning_rate": 9.917015525953815e-06,
      "loss": 0.9669,
      "step": 2997
    },
    {
      "epoch": 0.08638275802454908,
      "grad_norm": 1.9098289012908936,
      "learning_rate": 9.91693084555453e-06,
      "loss": 0.9614,
      "step": 2998
    },
    {
      "epoch": 0.08641157148619835,
      "grad_norm": 2.223005533218384,
      "learning_rate": 9.916846122333693e-06,
      "loss": 0.9821,
      "step": 2999
    },
    {
      "epoch": 0.08644038494784763,
      "grad_norm": 1.9201546907424927,
      "learning_rate": 9.916761356292039e-06,
      "loss": 0.9926,
      "step": 3000
    },
    {
      "epoch": 0.08646919840949692,
      "grad_norm": 1.8909633159637451,
      "learning_rate": 9.916676547430311e-06,
      "loss": 0.9582,
      "step": 3001
    },
    {
      "epoch": 0.0864980118711462,
      "grad_norm": 2.0139596462249756,
      "learning_rate": 9.916591695749244e-06,
      "loss": 0.9648,
      "step": 3002
    },
    {
      "epoch": 0.08652682533279549,
      "grad_norm": 2.037804365158081,
      "learning_rate": 9.916506801249579e-06,
      "loss": 0.9669,
      "step": 3003
    },
    {
      "epoch": 0.08655563879444476,
      "grad_norm": 2.361933946609497,
      "learning_rate": 9.916421863932054e-06,
      "loss": 0.958,
      "step": 3004
    },
    {
      "epoch": 0.08658445225609404,
      "grad_norm": 1.8589818477630615,
      "learning_rate": 9.91633688379741e-06,
      "loss": 0.9476,
      "step": 3005
    },
    {
      "epoch": 0.08661326571774333,
      "grad_norm": 2.156130790710449,
      "learning_rate": 9.916251860846385e-06,
      "loss": 0.9511,
      "step": 3006
    },
    {
      "epoch": 0.08664207917939261,
      "grad_norm": 1.9139324426651,
      "learning_rate": 9.916166795079721e-06,
      "loss": 0.9783,
      "step": 3007
    },
    {
      "epoch": 0.0866708926410419,
      "grad_norm": 2.0348780155181885,
      "learning_rate": 9.91608168649816e-06,
      "loss": 0.9138,
      "step": 3008
    },
    {
      "epoch": 0.08669970610269118,
      "grad_norm": 1.9156923294067383,
      "learning_rate": 9.91599653510244e-06,
      "loss": 0.9448,
      "step": 3009
    },
    {
      "epoch": 0.08672851956434045,
      "grad_norm": 2.072732925415039,
      "learning_rate": 9.915911340893305e-06,
      "loss": 0.96,
      "step": 3010
    },
    {
      "epoch": 0.08675733302598974,
      "grad_norm": 1.98256516456604,
      "learning_rate": 9.915826103871498e-06,
      "loss": 0.9482,
      "step": 3011
    },
    {
      "epoch": 0.08678614648763902,
      "grad_norm": 2.0977072715759277,
      "learning_rate": 9.915740824037756e-06,
      "loss": 0.9768,
      "step": 3012
    },
    {
      "epoch": 0.08681495994928831,
      "grad_norm": 2.0501151084899902,
      "learning_rate": 9.91565550139283e-06,
      "loss": 0.9777,
      "step": 3013
    },
    {
      "epoch": 0.08684377341093759,
      "grad_norm": 2.1659951210021973,
      "learning_rate": 9.915570135937455e-06,
      "loss": 0.9681,
      "step": 3014
    },
    {
      "epoch": 0.08687258687258688,
      "grad_norm": 2.107836961746216,
      "learning_rate": 9.915484727672378e-06,
      "loss": 0.9732,
      "step": 3015
    },
    {
      "epoch": 0.08690140033423616,
      "grad_norm": 2.096452474594116,
      "learning_rate": 9.915399276598344e-06,
      "loss": 0.9987,
      "step": 3016
    },
    {
      "epoch": 0.08693021379588543,
      "grad_norm": 2.046869993209839,
      "learning_rate": 9.915313782716094e-06,
      "loss": 0.9786,
      "step": 3017
    },
    {
      "epoch": 0.08695902725753472,
      "grad_norm": 2.2091851234436035,
      "learning_rate": 9.915228246026375e-06,
      "loss": 0.9517,
      "step": 3018
    },
    {
      "epoch": 0.086987840719184,
      "grad_norm": 1.9866540431976318,
      "learning_rate": 9.915142666529934e-06,
      "loss": 0.9735,
      "step": 3019
    },
    {
      "epoch": 0.08701665418083329,
      "grad_norm": 2.1686935424804688,
      "learning_rate": 9.915057044227512e-06,
      "loss": 0.9482,
      "step": 3020
    },
    {
      "epoch": 0.08704546764248257,
      "grad_norm": 2.274658203125,
      "learning_rate": 9.914971379119856e-06,
      "loss": 0.9789,
      "step": 3021
    },
    {
      "epoch": 0.08707428110413185,
      "grad_norm": 2.062133550643921,
      "learning_rate": 9.914885671207713e-06,
      "loss": 0.9883,
      "step": 3022
    },
    {
      "epoch": 0.08710309456578114,
      "grad_norm": 2.324540376663208,
      "learning_rate": 9.914799920491828e-06,
      "loss": 0.9538,
      "step": 3023
    },
    {
      "epoch": 0.08713190802743041,
      "grad_norm": 2.1084485054016113,
      "learning_rate": 9.914714126972951e-06,
      "loss": 0.9591,
      "step": 3024
    },
    {
      "epoch": 0.0871607214890797,
      "grad_norm": 1.975724220275879,
      "learning_rate": 9.914628290651824e-06,
      "loss": 0.9711,
      "step": 3025
    },
    {
      "epoch": 0.08718953495072898,
      "grad_norm": 2.297934055328369,
      "learning_rate": 9.9145424115292e-06,
      "loss": 0.9622,
      "step": 3026
    },
    {
      "epoch": 0.08721834841237826,
      "grad_norm": 2.090470552444458,
      "learning_rate": 9.914456489605822e-06,
      "loss": 0.9361,
      "step": 3027
    },
    {
      "epoch": 0.08724716187402755,
      "grad_norm": 2.172713279724121,
      "learning_rate": 9.914370524882442e-06,
      "loss": 0.9715,
      "step": 3028
    },
    {
      "epoch": 0.08727597533567683,
      "grad_norm": 1.970619797706604,
      "learning_rate": 9.914284517359807e-06,
      "loss": 0.9818,
      "step": 3029
    },
    {
      "epoch": 0.08730478879732612,
      "grad_norm": 2.063762664794922,
      "learning_rate": 9.914198467038667e-06,
      "loss": 0.9459,
      "step": 3030
    },
    {
      "epoch": 0.08733360225897539,
      "grad_norm": 2.062643051147461,
      "learning_rate": 9.91411237391977e-06,
      "loss": 0.9458,
      "step": 3031
    },
    {
      "epoch": 0.08736241572062467,
      "grad_norm": 2.0670738220214844,
      "learning_rate": 9.914026238003866e-06,
      "loss": 0.9568,
      "step": 3032
    },
    {
      "epoch": 0.08739122918227396,
      "grad_norm": 2.2724549770355225,
      "learning_rate": 9.913940059291706e-06,
      "loss": 0.999,
      "step": 3033
    },
    {
      "epoch": 0.08742004264392324,
      "grad_norm": 2.2126262187957764,
      "learning_rate": 9.913853837784042e-06,
      "loss": 0.9664,
      "step": 3034
    },
    {
      "epoch": 0.08744885610557253,
      "grad_norm": 2.0224030017852783,
      "learning_rate": 9.91376757348162e-06,
      "loss": 0.9512,
      "step": 3035
    },
    {
      "epoch": 0.0874776695672218,
      "grad_norm": 2.1190383434295654,
      "learning_rate": 9.913681266385198e-06,
      "loss": 0.9288,
      "step": 3036
    },
    {
      "epoch": 0.08750648302887108,
      "grad_norm": 2.135333776473999,
      "learning_rate": 9.913594916495522e-06,
      "loss": 0.9832,
      "step": 3037
    },
    {
      "epoch": 0.08753529649052037,
      "grad_norm": 1.981179118156433,
      "learning_rate": 9.913508523813346e-06,
      "loss": 0.9641,
      "step": 3038
    },
    {
      "epoch": 0.08756410995216965,
      "grad_norm": 2.367880344390869,
      "learning_rate": 9.913422088339422e-06,
      "loss": 0.9868,
      "step": 3039
    },
    {
      "epoch": 0.08759292341381894,
      "grad_norm": 2.0660400390625,
      "learning_rate": 9.913335610074504e-06,
      "loss": 0.9639,
      "step": 3040
    },
    {
      "epoch": 0.08762173687546822,
      "grad_norm": 2.235694408416748,
      "learning_rate": 9.913249089019345e-06,
      "loss": 0.9572,
      "step": 3041
    },
    {
      "epoch": 0.08765055033711751,
      "grad_norm": 2.0166940689086914,
      "learning_rate": 9.913162525174696e-06,
      "loss": 0.9863,
      "step": 3042
    },
    {
      "epoch": 0.08767936379876679,
      "grad_norm": 2.075082778930664,
      "learning_rate": 9.913075918541315e-06,
      "loss": 0.9467,
      "step": 3043
    },
    {
      "epoch": 0.08770817726041606,
      "grad_norm": 1.948862910270691,
      "learning_rate": 9.912989269119953e-06,
      "loss": 0.9648,
      "step": 3044
    },
    {
      "epoch": 0.08773699072206535,
      "grad_norm": 1.9984921216964722,
      "learning_rate": 9.912902576911366e-06,
      "loss": 0.9442,
      "step": 3045
    },
    {
      "epoch": 0.08776580418371463,
      "grad_norm": 2.0707693099975586,
      "learning_rate": 9.912815841916308e-06,
      "loss": 0.9238,
      "step": 3046
    },
    {
      "epoch": 0.08779461764536392,
      "grad_norm": 1.9932421445846558,
      "learning_rate": 9.912729064135534e-06,
      "loss": 0.9546,
      "step": 3047
    },
    {
      "epoch": 0.0878234311070132,
      "grad_norm": 2.2063493728637695,
      "learning_rate": 9.912642243569803e-06,
      "loss": 0.9644,
      "step": 3048
    },
    {
      "epoch": 0.08785224456866247,
      "grad_norm": 2.1390492916107178,
      "learning_rate": 9.912555380219869e-06,
      "loss": 0.9679,
      "step": 3049
    },
    {
      "epoch": 0.08788105803031176,
      "grad_norm": 1.9956403970718384,
      "learning_rate": 9.912468474086486e-06,
      "loss": 0.928,
      "step": 3050
    },
    {
      "epoch": 0.08790987149196104,
      "grad_norm": 1.989311933517456,
      "learning_rate": 9.912381525170415e-06,
      "loss": 0.9635,
      "step": 3051
    },
    {
      "epoch": 0.08793868495361033,
      "grad_norm": 2.051248550415039,
      "learning_rate": 9.912294533472411e-06,
      "loss": 0.955,
      "step": 3052
    },
    {
      "epoch": 0.08796749841525961,
      "grad_norm": 2.0062003135681152,
      "learning_rate": 9.912207498993232e-06,
      "loss": 0.9405,
      "step": 3053
    },
    {
      "epoch": 0.08799631187690889,
      "grad_norm": 1.8630883693695068,
      "learning_rate": 9.912120421733638e-06,
      "loss": 0.9403,
      "step": 3054
    },
    {
      "epoch": 0.08802512533855818,
      "grad_norm": 2.1252355575561523,
      "learning_rate": 9.912033301694382e-06,
      "loss": 0.9677,
      "step": 3055
    },
    {
      "epoch": 0.08805393880020745,
      "grad_norm": 2.0890440940856934,
      "learning_rate": 9.911946138876229e-06,
      "loss": 0.962,
      "step": 3056
    },
    {
      "epoch": 0.08808275226185674,
      "grad_norm": 2.0525364875793457,
      "learning_rate": 9.911858933279935e-06,
      "loss": 0.9547,
      "step": 3057
    },
    {
      "epoch": 0.08811156572350602,
      "grad_norm": 2.0523383617401123,
      "learning_rate": 9.911771684906257e-06,
      "loss": 0.9793,
      "step": 3058
    },
    {
      "epoch": 0.0881403791851553,
      "grad_norm": 2.13288950920105,
      "learning_rate": 9.91168439375596e-06,
      "loss": 0.9583,
      "step": 3059
    },
    {
      "epoch": 0.08816919264680459,
      "grad_norm": 2.0675833225250244,
      "learning_rate": 9.911597059829803e-06,
      "loss": 0.9586,
      "step": 3060
    },
    {
      "epoch": 0.08819800610845387,
      "grad_norm": 2.145643949508667,
      "learning_rate": 9.911509683128543e-06,
      "loss": 0.9693,
      "step": 3061
    },
    {
      "epoch": 0.08822681957010316,
      "grad_norm": 1.973827838897705,
      "learning_rate": 9.911422263652943e-06,
      "loss": 0.9739,
      "step": 3062
    },
    {
      "epoch": 0.08825563303175243,
      "grad_norm": 1.9910190105438232,
      "learning_rate": 9.911334801403768e-06,
      "loss": 0.9369,
      "step": 3063
    },
    {
      "epoch": 0.08828444649340171,
      "grad_norm": 1.9133224487304688,
      "learning_rate": 9.911247296381774e-06,
      "loss": 0.9254,
      "step": 3064
    },
    {
      "epoch": 0.088313259955051,
      "grad_norm": 2.0360660552978516,
      "learning_rate": 9.911159748587726e-06,
      "loss": 0.9342,
      "step": 3065
    },
    {
      "epoch": 0.08834207341670028,
      "grad_norm": 1.961956262588501,
      "learning_rate": 9.911072158022386e-06,
      "loss": 0.9671,
      "step": 3066
    },
    {
      "epoch": 0.08837088687834957,
      "grad_norm": 2.0310580730438232,
      "learning_rate": 9.910984524686516e-06,
      "loss": 0.9556,
      "step": 3067
    },
    {
      "epoch": 0.08839970033999885,
      "grad_norm": 2.068843126296997,
      "learning_rate": 9.91089684858088e-06,
      "loss": 0.991,
      "step": 3068
    },
    {
      "epoch": 0.08842851380164812,
      "grad_norm": 1.9960427284240723,
      "learning_rate": 9.910809129706243e-06,
      "loss": 0.9919,
      "step": 3069
    },
    {
      "epoch": 0.08845732726329741,
      "grad_norm": 2.022101402282715,
      "learning_rate": 9.910721368063367e-06,
      "loss": 0.9593,
      "step": 3070
    },
    {
      "epoch": 0.08848614072494669,
      "grad_norm": 1.9504450559616089,
      "learning_rate": 9.910633563653016e-06,
      "loss": 0.9663,
      "step": 3071
    },
    {
      "epoch": 0.08851495418659598,
      "grad_norm": 2.0221056938171387,
      "learning_rate": 9.910545716475955e-06,
      "loss": 0.9545,
      "step": 3072
    },
    {
      "epoch": 0.08854376764824526,
      "grad_norm": 1.9357385635375977,
      "learning_rate": 9.910457826532951e-06,
      "loss": 0.956,
      "step": 3073
    },
    {
      "epoch": 0.08857258110989455,
      "grad_norm": 1.9318177700042725,
      "learning_rate": 9.910369893824768e-06,
      "loss": 0.9478,
      "step": 3074
    },
    {
      "epoch": 0.08860139457154383,
      "grad_norm": 1.9734296798706055,
      "learning_rate": 9.91028191835217e-06,
      "loss": 0.9865,
      "step": 3075
    },
    {
      "epoch": 0.0886302080331931,
      "grad_norm": 1.987586259841919,
      "learning_rate": 9.910193900115927e-06,
      "loss": 0.9398,
      "step": 3076
    },
    {
      "epoch": 0.0886590214948424,
      "grad_norm": 1.9844826459884644,
      "learning_rate": 9.910105839116803e-06,
      "loss": 0.9358,
      "step": 3077
    },
    {
      "epoch": 0.08868783495649167,
      "grad_norm": 1.9953378438949585,
      "learning_rate": 9.910017735355565e-06,
      "loss": 0.9815,
      "step": 3078
    },
    {
      "epoch": 0.08871664841814096,
      "grad_norm": 2.0922744274139404,
      "learning_rate": 9.90992958883298e-06,
      "loss": 0.9545,
      "step": 3079
    },
    {
      "epoch": 0.08874546187979024,
      "grad_norm": 2.077867269515991,
      "learning_rate": 9.909841399549817e-06,
      "loss": 0.9604,
      "step": 3080
    },
    {
      "epoch": 0.08877427534143952,
      "grad_norm": 2.1341779232025146,
      "learning_rate": 9.909753167506845e-06,
      "loss": 0.9543,
      "step": 3081
    },
    {
      "epoch": 0.0888030888030888,
      "grad_norm": 1.9568259716033936,
      "learning_rate": 9.90966489270483e-06,
      "loss": 0.9654,
      "step": 3082
    },
    {
      "epoch": 0.08883190226473808,
      "grad_norm": 1.9985028505325317,
      "learning_rate": 9.909576575144541e-06,
      "loss": 0.9351,
      "step": 3083
    },
    {
      "epoch": 0.08886071572638737,
      "grad_norm": 1.9707353115081787,
      "learning_rate": 9.909488214826749e-06,
      "loss": 0.9376,
      "step": 3084
    },
    {
      "epoch": 0.08888952918803665,
      "grad_norm": 2.219151735305786,
      "learning_rate": 9.90939981175222e-06,
      "loss": 0.9459,
      "step": 3085
    },
    {
      "epoch": 0.08891834264968593,
      "grad_norm": 2.126739740371704,
      "learning_rate": 9.909311365921728e-06,
      "loss": 0.9437,
      "step": 3086
    },
    {
      "epoch": 0.08894715611133522,
      "grad_norm": 2.1086883544921875,
      "learning_rate": 9.909222877336042e-06,
      "loss": 0.9528,
      "step": 3087
    },
    {
      "epoch": 0.0889759695729845,
      "grad_norm": 2.2851855754852295,
      "learning_rate": 9.909134345995931e-06,
      "loss": 0.9353,
      "step": 3088
    },
    {
      "epoch": 0.08900478303463379,
      "grad_norm": 1.9948441982269287,
      "learning_rate": 9.909045771902168e-06,
      "loss": 0.977,
      "step": 3089
    },
    {
      "epoch": 0.08903359649628306,
      "grad_norm": 2.0544629096984863,
      "learning_rate": 9.908957155055522e-06,
      "loss": 0.9408,
      "step": 3090
    },
    {
      "epoch": 0.08906240995793234,
      "grad_norm": 1.9483261108398438,
      "learning_rate": 9.90886849545677e-06,
      "loss": 0.9364,
      "step": 3091
    },
    {
      "epoch": 0.08909122341958163,
      "grad_norm": 2.1162607669830322,
      "learning_rate": 9.908779793106676e-06,
      "loss": 0.9355,
      "step": 3092
    },
    {
      "epoch": 0.08912003688123091,
      "grad_norm": 1.9167580604553223,
      "learning_rate": 9.90869104800602e-06,
      "loss": 0.9317,
      "step": 3093
    },
    {
      "epoch": 0.0891488503428802,
      "grad_norm": 2.16056227684021,
      "learning_rate": 9.90860226015557e-06,
      "loss": 0.9563,
      "step": 3094
    },
    {
      "epoch": 0.08917766380452948,
      "grad_norm": 1.9933465719223022,
      "learning_rate": 9.908513429556102e-06,
      "loss": 0.9561,
      "step": 3095
    },
    {
      "epoch": 0.08920647726617875,
      "grad_norm": 2.1252052783966064,
      "learning_rate": 9.908424556208388e-06,
      "loss": 0.9653,
      "step": 3096
    },
    {
      "epoch": 0.08923529072782804,
      "grad_norm": 2.0002493858337402,
      "learning_rate": 9.908335640113205e-06,
      "loss": 0.9568,
      "step": 3097
    },
    {
      "epoch": 0.08926410418947732,
      "grad_norm": 1.9334245920181274,
      "learning_rate": 9.908246681271322e-06,
      "loss": 0.9757,
      "step": 3098
    },
    {
      "epoch": 0.08929291765112661,
      "grad_norm": 2.0294454097747803,
      "learning_rate": 9.90815767968352e-06,
      "loss": 0.9514,
      "step": 3099
    },
    {
      "epoch": 0.08932173111277589,
      "grad_norm": 2.0093469619750977,
      "learning_rate": 9.908068635350569e-06,
      "loss": 0.9532,
      "step": 3100
    },
    {
      "epoch": 0.08935054457442518,
      "grad_norm": 2.0003268718719482,
      "learning_rate": 9.907979548273248e-06,
      "loss": 0.9682,
      "step": 3101
    },
    {
      "epoch": 0.08937935803607445,
      "grad_norm": 1.9096208810806274,
      "learning_rate": 9.90789041845233e-06,
      "loss": 0.938,
      "step": 3102
    },
    {
      "epoch": 0.08940817149772373,
      "grad_norm": 2.0430448055267334,
      "learning_rate": 9.907801245888592e-06,
      "loss": 0.9458,
      "step": 3103
    },
    {
      "epoch": 0.08943698495937302,
      "grad_norm": 1.911338448524475,
      "learning_rate": 9.907712030582812e-06,
      "loss": 0.9236,
      "step": 3104
    },
    {
      "epoch": 0.0894657984210223,
      "grad_norm": 2.0509846210479736,
      "learning_rate": 9.907622772535766e-06,
      "loss": 0.9846,
      "step": 3105
    },
    {
      "epoch": 0.08949461188267159,
      "grad_norm": 2.205904483795166,
      "learning_rate": 9.907533471748232e-06,
      "loss": 0.9441,
      "step": 3106
    },
    {
      "epoch": 0.08952342534432087,
      "grad_norm": 1.9904167652130127,
      "learning_rate": 9.907444128220986e-06,
      "loss": 0.9666,
      "step": 3107
    },
    {
      "epoch": 0.08955223880597014,
      "grad_norm": 1.9470664262771606,
      "learning_rate": 9.907354741954809e-06,
      "loss": 0.9511,
      "step": 3108
    },
    {
      "epoch": 0.08958105226761943,
      "grad_norm": 1.9798859357833862,
      "learning_rate": 9.907265312950476e-06,
      "loss": 0.9405,
      "step": 3109
    },
    {
      "epoch": 0.08960986572926871,
      "grad_norm": 1.876659870147705,
      "learning_rate": 9.907175841208768e-06,
      "loss": 0.9063,
      "step": 3110
    },
    {
      "epoch": 0.089638679190918,
      "grad_norm": 1.9621331691741943,
      "learning_rate": 9.907086326730464e-06,
      "loss": 0.9645,
      "step": 3111
    },
    {
      "epoch": 0.08966749265256728,
      "grad_norm": 2.0185024738311768,
      "learning_rate": 9.906996769516343e-06,
      "loss": 0.9768,
      "step": 3112
    },
    {
      "epoch": 0.08969630611421656,
      "grad_norm": 1.8413786888122559,
      "learning_rate": 9.906907169567186e-06,
      "loss": 0.9837,
      "step": 3113
    },
    {
      "epoch": 0.08972511957586585,
      "grad_norm": 2.0174686908721924,
      "learning_rate": 9.906817526883775e-06,
      "loss": 0.945,
      "step": 3114
    },
    {
      "epoch": 0.08975393303751512,
      "grad_norm": 2.0379974842071533,
      "learning_rate": 9.906727841466884e-06,
      "loss": 0.9257,
      "step": 3115
    },
    {
      "epoch": 0.08978274649916441,
      "grad_norm": 1.925796627998352,
      "learning_rate": 9.906638113317301e-06,
      "loss": 0.9337,
      "step": 3116
    },
    {
      "epoch": 0.08981155996081369,
      "grad_norm": 2.012847423553467,
      "learning_rate": 9.906548342435804e-06,
      "loss": 0.952,
      "step": 3117
    },
    {
      "epoch": 0.08984037342246297,
      "grad_norm": 2.220292329788208,
      "learning_rate": 9.906458528823175e-06,
      "loss": 0.9587,
      "step": 3118
    },
    {
      "epoch": 0.08986918688411226,
      "grad_norm": 2.0126843452453613,
      "learning_rate": 9.906368672480197e-06,
      "loss": 0.9714,
      "step": 3119
    },
    {
      "epoch": 0.08989800034576154,
      "grad_norm": 2.218475580215454,
      "learning_rate": 9.906278773407654e-06,
      "loss": 0.9657,
      "step": 3120
    },
    {
      "epoch": 0.08992681380741083,
      "grad_norm": 1.9537837505340576,
      "learning_rate": 9.906188831606326e-06,
      "loss": 0.981,
      "step": 3121
    },
    {
      "epoch": 0.0899556272690601,
      "grad_norm": 2.2560527324676514,
      "learning_rate": 9.906098847077e-06,
      "loss": 0.9663,
      "step": 3122
    },
    {
      "epoch": 0.08998444073070938,
      "grad_norm": 2.205634117126465,
      "learning_rate": 9.906008819820455e-06,
      "loss": 0.956,
      "step": 3123
    },
    {
      "epoch": 0.09001325419235867,
      "grad_norm": 1.8758374452590942,
      "learning_rate": 9.905918749837477e-06,
      "loss": 0.9197,
      "step": 3124
    },
    {
      "epoch": 0.09004206765400795,
      "grad_norm": 2.122739553451538,
      "learning_rate": 9.905828637128853e-06,
      "loss": 0.9554,
      "step": 3125
    },
    {
      "epoch": 0.09007088111565724,
      "grad_norm": 2.0056872367858887,
      "learning_rate": 9.905738481695363e-06,
      "loss": 0.9413,
      "step": 3126
    },
    {
      "epoch": 0.09009969457730652,
      "grad_norm": 2.0360257625579834,
      "learning_rate": 9.905648283537798e-06,
      "loss": 0.92,
      "step": 3127
    },
    {
      "epoch": 0.0901285080389558,
      "grad_norm": 2.03783917427063,
      "learning_rate": 9.905558042656937e-06,
      "loss": 0.962,
      "step": 3128
    },
    {
      "epoch": 0.09015732150060508,
      "grad_norm": 2.0147106647491455,
      "learning_rate": 9.905467759053571e-06,
      "loss": 0.9475,
      "step": 3129
    },
    {
      "epoch": 0.09018613496225436,
      "grad_norm": 2.0511386394500732,
      "learning_rate": 9.905377432728484e-06,
      "loss": 0.9875,
      "step": 3130
    },
    {
      "epoch": 0.09021494842390365,
      "grad_norm": 2.0411887168884277,
      "learning_rate": 9.905287063682465e-06,
      "loss": 0.9558,
      "step": 3131
    },
    {
      "epoch": 0.09024376188555293,
      "grad_norm": 1.83855402469635,
      "learning_rate": 9.905196651916296e-06,
      "loss": 0.9414,
      "step": 3132
    },
    {
      "epoch": 0.09027257534720222,
      "grad_norm": 2.033005475997925,
      "learning_rate": 9.90510619743077e-06,
      "loss": 0.9404,
      "step": 3133
    },
    {
      "epoch": 0.0903013888088515,
      "grad_norm": 2.10947847366333,
      "learning_rate": 9.905015700226671e-06,
      "loss": 0.9315,
      "step": 3134
    },
    {
      "epoch": 0.09033020227050077,
      "grad_norm": 1.9720362424850464,
      "learning_rate": 9.90492516030479e-06,
      "loss": 0.9752,
      "step": 3135
    },
    {
      "epoch": 0.09035901573215006,
      "grad_norm": 1.9946376085281372,
      "learning_rate": 9.904834577665911e-06,
      "loss": 0.9369,
      "step": 3136
    },
    {
      "epoch": 0.09038782919379934,
      "grad_norm": 2.025519847869873,
      "learning_rate": 9.90474395231083e-06,
      "loss": 0.9581,
      "step": 3137
    },
    {
      "epoch": 0.09041664265544863,
      "grad_norm": 2.0132675170898438,
      "learning_rate": 9.904653284240328e-06,
      "loss": 0.9727,
      "step": 3138
    },
    {
      "epoch": 0.09044545611709791,
      "grad_norm": 1.9500043392181396,
      "learning_rate": 9.904562573455201e-06,
      "loss": 0.9424,
      "step": 3139
    },
    {
      "epoch": 0.09047426957874719,
      "grad_norm": 1.9553344249725342,
      "learning_rate": 9.904471819956237e-06,
      "loss": 0.9585,
      "step": 3140
    },
    {
      "epoch": 0.09050308304039648,
      "grad_norm": 1.9720075130462646,
      "learning_rate": 9.904381023744225e-06,
      "loss": 0.9617,
      "step": 3141
    },
    {
      "epoch": 0.09053189650204575,
      "grad_norm": 1.8783977031707764,
      "learning_rate": 9.904290184819957e-06,
      "loss": 0.9536,
      "step": 3142
    },
    {
      "epoch": 0.09056070996369504,
      "grad_norm": 2.027763605117798,
      "learning_rate": 9.904199303184225e-06,
      "loss": 0.9526,
      "step": 3143
    },
    {
      "epoch": 0.09058952342534432,
      "grad_norm": 2.079155683517456,
      "learning_rate": 9.904108378837818e-06,
      "loss": 0.9521,
      "step": 3144
    },
    {
      "epoch": 0.0906183368869936,
      "grad_norm": 2.051393508911133,
      "learning_rate": 9.90401741178153e-06,
      "loss": 0.9639,
      "step": 3145
    },
    {
      "epoch": 0.09064715034864289,
      "grad_norm": 2.238332748413086,
      "learning_rate": 9.903926402016153e-06,
      "loss": 0.9621,
      "step": 3146
    },
    {
      "epoch": 0.09067596381029217,
      "grad_norm": 2.0128610134124756,
      "learning_rate": 9.903835349542478e-06,
      "loss": 0.9426,
      "step": 3147
    },
    {
      "epoch": 0.09070477727194146,
      "grad_norm": 2.0484182834625244,
      "learning_rate": 9.9037442543613e-06,
      "loss": 0.9756,
      "step": 3148
    },
    {
      "epoch": 0.09073359073359073,
      "grad_norm": 2.066399574279785,
      "learning_rate": 9.903653116473413e-06,
      "loss": 0.9723,
      "step": 3149
    },
    {
      "epoch": 0.09076240419524001,
      "grad_norm": 2.029053211212158,
      "learning_rate": 9.903561935879606e-06,
      "loss": 0.9142,
      "step": 3150
    },
    {
      "epoch": 0.0907912176568893,
      "grad_norm": 2.165544271469116,
      "learning_rate": 9.903470712580679e-06,
      "loss": 0.9606,
      "step": 3151
    },
    {
      "epoch": 0.09082003111853858,
      "grad_norm": 2.1121532917022705,
      "learning_rate": 9.903379446577423e-06,
      "loss": 0.9141,
      "step": 3152
    },
    {
      "epoch": 0.09084884458018787,
      "grad_norm": 2.258327007293701,
      "learning_rate": 9.903288137870634e-06,
      "loss": 0.9627,
      "step": 3153
    },
    {
      "epoch": 0.09087765804183714,
      "grad_norm": 2.0472395420074463,
      "learning_rate": 9.903196786461106e-06,
      "loss": 0.9416,
      "step": 3154
    },
    {
      "epoch": 0.09090647150348644,
      "grad_norm": 2.129962682723999,
      "learning_rate": 9.903105392349634e-06,
      "loss": 0.9674,
      "step": 3155
    },
    {
      "epoch": 0.09093528496513571,
      "grad_norm": 2.2145073413848877,
      "learning_rate": 9.903013955537018e-06,
      "loss": 0.93,
      "step": 3156
    },
    {
      "epoch": 0.09096409842678499,
      "grad_norm": 2.0945472717285156,
      "learning_rate": 9.902922476024051e-06,
      "loss": 0.9276,
      "step": 3157
    },
    {
      "epoch": 0.09099291188843428,
      "grad_norm": 2.041144609451294,
      "learning_rate": 9.90283095381153e-06,
      "loss": 0.9666,
      "step": 3158
    },
    {
      "epoch": 0.09102172535008356,
      "grad_norm": 2.2832674980163574,
      "learning_rate": 9.902739388900252e-06,
      "loss": 0.9617,
      "step": 3159
    },
    {
      "epoch": 0.09105053881173285,
      "grad_norm": 2.081639528274536,
      "learning_rate": 9.902647781291015e-06,
      "loss": 0.9324,
      "step": 3160
    },
    {
      "epoch": 0.09107935227338212,
      "grad_norm": 2.225148916244507,
      "learning_rate": 9.902556130984618e-06,
      "loss": 0.9401,
      "step": 3161
    },
    {
      "epoch": 0.0911081657350314,
      "grad_norm": 2.100865125656128,
      "learning_rate": 9.902464437981855e-06,
      "loss": 0.968,
      "step": 3162
    },
    {
      "epoch": 0.09113697919668069,
      "grad_norm": 1.9050719738006592,
      "learning_rate": 9.902372702283531e-06,
      "loss": 0.9337,
      "step": 3163
    },
    {
      "epoch": 0.09116579265832997,
      "grad_norm": 2.0873022079467773,
      "learning_rate": 9.902280923890439e-06,
      "loss": 0.9889,
      "step": 3164
    },
    {
      "epoch": 0.09119460611997926,
      "grad_norm": 2.1199820041656494,
      "learning_rate": 9.90218910280338e-06,
      "loss": 0.9622,
      "step": 3165
    },
    {
      "epoch": 0.09122341958162854,
      "grad_norm": 2.0343520641326904,
      "learning_rate": 9.902097239023156e-06,
      "loss": 0.9622,
      "step": 3166
    },
    {
      "epoch": 0.09125223304327781,
      "grad_norm": 2.10593318939209,
      "learning_rate": 9.902005332550566e-06,
      "loss": 0.9623,
      "step": 3167
    },
    {
      "epoch": 0.0912810465049271,
      "grad_norm": 1.9231290817260742,
      "learning_rate": 9.901913383386407e-06,
      "loss": 0.9404,
      "step": 3168
    },
    {
      "epoch": 0.09130985996657638,
      "grad_norm": 2.069669723510742,
      "learning_rate": 9.901821391531486e-06,
      "loss": 0.9447,
      "step": 3169
    },
    {
      "epoch": 0.09133867342822567,
      "grad_norm": 1.989315390586853,
      "learning_rate": 9.901729356986597e-06,
      "loss": 0.9449,
      "step": 3170
    },
    {
      "epoch": 0.09136748688987495,
      "grad_norm": 2.1216981410980225,
      "learning_rate": 9.90163727975255e-06,
      "loss": 0.962,
      "step": 3171
    },
    {
      "epoch": 0.09139630035152423,
      "grad_norm": 2.1171317100524902,
      "learning_rate": 9.901545159830138e-06,
      "loss": 0.9711,
      "step": 3172
    },
    {
      "epoch": 0.09142511381317352,
      "grad_norm": 1.8638598918914795,
      "learning_rate": 9.90145299722017e-06,
      "loss": 0.9324,
      "step": 3173
    },
    {
      "epoch": 0.0914539272748228,
      "grad_norm": 2.040616750717163,
      "learning_rate": 9.901360791923444e-06,
      "loss": 0.9472,
      "step": 3174
    },
    {
      "epoch": 0.09148274073647208,
      "grad_norm": 1.9437251091003418,
      "learning_rate": 9.901268543940766e-06,
      "loss": 0.9298,
      "step": 3175
    },
    {
      "epoch": 0.09151155419812136,
      "grad_norm": 1.9647070169448853,
      "learning_rate": 9.901176253272936e-06,
      "loss": 0.967,
      "step": 3176
    },
    {
      "epoch": 0.09154036765977064,
      "grad_norm": 1.9574944972991943,
      "learning_rate": 9.901083919920764e-06,
      "loss": 0.9546,
      "step": 3177
    },
    {
      "epoch": 0.09156918112141993,
      "grad_norm": 1.8806889057159424,
      "learning_rate": 9.900991543885049e-06,
      "loss": 0.9462,
      "step": 3178
    },
    {
      "epoch": 0.0915979945830692,
      "grad_norm": 1.9947988986968994,
      "learning_rate": 9.900899125166597e-06,
      "loss": 0.9406,
      "step": 3179
    },
    {
      "epoch": 0.0916268080447185,
      "grad_norm": 1.8767669200897217,
      "learning_rate": 9.900806663766212e-06,
      "loss": 0.9446,
      "step": 3180
    },
    {
      "epoch": 0.09165562150636777,
      "grad_norm": 2.0454583168029785,
      "learning_rate": 9.900714159684701e-06,
      "loss": 0.9478,
      "step": 3181
    },
    {
      "epoch": 0.09168443496801706,
      "grad_norm": 2.1689391136169434,
      "learning_rate": 9.900621612922867e-06,
      "loss": 0.9653,
      "step": 3182
    },
    {
      "epoch": 0.09171324842966634,
      "grad_norm": 2.1004488468170166,
      "learning_rate": 9.900529023481519e-06,
      "loss": 0.9269,
      "step": 3183
    },
    {
      "epoch": 0.09174206189131562,
      "grad_norm": 2.164881706237793,
      "learning_rate": 9.90043639136146e-06,
      "loss": 0.922,
      "step": 3184
    },
    {
      "epoch": 0.09177087535296491,
      "grad_norm": 2.1136858463287354,
      "learning_rate": 9.9003437165635e-06,
      "loss": 0.9564,
      "step": 3185
    },
    {
      "epoch": 0.09179968881461419,
      "grad_norm": 2.1293623447418213,
      "learning_rate": 9.900250999088447e-06,
      "loss": 0.9603,
      "step": 3186
    },
    {
      "epoch": 0.09182850227626348,
      "grad_norm": 1.9663716554641724,
      "learning_rate": 9.900158238937104e-06,
      "loss": 0.9709,
      "step": 3187
    },
    {
      "epoch": 0.09185731573791275,
      "grad_norm": 1.9852104187011719,
      "learning_rate": 9.900065436110281e-06,
      "loss": 0.9907,
      "step": 3188
    },
    {
      "epoch": 0.09188612919956203,
      "grad_norm": 1.9333959817886353,
      "learning_rate": 9.899972590608786e-06,
      "loss": 0.9391,
      "step": 3189
    },
    {
      "epoch": 0.09191494266121132,
      "grad_norm": 1.949794054031372,
      "learning_rate": 9.89987970243343e-06,
      "loss": 0.9289,
      "step": 3190
    },
    {
      "epoch": 0.0919437561228606,
      "grad_norm": 1.8315153121948242,
      "learning_rate": 9.899786771585019e-06,
      "loss": 0.9517,
      "step": 3191
    },
    {
      "epoch": 0.09197256958450989,
      "grad_norm": 1.9260867834091187,
      "learning_rate": 9.899693798064362e-06,
      "loss": 0.945,
      "step": 3192
    },
    {
      "epoch": 0.09200138304615917,
      "grad_norm": 2.0449090003967285,
      "learning_rate": 9.899600781872269e-06,
      "loss": 0.939,
      "step": 3193
    },
    {
      "epoch": 0.09203019650780844,
      "grad_norm": 1.9276026487350464,
      "learning_rate": 9.899507723009555e-06,
      "loss": 0.9418,
      "step": 3194
    },
    {
      "epoch": 0.09205900996945773,
      "grad_norm": 1.99655020236969,
      "learning_rate": 9.899414621477022e-06,
      "loss": 0.938,
      "step": 3195
    },
    {
      "epoch": 0.09208782343110701,
      "grad_norm": 1.833849310874939,
      "learning_rate": 9.899321477275487e-06,
      "loss": 0.9349,
      "step": 3196
    },
    {
      "epoch": 0.0921166368927563,
      "grad_norm": 2.0683681964874268,
      "learning_rate": 9.899228290405761e-06,
      "loss": 0.9857,
      "step": 3197
    },
    {
      "epoch": 0.09214545035440558,
      "grad_norm": 1.9474939107894897,
      "learning_rate": 9.899135060868652e-06,
      "loss": 0.9401,
      "step": 3198
    },
    {
      "epoch": 0.09217426381605485,
      "grad_norm": 2.1995067596435547,
      "learning_rate": 9.899041788664974e-06,
      "loss": 0.9573,
      "step": 3199
    },
    {
      "epoch": 0.09220307727770415,
      "grad_norm": 1.9182450771331787,
      "learning_rate": 9.89894847379554e-06,
      "loss": 0.9531,
      "step": 3200
    },
    {
      "epoch": 0.09223189073935342,
      "grad_norm": 1.760501742362976,
      "learning_rate": 9.898855116261164e-06,
      "loss": 0.9481,
      "step": 3201
    },
    {
      "epoch": 0.09226070420100271,
      "grad_norm": 1.940032958984375,
      "learning_rate": 9.898761716062654e-06,
      "loss": 0.9286,
      "step": 3202
    },
    {
      "epoch": 0.09228951766265199,
      "grad_norm": 2.0962445735931396,
      "learning_rate": 9.898668273200826e-06,
      "loss": 0.9461,
      "step": 3203
    },
    {
      "epoch": 0.09231833112430127,
      "grad_norm": 1.7665210962295532,
      "learning_rate": 9.898574787676498e-06,
      "loss": 0.9734,
      "step": 3204
    },
    {
      "epoch": 0.09234714458595056,
      "grad_norm": 2.0952584743499756,
      "learning_rate": 9.898481259490476e-06,
      "loss": 0.9726,
      "step": 3205
    },
    {
      "epoch": 0.09237595804759983,
      "grad_norm": 2.0252301692962646,
      "learning_rate": 9.898387688643581e-06,
      "loss": 0.9538,
      "step": 3206
    },
    {
      "epoch": 0.09240477150924913,
      "grad_norm": 2.076385259628296,
      "learning_rate": 9.898294075136627e-06,
      "loss": 0.964,
      "step": 3207
    },
    {
      "epoch": 0.0924335849708984,
      "grad_norm": 1.9762390851974487,
      "learning_rate": 9.898200418970427e-06,
      "loss": 0.9394,
      "step": 3208
    },
    {
      "epoch": 0.09246239843254768,
      "grad_norm": 1.980067491531372,
      "learning_rate": 9.898106720145796e-06,
      "loss": 0.9529,
      "step": 3209
    },
    {
      "epoch": 0.09249121189419697,
      "grad_norm": 2.0616211891174316,
      "learning_rate": 9.898012978663553e-06,
      "loss": 0.9265,
      "step": 3210
    },
    {
      "epoch": 0.09252002535584625,
      "grad_norm": 2.1832354068756104,
      "learning_rate": 9.897919194524514e-06,
      "loss": 0.9828,
      "step": 3211
    },
    {
      "epoch": 0.09254883881749554,
      "grad_norm": 1.9207563400268555,
      "learning_rate": 9.897825367729495e-06,
      "loss": 0.9589,
      "step": 3212
    },
    {
      "epoch": 0.09257765227914481,
      "grad_norm": 2.1089367866516113,
      "learning_rate": 9.897731498279313e-06,
      "loss": 0.9917,
      "step": 3213
    },
    {
      "epoch": 0.0926064657407941,
      "grad_norm": 1.9431302547454834,
      "learning_rate": 9.897637586174785e-06,
      "loss": 0.9321,
      "step": 3214
    },
    {
      "epoch": 0.09263527920244338,
      "grad_norm": 2.1026411056518555,
      "learning_rate": 9.89754363141673e-06,
      "loss": 0.9567,
      "step": 3215
    },
    {
      "epoch": 0.09266409266409266,
      "grad_norm": 1.96564781665802,
      "learning_rate": 9.897449634005964e-06,
      "loss": 0.9559,
      "step": 3216
    },
    {
      "epoch": 0.09269290612574195,
      "grad_norm": 2.12361478805542,
      "learning_rate": 9.897355593943309e-06,
      "loss": 0.922,
      "step": 3217
    },
    {
      "epoch": 0.09272171958739123,
      "grad_norm": 1.9577974081039429,
      "learning_rate": 9.897261511229584e-06,
      "loss": 0.946,
      "step": 3218
    },
    {
      "epoch": 0.09275053304904052,
      "grad_norm": 2.002955198287964,
      "learning_rate": 9.897167385865602e-06,
      "loss": 0.9344,
      "step": 3219
    },
    {
      "epoch": 0.0927793465106898,
      "grad_norm": 1.9101380109786987,
      "learning_rate": 9.897073217852192e-06,
      "loss": 0.9427,
      "step": 3220
    },
    {
      "epoch": 0.09280815997233907,
      "grad_norm": 2.017669677734375,
      "learning_rate": 9.896979007190167e-06,
      "loss": 0.9634,
      "step": 3221
    },
    {
      "epoch": 0.09283697343398836,
      "grad_norm": 2.089278221130371,
      "learning_rate": 9.896884753880351e-06,
      "loss": 0.954,
      "step": 3222
    },
    {
      "epoch": 0.09286578689563764,
      "grad_norm": 1.938965916633606,
      "learning_rate": 9.896790457923564e-06,
      "loss": 0.9319,
      "step": 3223
    },
    {
      "epoch": 0.09289460035728693,
      "grad_norm": 1.971459984779358,
      "learning_rate": 9.896696119320625e-06,
      "loss": 0.9785,
      "step": 3224
    },
    {
      "epoch": 0.0929234138189362,
      "grad_norm": 2.150435447692871,
      "learning_rate": 9.89660173807236e-06,
      "loss": 0.9384,
      "step": 3225
    },
    {
      "epoch": 0.09295222728058548,
      "grad_norm": 1.8452105522155762,
      "learning_rate": 9.896507314179589e-06,
      "loss": 0.9505,
      "step": 3226
    },
    {
      "epoch": 0.09298104074223477,
      "grad_norm": 2.2417874336242676,
      "learning_rate": 9.896412847643133e-06,
      "loss": 0.9412,
      "step": 3227
    },
    {
      "epoch": 0.09300985420388405,
      "grad_norm": 1.9589775800704956,
      "learning_rate": 9.896318338463816e-06,
      "loss": 0.9133,
      "step": 3228
    },
    {
      "epoch": 0.09303866766553334,
      "grad_norm": 2.161224365234375,
      "learning_rate": 9.89622378664246e-06,
      "loss": 0.9586,
      "step": 3229
    },
    {
      "epoch": 0.09306748112718262,
      "grad_norm": 2.034222364425659,
      "learning_rate": 9.896129192179891e-06,
      "loss": 0.9592,
      "step": 3230
    },
    {
      "epoch": 0.0930962945888319,
      "grad_norm": 1.977239727973938,
      "learning_rate": 9.896034555076929e-06,
      "loss": 0.9408,
      "step": 3231
    },
    {
      "epoch": 0.09312510805048119,
      "grad_norm": 2.1309375762939453,
      "learning_rate": 9.895939875334401e-06,
      "loss": 0.9376,
      "step": 3232
    },
    {
      "epoch": 0.09315392151213046,
      "grad_norm": 1.889959692955017,
      "learning_rate": 9.895845152953131e-06,
      "loss": 0.9532,
      "step": 3233
    },
    {
      "epoch": 0.09318273497377975,
      "grad_norm": 2.10201096534729,
      "learning_rate": 9.895750387933945e-06,
      "loss": 0.9403,
      "step": 3234
    },
    {
      "epoch": 0.09321154843542903,
      "grad_norm": 2.0195884704589844,
      "learning_rate": 9.895655580277665e-06,
      "loss": 0.9575,
      "step": 3235
    },
    {
      "epoch": 0.09324036189707831,
      "grad_norm": 1.9641032218933105,
      "learning_rate": 9.895560729985121e-06,
      "loss": 0.933,
      "step": 3236
    },
    {
      "epoch": 0.0932691753587276,
      "grad_norm": 2.0543336868286133,
      "learning_rate": 9.895465837057135e-06,
      "loss": 0.9247,
      "step": 3237
    },
    {
      "epoch": 0.09329798882037688,
      "grad_norm": 2.069242477416992,
      "learning_rate": 9.895370901494536e-06,
      "loss": 0.9534,
      "step": 3238
    },
    {
      "epoch": 0.09332680228202617,
      "grad_norm": 2.034776210784912,
      "learning_rate": 9.895275923298149e-06,
      "loss": 0.9707,
      "step": 3239
    },
    {
      "epoch": 0.09335561574367544,
      "grad_norm": 1.9377551078796387,
      "learning_rate": 9.895180902468804e-06,
      "loss": 0.9068,
      "step": 3240
    },
    {
      "epoch": 0.09338442920532473,
      "grad_norm": 1.8797022104263306,
      "learning_rate": 9.895085839007325e-06,
      "loss": 0.9342,
      "step": 3241
    },
    {
      "epoch": 0.09341324266697401,
      "grad_norm": 2.0258641242980957,
      "learning_rate": 9.894990732914541e-06,
      "loss": 0.9237,
      "step": 3242
    },
    {
      "epoch": 0.09344205612862329,
      "grad_norm": 1.9156224727630615,
      "learning_rate": 9.894895584191284e-06,
      "loss": 0.9326,
      "step": 3243
    },
    {
      "epoch": 0.09347086959027258,
      "grad_norm": 2.113102436065674,
      "learning_rate": 9.894800392838377e-06,
      "loss": 0.9838,
      "step": 3244
    },
    {
      "epoch": 0.09349968305192186,
      "grad_norm": 2.0105433464050293,
      "learning_rate": 9.894705158856652e-06,
      "loss": 0.9662,
      "step": 3245
    },
    {
      "epoch": 0.09352849651357115,
      "grad_norm": 2.125354528427124,
      "learning_rate": 9.894609882246939e-06,
      "loss": 0.954,
      "step": 3246
    },
    {
      "epoch": 0.09355730997522042,
      "grad_norm": 1.9025497436523438,
      "learning_rate": 9.894514563010067e-06,
      "loss": 0.9508,
      "step": 3247
    },
    {
      "epoch": 0.0935861234368697,
      "grad_norm": 2.0353782176971436,
      "learning_rate": 9.894419201146866e-06,
      "loss": 0.928,
      "step": 3248
    },
    {
      "epoch": 0.09361493689851899,
      "grad_norm": 2.0223848819732666,
      "learning_rate": 9.894323796658164e-06,
      "loss": 0.9559,
      "step": 3249
    },
    {
      "epoch": 0.09364375036016827,
      "grad_norm": 2.1058247089385986,
      "learning_rate": 9.894228349544796e-06,
      "loss": 0.9607,
      "step": 3250
    },
    {
      "epoch": 0.09367256382181756,
      "grad_norm": 1.933660864830017,
      "learning_rate": 9.89413285980759e-06,
      "loss": 0.9185,
      "step": 3251
    },
    {
      "epoch": 0.09370137728346684,
      "grad_norm": 2.1478540897369385,
      "learning_rate": 9.89403732744738e-06,
      "loss": 0.9355,
      "step": 3252
    },
    {
      "epoch": 0.09373019074511611,
      "grad_norm": 1.9737927913665771,
      "learning_rate": 9.893941752465e-06,
      "loss": 0.9418,
      "step": 3253
    },
    {
      "epoch": 0.0937590042067654,
      "grad_norm": 1.9913651943206787,
      "learning_rate": 9.893846134861277e-06,
      "loss": 0.9189,
      "step": 3254
    },
    {
      "epoch": 0.09378781766841468,
      "grad_norm": 2.092437982559204,
      "learning_rate": 9.893750474637045e-06,
      "loss": 0.9545,
      "step": 3255
    },
    {
      "epoch": 0.09381663113006397,
      "grad_norm": 2.0365898609161377,
      "learning_rate": 9.893654771793139e-06,
      "loss": 0.9636,
      "step": 3256
    },
    {
      "epoch": 0.09384544459171325,
      "grad_norm": 2.015681743621826,
      "learning_rate": 9.893559026330393e-06,
      "loss": 0.9641,
      "step": 3257
    },
    {
      "epoch": 0.09387425805336252,
      "grad_norm": 2.0130984783172607,
      "learning_rate": 9.89346323824964e-06,
      "loss": 0.9433,
      "step": 3258
    },
    {
      "epoch": 0.09390307151501182,
      "grad_norm": 1.9091780185699463,
      "learning_rate": 9.89336740755171e-06,
      "loss": 0.961,
      "step": 3259
    },
    {
      "epoch": 0.09393188497666109,
      "grad_norm": 2.1439785957336426,
      "learning_rate": 9.893271534237445e-06,
      "loss": 0.9618,
      "step": 3260
    },
    {
      "epoch": 0.09396069843831038,
      "grad_norm": 2.1595144271850586,
      "learning_rate": 9.893175618307675e-06,
      "loss": 0.9417,
      "step": 3261
    },
    {
      "epoch": 0.09398951189995966,
      "grad_norm": 2.1224045753479004,
      "learning_rate": 9.893079659763239e-06,
      "loss": 0.9593,
      "step": 3262
    },
    {
      "epoch": 0.09401832536160894,
      "grad_norm": 2.0844266414642334,
      "learning_rate": 9.89298365860497e-06,
      "loss": 0.9613,
      "step": 3263
    },
    {
      "epoch": 0.09404713882325823,
      "grad_norm": 2.0072238445281982,
      "learning_rate": 9.892887614833702e-06,
      "loss": 0.9509,
      "step": 3264
    },
    {
      "epoch": 0.0940759522849075,
      "grad_norm": 2.1071929931640625,
      "learning_rate": 9.892791528450274e-06,
      "loss": 0.9607,
      "step": 3265
    },
    {
      "epoch": 0.0941047657465568,
      "grad_norm": 2.0137100219726562,
      "learning_rate": 9.892695399455526e-06,
      "loss": 0.9192,
      "step": 3266
    },
    {
      "epoch": 0.09413357920820607,
      "grad_norm": 1.958502173423767,
      "learning_rate": 9.892599227850289e-06,
      "loss": 0.951,
      "step": 3267
    },
    {
      "epoch": 0.09416239266985536,
      "grad_norm": 1.9522265195846558,
      "learning_rate": 9.892503013635405e-06,
      "loss": 0.9283,
      "step": 3268
    },
    {
      "epoch": 0.09419120613150464,
      "grad_norm": 2.0963666439056396,
      "learning_rate": 9.89240675681171e-06,
      "loss": 0.9548,
      "step": 3269
    },
    {
      "epoch": 0.09422001959315392,
      "grad_norm": 2.133012533187866,
      "learning_rate": 9.892310457380044e-06,
      "loss": 0.9756,
      "step": 3270
    },
    {
      "epoch": 0.09424883305480321,
      "grad_norm": 1.9478896856307983,
      "learning_rate": 9.892214115341242e-06,
      "loss": 0.8888,
      "step": 3271
    },
    {
      "epoch": 0.09427764651645248,
      "grad_norm": 1.9918280839920044,
      "learning_rate": 9.892117730696146e-06,
      "loss": 0.9474,
      "step": 3272
    },
    {
      "epoch": 0.09430645997810178,
      "grad_norm": 1.9296833276748657,
      "learning_rate": 9.892021303445596e-06,
      "loss": 0.9072,
      "step": 3273
    },
    {
      "epoch": 0.09433527343975105,
      "grad_norm": 2.07063889503479,
      "learning_rate": 9.891924833590431e-06,
      "loss": 0.9442,
      "step": 3274
    },
    {
      "epoch": 0.09436408690140033,
      "grad_norm": 1.966428279876709,
      "learning_rate": 9.89182832113149e-06,
      "loss": 0.9362,
      "step": 3275
    },
    {
      "epoch": 0.09439290036304962,
      "grad_norm": 1.9651129245758057,
      "learning_rate": 9.891731766069615e-06,
      "loss": 0.9514,
      "step": 3276
    },
    {
      "epoch": 0.0944217138246989,
      "grad_norm": 1.9702469110488892,
      "learning_rate": 9.891635168405647e-06,
      "loss": 0.9483,
      "step": 3277
    },
    {
      "epoch": 0.09445052728634819,
      "grad_norm": 1.9076486825942993,
      "learning_rate": 9.891538528140424e-06,
      "loss": 0.9246,
      "step": 3278
    },
    {
      "epoch": 0.09447934074799746,
      "grad_norm": 1.8956414461135864,
      "learning_rate": 9.891441845274792e-06,
      "loss": 0.9239,
      "step": 3279
    },
    {
      "epoch": 0.09450815420964674,
      "grad_norm": 1.8837816715240479,
      "learning_rate": 9.89134511980959e-06,
      "loss": 0.9623,
      "step": 3280
    },
    {
      "epoch": 0.09453696767129603,
      "grad_norm": 2.039185047149658,
      "learning_rate": 9.891248351745662e-06,
      "loss": 0.9291,
      "step": 3281
    },
    {
      "epoch": 0.09456578113294531,
      "grad_norm": 1.8880172967910767,
      "learning_rate": 9.891151541083853e-06,
      "loss": 0.9782,
      "step": 3282
    },
    {
      "epoch": 0.0945945945945946,
      "grad_norm": 1.9254212379455566,
      "learning_rate": 9.891054687824999e-06,
      "loss": 0.941,
      "step": 3283
    },
    {
      "epoch": 0.09462340805624388,
      "grad_norm": 2.0273032188415527,
      "learning_rate": 9.89095779196995e-06,
      "loss": 0.9343,
      "step": 3284
    },
    {
      "epoch": 0.09465222151789315,
      "grad_norm": 2.032919406890869,
      "learning_rate": 9.890860853519549e-06,
      "loss": 0.9248,
      "step": 3285
    },
    {
      "epoch": 0.09468103497954244,
      "grad_norm": 1.9991962909698486,
      "learning_rate": 9.890763872474636e-06,
      "loss": 0.9583,
      "step": 3286
    },
    {
      "epoch": 0.09470984844119172,
      "grad_norm": 1.9796594381332397,
      "learning_rate": 9.89066684883606e-06,
      "loss": 0.9514,
      "step": 3287
    },
    {
      "epoch": 0.09473866190284101,
      "grad_norm": 1.99705970287323,
      "learning_rate": 9.890569782604667e-06,
      "loss": 0.9493,
      "step": 3288
    },
    {
      "epoch": 0.09476747536449029,
      "grad_norm": 1.9296492338180542,
      "learning_rate": 9.890472673781297e-06,
      "loss": 0.9386,
      "step": 3289
    },
    {
      "epoch": 0.09479628882613957,
      "grad_norm": 1.834518551826477,
      "learning_rate": 9.8903755223668e-06,
      "loss": 0.9455,
      "step": 3290
    },
    {
      "epoch": 0.09482510228778886,
      "grad_norm": 1.9688471555709839,
      "learning_rate": 9.890278328362021e-06,
      "loss": 0.9272,
      "step": 3291
    },
    {
      "epoch": 0.09485391574943813,
      "grad_norm": 1.992104172706604,
      "learning_rate": 9.890181091767806e-06,
      "loss": 0.9548,
      "step": 3292
    },
    {
      "epoch": 0.09488272921108742,
      "grad_norm": 1.9893232583999634,
      "learning_rate": 9.890083812585e-06,
      "loss": 0.9317,
      "step": 3293
    },
    {
      "epoch": 0.0949115426727367,
      "grad_norm": 1.9082530736923218,
      "learning_rate": 9.889986490814456e-06,
      "loss": 0.9387,
      "step": 3294
    },
    {
      "epoch": 0.09494035613438599,
      "grad_norm": 2.095548391342163,
      "learning_rate": 9.889889126457015e-06,
      "loss": 0.9463,
      "step": 3295
    },
    {
      "epoch": 0.09496916959603527,
      "grad_norm": 2.0637497901916504,
      "learning_rate": 9.889791719513528e-06,
      "loss": 0.9171,
      "step": 3296
    },
    {
      "epoch": 0.09499798305768455,
      "grad_norm": 1.975091814994812,
      "learning_rate": 9.889694269984845e-06,
      "loss": 0.9315,
      "step": 3297
    },
    {
      "epoch": 0.09502679651933384,
      "grad_norm": 2.0285162925720215,
      "learning_rate": 9.889596777871813e-06,
      "loss": 0.9493,
      "step": 3298
    },
    {
      "epoch": 0.09505560998098311,
      "grad_norm": 2.0465402603149414,
      "learning_rate": 9.889499243175278e-06,
      "loss": 0.9711,
      "step": 3299
    },
    {
      "epoch": 0.0950844234426324,
      "grad_norm": 1.9163495302200317,
      "learning_rate": 9.889401665896095e-06,
      "loss": 0.9306,
      "step": 3300
    },
    {
      "epoch": 0.09511323690428168,
      "grad_norm": 2.025444984436035,
      "learning_rate": 9.88930404603511e-06,
      "loss": 0.9481,
      "step": 3301
    },
    {
      "epoch": 0.09514205036593096,
      "grad_norm": 2.0343222618103027,
      "learning_rate": 9.889206383593175e-06,
      "loss": 0.9382,
      "step": 3302
    },
    {
      "epoch": 0.09517086382758025,
      "grad_norm": 1.9741519689559937,
      "learning_rate": 9.889108678571138e-06,
      "loss": 0.9022,
      "step": 3303
    },
    {
      "epoch": 0.09519967728922953,
      "grad_norm": 1.9531043767929077,
      "learning_rate": 9.889010930969854e-06,
      "loss": 0.9239,
      "step": 3304
    },
    {
      "epoch": 0.09522849075087882,
      "grad_norm": 2.0596821308135986,
      "learning_rate": 9.88891314079017e-06,
      "loss": 0.9511,
      "step": 3305
    },
    {
      "epoch": 0.0952573042125281,
      "grad_norm": 2.0650177001953125,
      "learning_rate": 9.88881530803294e-06,
      "loss": 0.9323,
      "step": 3306
    },
    {
      "epoch": 0.09528611767417737,
      "grad_norm": 1.9960849285125732,
      "learning_rate": 9.888717432699016e-06,
      "loss": 0.9235,
      "step": 3307
    },
    {
      "epoch": 0.09531493113582666,
      "grad_norm": 2.0353877544403076,
      "learning_rate": 9.88861951478925e-06,
      "loss": 0.9598,
      "step": 3308
    },
    {
      "epoch": 0.09534374459747594,
      "grad_norm": 2.0041143894195557,
      "learning_rate": 9.888521554304496e-06,
      "loss": 0.9314,
      "step": 3309
    },
    {
      "epoch": 0.09537255805912523,
      "grad_norm": 2.0552256107330322,
      "learning_rate": 9.888423551245604e-06,
      "loss": 0.9447,
      "step": 3310
    },
    {
      "epoch": 0.0954013715207745,
      "grad_norm": 1.9988974332809448,
      "learning_rate": 9.888325505613433e-06,
      "loss": 0.9281,
      "step": 3311
    },
    {
      "epoch": 0.09543018498242378,
      "grad_norm": 2.156590700149536,
      "learning_rate": 9.88822741740883e-06,
      "loss": 0.9183,
      "step": 3312
    },
    {
      "epoch": 0.09545899844407307,
      "grad_norm": 1.9881151914596558,
      "learning_rate": 9.888129286632653e-06,
      "loss": 0.8904,
      "step": 3313
    },
    {
      "epoch": 0.09548781190572235,
      "grad_norm": 1.968496561050415,
      "learning_rate": 9.888031113285758e-06,
      "loss": 0.9372,
      "step": 3314
    },
    {
      "epoch": 0.09551662536737164,
      "grad_norm": 1.9466702938079834,
      "learning_rate": 9.887932897368996e-06,
      "loss": 0.9619,
      "step": 3315
    },
    {
      "epoch": 0.09554543882902092,
      "grad_norm": 1.9962385892868042,
      "learning_rate": 9.887834638883227e-06,
      "loss": 0.9237,
      "step": 3316
    },
    {
      "epoch": 0.0955742522906702,
      "grad_norm": 2.1214983463287354,
      "learning_rate": 9.887736337829303e-06,
      "loss": 0.9196,
      "step": 3317
    },
    {
      "epoch": 0.09560306575231949,
      "grad_norm": 1.9150125980377197,
      "learning_rate": 9.88763799420808e-06,
      "loss": 0.9185,
      "step": 3318
    },
    {
      "epoch": 0.09563187921396876,
      "grad_norm": 2.0800437927246094,
      "learning_rate": 9.887539608020417e-06,
      "loss": 0.9488,
      "step": 3319
    },
    {
      "epoch": 0.09566069267561805,
      "grad_norm": 1.9465967416763306,
      "learning_rate": 9.887441179267173e-06,
      "loss": 0.9091,
      "step": 3320
    },
    {
      "epoch": 0.09568950613726733,
      "grad_norm": 2.0825681686401367,
      "learning_rate": 9.887342707949198e-06,
      "loss": 0.9356,
      "step": 3321
    },
    {
      "epoch": 0.0957183195989166,
      "grad_norm": 1.8715753555297852,
      "learning_rate": 9.887244194067355e-06,
      "loss": 0.9242,
      "step": 3322
    },
    {
      "epoch": 0.0957471330605659,
      "grad_norm": 2.1533491611480713,
      "learning_rate": 9.8871456376225e-06,
      "loss": 0.9143,
      "step": 3323
    },
    {
      "epoch": 0.09577594652221517,
      "grad_norm": 2.0891811847686768,
      "learning_rate": 9.887047038615492e-06,
      "loss": 0.9619,
      "step": 3324
    },
    {
      "epoch": 0.09580475998386447,
      "grad_norm": 2.0524165630340576,
      "learning_rate": 9.88694839704719e-06,
      "loss": 0.9422,
      "step": 3325
    },
    {
      "epoch": 0.09583357344551374,
      "grad_norm": 2.167574882507324,
      "learning_rate": 9.886849712918452e-06,
      "loss": 0.9366,
      "step": 3326
    },
    {
      "epoch": 0.09586238690716303,
      "grad_norm": 2.0903124809265137,
      "learning_rate": 9.886750986230139e-06,
      "loss": 0.923,
      "step": 3327
    },
    {
      "epoch": 0.09589120036881231,
      "grad_norm": 1.997204065322876,
      "learning_rate": 9.886652216983109e-06,
      "loss": 0.9153,
      "step": 3328
    },
    {
      "epoch": 0.09592001383046159,
      "grad_norm": 2.097384214401245,
      "learning_rate": 9.886553405178223e-06,
      "loss": 0.9389,
      "step": 3329
    },
    {
      "epoch": 0.09594882729211088,
      "grad_norm": 1.9117677211761475,
      "learning_rate": 9.886454550816342e-06,
      "loss": 0.9282,
      "step": 3330
    },
    {
      "epoch": 0.09597764075376015,
      "grad_norm": 2.0812435150146484,
      "learning_rate": 9.886355653898327e-06,
      "loss": 0.9142,
      "step": 3331
    },
    {
      "epoch": 0.09600645421540945,
      "grad_norm": 2.069981575012207,
      "learning_rate": 9.886256714425037e-06,
      "loss": 0.9254,
      "step": 3332
    },
    {
      "epoch": 0.09603526767705872,
      "grad_norm": 2.200965642929077,
      "learning_rate": 9.886157732397339e-06,
      "loss": 0.9125,
      "step": 3333
    },
    {
      "epoch": 0.096064081138708,
      "grad_norm": 1.9855388402938843,
      "learning_rate": 9.886058707816088e-06,
      "loss": 0.9081,
      "step": 3334
    },
    {
      "epoch": 0.09609289460035729,
      "grad_norm": 1.9308127164840698,
      "learning_rate": 9.885959640682153e-06,
      "loss": 0.9273,
      "step": 3335
    },
    {
      "epoch": 0.09612170806200657,
      "grad_norm": 1.9506994485855103,
      "learning_rate": 9.885860530996391e-06,
      "loss": 0.9158,
      "step": 3336
    },
    {
      "epoch": 0.09615052152365586,
      "grad_norm": 2.0544865131378174,
      "learning_rate": 9.885761378759669e-06,
      "loss": 0.9494,
      "step": 3337
    },
    {
      "epoch": 0.09617933498530513,
      "grad_norm": 1.8901325464248657,
      "learning_rate": 9.885662183972848e-06,
      "loss": 0.9483,
      "step": 3338
    },
    {
      "epoch": 0.09620814844695441,
      "grad_norm": 2.011967182159424,
      "learning_rate": 9.885562946636795e-06,
      "loss": 0.9176,
      "step": 3339
    },
    {
      "epoch": 0.0962369619086037,
      "grad_norm": 2.0593292713165283,
      "learning_rate": 9.885463666752372e-06,
      "loss": 0.9441,
      "step": 3340
    },
    {
      "epoch": 0.09626577537025298,
      "grad_norm": 1.929129719734192,
      "learning_rate": 9.885364344320443e-06,
      "loss": 0.9101,
      "step": 3341
    },
    {
      "epoch": 0.09629458883190227,
      "grad_norm": 2.0251898765563965,
      "learning_rate": 9.885264979341877e-06,
      "loss": 0.9466,
      "step": 3342
    },
    {
      "epoch": 0.09632340229355155,
      "grad_norm": 2.1299386024475098,
      "learning_rate": 9.885165571817534e-06,
      "loss": 0.9407,
      "step": 3343
    },
    {
      "epoch": 0.09635221575520082,
      "grad_norm": 1.7970824241638184,
      "learning_rate": 9.885066121748284e-06,
      "loss": 0.9333,
      "step": 3344
    },
    {
      "epoch": 0.09638102921685011,
      "grad_norm": 1.953701138496399,
      "learning_rate": 9.884966629134988e-06,
      "loss": 0.9284,
      "step": 3345
    },
    {
      "epoch": 0.09640984267849939,
      "grad_norm": 2.098435878753662,
      "learning_rate": 9.88486709397852e-06,
      "loss": 0.9046,
      "step": 3346
    },
    {
      "epoch": 0.09643865614014868,
      "grad_norm": 1.8374733924865723,
      "learning_rate": 9.884767516279741e-06,
      "loss": 0.9351,
      "step": 3347
    },
    {
      "epoch": 0.09646746960179796,
      "grad_norm": 2.087475538253784,
      "learning_rate": 9.88466789603952e-06,
      "loss": 0.9082,
      "step": 3348
    },
    {
      "epoch": 0.09649628306344724,
      "grad_norm": 1.9744068384170532,
      "learning_rate": 9.884568233258724e-06,
      "loss": 0.9165,
      "step": 3349
    },
    {
      "epoch": 0.09652509652509653,
      "grad_norm": 1.8365837335586548,
      "learning_rate": 9.88446852793822e-06,
      "loss": 0.9014,
      "step": 3350
    },
    {
      "epoch": 0.0965539099867458,
      "grad_norm": 1.9563862085342407,
      "learning_rate": 9.884368780078881e-06,
      "loss": 0.9568,
      "step": 3351
    },
    {
      "epoch": 0.0965827234483951,
      "grad_norm": 2.000056505203247,
      "learning_rate": 9.884268989681571e-06,
      "loss": 0.9314,
      "step": 3352
    },
    {
      "epoch": 0.09661153691004437,
      "grad_norm": 2.040363073348999,
      "learning_rate": 9.884169156747161e-06,
      "loss": 0.9074,
      "step": 3353
    },
    {
      "epoch": 0.09664035037169366,
      "grad_norm": 2.0219712257385254,
      "learning_rate": 9.884069281276519e-06,
      "loss": 0.8876,
      "step": 3354
    },
    {
      "epoch": 0.09666916383334294,
      "grad_norm": 2.187202215194702,
      "learning_rate": 9.883969363270516e-06,
      "loss": 0.9354,
      "step": 3355
    },
    {
      "epoch": 0.09669797729499222,
      "grad_norm": 2.0332601070404053,
      "learning_rate": 9.883869402730023e-06,
      "loss": 0.931,
      "step": 3356
    },
    {
      "epoch": 0.0967267907566415,
      "grad_norm": 1.9339216947555542,
      "learning_rate": 9.88376939965591e-06,
      "loss": 0.9093,
      "step": 3357
    },
    {
      "epoch": 0.09675560421829078,
      "grad_norm": 2.035679340362549,
      "learning_rate": 9.883669354049045e-06,
      "loss": 0.9108,
      "step": 3358
    },
    {
      "epoch": 0.09678441767994007,
      "grad_norm": 2.061583995819092,
      "learning_rate": 9.883569265910305e-06,
      "loss": 0.9287,
      "step": 3359
    },
    {
      "epoch": 0.09681323114158935,
      "grad_norm": 2.193763017654419,
      "learning_rate": 9.883469135240555e-06,
      "loss": 0.8951,
      "step": 3360
    },
    {
      "epoch": 0.09684204460323863,
      "grad_norm": 2.041391611099243,
      "learning_rate": 9.883368962040674e-06,
      "loss": 0.9074,
      "step": 3361
    },
    {
      "epoch": 0.09687085806488792,
      "grad_norm": 2.169782876968384,
      "learning_rate": 9.88326874631153e-06,
      "loss": 0.9062,
      "step": 3362
    },
    {
      "epoch": 0.0968996715265372,
      "grad_norm": 2.074995517730713,
      "learning_rate": 9.883168488053994e-06,
      "loss": 0.9443,
      "step": 3363
    },
    {
      "epoch": 0.09692848498818649,
      "grad_norm": 1.8967053890228271,
      "learning_rate": 9.883068187268945e-06,
      "loss": 0.904,
      "step": 3364
    },
    {
      "epoch": 0.09695729844983576,
      "grad_norm": 2.067981004714966,
      "learning_rate": 9.882967843957252e-06,
      "loss": 0.9099,
      "step": 3365
    },
    {
      "epoch": 0.09698611191148504,
      "grad_norm": 1.9427815675735474,
      "learning_rate": 9.88286745811979e-06,
      "loss": 0.9171,
      "step": 3366
    },
    {
      "epoch": 0.09701492537313433,
      "grad_norm": 2.0782361030578613,
      "learning_rate": 9.882767029757434e-06,
      "loss": 0.894,
      "step": 3367
    },
    {
      "epoch": 0.09704373883478361,
      "grad_norm": 1.969459056854248,
      "learning_rate": 9.88266655887106e-06,
      "loss": 0.9208,
      "step": 3368
    },
    {
      "epoch": 0.0970725522964329,
      "grad_norm": 2.1588408946990967,
      "learning_rate": 9.882566045461539e-06,
      "loss": 0.902,
      "step": 3369
    },
    {
      "epoch": 0.09710136575808218,
      "grad_norm": 2.098170280456543,
      "learning_rate": 9.882465489529747e-06,
      "loss": 0.9414,
      "step": 3370
    },
    {
      "epoch": 0.09713017921973145,
      "grad_norm": 2.1600961685180664,
      "learning_rate": 9.882364891076564e-06,
      "loss": 0.928,
      "step": 3371
    },
    {
      "epoch": 0.09715899268138074,
      "grad_norm": 2.024040460586548,
      "learning_rate": 9.882264250102863e-06,
      "loss": 0.902,
      "step": 3372
    },
    {
      "epoch": 0.09718780614303002,
      "grad_norm": 1.990200400352478,
      "learning_rate": 9.88216356660952e-06,
      "loss": 0.9009,
      "step": 3373
    },
    {
      "epoch": 0.09721661960467931,
      "grad_norm": 1.936556100845337,
      "learning_rate": 9.882062840597412e-06,
      "loss": 0.9368,
      "step": 3374
    },
    {
      "epoch": 0.09724543306632859,
      "grad_norm": 1.9720712900161743,
      "learning_rate": 9.881962072067419e-06,
      "loss": 0.8808,
      "step": 3375
    },
    {
      "epoch": 0.09727424652797786,
      "grad_norm": 1.8746494054794312,
      "learning_rate": 9.881861261020415e-06,
      "loss": 0.9116,
      "step": 3376
    },
    {
      "epoch": 0.09730305998962716,
      "grad_norm": 2.125533103942871,
      "learning_rate": 9.881760407457281e-06,
      "loss": 0.9246,
      "step": 3377
    },
    {
      "epoch": 0.09733187345127643,
      "grad_norm": 2.0698492527008057,
      "learning_rate": 9.881659511378893e-06,
      "loss": 0.9141,
      "step": 3378
    },
    {
      "epoch": 0.09736068691292572,
      "grad_norm": 2.008794069290161,
      "learning_rate": 9.88155857278613e-06,
      "loss": 0.9017,
      "step": 3379
    },
    {
      "epoch": 0.097389500374575,
      "grad_norm": 2.0166704654693604,
      "learning_rate": 9.881457591679872e-06,
      "loss": 0.926,
      "step": 3380
    },
    {
      "epoch": 0.09741831383622429,
      "grad_norm": 1.9826046228408813,
      "learning_rate": 9.881356568060997e-06,
      "loss": 0.9119,
      "step": 3381
    },
    {
      "epoch": 0.09744712729787357,
      "grad_norm": 2.0751752853393555,
      "learning_rate": 9.881255501930388e-06,
      "loss": 0.9098,
      "step": 3382
    },
    {
      "epoch": 0.09747594075952284,
      "grad_norm": 1.918689250946045,
      "learning_rate": 9.88115439328892e-06,
      "loss": 0.9075,
      "step": 3383
    },
    {
      "epoch": 0.09750475422117214,
      "grad_norm": 1.9713072776794434,
      "learning_rate": 9.88105324213748e-06,
      "loss": 0.9015,
      "step": 3384
    },
    {
      "epoch": 0.09753356768282141,
      "grad_norm": 2.0504965782165527,
      "learning_rate": 9.880952048476944e-06,
      "loss": 0.9108,
      "step": 3385
    },
    {
      "epoch": 0.0975623811444707,
      "grad_norm": 2.0077970027923584,
      "learning_rate": 9.880850812308195e-06,
      "loss": 0.9298,
      "step": 3386
    },
    {
      "epoch": 0.09759119460611998,
      "grad_norm": 2.130924940109253,
      "learning_rate": 9.880749533632115e-06,
      "loss": 0.8952,
      "step": 3387
    },
    {
      "epoch": 0.09762000806776926,
      "grad_norm": 2.086573362350464,
      "learning_rate": 9.880648212449585e-06,
      "loss": 0.9046,
      "step": 3388
    },
    {
      "epoch": 0.09764882152941855,
      "grad_norm": 2.1122474670410156,
      "learning_rate": 9.880546848761488e-06,
      "loss": 0.8763,
      "step": 3389
    },
    {
      "epoch": 0.09767763499106782,
      "grad_norm": 1.9835577011108398,
      "learning_rate": 9.880445442568705e-06,
      "loss": 0.927,
      "step": 3390
    },
    {
      "epoch": 0.09770644845271712,
      "grad_norm": 2.1141533851623535,
      "learning_rate": 9.880343993872123e-06,
      "loss": 0.8998,
      "step": 3391
    },
    {
      "epoch": 0.09773526191436639,
      "grad_norm": 1.8995959758758545,
      "learning_rate": 9.880242502672623e-06,
      "loss": 0.8831,
      "step": 3392
    },
    {
      "epoch": 0.09776407537601567,
      "grad_norm": 2.0181407928466797,
      "learning_rate": 9.880140968971089e-06,
      "loss": 0.928,
      "step": 3393
    },
    {
      "epoch": 0.09779288883766496,
      "grad_norm": 2.0286314487457275,
      "learning_rate": 9.880039392768405e-06,
      "loss": 0.9066,
      "step": 3394
    },
    {
      "epoch": 0.09782170229931424,
      "grad_norm": 1.859434723854065,
      "learning_rate": 9.879937774065458e-06,
      "loss": 0.8863,
      "step": 3395
    },
    {
      "epoch": 0.09785051576096353,
      "grad_norm": 2.0234203338623047,
      "learning_rate": 9.879836112863129e-06,
      "loss": 0.9015,
      "step": 3396
    },
    {
      "epoch": 0.0978793292226128,
      "grad_norm": 1.8865516185760498,
      "learning_rate": 9.879734409162307e-06,
      "loss": 0.875,
      "step": 3397
    },
    {
      "epoch": 0.09790814268426208,
      "grad_norm": 1.9625020027160645,
      "learning_rate": 9.879632662963876e-06,
      "loss": 0.9099,
      "step": 3398
    },
    {
      "epoch": 0.09793695614591137,
      "grad_norm": 1.9787262678146362,
      "learning_rate": 9.879530874268722e-06,
      "loss": 0.8993,
      "step": 3399
    },
    {
      "epoch": 0.09796576960756065,
      "grad_norm": 1.950812578201294,
      "learning_rate": 9.87942904307773e-06,
      "loss": 0.8679,
      "step": 3400
    },
    {
      "epoch": 0.09799458306920994,
      "grad_norm": 1.8527393341064453,
      "learning_rate": 9.879327169391791e-06,
      "loss": 0.915,
      "step": 3401
    },
    {
      "epoch": 0.09802339653085922,
      "grad_norm": 2.013498306274414,
      "learning_rate": 9.87922525321179e-06,
      "loss": 0.9011,
      "step": 3402
    },
    {
      "epoch": 0.0980522099925085,
      "grad_norm": 1.8707618713378906,
      "learning_rate": 9.879123294538614e-06,
      "loss": 0.9068,
      "step": 3403
    },
    {
      "epoch": 0.09808102345415778,
      "grad_norm": 2.013155698776245,
      "learning_rate": 9.879021293373152e-06,
      "loss": 0.9071,
      "step": 3404
    },
    {
      "epoch": 0.09810983691580706,
      "grad_norm": 1.871360421180725,
      "learning_rate": 9.87891924971629e-06,
      "loss": 0.9225,
      "step": 3405
    },
    {
      "epoch": 0.09813865037745635,
      "grad_norm": 1.8413981199264526,
      "learning_rate": 9.878817163568919e-06,
      "loss": 0.9032,
      "step": 3406
    },
    {
      "epoch": 0.09816746383910563,
      "grad_norm": 1.9819424152374268,
      "learning_rate": 9.878715034931929e-06,
      "loss": 0.8953,
      "step": 3407
    },
    {
      "epoch": 0.09819627730075492,
      "grad_norm": 2.0818798542022705,
      "learning_rate": 9.878612863806207e-06,
      "loss": 0.9226,
      "step": 3408
    },
    {
      "epoch": 0.0982250907624042,
      "grad_norm": 2.063422203063965,
      "learning_rate": 9.878510650192644e-06,
      "loss": 0.8884,
      "step": 3409
    },
    {
      "epoch": 0.09825390422405347,
      "grad_norm": 1.9856919050216675,
      "learning_rate": 9.87840839409213e-06,
      "loss": 0.9006,
      "step": 3410
    },
    {
      "epoch": 0.09828271768570276,
      "grad_norm": 2.2399497032165527,
      "learning_rate": 9.878306095505555e-06,
      "loss": 0.8974,
      "step": 3411
    },
    {
      "epoch": 0.09831153114735204,
      "grad_norm": 1.9851666688919067,
      "learning_rate": 9.878203754433811e-06,
      "loss": 0.8819,
      "step": 3412
    },
    {
      "epoch": 0.09834034460900133,
      "grad_norm": 1.9919911623001099,
      "learning_rate": 9.878101370877788e-06,
      "loss": 0.8918,
      "step": 3413
    },
    {
      "epoch": 0.09836915807065061,
      "grad_norm": 2.1469948291778564,
      "learning_rate": 9.877998944838378e-06,
      "loss": 0.8892,
      "step": 3414
    },
    {
      "epoch": 0.09839797153229989,
      "grad_norm": 1.9806081056594849,
      "learning_rate": 9.877896476316475e-06,
      "loss": 0.8796,
      "step": 3415
    },
    {
      "epoch": 0.09842678499394918,
      "grad_norm": 2.1945292949676514,
      "learning_rate": 9.87779396531297e-06,
      "loss": 0.8747,
      "step": 3416
    },
    {
      "epoch": 0.09845559845559845,
      "grad_norm": 2.0063438415527344,
      "learning_rate": 9.877691411828753e-06,
      "loss": 0.9042,
      "step": 3417
    },
    {
      "epoch": 0.09848441191724774,
      "grad_norm": 2.4192330837249756,
      "learning_rate": 9.877588815864722e-06,
      "loss": 0.9112,
      "step": 3418
    },
    {
      "epoch": 0.09851322537889702,
      "grad_norm": 1.893815279006958,
      "learning_rate": 9.877486177421767e-06,
      "loss": 0.8952,
      "step": 3419
    },
    {
      "epoch": 0.0985420388405463,
      "grad_norm": 2.3448586463928223,
      "learning_rate": 9.877383496500783e-06,
      "loss": 0.8811,
      "step": 3420
    },
    {
      "epoch": 0.09857085230219559,
      "grad_norm": 2.0124292373657227,
      "learning_rate": 9.877280773102666e-06,
      "loss": 0.8769,
      "step": 3421
    },
    {
      "epoch": 0.09859966576384487,
      "grad_norm": 2.3491270542144775,
      "learning_rate": 9.877178007228307e-06,
      "loss": 0.9016,
      "step": 3422
    },
    {
      "epoch": 0.09862847922549416,
      "grad_norm": 2.0415971279144287,
      "learning_rate": 9.877075198878605e-06,
      "loss": 0.8666,
      "step": 3423
    },
    {
      "epoch": 0.09865729268714343,
      "grad_norm": 2.0681867599487305,
      "learning_rate": 9.876972348054453e-06,
      "loss": 0.8629,
      "step": 3424
    },
    {
      "epoch": 0.09868610614879271,
      "grad_norm": 2.0479907989501953,
      "learning_rate": 9.876869454756745e-06,
      "loss": 0.8767,
      "step": 3425
    },
    {
      "epoch": 0.098714919610442,
      "grad_norm": 1.9650764465332031,
      "learning_rate": 9.87676651898638e-06,
      "loss": 0.8772,
      "step": 3426
    },
    {
      "epoch": 0.09874373307209128,
      "grad_norm": 1.9945192337036133,
      "learning_rate": 9.876663540744255e-06,
      "loss": 0.8932,
      "step": 3427
    },
    {
      "epoch": 0.09877254653374057,
      "grad_norm": 2.119976043701172,
      "learning_rate": 9.876560520031264e-06,
      "loss": 0.908,
      "step": 3428
    },
    {
      "epoch": 0.09880135999538985,
      "grad_norm": 1.782120943069458,
      "learning_rate": 9.876457456848304e-06,
      "loss": 0.8781,
      "step": 3429
    },
    {
      "epoch": 0.09883017345703912,
      "grad_norm": 2.074291944503784,
      "learning_rate": 9.876354351196277e-06,
      "loss": 0.9161,
      "step": 3430
    },
    {
      "epoch": 0.09885898691868841,
      "grad_norm": 1.903593897819519,
      "learning_rate": 9.87625120307608e-06,
      "loss": 0.9014,
      "step": 3431
    },
    {
      "epoch": 0.09888780038033769,
      "grad_norm": 1.8574788570404053,
      "learning_rate": 9.876148012488606e-06,
      "loss": 0.8738,
      "step": 3432
    },
    {
      "epoch": 0.09891661384198698,
      "grad_norm": 1.9547410011291504,
      "learning_rate": 9.876044779434756e-06,
      "loss": 0.9062,
      "step": 3433
    },
    {
      "epoch": 0.09894542730363626,
      "grad_norm": 1.7961711883544922,
      "learning_rate": 9.875941503915433e-06,
      "loss": 0.8751,
      "step": 3434
    },
    {
      "epoch": 0.09897424076528555,
      "grad_norm": 2.0101826190948486,
      "learning_rate": 9.875838185931535e-06,
      "loss": 0.9063,
      "step": 3435
    },
    {
      "epoch": 0.09900305422693483,
      "grad_norm": 1.7834346294403076,
      "learning_rate": 9.875734825483959e-06,
      "loss": 0.8743,
      "step": 3436
    },
    {
      "epoch": 0.0990318676885841,
      "grad_norm": 2.001819133758545,
      "learning_rate": 9.875631422573605e-06,
      "loss": 0.8639,
      "step": 3437
    },
    {
      "epoch": 0.09906068115023339,
      "grad_norm": 1.8624924421310425,
      "learning_rate": 9.875527977201375e-06,
      "loss": 0.8717,
      "step": 3438
    },
    {
      "epoch": 0.09908949461188267,
      "grad_norm": 1.9407758712768555,
      "learning_rate": 9.875424489368174e-06,
      "loss": 0.8747,
      "step": 3439
    },
    {
      "epoch": 0.09911830807353196,
      "grad_norm": 1.7864325046539307,
      "learning_rate": 9.875320959074894e-06,
      "loss": 0.8569,
      "step": 3440
    },
    {
      "epoch": 0.09914712153518124,
      "grad_norm": 1.8634628057479858,
      "learning_rate": 9.875217386322445e-06,
      "loss": 0.8824,
      "step": 3441
    },
    {
      "epoch": 0.09917593499683051,
      "grad_norm": 1.9812407493591309,
      "learning_rate": 9.875113771111726e-06,
      "loss": 0.8962,
      "step": 3442
    },
    {
      "epoch": 0.0992047484584798,
      "grad_norm": 1.8014535903930664,
      "learning_rate": 9.87501011344364e-06,
      "loss": 0.8961,
      "step": 3443
    },
    {
      "epoch": 0.09923356192012908,
      "grad_norm": 1.8919494152069092,
      "learning_rate": 9.874906413319087e-06,
      "loss": 0.8522,
      "step": 3444
    },
    {
      "epoch": 0.09926237538177837,
      "grad_norm": 1.803529977798462,
      "learning_rate": 9.874802670738973e-06,
      "loss": 0.8775,
      "step": 3445
    },
    {
      "epoch": 0.09929118884342765,
      "grad_norm": 2.013094186782837,
      "learning_rate": 9.874698885704201e-06,
      "loss": 0.861,
      "step": 3446
    },
    {
      "epoch": 0.09932000230507693,
      "grad_norm": 1.885043978691101,
      "learning_rate": 9.874595058215675e-06,
      "loss": 0.8463,
      "step": 3447
    },
    {
      "epoch": 0.09934881576672622,
      "grad_norm": 1.7961745262145996,
      "learning_rate": 9.8744911882743e-06,
      "loss": 0.8685,
      "step": 3448
    },
    {
      "epoch": 0.0993776292283755,
      "grad_norm": 1.7767610549926758,
      "learning_rate": 9.874387275880976e-06,
      "loss": 0.8708,
      "step": 3449
    },
    {
      "epoch": 0.09940644269002478,
      "grad_norm": 1.9071216583251953,
      "learning_rate": 9.874283321036615e-06,
      "loss": 0.8561,
      "step": 3450
    },
    {
      "epoch": 0.09943525615167406,
      "grad_norm": 1.7345690727233887,
      "learning_rate": 9.874179323742117e-06,
      "loss": 0.8648,
      "step": 3451
    },
    {
      "epoch": 0.09946406961332334,
      "grad_norm": 1.7970930337905884,
      "learning_rate": 9.87407528399839e-06,
      "loss": 0.8673,
      "step": 3452
    },
    {
      "epoch": 0.09949288307497263,
      "grad_norm": 1.8854880332946777,
      "learning_rate": 9.873971201806338e-06,
      "loss": 0.8613,
      "step": 3453
    },
    {
      "epoch": 0.0995216965366219,
      "grad_norm": 1.806390643119812,
      "learning_rate": 9.873867077166872e-06,
      "loss": 0.8599,
      "step": 3454
    },
    {
      "epoch": 0.0995505099982712,
      "grad_norm": 1.7840232849121094,
      "learning_rate": 9.873762910080892e-06,
      "loss": 0.8857,
      "step": 3455
    },
    {
      "epoch": 0.09957932345992047,
      "grad_norm": 1.8509951829910278,
      "learning_rate": 9.873658700549314e-06,
      "loss": 0.8532,
      "step": 3456
    },
    {
      "epoch": 0.09960813692156975,
      "grad_norm": 1.6840256452560425,
      "learning_rate": 9.873554448573037e-06,
      "loss": 0.8486,
      "step": 3457
    },
    {
      "epoch": 0.09963695038321904,
      "grad_norm": 1.8890657424926758,
      "learning_rate": 9.873450154152972e-06,
      "loss": 0.8628,
      "step": 3458
    },
    {
      "epoch": 0.09966576384486832,
      "grad_norm": 1.8601018190383911,
      "learning_rate": 9.87334581729003e-06,
      "loss": 0.8853,
      "step": 3459
    },
    {
      "epoch": 0.09969457730651761,
      "grad_norm": 1.7228907346725464,
      "learning_rate": 9.873241437985117e-06,
      "loss": 0.8982,
      "step": 3460
    },
    {
      "epoch": 0.09972339076816689,
      "grad_norm": 1.8872003555297852,
      "learning_rate": 9.873137016239143e-06,
      "loss": 0.889,
      "step": 3461
    },
    {
      "epoch": 0.09975220422981616,
      "grad_norm": 1.8351627588272095,
      "learning_rate": 9.873032552053015e-06,
      "loss": 0.8801,
      "step": 3462
    },
    {
      "epoch": 0.09978101769146545,
      "grad_norm": 1.8194938898086548,
      "learning_rate": 9.872928045427646e-06,
      "loss": 0.8599,
      "step": 3463
    },
    {
      "epoch": 0.09980983115311473,
      "grad_norm": 1.8260804414749146,
      "learning_rate": 9.872823496363944e-06,
      "loss": 0.8737,
      "step": 3464
    },
    {
      "epoch": 0.09983864461476402,
      "grad_norm": 1.7754546403884888,
      "learning_rate": 9.872718904862821e-06,
      "loss": 0.8921,
      "step": 3465
    },
    {
      "epoch": 0.0998674580764133,
      "grad_norm": 1.7838706970214844,
      "learning_rate": 9.872614270925188e-06,
      "loss": 0.8592,
      "step": 3466
    },
    {
      "epoch": 0.09989627153806259,
      "grad_norm": 1.8173714876174927,
      "learning_rate": 9.872509594551955e-06,
      "loss": 0.8572,
      "step": 3467
    },
    {
      "epoch": 0.09992508499971187,
      "grad_norm": 1.7657098770141602,
      "learning_rate": 9.872404875744035e-06,
      "loss": 0.8879,
      "step": 3468
    },
    {
      "epoch": 0.09995389846136114,
      "grad_norm": 1.8967220783233643,
      "learning_rate": 9.872300114502338e-06,
      "loss": 0.8443,
      "step": 3469
    },
    {
      "epoch": 0.09998271192301043,
      "grad_norm": 1.8011430501937866,
      "learning_rate": 9.872195310827778e-06,
      "loss": 0.8742,
      "step": 3470
    },
    {
      "epoch": 0.10001152538465971,
      "grad_norm": 1.9080601930618286,
      "learning_rate": 9.872090464721266e-06,
      "loss": 0.8887,
      "step": 3471
    },
    {
      "epoch": 0.100040338846309,
      "grad_norm": 1.7664368152618408,
      "learning_rate": 9.871985576183719e-06,
      "loss": 0.8621,
      "step": 3472
    },
    {
      "epoch": 0.10006915230795828,
      "grad_norm": 1.8228800296783447,
      "learning_rate": 9.871880645216046e-06,
      "loss": 0.8981,
      "step": 3473
    },
    {
      "epoch": 0.10009796576960756,
      "grad_norm": 1.8676880598068237,
      "learning_rate": 9.871775671819162e-06,
      "loss": 0.8678,
      "step": 3474
    },
    {
      "epoch": 0.10012677923125685,
      "grad_norm": 1.8089056015014648,
      "learning_rate": 9.871670655993982e-06,
      "loss": 0.8735,
      "step": 3475
    },
    {
      "epoch": 0.10015559269290612,
      "grad_norm": 1.7557607889175415,
      "learning_rate": 9.871565597741422e-06,
      "loss": 0.8543,
      "step": 3476
    },
    {
      "epoch": 0.10018440615455541,
      "grad_norm": 1.8554929494857788,
      "learning_rate": 9.871460497062395e-06,
      "loss": 0.8475,
      "step": 3477
    },
    {
      "epoch": 0.10021321961620469,
      "grad_norm": 1.8188633918762207,
      "learning_rate": 9.871355353957817e-06,
      "loss": 0.8762,
      "step": 3478
    },
    {
      "epoch": 0.10024203307785397,
      "grad_norm": 1.8653684854507446,
      "learning_rate": 9.871250168428602e-06,
      "loss": 0.8758,
      "step": 3479
    },
    {
      "epoch": 0.10027084653950326,
      "grad_norm": 1.8118706941604614,
      "learning_rate": 9.871144940475667e-06,
      "loss": 0.8524,
      "step": 3480
    },
    {
      "epoch": 0.10029966000115254,
      "grad_norm": 1.6774498224258423,
      "learning_rate": 9.87103967009993e-06,
      "loss": 0.8865,
      "step": 3481
    },
    {
      "epoch": 0.10032847346280183,
      "grad_norm": 1.757202386856079,
      "learning_rate": 9.87093435730231e-06,
      "loss": 0.8463,
      "step": 3482
    },
    {
      "epoch": 0.1003572869244511,
      "grad_norm": 1.7202776670455933,
      "learning_rate": 9.870829002083718e-06,
      "loss": 0.8729,
      "step": 3483
    },
    {
      "epoch": 0.10038610038610038,
      "grad_norm": 1.6667128801345825,
      "learning_rate": 9.870723604445072e-06,
      "loss": 0.8675,
      "step": 3484
    },
    {
      "epoch": 0.10041491384774967,
      "grad_norm": 1.6806440353393555,
      "learning_rate": 9.870618164387295e-06,
      "loss": 0.8412,
      "step": 3485
    },
    {
      "epoch": 0.10044372730939895,
      "grad_norm": 1.7013185024261475,
      "learning_rate": 9.870512681911304e-06,
      "loss": 0.8621,
      "step": 3486
    },
    {
      "epoch": 0.10047254077104824,
      "grad_norm": 1.7887297868728638,
      "learning_rate": 9.870407157018013e-06,
      "loss": 0.8477,
      "step": 3487
    },
    {
      "epoch": 0.10050135423269752,
      "grad_norm": 1.6336597204208374,
      "learning_rate": 9.870301589708346e-06,
      "loss": 0.8591,
      "step": 3488
    },
    {
      "epoch": 0.10053016769434679,
      "grad_norm": 1.772367238998413,
      "learning_rate": 9.870195979983222e-06,
      "loss": 0.8586,
      "step": 3489
    },
    {
      "epoch": 0.10055898115599608,
      "grad_norm": 1.7718095779418945,
      "learning_rate": 9.870090327843558e-06,
      "loss": 0.8761,
      "step": 3490
    },
    {
      "epoch": 0.10058779461764536,
      "grad_norm": 1.7214670181274414,
      "learning_rate": 9.869984633290275e-06,
      "loss": 0.8677,
      "step": 3491
    },
    {
      "epoch": 0.10061660807929465,
      "grad_norm": 1.8102855682373047,
      "learning_rate": 9.869878896324295e-06,
      "loss": 0.8524,
      "step": 3492
    },
    {
      "epoch": 0.10064542154094393,
      "grad_norm": 1.7590163946151733,
      "learning_rate": 9.869773116946537e-06,
      "loss": 0.8686,
      "step": 3493
    },
    {
      "epoch": 0.10067423500259322,
      "grad_norm": 1.7149041891098022,
      "learning_rate": 9.869667295157924e-06,
      "loss": 0.8507,
      "step": 3494
    },
    {
      "epoch": 0.1007030484642425,
      "grad_norm": 1.700782060623169,
      "learning_rate": 9.869561430959375e-06,
      "loss": 0.8655,
      "step": 3495
    },
    {
      "epoch": 0.10073186192589177,
      "grad_norm": 1.7358636856079102,
      "learning_rate": 9.869455524351817e-06,
      "loss": 0.8499,
      "step": 3496
    },
    {
      "epoch": 0.10076067538754106,
      "grad_norm": 1.708945870399475,
      "learning_rate": 9.869349575336167e-06,
      "loss": 0.8514,
      "step": 3497
    },
    {
      "epoch": 0.10078948884919034,
      "grad_norm": 1.6849849224090576,
      "learning_rate": 9.869243583913348e-06,
      "loss": 0.8642,
      "step": 3498
    },
    {
      "epoch": 0.10081830231083963,
      "grad_norm": 1.6762475967407227,
      "learning_rate": 9.869137550084288e-06,
      "loss": 0.8677,
      "step": 3499
    },
    {
      "epoch": 0.10084711577248891,
      "grad_norm": 1.5960718393325806,
      "learning_rate": 9.869031473849905e-06,
      "loss": 0.8511,
      "step": 3500
    },
    {
      "epoch": 0.10087592923413818,
      "grad_norm": 1.7921736240386963,
      "learning_rate": 9.868925355211128e-06,
      "loss": 0.8666,
      "step": 3501
    },
    {
      "epoch": 0.10090474269578747,
      "grad_norm": 1.785818099975586,
      "learning_rate": 9.868819194168876e-06,
      "loss": 0.868,
      "step": 3502
    },
    {
      "epoch": 0.10093355615743675,
      "grad_norm": 1.77895987033844,
      "learning_rate": 9.868712990724076e-06,
      "loss": 0.8538,
      "step": 3503
    },
    {
      "epoch": 0.10096236961908604,
      "grad_norm": 1.7838454246520996,
      "learning_rate": 9.868606744877654e-06,
      "loss": 0.8733,
      "step": 3504
    },
    {
      "epoch": 0.10099118308073532,
      "grad_norm": 1.7571932077407837,
      "learning_rate": 9.868500456630533e-06,
      "loss": 0.857,
      "step": 3505
    },
    {
      "epoch": 0.1010199965423846,
      "grad_norm": 1.7693991661071777,
      "learning_rate": 9.868394125983639e-06,
      "loss": 0.8763,
      "step": 3506
    },
    {
      "epoch": 0.10104881000403389,
      "grad_norm": 1.7846581935882568,
      "learning_rate": 9.8682877529379e-06,
      "loss": 0.8578,
      "step": 3507
    },
    {
      "epoch": 0.10107762346568316,
      "grad_norm": 1.887683391571045,
      "learning_rate": 9.86818133749424e-06,
      "loss": 0.8631,
      "step": 3508
    },
    {
      "epoch": 0.10110643692733245,
      "grad_norm": 1.648751974105835,
      "learning_rate": 9.868074879653589e-06,
      "loss": 0.8554,
      "step": 3509
    },
    {
      "epoch": 0.10113525038898173,
      "grad_norm": 1.823019027709961,
      "learning_rate": 9.86796837941687e-06,
      "loss": 0.8724,
      "step": 3510
    },
    {
      "epoch": 0.10116406385063101,
      "grad_norm": 1.7538857460021973,
      "learning_rate": 9.867861836785015e-06,
      "loss": 0.869,
      "step": 3511
    },
    {
      "epoch": 0.1011928773122803,
      "grad_norm": 1.7005542516708374,
      "learning_rate": 9.867755251758947e-06,
      "loss": 0.8623,
      "step": 3512
    },
    {
      "epoch": 0.10122169077392958,
      "grad_norm": 1.6313765048980713,
      "learning_rate": 9.867648624339598e-06,
      "loss": 0.8646,
      "step": 3513
    },
    {
      "epoch": 0.10125050423557887,
      "grad_norm": 1.7355005741119385,
      "learning_rate": 9.867541954527895e-06,
      "loss": 0.8566,
      "step": 3514
    },
    {
      "epoch": 0.10127931769722814,
      "grad_norm": 1.7207400798797607,
      "learning_rate": 9.867435242324768e-06,
      "loss": 0.8602,
      "step": 3515
    },
    {
      "epoch": 0.10130813115887742,
      "grad_norm": 1.7084550857543945,
      "learning_rate": 9.867328487731145e-06,
      "loss": 0.8546,
      "step": 3516
    },
    {
      "epoch": 0.10133694462052671,
      "grad_norm": 1.622599720954895,
      "learning_rate": 9.867221690747957e-06,
      "loss": 0.8691,
      "step": 3517
    },
    {
      "epoch": 0.10136575808217599,
      "grad_norm": 1.7157702445983887,
      "learning_rate": 9.867114851376131e-06,
      "loss": 0.8404,
      "step": 3518
    },
    {
      "epoch": 0.10139457154382528,
      "grad_norm": 1.643898844718933,
      "learning_rate": 9.867007969616604e-06,
      "loss": 0.8465,
      "step": 3519
    },
    {
      "epoch": 0.10142338500547456,
      "grad_norm": 1.788000226020813,
      "learning_rate": 9.8669010454703e-06,
      "loss": 0.8472,
      "step": 3520
    },
    {
      "epoch": 0.10145219846712385,
      "grad_norm": 1.5703120231628418,
      "learning_rate": 9.866794078938155e-06,
      "loss": 0.8578,
      "step": 3521
    },
    {
      "epoch": 0.10148101192877312,
      "grad_norm": 1.7748278379440308,
      "learning_rate": 9.866687070021098e-06,
      "loss": 0.8492,
      "step": 3522
    },
    {
      "epoch": 0.1015098253904224,
      "grad_norm": 1.7913118600845337,
      "learning_rate": 9.86658001872006e-06,
      "loss": 0.8677,
      "step": 3523
    },
    {
      "epoch": 0.10153863885207169,
      "grad_norm": 1.6881659030914307,
      "learning_rate": 9.866472925035975e-06,
      "loss": 0.8602,
      "step": 3524
    },
    {
      "epoch": 0.10156745231372097,
      "grad_norm": 1.681765079498291,
      "learning_rate": 9.866365788969776e-06,
      "loss": 0.8593,
      "step": 3525
    },
    {
      "epoch": 0.10159626577537026,
      "grad_norm": 1.694248914718628,
      "learning_rate": 9.866258610522396e-06,
      "loss": 0.845,
      "step": 3526
    },
    {
      "epoch": 0.10162507923701954,
      "grad_norm": 1.7573050260543823,
      "learning_rate": 9.86615138969477e-06,
      "loss": 0.8344,
      "step": 3527
    },
    {
      "epoch": 0.10165389269866881,
      "grad_norm": 1.6975595951080322,
      "learning_rate": 9.866044126487825e-06,
      "loss": 0.8416,
      "step": 3528
    },
    {
      "epoch": 0.1016827061603181,
      "grad_norm": 1.7508665323257446,
      "learning_rate": 9.865936820902504e-06,
      "loss": 0.8441,
      "step": 3529
    },
    {
      "epoch": 0.10171151962196738,
      "grad_norm": 1.676964282989502,
      "learning_rate": 9.865829472939737e-06,
      "loss": 0.8604,
      "step": 3530
    },
    {
      "epoch": 0.10174033308361667,
      "grad_norm": 1.7461318969726562,
      "learning_rate": 9.865722082600457e-06,
      "loss": 0.8483,
      "step": 3531
    },
    {
      "epoch": 0.10176914654526595,
      "grad_norm": 1.7968344688415527,
      "learning_rate": 9.865614649885603e-06,
      "loss": 0.8455,
      "step": 3532
    },
    {
      "epoch": 0.10179796000691523,
      "grad_norm": 1.708844542503357,
      "learning_rate": 9.865507174796109e-06,
      "loss": 0.8777,
      "step": 3533
    },
    {
      "epoch": 0.10182677346856452,
      "grad_norm": 1.7062216997146606,
      "learning_rate": 9.865399657332913e-06,
      "loss": 0.8442,
      "step": 3534
    },
    {
      "epoch": 0.10185558693021379,
      "grad_norm": 1.630431890487671,
      "learning_rate": 9.865292097496946e-06,
      "loss": 0.8526,
      "step": 3535
    },
    {
      "epoch": 0.10188440039186308,
      "grad_norm": 1.636920690536499,
      "learning_rate": 9.865184495289151e-06,
      "loss": 0.8519,
      "step": 3536
    },
    {
      "epoch": 0.10191321385351236,
      "grad_norm": 1.7650922536849976,
      "learning_rate": 9.865076850710462e-06,
      "loss": 0.8248,
      "step": 3537
    },
    {
      "epoch": 0.10194202731516164,
      "grad_norm": 1.7083040475845337,
      "learning_rate": 9.864969163761818e-06,
      "loss": 0.8426,
      "step": 3538
    },
    {
      "epoch": 0.10197084077681093,
      "grad_norm": 1.6558364629745483,
      "learning_rate": 9.864861434444153e-06,
      "loss": 0.8398,
      "step": 3539
    },
    {
      "epoch": 0.1019996542384602,
      "grad_norm": 1.7408392429351807,
      "learning_rate": 9.864753662758411e-06,
      "loss": 0.8868,
      "step": 3540
    },
    {
      "epoch": 0.1020284677001095,
      "grad_norm": 1.7402812242507935,
      "learning_rate": 9.864645848705526e-06,
      "loss": 0.8397,
      "step": 3541
    },
    {
      "epoch": 0.10205728116175877,
      "grad_norm": 1.724117636680603,
      "learning_rate": 9.864537992286438e-06,
      "loss": 0.8555,
      "step": 3542
    },
    {
      "epoch": 0.10208609462340805,
      "grad_norm": 1.721794843673706,
      "learning_rate": 9.864430093502087e-06,
      "loss": 0.8766,
      "step": 3543
    },
    {
      "epoch": 0.10211490808505734,
      "grad_norm": 1.8063048124313354,
      "learning_rate": 9.864322152353414e-06,
      "loss": 0.8659,
      "step": 3544
    },
    {
      "epoch": 0.10214372154670662,
      "grad_norm": 1.6327346563339233,
      "learning_rate": 9.864214168841355e-06,
      "loss": 0.8381,
      "step": 3545
    },
    {
      "epoch": 0.10217253500835591,
      "grad_norm": 1.7325993776321411,
      "learning_rate": 9.864106142966857e-06,
      "loss": 0.8746,
      "step": 3546
    },
    {
      "epoch": 0.10220134847000518,
      "grad_norm": 1.6883794069290161,
      "learning_rate": 9.863998074730853e-06,
      "loss": 0.8412,
      "step": 3547
    },
    {
      "epoch": 0.10223016193165448,
      "grad_norm": 1.802162766456604,
      "learning_rate": 9.86388996413429e-06,
      "loss": 0.8689,
      "step": 3548
    },
    {
      "epoch": 0.10225897539330375,
      "grad_norm": 1.6085032224655151,
      "learning_rate": 9.863781811178105e-06,
      "loss": 0.8574,
      "step": 3549
    },
    {
      "epoch": 0.10228778885495303,
      "grad_norm": 1.5453455448150635,
      "learning_rate": 9.863673615863246e-06,
      "loss": 0.8622,
      "step": 3550
    },
    {
      "epoch": 0.10231660231660232,
      "grad_norm": 1.6633617877960205,
      "learning_rate": 9.863565378190649e-06,
      "loss": 0.8424,
      "step": 3551
    },
    {
      "epoch": 0.1023454157782516,
      "grad_norm": 1.6959691047668457,
      "learning_rate": 9.863457098161261e-06,
      "loss": 0.8483,
      "step": 3552
    },
    {
      "epoch": 0.10237422923990089,
      "grad_norm": 1.6787033081054688,
      "learning_rate": 9.863348775776023e-06,
      "loss": 0.8405,
      "step": 3553
    },
    {
      "epoch": 0.10240304270155016,
      "grad_norm": 1.7118746042251587,
      "learning_rate": 9.863240411035879e-06,
      "loss": 0.8426,
      "step": 3554
    },
    {
      "epoch": 0.10243185616319944,
      "grad_norm": 1.6686999797821045,
      "learning_rate": 9.86313200394177e-06,
      "loss": 0.8735,
      "step": 3555
    },
    {
      "epoch": 0.10246066962484873,
      "grad_norm": 1.7959420680999756,
      "learning_rate": 9.863023554494647e-06,
      "loss": 0.8654,
      "step": 3556
    },
    {
      "epoch": 0.10248948308649801,
      "grad_norm": 1.6631029844284058,
      "learning_rate": 9.862915062695447e-06,
      "loss": 0.8608,
      "step": 3557
    },
    {
      "epoch": 0.1025182965481473,
      "grad_norm": 1.7495498657226562,
      "learning_rate": 9.862806528545118e-06,
      "loss": 0.8699,
      "step": 3558
    },
    {
      "epoch": 0.10254711000979658,
      "grad_norm": 1.6688908338546753,
      "learning_rate": 9.862697952044607e-06,
      "loss": 0.8399,
      "step": 3559
    },
    {
      "epoch": 0.10257592347144585,
      "grad_norm": 1.7440887689590454,
      "learning_rate": 9.862589333194854e-06,
      "loss": 0.852,
      "step": 3560
    },
    {
      "epoch": 0.10260473693309514,
      "grad_norm": 1.6790822744369507,
      "learning_rate": 9.862480671996812e-06,
      "loss": 0.8515,
      "step": 3561
    },
    {
      "epoch": 0.10263355039474442,
      "grad_norm": 1.5963499546051025,
      "learning_rate": 9.862371968451423e-06,
      "loss": 0.8336,
      "step": 3562
    },
    {
      "epoch": 0.10266236385639371,
      "grad_norm": 1.6947143077850342,
      "learning_rate": 9.862263222559636e-06,
      "loss": 0.8521,
      "step": 3563
    },
    {
      "epoch": 0.10269117731804299,
      "grad_norm": 1.5922499895095825,
      "learning_rate": 9.862154434322396e-06,
      "loss": 0.839,
      "step": 3564
    },
    {
      "epoch": 0.10271999077969227,
      "grad_norm": 1.735298991203308,
      "learning_rate": 9.86204560374065e-06,
      "loss": 0.8303,
      "step": 3565
    },
    {
      "epoch": 0.10274880424134156,
      "grad_norm": 1.6065977811813354,
      "learning_rate": 9.86193673081535e-06,
      "loss": 0.8606,
      "step": 3566
    },
    {
      "epoch": 0.10277761770299083,
      "grad_norm": 1.6784800291061401,
      "learning_rate": 9.861827815547438e-06,
      "loss": 0.8233,
      "step": 3567
    },
    {
      "epoch": 0.10280643116464012,
      "grad_norm": 1.6551679372787476,
      "learning_rate": 9.861718857937868e-06,
      "loss": 0.8411,
      "step": 3568
    },
    {
      "epoch": 0.1028352446262894,
      "grad_norm": 1.7678154706954956,
      "learning_rate": 9.861609857987586e-06,
      "loss": 0.879,
      "step": 3569
    },
    {
      "epoch": 0.10286405808793868,
      "grad_norm": 1.8133032321929932,
      "learning_rate": 9.861500815697541e-06,
      "loss": 0.8336,
      "step": 3570
    },
    {
      "epoch": 0.10289287154958797,
      "grad_norm": 1.7413609027862549,
      "learning_rate": 9.861391731068684e-06,
      "loss": 0.8513,
      "step": 3571
    },
    {
      "epoch": 0.10292168501123725,
      "grad_norm": 1.7869902849197388,
      "learning_rate": 9.861282604101966e-06,
      "loss": 0.8243,
      "step": 3572
    },
    {
      "epoch": 0.10295049847288654,
      "grad_norm": 1.7807890176773071,
      "learning_rate": 9.861173434798335e-06,
      "loss": 0.8401,
      "step": 3573
    },
    {
      "epoch": 0.10297931193453581,
      "grad_norm": 1.7546985149383545,
      "learning_rate": 9.861064223158742e-06,
      "loss": 0.8589,
      "step": 3574
    },
    {
      "epoch": 0.1030081253961851,
      "grad_norm": 1.831992506980896,
      "learning_rate": 9.86095496918414e-06,
      "loss": 0.8748,
      "step": 3575
    },
    {
      "epoch": 0.10303693885783438,
      "grad_norm": 1.7333258390426636,
      "learning_rate": 9.860845672875479e-06,
      "loss": 0.8649,
      "step": 3576
    },
    {
      "epoch": 0.10306575231948366,
      "grad_norm": 1.7324597835540771,
      "learning_rate": 9.860736334233711e-06,
      "loss": 0.8551,
      "step": 3577
    },
    {
      "epoch": 0.10309456578113295,
      "grad_norm": 1.870054006576538,
      "learning_rate": 9.860626953259791e-06,
      "loss": 0.8327,
      "step": 3578
    },
    {
      "epoch": 0.10312337924278223,
      "grad_norm": 1.7475676536560059,
      "learning_rate": 9.860517529954666e-06,
      "loss": 0.8569,
      "step": 3579
    },
    {
      "epoch": 0.10315219270443152,
      "grad_norm": 1.645728349685669,
      "learning_rate": 9.860408064319294e-06,
      "loss": 0.8564,
      "step": 3580
    },
    {
      "epoch": 0.1031810061660808,
      "grad_norm": 1.764669418334961,
      "learning_rate": 9.860298556354625e-06,
      "loss": 0.8492,
      "step": 3581
    },
    {
      "epoch": 0.10320981962773007,
      "grad_norm": 1.6458840370178223,
      "learning_rate": 9.860189006061614e-06,
      "loss": 0.8381,
      "step": 3582
    },
    {
      "epoch": 0.10323863308937936,
      "grad_norm": 1.784116268157959,
      "learning_rate": 9.860079413441217e-06,
      "loss": 0.8695,
      "step": 3583
    },
    {
      "epoch": 0.10326744655102864,
      "grad_norm": 1.7007182836532593,
      "learning_rate": 9.859969778494384e-06,
      "loss": 0.8143,
      "step": 3584
    },
    {
      "epoch": 0.10329626001267793,
      "grad_norm": 1.6422739028930664,
      "learning_rate": 9.859860101222074e-06,
      "loss": 0.8135,
      "step": 3585
    },
    {
      "epoch": 0.1033250734743272,
      "grad_norm": 1.6470646858215332,
      "learning_rate": 9.859750381625241e-06,
      "loss": 0.8357,
      "step": 3586
    },
    {
      "epoch": 0.10335388693597648,
      "grad_norm": 1.7495673894882202,
      "learning_rate": 9.85964061970484e-06,
      "loss": 0.8368,
      "step": 3587
    },
    {
      "epoch": 0.10338270039762577,
      "grad_norm": 1.6548418998718262,
      "learning_rate": 9.859530815461825e-06,
      "loss": 0.8291,
      "step": 3588
    },
    {
      "epoch": 0.10341151385927505,
      "grad_norm": 1.5976358652114868,
      "learning_rate": 9.859420968897157e-06,
      "loss": 0.8505,
      "step": 3589
    },
    {
      "epoch": 0.10344032732092434,
      "grad_norm": 1.5638999938964844,
      "learning_rate": 9.859311080011788e-06,
      "loss": 0.871,
      "step": 3590
    },
    {
      "epoch": 0.10346914078257362,
      "grad_norm": 1.6692031621932983,
      "learning_rate": 9.859201148806679e-06,
      "loss": 0.866,
      "step": 3591
    },
    {
      "epoch": 0.1034979542442229,
      "grad_norm": 1.6360859870910645,
      "learning_rate": 9.859091175282783e-06,
      "loss": 0.8254,
      "step": 3592
    },
    {
      "epoch": 0.10352676770587219,
      "grad_norm": 1.6126335859298706,
      "learning_rate": 9.858981159441062e-06,
      "loss": 0.8661,
      "step": 3593
    },
    {
      "epoch": 0.10355558116752146,
      "grad_norm": 1.6843229532241821,
      "learning_rate": 9.858871101282473e-06,
      "loss": 0.8524,
      "step": 3594
    },
    {
      "epoch": 0.10358439462917075,
      "grad_norm": 1.6106549501419067,
      "learning_rate": 9.85876100080797e-06,
      "loss": 0.8731,
      "step": 3595
    },
    {
      "epoch": 0.10361320809082003,
      "grad_norm": 1.7372921705245972,
      "learning_rate": 9.85865085801852e-06,
      "loss": 0.8614,
      "step": 3596
    },
    {
      "epoch": 0.10364202155246931,
      "grad_norm": 1.6721168756484985,
      "learning_rate": 9.858540672915076e-06,
      "loss": 0.8427,
      "step": 3597
    },
    {
      "epoch": 0.1036708350141186,
      "grad_norm": 1.7288726568222046,
      "learning_rate": 9.858430445498599e-06,
      "loss": 0.8785,
      "step": 3598
    },
    {
      "epoch": 0.10369964847576787,
      "grad_norm": 1.73403000831604,
      "learning_rate": 9.858320175770051e-06,
      "loss": 0.8784,
      "step": 3599
    },
    {
      "epoch": 0.10372846193741717,
      "grad_norm": 1.68582284450531,
      "learning_rate": 9.85820986373039e-06,
      "loss": 0.8685,
      "step": 3600
    },
    {
      "epoch": 0.10375727539906644,
      "grad_norm": 1.6927789449691772,
      "learning_rate": 9.858099509380578e-06,
      "loss": 0.8281,
      "step": 3601
    },
    {
      "epoch": 0.10378608886071572,
      "grad_norm": 1.7771344184875488,
      "learning_rate": 9.857989112721574e-06,
      "loss": 0.8595,
      "step": 3602
    },
    {
      "epoch": 0.10381490232236501,
      "grad_norm": 1.597956657409668,
      "learning_rate": 9.857878673754342e-06,
      "loss": 0.8533,
      "step": 3603
    },
    {
      "epoch": 0.10384371578401429,
      "grad_norm": 1.683671474456787,
      "learning_rate": 9.857768192479845e-06,
      "loss": 0.8397,
      "step": 3604
    },
    {
      "epoch": 0.10387252924566358,
      "grad_norm": 1.7588895559310913,
      "learning_rate": 9.857657668899038e-06,
      "loss": 0.862,
      "step": 3605
    },
    {
      "epoch": 0.10390134270731285,
      "grad_norm": 1.6074150800704956,
      "learning_rate": 9.857547103012893e-06,
      "loss": 0.8529,
      "step": 3606
    },
    {
      "epoch": 0.10393015616896215,
      "grad_norm": 1.7516224384307861,
      "learning_rate": 9.857436494822368e-06,
      "loss": 0.8494,
      "step": 3607
    },
    {
      "epoch": 0.10395896963061142,
      "grad_norm": 1.760820984840393,
      "learning_rate": 9.857325844328425e-06,
      "loss": 0.8697,
      "step": 3608
    },
    {
      "epoch": 0.1039877830922607,
      "grad_norm": 1.64093816280365,
      "learning_rate": 9.857215151532031e-06,
      "loss": 0.8488,
      "step": 3609
    },
    {
      "epoch": 0.10401659655390999,
      "grad_norm": 1.648193359375,
      "learning_rate": 9.857104416434147e-06,
      "loss": 0.8522,
      "step": 3610
    },
    {
      "epoch": 0.10404541001555927,
      "grad_norm": 1.6562930345535278,
      "learning_rate": 9.85699363903574e-06,
      "loss": 0.8607,
      "step": 3611
    },
    {
      "epoch": 0.10407422347720856,
      "grad_norm": 1.5650030374526978,
      "learning_rate": 9.856882819337775e-06,
      "loss": 0.8329,
      "step": 3612
    },
    {
      "epoch": 0.10410303693885783,
      "grad_norm": 1.6273261308670044,
      "learning_rate": 9.856771957341212e-06,
      "loss": 0.8422,
      "step": 3613
    },
    {
      "epoch": 0.10413185040050711,
      "grad_norm": 1.5779948234558105,
      "learning_rate": 9.856661053047022e-06,
      "loss": 0.832,
      "step": 3614
    },
    {
      "epoch": 0.1041606638621564,
      "grad_norm": 1.6620999574661255,
      "learning_rate": 9.856550106456171e-06,
      "loss": 0.8321,
      "step": 3615
    },
    {
      "epoch": 0.10418947732380568,
      "grad_norm": 1.7151002883911133,
      "learning_rate": 9.856439117569621e-06,
      "loss": 0.8761,
      "step": 3616
    },
    {
      "epoch": 0.10421829078545497,
      "grad_norm": 1.631728172302246,
      "learning_rate": 9.856328086388342e-06,
      "loss": 0.8794,
      "step": 3617
    },
    {
      "epoch": 0.10424710424710425,
      "grad_norm": 1.7189613580703735,
      "learning_rate": 9.856217012913299e-06,
      "loss": 0.8255,
      "step": 3618
    },
    {
      "epoch": 0.10427591770875352,
      "grad_norm": 1.5654942989349365,
      "learning_rate": 9.85610589714546e-06,
      "loss": 0.8388,
      "step": 3619
    },
    {
      "epoch": 0.10430473117040281,
      "grad_norm": 1.6075128316879272,
      "learning_rate": 9.855994739085795e-06,
      "loss": 0.8319,
      "step": 3620
    },
    {
      "epoch": 0.10433354463205209,
      "grad_norm": 1.722394347190857,
      "learning_rate": 9.855883538735269e-06,
      "loss": 0.8603,
      "step": 3621
    },
    {
      "epoch": 0.10436235809370138,
      "grad_norm": 1.70301353931427,
      "learning_rate": 9.85577229609485e-06,
      "loss": 0.8562,
      "step": 3622
    },
    {
      "epoch": 0.10439117155535066,
      "grad_norm": 1.6747721433639526,
      "learning_rate": 9.855661011165512e-06,
      "loss": 0.8468,
      "step": 3623
    },
    {
      "epoch": 0.10441998501699994,
      "grad_norm": 1.664692759513855,
      "learning_rate": 9.855549683948217e-06,
      "loss": 0.8701,
      "step": 3624
    },
    {
      "epoch": 0.10444879847864923,
      "grad_norm": 1.6694267988204956,
      "learning_rate": 9.855438314443939e-06,
      "loss": 0.853,
      "step": 3625
    },
    {
      "epoch": 0.1044776119402985,
      "grad_norm": 1.7291373014450073,
      "learning_rate": 9.855326902653648e-06,
      "loss": 0.8487,
      "step": 3626
    },
    {
      "epoch": 0.1045064254019478,
      "grad_norm": 1.690308928489685,
      "learning_rate": 9.855215448578311e-06,
      "loss": 0.8677,
      "step": 3627
    },
    {
      "epoch": 0.10453523886359707,
      "grad_norm": 1.7892675399780273,
      "learning_rate": 9.855103952218903e-06,
      "loss": 0.8678,
      "step": 3628
    },
    {
      "epoch": 0.10456405232524635,
      "grad_norm": 1.871652364730835,
      "learning_rate": 9.854992413576392e-06,
      "loss": 0.8538,
      "step": 3629
    },
    {
      "epoch": 0.10459286578689564,
      "grad_norm": 1.5681389570236206,
      "learning_rate": 9.85488083265175e-06,
      "loss": 0.8367,
      "step": 3630
    },
    {
      "epoch": 0.10462167924854492,
      "grad_norm": 1.826711893081665,
      "learning_rate": 9.85476920944595e-06,
      "loss": 0.8541,
      "step": 3631
    },
    {
      "epoch": 0.1046504927101942,
      "grad_norm": 1.7160547971725464,
      "learning_rate": 9.854657543959962e-06,
      "loss": 0.8461,
      "step": 3632
    },
    {
      "epoch": 0.10467930617184348,
      "grad_norm": 1.6516985893249512,
      "learning_rate": 9.85454583619476e-06,
      "loss": 0.8446,
      "step": 3633
    },
    {
      "epoch": 0.10470811963349277,
      "grad_norm": 1.6897001266479492,
      "learning_rate": 9.854434086151318e-06,
      "loss": 0.8182,
      "step": 3634
    },
    {
      "epoch": 0.10473693309514205,
      "grad_norm": 1.590396523475647,
      "learning_rate": 9.854322293830607e-06,
      "loss": 0.848,
      "step": 3635
    },
    {
      "epoch": 0.10476574655679133,
      "grad_norm": 1.7552076578140259,
      "learning_rate": 9.8542104592336e-06,
      "loss": 0.8336,
      "step": 3636
    },
    {
      "epoch": 0.10479456001844062,
      "grad_norm": 1.6270533800125122,
      "learning_rate": 9.854098582361273e-06,
      "loss": 0.8342,
      "step": 3637
    },
    {
      "epoch": 0.1048233734800899,
      "grad_norm": 1.7119321823120117,
      "learning_rate": 9.853986663214601e-06,
      "loss": 0.8202,
      "step": 3638
    },
    {
      "epoch": 0.10485218694173919,
      "grad_norm": 1.737391710281372,
      "learning_rate": 9.853874701794555e-06,
      "loss": 0.8523,
      "step": 3639
    },
    {
      "epoch": 0.10488100040338846,
      "grad_norm": 1.669867992401123,
      "learning_rate": 9.853762698102114e-06,
      "loss": 0.845,
      "step": 3640
    },
    {
      "epoch": 0.10490981386503774,
      "grad_norm": 1.6705471277236938,
      "learning_rate": 9.853650652138252e-06,
      "loss": 0.8594,
      "step": 3641
    },
    {
      "epoch": 0.10493862732668703,
      "grad_norm": 1.5823737382888794,
      "learning_rate": 9.853538563903945e-06,
      "loss": 0.8486,
      "step": 3642
    },
    {
      "epoch": 0.10496744078833631,
      "grad_norm": 1.7363685369491577,
      "learning_rate": 9.853426433400168e-06,
      "loss": 0.8173,
      "step": 3643
    },
    {
      "epoch": 0.1049962542499856,
      "grad_norm": 1.8041306734085083,
      "learning_rate": 9.853314260627899e-06,
      "loss": 0.8581,
      "step": 3644
    },
    {
      "epoch": 0.10502506771163488,
      "grad_norm": 1.6110655069351196,
      "learning_rate": 9.853202045588114e-06,
      "loss": 0.8727,
      "step": 3645
    },
    {
      "epoch": 0.10505388117328415,
      "grad_norm": 1.6064043045043945,
      "learning_rate": 9.85308978828179e-06,
      "loss": 0.8587,
      "step": 3646
    },
    {
      "epoch": 0.10508269463493344,
      "grad_norm": 1.5903584957122803,
      "learning_rate": 9.852977488709907e-06,
      "loss": 0.8084,
      "step": 3647
    },
    {
      "epoch": 0.10511150809658272,
      "grad_norm": 1.6907477378845215,
      "learning_rate": 9.852865146873439e-06,
      "loss": 0.8471,
      "step": 3648
    },
    {
      "epoch": 0.10514032155823201,
      "grad_norm": 1.5756006240844727,
      "learning_rate": 9.852752762773368e-06,
      "loss": 0.8403,
      "step": 3649
    },
    {
      "epoch": 0.10516913501988129,
      "grad_norm": 1.584954023361206,
      "learning_rate": 9.852640336410672e-06,
      "loss": 0.816,
      "step": 3650
    },
    {
      "epoch": 0.10519794848153056,
      "grad_norm": 1.6741538047790527,
      "learning_rate": 9.852527867786328e-06,
      "loss": 0.8389,
      "step": 3651
    },
    {
      "epoch": 0.10522676194317986,
      "grad_norm": 1.7912132740020752,
      "learning_rate": 9.852415356901317e-06,
      "loss": 0.8417,
      "step": 3652
    },
    {
      "epoch": 0.10525557540482913,
      "grad_norm": 1.6231318712234497,
      "learning_rate": 9.85230280375662e-06,
      "loss": 0.839,
      "step": 3653
    },
    {
      "epoch": 0.10528438886647842,
      "grad_norm": 1.5536706447601318,
      "learning_rate": 9.852190208353216e-06,
      "loss": 0.8383,
      "step": 3654
    },
    {
      "epoch": 0.1053132023281277,
      "grad_norm": 1.600934624671936,
      "learning_rate": 9.852077570692086e-06,
      "loss": 0.8154,
      "step": 3655
    },
    {
      "epoch": 0.10534201578977698,
      "grad_norm": 1.7473340034484863,
      "learning_rate": 9.851964890774211e-06,
      "loss": 0.865,
      "step": 3656
    },
    {
      "epoch": 0.10537082925142627,
      "grad_norm": 1.6563040018081665,
      "learning_rate": 9.851852168600572e-06,
      "loss": 0.8518,
      "step": 3657
    },
    {
      "epoch": 0.10539964271307554,
      "grad_norm": 1.7137746810913086,
      "learning_rate": 9.851739404172148e-06,
      "loss": 0.8317,
      "step": 3658
    },
    {
      "epoch": 0.10542845617472484,
      "grad_norm": 1.620337724685669,
      "learning_rate": 9.851626597489924e-06,
      "loss": 0.8559,
      "step": 3659
    },
    {
      "epoch": 0.10545726963637411,
      "grad_norm": 1.7269304990768433,
      "learning_rate": 9.851513748554885e-06,
      "loss": 0.8475,
      "step": 3660
    },
    {
      "epoch": 0.1054860830980234,
      "grad_norm": 1.8032379150390625,
      "learning_rate": 9.85140085736801e-06,
      "loss": 0.8483,
      "step": 3661
    },
    {
      "epoch": 0.10551489655967268,
      "grad_norm": 1.682546615600586,
      "learning_rate": 9.851287923930282e-06,
      "loss": 0.8475,
      "step": 3662
    },
    {
      "epoch": 0.10554371002132196,
      "grad_norm": 1.6893106698989868,
      "learning_rate": 9.851174948242688e-06,
      "loss": 0.8033,
      "step": 3663
    },
    {
      "epoch": 0.10557252348297125,
      "grad_norm": 1.6527252197265625,
      "learning_rate": 9.851061930306208e-06,
      "loss": 0.8192,
      "step": 3664
    },
    {
      "epoch": 0.10560133694462052,
      "grad_norm": 1.5835292339324951,
      "learning_rate": 9.850948870121828e-06,
      "loss": 0.8512,
      "step": 3665
    },
    {
      "epoch": 0.10563015040626982,
      "grad_norm": 1.749461054801941,
      "learning_rate": 9.850835767690532e-06,
      "loss": 0.8241,
      "step": 3666
    },
    {
      "epoch": 0.10565896386791909,
      "grad_norm": 1.6832869052886963,
      "learning_rate": 9.850722623013306e-06,
      "loss": 0.8576,
      "step": 3667
    },
    {
      "epoch": 0.10568777732956837,
      "grad_norm": 1.5972836017608643,
      "learning_rate": 9.850609436091135e-06,
      "loss": 0.8295,
      "step": 3668
    },
    {
      "epoch": 0.10571659079121766,
      "grad_norm": 1.667191743850708,
      "learning_rate": 9.850496206925005e-06,
      "loss": 0.8206,
      "step": 3669
    },
    {
      "epoch": 0.10574540425286694,
      "grad_norm": 1.7368297576904297,
      "learning_rate": 9.8503829355159e-06,
      "loss": 0.8655,
      "step": 3670
    },
    {
      "epoch": 0.10577421771451623,
      "grad_norm": 1.6395618915557861,
      "learning_rate": 9.850269621864812e-06,
      "loss": 0.8484,
      "step": 3671
    },
    {
      "epoch": 0.1058030311761655,
      "grad_norm": 1.724815845489502,
      "learning_rate": 9.850156265972722e-06,
      "loss": 0.8322,
      "step": 3672
    },
    {
      "epoch": 0.10583184463781478,
      "grad_norm": 1.6665161848068237,
      "learning_rate": 9.850042867840619e-06,
      "loss": 0.8261,
      "step": 3673
    },
    {
      "epoch": 0.10586065809946407,
      "grad_norm": 1.5707154273986816,
      "learning_rate": 9.84992942746949e-06,
      "loss": 0.8329,
      "step": 3674
    },
    {
      "epoch": 0.10588947156111335,
      "grad_norm": 1.630228042602539,
      "learning_rate": 9.849815944860324e-06,
      "loss": 0.84,
      "step": 3675
    },
    {
      "epoch": 0.10591828502276264,
      "grad_norm": 1.5516172647476196,
      "learning_rate": 9.849702420014111e-06,
      "loss": 0.827,
      "step": 3676
    },
    {
      "epoch": 0.10594709848441192,
      "grad_norm": 1.5345747470855713,
      "learning_rate": 9.849588852931836e-06,
      "loss": 0.8307,
      "step": 3677
    },
    {
      "epoch": 0.1059759119460612,
      "grad_norm": 1.6132009029388428,
      "learning_rate": 9.84947524361449e-06,
      "loss": 0.8586,
      "step": 3678
    },
    {
      "epoch": 0.10600472540771048,
      "grad_norm": 1.6226160526275635,
      "learning_rate": 9.849361592063065e-06,
      "loss": 0.8386,
      "step": 3679
    },
    {
      "epoch": 0.10603353886935976,
      "grad_norm": 1.5412073135375977,
      "learning_rate": 9.849247898278545e-06,
      "loss": 0.8107,
      "step": 3680
    },
    {
      "epoch": 0.10606235233100905,
      "grad_norm": 1.5453284978866577,
      "learning_rate": 9.849134162261924e-06,
      "loss": 0.8369,
      "step": 3681
    },
    {
      "epoch": 0.10609116579265833,
      "grad_norm": 1.6367723941802979,
      "learning_rate": 9.849020384014192e-06,
      "loss": 0.8487,
      "step": 3682
    },
    {
      "epoch": 0.1061199792543076,
      "grad_norm": 1.6244038343429565,
      "learning_rate": 9.84890656353634e-06,
      "loss": 0.8618,
      "step": 3683
    },
    {
      "epoch": 0.1061487927159569,
      "grad_norm": 1.5752369165420532,
      "learning_rate": 9.84879270082936e-06,
      "loss": 0.8547,
      "step": 3684
    },
    {
      "epoch": 0.10617760617760617,
      "grad_norm": 1.6061689853668213,
      "learning_rate": 9.84867879589424e-06,
      "loss": 0.8245,
      "step": 3685
    },
    {
      "epoch": 0.10620641963925546,
      "grad_norm": 1.4644408226013184,
      "learning_rate": 9.848564848731977e-06,
      "loss": 0.8364,
      "step": 3686
    },
    {
      "epoch": 0.10623523310090474,
      "grad_norm": 1.6833739280700684,
      "learning_rate": 9.84845085934356e-06,
      "loss": 0.8239,
      "step": 3687
    },
    {
      "epoch": 0.10626404656255403,
      "grad_norm": 1.6750632524490356,
      "learning_rate": 9.848336827729983e-06,
      "loss": 0.8534,
      "step": 3688
    },
    {
      "epoch": 0.10629286002420331,
      "grad_norm": 1.6300230026245117,
      "learning_rate": 9.848222753892239e-06,
      "loss": 0.8431,
      "step": 3689
    },
    {
      "epoch": 0.10632167348585259,
      "grad_norm": 1.5330699682235718,
      "learning_rate": 9.84810863783132e-06,
      "loss": 0.8237,
      "step": 3690
    },
    {
      "epoch": 0.10635048694750188,
      "grad_norm": 1.6745483875274658,
      "learning_rate": 9.847994479548221e-06,
      "loss": 0.8272,
      "step": 3691
    },
    {
      "epoch": 0.10637930040915115,
      "grad_norm": 1.5470422506332397,
      "learning_rate": 9.847880279043937e-06,
      "loss": 0.8629,
      "step": 3692
    },
    {
      "epoch": 0.10640811387080044,
      "grad_norm": 1.539513349533081,
      "learning_rate": 9.847766036319464e-06,
      "loss": 0.8429,
      "step": 3693
    },
    {
      "epoch": 0.10643692733244972,
      "grad_norm": 1.6192734241485596,
      "learning_rate": 9.847651751375794e-06,
      "loss": 0.8563,
      "step": 3694
    },
    {
      "epoch": 0.106465740794099,
      "grad_norm": 1.5180325508117676,
      "learning_rate": 9.84753742421392e-06,
      "loss": 0.8426,
      "step": 3695
    },
    {
      "epoch": 0.10649455425574829,
      "grad_norm": 1.5091744661331177,
      "learning_rate": 9.847423054834844e-06,
      "loss": 0.8269,
      "step": 3696
    },
    {
      "epoch": 0.10652336771739757,
      "grad_norm": 1.6074687242507935,
      "learning_rate": 9.847308643239558e-06,
      "loss": 0.8442,
      "step": 3697
    },
    {
      "epoch": 0.10655218117904686,
      "grad_norm": 1.5677977800369263,
      "learning_rate": 9.847194189429057e-06,
      "loss": 0.8543,
      "step": 3698
    },
    {
      "epoch": 0.10658099464069613,
      "grad_norm": 1.566673755645752,
      "learning_rate": 9.847079693404342e-06,
      "loss": 0.8588,
      "step": 3699
    },
    {
      "epoch": 0.10660980810234541,
      "grad_norm": 1.5811363458633423,
      "learning_rate": 9.846965155166409e-06,
      "loss": 0.8562,
      "step": 3700
    },
    {
      "epoch": 0.1066386215639947,
      "grad_norm": 1.4800628423690796,
      "learning_rate": 9.846850574716253e-06,
      "loss": 0.8396,
      "step": 3701
    },
    {
      "epoch": 0.10666743502564398,
      "grad_norm": 1.4846380949020386,
      "learning_rate": 9.846735952054876e-06,
      "loss": 0.8195,
      "step": 3702
    },
    {
      "epoch": 0.10669624848729327,
      "grad_norm": 1.568955659866333,
      "learning_rate": 9.84662128718327e-06,
      "loss": 0.8456,
      "step": 3703
    },
    {
      "epoch": 0.10672506194894255,
      "grad_norm": 1.5582911968231201,
      "learning_rate": 9.846506580102441e-06,
      "loss": 0.8288,
      "step": 3704
    },
    {
      "epoch": 0.10675387541059182,
      "grad_norm": 1.5950886011123657,
      "learning_rate": 9.846391830813383e-06,
      "loss": 0.8727,
      "step": 3705
    },
    {
      "epoch": 0.10678268887224111,
      "grad_norm": 1.5372105836868286,
      "learning_rate": 9.846277039317096e-06,
      "loss": 0.8418,
      "step": 3706
    },
    {
      "epoch": 0.10681150233389039,
      "grad_norm": 1.6352959871292114,
      "learning_rate": 9.846162205614582e-06,
      "loss": 0.8792,
      "step": 3707
    },
    {
      "epoch": 0.10684031579553968,
      "grad_norm": 1.5756255388259888,
      "learning_rate": 9.846047329706838e-06,
      "loss": 0.8392,
      "step": 3708
    },
    {
      "epoch": 0.10686912925718896,
      "grad_norm": 1.5362392663955688,
      "learning_rate": 9.845932411594867e-06,
      "loss": 0.8597,
      "step": 3709
    },
    {
      "epoch": 0.10689794271883823,
      "grad_norm": 1.6231380701065063,
      "learning_rate": 9.845817451279669e-06,
      "loss": 0.8441,
      "step": 3710
    },
    {
      "epoch": 0.10692675618048753,
      "grad_norm": 1.693291425704956,
      "learning_rate": 9.845702448762245e-06,
      "loss": 0.8533,
      "step": 3711
    },
    {
      "epoch": 0.1069555696421368,
      "grad_norm": 1.598160982131958,
      "learning_rate": 9.845587404043597e-06,
      "loss": 0.8344,
      "step": 3712
    },
    {
      "epoch": 0.1069843831037861,
      "grad_norm": 1.6374928951263428,
      "learning_rate": 9.845472317124725e-06,
      "loss": 0.8082,
      "step": 3713
    },
    {
      "epoch": 0.10701319656543537,
      "grad_norm": 1.7027136087417603,
      "learning_rate": 9.845357188006635e-06,
      "loss": 0.8381,
      "step": 3714
    },
    {
      "epoch": 0.10704201002708466,
      "grad_norm": 1.6097004413604736,
      "learning_rate": 9.845242016690327e-06,
      "loss": 0.8502,
      "step": 3715
    },
    {
      "epoch": 0.10707082348873394,
      "grad_norm": 1.593470811843872,
      "learning_rate": 9.845126803176803e-06,
      "loss": 0.8367,
      "step": 3716
    },
    {
      "epoch": 0.10709963695038321,
      "grad_norm": 1.6364493370056152,
      "learning_rate": 9.845011547467069e-06,
      "loss": 0.8226,
      "step": 3717
    },
    {
      "epoch": 0.1071284504120325,
      "grad_norm": 1.5794103145599365,
      "learning_rate": 9.844896249562126e-06,
      "loss": 0.8432,
      "step": 3718
    },
    {
      "epoch": 0.10715726387368178,
      "grad_norm": 1.580935001373291,
      "learning_rate": 9.844780909462982e-06,
      "loss": 0.8329,
      "step": 3719
    },
    {
      "epoch": 0.10718607733533107,
      "grad_norm": 1.6783488988876343,
      "learning_rate": 9.844665527170638e-06,
      "loss": 0.839,
      "step": 3720
    },
    {
      "epoch": 0.10721489079698035,
      "grad_norm": 1.5330368280410767,
      "learning_rate": 9.8445501026861e-06,
      "loss": 0.839,
      "step": 3721
    },
    {
      "epoch": 0.10724370425862963,
      "grad_norm": 1.5903900861740112,
      "learning_rate": 9.844434636010374e-06,
      "loss": 0.8477,
      "step": 3722
    },
    {
      "epoch": 0.10727251772027892,
      "grad_norm": 1.6568366289138794,
      "learning_rate": 9.844319127144465e-06,
      "loss": 0.844,
      "step": 3723
    },
    {
      "epoch": 0.1073013311819282,
      "grad_norm": 1.5100607872009277,
      "learning_rate": 9.844203576089379e-06,
      "loss": 0.8291,
      "step": 3724
    },
    {
      "epoch": 0.10733014464357749,
      "grad_norm": 1.5396531820297241,
      "learning_rate": 9.844087982846123e-06,
      "loss": 0.8467,
      "step": 3725
    },
    {
      "epoch": 0.10735895810522676,
      "grad_norm": 1.5875189304351807,
      "learning_rate": 9.8439723474157e-06,
      "loss": 0.8703,
      "step": 3726
    },
    {
      "epoch": 0.10738777156687604,
      "grad_norm": 1.6062183380126953,
      "learning_rate": 9.843856669799123e-06,
      "loss": 0.8407,
      "step": 3727
    },
    {
      "epoch": 0.10741658502852533,
      "grad_norm": 1.4827316999435425,
      "learning_rate": 9.843740949997396e-06,
      "loss": 0.8545,
      "step": 3728
    },
    {
      "epoch": 0.1074453984901746,
      "grad_norm": 1.6135468482971191,
      "learning_rate": 9.843625188011528e-06,
      "loss": 0.839,
      "step": 3729
    },
    {
      "epoch": 0.1074742119518239,
      "grad_norm": 1.6541045904159546,
      "learning_rate": 9.843509383842524e-06,
      "loss": 0.8435,
      "step": 3730
    },
    {
      "epoch": 0.10750302541347317,
      "grad_norm": 1.60893714427948,
      "learning_rate": 9.843393537491396e-06,
      "loss": 0.8344,
      "step": 3731
    },
    {
      "epoch": 0.10753183887512245,
      "grad_norm": 1.491101861000061,
      "learning_rate": 9.843277648959153e-06,
      "loss": 0.8205,
      "step": 3732
    },
    {
      "epoch": 0.10756065233677174,
      "grad_norm": 1.5470198392868042,
      "learning_rate": 9.843161718246803e-06,
      "loss": 0.8648,
      "step": 3733
    },
    {
      "epoch": 0.10758946579842102,
      "grad_norm": 1.5499194860458374,
      "learning_rate": 9.843045745355355e-06,
      "loss": 0.8378,
      "step": 3734
    },
    {
      "epoch": 0.10761827926007031,
      "grad_norm": 1.5808908939361572,
      "learning_rate": 9.84292973028582e-06,
      "loss": 0.8159,
      "step": 3735
    },
    {
      "epoch": 0.10764709272171959,
      "grad_norm": 1.4733729362487793,
      "learning_rate": 9.842813673039209e-06,
      "loss": 0.8577,
      "step": 3736
    },
    {
      "epoch": 0.10767590618336886,
      "grad_norm": 1.6074177026748657,
      "learning_rate": 9.842697573616531e-06,
      "loss": 0.8418,
      "step": 3737
    },
    {
      "epoch": 0.10770471964501815,
      "grad_norm": 1.6022995710372925,
      "learning_rate": 9.842581432018798e-06,
      "loss": 0.865,
      "step": 3738
    },
    {
      "epoch": 0.10773353310666743,
      "grad_norm": 1.5348660945892334,
      "learning_rate": 9.842465248247022e-06,
      "loss": 0.845,
      "step": 3739
    },
    {
      "epoch": 0.10776234656831672,
      "grad_norm": 1.4895989894866943,
      "learning_rate": 9.842349022302215e-06,
      "loss": 0.8469,
      "step": 3740
    },
    {
      "epoch": 0.107791160029966,
      "grad_norm": 1.5950628519058228,
      "learning_rate": 9.842232754185388e-06,
      "loss": 0.8176,
      "step": 3741
    },
    {
      "epoch": 0.10781997349161528,
      "grad_norm": 1.5214678049087524,
      "learning_rate": 9.842116443897554e-06,
      "loss": 0.8186,
      "step": 3742
    },
    {
      "epoch": 0.10784878695326457,
      "grad_norm": 1.5894384384155273,
      "learning_rate": 9.842000091439726e-06,
      "loss": 0.8481,
      "step": 3743
    },
    {
      "epoch": 0.10787760041491384,
      "grad_norm": 1.633432149887085,
      "learning_rate": 9.841883696812918e-06,
      "loss": 0.867,
      "step": 3744
    },
    {
      "epoch": 0.10790641387656313,
      "grad_norm": 1.5695297718048096,
      "learning_rate": 9.841767260018143e-06,
      "loss": 0.8436,
      "step": 3745
    },
    {
      "epoch": 0.10793522733821241,
      "grad_norm": 1.5157032012939453,
      "learning_rate": 9.841650781056413e-06,
      "loss": 0.8546,
      "step": 3746
    },
    {
      "epoch": 0.1079640407998617,
      "grad_norm": 1.7324897050857544,
      "learning_rate": 9.841534259928747e-06,
      "loss": 0.8442,
      "step": 3747
    },
    {
      "epoch": 0.10799285426151098,
      "grad_norm": 1.5462647676467896,
      "learning_rate": 9.841417696636157e-06,
      "loss": 0.8427,
      "step": 3748
    },
    {
      "epoch": 0.10802166772316026,
      "grad_norm": 1.5462349653244019,
      "learning_rate": 9.841301091179657e-06,
      "loss": 0.8533,
      "step": 3749
    },
    {
      "epoch": 0.10805048118480955,
      "grad_norm": 1.6046944856643677,
      "learning_rate": 9.841184443560266e-06,
      "loss": 0.8712,
      "step": 3750
    },
    {
      "epoch": 0.10807929464645882,
      "grad_norm": 1.596147894859314,
      "learning_rate": 9.841067753778998e-06,
      "loss": 0.8708,
      "step": 3751
    },
    {
      "epoch": 0.10810810810810811,
      "grad_norm": 1.6062220335006714,
      "learning_rate": 9.840951021836869e-06,
      "loss": 0.8283,
      "step": 3752
    },
    {
      "epoch": 0.10813692156975739,
      "grad_norm": 1.645391821861267,
      "learning_rate": 9.840834247734894e-06,
      "loss": 0.8267,
      "step": 3753
    },
    {
      "epoch": 0.10816573503140667,
      "grad_norm": 1.5011353492736816,
      "learning_rate": 9.840717431474094e-06,
      "loss": 0.8314,
      "step": 3754
    },
    {
      "epoch": 0.10819454849305596,
      "grad_norm": 1.5980560779571533,
      "learning_rate": 9.840600573055484e-06,
      "loss": 0.8513,
      "step": 3755
    },
    {
      "epoch": 0.10822336195470524,
      "grad_norm": 1.5551871061325073,
      "learning_rate": 9.840483672480079e-06,
      "loss": 0.8523,
      "step": 3756
    },
    {
      "epoch": 0.10825217541635453,
      "grad_norm": 1.4604641199111938,
      "learning_rate": 9.840366729748903e-06,
      "loss": 0.8525,
      "step": 3757
    },
    {
      "epoch": 0.1082809888780038,
      "grad_norm": 1.5910693407058716,
      "learning_rate": 9.84024974486297e-06,
      "loss": 0.827,
      "step": 3758
    },
    {
      "epoch": 0.10830980233965308,
      "grad_norm": 1.5598084926605225,
      "learning_rate": 9.8401327178233e-06,
      "loss": 0.8436,
      "step": 3759
    },
    {
      "epoch": 0.10833861580130237,
      "grad_norm": 1.5341529846191406,
      "learning_rate": 9.840015648630912e-06,
      "loss": 0.834,
      "step": 3760
    },
    {
      "epoch": 0.10836742926295165,
      "grad_norm": 1.6464966535568237,
      "learning_rate": 9.839898537286827e-06,
      "loss": 0.8336,
      "step": 3761
    },
    {
      "epoch": 0.10839624272460094,
      "grad_norm": 1.5996899604797363,
      "learning_rate": 9.839781383792065e-06,
      "loss": 0.8387,
      "step": 3762
    },
    {
      "epoch": 0.10842505618625022,
      "grad_norm": 1.5154184103012085,
      "learning_rate": 9.839664188147644e-06,
      "loss": 0.8414,
      "step": 3763
    },
    {
      "epoch": 0.10845386964789949,
      "grad_norm": 1.5575950145721436,
      "learning_rate": 9.839546950354586e-06,
      "loss": 0.8477,
      "step": 3764
    },
    {
      "epoch": 0.10848268310954878,
      "grad_norm": 1.5337013006210327,
      "learning_rate": 9.83942967041391e-06,
      "loss": 0.8113,
      "step": 3765
    },
    {
      "epoch": 0.10851149657119806,
      "grad_norm": 1.5732738971710205,
      "learning_rate": 9.83931234832664e-06,
      "loss": 0.8634,
      "step": 3766
    },
    {
      "epoch": 0.10854031003284735,
      "grad_norm": 1.4842013120651245,
      "learning_rate": 9.839194984093799e-06,
      "loss": 0.8459,
      "step": 3767
    },
    {
      "epoch": 0.10856912349449663,
      "grad_norm": 1.5341798067092896,
      "learning_rate": 9.839077577716405e-06,
      "loss": 0.8321,
      "step": 3768
    },
    {
      "epoch": 0.1085979369561459,
      "grad_norm": 1.5190361738204956,
      "learning_rate": 9.838960129195483e-06,
      "loss": 0.8293,
      "step": 3769
    },
    {
      "epoch": 0.1086267504177952,
      "grad_norm": 1.5424418449401855,
      "learning_rate": 9.838842638532057e-06,
      "loss": 0.8552,
      "step": 3770
    },
    {
      "epoch": 0.10865556387944447,
      "grad_norm": 1.4493505954742432,
      "learning_rate": 9.838725105727147e-06,
      "loss": 0.8556,
      "step": 3771
    },
    {
      "epoch": 0.10868437734109376,
      "grad_norm": 1.5077873468399048,
      "learning_rate": 9.838607530781779e-06,
      "loss": 0.8234,
      "step": 3772
    },
    {
      "epoch": 0.10871319080274304,
      "grad_norm": 1.5369114875793457,
      "learning_rate": 9.838489913696975e-06,
      "loss": 0.8267,
      "step": 3773
    },
    {
      "epoch": 0.10874200426439233,
      "grad_norm": 1.6001619100570679,
      "learning_rate": 9.838372254473763e-06,
      "loss": 0.8266,
      "step": 3774
    },
    {
      "epoch": 0.10877081772604161,
      "grad_norm": 1.5201115608215332,
      "learning_rate": 9.838254553113163e-06,
      "loss": 0.856,
      "step": 3775
    },
    {
      "epoch": 0.10879963118769088,
      "grad_norm": 1.575313925743103,
      "learning_rate": 9.838136809616204e-06,
      "loss": 0.8216,
      "step": 3776
    },
    {
      "epoch": 0.10882844464934018,
      "grad_norm": 1.687484860420227,
      "learning_rate": 9.838019023983909e-06,
      "loss": 0.8605,
      "step": 3777
    },
    {
      "epoch": 0.10885725811098945,
      "grad_norm": 2.0154459476470947,
      "learning_rate": 9.837901196217304e-06,
      "loss": 0.8155,
      "step": 3778
    },
    {
      "epoch": 0.10888607157263874,
      "grad_norm": 1.4923590421676636,
      "learning_rate": 9.837783326317416e-06,
      "loss": 0.8133,
      "step": 3779
    },
    {
      "epoch": 0.10891488503428802,
      "grad_norm": 1.7108967304229736,
      "learning_rate": 9.837665414285271e-06,
      "loss": 0.818,
      "step": 3780
    },
    {
      "epoch": 0.1089436984959373,
      "grad_norm": 1.5851019620895386,
      "learning_rate": 9.837547460121897e-06,
      "loss": 0.8402,
      "step": 3781
    },
    {
      "epoch": 0.10897251195758659,
      "grad_norm": 1.5840286016464233,
      "learning_rate": 9.837429463828321e-06,
      "loss": 0.8408,
      "step": 3782
    },
    {
      "epoch": 0.10900132541923586,
      "grad_norm": 1.746128797531128,
      "learning_rate": 9.837311425405569e-06,
      "loss": 0.8358,
      "step": 3783
    },
    {
      "epoch": 0.10903013888088516,
      "grad_norm": 1.5056267976760864,
      "learning_rate": 9.837193344854669e-06,
      "loss": 0.8262,
      "step": 3784
    },
    {
      "epoch": 0.10905895234253443,
      "grad_norm": 1.6376173496246338,
      "learning_rate": 9.837075222176653e-06,
      "loss": 0.8276,
      "step": 3785
    },
    {
      "epoch": 0.10908776580418371,
      "grad_norm": 1.7046918869018555,
      "learning_rate": 9.836957057372545e-06,
      "loss": 0.8377,
      "step": 3786
    },
    {
      "epoch": 0.109116579265833,
      "grad_norm": 1.5258338451385498,
      "learning_rate": 9.836838850443376e-06,
      "loss": 0.8221,
      "step": 3787
    },
    {
      "epoch": 0.10914539272748228,
      "grad_norm": 1.689644455909729,
      "learning_rate": 9.836720601390177e-06,
      "loss": 0.8256,
      "step": 3788
    },
    {
      "epoch": 0.10917420618913157,
      "grad_norm": 1.7481857538223267,
      "learning_rate": 9.836602310213976e-06,
      "loss": 0.8562,
      "step": 3789
    },
    {
      "epoch": 0.10920301965078084,
      "grad_norm": 1.5660901069641113,
      "learning_rate": 9.836483976915804e-06,
      "loss": 0.8371,
      "step": 3790
    },
    {
      "epoch": 0.10923183311243012,
      "grad_norm": 1.6795417070388794,
      "learning_rate": 9.836365601496692e-06,
      "loss": 0.8566,
      "step": 3791
    },
    {
      "epoch": 0.10926064657407941,
      "grad_norm": 1.580820918083191,
      "learning_rate": 9.836247183957668e-06,
      "loss": 0.834,
      "step": 3792
    },
    {
      "epoch": 0.10928946003572869,
      "grad_norm": 1.6391915082931519,
      "learning_rate": 9.836128724299768e-06,
      "loss": 0.8241,
      "step": 3793
    },
    {
      "epoch": 0.10931827349737798,
      "grad_norm": 1.5605305433273315,
      "learning_rate": 9.836010222524018e-06,
      "loss": 0.8385,
      "step": 3794
    },
    {
      "epoch": 0.10934708695902726,
      "grad_norm": 1.6163818836212158,
      "learning_rate": 9.835891678631454e-06,
      "loss": 0.8303,
      "step": 3795
    },
    {
      "epoch": 0.10937590042067653,
      "grad_norm": 1.552976369857788,
      "learning_rate": 9.835773092623111e-06,
      "loss": 0.8136,
      "step": 3796
    },
    {
      "epoch": 0.10940471388232582,
      "grad_norm": 1.5299913883209229,
      "learning_rate": 9.835654464500016e-06,
      "loss": 0.8313,
      "step": 3797
    },
    {
      "epoch": 0.1094335273439751,
      "grad_norm": 1.5169061422348022,
      "learning_rate": 9.835535794263205e-06,
      "loss": 0.8226,
      "step": 3798
    },
    {
      "epoch": 0.10946234080562439,
      "grad_norm": 1.5852683782577515,
      "learning_rate": 9.83541708191371e-06,
      "loss": 0.844,
      "step": 3799
    },
    {
      "epoch": 0.10949115426727367,
      "grad_norm": 1.5323898792266846,
      "learning_rate": 9.835298327452566e-06,
      "loss": 0.8097,
      "step": 3800
    },
    {
      "epoch": 0.10951996772892296,
      "grad_norm": 1.502557635307312,
      "learning_rate": 9.835179530880808e-06,
      "loss": 0.8156,
      "step": 3801
    },
    {
      "epoch": 0.10954878119057224,
      "grad_norm": 1.4895102977752686,
      "learning_rate": 9.835060692199469e-06,
      "loss": 0.863,
      "step": 3802
    },
    {
      "epoch": 0.10957759465222151,
      "grad_norm": 1.429476022720337,
      "learning_rate": 9.834941811409584e-06,
      "loss": 0.8433,
      "step": 3803
    },
    {
      "epoch": 0.1096064081138708,
      "grad_norm": 1.545976996421814,
      "learning_rate": 9.83482288851219e-06,
      "loss": 0.8443,
      "step": 3804
    },
    {
      "epoch": 0.10963522157552008,
      "grad_norm": 1.5181679725646973,
      "learning_rate": 9.834703923508322e-06,
      "loss": 0.8494,
      "step": 3805
    },
    {
      "epoch": 0.10966403503716937,
      "grad_norm": 1.4625786542892456,
      "learning_rate": 9.834584916399015e-06,
      "loss": 0.8436,
      "step": 3806
    },
    {
      "epoch": 0.10969284849881865,
      "grad_norm": 1.478067398071289,
      "learning_rate": 9.834465867185306e-06,
      "loss": 0.848,
      "step": 3807
    },
    {
      "epoch": 0.10972166196046793,
      "grad_norm": 1.4533579349517822,
      "learning_rate": 9.834346775868233e-06,
      "loss": 0.8424,
      "step": 3808
    },
    {
      "epoch": 0.10975047542211722,
      "grad_norm": 1.543486475944519,
      "learning_rate": 9.834227642448833e-06,
      "loss": 0.8375,
      "step": 3809
    },
    {
      "epoch": 0.1097792888837665,
      "grad_norm": 1.6137702465057373,
      "learning_rate": 9.83410846692814e-06,
      "loss": 0.8459,
      "step": 3810
    },
    {
      "epoch": 0.10980810234541578,
      "grad_norm": 1.501124620437622,
      "learning_rate": 9.833989249307196e-06,
      "loss": 0.8193,
      "step": 3811
    },
    {
      "epoch": 0.10983691580706506,
      "grad_norm": 1.6056060791015625,
      "learning_rate": 9.833869989587036e-06,
      "loss": 0.8402,
      "step": 3812
    },
    {
      "epoch": 0.10986572926871434,
      "grad_norm": 1.4985804557800293,
      "learning_rate": 9.833750687768704e-06,
      "loss": 0.8372,
      "step": 3813
    },
    {
      "epoch": 0.10989454273036363,
      "grad_norm": 1.5901823043823242,
      "learning_rate": 9.833631343853233e-06,
      "loss": 0.8249,
      "step": 3814
    },
    {
      "epoch": 0.1099233561920129,
      "grad_norm": 1.4806606769561768,
      "learning_rate": 9.833511957841665e-06,
      "loss": 0.8287,
      "step": 3815
    },
    {
      "epoch": 0.1099521696536622,
      "grad_norm": 1.588112473487854,
      "learning_rate": 9.833392529735042e-06,
      "loss": 0.8245,
      "step": 3816
    },
    {
      "epoch": 0.10998098311531147,
      "grad_norm": 1.4905529022216797,
      "learning_rate": 9.8332730595344e-06,
      "loss": 0.8186,
      "step": 3817
    },
    {
      "epoch": 0.11000979657696075,
      "grad_norm": 1.5369318723678589,
      "learning_rate": 9.833153547240781e-06,
      "loss": 0.806,
      "step": 3818
    },
    {
      "epoch": 0.11003861003861004,
      "grad_norm": 1.5645902156829834,
      "learning_rate": 9.833033992855225e-06,
      "loss": 0.8457,
      "step": 3819
    },
    {
      "epoch": 0.11006742350025932,
      "grad_norm": 1.586631417274475,
      "learning_rate": 9.832914396378777e-06,
      "loss": 0.833,
      "step": 3820
    },
    {
      "epoch": 0.11009623696190861,
      "grad_norm": 1.4414846897125244,
      "learning_rate": 9.832794757812474e-06,
      "loss": 0.8179,
      "step": 3821
    },
    {
      "epoch": 0.11012505042355789,
      "grad_norm": 1.4983254671096802,
      "learning_rate": 9.83267507715736e-06,
      "loss": 0.8637,
      "step": 3822
    },
    {
      "epoch": 0.11015386388520716,
      "grad_norm": 1.6015511751174927,
      "learning_rate": 9.832555354414475e-06,
      "loss": 0.8472,
      "step": 3823
    },
    {
      "epoch": 0.11018267734685645,
      "grad_norm": 1.5532610416412354,
      "learning_rate": 9.832435589584866e-06,
      "loss": 0.8486,
      "step": 3824
    },
    {
      "epoch": 0.11021149080850573,
      "grad_norm": 1.617685079574585,
      "learning_rate": 9.832315782669573e-06,
      "loss": 0.8534,
      "step": 3825
    },
    {
      "epoch": 0.11024030427015502,
      "grad_norm": 1.562787652015686,
      "learning_rate": 9.832195933669639e-06,
      "loss": 0.8233,
      "step": 3826
    },
    {
      "epoch": 0.1102691177318043,
      "grad_norm": 1.608658790588379,
      "learning_rate": 9.832076042586112e-06,
      "loss": 0.8421,
      "step": 3827
    },
    {
      "epoch": 0.11029793119345359,
      "grad_norm": 1.5297026634216309,
      "learning_rate": 9.83195610942003e-06,
      "loss": 0.8388,
      "step": 3828
    },
    {
      "epoch": 0.11032674465510287,
      "grad_norm": 1.468101143836975,
      "learning_rate": 9.831836134172441e-06,
      "loss": 0.8446,
      "step": 3829
    },
    {
      "epoch": 0.11035555811675214,
      "grad_norm": 1.4716248512268066,
      "learning_rate": 9.83171611684439e-06,
      "loss": 0.8145,
      "step": 3830
    },
    {
      "epoch": 0.11038437157840143,
      "grad_norm": 1.4658539295196533,
      "learning_rate": 9.831596057436921e-06,
      "loss": 0.8188,
      "step": 3831
    },
    {
      "epoch": 0.11041318504005071,
      "grad_norm": 1.607608437538147,
      "learning_rate": 9.831475955951082e-06,
      "loss": 0.8464,
      "step": 3832
    },
    {
      "epoch": 0.1104419985017,
      "grad_norm": 1.5136444568634033,
      "learning_rate": 9.831355812387915e-06,
      "loss": 0.8128,
      "step": 3833
    },
    {
      "epoch": 0.11047081196334928,
      "grad_norm": 1.4741950035095215,
      "learning_rate": 9.831235626748467e-06,
      "loss": 0.8163,
      "step": 3834
    },
    {
      "epoch": 0.11049962542499855,
      "grad_norm": 1.526097059249878,
      "learning_rate": 9.83111539903379e-06,
      "loss": 0.8413,
      "step": 3835
    },
    {
      "epoch": 0.11052843888664785,
      "grad_norm": 1.5781962871551514,
      "learning_rate": 9.830995129244925e-06,
      "loss": 0.8313,
      "step": 3836
    },
    {
      "epoch": 0.11055725234829712,
      "grad_norm": 1.553823471069336,
      "learning_rate": 9.830874817382923e-06,
      "loss": 0.8328,
      "step": 3837
    },
    {
      "epoch": 0.11058606580994641,
      "grad_norm": 1.528343915939331,
      "learning_rate": 9.830754463448829e-06,
      "loss": 0.8584,
      "step": 3838
    },
    {
      "epoch": 0.11061487927159569,
      "grad_norm": 1.4933297634124756,
      "learning_rate": 9.830634067443692e-06,
      "loss": 0.8539,
      "step": 3839
    },
    {
      "epoch": 0.11064369273324497,
      "grad_norm": 1.4883335828781128,
      "learning_rate": 9.830513629368564e-06,
      "loss": 0.8224,
      "step": 3840
    },
    {
      "epoch": 0.11067250619489426,
      "grad_norm": 1.6272577047348022,
      "learning_rate": 9.830393149224489e-06,
      "loss": 0.8385,
      "step": 3841
    },
    {
      "epoch": 0.11070131965654353,
      "grad_norm": 1.4825479984283447,
      "learning_rate": 9.830272627012519e-06,
      "loss": 0.8569,
      "step": 3842
    },
    {
      "epoch": 0.11073013311819283,
      "grad_norm": 1.537581443786621,
      "learning_rate": 9.830152062733703e-06,
      "loss": 0.8296,
      "step": 3843
    },
    {
      "epoch": 0.1107589465798421,
      "grad_norm": 1.6430162191390991,
      "learning_rate": 9.83003145638909e-06,
      "loss": 0.8435,
      "step": 3844
    },
    {
      "epoch": 0.11078776004149138,
      "grad_norm": 1.5033560991287231,
      "learning_rate": 9.82991080797973e-06,
      "loss": 0.8373,
      "step": 3845
    },
    {
      "epoch": 0.11081657350314067,
      "grad_norm": 1.512586236000061,
      "learning_rate": 9.829790117506678e-06,
      "loss": 0.8361,
      "step": 3846
    },
    {
      "epoch": 0.11084538696478995,
      "grad_norm": 1.5250893831253052,
      "learning_rate": 9.829669384970981e-06,
      "loss": 0.8511,
      "step": 3847
    },
    {
      "epoch": 0.11087420042643924,
      "grad_norm": 1.525219440460205,
      "learning_rate": 9.829548610373691e-06,
      "loss": 0.802,
      "step": 3848
    },
    {
      "epoch": 0.11090301388808851,
      "grad_norm": 1.5379276275634766,
      "learning_rate": 9.82942779371586e-06,
      "loss": 0.8298,
      "step": 3849
    },
    {
      "epoch": 0.11093182734973779,
      "grad_norm": 1.6127889156341553,
      "learning_rate": 9.829306934998541e-06,
      "loss": 0.8339,
      "step": 3850
    },
    {
      "epoch": 0.11096064081138708,
      "grad_norm": 1.5134098529815674,
      "learning_rate": 9.829186034222788e-06,
      "loss": 0.8402,
      "step": 3851
    },
    {
      "epoch": 0.11098945427303636,
      "grad_norm": 1.461622953414917,
      "learning_rate": 9.82906509138965e-06,
      "loss": 0.8367,
      "step": 3852
    },
    {
      "epoch": 0.11101826773468565,
      "grad_norm": 1.5426857471466064,
      "learning_rate": 9.828944106500182e-06,
      "loss": 0.8578,
      "step": 3853
    },
    {
      "epoch": 0.11104708119633493,
      "grad_norm": 1.5583025217056274,
      "learning_rate": 9.828823079555441e-06,
      "loss": 0.8377,
      "step": 3854
    },
    {
      "epoch": 0.11107589465798422,
      "grad_norm": 1.4911831617355347,
      "learning_rate": 9.828702010556475e-06,
      "loss": 0.8417,
      "step": 3855
    },
    {
      "epoch": 0.1111047081196335,
      "grad_norm": 1.5397087335586548,
      "learning_rate": 9.828580899504343e-06,
      "loss": 0.8239,
      "step": 3856
    },
    {
      "epoch": 0.11113352158128277,
      "grad_norm": 1.4959298372268677,
      "learning_rate": 9.828459746400097e-06,
      "loss": 0.8282,
      "step": 3857
    },
    {
      "epoch": 0.11116233504293206,
      "grad_norm": 1.5079752206802368,
      "learning_rate": 9.828338551244794e-06,
      "loss": 0.8448,
      "step": 3858
    },
    {
      "epoch": 0.11119114850458134,
      "grad_norm": 1.5156313180923462,
      "learning_rate": 9.828217314039488e-06,
      "loss": 0.8239,
      "step": 3859
    },
    {
      "epoch": 0.11121996196623063,
      "grad_norm": 1.627464771270752,
      "learning_rate": 9.828096034785237e-06,
      "loss": 0.8514,
      "step": 3860
    },
    {
      "epoch": 0.1112487754278799,
      "grad_norm": 1.488652229309082,
      "learning_rate": 9.827974713483096e-06,
      "loss": 0.817,
      "step": 3861
    },
    {
      "epoch": 0.11127758888952918,
      "grad_norm": 1.5238165855407715,
      "learning_rate": 9.82785335013412e-06,
      "loss": 0.8432,
      "step": 3862
    },
    {
      "epoch": 0.11130640235117847,
      "grad_norm": 1.6072129011154175,
      "learning_rate": 9.827731944739367e-06,
      "loss": 0.8404,
      "step": 3863
    },
    {
      "epoch": 0.11133521581282775,
      "grad_norm": 1.4458025693893433,
      "learning_rate": 9.827610497299897e-06,
      "loss": 0.8523,
      "step": 3864
    },
    {
      "epoch": 0.11136402927447704,
      "grad_norm": 1.5665363073349,
      "learning_rate": 9.827489007816764e-06,
      "loss": 0.8397,
      "step": 3865
    },
    {
      "epoch": 0.11139284273612632,
      "grad_norm": 1.551185131072998,
      "learning_rate": 9.827367476291027e-06,
      "loss": 0.8207,
      "step": 3866
    },
    {
      "epoch": 0.1114216561977756,
      "grad_norm": 1.4874237775802612,
      "learning_rate": 9.827245902723745e-06,
      "loss": 0.8445,
      "step": 3867
    },
    {
      "epoch": 0.11145046965942489,
      "grad_norm": 1.4885398149490356,
      "learning_rate": 9.827124287115978e-06,
      "loss": 0.8164,
      "step": 3868
    },
    {
      "epoch": 0.11147928312107416,
      "grad_norm": 1.523769736289978,
      "learning_rate": 9.827002629468782e-06,
      "loss": 0.8014,
      "step": 3869
    },
    {
      "epoch": 0.11150809658272345,
      "grad_norm": 1.5043785572052002,
      "learning_rate": 9.82688092978322e-06,
      "loss": 0.8161,
      "step": 3870
    },
    {
      "epoch": 0.11153691004437273,
      "grad_norm": 1.515223503112793,
      "learning_rate": 9.826759188060349e-06,
      "loss": 0.8387,
      "step": 3871
    },
    {
      "epoch": 0.11156572350602201,
      "grad_norm": 1.5263389348983765,
      "learning_rate": 9.826637404301231e-06,
      "loss": 0.8041,
      "step": 3872
    },
    {
      "epoch": 0.1115945369676713,
      "grad_norm": 1.6095457077026367,
      "learning_rate": 9.826515578506927e-06,
      "loss": 0.8154,
      "step": 3873
    },
    {
      "epoch": 0.11162335042932058,
      "grad_norm": 1.4802623987197876,
      "learning_rate": 9.826393710678497e-06,
      "loss": 0.8231,
      "step": 3874
    },
    {
      "epoch": 0.11165216389096987,
      "grad_norm": 1.5442192554473877,
      "learning_rate": 9.826271800817e-06,
      "loss": 0.8239,
      "step": 3875
    },
    {
      "epoch": 0.11168097735261914,
      "grad_norm": 1.5798859596252441,
      "learning_rate": 9.826149848923502e-06,
      "loss": 0.8106,
      "step": 3876
    },
    {
      "epoch": 0.11170979081426842,
      "grad_norm": 1.5553959608078003,
      "learning_rate": 9.826027854999063e-06,
      "loss": 0.8308,
      "step": 3877
    },
    {
      "epoch": 0.11173860427591771,
      "grad_norm": 1.5488827228546143,
      "learning_rate": 9.825905819044745e-06,
      "loss": 0.8352,
      "step": 3878
    },
    {
      "epoch": 0.11176741773756699,
      "grad_norm": 1.4500094652175903,
      "learning_rate": 9.825783741061613e-06,
      "loss": 0.8454,
      "step": 3879
    },
    {
      "epoch": 0.11179623119921628,
      "grad_norm": 1.549766182899475,
      "learning_rate": 9.825661621050726e-06,
      "loss": 0.845,
      "step": 3880
    },
    {
      "epoch": 0.11182504466086556,
      "grad_norm": 1.4392136335372925,
      "learning_rate": 9.825539459013154e-06,
      "loss": 0.8198,
      "step": 3881
    },
    {
      "epoch": 0.11185385812251483,
      "grad_norm": 1.4113863706588745,
      "learning_rate": 9.825417254949953e-06,
      "loss": 0.8082,
      "step": 3882
    },
    {
      "epoch": 0.11188267158416412,
      "grad_norm": 1.5192162990570068,
      "learning_rate": 9.825295008862194e-06,
      "loss": 0.8579,
      "step": 3883
    },
    {
      "epoch": 0.1119114850458134,
      "grad_norm": 1.4277467727661133,
      "learning_rate": 9.825172720750938e-06,
      "loss": 0.8367,
      "step": 3884
    },
    {
      "epoch": 0.11194029850746269,
      "grad_norm": 1.4927068948745728,
      "learning_rate": 9.825050390617252e-06,
      "loss": 0.8207,
      "step": 3885
    },
    {
      "epoch": 0.11196911196911197,
      "grad_norm": 1.4707998037338257,
      "learning_rate": 9.824928018462199e-06,
      "loss": 0.8194,
      "step": 3886
    },
    {
      "epoch": 0.11199792543076126,
      "grad_norm": 1.4285420179367065,
      "learning_rate": 9.824805604286847e-06,
      "loss": 0.8064,
      "step": 3887
    },
    {
      "epoch": 0.11202673889241054,
      "grad_norm": 1.4623082876205444,
      "learning_rate": 9.824683148092262e-06,
      "loss": 0.8539,
      "step": 3888
    },
    {
      "epoch": 0.11205555235405981,
      "grad_norm": 1.4520877599716187,
      "learning_rate": 9.824560649879508e-06,
      "loss": 0.8474,
      "step": 3889
    },
    {
      "epoch": 0.1120843658157091,
      "grad_norm": 1.563666820526123,
      "learning_rate": 9.824438109649654e-06,
      "loss": 0.8455,
      "step": 3890
    },
    {
      "epoch": 0.11211317927735838,
      "grad_norm": 1.5498594045639038,
      "learning_rate": 9.824315527403766e-06,
      "loss": 0.8043,
      "step": 3891
    },
    {
      "epoch": 0.11214199273900767,
      "grad_norm": 1.5061144828796387,
      "learning_rate": 9.824192903142913e-06,
      "loss": 0.8157,
      "step": 3892
    },
    {
      "epoch": 0.11217080620065695,
      "grad_norm": 1.5632917881011963,
      "learning_rate": 9.824070236868164e-06,
      "loss": 0.8366,
      "step": 3893
    },
    {
      "epoch": 0.11219961966230622,
      "grad_norm": 1.477961778640747,
      "learning_rate": 9.823947528580585e-06,
      "loss": 0.8163,
      "step": 3894
    },
    {
      "epoch": 0.11222843312395551,
      "grad_norm": 1.7422068119049072,
      "learning_rate": 9.823824778281243e-06,
      "loss": 0.8489,
      "step": 3895
    },
    {
      "epoch": 0.11225724658560479,
      "grad_norm": 1.5496940612792969,
      "learning_rate": 9.823701985971211e-06,
      "loss": 0.8495,
      "step": 3896
    },
    {
      "epoch": 0.11228606004725408,
      "grad_norm": 1.5710474252700806,
      "learning_rate": 9.823579151651555e-06,
      "loss": 0.8413,
      "step": 3897
    },
    {
      "epoch": 0.11231487350890336,
      "grad_norm": 1.5081534385681152,
      "learning_rate": 9.823456275323348e-06,
      "loss": 0.8341,
      "step": 3898
    },
    {
      "epoch": 0.11234368697055264,
      "grad_norm": 1.4890543222427368,
      "learning_rate": 9.823333356987658e-06,
      "loss": 0.8177,
      "step": 3899
    },
    {
      "epoch": 0.11237250043220193,
      "grad_norm": 1.537145733833313,
      "learning_rate": 9.823210396645556e-06,
      "loss": 0.8489,
      "step": 3900
    },
    {
      "epoch": 0.1124013138938512,
      "grad_norm": 1.588811993598938,
      "learning_rate": 9.823087394298113e-06,
      "loss": 0.837,
      "step": 3901
    },
    {
      "epoch": 0.1124301273555005,
      "grad_norm": 1.4213497638702393,
      "learning_rate": 9.8229643499464e-06,
      "loss": 0.8306,
      "step": 3902
    },
    {
      "epoch": 0.11245894081714977,
      "grad_norm": 1.540778398513794,
      "learning_rate": 9.822841263591489e-06,
      "loss": 0.84,
      "step": 3903
    },
    {
      "epoch": 0.11248775427879905,
      "grad_norm": 1.5935157537460327,
      "learning_rate": 9.822718135234452e-06,
      "loss": 0.8332,
      "step": 3904
    },
    {
      "epoch": 0.11251656774044834,
      "grad_norm": 1.5506781339645386,
      "learning_rate": 9.82259496487636e-06,
      "loss": 0.8444,
      "step": 3905
    },
    {
      "epoch": 0.11254538120209762,
      "grad_norm": 1.5654460191726685,
      "learning_rate": 9.822471752518287e-06,
      "loss": 0.8336,
      "step": 3906
    },
    {
      "epoch": 0.11257419466374691,
      "grad_norm": 1.5139819383621216,
      "learning_rate": 9.822348498161307e-06,
      "loss": 0.8225,
      "step": 3907
    },
    {
      "epoch": 0.11260300812539618,
      "grad_norm": 1.5778192281723022,
      "learning_rate": 9.82222520180649e-06,
      "loss": 0.8195,
      "step": 3908
    },
    {
      "epoch": 0.11263182158704546,
      "grad_norm": 1.6607329845428467,
      "learning_rate": 9.822101863454913e-06,
      "loss": 0.8156,
      "step": 3909
    },
    {
      "epoch": 0.11266063504869475,
      "grad_norm": 1.5353634357452393,
      "learning_rate": 9.82197848310765e-06,
      "loss": 0.8237,
      "step": 3910
    },
    {
      "epoch": 0.11268944851034403,
      "grad_norm": 1.5736193656921387,
      "learning_rate": 9.821855060765773e-06,
      "loss": 0.854,
      "step": 3911
    },
    {
      "epoch": 0.11271826197199332,
      "grad_norm": 1.658496618270874,
      "learning_rate": 9.82173159643036e-06,
      "loss": 0.8353,
      "step": 3912
    },
    {
      "epoch": 0.1127470754336426,
      "grad_norm": 1.4876806735992432,
      "learning_rate": 9.821608090102484e-06,
      "loss": 0.8398,
      "step": 3913
    },
    {
      "epoch": 0.11277588889529189,
      "grad_norm": 1.5961508750915527,
      "learning_rate": 9.821484541783222e-06,
      "loss": 0.8395,
      "step": 3914
    },
    {
      "epoch": 0.11280470235694116,
      "grad_norm": 1.597923994064331,
      "learning_rate": 9.821360951473649e-06,
      "loss": 0.8504,
      "step": 3915
    },
    {
      "epoch": 0.11283351581859044,
      "grad_norm": 1.4044469594955444,
      "learning_rate": 9.821237319174842e-06,
      "loss": 0.8306,
      "step": 3916
    },
    {
      "epoch": 0.11286232928023973,
      "grad_norm": 1.6936229467391968,
      "learning_rate": 9.821113644887876e-06,
      "loss": 0.8134,
      "step": 3917
    },
    {
      "epoch": 0.11289114274188901,
      "grad_norm": 1.5720512866973877,
      "learning_rate": 9.82098992861383e-06,
      "loss": 0.8513,
      "step": 3918
    },
    {
      "epoch": 0.1129199562035383,
      "grad_norm": 1.5323195457458496,
      "learning_rate": 9.82086617035378e-06,
      "loss": 0.8365,
      "step": 3919
    },
    {
      "epoch": 0.11294876966518758,
      "grad_norm": 1.6022568941116333,
      "learning_rate": 9.820742370108807e-06,
      "loss": 0.8189,
      "step": 3920
    },
    {
      "epoch": 0.11297758312683685,
      "grad_norm": 1.6009198427200317,
      "learning_rate": 9.820618527879986e-06,
      "loss": 0.8572,
      "step": 3921
    },
    {
      "epoch": 0.11300639658848614,
      "grad_norm": 1.6330666542053223,
      "learning_rate": 9.820494643668397e-06,
      "loss": 0.7929,
      "step": 3922
    },
    {
      "epoch": 0.11303521005013542,
      "grad_norm": 1.5039258003234863,
      "learning_rate": 9.820370717475119e-06,
      "loss": 0.8321,
      "step": 3923
    },
    {
      "epoch": 0.11306402351178471,
      "grad_norm": 1.5794099569320679,
      "learning_rate": 9.820246749301227e-06,
      "loss": 0.8507,
      "step": 3924
    },
    {
      "epoch": 0.11309283697343399,
      "grad_norm": 1.610490083694458,
      "learning_rate": 9.820122739147809e-06,
      "loss": 0.8379,
      "step": 3925
    },
    {
      "epoch": 0.11312165043508327,
      "grad_norm": 1.5522148609161377,
      "learning_rate": 9.819998687015937e-06,
      "loss": 0.8428,
      "step": 3926
    },
    {
      "epoch": 0.11315046389673256,
      "grad_norm": 1.587267518043518,
      "learning_rate": 9.819874592906696e-06,
      "loss": 0.8198,
      "step": 3927
    },
    {
      "epoch": 0.11317927735838183,
      "grad_norm": 1.4935377836227417,
      "learning_rate": 9.819750456821166e-06,
      "loss": 0.8209,
      "step": 3928
    },
    {
      "epoch": 0.11320809082003112,
      "grad_norm": 1.4623336791992188,
      "learning_rate": 9.819626278760426e-06,
      "loss": 0.8472,
      "step": 3929
    },
    {
      "epoch": 0.1132369042816804,
      "grad_norm": 1.5968180894851685,
      "learning_rate": 9.81950205872556e-06,
      "loss": 0.829,
      "step": 3930
    },
    {
      "epoch": 0.11326571774332968,
      "grad_norm": 1.5083435773849487,
      "learning_rate": 9.819377796717649e-06,
      "loss": 0.822,
      "step": 3931
    },
    {
      "epoch": 0.11329453120497897,
      "grad_norm": 1.6674022674560547,
      "learning_rate": 9.819253492737775e-06,
      "loss": 0.857,
      "step": 3932
    },
    {
      "epoch": 0.11332334466662825,
      "grad_norm": 1.5361860990524292,
      "learning_rate": 9.819129146787018e-06,
      "loss": 0.8419,
      "step": 3933
    },
    {
      "epoch": 0.11335215812827754,
      "grad_norm": 1.5784605741500854,
      "learning_rate": 9.819004758866466e-06,
      "loss": 0.8211,
      "step": 3934
    },
    {
      "epoch": 0.11338097158992681,
      "grad_norm": 1.6516984701156616,
      "learning_rate": 9.8188803289772e-06,
      "loss": 0.8239,
      "step": 3935
    },
    {
      "epoch": 0.11340978505157609,
      "grad_norm": 1.4789471626281738,
      "learning_rate": 9.818755857120303e-06,
      "loss": 0.8364,
      "step": 3936
    },
    {
      "epoch": 0.11343859851322538,
      "grad_norm": 1.549001932144165,
      "learning_rate": 9.818631343296859e-06,
      "loss": 0.8434,
      "step": 3937
    },
    {
      "epoch": 0.11346741197487466,
      "grad_norm": 1.5994116067886353,
      "learning_rate": 9.818506787507953e-06,
      "loss": 0.8371,
      "step": 3938
    },
    {
      "epoch": 0.11349622543652395,
      "grad_norm": 1.6013238430023193,
      "learning_rate": 9.81838218975467e-06,
      "loss": 0.8416,
      "step": 3939
    },
    {
      "epoch": 0.11352503889817323,
      "grad_norm": 1.568062424659729,
      "learning_rate": 9.818257550038095e-06,
      "loss": 0.8494,
      "step": 3940
    },
    {
      "epoch": 0.11355385235982252,
      "grad_norm": 1.4679855108261108,
      "learning_rate": 9.818132868359313e-06,
      "loss": 0.8168,
      "step": 3941
    },
    {
      "epoch": 0.11358266582147179,
      "grad_norm": 1.5168668031692505,
      "learning_rate": 9.81800814471941e-06,
      "loss": 0.8287,
      "step": 3942
    },
    {
      "epoch": 0.11361147928312107,
      "grad_norm": 1.5376478433609009,
      "learning_rate": 9.817883379119472e-06,
      "loss": 0.8036,
      "step": 3943
    },
    {
      "epoch": 0.11364029274477036,
      "grad_norm": 1.5119789838790894,
      "learning_rate": 9.817758571560587e-06,
      "loss": 0.814,
      "step": 3944
    },
    {
      "epoch": 0.11366910620641964,
      "grad_norm": 1.5266039371490479,
      "learning_rate": 9.817633722043838e-06,
      "loss": 0.8026,
      "step": 3945
    },
    {
      "epoch": 0.11369791966806893,
      "grad_norm": 1.499445915222168,
      "learning_rate": 9.817508830570319e-06,
      "loss": 0.8401,
      "step": 3946
    },
    {
      "epoch": 0.1137267331297182,
      "grad_norm": 1.5906617641448975,
      "learning_rate": 9.817383897141111e-06,
      "loss": 0.8338,
      "step": 3947
    },
    {
      "epoch": 0.11375554659136748,
      "grad_norm": 1.5213415622711182,
      "learning_rate": 9.817258921757306e-06,
      "loss": 0.825,
      "step": 3948
    },
    {
      "epoch": 0.11378436005301677,
      "grad_norm": 1.5658601522445679,
      "learning_rate": 9.817133904419992e-06,
      "loss": 0.8318,
      "step": 3949
    },
    {
      "epoch": 0.11381317351466605,
      "grad_norm": 1.6097326278686523,
      "learning_rate": 9.817008845130255e-06,
      "loss": 0.8165,
      "step": 3950
    },
    {
      "epoch": 0.11384198697631534,
      "grad_norm": 1.5064618587493896,
      "learning_rate": 9.816883743889187e-06,
      "loss": 0.8312,
      "step": 3951
    },
    {
      "epoch": 0.11387080043796462,
      "grad_norm": 1.5128928422927856,
      "learning_rate": 9.816758600697878e-06,
      "loss": 0.8233,
      "step": 3952
    },
    {
      "epoch": 0.1138996138996139,
      "grad_norm": 1.5894043445587158,
      "learning_rate": 9.816633415557415e-06,
      "loss": 0.8269,
      "step": 3953
    },
    {
      "epoch": 0.11392842736126318,
      "grad_norm": 1.4618698358535767,
      "learning_rate": 9.81650818846889e-06,
      "loss": 0.8322,
      "step": 3954
    },
    {
      "epoch": 0.11395724082291246,
      "grad_norm": 1.5975980758666992,
      "learning_rate": 9.816382919433394e-06,
      "loss": 0.8287,
      "step": 3955
    },
    {
      "epoch": 0.11398605428456175,
      "grad_norm": 1.5381133556365967,
      "learning_rate": 9.816257608452018e-06,
      "loss": 0.8299,
      "step": 3956
    },
    {
      "epoch": 0.11401486774621103,
      "grad_norm": 1.5514031648635864,
      "learning_rate": 9.816132255525852e-06,
      "loss": 0.81,
      "step": 3957
    },
    {
      "epoch": 0.1140436812078603,
      "grad_norm": 1.5137627124786377,
      "learning_rate": 9.816006860655988e-06,
      "loss": 0.8395,
      "step": 3958
    },
    {
      "epoch": 0.1140724946695096,
      "grad_norm": 1.5337897539138794,
      "learning_rate": 9.815881423843518e-06,
      "loss": 0.824,
      "step": 3959
    },
    {
      "epoch": 0.11410130813115887,
      "grad_norm": 1.566625952720642,
      "learning_rate": 9.815755945089537e-06,
      "loss": 0.8301,
      "step": 3960
    },
    {
      "epoch": 0.11413012159280816,
      "grad_norm": 1.5313565731048584,
      "learning_rate": 9.815630424395135e-06,
      "loss": 0.8429,
      "step": 3961
    },
    {
      "epoch": 0.11415893505445744,
      "grad_norm": 1.5030388832092285,
      "learning_rate": 9.815504861761404e-06,
      "loss": 0.8234,
      "step": 3962
    },
    {
      "epoch": 0.11418774851610672,
      "grad_norm": 1.5925670862197876,
      "learning_rate": 9.815379257189442e-06,
      "loss": 0.8188,
      "step": 3963
    },
    {
      "epoch": 0.11421656197775601,
      "grad_norm": 1.4635233879089355,
      "learning_rate": 9.815253610680338e-06,
      "loss": 0.8282,
      "step": 3964
    },
    {
      "epoch": 0.11424537543940529,
      "grad_norm": 1.551346778869629,
      "learning_rate": 9.815127922235191e-06,
      "loss": 0.82,
      "step": 3965
    },
    {
      "epoch": 0.11427418890105458,
      "grad_norm": 1.5182673931121826,
      "learning_rate": 9.81500219185509e-06,
      "loss": 0.8243,
      "step": 3966
    },
    {
      "epoch": 0.11430300236270385,
      "grad_norm": 1.5850154161453247,
      "learning_rate": 9.814876419541135e-06,
      "loss": 0.8278,
      "step": 3967
    },
    {
      "epoch": 0.11433181582435314,
      "grad_norm": 1.588485598564148,
      "learning_rate": 9.814750605294421e-06,
      "loss": 0.8457,
      "step": 3968
    },
    {
      "epoch": 0.11436062928600242,
      "grad_norm": 1.4886637926101685,
      "learning_rate": 9.81462474911604e-06,
      "loss": 0.8264,
      "step": 3969
    },
    {
      "epoch": 0.1143894427476517,
      "grad_norm": 1.5134994983673096,
      "learning_rate": 9.81449885100709e-06,
      "loss": 0.8081,
      "step": 3970
    },
    {
      "epoch": 0.11441825620930099,
      "grad_norm": 1.4818391799926758,
      "learning_rate": 9.81437291096867e-06,
      "loss": 0.8104,
      "step": 3971
    },
    {
      "epoch": 0.11444706967095027,
      "grad_norm": 1.3845741748809814,
      "learning_rate": 9.814246929001873e-06,
      "loss": 0.8093,
      "step": 3972
    },
    {
      "epoch": 0.11447588313259956,
      "grad_norm": 1.51103937625885,
      "learning_rate": 9.814120905107797e-06,
      "loss": 0.827,
      "step": 3973
    },
    {
      "epoch": 0.11450469659424883,
      "grad_norm": 1.5771989822387695,
      "learning_rate": 9.813994839287542e-06,
      "loss": 0.8537,
      "step": 3974
    },
    {
      "epoch": 0.11453351005589811,
      "grad_norm": 1.4323829412460327,
      "learning_rate": 9.813868731542204e-06,
      "loss": 0.8064,
      "step": 3975
    },
    {
      "epoch": 0.1145623235175474,
      "grad_norm": 1.5442090034484863,
      "learning_rate": 9.81374258187288e-06,
      "loss": 0.8293,
      "step": 3976
    },
    {
      "epoch": 0.11459113697919668,
      "grad_norm": 1.4949495792388916,
      "learning_rate": 9.813616390280672e-06,
      "loss": 0.811,
      "step": 3977
    },
    {
      "epoch": 0.11461995044084597,
      "grad_norm": 1.5258866548538208,
      "learning_rate": 9.813490156766677e-06,
      "loss": 0.8257,
      "step": 3978
    },
    {
      "epoch": 0.11464876390249525,
      "grad_norm": 1.4837158918380737,
      "learning_rate": 9.813363881331992e-06,
      "loss": 0.8249,
      "step": 3979
    },
    {
      "epoch": 0.11467757736414452,
      "grad_norm": 1.54611337184906,
      "learning_rate": 9.81323756397772e-06,
      "loss": 0.8282,
      "step": 3980
    },
    {
      "epoch": 0.11470639082579381,
      "grad_norm": 1.490296721458435,
      "learning_rate": 9.81311120470496e-06,
      "loss": 0.8277,
      "step": 3981
    },
    {
      "epoch": 0.11473520428744309,
      "grad_norm": 1.529383659362793,
      "learning_rate": 9.812984803514816e-06,
      "loss": 0.8227,
      "step": 3982
    },
    {
      "epoch": 0.11476401774909238,
      "grad_norm": 1.6009713411331177,
      "learning_rate": 9.812858360408384e-06,
      "loss": 0.8291,
      "step": 3983
    },
    {
      "epoch": 0.11479283121074166,
      "grad_norm": 1.468171238899231,
      "learning_rate": 9.812731875386765e-06,
      "loss": 0.831,
      "step": 3984
    },
    {
      "epoch": 0.11482164467239094,
      "grad_norm": 1.7646138668060303,
      "learning_rate": 9.812605348451064e-06,
      "loss": 0.8386,
      "step": 3985
    },
    {
      "epoch": 0.11485045813404023,
      "grad_norm": 1.560049295425415,
      "learning_rate": 9.812478779602382e-06,
      "loss": 0.8369,
      "step": 3986
    },
    {
      "epoch": 0.1148792715956895,
      "grad_norm": 1.5011478662490845,
      "learning_rate": 9.812352168841818e-06,
      "loss": 0.8251,
      "step": 3987
    },
    {
      "epoch": 0.1149080850573388,
      "grad_norm": 1.5426133871078491,
      "learning_rate": 9.81222551617048e-06,
      "loss": 0.8218,
      "step": 3988
    },
    {
      "epoch": 0.11493689851898807,
      "grad_norm": 1.5401848554611206,
      "learning_rate": 9.812098821589467e-06,
      "loss": 0.7881,
      "step": 3989
    },
    {
      "epoch": 0.11496571198063735,
      "grad_norm": 1.4996120929718018,
      "learning_rate": 9.811972085099881e-06,
      "loss": 0.8372,
      "step": 3990
    },
    {
      "epoch": 0.11499452544228664,
      "grad_norm": 1.5601885318756104,
      "learning_rate": 9.811845306702832e-06,
      "loss": 0.8047,
      "step": 3991
    },
    {
      "epoch": 0.11502333890393591,
      "grad_norm": 1.5171566009521484,
      "learning_rate": 9.81171848639942e-06,
      "loss": 0.8271,
      "step": 3992
    },
    {
      "epoch": 0.1150521523655852,
      "grad_norm": 1.6500279903411865,
      "learning_rate": 9.811591624190749e-06,
      "loss": 0.8132,
      "step": 3993
    },
    {
      "epoch": 0.11508096582723448,
      "grad_norm": 1.5808403491973877,
      "learning_rate": 9.811464720077924e-06,
      "loss": 0.8142,
      "step": 3994
    },
    {
      "epoch": 0.11510977928888377,
      "grad_norm": 1.4614983797073364,
      "learning_rate": 9.811337774062052e-06,
      "loss": 0.8184,
      "step": 3995
    },
    {
      "epoch": 0.11513859275053305,
      "grad_norm": 1.6824672222137451,
      "learning_rate": 9.811210786144235e-06,
      "loss": 0.8495,
      "step": 3996
    },
    {
      "epoch": 0.11516740621218233,
      "grad_norm": 1.500536561012268,
      "learning_rate": 9.811083756325585e-06,
      "loss": 0.8021,
      "step": 3997
    },
    {
      "epoch": 0.11519621967383162,
      "grad_norm": 1.4461562633514404,
      "learning_rate": 9.810956684607205e-06,
      "loss": 0.8151,
      "step": 3998
    },
    {
      "epoch": 0.1152250331354809,
      "grad_norm": 1.6606502532958984,
      "learning_rate": 9.8108295709902e-06,
      "loss": 0.8521,
      "step": 3999
    },
    {
      "epoch": 0.11525384659713019,
      "grad_norm": 1.436102032661438,
      "learning_rate": 9.81070241547568e-06,
      "loss": 0.8098,
      "step": 4000
    },
    {
      "epoch": 0.11528266005877946,
      "grad_norm": 1.4665278196334839,
      "learning_rate": 9.810575218064748e-06,
      "loss": 0.8273,
      "step": 4001
    },
    {
      "epoch": 0.11531147352042874,
      "grad_norm": 1.5570106506347656,
      "learning_rate": 9.810447978758517e-06,
      "loss": 0.8278,
      "step": 4002
    },
    {
      "epoch": 0.11534028698207803,
      "grad_norm": 1.459897518157959,
      "learning_rate": 9.810320697558091e-06,
      "loss": 0.8244,
      "step": 4003
    },
    {
      "epoch": 0.11536910044372731,
      "grad_norm": 1.402036428451538,
      "learning_rate": 9.810193374464583e-06,
      "loss": 0.8351,
      "step": 4004
    },
    {
      "epoch": 0.1153979139053766,
      "grad_norm": 1.4549959897994995,
      "learning_rate": 9.810066009479096e-06,
      "loss": 0.8109,
      "step": 4005
    },
    {
      "epoch": 0.11542672736702587,
      "grad_norm": 1.6705129146575928,
      "learning_rate": 9.809938602602743e-06,
      "loss": 0.8244,
      "step": 4006
    },
    {
      "epoch": 0.11545554082867515,
      "grad_norm": 1.4498329162597656,
      "learning_rate": 9.809811153836635e-06,
      "loss": 0.8115,
      "step": 4007
    },
    {
      "epoch": 0.11548435429032444,
      "grad_norm": 1.5473606586456299,
      "learning_rate": 9.80968366318188e-06,
      "loss": 0.8185,
      "step": 4008
    },
    {
      "epoch": 0.11551316775197372,
      "grad_norm": 1.5606496334075928,
      "learning_rate": 9.809556130639585e-06,
      "loss": 0.8226,
      "step": 4009
    },
    {
      "epoch": 0.11554198121362301,
      "grad_norm": 1.4787523746490479,
      "learning_rate": 9.809428556210867e-06,
      "loss": 0.8484,
      "step": 4010
    },
    {
      "epoch": 0.11557079467527229,
      "grad_norm": 1.5563000440597534,
      "learning_rate": 9.809300939896833e-06,
      "loss": 0.7949,
      "step": 4011
    },
    {
      "epoch": 0.11559960813692156,
      "grad_norm": 1.4073975086212158,
      "learning_rate": 9.809173281698595e-06,
      "loss": 0.8376,
      "step": 4012
    },
    {
      "epoch": 0.11562842159857085,
      "grad_norm": 1.4331685304641724,
      "learning_rate": 9.809045581617264e-06,
      "loss": 0.8079,
      "step": 4013
    },
    {
      "epoch": 0.11565723506022013,
      "grad_norm": 1.4953404664993286,
      "learning_rate": 9.808917839653955e-06,
      "loss": 0.8238,
      "step": 4014
    },
    {
      "epoch": 0.11568604852186942,
      "grad_norm": 1.4823724031448364,
      "learning_rate": 9.808790055809779e-06,
      "loss": 0.8236,
      "step": 4015
    },
    {
      "epoch": 0.1157148619835187,
      "grad_norm": 1.5547178983688354,
      "learning_rate": 9.808662230085849e-06,
      "loss": 0.8303,
      "step": 4016
    },
    {
      "epoch": 0.11574367544516798,
      "grad_norm": 1.454906702041626,
      "learning_rate": 9.808534362483275e-06,
      "loss": 0.8411,
      "step": 4017
    },
    {
      "epoch": 0.11577248890681727,
      "grad_norm": 1.4528752565383911,
      "learning_rate": 9.808406453003177e-06,
      "loss": 0.8204,
      "step": 4018
    },
    {
      "epoch": 0.11580130236846654,
      "grad_norm": 1.4905754327774048,
      "learning_rate": 9.808278501646663e-06,
      "loss": 0.8204,
      "step": 4019
    },
    {
      "epoch": 0.11583011583011583,
      "grad_norm": 1.4781506061553955,
      "learning_rate": 9.80815050841485e-06,
      "loss": 0.8222,
      "step": 4020
    },
    {
      "epoch": 0.11585892929176511,
      "grad_norm": 1.6708576679229736,
      "learning_rate": 9.808022473308853e-06,
      "loss": 0.8167,
      "step": 4021
    },
    {
      "epoch": 0.11588774275341439,
      "grad_norm": 1.5179038047790527,
      "learning_rate": 9.807894396329787e-06,
      "loss": 0.8156,
      "step": 4022
    },
    {
      "epoch": 0.11591655621506368,
      "grad_norm": 1.5182924270629883,
      "learning_rate": 9.807766277478767e-06,
      "loss": 0.8461,
      "step": 4023
    },
    {
      "epoch": 0.11594536967671296,
      "grad_norm": 1.4147406816482544,
      "learning_rate": 9.807638116756908e-06,
      "loss": 0.822,
      "step": 4024
    },
    {
      "epoch": 0.11597418313836225,
      "grad_norm": 1.4831773042678833,
      "learning_rate": 9.807509914165328e-06,
      "loss": 0.781,
      "step": 4025
    },
    {
      "epoch": 0.11600299660001152,
      "grad_norm": 1.5049790143966675,
      "learning_rate": 9.807381669705141e-06,
      "loss": 0.8365,
      "step": 4026
    },
    {
      "epoch": 0.11603181006166081,
      "grad_norm": 1.4476957321166992,
      "learning_rate": 9.807253383377466e-06,
      "loss": 0.8354,
      "step": 4027
    },
    {
      "epoch": 0.11606062352331009,
      "grad_norm": 1.4623342752456665,
      "learning_rate": 9.807125055183419e-06,
      "loss": 0.816,
      "step": 4028
    },
    {
      "epoch": 0.11608943698495937,
      "grad_norm": 1.5224717855453491,
      "learning_rate": 9.806996685124117e-06,
      "loss": 0.8123,
      "step": 4029
    },
    {
      "epoch": 0.11611825044660866,
      "grad_norm": 1.5114861726760864,
      "learning_rate": 9.806868273200682e-06,
      "loss": 0.8269,
      "step": 4030
    },
    {
      "epoch": 0.11614706390825794,
      "grad_norm": 1.4615683555603027,
      "learning_rate": 9.806739819414227e-06,
      "loss": 0.8086,
      "step": 4031
    },
    {
      "epoch": 0.11617587736990723,
      "grad_norm": 1.41712486743927,
      "learning_rate": 9.806611323765874e-06,
      "loss": 0.808,
      "step": 4032
    },
    {
      "epoch": 0.1162046908315565,
      "grad_norm": 1.4909543991088867,
      "learning_rate": 9.806482786256742e-06,
      "loss": 0.8396,
      "step": 4033
    },
    {
      "epoch": 0.11623350429320578,
      "grad_norm": 1.3219084739685059,
      "learning_rate": 9.806354206887949e-06,
      "loss": 0.8225,
      "step": 4034
    },
    {
      "epoch": 0.11626231775485507,
      "grad_norm": 1.4308664798736572,
      "learning_rate": 9.806225585660616e-06,
      "loss": 0.8511,
      "step": 4035
    },
    {
      "epoch": 0.11629113121650435,
      "grad_norm": 1.437782645225525,
      "learning_rate": 9.80609692257586e-06,
      "loss": 0.8194,
      "step": 4036
    },
    {
      "epoch": 0.11631994467815364,
      "grad_norm": 1.4235845804214478,
      "learning_rate": 9.805968217634807e-06,
      "loss": 0.8378,
      "step": 4037
    },
    {
      "epoch": 0.11634875813980292,
      "grad_norm": 1.48442542552948,
      "learning_rate": 9.805839470838573e-06,
      "loss": 0.805,
      "step": 4038
    },
    {
      "epoch": 0.11637757160145219,
      "grad_norm": 1.4959659576416016,
      "learning_rate": 9.805710682188285e-06,
      "loss": 0.8385,
      "step": 4039
    },
    {
      "epoch": 0.11640638506310148,
      "grad_norm": 1.5939092636108398,
      "learning_rate": 9.805581851685058e-06,
      "loss": 0.844,
      "step": 4040
    },
    {
      "epoch": 0.11643519852475076,
      "grad_norm": 1.5402803421020508,
      "learning_rate": 9.805452979330017e-06,
      "loss": 0.8026,
      "step": 4041
    },
    {
      "epoch": 0.11646401198640005,
      "grad_norm": 1.4670230150222778,
      "learning_rate": 9.805324065124284e-06,
      "loss": 0.8483,
      "step": 4042
    },
    {
      "epoch": 0.11649282544804933,
      "grad_norm": 1.5425652265548706,
      "learning_rate": 9.805195109068984e-06,
      "loss": 0.8191,
      "step": 4043
    },
    {
      "epoch": 0.1165216389096986,
      "grad_norm": 1.532217264175415,
      "learning_rate": 9.805066111165235e-06,
      "loss": 0.8339,
      "step": 4044
    },
    {
      "epoch": 0.1165504523713479,
      "grad_norm": 1.384225845336914,
      "learning_rate": 9.804937071414167e-06,
      "loss": 0.8302,
      "step": 4045
    },
    {
      "epoch": 0.11657926583299717,
      "grad_norm": 1.5581008195877075,
      "learning_rate": 9.804807989816899e-06,
      "loss": 0.8366,
      "step": 4046
    },
    {
      "epoch": 0.11660807929464646,
      "grad_norm": 1.5434284210205078,
      "learning_rate": 9.804678866374554e-06,
      "loss": 0.8391,
      "step": 4047
    },
    {
      "epoch": 0.11663689275629574,
      "grad_norm": 1.43949556350708,
      "learning_rate": 9.804549701088262e-06,
      "loss": 0.8257,
      "step": 4048
    },
    {
      "epoch": 0.11666570621794502,
      "grad_norm": 1.5678471326828003,
      "learning_rate": 9.804420493959142e-06,
      "loss": 0.8361,
      "step": 4049
    },
    {
      "epoch": 0.11669451967959431,
      "grad_norm": 1.5482745170593262,
      "learning_rate": 9.804291244988324e-06,
      "loss": 0.8452,
      "step": 4050
    },
    {
      "epoch": 0.11672333314124358,
      "grad_norm": 1.483383297920227,
      "learning_rate": 9.804161954176932e-06,
      "loss": 0.8239,
      "step": 4051
    },
    {
      "epoch": 0.11675214660289288,
      "grad_norm": 1.4977145195007324,
      "learning_rate": 9.804032621526092e-06,
      "loss": 0.8292,
      "step": 4052
    },
    {
      "epoch": 0.11678096006454215,
      "grad_norm": 1.3915623426437378,
      "learning_rate": 9.80390324703693e-06,
      "loss": 0.7879,
      "step": 4053
    },
    {
      "epoch": 0.11680977352619144,
      "grad_norm": 1.3848520517349243,
      "learning_rate": 9.803773830710572e-06,
      "loss": 0.8094,
      "step": 4054
    },
    {
      "epoch": 0.11683858698784072,
      "grad_norm": 1.499751329421997,
      "learning_rate": 9.803644372548147e-06,
      "loss": 0.8054,
      "step": 4055
    },
    {
      "epoch": 0.11686740044949,
      "grad_norm": 1.5168293714523315,
      "learning_rate": 9.80351487255078e-06,
      "loss": 0.8208,
      "step": 4056
    },
    {
      "epoch": 0.11689621391113929,
      "grad_norm": 1.4487971067428589,
      "learning_rate": 9.803385330719602e-06,
      "loss": 0.8174,
      "step": 4057
    },
    {
      "epoch": 0.11692502737278856,
      "grad_norm": 1.442983865737915,
      "learning_rate": 9.803255747055738e-06,
      "loss": 0.8293,
      "step": 4058
    },
    {
      "epoch": 0.11695384083443786,
      "grad_norm": 1.4889510869979858,
      "learning_rate": 9.803126121560318e-06,
      "loss": 0.8085,
      "step": 4059
    },
    {
      "epoch": 0.11698265429608713,
      "grad_norm": 1.446043848991394,
      "learning_rate": 9.80299645423447e-06,
      "loss": 0.7985,
      "step": 4060
    },
    {
      "epoch": 0.11701146775773641,
      "grad_norm": 1.525867223739624,
      "learning_rate": 9.802866745079326e-06,
      "loss": 0.8306,
      "step": 4061
    },
    {
      "epoch": 0.1170402812193857,
      "grad_norm": 1.458633303642273,
      "learning_rate": 9.802736994096014e-06,
      "loss": 0.8301,
      "step": 4062
    },
    {
      "epoch": 0.11706909468103498,
      "grad_norm": 1.751839518547058,
      "learning_rate": 9.802607201285663e-06,
      "loss": 0.8452,
      "step": 4063
    },
    {
      "epoch": 0.11709790814268427,
      "grad_norm": 1.3937009572982788,
      "learning_rate": 9.802477366649404e-06,
      "loss": 0.8112,
      "step": 4064
    },
    {
      "epoch": 0.11712672160433354,
      "grad_norm": 1.4789165258407593,
      "learning_rate": 9.802347490188366e-06,
      "loss": 0.8449,
      "step": 4065
    },
    {
      "epoch": 0.11715553506598282,
      "grad_norm": 1.5984163284301758,
      "learning_rate": 9.802217571903685e-06,
      "loss": 0.8208,
      "step": 4066
    },
    {
      "epoch": 0.11718434852763211,
      "grad_norm": 1.359556794166565,
      "learning_rate": 9.802087611796489e-06,
      "loss": 0.8255,
      "step": 4067
    },
    {
      "epoch": 0.11721316198928139,
      "grad_norm": 1.5876981019973755,
      "learning_rate": 9.801957609867907e-06,
      "loss": 0.8415,
      "step": 4068
    },
    {
      "epoch": 0.11724197545093068,
      "grad_norm": 1.5278843641281128,
      "learning_rate": 9.801827566119077e-06,
      "loss": 0.8079,
      "step": 4069
    },
    {
      "epoch": 0.11727078891257996,
      "grad_norm": 1.3732271194458008,
      "learning_rate": 9.80169748055113e-06,
      "loss": 0.8279,
      "step": 4070
    },
    {
      "epoch": 0.11729960237422923,
      "grad_norm": 1.5408574342727661,
      "learning_rate": 9.801567353165196e-06,
      "loss": 0.8309,
      "step": 4071
    },
    {
      "epoch": 0.11732841583587852,
      "grad_norm": 1.552565097808838,
      "learning_rate": 9.801437183962412e-06,
      "loss": 0.8318,
      "step": 4072
    },
    {
      "epoch": 0.1173572292975278,
      "grad_norm": 1.3532752990722656,
      "learning_rate": 9.801306972943909e-06,
      "loss": 0.8021,
      "step": 4073
    },
    {
      "epoch": 0.11738604275917709,
      "grad_norm": 1.5727777481079102,
      "learning_rate": 9.801176720110821e-06,
      "loss": 0.8373,
      "step": 4074
    },
    {
      "epoch": 0.11741485622082637,
      "grad_norm": 1.5581334829330444,
      "learning_rate": 9.801046425464284e-06,
      "loss": 0.816,
      "step": 4075
    },
    {
      "epoch": 0.11744366968247565,
      "grad_norm": 1.4229344129562378,
      "learning_rate": 9.800916089005431e-06,
      "loss": 0.8154,
      "step": 4076
    },
    {
      "epoch": 0.11747248314412494,
      "grad_norm": 1.5368571281433105,
      "learning_rate": 9.800785710735399e-06,
      "loss": 0.8007,
      "step": 4077
    },
    {
      "epoch": 0.11750129660577421,
      "grad_norm": 1.4891289472579956,
      "learning_rate": 9.800655290655322e-06,
      "loss": 0.8158,
      "step": 4078
    },
    {
      "epoch": 0.1175301100674235,
      "grad_norm": 1.5618486404418945,
      "learning_rate": 9.800524828766336e-06,
      "loss": 0.8313,
      "step": 4079
    },
    {
      "epoch": 0.11755892352907278,
      "grad_norm": 1.4810060262680054,
      "learning_rate": 9.800394325069577e-06,
      "loss": 0.8182,
      "step": 4080
    },
    {
      "epoch": 0.11758773699072207,
      "grad_norm": 1.4605231285095215,
      "learning_rate": 9.800263779566184e-06,
      "loss": 0.8449,
      "step": 4081
    },
    {
      "epoch": 0.11761655045237135,
      "grad_norm": 1.4880083799362183,
      "learning_rate": 9.80013319225729e-06,
      "loss": 0.8309,
      "step": 4082
    },
    {
      "epoch": 0.11764536391402063,
      "grad_norm": 1.4462472200393677,
      "learning_rate": 9.800002563144036e-06,
      "loss": 0.8126,
      "step": 4083
    },
    {
      "epoch": 0.11767417737566992,
      "grad_norm": 1.4035141468048096,
      "learning_rate": 9.799871892227557e-06,
      "loss": 0.8412,
      "step": 4084
    },
    {
      "epoch": 0.1177029908373192,
      "grad_norm": 1.427209734916687,
      "learning_rate": 9.799741179508993e-06,
      "loss": 0.8364,
      "step": 4085
    },
    {
      "epoch": 0.11773180429896848,
      "grad_norm": 1.3554978370666504,
      "learning_rate": 9.79961042498948e-06,
      "loss": 0.8437,
      "step": 4086
    },
    {
      "epoch": 0.11776061776061776,
      "grad_norm": 1.5385478734970093,
      "learning_rate": 9.799479628670158e-06,
      "loss": 0.8576,
      "step": 4087
    },
    {
      "epoch": 0.11778943122226704,
      "grad_norm": 1.4312443733215332,
      "learning_rate": 9.799348790552168e-06,
      "loss": 0.817,
      "step": 4088
    },
    {
      "epoch": 0.11781824468391633,
      "grad_norm": 1.3912452459335327,
      "learning_rate": 9.799217910636645e-06,
      "loss": 0.8081,
      "step": 4089
    },
    {
      "epoch": 0.1178470581455656,
      "grad_norm": 1.436722993850708,
      "learning_rate": 9.799086988924733e-06,
      "loss": 0.8257,
      "step": 4090
    },
    {
      "epoch": 0.1178758716072149,
      "grad_norm": 1.4020802974700928,
      "learning_rate": 9.79895602541757e-06,
      "loss": 0.8338,
      "step": 4091
    },
    {
      "epoch": 0.11790468506886417,
      "grad_norm": 1.4164447784423828,
      "learning_rate": 9.798825020116297e-06,
      "loss": 0.818,
      "step": 4092
    },
    {
      "epoch": 0.11793349853051345,
      "grad_norm": 1.443326473236084,
      "learning_rate": 9.798693973022058e-06,
      "loss": 0.8248,
      "step": 4093
    },
    {
      "epoch": 0.11796231199216274,
      "grad_norm": 1.5070992708206177,
      "learning_rate": 9.798562884135988e-06,
      "loss": 0.8057,
      "step": 4094
    },
    {
      "epoch": 0.11799112545381202,
      "grad_norm": 1.4662853479385376,
      "learning_rate": 9.798431753459233e-06,
      "loss": 0.8205,
      "step": 4095
    },
    {
      "epoch": 0.11801993891546131,
      "grad_norm": 1.4808262586593628,
      "learning_rate": 9.798300580992935e-06,
      "loss": 0.8313,
      "step": 4096
    },
    {
      "epoch": 0.11804875237711059,
      "grad_norm": 1.4496630430221558,
      "learning_rate": 9.798169366738235e-06,
      "loss": 0.7916,
      "step": 4097
    },
    {
      "epoch": 0.11807756583875986,
      "grad_norm": 1.426769733428955,
      "learning_rate": 9.798038110696275e-06,
      "loss": 0.8071,
      "step": 4098
    },
    {
      "epoch": 0.11810637930040915,
      "grad_norm": 1.4533661603927612,
      "learning_rate": 9.797906812868202e-06,
      "loss": 0.8124,
      "step": 4099
    },
    {
      "epoch": 0.11813519276205843,
      "grad_norm": 1.4691237211227417,
      "learning_rate": 9.797775473255154e-06,
      "loss": 0.8394,
      "step": 4100
    },
    {
      "epoch": 0.11816400622370772,
      "grad_norm": 1.495896577835083,
      "learning_rate": 9.79764409185828e-06,
      "loss": 0.814,
      "step": 4101
    },
    {
      "epoch": 0.118192819685357,
      "grad_norm": 1.5615626573562622,
      "learning_rate": 9.797512668678719e-06,
      "loss": 0.8182,
      "step": 4102
    },
    {
      "epoch": 0.11822163314700627,
      "grad_norm": 1.4433766603469849,
      "learning_rate": 9.79738120371762e-06,
      "loss": 0.8231,
      "step": 4103
    },
    {
      "epoch": 0.11825044660865557,
      "grad_norm": 1.5956417322158813,
      "learning_rate": 9.797249696976128e-06,
      "loss": 0.829,
      "step": 4104
    },
    {
      "epoch": 0.11827926007030484,
      "grad_norm": 1.4344408512115479,
      "learning_rate": 9.797118148455384e-06,
      "loss": 0.8232,
      "step": 4105
    },
    {
      "epoch": 0.11830807353195413,
      "grad_norm": 1.496087908744812,
      "learning_rate": 9.796986558156538e-06,
      "loss": 0.8312,
      "step": 4106
    },
    {
      "epoch": 0.11833688699360341,
      "grad_norm": 1.484470009803772,
      "learning_rate": 9.796854926080734e-06,
      "loss": 0.7973,
      "step": 4107
    },
    {
      "epoch": 0.1183657004552527,
      "grad_norm": 1.4455326795578003,
      "learning_rate": 9.796723252229117e-06,
      "loss": 0.8002,
      "step": 4108
    },
    {
      "epoch": 0.11839451391690198,
      "grad_norm": 1.5342849493026733,
      "learning_rate": 9.796591536602837e-06,
      "loss": 0.8484,
      "step": 4109
    },
    {
      "epoch": 0.11842332737855125,
      "grad_norm": 1.5316029787063599,
      "learning_rate": 9.79645977920304e-06,
      "loss": 0.8576,
      "step": 4110
    },
    {
      "epoch": 0.11845214084020055,
      "grad_norm": 1.3673385381698608,
      "learning_rate": 9.796327980030873e-06,
      "loss": 0.8161,
      "step": 4111
    },
    {
      "epoch": 0.11848095430184982,
      "grad_norm": 1.476812720298767,
      "learning_rate": 9.796196139087484e-06,
      "loss": 0.8207,
      "step": 4112
    },
    {
      "epoch": 0.11850976776349911,
      "grad_norm": 1.4097282886505127,
      "learning_rate": 9.79606425637402e-06,
      "loss": 0.844,
      "step": 4113
    },
    {
      "epoch": 0.11853858122514839,
      "grad_norm": 1.4378185272216797,
      "learning_rate": 9.79593233189163e-06,
      "loss": 0.7968,
      "step": 4114
    },
    {
      "epoch": 0.11856739468679767,
      "grad_norm": 1.3978147506713867,
      "learning_rate": 9.795800365641465e-06,
      "loss": 0.8109,
      "step": 4115
    },
    {
      "epoch": 0.11859620814844696,
      "grad_norm": 1.436678171157837,
      "learning_rate": 9.795668357624672e-06,
      "loss": 0.8152,
      "step": 4116
    },
    {
      "epoch": 0.11862502161009623,
      "grad_norm": 1.4178051948547363,
      "learning_rate": 9.795536307842403e-06,
      "loss": 0.8287,
      "step": 4117
    },
    {
      "epoch": 0.11865383507174553,
      "grad_norm": 1.3825818300247192,
      "learning_rate": 9.795404216295805e-06,
      "loss": 0.8186,
      "step": 4118
    },
    {
      "epoch": 0.1186826485333948,
      "grad_norm": 1.4459370374679565,
      "learning_rate": 9.79527208298603e-06,
      "loss": 0.8199,
      "step": 4119
    },
    {
      "epoch": 0.11871146199504408,
      "grad_norm": 1.3656014204025269,
      "learning_rate": 9.79513990791423e-06,
      "loss": 0.8031,
      "step": 4120
    },
    {
      "epoch": 0.11874027545669337,
      "grad_norm": 1.4201586246490479,
      "learning_rate": 9.795007691081553e-06,
      "loss": 0.8142,
      "step": 4121
    },
    {
      "epoch": 0.11876908891834265,
      "grad_norm": 1.4313970804214478,
      "learning_rate": 9.794875432489153e-06,
      "loss": 0.83,
      "step": 4122
    },
    {
      "epoch": 0.11879790237999194,
      "grad_norm": 1.4294006824493408,
      "learning_rate": 9.794743132138182e-06,
      "loss": 0.8081,
      "step": 4123
    },
    {
      "epoch": 0.11882671584164121,
      "grad_norm": 1.4559180736541748,
      "learning_rate": 9.79461079002979e-06,
      "loss": 0.8251,
      "step": 4124
    },
    {
      "epoch": 0.11885552930329049,
      "grad_norm": 1.468894124031067,
      "learning_rate": 9.794478406165131e-06,
      "loss": 0.8148,
      "step": 4125
    },
    {
      "epoch": 0.11888434276493978,
      "grad_norm": 1.379024863243103,
      "learning_rate": 9.794345980545359e-06,
      "loss": 0.8269,
      "step": 4126
    },
    {
      "epoch": 0.11891315622658906,
      "grad_norm": 1.4123456478118896,
      "learning_rate": 9.794213513171625e-06,
      "loss": 0.8191,
      "step": 4127
    },
    {
      "epoch": 0.11894196968823835,
      "grad_norm": 1.430040717124939,
      "learning_rate": 9.794081004045083e-06,
      "loss": 0.8011,
      "step": 4128
    },
    {
      "epoch": 0.11897078314988763,
      "grad_norm": 1.4144402742385864,
      "learning_rate": 9.79394845316689e-06,
      "loss": 0.8314,
      "step": 4129
    },
    {
      "epoch": 0.1189995966115369,
      "grad_norm": 1.421124815940857,
      "learning_rate": 9.793815860538197e-06,
      "loss": 0.8202,
      "step": 4130
    },
    {
      "epoch": 0.1190284100731862,
      "grad_norm": 1.4522912502288818,
      "learning_rate": 9.79368322616016e-06,
      "loss": 0.8233,
      "step": 4131
    },
    {
      "epoch": 0.11905722353483547,
      "grad_norm": 1.3422414064407349,
      "learning_rate": 9.793550550033935e-06,
      "loss": 0.8193,
      "step": 4132
    },
    {
      "epoch": 0.11908603699648476,
      "grad_norm": 1.4354873895645142,
      "learning_rate": 9.793417832160675e-06,
      "loss": 0.8192,
      "step": 4133
    },
    {
      "epoch": 0.11911485045813404,
      "grad_norm": 1.358102798461914,
      "learning_rate": 9.793285072541537e-06,
      "loss": 0.8315,
      "step": 4134
    },
    {
      "epoch": 0.11914366391978332,
      "grad_norm": 1.3639332056045532,
      "learning_rate": 9.79315227117768e-06,
      "loss": 0.7907,
      "step": 4135
    },
    {
      "epoch": 0.1191724773814326,
      "grad_norm": 1.3913578987121582,
      "learning_rate": 9.793019428070256e-06,
      "loss": 0.8306,
      "step": 4136
    },
    {
      "epoch": 0.11920129084308188,
      "grad_norm": 1.4525946378707886,
      "learning_rate": 9.792886543220424e-06,
      "loss": 0.8195,
      "step": 4137
    },
    {
      "epoch": 0.11923010430473117,
      "grad_norm": 1.4534778594970703,
      "learning_rate": 9.792753616629342e-06,
      "loss": 0.7979,
      "step": 4138
    },
    {
      "epoch": 0.11925891776638045,
      "grad_norm": 1.4147086143493652,
      "learning_rate": 9.792620648298167e-06,
      "loss": 0.8199,
      "step": 4139
    },
    {
      "epoch": 0.11928773122802974,
      "grad_norm": 1.3340873718261719,
      "learning_rate": 9.792487638228056e-06,
      "loss": 0.8138,
      "step": 4140
    },
    {
      "epoch": 0.11931654468967902,
      "grad_norm": 1.4400705099105835,
      "learning_rate": 9.79235458642017e-06,
      "loss": 0.8388,
      "step": 4141
    },
    {
      "epoch": 0.1193453581513283,
      "grad_norm": 1.4696089029312134,
      "learning_rate": 9.792221492875665e-06,
      "loss": 0.8251,
      "step": 4142
    },
    {
      "epoch": 0.11937417161297759,
      "grad_norm": 1.4497408866882324,
      "learning_rate": 9.792088357595702e-06,
      "loss": 0.8234,
      "step": 4143
    },
    {
      "epoch": 0.11940298507462686,
      "grad_norm": 1.430241584777832,
      "learning_rate": 9.79195518058144e-06,
      "loss": 0.8438,
      "step": 4144
    },
    {
      "epoch": 0.11943179853627615,
      "grad_norm": 1.3752824068069458,
      "learning_rate": 9.791821961834039e-06,
      "loss": 0.8226,
      "step": 4145
    },
    {
      "epoch": 0.11946061199792543,
      "grad_norm": 1.5299161672592163,
      "learning_rate": 9.791688701354656e-06,
      "loss": 0.8282,
      "step": 4146
    },
    {
      "epoch": 0.11948942545957471,
      "grad_norm": 1.4035862684249878,
      "learning_rate": 9.791555399144457e-06,
      "loss": 0.826,
      "step": 4147
    },
    {
      "epoch": 0.119518238921224,
      "grad_norm": 1.3769453763961792,
      "learning_rate": 9.791422055204599e-06,
      "loss": 0.8161,
      "step": 4148
    },
    {
      "epoch": 0.11954705238287328,
      "grad_norm": 1.4524455070495605,
      "learning_rate": 9.791288669536246e-06,
      "loss": 0.783,
      "step": 4149
    },
    {
      "epoch": 0.11957586584452257,
      "grad_norm": 1.5226705074310303,
      "learning_rate": 9.791155242140557e-06,
      "loss": 0.8514,
      "step": 4150
    },
    {
      "epoch": 0.11960467930617184,
      "grad_norm": 1.3916449546813965,
      "learning_rate": 9.791021773018696e-06,
      "loss": 0.8343,
      "step": 4151
    },
    {
      "epoch": 0.11963349276782112,
      "grad_norm": 1.436306118965149,
      "learning_rate": 9.790888262171822e-06,
      "loss": 0.8337,
      "step": 4152
    },
    {
      "epoch": 0.11966230622947041,
      "grad_norm": 1.393684983253479,
      "learning_rate": 9.790754709601103e-06,
      "loss": 0.8312,
      "step": 4153
    },
    {
      "epoch": 0.11969111969111969,
      "grad_norm": 1.4693940877914429,
      "learning_rate": 9.7906211153077e-06,
      "loss": 0.8125,
      "step": 4154
    },
    {
      "epoch": 0.11971993315276898,
      "grad_norm": 1.3683874607086182,
      "learning_rate": 9.790487479292773e-06,
      "loss": 0.8201,
      "step": 4155
    },
    {
      "epoch": 0.11974874661441826,
      "grad_norm": 1.4126055240631104,
      "learning_rate": 9.790353801557493e-06,
      "loss": 0.8316,
      "step": 4156
    },
    {
      "epoch": 0.11977756007606753,
      "grad_norm": 1.3760191202163696,
      "learning_rate": 9.790220082103017e-06,
      "loss": 0.8371,
      "step": 4157
    },
    {
      "epoch": 0.11980637353771682,
      "grad_norm": 1.375126838684082,
      "learning_rate": 9.790086320930514e-06,
      "loss": 0.8258,
      "step": 4158
    },
    {
      "epoch": 0.1198351869993661,
      "grad_norm": 1.4424365758895874,
      "learning_rate": 9.789952518041146e-06,
      "loss": 0.8224,
      "step": 4159
    },
    {
      "epoch": 0.11986400046101539,
      "grad_norm": 1.445020079612732,
      "learning_rate": 9.78981867343608e-06,
      "loss": 0.8128,
      "step": 4160
    },
    {
      "epoch": 0.11989281392266467,
      "grad_norm": 1.4244699478149414,
      "learning_rate": 9.789684787116483e-06,
      "loss": 0.8227,
      "step": 4161
    },
    {
      "epoch": 0.11992162738431394,
      "grad_norm": 1.4406815767288208,
      "learning_rate": 9.789550859083517e-06,
      "loss": 0.7993,
      "step": 4162
    },
    {
      "epoch": 0.11995044084596324,
      "grad_norm": 1.4081907272338867,
      "learning_rate": 9.789416889338352e-06,
      "loss": 0.8122,
      "step": 4163
    },
    {
      "epoch": 0.11997925430761251,
      "grad_norm": 1.5353708267211914,
      "learning_rate": 9.789282877882155e-06,
      "loss": 0.8209,
      "step": 4164
    },
    {
      "epoch": 0.1200080677692618,
      "grad_norm": 1.3336111307144165,
      "learning_rate": 9.789148824716091e-06,
      "loss": 0.8266,
      "step": 4165
    },
    {
      "epoch": 0.12003688123091108,
      "grad_norm": 1.4306033849716187,
      "learning_rate": 9.789014729841327e-06,
      "loss": 0.8121,
      "step": 4166
    },
    {
      "epoch": 0.12006569469256037,
      "grad_norm": 1.3701211214065552,
      "learning_rate": 9.788880593259031e-06,
      "loss": 0.8139,
      "step": 4167
    },
    {
      "epoch": 0.12009450815420965,
      "grad_norm": 1.4881443977355957,
      "learning_rate": 9.788746414970374e-06,
      "loss": 0.8384,
      "step": 4168
    },
    {
      "epoch": 0.12012332161585892,
      "grad_norm": 1.4315885305404663,
      "learning_rate": 9.788612194976523e-06,
      "loss": 0.8155,
      "step": 4169
    },
    {
      "epoch": 0.12015213507750822,
      "grad_norm": 1.483365535736084,
      "learning_rate": 9.788477933278647e-06,
      "loss": 0.8005,
      "step": 4170
    },
    {
      "epoch": 0.12018094853915749,
      "grad_norm": 1.4758542776107788,
      "learning_rate": 9.788343629877913e-06,
      "loss": 0.8082,
      "step": 4171
    },
    {
      "epoch": 0.12020976200080678,
      "grad_norm": 1.3819180727005005,
      "learning_rate": 9.788209284775494e-06,
      "loss": 0.8012,
      "step": 4172
    },
    {
      "epoch": 0.12023857546245606,
      "grad_norm": 1.360020399093628,
      "learning_rate": 9.788074897972558e-06,
      "loss": 0.8165,
      "step": 4173
    },
    {
      "epoch": 0.12026738892410534,
      "grad_norm": 1.3977608680725098,
      "learning_rate": 9.787940469470277e-06,
      "loss": 0.8105,
      "step": 4174
    },
    {
      "epoch": 0.12029620238575463,
      "grad_norm": 1.3584743738174438,
      "learning_rate": 9.78780599926982e-06,
      "loss": 0.7982,
      "step": 4175
    },
    {
      "epoch": 0.1203250158474039,
      "grad_norm": 1.4216499328613281,
      "learning_rate": 9.787671487372356e-06,
      "loss": 0.8351,
      "step": 4176
    },
    {
      "epoch": 0.1203538293090532,
      "grad_norm": 1.3967887163162231,
      "learning_rate": 9.787536933779063e-06,
      "loss": 0.8233,
      "step": 4177
    },
    {
      "epoch": 0.12038264277070247,
      "grad_norm": 1.3989925384521484,
      "learning_rate": 9.787402338491107e-06,
      "loss": 0.814,
      "step": 4178
    },
    {
      "epoch": 0.12041145623235175,
      "grad_norm": 1.460754632949829,
      "learning_rate": 9.787267701509665e-06,
      "loss": 0.8142,
      "step": 4179
    },
    {
      "epoch": 0.12044026969400104,
      "grad_norm": 1.382116436958313,
      "learning_rate": 9.787133022835903e-06,
      "loss": 0.8188,
      "step": 4180
    },
    {
      "epoch": 0.12046908315565032,
      "grad_norm": 1.500829815864563,
      "learning_rate": 9.786998302471e-06,
      "loss": 0.8104,
      "step": 4181
    },
    {
      "epoch": 0.12049789661729961,
      "grad_norm": 1.3734279870986938,
      "learning_rate": 9.786863540416128e-06,
      "loss": 0.8216,
      "step": 4182
    },
    {
      "epoch": 0.12052671007894888,
      "grad_norm": 1.4524632692337036,
      "learning_rate": 9.786728736672458e-06,
      "loss": 0.8077,
      "step": 4183
    },
    {
      "epoch": 0.12055552354059816,
      "grad_norm": 1.5213263034820557,
      "learning_rate": 9.786593891241167e-06,
      "loss": 0.818,
      "step": 4184
    },
    {
      "epoch": 0.12058433700224745,
      "grad_norm": 1.3975697755813599,
      "learning_rate": 9.786459004123427e-06,
      "loss": 0.8422,
      "step": 4185
    },
    {
      "epoch": 0.12061315046389673,
      "grad_norm": 1.49543297290802,
      "learning_rate": 9.786324075320415e-06,
      "loss": 0.8334,
      "step": 4186
    },
    {
      "epoch": 0.12064196392554602,
      "grad_norm": 1.4032883644104004,
      "learning_rate": 9.786189104833303e-06,
      "loss": 0.814,
      "step": 4187
    },
    {
      "epoch": 0.1206707773871953,
      "grad_norm": 1.4156733751296997,
      "learning_rate": 9.786054092663268e-06,
      "loss": 0.8292,
      "step": 4188
    },
    {
      "epoch": 0.12069959084884457,
      "grad_norm": 1.3898799419403076,
      "learning_rate": 9.78591903881149e-06,
      "loss": 0.814,
      "step": 4189
    },
    {
      "epoch": 0.12072840431049386,
      "grad_norm": 1.4285895824432373,
      "learning_rate": 9.785783943279137e-06,
      "loss": 0.8281,
      "step": 4190
    },
    {
      "epoch": 0.12075721777214314,
      "grad_norm": 1.4514673948287964,
      "learning_rate": 9.785648806067391e-06,
      "loss": 0.8016,
      "step": 4191
    },
    {
      "epoch": 0.12078603123379243,
      "grad_norm": 1.3831478357315063,
      "learning_rate": 9.78551362717743e-06,
      "loss": 0.7799,
      "step": 4192
    },
    {
      "epoch": 0.12081484469544171,
      "grad_norm": 1.4232800006866455,
      "learning_rate": 9.785378406610425e-06,
      "loss": 0.8236,
      "step": 4193
    },
    {
      "epoch": 0.120843658157091,
      "grad_norm": 1.3937088251113892,
      "learning_rate": 9.785243144367562e-06,
      "loss": 0.8108,
      "step": 4194
    },
    {
      "epoch": 0.12087247161874028,
      "grad_norm": 1.3895955085754395,
      "learning_rate": 9.785107840450012e-06,
      "loss": 0.8127,
      "step": 4195
    },
    {
      "epoch": 0.12090128508038955,
      "grad_norm": 1.3555567264556885,
      "learning_rate": 9.784972494858957e-06,
      "loss": 0.8254,
      "step": 4196
    },
    {
      "epoch": 0.12093009854203884,
      "grad_norm": 1.3320726156234741,
      "learning_rate": 9.784837107595574e-06,
      "loss": 0.8264,
      "step": 4197
    },
    {
      "epoch": 0.12095891200368812,
      "grad_norm": 1.4651048183441162,
      "learning_rate": 9.784701678661045e-06,
      "loss": 0.816,
      "step": 4198
    },
    {
      "epoch": 0.12098772546533741,
      "grad_norm": 1.4319429397583008,
      "learning_rate": 9.784566208056547e-06,
      "loss": 0.8035,
      "step": 4199
    },
    {
      "epoch": 0.12101653892698669,
      "grad_norm": 1.4244110584259033,
      "learning_rate": 9.784430695783257e-06,
      "loss": 0.8252,
      "step": 4200
    },
    {
      "epoch": 0.12104535238863597,
      "grad_norm": 1.4099422693252563,
      "learning_rate": 9.784295141842362e-06,
      "loss": 0.8294,
      "step": 4201
    },
    {
      "epoch": 0.12107416585028526,
      "grad_norm": 1.4238163232803345,
      "learning_rate": 9.784159546235037e-06,
      "loss": 0.7939,
      "step": 4202
    },
    {
      "epoch": 0.12110297931193453,
      "grad_norm": 1.3997597694396973,
      "learning_rate": 9.784023908962466e-06,
      "loss": 0.8059,
      "step": 4203
    },
    {
      "epoch": 0.12113179277358382,
      "grad_norm": 1.4560635089874268,
      "learning_rate": 9.783888230025827e-06,
      "loss": 0.8251,
      "step": 4204
    },
    {
      "epoch": 0.1211606062352331,
      "grad_norm": 1.5192121267318726,
      "learning_rate": 9.783752509426306e-06,
      "loss": 0.8435,
      "step": 4205
    },
    {
      "epoch": 0.12118941969688238,
      "grad_norm": 1.4056023359298706,
      "learning_rate": 9.78361674716508e-06,
      "loss": 0.8032,
      "step": 4206
    },
    {
      "epoch": 0.12121823315853167,
      "grad_norm": 1.4249180555343628,
      "learning_rate": 9.783480943243336e-06,
      "loss": 0.8322,
      "step": 4207
    },
    {
      "epoch": 0.12124704662018095,
      "grad_norm": 1.4094693660736084,
      "learning_rate": 9.783345097662253e-06,
      "loss": 0.8084,
      "step": 4208
    },
    {
      "epoch": 0.12127586008183024,
      "grad_norm": 1.4576928615570068,
      "learning_rate": 9.783209210423018e-06,
      "loss": 0.8177,
      "step": 4209
    },
    {
      "epoch": 0.12130467354347951,
      "grad_norm": 1.3874107599258423,
      "learning_rate": 9.78307328152681e-06,
      "loss": 0.8325,
      "step": 4210
    },
    {
      "epoch": 0.12133348700512879,
      "grad_norm": 1.421809434890747,
      "learning_rate": 9.782937310974816e-06,
      "loss": 0.8182,
      "step": 4211
    },
    {
      "epoch": 0.12136230046677808,
      "grad_norm": 1.464823842048645,
      "learning_rate": 9.78280129876822e-06,
      "loss": 0.8389,
      "step": 4212
    },
    {
      "epoch": 0.12139111392842736,
      "grad_norm": 1.4464792013168335,
      "learning_rate": 9.782665244908206e-06,
      "loss": 0.8374,
      "step": 4213
    },
    {
      "epoch": 0.12141992739007665,
      "grad_norm": 1.3602787256240845,
      "learning_rate": 9.782529149395955e-06,
      "loss": 0.806,
      "step": 4214
    },
    {
      "epoch": 0.12144874085172593,
      "grad_norm": 1.421167016029358,
      "learning_rate": 9.78239301223266e-06,
      "loss": 0.8205,
      "step": 4215
    },
    {
      "epoch": 0.1214775543133752,
      "grad_norm": 1.467236042022705,
      "learning_rate": 9.7822568334195e-06,
      "loss": 0.8228,
      "step": 4216
    },
    {
      "epoch": 0.12150636777502449,
      "grad_norm": 1.4739961624145508,
      "learning_rate": 9.782120612957666e-06,
      "loss": 0.8027,
      "step": 4217
    },
    {
      "epoch": 0.12153518123667377,
      "grad_norm": 1.3619449138641357,
      "learning_rate": 9.781984350848341e-06,
      "loss": 0.8313,
      "step": 4218
    },
    {
      "epoch": 0.12156399469832306,
      "grad_norm": 1.4406895637512207,
      "learning_rate": 9.781848047092712e-06,
      "loss": 0.8323,
      "step": 4219
    },
    {
      "epoch": 0.12159280815997234,
      "grad_norm": 1.4246010780334473,
      "learning_rate": 9.781711701691966e-06,
      "loss": 0.8255,
      "step": 4220
    },
    {
      "epoch": 0.12162162162162163,
      "grad_norm": 1.4335949420928955,
      "learning_rate": 9.781575314647291e-06,
      "loss": 0.8042,
      "step": 4221
    },
    {
      "epoch": 0.1216504350832709,
      "grad_norm": 1.405441403388977,
      "learning_rate": 9.781438885959876e-06,
      "loss": 0.8138,
      "step": 4222
    },
    {
      "epoch": 0.12167924854492018,
      "grad_norm": 1.408002257347107,
      "learning_rate": 9.781302415630908e-06,
      "loss": 0.8147,
      "step": 4223
    },
    {
      "epoch": 0.12170806200656947,
      "grad_norm": 1.3441075086593628,
      "learning_rate": 9.781165903661575e-06,
      "loss": 0.7983,
      "step": 4224
    },
    {
      "epoch": 0.12173687546821875,
      "grad_norm": 1.4292436838150024,
      "learning_rate": 9.781029350053067e-06,
      "loss": 0.8155,
      "step": 4225
    },
    {
      "epoch": 0.12176568892986804,
      "grad_norm": 1.4494497776031494,
      "learning_rate": 9.780892754806571e-06,
      "loss": 0.8339,
      "step": 4226
    },
    {
      "epoch": 0.12179450239151732,
      "grad_norm": 1.4433443546295166,
      "learning_rate": 9.78075611792328e-06,
      "loss": 0.8038,
      "step": 4227
    },
    {
      "epoch": 0.1218233158531666,
      "grad_norm": 1.416361689567566,
      "learning_rate": 9.780619439404382e-06,
      "loss": 0.8127,
      "step": 4228
    },
    {
      "epoch": 0.12185212931481589,
      "grad_norm": 1.438289999961853,
      "learning_rate": 9.780482719251066e-06,
      "loss": 0.8007,
      "step": 4229
    },
    {
      "epoch": 0.12188094277646516,
      "grad_norm": 1.5003483295440674,
      "learning_rate": 9.780345957464526e-06,
      "loss": 0.8111,
      "step": 4230
    },
    {
      "epoch": 0.12190975623811445,
      "grad_norm": 1.4187445640563965,
      "learning_rate": 9.78020915404595e-06,
      "loss": 0.8055,
      "step": 4231
    },
    {
      "epoch": 0.12193856969976373,
      "grad_norm": 1.4424866437911987,
      "learning_rate": 9.780072308996531e-06,
      "loss": 0.8181,
      "step": 4232
    },
    {
      "epoch": 0.121967383161413,
      "grad_norm": 1.4247426986694336,
      "learning_rate": 9.779935422317461e-06,
      "loss": 0.8271,
      "step": 4233
    },
    {
      "epoch": 0.1219961966230623,
      "grad_norm": 1.4060351848602295,
      "learning_rate": 9.77979849400993e-06,
      "loss": 0.7838,
      "step": 4234
    },
    {
      "epoch": 0.12202501008471157,
      "grad_norm": 1.5211173295974731,
      "learning_rate": 9.779661524075135e-06,
      "loss": 0.8202,
      "step": 4235
    },
    {
      "epoch": 0.12205382354636087,
      "grad_norm": 1.424377202987671,
      "learning_rate": 9.779524512514265e-06,
      "loss": 0.8442,
      "step": 4236
    },
    {
      "epoch": 0.12208263700801014,
      "grad_norm": 1.503057837486267,
      "learning_rate": 9.779387459328513e-06,
      "loss": 0.801,
      "step": 4237
    },
    {
      "epoch": 0.12211145046965942,
      "grad_norm": 1.4771602153778076,
      "learning_rate": 9.779250364519075e-06,
      "loss": 0.7982,
      "step": 4238
    },
    {
      "epoch": 0.12214026393130871,
      "grad_norm": 1.4989913702011108,
      "learning_rate": 9.779113228087145e-06,
      "loss": 0.8397,
      "step": 4239
    },
    {
      "epoch": 0.12216907739295799,
      "grad_norm": 1.4468543529510498,
      "learning_rate": 9.778976050033914e-06,
      "loss": 0.817,
      "step": 4240
    },
    {
      "epoch": 0.12219789085460728,
      "grad_norm": 1.3602161407470703,
      "learning_rate": 9.77883883036058e-06,
      "loss": 0.819,
      "step": 4241
    },
    {
      "epoch": 0.12222670431625655,
      "grad_norm": 1.4085792303085327,
      "learning_rate": 9.778701569068336e-06,
      "loss": 0.837,
      "step": 4242
    },
    {
      "epoch": 0.12225551777790583,
      "grad_norm": 1.4442414045333862,
      "learning_rate": 9.778564266158379e-06,
      "loss": 0.8413,
      "step": 4243
    },
    {
      "epoch": 0.12228433123955512,
      "grad_norm": 1.4355357885360718,
      "learning_rate": 9.778426921631903e-06,
      "loss": 0.8354,
      "step": 4244
    },
    {
      "epoch": 0.1223131447012044,
      "grad_norm": 1.3684805631637573,
      "learning_rate": 9.778289535490107e-06,
      "loss": 0.8184,
      "step": 4245
    },
    {
      "epoch": 0.12234195816285369,
      "grad_norm": 1.3278104066848755,
      "learning_rate": 9.778152107734184e-06,
      "loss": 0.8357,
      "step": 4246
    },
    {
      "epoch": 0.12237077162450297,
      "grad_norm": 1.4776643514633179,
      "learning_rate": 9.778014638365334e-06,
      "loss": 0.8455,
      "step": 4247
    },
    {
      "epoch": 0.12239958508615226,
      "grad_norm": 1.386718988418579,
      "learning_rate": 9.777877127384752e-06,
      "loss": 0.8265,
      "step": 4248
    },
    {
      "epoch": 0.12242839854780153,
      "grad_norm": 1.3434103727340698,
      "learning_rate": 9.777739574793637e-06,
      "loss": 0.8034,
      "step": 4249
    },
    {
      "epoch": 0.12245721200945081,
      "grad_norm": 1.4468289613723755,
      "learning_rate": 9.777601980593186e-06,
      "loss": 0.8228,
      "step": 4250
    },
    {
      "epoch": 0.1224860254711001,
      "grad_norm": 1.3880646228790283,
      "learning_rate": 9.777464344784598e-06,
      "loss": 0.8097,
      "step": 4251
    },
    {
      "epoch": 0.12251483893274938,
      "grad_norm": 1.4214439392089844,
      "learning_rate": 9.77732666736907e-06,
      "loss": 0.8122,
      "step": 4252
    },
    {
      "epoch": 0.12254365239439867,
      "grad_norm": 1.3505632877349854,
      "learning_rate": 9.777188948347802e-06,
      "loss": 0.834,
      "step": 4253
    },
    {
      "epoch": 0.12257246585604795,
      "grad_norm": 1.3497648239135742,
      "learning_rate": 9.777051187721996e-06,
      "loss": 0.8215,
      "step": 4254
    },
    {
      "epoch": 0.12260127931769722,
      "grad_norm": 1.3244696855545044,
      "learning_rate": 9.776913385492846e-06,
      "loss": 0.8405,
      "step": 4255
    },
    {
      "epoch": 0.12263009277934651,
      "grad_norm": 1.3705182075500488,
      "learning_rate": 9.776775541661558e-06,
      "loss": 0.847,
      "step": 4256
    },
    {
      "epoch": 0.12265890624099579,
      "grad_norm": 1.4318625926971436,
      "learning_rate": 9.77663765622933e-06,
      "loss": 0.8111,
      "step": 4257
    },
    {
      "epoch": 0.12268771970264508,
      "grad_norm": 1.498822808265686,
      "learning_rate": 9.776499729197362e-06,
      "loss": 0.8063,
      "step": 4258
    },
    {
      "epoch": 0.12271653316429436,
      "grad_norm": 1.4625681638717651,
      "learning_rate": 9.776361760566857e-06,
      "loss": 0.8135,
      "step": 4259
    },
    {
      "epoch": 0.12274534662594364,
      "grad_norm": 1.3615719079971313,
      "learning_rate": 9.776223750339016e-06,
      "loss": 0.7917,
      "step": 4260
    },
    {
      "epoch": 0.12277416008759293,
      "grad_norm": 1.3469115495681763,
      "learning_rate": 9.776085698515039e-06,
      "loss": 0.8116,
      "step": 4261
    },
    {
      "epoch": 0.1228029735492422,
      "grad_norm": 1.4218871593475342,
      "learning_rate": 9.775947605096132e-06,
      "loss": 0.8139,
      "step": 4262
    },
    {
      "epoch": 0.1228317870108915,
      "grad_norm": 1.4367015361785889,
      "learning_rate": 9.775809470083494e-06,
      "loss": 0.7969,
      "step": 4263
    },
    {
      "epoch": 0.12286060047254077,
      "grad_norm": 1.4123457670211792,
      "learning_rate": 9.77567129347833e-06,
      "loss": 0.8433,
      "step": 4264
    },
    {
      "epoch": 0.12288941393419005,
      "grad_norm": 1.3900383710861206,
      "learning_rate": 9.775533075281842e-06,
      "loss": 0.8178,
      "step": 4265
    },
    {
      "epoch": 0.12291822739583934,
      "grad_norm": 1.4611669778823853,
      "learning_rate": 9.775394815495236e-06,
      "loss": 0.8319,
      "step": 4266
    },
    {
      "epoch": 0.12294704085748862,
      "grad_norm": 1.3550388813018799,
      "learning_rate": 9.775256514119716e-06,
      "loss": 0.8243,
      "step": 4267
    },
    {
      "epoch": 0.1229758543191379,
      "grad_norm": 1.395760178565979,
      "learning_rate": 9.775118171156483e-06,
      "loss": 0.7958,
      "step": 4268
    },
    {
      "epoch": 0.12300466778078718,
      "grad_norm": 1.3888192176818848,
      "learning_rate": 9.774979786606745e-06,
      "loss": 0.8002,
      "step": 4269
    },
    {
      "epoch": 0.12303348124243646,
      "grad_norm": 1.3578197956085205,
      "learning_rate": 9.774841360471705e-06,
      "loss": 0.8047,
      "step": 4270
    },
    {
      "epoch": 0.12306229470408575,
      "grad_norm": 1.4163790941238403,
      "learning_rate": 9.774702892752572e-06,
      "loss": 0.8,
      "step": 4271
    },
    {
      "epoch": 0.12309110816573503,
      "grad_norm": 1.3234823942184448,
      "learning_rate": 9.77456438345055e-06,
      "loss": 0.7935,
      "step": 4272
    },
    {
      "epoch": 0.12311992162738432,
      "grad_norm": 1.435966968536377,
      "learning_rate": 9.774425832566843e-06,
      "loss": 0.8297,
      "step": 4273
    },
    {
      "epoch": 0.1231487350890336,
      "grad_norm": 1.4207277297973633,
      "learning_rate": 9.77428724010266e-06,
      "loss": 0.8073,
      "step": 4274
    },
    {
      "epoch": 0.12317754855068287,
      "grad_norm": 1.3644397258758545,
      "learning_rate": 9.774148606059209e-06,
      "loss": 0.8363,
      "step": 4275
    },
    {
      "epoch": 0.12320636201233216,
      "grad_norm": 1.3434873819351196,
      "learning_rate": 9.774009930437693e-06,
      "loss": 0.8229,
      "step": 4276
    },
    {
      "epoch": 0.12323517547398144,
      "grad_norm": 1.3900203704833984,
      "learning_rate": 9.773871213239325e-06,
      "loss": 0.7983,
      "step": 4277
    },
    {
      "epoch": 0.12326398893563073,
      "grad_norm": 1.3767669200897217,
      "learning_rate": 9.773732454465313e-06,
      "loss": 0.8066,
      "step": 4278
    },
    {
      "epoch": 0.12329280239728001,
      "grad_norm": 1.3660787343978882,
      "learning_rate": 9.77359365411686e-06,
      "loss": 0.8019,
      "step": 4279
    },
    {
      "epoch": 0.1233216158589293,
      "grad_norm": 1.3784899711608887,
      "learning_rate": 9.77345481219518e-06,
      "loss": 0.8353,
      "step": 4280
    },
    {
      "epoch": 0.12335042932057858,
      "grad_norm": 1.4264670610427856,
      "learning_rate": 9.773315928701478e-06,
      "loss": 0.8224,
      "step": 4281
    },
    {
      "epoch": 0.12337924278222785,
      "grad_norm": 1.348930835723877,
      "learning_rate": 9.773177003636968e-06,
      "loss": 0.7964,
      "step": 4282
    },
    {
      "epoch": 0.12340805624387714,
      "grad_norm": 1.3159431219100952,
      "learning_rate": 9.773038037002858e-06,
      "loss": 0.8238,
      "step": 4283
    },
    {
      "epoch": 0.12343686970552642,
      "grad_norm": 1.4315217733383179,
      "learning_rate": 9.772899028800355e-06,
      "loss": 0.8279,
      "step": 4284
    },
    {
      "epoch": 0.12346568316717571,
      "grad_norm": 1.358641505241394,
      "learning_rate": 9.772759979030676e-06,
      "loss": 0.8052,
      "step": 4285
    },
    {
      "epoch": 0.12349449662882499,
      "grad_norm": 1.3810803890228271,
      "learning_rate": 9.772620887695028e-06,
      "loss": 0.8277,
      "step": 4286
    },
    {
      "epoch": 0.12352331009047426,
      "grad_norm": 1.454258918762207,
      "learning_rate": 9.772481754794622e-06,
      "loss": 0.8287,
      "step": 4287
    },
    {
      "epoch": 0.12355212355212356,
      "grad_norm": 1.3526324033737183,
      "learning_rate": 9.772342580330671e-06,
      "loss": 0.8003,
      "step": 4288
    },
    {
      "epoch": 0.12358093701377283,
      "grad_norm": 1.3849809169769287,
      "learning_rate": 9.772203364304387e-06,
      "loss": 0.8151,
      "step": 4289
    },
    {
      "epoch": 0.12360975047542212,
      "grad_norm": 1.3367847204208374,
      "learning_rate": 9.77206410671698e-06,
      "loss": 0.8236,
      "step": 4290
    },
    {
      "epoch": 0.1236385639370714,
      "grad_norm": 1.3458330631256104,
      "learning_rate": 9.771924807569668e-06,
      "loss": 0.8112,
      "step": 4291
    },
    {
      "epoch": 0.12366737739872068,
      "grad_norm": 1.3558404445648193,
      "learning_rate": 9.77178546686366e-06,
      "loss": 0.8049,
      "step": 4292
    },
    {
      "epoch": 0.12369619086036997,
      "grad_norm": 1.2652479410171509,
      "learning_rate": 9.77164608460017e-06,
      "loss": 0.8035,
      "step": 4293
    },
    {
      "epoch": 0.12372500432201924,
      "grad_norm": 1.31302011013031,
      "learning_rate": 9.771506660780412e-06,
      "loss": 0.8109,
      "step": 4294
    },
    {
      "epoch": 0.12375381778366853,
      "grad_norm": 1.3107906579971313,
      "learning_rate": 9.771367195405602e-06,
      "loss": 0.8128,
      "step": 4295
    },
    {
      "epoch": 0.12378263124531781,
      "grad_norm": 1.4110796451568604,
      "learning_rate": 9.771227688476954e-06,
      "loss": 0.8298,
      "step": 4296
    },
    {
      "epoch": 0.12381144470696709,
      "grad_norm": 1.353963017463684,
      "learning_rate": 9.771088139995682e-06,
      "loss": 0.8229,
      "step": 4297
    },
    {
      "epoch": 0.12384025816861638,
      "grad_norm": 1.3183140754699707,
      "learning_rate": 9.770948549963e-06,
      "loss": 0.8358,
      "step": 4298
    },
    {
      "epoch": 0.12386907163026566,
      "grad_norm": 1.467310905456543,
      "learning_rate": 9.770808918380126e-06,
      "loss": 0.8086,
      "step": 4299
    },
    {
      "epoch": 0.12389788509191495,
      "grad_norm": 1.3318430185317993,
      "learning_rate": 9.770669245248276e-06,
      "loss": 0.8278,
      "step": 4300
    },
    {
      "epoch": 0.12392669855356422,
      "grad_norm": 1.3771713972091675,
      "learning_rate": 9.770529530568665e-06,
      "loss": 0.8055,
      "step": 4301
    },
    {
      "epoch": 0.1239555120152135,
      "grad_norm": 1.313978910446167,
      "learning_rate": 9.770389774342511e-06,
      "loss": 0.8311,
      "step": 4302
    },
    {
      "epoch": 0.12398432547686279,
      "grad_norm": 1.3888635635375977,
      "learning_rate": 9.770249976571032e-06,
      "loss": 0.8063,
      "step": 4303
    },
    {
      "epoch": 0.12401313893851207,
      "grad_norm": 1.3872463703155518,
      "learning_rate": 9.770110137255442e-06,
      "loss": 0.8232,
      "step": 4304
    },
    {
      "epoch": 0.12404195240016136,
      "grad_norm": 1.3262810707092285,
      "learning_rate": 9.769970256396962e-06,
      "loss": 0.8361,
      "step": 4305
    },
    {
      "epoch": 0.12407076586181064,
      "grad_norm": 1.4002997875213623,
      "learning_rate": 9.76983033399681e-06,
      "loss": 0.8079,
      "step": 4306
    },
    {
      "epoch": 0.12409957932345993,
      "grad_norm": 1.409175157546997,
      "learning_rate": 9.769690370056203e-06,
      "loss": 0.8205,
      "step": 4307
    },
    {
      "epoch": 0.1241283927851092,
      "grad_norm": 1.4114680290222168,
      "learning_rate": 9.76955036457636e-06,
      "loss": 0.799,
      "step": 4308
    },
    {
      "epoch": 0.12415720624675848,
      "grad_norm": 1.3919386863708496,
      "learning_rate": 9.769410317558504e-06,
      "loss": 0.8378,
      "step": 4309
    },
    {
      "epoch": 0.12418601970840777,
      "grad_norm": 1.3986990451812744,
      "learning_rate": 9.769270229003849e-06,
      "loss": 0.8005,
      "step": 4310
    },
    {
      "epoch": 0.12421483317005705,
      "grad_norm": 1.4359281063079834,
      "learning_rate": 9.769130098913619e-06,
      "loss": 0.8384,
      "step": 4311
    },
    {
      "epoch": 0.12424364663170634,
      "grad_norm": 1.4527860879898071,
      "learning_rate": 9.768989927289033e-06,
      "loss": 0.8184,
      "step": 4312
    },
    {
      "epoch": 0.12427246009335562,
      "grad_norm": 1.38182532787323,
      "learning_rate": 9.768849714131312e-06,
      "loss": 0.8018,
      "step": 4313
    },
    {
      "epoch": 0.12430127355500489,
      "grad_norm": 1.464151382446289,
      "learning_rate": 9.768709459441677e-06,
      "loss": 0.8017,
      "step": 4314
    },
    {
      "epoch": 0.12433008701665418,
      "grad_norm": 1.649126648902893,
      "learning_rate": 9.76856916322135e-06,
      "loss": 0.7887,
      "step": 4315
    },
    {
      "epoch": 0.12435890047830346,
      "grad_norm": 1.410646915435791,
      "learning_rate": 9.768428825471552e-06,
      "loss": 0.7916,
      "step": 4316
    },
    {
      "epoch": 0.12438771393995275,
      "grad_norm": 1.4584026336669922,
      "learning_rate": 9.768288446193505e-06,
      "loss": 0.8346,
      "step": 4317
    },
    {
      "epoch": 0.12441652740160203,
      "grad_norm": 1.415579915046692,
      "learning_rate": 9.768148025388434e-06,
      "loss": 0.7953,
      "step": 4318
    },
    {
      "epoch": 0.1244453408632513,
      "grad_norm": 1.4379197359085083,
      "learning_rate": 9.768007563057558e-06,
      "loss": 0.8373,
      "step": 4319
    },
    {
      "epoch": 0.1244741543249006,
      "grad_norm": 1.368126630783081,
      "learning_rate": 9.767867059202102e-06,
      "loss": 0.8051,
      "step": 4320
    },
    {
      "epoch": 0.12450296778654987,
      "grad_norm": 1.423019289970398,
      "learning_rate": 9.767726513823294e-06,
      "loss": 0.8285,
      "step": 4321
    },
    {
      "epoch": 0.12453178124819916,
      "grad_norm": 1.3982319831848145,
      "learning_rate": 9.76758592692235e-06,
      "loss": 0.8224,
      "step": 4322
    },
    {
      "epoch": 0.12456059470984844,
      "grad_norm": 1.321178674697876,
      "learning_rate": 9.767445298500501e-06,
      "loss": 0.8271,
      "step": 4323
    },
    {
      "epoch": 0.12458940817149772,
      "grad_norm": 1.416747808456421,
      "learning_rate": 9.767304628558967e-06,
      "loss": 0.8153,
      "step": 4324
    },
    {
      "epoch": 0.12461822163314701,
      "grad_norm": 1.4245522022247314,
      "learning_rate": 9.767163917098976e-06,
      "loss": 0.823,
      "step": 4325
    },
    {
      "epoch": 0.12464703509479629,
      "grad_norm": 1.3476732969284058,
      "learning_rate": 9.767023164121754e-06,
      "loss": 0.8088,
      "step": 4326
    },
    {
      "epoch": 0.12467584855644558,
      "grad_norm": 1.4267048835754395,
      "learning_rate": 9.766882369628524e-06,
      "loss": 0.8147,
      "step": 4327
    },
    {
      "epoch": 0.12470466201809485,
      "grad_norm": 1.3506534099578857,
      "learning_rate": 9.766741533620514e-06,
      "loss": 0.8059,
      "step": 4328
    },
    {
      "epoch": 0.12473347547974413,
      "grad_norm": 1.325116515159607,
      "learning_rate": 9.766600656098952e-06,
      "loss": 0.8226,
      "step": 4329
    },
    {
      "epoch": 0.12476228894139342,
      "grad_norm": 1.3613018989562988,
      "learning_rate": 9.76645973706506e-06,
      "loss": 0.8211,
      "step": 4330
    },
    {
      "epoch": 0.1247911024030427,
      "grad_norm": 1.4021272659301758,
      "learning_rate": 9.766318776520071e-06,
      "loss": 0.8379,
      "step": 4331
    },
    {
      "epoch": 0.12481991586469199,
      "grad_norm": 1.3411500453948975,
      "learning_rate": 9.766177774465209e-06,
      "loss": 0.8143,
      "step": 4332
    },
    {
      "epoch": 0.12484872932634127,
      "grad_norm": 1.3858520984649658,
      "learning_rate": 9.766036730901702e-06,
      "loss": 0.8295,
      "step": 4333
    },
    {
      "epoch": 0.12487754278799056,
      "grad_norm": 1.3550363779067993,
      "learning_rate": 9.765895645830781e-06,
      "loss": 0.8147,
      "step": 4334
    },
    {
      "epoch": 0.12490635624963983,
      "grad_norm": 1.4100061655044556,
      "learning_rate": 9.765754519253672e-06,
      "loss": 0.8196,
      "step": 4335
    },
    {
      "epoch": 0.12493516971128911,
      "grad_norm": 1.403235912322998,
      "learning_rate": 9.765613351171605e-06,
      "loss": 0.8231,
      "step": 4336
    },
    {
      "epoch": 0.1249639831729384,
      "grad_norm": 1.3785446882247925,
      "learning_rate": 9.765472141585813e-06,
      "loss": 0.8097,
      "step": 4337
    },
    {
      "epoch": 0.12499279663458768,
      "grad_norm": 1.394309639930725,
      "learning_rate": 9.765330890497518e-06,
      "loss": 0.8144,
      "step": 4338
    },
    {
      "epoch": 0.12502161009623697,
      "grad_norm": 1.3701808452606201,
      "learning_rate": 9.765189597907956e-06,
      "loss": 0.7846,
      "step": 4339
    },
    {
      "epoch": 0.12505042355788623,
      "grad_norm": 1.3073307275772095,
      "learning_rate": 9.765048263818356e-06,
      "loss": 0.8117,
      "step": 4340
    },
    {
      "epoch": 0.12507923701953552,
      "grad_norm": 1.4408283233642578,
      "learning_rate": 9.76490688822995e-06,
      "loss": 0.8497,
      "step": 4341
    },
    {
      "epoch": 0.1251080504811848,
      "grad_norm": 1.3545986413955688,
      "learning_rate": 9.764765471143967e-06,
      "loss": 0.83,
      "step": 4342
    },
    {
      "epoch": 0.1251368639428341,
      "grad_norm": 1.409261703491211,
      "learning_rate": 9.76462401256164e-06,
      "loss": 0.7797,
      "step": 4343
    },
    {
      "epoch": 0.12516567740448337,
      "grad_norm": 1.3997719287872314,
      "learning_rate": 9.7644825124842e-06,
      "loss": 0.8246,
      "step": 4344
    },
    {
      "epoch": 0.12519449086613266,
      "grad_norm": 1.3361351490020752,
      "learning_rate": 9.764340970912881e-06,
      "loss": 0.8296,
      "step": 4345
    },
    {
      "epoch": 0.12522330432778195,
      "grad_norm": 1.429281234741211,
      "learning_rate": 9.764199387848915e-06,
      "loss": 0.8353,
      "step": 4346
    },
    {
      "epoch": 0.1252521177894312,
      "grad_norm": 1.3681437969207764,
      "learning_rate": 9.764057763293535e-06,
      "loss": 0.8245,
      "step": 4347
    },
    {
      "epoch": 0.1252809312510805,
      "grad_norm": 1.338557243347168,
      "learning_rate": 9.763916097247972e-06,
      "loss": 0.8245,
      "step": 4348
    },
    {
      "epoch": 0.1253097447127298,
      "grad_norm": 1.3786226511001587,
      "learning_rate": 9.763774389713464e-06,
      "loss": 0.8062,
      "step": 4349
    },
    {
      "epoch": 0.12533855817437908,
      "grad_norm": 1.3090920448303223,
      "learning_rate": 9.763632640691242e-06,
      "loss": 0.8332,
      "step": 4350
    },
    {
      "epoch": 0.12536737163602835,
      "grad_norm": 1.3846261501312256,
      "learning_rate": 9.763490850182543e-06,
      "loss": 0.8282,
      "step": 4351
    },
    {
      "epoch": 0.12539618509767764,
      "grad_norm": 1.4644635915756226,
      "learning_rate": 9.763349018188599e-06,
      "loss": 0.8159,
      "step": 4352
    },
    {
      "epoch": 0.12542499855932693,
      "grad_norm": 1.3363094329833984,
      "learning_rate": 9.763207144710648e-06,
      "loss": 0.7932,
      "step": 4353
    },
    {
      "epoch": 0.1254538120209762,
      "grad_norm": 1.2997665405273438,
      "learning_rate": 9.763065229749924e-06,
      "loss": 0.8239,
      "step": 4354
    },
    {
      "epoch": 0.12548262548262548,
      "grad_norm": 1.462196946144104,
      "learning_rate": 9.762923273307663e-06,
      "loss": 0.8059,
      "step": 4355
    },
    {
      "epoch": 0.12551143894427477,
      "grad_norm": 1.3730812072753906,
      "learning_rate": 9.762781275385101e-06,
      "loss": 0.7835,
      "step": 4356
    },
    {
      "epoch": 0.12554025240592404,
      "grad_norm": 1.2827173471450806,
      "learning_rate": 9.762639235983476e-06,
      "loss": 0.7905,
      "step": 4357
    },
    {
      "epoch": 0.12556906586757333,
      "grad_norm": 1.4012292623519897,
      "learning_rate": 9.762497155104023e-06,
      "loss": 0.7987,
      "step": 4358
    },
    {
      "epoch": 0.12559787932922262,
      "grad_norm": 1.3722591400146484,
      "learning_rate": 9.76235503274798e-06,
      "loss": 0.8391,
      "step": 4359
    },
    {
      "epoch": 0.1256266927908719,
      "grad_norm": 1.3252407312393188,
      "learning_rate": 9.762212868916586e-06,
      "loss": 0.8208,
      "step": 4360
    },
    {
      "epoch": 0.12565550625252117,
      "grad_norm": 1.441201090812683,
      "learning_rate": 9.762070663611078e-06,
      "loss": 0.8326,
      "step": 4361
    },
    {
      "epoch": 0.12568431971417046,
      "grad_norm": 1.362013339996338,
      "learning_rate": 9.761928416832696e-06,
      "loss": 0.8026,
      "step": 4362
    },
    {
      "epoch": 0.12571313317581975,
      "grad_norm": 1.4402843713760376,
      "learning_rate": 9.761786128582678e-06,
      "loss": 0.8159,
      "step": 4363
    },
    {
      "epoch": 0.12574194663746902,
      "grad_norm": 1.451636791229248,
      "learning_rate": 9.76164379886226e-06,
      "loss": 0.8266,
      "step": 4364
    },
    {
      "epoch": 0.1257707600991183,
      "grad_norm": 1.4359359741210938,
      "learning_rate": 9.761501427672689e-06,
      "loss": 0.827,
      "step": 4365
    },
    {
      "epoch": 0.1257995735607676,
      "grad_norm": 1.3729755878448486,
      "learning_rate": 9.761359015015198e-06,
      "loss": 0.8477,
      "step": 4366
    },
    {
      "epoch": 0.12582838702241686,
      "grad_norm": 1.477636456489563,
      "learning_rate": 9.761216560891028e-06,
      "loss": 0.8306,
      "step": 4367
    },
    {
      "epoch": 0.12585720048406615,
      "grad_norm": 1.3849284648895264,
      "learning_rate": 9.761074065301422e-06,
      "loss": 0.8095,
      "step": 4368
    },
    {
      "epoch": 0.12588601394571544,
      "grad_norm": 1.3358359336853027,
      "learning_rate": 9.760931528247622e-06,
      "loss": 0.7971,
      "step": 4369
    },
    {
      "epoch": 0.12591482740736473,
      "grad_norm": 1.5124173164367676,
      "learning_rate": 9.760788949730866e-06,
      "loss": 0.8001,
      "step": 4370
    },
    {
      "epoch": 0.125943640869014,
      "grad_norm": 1.325782060623169,
      "learning_rate": 9.760646329752397e-06,
      "loss": 0.8082,
      "step": 4371
    },
    {
      "epoch": 0.12597245433066329,
      "grad_norm": 1.322878122329712,
      "learning_rate": 9.760503668313458e-06,
      "loss": 0.8014,
      "step": 4372
    },
    {
      "epoch": 0.12600126779231258,
      "grad_norm": 1.4551342725753784,
      "learning_rate": 9.760360965415291e-06,
      "loss": 0.7969,
      "step": 4373
    },
    {
      "epoch": 0.12603008125396184,
      "grad_norm": 1.335073471069336,
      "learning_rate": 9.760218221059139e-06,
      "loss": 0.828,
      "step": 4374
    },
    {
      "epoch": 0.12605889471561113,
      "grad_norm": 1.4396151304244995,
      "learning_rate": 9.760075435246243e-06,
      "loss": 0.8392,
      "step": 4375
    },
    {
      "epoch": 0.12608770817726042,
      "grad_norm": 1.4241559505462646,
      "learning_rate": 9.759932607977851e-06,
      "loss": 0.8196,
      "step": 4376
    },
    {
      "epoch": 0.1261165216389097,
      "grad_norm": 1.3829185962677002,
      "learning_rate": 9.759789739255204e-06,
      "loss": 0.8104,
      "step": 4377
    },
    {
      "epoch": 0.12614533510055898,
      "grad_norm": 1.3255919218063354,
      "learning_rate": 9.759646829079543e-06,
      "loss": 0.8295,
      "step": 4378
    },
    {
      "epoch": 0.12617414856220827,
      "grad_norm": 1.3494678735733032,
      "learning_rate": 9.759503877452118e-06,
      "loss": 0.8093,
      "step": 4379
    },
    {
      "epoch": 0.12620296202385756,
      "grad_norm": 1.3604559898376465,
      "learning_rate": 9.759360884374174e-06,
      "loss": 0.841,
      "step": 4380
    },
    {
      "epoch": 0.12623177548550682,
      "grad_norm": 1.3919429779052734,
      "learning_rate": 9.759217849846951e-06,
      "loss": 0.7995,
      "step": 4381
    },
    {
      "epoch": 0.1262605889471561,
      "grad_norm": 1.3833519220352173,
      "learning_rate": 9.7590747738717e-06,
      "loss": 0.7951,
      "step": 4382
    },
    {
      "epoch": 0.1262894024088054,
      "grad_norm": 1.3383092880249023,
      "learning_rate": 9.758931656449664e-06,
      "loss": 0.8007,
      "step": 4383
    },
    {
      "epoch": 0.12631821587045466,
      "grad_norm": 1.36739981174469,
      "learning_rate": 9.758788497582091e-06,
      "loss": 0.8114,
      "step": 4384
    },
    {
      "epoch": 0.12634702933210396,
      "grad_norm": 1.396626353263855,
      "learning_rate": 9.758645297270228e-06,
      "loss": 0.8049,
      "step": 4385
    },
    {
      "epoch": 0.12637584279375325,
      "grad_norm": 1.3352946043014526,
      "learning_rate": 9.75850205551532e-06,
      "loss": 0.8143,
      "step": 4386
    },
    {
      "epoch": 0.12640465625540254,
      "grad_norm": 1.3149458169937134,
      "learning_rate": 9.758358772318617e-06,
      "loss": 0.8157,
      "step": 4387
    },
    {
      "epoch": 0.1264334697170518,
      "grad_norm": 1.371046543121338,
      "learning_rate": 9.758215447681365e-06,
      "loss": 0.8151,
      "step": 4388
    },
    {
      "epoch": 0.1264622831787011,
      "grad_norm": 1.3678374290466309,
      "learning_rate": 9.758072081604812e-06,
      "loss": 0.8137,
      "step": 4389
    },
    {
      "epoch": 0.12649109664035038,
      "grad_norm": 1.3215093612670898,
      "learning_rate": 9.757928674090209e-06,
      "loss": 0.843,
      "step": 4390
    },
    {
      "epoch": 0.12651991010199964,
      "grad_norm": 1.2550458908081055,
      "learning_rate": 9.757785225138803e-06,
      "loss": 0.7909,
      "step": 4391
    },
    {
      "epoch": 0.12654872356364893,
      "grad_norm": 1.3324471712112427,
      "learning_rate": 9.757641734751844e-06,
      "loss": 0.7858,
      "step": 4392
    },
    {
      "epoch": 0.12657753702529823,
      "grad_norm": 1.3329172134399414,
      "learning_rate": 9.757498202930582e-06,
      "loss": 0.8328,
      "step": 4393
    },
    {
      "epoch": 0.1266063504869475,
      "grad_norm": 1.3870060443878174,
      "learning_rate": 9.757354629676265e-06,
      "loss": 0.8047,
      "step": 4394
    },
    {
      "epoch": 0.12663516394859678,
      "grad_norm": 1.277984380722046,
      "learning_rate": 9.757211014990144e-06,
      "loss": 0.7646,
      "step": 4395
    },
    {
      "epoch": 0.12666397741024607,
      "grad_norm": 1.5115931034088135,
      "learning_rate": 9.757067358873474e-06,
      "loss": 0.8161,
      "step": 4396
    },
    {
      "epoch": 0.12669279087189536,
      "grad_norm": 1.367793321609497,
      "learning_rate": 9.7569236613275e-06,
      "loss": 0.796,
      "step": 4397
    },
    {
      "epoch": 0.12672160433354462,
      "grad_norm": 1.306928277015686,
      "learning_rate": 9.756779922353477e-06,
      "loss": 0.8125,
      "step": 4398
    },
    {
      "epoch": 0.12675041779519391,
      "grad_norm": 1.3549787998199463,
      "learning_rate": 9.756636141952655e-06,
      "loss": 0.813,
      "step": 4399
    },
    {
      "epoch": 0.1267792312568432,
      "grad_norm": 1.372145175933838,
      "learning_rate": 9.756492320126287e-06,
      "loss": 0.8237,
      "step": 4400
    },
    {
      "epoch": 0.12680804471849247,
      "grad_norm": 1.4616492986679077,
      "learning_rate": 9.756348456875626e-06,
      "loss": 0.8387,
      "step": 4401
    },
    {
      "epoch": 0.12683685818014176,
      "grad_norm": 1.2723666429519653,
      "learning_rate": 9.756204552201926e-06,
      "loss": 0.8353,
      "step": 4402
    },
    {
      "epoch": 0.12686567164179105,
      "grad_norm": 1.2960985898971558,
      "learning_rate": 9.756060606106437e-06,
      "loss": 0.8106,
      "step": 4403
    },
    {
      "epoch": 0.12689448510344034,
      "grad_norm": 1.3974261283874512,
      "learning_rate": 9.755916618590416e-06,
      "loss": 0.8094,
      "step": 4404
    },
    {
      "epoch": 0.1269232985650896,
      "grad_norm": 1.3131699562072754,
      "learning_rate": 9.755772589655114e-06,
      "loss": 0.8184,
      "step": 4405
    },
    {
      "epoch": 0.1269521120267389,
      "grad_norm": 1.3488775491714478,
      "learning_rate": 9.755628519301787e-06,
      "loss": 0.8105,
      "step": 4406
    },
    {
      "epoch": 0.12698092548838819,
      "grad_norm": 1.350017786026001,
      "learning_rate": 9.755484407531691e-06,
      "loss": 0.7848,
      "step": 4407
    },
    {
      "epoch": 0.12700973895003745,
      "grad_norm": 1.3560642004013062,
      "learning_rate": 9.755340254346077e-06,
      "loss": 0.8067,
      "step": 4408
    },
    {
      "epoch": 0.12703855241168674,
      "grad_norm": 1.3483901023864746,
      "learning_rate": 9.755196059746204e-06,
      "loss": 0.8387,
      "step": 4409
    },
    {
      "epoch": 0.12706736587333603,
      "grad_norm": 1.431336522102356,
      "learning_rate": 9.755051823733327e-06,
      "loss": 0.8034,
      "step": 4410
    },
    {
      "epoch": 0.1270961793349853,
      "grad_norm": 1.3956013917922974,
      "learning_rate": 9.754907546308703e-06,
      "loss": 0.789,
      "step": 4411
    },
    {
      "epoch": 0.12712499279663458,
      "grad_norm": 1.3238147497177124,
      "learning_rate": 9.754763227473585e-06,
      "loss": 0.8059,
      "step": 4412
    },
    {
      "epoch": 0.12715380625828387,
      "grad_norm": 1.3319231271743774,
      "learning_rate": 9.754618867229234e-06,
      "loss": 0.8091,
      "step": 4413
    },
    {
      "epoch": 0.12718261971993317,
      "grad_norm": 1.407665729522705,
      "learning_rate": 9.754474465576906e-06,
      "loss": 0.8084,
      "step": 4414
    },
    {
      "epoch": 0.12721143318158243,
      "grad_norm": 1.3505380153656006,
      "learning_rate": 9.754330022517857e-06,
      "loss": 0.7879,
      "step": 4415
    },
    {
      "epoch": 0.12724024664323172,
      "grad_norm": 1.3470585346221924,
      "learning_rate": 9.754185538053347e-06,
      "loss": 0.8182,
      "step": 4416
    },
    {
      "epoch": 0.127269060104881,
      "grad_norm": 1.3169735670089722,
      "learning_rate": 9.754041012184632e-06,
      "loss": 0.8411,
      "step": 4417
    },
    {
      "epoch": 0.12729787356653027,
      "grad_norm": 1.3790346384048462,
      "learning_rate": 9.753896444912974e-06,
      "loss": 0.8035,
      "step": 4418
    },
    {
      "epoch": 0.12732668702817956,
      "grad_norm": 1.3224685192108154,
      "learning_rate": 9.753751836239629e-06,
      "loss": 0.8142,
      "step": 4419
    },
    {
      "epoch": 0.12735550048982885,
      "grad_norm": 1.4550931453704834,
      "learning_rate": 9.753607186165856e-06,
      "loss": 0.8334,
      "step": 4420
    },
    {
      "epoch": 0.12738431395147812,
      "grad_norm": 1.4864962100982666,
      "learning_rate": 9.75346249469292e-06,
      "loss": 0.8112,
      "step": 4421
    },
    {
      "epoch": 0.1274131274131274,
      "grad_norm": 1.333570957183838,
      "learning_rate": 9.753317761822075e-06,
      "loss": 0.8139,
      "step": 4422
    },
    {
      "epoch": 0.1274419408747767,
      "grad_norm": 1.3507391214370728,
      "learning_rate": 9.753172987554585e-06,
      "loss": 0.8091,
      "step": 4423
    },
    {
      "epoch": 0.127470754336426,
      "grad_norm": 1.4602346420288086,
      "learning_rate": 9.753028171891707e-06,
      "loss": 0.8111,
      "step": 4424
    },
    {
      "epoch": 0.12749956779807525,
      "grad_norm": 1.3634371757507324,
      "learning_rate": 9.752883314834707e-06,
      "loss": 0.7966,
      "step": 4425
    },
    {
      "epoch": 0.12752838125972454,
      "grad_norm": 1.4039390087127686,
      "learning_rate": 9.752738416384845e-06,
      "loss": 0.8236,
      "step": 4426
    },
    {
      "epoch": 0.12755719472137383,
      "grad_norm": 1.4375190734863281,
      "learning_rate": 9.752593476543382e-06,
      "loss": 0.8143,
      "step": 4427
    },
    {
      "epoch": 0.1275860081830231,
      "grad_norm": 1.3732445240020752,
      "learning_rate": 9.752448495311579e-06,
      "loss": 0.8243,
      "step": 4428
    },
    {
      "epoch": 0.1276148216446724,
      "grad_norm": 1.415186882019043,
      "learning_rate": 9.752303472690703e-06,
      "loss": 0.7912,
      "step": 4429
    },
    {
      "epoch": 0.12764363510632168,
      "grad_norm": 1.3897913694381714,
      "learning_rate": 9.752158408682013e-06,
      "loss": 0.8205,
      "step": 4430
    },
    {
      "epoch": 0.12767244856797097,
      "grad_norm": 1.3505498170852661,
      "learning_rate": 9.752013303286774e-06,
      "loss": 0.8087,
      "step": 4431
    },
    {
      "epoch": 0.12770126202962023,
      "grad_norm": 1.4802745580673218,
      "learning_rate": 9.751868156506249e-06,
      "loss": 0.8085,
      "step": 4432
    },
    {
      "epoch": 0.12773007549126952,
      "grad_norm": 1.3979412317276,
      "learning_rate": 9.751722968341701e-06,
      "loss": 0.8122,
      "step": 4433
    },
    {
      "epoch": 0.12775888895291881,
      "grad_norm": 1.3012524843215942,
      "learning_rate": 9.751577738794397e-06,
      "loss": 0.8011,
      "step": 4434
    },
    {
      "epoch": 0.12778770241456808,
      "grad_norm": 1.3722965717315674,
      "learning_rate": 9.751432467865603e-06,
      "loss": 0.8578,
      "step": 4435
    },
    {
      "epoch": 0.12781651587621737,
      "grad_norm": 1.466106653213501,
      "learning_rate": 9.75128715555658e-06,
      "loss": 0.8111,
      "step": 4436
    },
    {
      "epoch": 0.12784532933786666,
      "grad_norm": 1.5074971914291382,
      "learning_rate": 9.751141801868596e-06,
      "loss": 0.8057,
      "step": 4437
    },
    {
      "epoch": 0.12787414279951592,
      "grad_norm": 1.3546777963638306,
      "learning_rate": 9.750996406802915e-06,
      "loss": 0.8082,
      "step": 4438
    },
    {
      "epoch": 0.1279029562611652,
      "grad_norm": 1.3665218353271484,
      "learning_rate": 9.750850970360805e-06,
      "loss": 0.8095,
      "step": 4439
    },
    {
      "epoch": 0.1279317697228145,
      "grad_norm": 1.3769145011901855,
      "learning_rate": 9.750705492543533e-06,
      "loss": 0.8236,
      "step": 4440
    },
    {
      "epoch": 0.1279605831844638,
      "grad_norm": 1.3538416624069214,
      "learning_rate": 9.750559973352366e-06,
      "loss": 0.8233,
      "step": 4441
    },
    {
      "epoch": 0.12798939664611306,
      "grad_norm": 1.4097014665603638,
      "learning_rate": 9.750414412788568e-06,
      "loss": 0.8066,
      "step": 4442
    },
    {
      "epoch": 0.12801821010776235,
      "grad_norm": 1.345666527748108,
      "learning_rate": 9.75026881085341e-06,
      "loss": 0.8198,
      "step": 4443
    },
    {
      "epoch": 0.12804702356941164,
      "grad_norm": 1.2848701477050781,
      "learning_rate": 9.75012316754816e-06,
      "loss": 0.7928,
      "step": 4444
    },
    {
      "epoch": 0.1280758370310609,
      "grad_norm": 1.3527541160583496,
      "learning_rate": 9.749977482874085e-06,
      "loss": 0.8392,
      "step": 4445
    },
    {
      "epoch": 0.1281046504927102,
      "grad_norm": 1.339941382408142,
      "learning_rate": 9.749831756832455e-06,
      "loss": 0.8076,
      "step": 4446
    },
    {
      "epoch": 0.12813346395435948,
      "grad_norm": 1.368994951248169,
      "learning_rate": 9.749685989424537e-06,
      "loss": 0.809,
      "step": 4447
    },
    {
      "epoch": 0.12816227741600875,
      "grad_norm": 1.379884123802185,
      "learning_rate": 9.749540180651604e-06,
      "loss": 0.8292,
      "step": 4448
    },
    {
      "epoch": 0.12819109087765804,
      "grad_norm": 1.3313522338867188,
      "learning_rate": 9.749394330514921e-06,
      "loss": 0.813,
      "step": 4449
    },
    {
      "epoch": 0.12821990433930733,
      "grad_norm": 1.319328784942627,
      "learning_rate": 9.749248439015763e-06,
      "loss": 0.8087,
      "step": 4450
    },
    {
      "epoch": 0.12824871780095662,
      "grad_norm": 1.2930752038955688,
      "learning_rate": 9.749102506155398e-06,
      "loss": 0.8206,
      "step": 4451
    },
    {
      "epoch": 0.12827753126260588,
      "grad_norm": 1.2910150289535522,
      "learning_rate": 9.748956531935099e-06,
      "loss": 0.8418,
      "step": 4452
    },
    {
      "epoch": 0.12830634472425517,
      "grad_norm": 1.3341689109802246,
      "learning_rate": 9.748810516356136e-06,
      "loss": 0.8196,
      "step": 4453
    },
    {
      "epoch": 0.12833515818590446,
      "grad_norm": 1.2933412790298462,
      "learning_rate": 9.748664459419778e-06,
      "loss": 0.8252,
      "step": 4454
    },
    {
      "epoch": 0.12836397164755373,
      "grad_norm": 1.2828271389007568,
      "learning_rate": 9.748518361127302e-06,
      "loss": 0.8219,
      "step": 4455
    },
    {
      "epoch": 0.12839278510920302,
      "grad_norm": 1.3021329641342163,
      "learning_rate": 9.748372221479976e-06,
      "loss": 0.8215,
      "step": 4456
    },
    {
      "epoch": 0.1284215985708523,
      "grad_norm": 1.3556185960769653,
      "learning_rate": 9.748226040479075e-06,
      "loss": 0.8361,
      "step": 4457
    },
    {
      "epoch": 0.1284504120325016,
      "grad_norm": 1.338245153427124,
      "learning_rate": 9.748079818125871e-06,
      "loss": 0.798,
      "step": 4458
    },
    {
      "epoch": 0.12847922549415086,
      "grad_norm": 1.3368607759475708,
      "learning_rate": 9.747933554421638e-06,
      "loss": 0.8213,
      "step": 4459
    },
    {
      "epoch": 0.12850803895580015,
      "grad_norm": 1.4597785472869873,
      "learning_rate": 9.747787249367653e-06,
      "loss": 0.8265,
      "step": 4460
    },
    {
      "epoch": 0.12853685241744944,
      "grad_norm": 1.3935624361038208,
      "learning_rate": 9.747640902965185e-06,
      "loss": 0.809,
      "step": 4461
    },
    {
      "epoch": 0.1285656658790987,
      "grad_norm": 1.410588026046753,
      "learning_rate": 9.74749451521551e-06,
      "loss": 0.8043,
      "step": 4462
    },
    {
      "epoch": 0.128594479340748,
      "grad_norm": 1.3696215152740479,
      "learning_rate": 9.747348086119903e-06,
      "loss": 0.7991,
      "step": 4463
    },
    {
      "epoch": 0.1286232928023973,
      "grad_norm": 1.3531494140625,
      "learning_rate": 9.747201615679642e-06,
      "loss": 0.8127,
      "step": 4464
    },
    {
      "epoch": 0.12865210626404655,
      "grad_norm": 1.4179248809814453,
      "learning_rate": 9.747055103896e-06,
      "loss": 0.8007,
      "step": 4465
    },
    {
      "epoch": 0.12868091972569584,
      "grad_norm": 1.377912163734436,
      "learning_rate": 9.746908550770252e-06,
      "loss": 0.8198,
      "step": 4466
    },
    {
      "epoch": 0.12870973318734513,
      "grad_norm": 1.356845498085022,
      "learning_rate": 9.746761956303677e-06,
      "loss": 0.8187,
      "step": 4467
    },
    {
      "epoch": 0.12873854664899442,
      "grad_norm": 1.3529331684112549,
      "learning_rate": 9.746615320497549e-06,
      "loss": 0.8182,
      "step": 4468
    },
    {
      "epoch": 0.12876736011064369,
      "grad_norm": 1.4130690097808838,
      "learning_rate": 9.746468643353148e-06,
      "loss": 0.8046,
      "step": 4469
    },
    {
      "epoch": 0.12879617357229298,
      "grad_norm": 1.3003544807434082,
      "learning_rate": 9.746321924871749e-06,
      "loss": 0.8049,
      "step": 4470
    },
    {
      "epoch": 0.12882498703394227,
      "grad_norm": 1.3458144664764404,
      "learning_rate": 9.74617516505463e-06,
      "loss": 0.7844,
      "step": 4471
    },
    {
      "epoch": 0.12885380049559153,
      "grad_norm": 1.355210304260254,
      "learning_rate": 9.74602836390307e-06,
      "loss": 0.7922,
      "step": 4472
    },
    {
      "epoch": 0.12888261395724082,
      "grad_norm": 6.971841812133789,
      "learning_rate": 9.745881521418348e-06,
      "loss": 0.8373,
      "step": 4473
    },
    {
      "epoch": 0.1289114274188901,
      "grad_norm": 1.3422720432281494,
      "learning_rate": 9.745734637601742e-06,
      "loss": 0.7977,
      "step": 4474
    },
    {
      "epoch": 0.12894024088053938,
      "grad_norm": 1.4299565553665161,
      "learning_rate": 9.745587712454531e-06,
      "loss": 0.7955,
      "step": 4475
    },
    {
      "epoch": 0.12896905434218867,
      "grad_norm": 1.394027829170227,
      "learning_rate": 9.745440745977994e-06,
      "loss": 0.8122,
      "step": 4476
    },
    {
      "epoch": 0.12899786780383796,
      "grad_norm": 1.336716890335083,
      "learning_rate": 9.745293738173414e-06,
      "loss": 0.7974,
      "step": 4477
    },
    {
      "epoch": 0.12902668126548725,
      "grad_norm": 1.4303876161575317,
      "learning_rate": 9.745146689042068e-06,
      "loss": 0.8432,
      "step": 4478
    },
    {
      "epoch": 0.1290554947271365,
      "grad_norm": 1.5277177095413208,
      "learning_rate": 9.744999598585237e-06,
      "loss": 0.804,
      "step": 4479
    },
    {
      "epoch": 0.1290843081887858,
      "grad_norm": 1.381229043006897,
      "learning_rate": 9.744852466804205e-06,
      "loss": 0.8178,
      "step": 4480
    },
    {
      "epoch": 0.1291131216504351,
      "grad_norm": 1.3194814920425415,
      "learning_rate": 9.744705293700249e-06,
      "loss": 0.7968,
      "step": 4481
    },
    {
      "epoch": 0.12914193511208436,
      "grad_norm": 1.4000928401947021,
      "learning_rate": 9.744558079274654e-06,
      "loss": 0.8195,
      "step": 4482
    },
    {
      "epoch": 0.12917074857373365,
      "grad_norm": 1.4072520732879639,
      "learning_rate": 9.744410823528698e-06,
      "loss": 0.8489,
      "step": 4483
    },
    {
      "epoch": 0.12919956203538294,
      "grad_norm": 1.3475475311279297,
      "learning_rate": 9.74426352646367e-06,
      "loss": 0.8336,
      "step": 4484
    },
    {
      "epoch": 0.1292283754970322,
      "grad_norm": 1.34982430934906,
      "learning_rate": 9.744116188080848e-06,
      "loss": 0.8035,
      "step": 4485
    },
    {
      "epoch": 0.1292571889586815,
      "grad_norm": 1.3814722299575806,
      "learning_rate": 9.743968808381517e-06,
      "loss": 0.8193,
      "step": 4486
    },
    {
      "epoch": 0.12928600242033078,
      "grad_norm": 1.3402875661849976,
      "learning_rate": 9.743821387366958e-06,
      "loss": 0.821,
      "step": 4487
    },
    {
      "epoch": 0.12931481588198007,
      "grad_norm": 1.2942482233047485,
      "learning_rate": 9.743673925038458e-06,
      "loss": 0.8288,
      "step": 4488
    },
    {
      "epoch": 0.12934362934362933,
      "grad_norm": 1.3337851762771606,
      "learning_rate": 9.7435264213973e-06,
      "loss": 0.8325,
      "step": 4489
    },
    {
      "epoch": 0.12937244280527863,
      "grad_norm": 1.28263258934021,
      "learning_rate": 9.74337887644477e-06,
      "loss": 0.835,
      "step": 4490
    },
    {
      "epoch": 0.12940125626692792,
      "grad_norm": 1.3127515316009521,
      "learning_rate": 9.74323129018215e-06,
      "loss": 0.8174,
      "step": 4491
    },
    {
      "epoch": 0.12943006972857718,
      "grad_norm": 1.3211302757263184,
      "learning_rate": 9.743083662610725e-06,
      "loss": 0.8129,
      "step": 4492
    },
    {
      "epoch": 0.12945888319022647,
      "grad_norm": 1.280363917350769,
      "learning_rate": 9.742935993731786e-06,
      "loss": 0.7921,
      "step": 4493
    },
    {
      "epoch": 0.12948769665187576,
      "grad_norm": 1.2986947298049927,
      "learning_rate": 9.742788283546617e-06,
      "loss": 0.7984,
      "step": 4494
    },
    {
      "epoch": 0.12951651011352505,
      "grad_norm": 1.276759386062622,
      "learning_rate": 9.7426405320565e-06,
      "loss": 0.8016,
      "step": 4495
    },
    {
      "epoch": 0.12954532357517431,
      "grad_norm": 1.2845121622085571,
      "learning_rate": 9.742492739262725e-06,
      "loss": 0.8207,
      "step": 4496
    },
    {
      "epoch": 0.1295741370368236,
      "grad_norm": 1.330251932144165,
      "learning_rate": 9.742344905166579e-06,
      "loss": 0.8073,
      "step": 4497
    },
    {
      "epoch": 0.1296029504984729,
      "grad_norm": 1.27349054813385,
      "learning_rate": 9.74219702976935e-06,
      "loss": 0.8111,
      "step": 4498
    },
    {
      "epoch": 0.12963176396012216,
      "grad_norm": 1.2943509817123413,
      "learning_rate": 9.742049113072325e-06,
      "loss": 0.8155,
      "step": 4499
    },
    {
      "epoch": 0.12966057742177145,
      "grad_norm": 1.3300447463989258,
      "learning_rate": 9.741901155076794e-06,
      "loss": 0.8123,
      "step": 4500
    },
    {
      "epoch": 0.12968939088342074,
      "grad_norm": 1.3338491916656494,
      "learning_rate": 9.741753155784043e-06,
      "loss": 0.8012,
      "step": 4501
    },
    {
      "epoch": 0.12971820434507,
      "grad_norm": 1.3102906942367554,
      "learning_rate": 9.741605115195363e-06,
      "loss": 0.7784,
      "step": 4502
    },
    {
      "epoch": 0.1297470178067193,
      "grad_norm": 1.3776615858078003,
      "learning_rate": 9.741457033312041e-06,
      "loss": 0.8012,
      "step": 4503
    },
    {
      "epoch": 0.12977583126836859,
      "grad_norm": 1.3869929313659668,
      "learning_rate": 9.741308910135368e-06,
      "loss": 0.7907,
      "step": 4504
    },
    {
      "epoch": 0.12980464473001788,
      "grad_norm": 1.3670278787612915,
      "learning_rate": 9.741160745666634e-06,
      "loss": 0.8152,
      "step": 4505
    },
    {
      "epoch": 0.12983345819166714,
      "grad_norm": 1.3548425436019897,
      "learning_rate": 9.74101253990713e-06,
      "loss": 0.8534,
      "step": 4506
    },
    {
      "epoch": 0.12986227165331643,
      "grad_norm": 1.3248355388641357,
      "learning_rate": 9.740864292858145e-06,
      "loss": 0.8585,
      "step": 4507
    },
    {
      "epoch": 0.12989108511496572,
      "grad_norm": 1.3852208852767944,
      "learning_rate": 9.740716004520974e-06,
      "loss": 0.8075,
      "step": 4508
    },
    {
      "epoch": 0.12991989857661498,
      "grad_norm": 1.3361841440200806,
      "learning_rate": 9.740567674896902e-06,
      "loss": 0.8194,
      "step": 4509
    },
    {
      "epoch": 0.12994871203826427,
      "grad_norm": 1.3421918153762817,
      "learning_rate": 9.740419303987226e-06,
      "loss": 0.8182,
      "step": 4510
    },
    {
      "epoch": 0.12997752549991357,
      "grad_norm": 1.305356502532959,
      "learning_rate": 9.740270891793236e-06,
      "loss": 0.8068,
      "step": 4511
    },
    {
      "epoch": 0.13000633896156283,
      "grad_norm": 1.3106664419174194,
      "learning_rate": 9.740122438316227e-06,
      "loss": 0.7928,
      "step": 4512
    },
    {
      "epoch": 0.13003515242321212,
      "grad_norm": 1.384433627128601,
      "learning_rate": 9.739973943557487e-06,
      "loss": 0.7998,
      "step": 4513
    },
    {
      "epoch": 0.1300639658848614,
      "grad_norm": 1.2768551111221313,
      "learning_rate": 9.739825407518315e-06,
      "loss": 0.7969,
      "step": 4514
    },
    {
      "epoch": 0.1300927793465107,
      "grad_norm": 1.3288565874099731,
      "learning_rate": 9.7396768302e-06,
      "loss": 0.788,
      "step": 4515
    },
    {
      "epoch": 0.13012159280815996,
      "grad_norm": 1.3604204654693604,
      "learning_rate": 9.739528211603837e-06,
      "loss": 0.8117,
      "step": 4516
    },
    {
      "epoch": 0.13015040626980925,
      "grad_norm": 1.3497904539108276,
      "learning_rate": 9.739379551731124e-06,
      "loss": 0.8049,
      "step": 4517
    },
    {
      "epoch": 0.13017921973145855,
      "grad_norm": 1.4093352556228638,
      "learning_rate": 9.73923085058315e-06,
      "loss": 0.8077,
      "step": 4518
    },
    {
      "epoch": 0.1302080331931078,
      "grad_norm": 1.847502589225769,
      "learning_rate": 9.739082108161214e-06,
      "loss": 0.8017,
      "step": 4519
    },
    {
      "epoch": 0.1302368466547571,
      "grad_norm": 1.3806869983673096,
      "learning_rate": 9.73893332446661e-06,
      "loss": 0.8098,
      "step": 4520
    },
    {
      "epoch": 0.1302656601164064,
      "grad_norm": 1.2917457818984985,
      "learning_rate": 9.738784499500633e-06,
      "loss": 0.8438,
      "step": 4521
    },
    {
      "epoch": 0.13029447357805568,
      "grad_norm": 1.3008726835250854,
      "learning_rate": 9.738635633264582e-06,
      "loss": 0.7859,
      "step": 4522
    },
    {
      "epoch": 0.13032328703970494,
      "grad_norm": 1.4595961570739746,
      "learning_rate": 9.73848672575975e-06,
      "loss": 0.8167,
      "step": 4523
    },
    {
      "epoch": 0.13035210050135423,
      "grad_norm": 1.3543113470077515,
      "learning_rate": 9.738337776987437e-06,
      "loss": 0.8478,
      "step": 4524
    },
    {
      "epoch": 0.13038091396300353,
      "grad_norm": 1.3042714595794678,
      "learning_rate": 9.738188786948936e-06,
      "loss": 0.7947,
      "step": 4525
    },
    {
      "epoch": 0.1304097274246528,
      "grad_norm": 1.387123703956604,
      "learning_rate": 9.73803975564555e-06,
      "loss": 0.8102,
      "step": 4526
    },
    {
      "epoch": 0.13043854088630208,
      "grad_norm": 1.2988255023956299,
      "learning_rate": 9.737890683078571e-06,
      "loss": 0.7852,
      "step": 4527
    },
    {
      "epoch": 0.13046735434795137,
      "grad_norm": 1.3475158214569092,
      "learning_rate": 9.737741569249302e-06,
      "loss": 0.8169,
      "step": 4528
    },
    {
      "epoch": 0.13049616780960063,
      "grad_norm": 1.3245408535003662,
      "learning_rate": 9.737592414159041e-06,
      "loss": 0.8228,
      "step": 4529
    },
    {
      "epoch": 0.13052498127124992,
      "grad_norm": 1.3357921838760376,
      "learning_rate": 9.737443217809083e-06,
      "loss": 0.808,
      "step": 4530
    },
    {
      "epoch": 0.13055379473289921,
      "grad_norm": 1.3033148050308228,
      "learning_rate": 9.737293980200733e-06,
      "loss": 0.8026,
      "step": 4531
    },
    {
      "epoch": 0.1305826081945485,
      "grad_norm": 1.2794930934906006,
      "learning_rate": 9.737144701335287e-06,
      "loss": 0.8347,
      "step": 4532
    },
    {
      "epoch": 0.13061142165619777,
      "grad_norm": 1.337975025177002,
      "learning_rate": 9.736995381214047e-06,
      "loss": 0.8275,
      "step": 4533
    },
    {
      "epoch": 0.13064023511784706,
      "grad_norm": 1.2666786909103394,
      "learning_rate": 9.73684601983831e-06,
      "loss": 0.8252,
      "step": 4534
    },
    {
      "epoch": 0.13066904857949635,
      "grad_norm": 1.3228096961975098,
      "learning_rate": 9.736696617209382e-06,
      "loss": 0.8169,
      "step": 4535
    },
    {
      "epoch": 0.1306978620411456,
      "grad_norm": 1.319887399673462,
      "learning_rate": 9.73654717332856e-06,
      "loss": 0.8239,
      "step": 4536
    },
    {
      "epoch": 0.1307266755027949,
      "grad_norm": 1.3490206003189087,
      "learning_rate": 9.736397688197148e-06,
      "loss": 0.8295,
      "step": 4537
    },
    {
      "epoch": 0.1307554889644442,
      "grad_norm": 1.301351547241211,
      "learning_rate": 9.736248161816445e-06,
      "loss": 0.8274,
      "step": 4538
    },
    {
      "epoch": 0.13078430242609346,
      "grad_norm": 1.3100649118423462,
      "learning_rate": 9.736098594187757e-06,
      "loss": 0.7892,
      "step": 4539
    },
    {
      "epoch": 0.13081311588774275,
      "grad_norm": 1.305128574371338,
      "learning_rate": 9.735948985312383e-06,
      "loss": 0.8008,
      "step": 4540
    },
    {
      "epoch": 0.13084192934939204,
      "grad_norm": 1.3388416767120361,
      "learning_rate": 9.735799335191628e-06,
      "loss": 0.8195,
      "step": 4541
    },
    {
      "epoch": 0.13087074281104133,
      "grad_norm": 1.2924903631210327,
      "learning_rate": 9.735649643826795e-06,
      "loss": 0.8332,
      "step": 4542
    },
    {
      "epoch": 0.1308995562726906,
      "grad_norm": 1.3647959232330322,
      "learning_rate": 9.735499911219189e-06,
      "loss": 0.8059,
      "step": 4543
    },
    {
      "epoch": 0.13092836973433988,
      "grad_norm": 1.3528379201889038,
      "learning_rate": 9.73535013737011e-06,
      "loss": 0.8267,
      "step": 4544
    },
    {
      "epoch": 0.13095718319598917,
      "grad_norm": 1.2976771593093872,
      "learning_rate": 9.735200322280867e-06,
      "loss": 0.8244,
      "step": 4545
    },
    {
      "epoch": 0.13098599665763844,
      "grad_norm": 1.409209966659546,
      "learning_rate": 9.73505046595276e-06,
      "loss": 0.8315,
      "step": 4546
    },
    {
      "epoch": 0.13101481011928773,
      "grad_norm": 1.260694146156311,
      "learning_rate": 9.7349005683871e-06,
      "loss": 0.7958,
      "step": 4547
    },
    {
      "epoch": 0.13104362358093702,
      "grad_norm": 1.2860699892044067,
      "learning_rate": 9.734750629585189e-06,
      "loss": 0.8139,
      "step": 4548
    },
    {
      "epoch": 0.1310724370425863,
      "grad_norm": 1.3481645584106445,
      "learning_rate": 9.73460064954833e-06,
      "loss": 0.8049,
      "step": 4549
    },
    {
      "epoch": 0.13110125050423557,
      "grad_norm": 1.2873749732971191,
      "learning_rate": 9.734450628277833e-06,
      "loss": 0.8127,
      "step": 4550
    },
    {
      "epoch": 0.13113006396588486,
      "grad_norm": 1.2946419715881348,
      "learning_rate": 9.734300565775006e-06,
      "loss": 0.8282,
      "step": 4551
    },
    {
      "epoch": 0.13115887742753415,
      "grad_norm": 1.2903951406478882,
      "learning_rate": 9.734150462041152e-06,
      "loss": 0.8266,
      "step": 4552
    },
    {
      "epoch": 0.13118769088918342,
      "grad_norm": 1.4207476377487183,
      "learning_rate": 9.73400031707758e-06,
      "loss": 0.8297,
      "step": 4553
    },
    {
      "epoch": 0.1312165043508327,
      "grad_norm": 1.301079273223877,
      "learning_rate": 9.733850130885598e-06,
      "loss": 0.8169,
      "step": 4554
    },
    {
      "epoch": 0.131245317812482,
      "grad_norm": 1.3328726291656494,
      "learning_rate": 9.733699903466514e-06,
      "loss": 0.8252,
      "step": 4555
    },
    {
      "epoch": 0.13127413127413126,
      "grad_norm": 1.3366141319274902,
      "learning_rate": 9.733549634821634e-06,
      "loss": 0.8039,
      "step": 4556
    },
    {
      "epoch": 0.13130294473578055,
      "grad_norm": 1.4125858545303345,
      "learning_rate": 9.73339932495227e-06,
      "loss": 0.8426,
      "step": 4557
    },
    {
      "epoch": 0.13133175819742984,
      "grad_norm": 1.2824479341506958,
      "learning_rate": 9.733248973859729e-06,
      "loss": 0.8189,
      "step": 4558
    },
    {
      "epoch": 0.13136057165907913,
      "grad_norm": 1.3405168056488037,
      "learning_rate": 9.733098581545321e-06,
      "loss": 0.7975,
      "step": 4559
    },
    {
      "epoch": 0.1313893851207284,
      "grad_norm": 1.336677074432373,
      "learning_rate": 9.732948148010357e-06,
      "loss": 0.8057,
      "step": 4560
    },
    {
      "epoch": 0.1314181985823777,
      "grad_norm": 1.3182926177978516,
      "learning_rate": 9.732797673256143e-06,
      "loss": 0.815,
      "step": 4561
    },
    {
      "epoch": 0.13144701204402698,
      "grad_norm": 1.3765004873275757,
      "learning_rate": 9.732647157283994e-06,
      "loss": 0.8063,
      "step": 4562
    },
    {
      "epoch": 0.13147582550567624,
      "grad_norm": 1.3243786096572876,
      "learning_rate": 9.732496600095219e-06,
      "loss": 0.8128,
      "step": 4563
    },
    {
      "epoch": 0.13150463896732553,
      "grad_norm": 1.2666937112808228,
      "learning_rate": 9.73234600169113e-06,
      "loss": 0.807,
      "step": 4564
    },
    {
      "epoch": 0.13153345242897482,
      "grad_norm": 1.3487755060195923,
      "learning_rate": 9.732195362073038e-06,
      "loss": 0.8167,
      "step": 4565
    },
    {
      "epoch": 0.13156226589062409,
      "grad_norm": 1.3069881200790405,
      "learning_rate": 9.732044681242255e-06,
      "loss": 0.8222,
      "step": 4566
    },
    {
      "epoch": 0.13159107935227338,
      "grad_norm": 1.2500306367874146,
      "learning_rate": 9.731893959200091e-06,
      "loss": 0.7762,
      "step": 4567
    },
    {
      "epoch": 0.13161989281392267,
      "grad_norm": 1.385407567024231,
      "learning_rate": 9.731743195947863e-06,
      "loss": 0.8192,
      "step": 4568
    },
    {
      "epoch": 0.13164870627557196,
      "grad_norm": 1.3154239654541016,
      "learning_rate": 9.73159239148688e-06,
      "loss": 0.8009,
      "step": 4569
    },
    {
      "epoch": 0.13167751973722122,
      "grad_norm": 1.3211565017700195,
      "learning_rate": 9.731441545818458e-06,
      "loss": 0.8129,
      "step": 4570
    },
    {
      "epoch": 0.1317063331988705,
      "grad_norm": 1.3768317699432373,
      "learning_rate": 9.73129065894391e-06,
      "loss": 0.8191,
      "step": 4571
    },
    {
      "epoch": 0.1317351466605198,
      "grad_norm": 1.3360689878463745,
      "learning_rate": 9.73113973086455e-06,
      "loss": 0.8105,
      "step": 4572
    },
    {
      "epoch": 0.13176396012216907,
      "grad_norm": 1.3206387758255005,
      "learning_rate": 9.730988761581693e-06,
      "loss": 0.7847,
      "step": 4573
    },
    {
      "epoch": 0.13179277358381836,
      "grad_norm": 1.3681144714355469,
      "learning_rate": 9.730837751096652e-06,
      "loss": 0.8034,
      "step": 4574
    },
    {
      "epoch": 0.13182158704546765,
      "grad_norm": 1.3177807331085205,
      "learning_rate": 9.730686699410742e-06,
      "loss": 0.8135,
      "step": 4575
    },
    {
      "epoch": 0.13185040050711694,
      "grad_norm": 1.3813153505325317,
      "learning_rate": 9.730535606525282e-06,
      "loss": 0.8307,
      "step": 4576
    },
    {
      "epoch": 0.1318792139687662,
      "grad_norm": 1.2747989892959595,
      "learning_rate": 9.730384472441587e-06,
      "loss": 0.8213,
      "step": 4577
    },
    {
      "epoch": 0.1319080274304155,
      "grad_norm": 1.3307044506072998,
      "learning_rate": 9.73023329716097e-06,
      "loss": 0.8003,
      "step": 4578
    },
    {
      "epoch": 0.13193684089206478,
      "grad_norm": 1.4085842370986938,
      "learning_rate": 9.730082080684748e-06,
      "loss": 0.8183,
      "step": 4579
    },
    {
      "epoch": 0.13196565435371405,
      "grad_norm": 1.2722728252410889,
      "learning_rate": 9.729930823014242e-06,
      "loss": 0.8253,
      "step": 4580
    },
    {
      "epoch": 0.13199446781536334,
      "grad_norm": 1.2772691249847412,
      "learning_rate": 9.729779524150765e-06,
      "loss": 0.8216,
      "step": 4581
    },
    {
      "epoch": 0.13202328127701263,
      "grad_norm": 1.400433897972107,
      "learning_rate": 9.729628184095637e-06,
      "loss": 0.799,
      "step": 4582
    },
    {
      "epoch": 0.1320520947386619,
      "grad_norm": 1.3248881101608276,
      "learning_rate": 9.729476802850176e-06,
      "loss": 0.8159,
      "step": 4583
    },
    {
      "epoch": 0.13208090820031118,
      "grad_norm": 1.3054497241973877,
      "learning_rate": 9.729325380415699e-06,
      "loss": 0.7876,
      "step": 4584
    },
    {
      "epoch": 0.13210972166196047,
      "grad_norm": 2.6942124366760254,
      "learning_rate": 9.729173916793527e-06,
      "loss": 0.8272,
      "step": 4585
    },
    {
      "epoch": 0.13213853512360976,
      "grad_norm": 1.3123832941055298,
      "learning_rate": 9.729022411984975e-06,
      "loss": 0.7938,
      "step": 4586
    },
    {
      "epoch": 0.13216734858525903,
      "grad_norm": 1.3372279405593872,
      "learning_rate": 9.728870865991366e-06,
      "loss": 0.8048,
      "step": 4587
    },
    {
      "epoch": 0.13219616204690832,
      "grad_norm": 1.3519114255905151,
      "learning_rate": 9.728719278814021e-06,
      "loss": 0.8391,
      "step": 4588
    },
    {
      "epoch": 0.1322249755085576,
      "grad_norm": 1.2959903478622437,
      "learning_rate": 9.728567650454256e-06,
      "loss": 0.7988,
      "step": 4589
    },
    {
      "epoch": 0.13225378897020687,
      "grad_norm": 1.3736180067062378,
      "learning_rate": 9.728415980913395e-06,
      "loss": 0.8054,
      "step": 4590
    },
    {
      "epoch": 0.13228260243185616,
      "grad_norm": 1.2938001155853271,
      "learning_rate": 9.728264270192756e-06,
      "loss": 0.8186,
      "step": 4591
    },
    {
      "epoch": 0.13231141589350545,
      "grad_norm": 1.268027901649475,
      "learning_rate": 9.728112518293662e-06,
      "loss": 0.8225,
      "step": 4592
    },
    {
      "epoch": 0.13234022935515471,
      "grad_norm": 1.297812819480896,
      "learning_rate": 9.727960725217434e-06,
      "loss": 0.8044,
      "step": 4593
    },
    {
      "epoch": 0.132369042816804,
      "grad_norm": 1.3507753610610962,
      "learning_rate": 9.727808890965396e-06,
      "loss": 0.8198,
      "step": 4594
    },
    {
      "epoch": 0.1323978562784533,
      "grad_norm": 1.277376413345337,
      "learning_rate": 9.727657015538868e-06,
      "loss": 0.8419,
      "step": 4595
    },
    {
      "epoch": 0.1324266697401026,
      "grad_norm": 1.3361306190490723,
      "learning_rate": 9.727505098939172e-06,
      "loss": 0.8066,
      "step": 4596
    },
    {
      "epoch": 0.13245548320175185,
      "grad_norm": 1.2799817323684692,
      "learning_rate": 9.727353141167633e-06,
      "loss": 0.8091,
      "step": 4597
    },
    {
      "epoch": 0.13248429666340114,
      "grad_norm": 1.276284098625183,
      "learning_rate": 9.727201142225575e-06,
      "loss": 0.778,
      "step": 4598
    },
    {
      "epoch": 0.13251311012505043,
      "grad_norm": 1.2511652708053589,
      "learning_rate": 9.727049102114319e-06,
      "loss": 0.808,
      "step": 4599
    },
    {
      "epoch": 0.1325419235866997,
      "grad_norm": 1.2898980379104614,
      "learning_rate": 9.726897020835191e-06,
      "loss": 0.8159,
      "step": 4600
    },
    {
      "epoch": 0.13257073704834899,
      "grad_norm": 1.3186392784118652,
      "learning_rate": 9.726744898389515e-06,
      "loss": 0.8069,
      "step": 4601
    },
    {
      "epoch": 0.13259955050999828,
      "grad_norm": 1.3486361503601074,
      "learning_rate": 9.726592734778618e-06,
      "loss": 0.8097,
      "step": 4602
    },
    {
      "epoch": 0.13262836397164757,
      "grad_norm": 1.3032699823379517,
      "learning_rate": 9.72644053000382e-06,
      "loss": 0.8129,
      "step": 4603
    },
    {
      "epoch": 0.13265717743329683,
      "grad_norm": 1.3114136457443237,
      "learning_rate": 9.726288284066452e-06,
      "loss": 0.8186,
      "step": 4604
    },
    {
      "epoch": 0.13268599089494612,
      "grad_norm": 1.4699448347091675,
      "learning_rate": 9.726135996967836e-06,
      "loss": 0.8333,
      "step": 4605
    },
    {
      "epoch": 0.1327148043565954,
      "grad_norm": 1.5466798543930054,
      "learning_rate": 9.725983668709301e-06,
      "loss": 0.8339,
      "step": 4606
    },
    {
      "epoch": 0.13274361781824467,
      "grad_norm": 1.2408764362335205,
      "learning_rate": 9.725831299292172e-06,
      "loss": 0.8209,
      "step": 4607
    },
    {
      "epoch": 0.13277243127989397,
      "grad_norm": 1.326765775680542,
      "learning_rate": 9.725678888717777e-06,
      "loss": 0.8073,
      "step": 4608
    },
    {
      "epoch": 0.13280124474154326,
      "grad_norm": 1.468146562576294,
      "learning_rate": 9.725526436987443e-06,
      "loss": 0.8284,
      "step": 4609
    },
    {
      "epoch": 0.13283005820319252,
      "grad_norm": 1.2894682884216309,
      "learning_rate": 9.725373944102496e-06,
      "loss": 0.8021,
      "step": 4610
    },
    {
      "epoch": 0.1328588716648418,
      "grad_norm": 1.330962896347046,
      "learning_rate": 9.725221410064268e-06,
      "loss": 0.8296,
      "step": 4611
    },
    {
      "epoch": 0.1328876851264911,
      "grad_norm": 1.3856463432312012,
      "learning_rate": 9.725068834874083e-06,
      "loss": 0.796,
      "step": 4612
    },
    {
      "epoch": 0.1329164985881404,
      "grad_norm": 1.404701828956604,
      "learning_rate": 9.724916218533272e-06,
      "loss": 0.8215,
      "step": 4613
    },
    {
      "epoch": 0.13294531204978965,
      "grad_norm": 1.3404741287231445,
      "learning_rate": 9.724763561043165e-06,
      "loss": 0.785,
      "step": 4614
    },
    {
      "epoch": 0.13297412551143895,
      "grad_norm": 1.3485311269760132,
      "learning_rate": 9.724610862405088e-06,
      "loss": 0.8028,
      "step": 4615
    },
    {
      "epoch": 0.13300293897308824,
      "grad_norm": 1.4084992408752441,
      "learning_rate": 9.724458122620376e-06,
      "loss": 0.812,
      "step": 4616
    },
    {
      "epoch": 0.1330317524347375,
      "grad_norm": 1.4024078845977783,
      "learning_rate": 9.724305341690355e-06,
      "loss": 0.8165,
      "step": 4617
    },
    {
      "epoch": 0.1330605658963868,
      "grad_norm": 1.3467410802841187,
      "learning_rate": 9.724152519616357e-06,
      "loss": 0.8215,
      "step": 4618
    },
    {
      "epoch": 0.13308937935803608,
      "grad_norm": 1.341854214668274,
      "learning_rate": 9.723999656399715e-06,
      "loss": 0.8404,
      "step": 4619
    },
    {
      "epoch": 0.13311819281968534,
      "grad_norm": 1.3458622694015503,
      "learning_rate": 9.723846752041757e-06,
      "loss": 0.8146,
      "step": 4620
    },
    {
      "epoch": 0.13314700628133463,
      "grad_norm": 1.3042423725128174,
      "learning_rate": 9.723693806543814e-06,
      "loss": 0.7971,
      "step": 4621
    },
    {
      "epoch": 0.13317581974298393,
      "grad_norm": 1.2970390319824219,
      "learning_rate": 9.723540819907222e-06,
      "loss": 0.7963,
      "step": 4622
    },
    {
      "epoch": 0.13320463320463322,
      "grad_norm": 1.3792259693145752,
      "learning_rate": 9.723387792133312e-06,
      "loss": 0.8081,
      "step": 4623
    },
    {
      "epoch": 0.13323344666628248,
      "grad_norm": 1.3128046989440918,
      "learning_rate": 9.723234723223414e-06,
      "loss": 0.802,
      "step": 4624
    },
    {
      "epoch": 0.13326226012793177,
      "grad_norm": 1.2950201034545898,
      "learning_rate": 9.723081613178865e-06,
      "loss": 0.7934,
      "step": 4625
    },
    {
      "epoch": 0.13329107358958106,
      "grad_norm": 1.3302671909332275,
      "learning_rate": 9.722928462000995e-06,
      "loss": 0.8004,
      "step": 4626
    },
    {
      "epoch": 0.13331988705123032,
      "grad_norm": 1.3339425325393677,
      "learning_rate": 9.72277526969114e-06,
      "loss": 0.8082,
      "step": 4627
    },
    {
      "epoch": 0.13334870051287961,
      "grad_norm": 1.2948393821716309,
      "learning_rate": 9.722622036250633e-06,
      "loss": 0.8042,
      "step": 4628
    },
    {
      "epoch": 0.1333775139745289,
      "grad_norm": 1.3286304473876953,
      "learning_rate": 9.722468761680808e-06,
      "loss": 0.8203,
      "step": 4629
    },
    {
      "epoch": 0.1334063274361782,
      "grad_norm": 1.287568211555481,
      "learning_rate": 9.722315445983002e-06,
      "loss": 0.7856,
      "step": 4630
    },
    {
      "epoch": 0.13343514089782746,
      "grad_norm": 1.3248144388198853,
      "learning_rate": 9.722162089158548e-06,
      "loss": 0.8012,
      "step": 4631
    },
    {
      "epoch": 0.13346395435947675,
      "grad_norm": 1.3871901035308838,
      "learning_rate": 9.722008691208785e-06,
      "loss": 0.8126,
      "step": 4632
    },
    {
      "epoch": 0.13349276782112604,
      "grad_norm": 1.38923978805542,
      "learning_rate": 9.721855252135045e-06,
      "loss": 0.7963,
      "step": 4633
    },
    {
      "epoch": 0.1335215812827753,
      "grad_norm": 1.2924314737319946,
      "learning_rate": 9.721701771938667e-06,
      "loss": 0.8071,
      "step": 4634
    },
    {
      "epoch": 0.1335503947444246,
      "grad_norm": 1.2997969388961792,
      "learning_rate": 9.721548250620984e-06,
      "loss": 0.8168,
      "step": 4635
    },
    {
      "epoch": 0.13357920820607389,
      "grad_norm": 1.3847116231918335,
      "learning_rate": 9.72139468818334e-06,
      "loss": 0.8219,
      "step": 4636
    },
    {
      "epoch": 0.13360802166772315,
      "grad_norm": 1.2561137676239014,
      "learning_rate": 9.721241084627063e-06,
      "loss": 0.8017,
      "step": 4637
    },
    {
      "epoch": 0.13363683512937244,
      "grad_norm": 1.2463270425796509,
      "learning_rate": 9.721087439953499e-06,
      "loss": 0.797,
      "step": 4638
    },
    {
      "epoch": 0.13366564859102173,
      "grad_norm": 1.292683720588684,
      "learning_rate": 9.720933754163981e-06,
      "loss": 0.8462,
      "step": 4639
    },
    {
      "epoch": 0.13369446205267102,
      "grad_norm": 1.3463006019592285,
      "learning_rate": 9.720780027259848e-06,
      "loss": 0.8062,
      "step": 4640
    },
    {
      "epoch": 0.13372327551432028,
      "grad_norm": 1.2977222204208374,
      "learning_rate": 9.720626259242442e-06,
      "loss": 0.8093,
      "step": 4641
    },
    {
      "epoch": 0.13375208897596957,
      "grad_norm": 1.2897021770477295,
      "learning_rate": 9.720472450113101e-06,
      "loss": 0.8049,
      "step": 4642
    },
    {
      "epoch": 0.13378090243761886,
      "grad_norm": 1.379249930381775,
      "learning_rate": 9.720318599873162e-06,
      "loss": 0.8151,
      "step": 4643
    },
    {
      "epoch": 0.13380971589926813,
      "grad_norm": 1.2439849376678467,
      "learning_rate": 9.720164708523969e-06,
      "loss": 0.7982,
      "step": 4644
    },
    {
      "epoch": 0.13383852936091742,
      "grad_norm": 1.654357671737671,
      "learning_rate": 9.720010776066857e-06,
      "loss": 0.7916,
      "step": 4645
    },
    {
      "epoch": 0.1338673428225667,
      "grad_norm": 1.322702407836914,
      "learning_rate": 9.719856802503171e-06,
      "loss": 0.8227,
      "step": 4646
    },
    {
      "epoch": 0.13389615628421597,
      "grad_norm": 1.323707938194275,
      "learning_rate": 9.719702787834248e-06,
      "loss": 0.8232,
      "step": 4647
    },
    {
      "epoch": 0.13392496974586526,
      "grad_norm": 1.2511656284332275,
      "learning_rate": 9.719548732061435e-06,
      "loss": 0.8244,
      "step": 4648
    },
    {
      "epoch": 0.13395378320751455,
      "grad_norm": 1.2830127477645874,
      "learning_rate": 9.719394635186067e-06,
      "loss": 0.8137,
      "step": 4649
    },
    {
      "epoch": 0.13398259666916384,
      "grad_norm": 1.3260769844055176,
      "learning_rate": 9.719240497209493e-06,
      "loss": 0.8172,
      "step": 4650
    },
    {
      "epoch": 0.1340114101308131,
      "grad_norm": 1.287997841835022,
      "learning_rate": 9.71908631813305e-06,
      "loss": 0.8211,
      "step": 4651
    },
    {
      "epoch": 0.1340402235924624,
      "grad_norm": 1.2934356927871704,
      "learning_rate": 9.718932097958083e-06,
      "loss": 0.7898,
      "step": 4652
    },
    {
      "epoch": 0.1340690370541117,
      "grad_norm": 1.9592537879943848,
      "learning_rate": 9.718777836685935e-06,
      "loss": 0.7807,
      "step": 4653
    },
    {
      "epoch": 0.13409785051576095,
      "grad_norm": 1.3172004222869873,
      "learning_rate": 9.718623534317948e-06,
      "loss": 0.7922,
      "step": 4654
    },
    {
      "epoch": 0.13412666397741024,
      "grad_norm": 1.2635765075683594,
      "learning_rate": 9.718469190855466e-06,
      "loss": 0.8166,
      "step": 4655
    },
    {
      "epoch": 0.13415547743905953,
      "grad_norm": 1.2358312606811523,
      "learning_rate": 9.718314806299836e-06,
      "loss": 0.8011,
      "step": 4656
    },
    {
      "epoch": 0.13418429090070882,
      "grad_norm": 1.2547969818115234,
      "learning_rate": 9.718160380652399e-06,
      "loss": 0.809,
      "step": 4657
    },
    {
      "epoch": 0.1342131043623581,
      "grad_norm": 1.2943906784057617,
      "learning_rate": 9.718005913914504e-06,
      "loss": 0.7975,
      "step": 4658
    },
    {
      "epoch": 0.13424191782400738,
      "grad_norm": 1.3242161273956299,
      "learning_rate": 9.717851406087492e-06,
      "loss": 0.8007,
      "step": 4659
    },
    {
      "epoch": 0.13427073128565667,
      "grad_norm": 1.2497127056121826,
      "learning_rate": 9.71769685717271e-06,
      "loss": 0.8001,
      "step": 4660
    },
    {
      "epoch": 0.13429954474730593,
      "grad_norm": 1.2199077606201172,
      "learning_rate": 9.717542267171505e-06,
      "loss": 0.8019,
      "step": 4661
    },
    {
      "epoch": 0.13432835820895522,
      "grad_norm": 1.2052557468414307,
      "learning_rate": 9.717387636085223e-06,
      "loss": 0.7813,
      "step": 4662
    },
    {
      "epoch": 0.1343571716706045,
      "grad_norm": 1.3426940441131592,
      "learning_rate": 9.71723296391521e-06,
      "loss": 0.824,
      "step": 4663
    },
    {
      "epoch": 0.13438598513225378,
      "grad_norm": 1.2959178686141968,
      "learning_rate": 9.717078250662813e-06,
      "loss": 0.823,
      "step": 4664
    },
    {
      "epoch": 0.13441479859390307,
      "grad_norm": 1.2761270999908447,
      "learning_rate": 9.71692349632938e-06,
      "loss": 0.7861,
      "step": 4665
    },
    {
      "epoch": 0.13444361205555236,
      "grad_norm": 1.3290001153945923,
      "learning_rate": 9.71676870091626e-06,
      "loss": 0.7971,
      "step": 4666
    },
    {
      "epoch": 0.13447242551720165,
      "grad_norm": 1.3187326192855835,
      "learning_rate": 9.716613864424797e-06,
      "loss": 0.8241,
      "step": 4667
    },
    {
      "epoch": 0.1345012389788509,
      "grad_norm": 1.2904760837554932,
      "learning_rate": 9.716458986856345e-06,
      "loss": 0.8019,
      "step": 4668
    },
    {
      "epoch": 0.1345300524405002,
      "grad_norm": 1.2791413068771362,
      "learning_rate": 9.716304068212247e-06,
      "loss": 0.7993,
      "step": 4669
    },
    {
      "epoch": 0.1345588659021495,
      "grad_norm": 1.2472429275512695,
      "learning_rate": 9.716149108493857e-06,
      "loss": 0.7791,
      "step": 4670
    },
    {
      "epoch": 0.13458767936379876,
      "grad_norm": 1.338564395904541,
      "learning_rate": 9.715994107702523e-06,
      "loss": 0.8293,
      "step": 4671
    },
    {
      "epoch": 0.13461649282544805,
      "grad_norm": 1.3290417194366455,
      "learning_rate": 9.715839065839594e-06,
      "loss": 0.8243,
      "step": 4672
    },
    {
      "epoch": 0.13464530628709734,
      "grad_norm": 1.2292007207870483,
      "learning_rate": 9.715683982906422e-06,
      "loss": 0.8034,
      "step": 4673
    },
    {
      "epoch": 0.1346741197487466,
      "grad_norm": 1.356940507888794,
      "learning_rate": 9.715528858904353e-06,
      "loss": 0.8199,
      "step": 4674
    },
    {
      "epoch": 0.1347029332103959,
      "grad_norm": 1.3356350660324097,
      "learning_rate": 9.715373693834745e-06,
      "loss": 0.7994,
      "step": 4675
    },
    {
      "epoch": 0.13473174667204518,
      "grad_norm": 1.2867985963821411,
      "learning_rate": 9.715218487698944e-06,
      "loss": 0.8378,
      "step": 4676
    },
    {
      "epoch": 0.13476056013369447,
      "grad_norm": 1.4420766830444336,
      "learning_rate": 9.715063240498304e-06,
      "loss": 0.8203,
      "step": 4677
    },
    {
      "epoch": 0.13478937359534374,
      "grad_norm": 1.3186694383621216,
      "learning_rate": 9.714907952234177e-06,
      "loss": 0.8102,
      "step": 4678
    },
    {
      "epoch": 0.13481818705699303,
      "grad_norm": 1.333284616470337,
      "learning_rate": 9.714752622907913e-06,
      "loss": 0.8034,
      "step": 4679
    },
    {
      "epoch": 0.13484700051864232,
      "grad_norm": 1.3887048959732056,
      "learning_rate": 9.71459725252087e-06,
      "loss": 0.7951,
      "step": 4680
    },
    {
      "epoch": 0.13487581398029158,
      "grad_norm": 1.350414514541626,
      "learning_rate": 9.714441841074394e-06,
      "loss": 0.8203,
      "step": 4681
    },
    {
      "epoch": 0.13490462744194087,
      "grad_norm": 1.3815134763717651,
      "learning_rate": 9.714286388569845e-06,
      "loss": 0.8183,
      "step": 4682
    },
    {
      "epoch": 0.13493344090359016,
      "grad_norm": 1.4057176113128662,
      "learning_rate": 9.714130895008573e-06,
      "loss": 0.807,
      "step": 4683
    },
    {
      "epoch": 0.13496225436523945,
      "grad_norm": 1.3142142295837402,
      "learning_rate": 9.713975360391933e-06,
      "loss": 0.819,
      "step": 4684
    },
    {
      "epoch": 0.13499106782688872,
      "grad_norm": 1.3425661325454712,
      "learning_rate": 9.71381978472128e-06,
      "loss": 0.7926,
      "step": 4685
    },
    {
      "epoch": 0.135019881288538,
      "grad_norm": 1.291646957397461,
      "learning_rate": 9.71366416799797e-06,
      "loss": 0.7965,
      "step": 4686
    },
    {
      "epoch": 0.1350486947501873,
      "grad_norm": 1.3667290210723877,
      "learning_rate": 9.713508510223357e-06,
      "loss": 0.8301,
      "step": 4687
    },
    {
      "epoch": 0.13507750821183656,
      "grad_norm": 1.258650779724121,
      "learning_rate": 9.713352811398796e-06,
      "loss": 0.7975,
      "step": 4688
    },
    {
      "epoch": 0.13510632167348585,
      "grad_norm": 1.4573814868927002,
      "learning_rate": 9.713197071525641e-06,
      "loss": 0.8122,
      "step": 4689
    },
    {
      "epoch": 0.13513513513513514,
      "grad_norm": 1.2735075950622559,
      "learning_rate": 9.713041290605255e-06,
      "loss": 0.8083,
      "step": 4690
    },
    {
      "epoch": 0.1351639485967844,
      "grad_norm": 1.3560287952423096,
      "learning_rate": 9.712885468638988e-06,
      "loss": 0.8091,
      "step": 4691
    },
    {
      "epoch": 0.1351927620584337,
      "grad_norm": 1.4046273231506348,
      "learning_rate": 9.712729605628202e-06,
      "loss": 0.8303,
      "step": 4692
    },
    {
      "epoch": 0.135221575520083,
      "grad_norm": 1.3248038291931152,
      "learning_rate": 9.712573701574251e-06,
      "loss": 0.8174,
      "step": 4693
    },
    {
      "epoch": 0.13525038898173228,
      "grad_norm": 1.3286585807800293,
      "learning_rate": 9.712417756478493e-06,
      "loss": 0.8121,
      "step": 4694
    },
    {
      "epoch": 0.13527920244338154,
      "grad_norm": 1.3405368328094482,
      "learning_rate": 9.712261770342288e-06,
      "loss": 0.829,
      "step": 4695
    },
    {
      "epoch": 0.13530801590503083,
      "grad_norm": 1.3167332410812378,
      "learning_rate": 9.712105743166996e-06,
      "loss": 0.796,
      "step": 4696
    },
    {
      "epoch": 0.13533682936668012,
      "grad_norm": 1.3195483684539795,
      "learning_rate": 9.711949674953969e-06,
      "loss": 0.8025,
      "step": 4697
    },
    {
      "epoch": 0.13536564282832939,
      "grad_norm": 1.4151029586791992,
      "learning_rate": 9.711793565704572e-06,
      "loss": 0.8155,
      "step": 4698
    },
    {
      "epoch": 0.13539445628997868,
      "grad_norm": 1.2299073934555054,
      "learning_rate": 9.711637415420165e-06,
      "loss": 0.7767,
      "step": 4699
    },
    {
      "epoch": 0.13542326975162797,
      "grad_norm": 1.2547181844711304,
      "learning_rate": 9.711481224102104e-06,
      "loss": 0.8355,
      "step": 4700
    },
    {
      "epoch": 0.13545208321327723,
      "grad_norm": 1.4240120649337769,
      "learning_rate": 9.711324991751752e-06,
      "loss": 0.8289,
      "step": 4701
    },
    {
      "epoch": 0.13548089667492652,
      "grad_norm": 1.2864296436309814,
      "learning_rate": 9.711168718370469e-06,
      "loss": 0.8101,
      "step": 4702
    },
    {
      "epoch": 0.1355097101365758,
      "grad_norm": 1.3185826539993286,
      "learning_rate": 9.711012403959615e-06,
      "loss": 0.8014,
      "step": 4703
    },
    {
      "epoch": 0.1355385235982251,
      "grad_norm": 1.4016278982162476,
      "learning_rate": 9.710856048520553e-06,
      "loss": 0.8055,
      "step": 4704
    },
    {
      "epoch": 0.13556733705987437,
      "grad_norm": 1.2725962400436401,
      "learning_rate": 9.710699652054644e-06,
      "loss": 0.8016,
      "step": 4705
    },
    {
      "epoch": 0.13559615052152366,
      "grad_norm": 1.2719662189483643,
      "learning_rate": 9.710543214563251e-06,
      "loss": 0.8188,
      "step": 4706
    },
    {
      "epoch": 0.13562496398317295,
      "grad_norm": 1.313650131225586,
      "learning_rate": 9.710386736047734e-06,
      "loss": 0.8053,
      "step": 4707
    },
    {
      "epoch": 0.1356537774448222,
      "grad_norm": 1.2564157247543335,
      "learning_rate": 9.710230216509456e-06,
      "loss": 0.7767,
      "step": 4708
    },
    {
      "epoch": 0.1356825909064715,
      "grad_norm": 1.2936159372329712,
      "learning_rate": 9.710073655949783e-06,
      "loss": 0.796,
      "step": 4709
    },
    {
      "epoch": 0.1357114043681208,
      "grad_norm": 1.3178764581680298,
      "learning_rate": 9.709917054370077e-06,
      "loss": 0.8182,
      "step": 4710
    },
    {
      "epoch": 0.13574021782977008,
      "grad_norm": 1.318058967590332,
      "learning_rate": 9.709760411771701e-06,
      "loss": 0.7813,
      "step": 4711
    },
    {
      "epoch": 0.13576903129141935,
      "grad_norm": 1.2610745429992676,
      "learning_rate": 9.709603728156021e-06,
      "loss": 0.8246,
      "step": 4712
    },
    {
      "epoch": 0.13579784475306864,
      "grad_norm": 1.247326374053955,
      "learning_rate": 9.7094470035244e-06,
      "loss": 0.7949,
      "step": 4713
    },
    {
      "epoch": 0.13582665821471793,
      "grad_norm": 1.2417181730270386,
      "learning_rate": 9.709290237878202e-06,
      "loss": 0.7908,
      "step": 4714
    },
    {
      "epoch": 0.1358554716763672,
      "grad_norm": 1.3560025691986084,
      "learning_rate": 9.709133431218793e-06,
      "loss": 0.7854,
      "step": 4715
    },
    {
      "epoch": 0.13588428513801648,
      "grad_norm": 1.3726829290390015,
      "learning_rate": 9.708976583547542e-06,
      "loss": 0.8386,
      "step": 4716
    },
    {
      "epoch": 0.13591309859966577,
      "grad_norm": 1.2304353713989258,
      "learning_rate": 9.70881969486581e-06,
      "loss": 0.7764,
      "step": 4717
    },
    {
      "epoch": 0.13594191206131503,
      "grad_norm": 1.3218406438827515,
      "learning_rate": 9.708662765174968e-06,
      "loss": 0.805,
      "step": 4718
    },
    {
      "epoch": 0.13597072552296433,
      "grad_norm": 1.3013919591903687,
      "learning_rate": 9.708505794476378e-06,
      "loss": 0.8077,
      "step": 4719
    },
    {
      "epoch": 0.13599953898461362,
      "grad_norm": 1.2770717144012451,
      "learning_rate": 9.70834878277141e-06,
      "loss": 0.8318,
      "step": 4720
    },
    {
      "epoch": 0.1360283524462629,
      "grad_norm": 1.3854235410690308,
      "learning_rate": 9.708191730061431e-06,
      "loss": 0.819,
      "step": 4721
    },
    {
      "epoch": 0.13605716590791217,
      "grad_norm": 1.2454415559768677,
      "learning_rate": 9.708034636347807e-06,
      "loss": 0.8301,
      "step": 4722
    },
    {
      "epoch": 0.13608597936956146,
      "grad_norm": 1.295049786567688,
      "learning_rate": 9.70787750163191e-06,
      "loss": 0.8169,
      "step": 4723
    },
    {
      "epoch": 0.13611479283121075,
      "grad_norm": 1.2588657140731812,
      "learning_rate": 9.707720325915105e-06,
      "loss": 0.7875,
      "step": 4724
    },
    {
      "epoch": 0.13614360629286001,
      "grad_norm": 1.2376585006713867,
      "learning_rate": 9.707563109198761e-06,
      "loss": 0.8149,
      "step": 4725
    },
    {
      "epoch": 0.1361724197545093,
      "grad_norm": 1.2604478597640991,
      "learning_rate": 9.707405851484251e-06,
      "loss": 0.8345,
      "step": 4726
    },
    {
      "epoch": 0.1362012332161586,
      "grad_norm": 1.254758596420288,
      "learning_rate": 9.70724855277294e-06,
      "loss": 0.7962,
      "step": 4727
    },
    {
      "epoch": 0.13623004667780786,
      "grad_norm": 1.274823784828186,
      "learning_rate": 9.707091213066197e-06,
      "loss": 0.8121,
      "step": 4728
    },
    {
      "epoch": 0.13625886013945715,
      "grad_norm": 1.3247411251068115,
      "learning_rate": 9.706933832365398e-06,
      "loss": 0.8073,
      "step": 4729
    },
    {
      "epoch": 0.13628767360110644,
      "grad_norm": 1.2623811960220337,
      "learning_rate": 9.70677641067191e-06,
      "loss": 0.7977,
      "step": 4730
    },
    {
      "epoch": 0.13631648706275573,
      "grad_norm": 1.2819455862045288,
      "learning_rate": 9.706618947987105e-06,
      "loss": 0.7978,
      "step": 4731
    },
    {
      "epoch": 0.136345300524405,
      "grad_norm": 1.3061342239379883,
      "learning_rate": 9.706461444312353e-06,
      "loss": 0.814,
      "step": 4732
    },
    {
      "epoch": 0.13637411398605429,
      "grad_norm": 1.2417875528335571,
      "learning_rate": 9.706303899649027e-06,
      "loss": 0.803,
      "step": 4733
    },
    {
      "epoch": 0.13640292744770358,
      "grad_norm": 1.2936357259750366,
      "learning_rate": 9.706146313998498e-06,
      "loss": 0.7964,
      "step": 4734
    },
    {
      "epoch": 0.13643174090935284,
      "grad_norm": 1.3547461032867432,
      "learning_rate": 9.70598868736214e-06,
      "loss": 0.7701,
      "step": 4735
    },
    {
      "epoch": 0.13646055437100213,
      "grad_norm": 1.2578837871551514,
      "learning_rate": 9.705831019741324e-06,
      "loss": 0.7788,
      "step": 4736
    },
    {
      "epoch": 0.13648936783265142,
      "grad_norm": 1.281528115272522,
      "learning_rate": 9.705673311137425e-06,
      "loss": 0.8043,
      "step": 4737
    },
    {
      "epoch": 0.1365181812943007,
      "grad_norm": 1.3425236940383911,
      "learning_rate": 9.705515561551814e-06,
      "loss": 0.8183,
      "step": 4738
    },
    {
      "epoch": 0.13654699475594997,
      "grad_norm": 1.3207179307937622,
      "learning_rate": 9.705357770985866e-06,
      "loss": 0.8018,
      "step": 4739
    },
    {
      "epoch": 0.13657580821759926,
      "grad_norm": 1.3353936672210693,
      "learning_rate": 9.705199939440957e-06,
      "loss": 0.8091,
      "step": 4740
    },
    {
      "epoch": 0.13660462167924856,
      "grad_norm": 1.2469546794891357,
      "learning_rate": 9.70504206691846e-06,
      "loss": 0.7989,
      "step": 4741
    },
    {
      "epoch": 0.13663343514089782,
      "grad_norm": 1.464715600013733,
      "learning_rate": 9.70488415341975e-06,
      "loss": 0.8285,
      "step": 4742
    },
    {
      "epoch": 0.1366622486025471,
      "grad_norm": 1.3461791276931763,
      "learning_rate": 9.7047261989462e-06,
      "loss": 0.828,
      "step": 4743
    },
    {
      "epoch": 0.1366910620641964,
      "grad_norm": 1.3076928853988647,
      "learning_rate": 9.704568203499189e-06,
      "loss": 0.7918,
      "step": 4744
    },
    {
      "epoch": 0.13671987552584566,
      "grad_norm": 1.3711395263671875,
      "learning_rate": 9.704410167080092e-06,
      "loss": 0.8359,
      "step": 4745
    },
    {
      "epoch": 0.13674868898749495,
      "grad_norm": 1.2751493453979492,
      "learning_rate": 9.704252089690284e-06,
      "loss": 0.8218,
      "step": 4746
    },
    {
      "epoch": 0.13677750244914424,
      "grad_norm": 1.311126947402954,
      "learning_rate": 9.704093971331146e-06,
      "loss": 0.8042,
      "step": 4747
    },
    {
      "epoch": 0.13680631591079354,
      "grad_norm": 1.3337390422821045,
      "learning_rate": 9.703935812004048e-06,
      "loss": 0.8088,
      "step": 4748
    },
    {
      "epoch": 0.1368351293724428,
      "grad_norm": 1.2692714929580688,
      "learning_rate": 9.703777611710375e-06,
      "loss": 0.7974,
      "step": 4749
    },
    {
      "epoch": 0.1368639428340921,
      "grad_norm": 1.3799164295196533,
      "learning_rate": 9.703619370451498e-06,
      "loss": 0.8065,
      "step": 4750
    },
    {
      "epoch": 0.13689275629574138,
      "grad_norm": 1.3320735692977905,
      "learning_rate": 9.703461088228801e-06,
      "loss": 0.7807,
      "step": 4751
    },
    {
      "epoch": 0.13692156975739064,
      "grad_norm": 1.353716492652893,
      "learning_rate": 9.703302765043656e-06,
      "loss": 0.8076,
      "step": 4752
    },
    {
      "epoch": 0.13695038321903993,
      "grad_norm": 1.3818289041519165,
      "learning_rate": 9.703144400897449e-06,
      "loss": 0.8044,
      "step": 4753
    },
    {
      "epoch": 0.13697919668068922,
      "grad_norm": 1.3081676959991455,
      "learning_rate": 9.702985995791554e-06,
      "loss": 0.7802,
      "step": 4754
    },
    {
      "epoch": 0.1370080101423385,
      "grad_norm": 1.339048981666565,
      "learning_rate": 9.702827549727351e-06,
      "loss": 0.8259,
      "step": 4755
    },
    {
      "epoch": 0.13703682360398778,
      "grad_norm": 1.291929006576538,
      "learning_rate": 9.702669062706223e-06,
      "loss": 0.7855,
      "step": 4756
    },
    {
      "epoch": 0.13706563706563707,
      "grad_norm": 1.3003777265548706,
      "learning_rate": 9.70251053472955e-06,
      "loss": 0.8068,
      "step": 4757
    },
    {
      "epoch": 0.13709445052728636,
      "grad_norm": 1.2418458461761475,
      "learning_rate": 9.702351965798708e-06,
      "loss": 0.8157,
      "step": 4758
    },
    {
      "epoch": 0.13712326398893562,
      "grad_norm": 1.2269953489303589,
      "learning_rate": 9.702193355915082e-06,
      "loss": 0.7965,
      "step": 4759
    },
    {
      "epoch": 0.1371520774505849,
      "grad_norm": 1.2312613725662231,
      "learning_rate": 9.702034705080052e-06,
      "loss": 0.8017,
      "step": 4760
    },
    {
      "epoch": 0.1371808909122342,
      "grad_norm": 1.297118902206421,
      "learning_rate": 9.701876013295e-06,
      "loss": 0.7955,
      "step": 4761
    },
    {
      "epoch": 0.13720970437388347,
      "grad_norm": 1.2681446075439453,
      "learning_rate": 9.701717280561309e-06,
      "loss": 0.8043,
      "step": 4762
    },
    {
      "epoch": 0.13723851783553276,
      "grad_norm": 1.2763137817382812,
      "learning_rate": 9.701558506880359e-06,
      "loss": 0.8349,
      "step": 4763
    },
    {
      "epoch": 0.13726733129718205,
      "grad_norm": 1.2943129539489746,
      "learning_rate": 9.701399692253537e-06,
      "loss": 0.8129,
      "step": 4764
    },
    {
      "epoch": 0.1372961447588313,
      "grad_norm": 1.2766033411026,
      "learning_rate": 9.701240836682222e-06,
      "loss": 0.8035,
      "step": 4765
    },
    {
      "epoch": 0.1373249582204806,
      "grad_norm": 1.3877280950546265,
      "learning_rate": 9.701081940167798e-06,
      "loss": 0.8169,
      "step": 4766
    },
    {
      "epoch": 0.1373537716821299,
      "grad_norm": 1.2767516374588013,
      "learning_rate": 9.700923002711649e-06,
      "loss": 0.8037,
      "step": 4767
    },
    {
      "epoch": 0.13738258514377918,
      "grad_norm": 1.2919213771820068,
      "learning_rate": 9.700764024315162e-06,
      "loss": 0.8235,
      "step": 4768
    },
    {
      "epoch": 0.13741139860542845,
      "grad_norm": 1.2702041864395142,
      "learning_rate": 9.700605004979719e-06,
      "loss": 0.7799,
      "step": 4769
    },
    {
      "epoch": 0.13744021206707774,
      "grad_norm": 1.2340362071990967,
      "learning_rate": 9.700445944706704e-06,
      "loss": 0.7917,
      "step": 4770
    },
    {
      "epoch": 0.13746902552872703,
      "grad_norm": 1.3694736957550049,
      "learning_rate": 9.700286843497504e-06,
      "loss": 0.8178,
      "step": 4771
    },
    {
      "epoch": 0.1374978389903763,
      "grad_norm": 1.2624797821044922,
      "learning_rate": 9.700127701353505e-06,
      "loss": 0.8312,
      "step": 4772
    },
    {
      "epoch": 0.13752665245202558,
      "grad_norm": 1.3259457349777222,
      "learning_rate": 9.69996851827609e-06,
      "loss": 0.8377,
      "step": 4773
    },
    {
      "epoch": 0.13755546591367487,
      "grad_norm": 1.359318733215332,
      "learning_rate": 9.69980929426665e-06,
      "loss": 0.7957,
      "step": 4774
    },
    {
      "epoch": 0.13758427937532416,
      "grad_norm": 1.2683541774749756,
      "learning_rate": 9.699650029326569e-06,
      "loss": 0.794,
      "step": 4775
    },
    {
      "epoch": 0.13761309283697343,
      "grad_norm": 1.267680287361145,
      "learning_rate": 9.699490723457232e-06,
      "loss": 0.804,
      "step": 4776
    },
    {
      "epoch": 0.13764190629862272,
      "grad_norm": 1.2818642854690552,
      "learning_rate": 9.699331376660031e-06,
      "loss": 0.7775,
      "step": 4777
    },
    {
      "epoch": 0.137670719760272,
      "grad_norm": 1.377102017402649,
      "learning_rate": 9.699171988936349e-06,
      "loss": 0.812,
      "step": 4778
    },
    {
      "epoch": 0.13769953322192127,
      "grad_norm": 1.3121304512023926,
      "learning_rate": 9.699012560287577e-06,
      "loss": 0.801,
      "step": 4779
    },
    {
      "epoch": 0.13772834668357056,
      "grad_norm": 1.3599828481674194,
      "learning_rate": 9.698853090715105e-06,
      "loss": 0.7998,
      "step": 4780
    },
    {
      "epoch": 0.13775716014521985,
      "grad_norm": 1.3131473064422607,
      "learning_rate": 9.698693580220317e-06,
      "loss": 0.7745,
      "step": 4781
    },
    {
      "epoch": 0.13778597360686912,
      "grad_norm": 1.2491075992584229,
      "learning_rate": 9.698534028804607e-06,
      "loss": 0.8111,
      "step": 4782
    },
    {
      "epoch": 0.1378147870685184,
      "grad_norm": 1.2654742002487183,
      "learning_rate": 9.698374436469361e-06,
      "loss": 0.8081,
      "step": 4783
    },
    {
      "epoch": 0.1378436005301677,
      "grad_norm": 1.2838577032089233,
      "learning_rate": 9.69821480321597e-06,
      "loss": 0.7994,
      "step": 4784
    },
    {
      "epoch": 0.137872413991817,
      "grad_norm": 1.2677812576293945,
      "learning_rate": 9.698055129045827e-06,
      "loss": 0.8401,
      "step": 4785
    },
    {
      "epoch": 0.13790122745346625,
      "grad_norm": 1.3159258365631104,
      "learning_rate": 9.69789541396032e-06,
      "loss": 0.8342,
      "step": 4786
    },
    {
      "epoch": 0.13793004091511554,
      "grad_norm": 1.3005445003509521,
      "learning_rate": 9.697735657960836e-06,
      "loss": 0.7886,
      "step": 4787
    },
    {
      "epoch": 0.13795885437676483,
      "grad_norm": 1.267246961593628,
      "learning_rate": 9.697575861048774e-06,
      "loss": 0.8043,
      "step": 4788
    },
    {
      "epoch": 0.1379876678384141,
      "grad_norm": 1.3077820539474487,
      "learning_rate": 9.697416023225522e-06,
      "loss": 0.7825,
      "step": 4789
    },
    {
      "epoch": 0.1380164813000634,
      "grad_norm": 1.3365007638931274,
      "learning_rate": 9.697256144492471e-06,
      "loss": 0.8062,
      "step": 4790
    },
    {
      "epoch": 0.13804529476171268,
      "grad_norm": 1.2825719118118286,
      "learning_rate": 9.697096224851015e-06,
      "loss": 0.7988,
      "step": 4791
    },
    {
      "epoch": 0.13807410822336194,
      "grad_norm": 1.4149651527404785,
      "learning_rate": 9.696936264302545e-06,
      "loss": 0.8187,
      "step": 4792
    },
    {
      "epoch": 0.13810292168501123,
      "grad_norm": 1.2357392311096191,
      "learning_rate": 9.696776262848458e-06,
      "loss": 0.8084,
      "step": 4793
    },
    {
      "epoch": 0.13813173514666052,
      "grad_norm": 1.3675792217254639,
      "learning_rate": 9.696616220490142e-06,
      "loss": 0.792,
      "step": 4794
    },
    {
      "epoch": 0.1381605486083098,
      "grad_norm": 1.3369061946868896,
      "learning_rate": 9.696456137228996e-06,
      "loss": 0.7821,
      "step": 4795
    },
    {
      "epoch": 0.13818936206995908,
      "grad_norm": 1.3132199048995972,
      "learning_rate": 9.69629601306641e-06,
      "loss": 0.7971,
      "step": 4796
    },
    {
      "epoch": 0.13821817553160837,
      "grad_norm": 1.210183024406433,
      "learning_rate": 9.69613584800378e-06,
      "loss": 0.777,
      "step": 4797
    },
    {
      "epoch": 0.13824698899325766,
      "grad_norm": 1.3275386095046997,
      "learning_rate": 9.695975642042504e-06,
      "loss": 0.8077,
      "step": 4798
    },
    {
      "epoch": 0.13827580245490692,
      "grad_norm": 1.355122447013855,
      "learning_rate": 9.695815395183971e-06,
      "loss": 0.8129,
      "step": 4799
    },
    {
      "epoch": 0.1383046159165562,
      "grad_norm": 1.283375859260559,
      "learning_rate": 9.695655107429583e-06,
      "loss": 0.7957,
      "step": 4800
    },
    {
      "epoch": 0.1383334293782055,
      "grad_norm": 1.2816718816757202,
      "learning_rate": 9.695494778780729e-06,
      "loss": 0.8044,
      "step": 4801
    },
    {
      "epoch": 0.1383622428398548,
      "grad_norm": 1.2780190706253052,
      "learning_rate": 9.695334409238813e-06,
      "loss": 0.7963,
      "step": 4802
    },
    {
      "epoch": 0.13839105630150406,
      "grad_norm": 1.2359527349472046,
      "learning_rate": 9.695173998805227e-06,
      "loss": 0.8062,
      "step": 4803
    },
    {
      "epoch": 0.13841986976315335,
      "grad_norm": 1.3237062692642212,
      "learning_rate": 9.695013547481368e-06,
      "loss": 0.7939,
      "step": 4804
    },
    {
      "epoch": 0.13844868322480264,
      "grad_norm": 1.3817020654678345,
      "learning_rate": 9.694853055268636e-06,
      "loss": 0.8074,
      "step": 4805
    },
    {
      "epoch": 0.1384774966864519,
      "grad_norm": 1.1930291652679443,
      "learning_rate": 9.694692522168424e-06,
      "loss": 0.8087,
      "step": 4806
    },
    {
      "epoch": 0.1385063101481012,
      "grad_norm": 1.3830739259719849,
      "learning_rate": 9.694531948182137e-06,
      "loss": 0.8144,
      "step": 4807
    },
    {
      "epoch": 0.13853512360975048,
      "grad_norm": 1.349410891532898,
      "learning_rate": 9.694371333311169e-06,
      "loss": 0.8106,
      "step": 4808
    },
    {
      "epoch": 0.13856393707139975,
      "grad_norm": 1.2364519834518433,
      "learning_rate": 9.694210677556916e-06,
      "loss": 0.7938,
      "step": 4809
    },
    {
      "epoch": 0.13859275053304904,
      "grad_norm": 1.3524283170700073,
      "learning_rate": 9.694049980920784e-06,
      "loss": 0.7739,
      "step": 4810
    },
    {
      "epoch": 0.13862156399469833,
      "grad_norm": 1.3414387702941895,
      "learning_rate": 9.693889243404166e-06,
      "loss": 0.8274,
      "step": 4811
    },
    {
      "epoch": 0.13865037745634762,
      "grad_norm": 1.3257454633712769,
      "learning_rate": 9.693728465008466e-06,
      "loss": 0.8186,
      "step": 4812
    },
    {
      "epoch": 0.13867919091799688,
      "grad_norm": 1.3277000188827515,
      "learning_rate": 9.693567645735084e-06,
      "loss": 0.8071,
      "step": 4813
    },
    {
      "epoch": 0.13870800437964617,
      "grad_norm": 1.272312879562378,
      "learning_rate": 9.693406785585419e-06,
      "loss": 0.8233,
      "step": 4814
    },
    {
      "epoch": 0.13873681784129546,
      "grad_norm": 1.2588675022125244,
      "learning_rate": 9.693245884560873e-06,
      "loss": 0.8006,
      "step": 4815
    },
    {
      "epoch": 0.13876563130294473,
      "grad_norm": 1.2590007781982422,
      "learning_rate": 9.693084942662847e-06,
      "loss": 0.8199,
      "step": 4816
    },
    {
      "epoch": 0.13879444476459402,
      "grad_norm": 1.2338881492614746,
      "learning_rate": 9.692923959892742e-06,
      "loss": 0.8114,
      "step": 4817
    },
    {
      "epoch": 0.1388232582262433,
      "grad_norm": 1.215193271636963,
      "learning_rate": 9.69276293625196e-06,
      "loss": 0.8014,
      "step": 4818
    },
    {
      "epoch": 0.13885207168789257,
      "grad_norm": 1.3581616878509521,
      "learning_rate": 9.692601871741905e-06,
      "loss": 0.8108,
      "step": 4819
    },
    {
      "epoch": 0.13888088514954186,
      "grad_norm": 1.3199782371520996,
      "learning_rate": 9.692440766363977e-06,
      "loss": 0.8016,
      "step": 4820
    },
    {
      "epoch": 0.13890969861119115,
      "grad_norm": 1.3391826152801514,
      "learning_rate": 9.692279620119581e-06,
      "loss": 0.8213,
      "step": 4821
    },
    {
      "epoch": 0.13893851207284044,
      "grad_norm": 1.2316166162490845,
      "learning_rate": 9.69211843301012e-06,
      "loss": 0.79,
      "step": 4822
    },
    {
      "epoch": 0.1389673255344897,
      "grad_norm": 1.3494757413864136,
      "learning_rate": 9.691957205036999e-06,
      "loss": 0.8263,
      "step": 4823
    },
    {
      "epoch": 0.138996138996139,
      "grad_norm": 1.325943946838379,
      "learning_rate": 9.69179593620162e-06,
      "loss": 0.8006,
      "step": 4824
    },
    {
      "epoch": 0.1390249524577883,
      "grad_norm": 1.2652841806411743,
      "learning_rate": 9.69163462650539e-06,
      "loss": 0.803,
      "step": 4825
    },
    {
      "epoch": 0.13905376591943755,
      "grad_norm": 1.396045446395874,
      "learning_rate": 9.691473275949712e-06,
      "loss": 0.8411,
      "step": 4826
    },
    {
      "epoch": 0.13908257938108684,
      "grad_norm": 1.3524894714355469,
      "learning_rate": 9.69131188453599e-06,
      "loss": 0.836,
      "step": 4827
    },
    {
      "epoch": 0.13911139284273613,
      "grad_norm": 1.4033235311508179,
      "learning_rate": 9.691150452265632e-06,
      "loss": 0.8162,
      "step": 4828
    },
    {
      "epoch": 0.13914020630438542,
      "grad_norm": 1.267081379890442,
      "learning_rate": 9.690988979140044e-06,
      "loss": 0.8275,
      "step": 4829
    },
    {
      "epoch": 0.13916901976603469,
      "grad_norm": 1.3044968843460083,
      "learning_rate": 9.69082746516063e-06,
      "loss": 0.7759,
      "step": 4830
    },
    {
      "epoch": 0.13919783322768398,
      "grad_norm": 1.349735140800476,
      "learning_rate": 9.690665910328798e-06,
      "loss": 0.777,
      "step": 4831
    },
    {
      "epoch": 0.13922664668933327,
      "grad_norm": 1.242240071296692,
      "learning_rate": 9.690504314645954e-06,
      "loss": 0.8044,
      "step": 4832
    },
    {
      "epoch": 0.13925546015098253,
      "grad_norm": 1.3813294172286987,
      "learning_rate": 9.690342678113508e-06,
      "loss": 0.7918,
      "step": 4833
    },
    {
      "epoch": 0.13928427361263182,
      "grad_norm": 1.3773962259292603,
      "learning_rate": 9.690181000732865e-06,
      "loss": 0.8275,
      "step": 4834
    },
    {
      "epoch": 0.1393130870742811,
      "grad_norm": 1.2464760541915894,
      "learning_rate": 9.690019282505435e-06,
      "loss": 0.8231,
      "step": 4835
    },
    {
      "epoch": 0.13934190053593037,
      "grad_norm": 1.344201922416687,
      "learning_rate": 9.689857523432623e-06,
      "loss": 0.8107,
      "step": 4836
    },
    {
      "epoch": 0.13937071399757966,
      "grad_norm": 1.382773995399475,
      "learning_rate": 9.689695723515841e-06,
      "loss": 0.7901,
      "step": 4837
    },
    {
      "epoch": 0.13939952745922896,
      "grad_norm": 1.3345681428909302,
      "learning_rate": 9.689533882756499e-06,
      "loss": 0.8238,
      "step": 4838
    },
    {
      "epoch": 0.13942834092087825,
      "grad_norm": 1.3057506084442139,
      "learning_rate": 9.689372001156003e-06,
      "loss": 0.7833,
      "step": 4839
    },
    {
      "epoch": 0.1394571543825275,
      "grad_norm": 1.4092609882354736,
      "learning_rate": 9.689210078715766e-06,
      "loss": 0.7994,
      "step": 4840
    },
    {
      "epoch": 0.1394859678441768,
      "grad_norm": 1.2608675956726074,
      "learning_rate": 9.689048115437194e-06,
      "loss": 0.8297,
      "step": 4841
    },
    {
      "epoch": 0.1395147813058261,
      "grad_norm": 1.2403110265731812,
      "learning_rate": 9.688886111321703e-06,
      "loss": 0.8128,
      "step": 4842
    },
    {
      "epoch": 0.13954359476747535,
      "grad_norm": 1.3151826858520508,
      "learning_rate": 9.6887240663707e-06,
      "loss": 0.8182,
      "step": 4843
    },
    {
      "epoch": 0.13957240822912464,
      "grad_norm": 1.245744228363037,
      "learning_rate": 9.688561980585597e-06,
      "loss": 0.8037,
      "step": 4844
    },
    {
      "epoch": 0.13960122169077394,
      "grad_norm": 1.277154803276062,
      "learning_rate": 9.688399853967805e-06,
      "loss": 0.8214,
      "step": 4845
    },
    {
      "epoch": 0.1396300351524232,
      "grad_norm": 1.3165501356124878,
      "learning_rate": 9.68823768651874e-06,
      "loss": 0.8284,
      "step": 4846
    },
    {
      "epoch": 0.1396588486140725,
      "grad_norm": 1.2060030698776245,
      "learning_rate": 9.68807547823981e-06,
      "loss": 0.7865,
      "step": 4847
    },
    {
      "epoch": 0.13968766207572178,
      "grad_norm": 1.3476464748382568,
      "learning_rate": 9.687913229132427e-06,
      "loss": 0.8285,
      "step": 4848
    },
    {
      "epoch": 0.13971647553737107,
      "grad_norm": 1.26659095287323,
      "learning_rate": 9.687750939198008e-06,
      "loss": 0.7921,
      "step": 4849
    },
    {
      "epoch": 0.13974528899902033,
      "grad_norm": 1.2794660329818726,
      "learning_rate": 9.687588608437963e-06,
      "loss": 0.7897,
      "step": 4850
    },
    {
      "epoch": 0.13977410246066962,
      "grad_norm": 1.23002290725708,
      "learning_rate": 9.68742623685371e-06,
      "loss": 0.784,
      "step": 4851
    },
    {
      "epoch": 0.13980291592231892,
      "grad_norm": 1.2497755289077759,
      "learning_rate": 9.687263824446658e-06,
      "loss": 0.7943,
      "step": 4852
    },
    {
      "epoch": 0.13983172938396818,
      "grad_norm": 1.2710611820220947,
      "learning_rate": 9.687101371218224e-06,
      "loss": 0.8075,
      "step": 4853
    },
    {
      "epoch": 0.13986054284561747,
      "grad_norm": 1.347102403640747,
      "learning_rate": 9.686938877169821e-06,
      "loss": 0.8336,
      "step": 4854
    },
    {
      "epoch": 0.13988935630726676,
      "grad_norm": 1.1707494258880615,
      "learning_rate": 9.686776342302868e-06,
      "loss": 0.7856,
      "step": 4855
    },
    {
      "epoch": 0.13991816976891605,
      "grad_norm": 1.222877025604248,
      "learning_rate": 9.686613766618777e-06,
      "loss": 0.782,
      "step": 4856
    },
    {
      "epoch": 0.1399469832305653,
      "grad_norm": 1.3199559450149536,
      "learning_rate": 9.686451150118965e-06,
      "loss": 0.7819,
      "step": 4857
    },
    {
      "epoch": 0.1399757966922146,
      "grad_norm": 1.210314393043518,
      "learning_rate": 9.686288492804847e-06,
      "loss": 0.8023,
      "step": 4858
    },
    {
      "epoch": 0.1400046101538639,
      "grad_norm": 1.3036351203918457,
      "learning_rate": 9.686125794677843e-06,
      "loss": 0.7998,
      "step": 4859
    },
    {
      "epoch": 0.14003342361551316,
      "grad_norm": 1.2065980434417725,
      "learning_rate": 9.685963055739366e-06,
      "loss": 0.8179,
      "step": 4860
    },
    {
      "epoch": 0.14006223707716245,
      "grad_norm": 1.3417726755142212,
      "learning_rate": 9.685800275990835e-06,
      "loss": 0.7823,
      "step": 4861
    },
    {
      "epoch": 0.14009105053881174,
      "grad_norm": 1.2163830995559692,
      "learning_rate": 9.68563745543367e-06,
      "loss": 0.7992,
      "step": 4862
    },
    {
      "epoch": 0.140119864000461,
      "grad_norm": 1.3483821153640747,
      "learning_rate": 9.685474594069285e-06,
      "loss": 0.8013,
      "step": 4863
    },
    {
      "epoch": 0.1401486774621103,
      "grad_norm": 1.2385449409484863,
      "learning_rate": 9.6853116918991e-06,
      "loss": 0.7841,
      "step": 4864
    },
    {
      "epoch": 0.14017749092375958,
      "grad_norm": 1.2890195846557617,
      "learning_rate": 9.685148748924534e-06,
      "loss": 0.8062,
      "step": 4865
    },
    {
      "epoch": 0.14020630438540888,
      "grad_norm": 1.2792260646820068,
      "learning_rate": 9.684985765147006e-06,
      "loss": 0.8202,
      "step": 4866
    },
    {
      "epoch": 0.14023511784705814,
      "grad_norm": 1.2891103029251099,
      "learning_rate": 9.684822740567934e-06,
      "loss": 0.7889,
      "step": 4867
    },
    {
      "epoch": 0.14026393130870743,
      "grad_norm": 1.2795811891555786,
      "learning_rate": 9.684659675188741e-06,
      "loss": 0.807,
      "step": 4868
    },
    {
      "epoch": 0.14029274477035672,
      "grad_norm": 1.2674559354782104,
      "learning_rate": 9.684496569010844e-06,
      "loss": 0.8042,
      "step": 4869
    },
    {
      "epoch": 0.14032155823200598,
      "grad_norm": 1.21268892288208,
      "learning_rate": 9.684333422035665e-06,
      "loss": 0.8033,
      "step": 4870
    },
    {
      "epoch": 0.14035037169365527,
      "grad_norm": 1.250778317451477,
      "learning_rate": 9.684170234264625e-06,
      "loss": 0.8152,
      "step": 4871
    },
    {
      "epoch": 0.14037918515530456,
      "grad_norm": 1.2975916862487793,
      "learning_rate": 9.684007005699143e-06,
      "loss": 0.8015,
      "step": 4872
    },
    {
      "epoch": 0.14040799861695383,
      "grad_norm": 1.2330516576766968,
      "learning_rate": 9.683843736340643e-06,
      "loss": 0.7916,
      "step": 4873
    },
    {
      "epoch": 0.14043681207860312,
      "grad_norm": 1.2783169746398926,
      "learning_rate": 9.683680426190547e-06,
      "loss": 0.8017,
      "step": 4874
    },
    {
      "epoch": 0.1404656255402524,
      "grad_norm": 1.3063541650772095,
      "learning_rate": 9.683517075250278e-06,
      "loss": 0.8156,
      "step": 4875
    },
    {
      "epoch": 0.1404944390019017,
      "grad_norm": 1.2538275718688965,
      "learning_rate": 9.683353683521254e-06,
      "loss": 0.8061,
      "step": 4876
    },
    {
      "epoch": 0.14052325246355096,
      "grad_norm": 1.2905099391937256,
      "learning_rate": 9.683190251004902e-06,
      "loss": 0.7994,
      "step": 4877
    },
    {
      "epoch": 0.14055206592520025,
      "grad_norm": 1.2630316019058228,
      "learning_rate": 9.683026777702645e-06,
      "loss": 0.78,
      "step": 4878
    },
    {
      "epoch": 0.14058087938684954,
      "grad_norm": 1.2342954874038696,
      "learning_rate": 9.682863263615905e-06,
      "loss": 0.8193,
      "step": 4879
    },
    {
      "epoch": 0.1406096928484988,
      "grad_norm": 1.3390275239944458,
      "learning_rate": 9.682699708746108e-06,
      "loss": 0.8089,
      "step": 4880
    },
    {
      "epoch": 0.1406385063101481,
      "grad_norm": 1.3070179224014282,
      "learning_rate": 9.682536113094678e-06,
      "loss": 0.8067,
      "step": 4881
    },
    {
      "epoch": 0.1406673197717974,
      "grad_norm": 1.305006980895996,
      "learning_rate": 9.682372476663038e-06,
      "loss": 0.8141,
      "step": 4882
    },
    {
      "epoch": 0.14069613323344668,
      "grad_norm": 1.4029717445373535,
      "learning_rate": 9.682208799452615e-06,
      "loss": 0.8162,
      "step": 4883
    },
    {
      "epoch": 0.14072494669509594,
      "grad_norm": 1.287441611289978,
      "learning_rate": 9.682045081464833e-06,
      "loss": 0.8079,
      "step": 4884
    },
    {
      "epoch": 0.14075376015674523,
      "grad_norm": 1.2365869283676147,
      "learning_rate": 9.68188132270112e-06,
      "loss": 0.8109,
      "step": 4885
    },
    {
      "epoch": 0.14078257361839452,
      "grad_norm": 1.3141827583312988,
      "learning_rate": 9.6817175231629e-06,
      "loss": 0.8212,
      "step": 4886
    },
    {
      "epoch": 0.1408113870800438,
      "grad_norm": 1.2772902250289917,
      "learning_rate": 9.681553682851601e-06,
      "loss": 0.7934,
      "step": 4887
    },
    {
      "epoch": 0.14084020054169308,
      "grad_norm": 1.3135148286819458,
      "learning_rate": 9.68138980176865e-06,
      "loss": 0.8134,
      "step": 4888
    },
    {
      "epoch": 0.14086901400334237,
      "grad_norm": 1.3198912143707275,
      "learning_rate": 9.681225879915472e-06,
      "loss": 0.7858,
      "step": 4889
    },
    {
      "epoch": 0.14089782746499163,
      "grad_norm": 1.2995270490646362,
      "learning_rate": 9.681061917293498e-06,
      "loss": 0.8227,
      "step": 4890
    },
    {
      "epoch": 0.14092664092664092,
      "grad_norm": 1.344009518623352,
      "learning_rate": 9.680897913904151e-06,
      "loss": 0.8301,
      "step": 4891
    },
    {
      "epoch": 0.1409554543882902,
      "grad_norm": 1.2921870946884155,
      "learning_rate": 9.680733869748864e-06,
      "loss": 0.8096,
      "step": 4892
    },
    {
      "epoch": 0.1409842678499395,
      "grad_norm": 1.3592978715896606,
      "learning_rate": 9.680569784829064e-06,
      "loss": 0.7795,
      "step": 4893
    },
    {
      "epoch": 0.14101308131158877,
      "grad_norm": 1.393542766571045,
      "learning_rate": 9.680405659146181e-06,
      "loss": 0.8093,
      "step": 4894
    },
    {
      "epoch": 0.14104189477323806,
      "grad_norm": 1.3610478639602661,
      "learning_rate": 9.680241492701642e-06,
      "loss": 0.8263,
      "step": 4895
    },
    {
      "epoch": 0.14107070823488735,
      "grad_norm": 1.3179271221160889,
      "learning_rate": 9.680077285496878e-06,
      "loss": 0.7976,
      "step": 4896
    },
    {
      "epoch": 0.1410995216965366,
      "grad_norm": 1.3265182971954346,
      "learning_rate": 9.67991303753332e-06,
      "loss": 0.7932,
      "step": 4897
    },
    {
      "epoch": 0.1411283351581859,
      "grad_norm": 1.402504324913025,
      "learning_rate": 9.679748748812397e-06,
      "loss": 0.8088,
      "step": 4898
    },
    {
      "epoch": 0.1411571486198352,
      "grad_norm": 1.2908706665039062,
      "learning_rate": 9.679584419335542e-06,
      "loss": 0.8276,
      "step": 4899
    },
    {
      "epoch": 0.14118596208148446,
      "grad_norm": 1.3211296796798706,
      "learning_rate": 9.679420049104183e-06,
      "loss": 0.7976,
      "step": 4900
    },
    {
      "epoch": 0.14121477554313375,
      "grad_norm": 1.3174219131469727,
      "learning_rate": 9.679255638119753e-06,
      "loss": 0.8004,
      "step": 4901
    },
    {
      "epoch": 0.14124358900478304,
      "grad_norm": 1.2361470460891724,
      "learning_rate": 9.679091186383685e-06,
      "loss": 0.7969,
      "step": 4902
    },
    {
      "epoch": 0.14127240246643233,
      "grad_norm": 1.2555344104766846,
      "learning_rate": 9.678926693897411e-06,
      "loss": 0.8104,
      "step": 4903
    },
    {
      "epoch": 0.1413012159280816,
      "grad_norm": 1.3202030658721924,
      "learning_rate": 9.67876216066236e-06,
      "loss": 0.782,
      "step": 4904
    },
    {
      "epoch": 0.14133002938973088,
      "grad_norm": 1.2985578775405884,
      "learning_rate": 9.678597586679969e-06,
      "loss": 0.7892,
      "step": 4905
    },
    {
      "epoch": 0.14135884285138017,
      "grad_norm": 1.2335941791534424,
      "learning_rate": 9.67843297195167e-06,
      "loss": 0.7982,
      "step": 4906
    },
    {
      "epoch": 0.14138765631302944,
      "grad_norm": 1.3198429346084595,
      "learning_rate": 9.678268316478896e-06,
      "loss": 0.8166,
      "step": 4907
    },
    {
      "epoch": 0.14141646977467873,
      "grad_norm": 1.2586804628372192,
      "learning_rate": 9.678103620263082e-06,
      "loss": 0.7911,
      "step": 4908
    },
    {
      "epoch": 0.14144528323632802,
      "grad_norm": 1.206203818321228,
      "learning_rate": 9.677938883305663e-06,
      "loss": 0.8308,
      "step": 4909
    },
    {
      "epoch": 0.1414740966979773,
      "grad_norm": 1.3077008724212646,
      "learning_rate": 9.67777410560807e-06,
      "loss": 0.8098,
      "step": 4910
    },
    {
      "epoch": 0.14150291015962657,
      "grad_norm": 1.2583147287368774,
      "learning_rate": 9.677609287171742e-06,
      "loss": 0.8177,
      "step": 4911
    },
    {
      "epoch": 0.14153172362127586,
      "grad_norm": 1.3289834260940552,
      "learning_rate": 9.677444427998113e-06,
      "loss": 0.8199,
      "step": 4912
    },
    {
      "epoch": 0.14156053708292515,
      "grad_norm": 1.2865784168243408,
      "learning_rate": 9.677279528088619e-06,
      "loss": 0.8199,
      "step": 4913
    },
    {
      "epoch": 0.14158935054457442,
      "grad_norm": 1.2756669521331787,
      "learning_rate": 9.677114587444695e-06,
      "loss": 0.8275,
      "step": 4914
    },
    {
      "epoch": 0.1416181640062237,
      "grad_norm": 1.302334189414978,
      "learning_rate": 9.676949606067779e-06,
      "loss": 0.7916,
      "step": 4915
    },
    {
      "epoch": 0.141646977467873,
      "grad_norm": 1.238903522491455,
      "learning_rate": 9.676784583959306e-06,
      "loss": 0.8123,
      "step": 4916
    },
    {
      "epoch": 0.14167579092952226,
      "grad_norm": 1.31707763671875,
      "learning_rate": 9.676619521120717e-06,
      "loss": 0.795,
      "step": 4917
    },
    {
      "epoch": 0.14170460439117155,
      "grad_norm": 1.2740706205368042,
      "learning_rate": 9.676454417553445e-06,
      "loss": 0.811,
      "step": 4918
    },
    {
      "epoch": 0.14173341785282084,
      "grad_norm": 1.2806487083435059,
      "learning_rate": 9.67628927325893e-06,
      "loss": 0.79,
      "step": 4919
    },
    {
      "epoch": 0.14176223131447013,
      "grad_norm": 1.373518705368042,
      "learning_rate": 9.67612408823861e-06,
      "loss": 0.8104,
      "step": 4920
    },
    {
      "epoch": 0.1417910447761194,
      "grad_norm": 1.234704613685608,
      "learning_rate": 9.675958862493923e-06,
      "loss": 0.7922,
      "step": 4921
    },
    {
      "epoch": 0.1418198582377687,
      "grad_norm": 1.281395673751831,
      "learning_rate": 9.675793596026311e-06,
      "loss": 0.8095,
      "step": 4922
    },
    {
      "epoch": 0.14184867169941798,
      "grad_norm": 1.2356231212615967,
      "learning_rate": 9.67562828883721e-06,
      "loss": 0.8048,
      "step": 4923
    },
    {
      "epoch": 0.14187748516106724,
      "grad_norm": 1.1936920881271362,
      "learning_rate": 9.675462940928057e-06,
      "loss": 0.7921,
      "step": 4924
    },
    {
      "epoch": 0.14190629862271653,
      "grad_norm": 1.1847294569015503,
      "learning_rate": 9.675297552300299e-06,
      "loss": 0.8002,
      "step": 4925
    },
    {
      "epoch": 0.14193511208436582,
      "grad_norm": 1.2614957094192505,
      "learning_rate": 9.675132122955373e-06,
      "loss": 0.803,
      "step": 4926
    },
    {
      "epoch": 0.14196392554601509,
      "grad_norm": 1.2810717821121216,
      "learning_rate": 9.674966652894717e-06,
      "loss": 0.8341,
      "step": 4927
    },
    {
      "epoch": 0.14199273900766438,
      "grad_norm": 1.2277518510818481,
      "learning_rate": 9.674801142119777e-06,
      "loss": 0.7957,
      "step": 4928
    },
    {
      "epoch": 0.14202155246931367,
      "grad_norm": 1.212155818939209,
      "learning_rate": 9.674635590631991e-06,
      "loss": 0.8078,
      "step": 4929
    },
    {
      "epoch": 0.14205036593096296,
      "grad_norm": 1.2154455184936523,
      "learning_rate": 9.674469998432803e-06,
      "loss": 0.802,
      "step": 4930
    },
    {
      "epoch": 0.14207917939261222,
      "grad_norm": 1.2258505821228027,
      "learning_rate": 9.674304365523654e-06,
      "loss": 0.7962,
      "step": 4931
    },
    {
      "epoch": 0.1421079928542615,
      "grad_norm": 1.2616331577301025,
      "learning_rate": 9.674138691905985e-06,
      "loss": 0.8043,
      "step": 4932
    },
    {
      "epoch": 0.1421368063159108,
      "grad_norm": 1.2673779726028442,
      "learning_rate": 9.673972977581242e-06,
      "loss": 0.8035,
      "step": 4933
    },
    {
      "epoch": 0.14216561977756006,
      "grad_norm": 1.1742838621139526,
      "learning_rate": 9.673807222550866e-06,
      "loss": 0.7831,
      "step": 4934
    },
    {
      "epoch": 0.14219443323920936,
      "grad_norm": 1.2424532175064087,
      "learning_rate": 9.673641426816303e-06,
      "loss": 0.7899,
      "step": 4935
    },
    {
      "epoch": 0.14222324670085865,
      "grad_norm": 1.2888916730880737,
      "learning_rate": 9.673475590378993e-06,
      "loss": 0.7867,
      "step": 4936
    },
    {
      "epoch": 0.14225206016250794,
      "grad_norm": 1.2324824333190918,
      "learning_rate": 9.673309713240382e-06,
      "loss": 0.7838,
      "step": 4937
    },
    {
      "epoch": 0.1422808736241572,
      "grad_norm": 1.250747561454773,
      "learning_rate": 9.673143795401917e-06,
      "loss": 0.807,
      "step": 4938
    },
    {
      "epoch": 0.1423096870858065,
      "grad_norm": 1.3023229837417603,
      "learning_rate": 9.67297783686504e-06,
      "loss": 0.7941,
      "step": 4939
    },
    {
      "epoch": 0.14233850054745578,
      "grad_norm": 1.240322470664978,
      "learning_rate": 9.672811837631197e-06,
      "loss": 0.8099,
      "step": 4940
    },
    {
      "epoch": 0.14236731400910504,
      "grad_norm": 1.2585355043411255,
      "learning_rate": 9.672645797701834e-06,
      "loss": 0.7939,
      "step": 4941
    },
    {
      "epoch": 0.14239612747075434,
      "grad_norm": 1.272895336151123,
      "learning_rate": 9.672479717078396e-06,
      "loss": 0.8045,
      "step": 4942
    },
    {
      "epoch": 0.14242494093240363,
      "grad_norm": 1.232157826423645,
      "learning_rate": 9.672313595762332e-06,
      "loss": 0.8181,
      "step": 4943
    },
    {
      "epoch": 0.1424537543940529,
      "grad_norm": 1.2587908506393433,
      "learning_rate": 9.672147433755086e-06,
      "loss": 0.7913,
      "step": 4944
    },
    {
      "epoch": 0.14248256785570218,
      "grad_norm": 1.2823090553283691,
      "learning_rate": 9.671981231058107e-06,
      "loss": 0.8255,
      "step": 4945
    },
    {
      "epoch": 0.14251138131735147,
      "grad_norm": 1.246468424797058,
      "learning_rate": 9.671814987672844e-06,
      "loss": 0.8112,
      "step": 4946
    },
    {
      "epoch": 0.14254019477900076,
      "grad_norm": 1.340595006942749,
      "learning_rate": 9.67164870360074e-06,
      "loss": 0.7977,
      "step": 4947
    },
    {
      "epoch": 0.14256900824065002,
      "grad_norm": 1.287240982055664,
      "learning_rate": 9.671482378843246e-06,
      "loss": 0.8012,
      "step": 4948
    },
    {
      "epoch": 0.14259782170229932,
      "grad_norm": 1.2617454528808594,
      "learning_rate": 9.67131601340181e-06,
      "loss": 0.8269,
      "step": 4949
    },
    {
      "epoch": 0.1426266351639486,
      "grad_norm": 1.2723010778427124,
      "learning_rate": 9.671149607277882e-06,
      "loss": 0.795,
      "step": 4950
    },
    {
      "epoch": 0.14265544862559787,
      "grad_norm": 1.2668243646621704,
      "learning_rate": 9.670983160472909e-06,
      "loss": 0.8194,
      "step": 4951
    },
    {
      "epoch": 0.14268426208724716,
      "grad_norm": 1.2170584201812744,
      "learning_rate": 9.670816672988344e-06,
      "loss": 0.8066,
      "step": 4952
    },
    {
      "epoch": 0.14271307554889645,
      "grad_norm": 1.2291278839111328,
      "learning_rate": 9.670650144825633e-06,
      "loss": 0.8066,
      "step": 4953
    },
    {
      "epoch": 0.14274188901054571,
      "grad_norm": 1.2796001434326172,
      "learning_rate": 9.67048357598623e-06,
      "loss": 0.8105,
      "step": 4954
    },
    {
      "epoch": 0.142770702472195,
      "grad_norm": 1.243789553642273,
      "learning_rate": 9.67031696647158e-06,
      "loss": 0.8155,
      "step": 4955
    },
    {
      "epoch": 0.1427995159338443,
      "grad_norm": 1.2557501792907715,
      "learning_rate": 9.670150316283142e-06,
      "loss": 0.8046,
      "step": 4956
    },
    {
      "epoch": 0.1428283293954936,
      "grad_norm": 1.2543843984603882,
      "learning_rate": 9.669983625422361e-06,
      "loss": 0.8263,
      "step": 4957
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 1.3042387962341309,
      "learning_rate": 9.669816893890692e-06,
      "loss": 0.8146,
      "step": 4958
    },
    {
      "epoch": 0.14288595631879214,
      "grad_norm": 1.2790685892105103,
      "learning_rate": 9.669650121689586e-06,
      "loss": 0.8078,
      "step": 4959
    },
    {
      "epoch": 0.14291476978044143,
      "grad_norm": 1.2458288669586182,
      "learning_rate": 9.669483308820496e-06,
      "loss": 0.8265,
      "step": 4960
    },
    {
      "epoch": 0.1429435832420907,
      "grad_norm": 1.2031961679458618,
      "learning_rate": 9.669316455284872e-06,
      "loss": 0.8006,
      "step": 4961
    },
    {
      "epoch": 0.14297239670373998,
      "grad_norm": 1.2305611371994019,
      "learning_rate": 9.66914956108417e-06,
      "loss": 0.816,
      "step": 4962
    },
    {
      "epoch": 0.14300121016538928,
      "grad_norm": 1.2154966592788696,
      "learning_rate": 9.668982626219843e-06,
      "loss": 0.7989,
      "step": 4963
    },
    {
      "epoch": 0.14303002362703857,
      "grad_norm": 1.2465656995773315,
      "learning_rate": 9.668815650693345e-06,
      "loss": 0.7895,
      "step": 4964
    },
    {
      "epoch": 0.14305883708868783,
      "grad_norm": 1.2592490911483765,
      "learning_rate": 9.66864863450613e-06,
      "loss": 0.7938,
      "step": 4965
    },
    {
      "epoch": 0.14308765055033712,
      "grad_norm": 1.1958832740783691,
      "learning_rate": 9.668481577659652e-06,
      "loss": 0.8247,
      "step": 4966
    },
    {
      "epoch": 0.1431164640119864,
      "grad_norm": 1.3463859558105469,
      "learning_rate": 9.668314480155365e-06,
      "loss": 0.7977,
      "step": 4967
    },
    {
      "epoch": 0.14314527747363567,
      "grad_norm": 1.2294542789459229,
      "learning_rate": 9.668147341994726e-06,
      "loss": 0.7902,
      "step": 4968
    },
    {
      "epoch": 0.14317409093528496,
      "grad_norm": 1.296130895614624,
      "learning_rate": 9.667980163179191e-06,
      "loss": 0.7846,
      "step": 4969
    },
    {
      "epoch": 0.14320290439693426,
      "grad_norm": 1.303221583366394,
      "learning_rate": 9.667812943710215e-06,
      "loss": 0.7868,
      "step": 4970
    },
    {
      "epoch": 0.14323171785858352,
      "grad_norm": 1.1664745807647705,
      "learning_rate": 9.667645683589255e-06,
      "loss": 0.8053,
      "step": 4971
    },
    {
      "epoch": 0.1432605313202328,
      "grad_norm": 1.2361587285995483,
      "learning_rate": 9.667478382817767e-06,
      "loss": 0.8113,
      "step": 4972
    },
    {
      "epoch": 0.1432893447818821,
      "grad_norm": 1.376214623451233,
      "learning_rate": 9.667311041397207e-06,
      "loss": 0.799,
      "step": 4973
    },
    {
      "epoch": 0.1433181582435314,
      "grad_norm": 1.2947651147842407,
      "learning_rate": 9.667143659329032e-06,
      "loss": 0.7982,
      "step": 4974
    },
    {
      "epoch": 0.14334697170518065,
      "grad_norm": 1.2430895566940308,
      "learning_rate": 9.666976236614705e-06,
      "loss": 0.7826,
      "step": 4975
    },
    {
      "epoch": 0.14337578516682994,
      "grad_norm": 1.3755460977554321,
      "learning_rate": 9.666808773255677e-06,
      "loss": 0.8102,
      "step": 4976
    },
    {
      "epoch": 0.14340459862847924,
      "grad_norm": 1.1978458166122437,
      "learning_rate": 9.66664126925341e-06,
      "loss": 0.7856,
      "step": 4977
    },
    {
      "epoch": 0.1434334120901285,
      "grad_norm": 1.3121546506881714,
      "learning_rate": 9.666473724609364e-06,
      "loss": 0.8148,
      "step": 4978
    },
    {
      "epoch": 0.1434622255517778,
      "grad_norm": 1.2005785703659058,
      "learning_rate": 9.666306139324996e-06,
      "loss": 0.8024,
      "step": 4979
    },
    {
      "epoch": 0.14349103901342708,
      "grad_norm": 1.2024774551391602,
      "learning_rate": 9.666138513401766e-06,
      "loss": 0.7894,
      "step": 4980
    },
    {
      "epoch": 0.14351985247507634,
      "grad_norm": 1.2942936420440674,
      "learning_rate": 9.665970846841133e-06,
      "loss": 0.7812,
      "step": 4981
    },
    {
      "epoch": 0.14354866593672563,
      "grad_norm": 1.3260722160339355,
      "learning_rate": 9.66580313964456e-06,
      "loss": 0.8062,
      "step": 4982
    },
    {
      "epoch": 0.14357747939837492,
      "grad_norm": 1.3314889669418335,
      "learning_rate": 9.665635391813504e-06,
      "loss": 0.8027,
      "step": 4983
    },
    {
      "epoch": 0.14360629286002422,
      "grad_norm": 1.1664973497390747,
      "learning_rate": 9.66546760334943e-06,
      "loss": 0.8024,
      "step": 4984
    },
    {
      "epoch": 0.14363510632167348,
      "grad_norm": 1.2688188552856445,
      "learning_rate": 9.665299774253794e-06,
      "loss": 0.8209,
      "step": 4985
    },
    {
      "epoch": 0.14366391978332277,
      "grad_norm": 1.3426854610443115,
      "learning_rate": 9.665131904528063e-06,
      "loss": 0.8317,
      "step": 4986
    },
    {
      "epoch": 0.14369273324497206,
      "grad_norm": 1.2312889099121094,
      "learning_rate": 9.664963994173695e-06,
      "loss": 0.8215,
      "step": 4987
    },
    {
      "epoch": 0.14372154670662132,
      "grad_norm": 1.1656450033187866,
      "learning_rate": 9.664796043192155e-06,
      "loss": 0.7579,
      "step": 4988
    },
    {
      "epoch": 0.1437503601682706,
      "grad_norm": 1.3166192770004272,
      "learning_rate": 9.664628051584903e-06,
      "loss": 0.8002,
      "step": 4989
    },
    {
      "epoch": 0.1437791736299199,
      "grad_norm": 1.2814044952392578,
      "learning_rate": 9.664460019353406e-06,
      "loss": 0.8084,
      "step": 4990
    },
    {
      "epoch": 0.1438079870915692,
      "grad_norm": 1.179755449295044,
      "learning_rate": 9.664291946499122e-06,
      "loss": 0.7966,
      "step": 4991
    },
    {
      "epoch": 0.14383680055321846,
      "grad_norm": 1.196044921875,
      "learning_rate": 9.66412383302352e-06,
      "loss": 0.8032,
      "step": 4992
    },
    {
      "epoch": 0.14386561401486775,
      "grad_norm": 1.2102775573730469,
      "learning_rate": 9.66395567892806e-06,
      "loss": 0.7937,
      "step": 4993
    },
    {
      "epoch": 0.14389442747651704,
      "grad_norm": 1.2362563610076904,
      "learning_rate": 9.663787484214211e-06,
      "loss": 0.7979,
      "step": 4994
    },
    {
      "epoch": 0.1439232409381663,
      "grad_norm": 1.2280904054641724,
      "learning_rate": 9.663619248883434e-06,
      "loss": 0.8028,
      "step": 4995
    },
    {
      "epoch": 0.1439520543998156,
      "grad_norm": 1.2159686088562012,
      "learning_rate": 9.663450972937193e-06,
      "loss": 0.8027,
      "step": 4996
    },
    {
      "epoch": 0.14398086786146488,
      "grad_norm": 1.2207695245742798,
      "learning_rate": 9.66328265637696e-06,
      "loss": 0.7969,
      "step": 4997
    },
    {
      "epoch": 0.14400968132311415,
      "grad_norm": 1.2003958225250244,
      "learning_rate": 9.663114299204193e-06,
      "loss": 0.8283,
      "step": 4998
    },
    {
      "epoch": 0.14403849478476344,
      "grad_norm": 1.1939539909362793,
      "learning_rate": 9.662945901420364e-06,
      "loss": 0.8031,
      "step": 4999
    },
    {
      "epoch": 0.14406730824641273,
      "grad_norm": 1.3108806610107422,
      "learning_rate": 9.662777463026936e-06,
      "loss": 0.8129,
      "step": 5000
    },
    {
      "epoch": 0.14409612170806202,
      "grad_norm": 1.2375285625457764,
      "learning_rate": 9.662608984025376e-06,
      "loss": 0.7933,
      "step": 5001
    },
    {
      "epoch": 0.14412493516971128,
      "grad_norm": 1.1715461015701294,
      "learning_rate": 9.662440464417156e-06,
      "loss": 0.8038,
      "step": 5002
    },
    {
      "epoch": 0.14415374863136057,
      "grad_norm": 1.357311725616455,
      "learning_rate": 9.662271904203739e-06,
      "loss": 0.8026,
      "step": 5003
    },
    {
      "epoch": 0.14418256209300986,
      "grad_norm": 1.2329838275909424,
      "learning_rate": 9.662103303386593e-06,
      "loss": 0.8025,
      "step": 5004
    },
    {
      "epoch": 0.14421137555465913,
      "grad_norm": 1.230563998222351,
      "learning_rate": 9.66193466196719e-06,
      "loss": 0.7994,
      "step": 5005
    },
    {
      "epoch": 0.14424018901630842,
      "grad_norm": 1.2736088037490845,
      "learning_rate": 9.661765979946995e-06,
      "loss": 0.8022,
      "step": 5006
    },
    {
      "epoch": 0.1442690024779577,
      "grad_norm": 1.266280174255371,
      "learning_rate": 9.661597257327477e-06,
      "loss": 0.8058,
      "step": 5007
    },
    {
      "epoch": 0.14429781593960697,
      "grad_norm": 1.2751259803771973,
      "learning_rate": 9.661428494110109e-06,
      "loss": 0.7875,
      "step": 5008
    },
    {
      "epoch": 0.14432662940125626,
      "grad_norm": 2.473926544189453,
      "learning_rate": 9.661259690296358e-06,
      "loss": 0.7937,
      "step": 5009
    },
    {
      "epoch": 0.14435544286290555,
      "grad_norm": 1.340908408164978,
      "learning_rate": 9.661090845887693e-06,
      "loss": 0.7995,
      "step": 5010
    },
    {
      "epoch": 0.14438425632455484,
      "grad_norm": 1.3656423091888428,
      "learning_rate": 9.660921960885588e-06,
      "loss": 0.7814,
      "step": 5011
    },
    {
      "epoch": 0.1444130697862041,
      "grad_norm": 1.2772164344787598,
      "learning_rate": 9.660753035291513e-06,
      "loss": 0.815,
      "step": 5012
    },
    {
      "epoch": 0.1444418832478534,
      "grad_norm": 1.2891055345535278,
      "learning_rate": 9.660584069106935e-06,
      "loss": 0.7924,
      "step": 5013
    },
    {
      "epoch": 0.1444706967095027,
      "grad_norm": 1.2611339092254639,
      "learning_rate": 9.66041506233333e-06,
      "loss": 0.797,
      "step": 5014
    },
    {
      "epoch": 0.14449951017115195,
      "grad_norm": 1.217200517654419,
      "learning_rate": 9.660246014972168e-06,
      "loss": 0.8105,
      "step": 5015
    },
    {
      "epoch": 0.14452832363280124,
      "grad_norm": 1.245303988456726,
      "learning_rate": 9.660076927024925e-06,
      "loss": 0.7928,
      "step": 5016
    },
    {
      "epoch": 0.14455713709445053,
      "grad_norm": 1.3345818519592285,
      "learning_rate": 9.659907798493067e-06,
      "loss": 0.7955,
      "step": 5017
    },
    {
      "epoch": 0.1445859505560998,
      "grad_norm": 1.2427008152008057,
      "learning_rate": 9.659738629378071e-06,
      "loss": 0.7739,
      "step": 5018
    },
    {
      "epoch": 0.1446147640177491,
      "grad_norm": 1.2778455018997192,
      "learning_rate": 9.65956941968141e-06,
      "loss": 0.8175,
      "step": 5019
    },
    {
      "epoch": 0.14464357747939838,
      "grad_norm": 1.2378736734390259,
      "learning_rate": 9.659400169404556e-06,
      "loss": 0.8143,
      "step": 5020
    },
    {
      "epoch": 0.14467239094104767,
      "grad_norm": 1.2599718570709229,
      "learning_rate": 9.659230878548984e-06,
      "loss": 0.808,
      "step": 5021
    },
    {
      "epoch": 0.14470120440269693,
      "grad_norm": 1.2614662647247314,
      "learning_rate": 9.65906154711617e-06,
      "loss": 0.8134,
      "step": 5022
    },
    {
      "epoch": 0.14473001786434622,
      "grad_norm": 1.2116410732269287,
      "learning_rate": 9.658892175107588e-06,
      "loss": 0.7975,
      "step": 5023
    },
    {
      "epoch": 0.1447588313259955,
      "grad_norm": 1.1831072568893433,
      "learning_rate": 9.65872276252471e-06,
      "loss": 0.7793,
      "step": 5024
    },
    {
      "epoch": 0.14478764478764478,
      "grad_norm": 1.2603486776351929,
      "learning_rate": 9.658553309369017e-06,
      "loss": 0.8103,
      "step": 5025
    },
    {
      "epoch": 0.14481645824929407,
      "grad_norm": 1.2738901376724243,
      "learning_rate": 9.658383815641978e-06,
      "loss": 0.8291,
      "step": 5026
    },
    {
      "epoch": 0.14484527171094336,
      "grad_norm": 1.2502479553222656,
      "learning_rate": 9.658214281345075e-06,
      "loss": 0.8066,
      "step": 5027
    },
    {
      "epoch": 0.14487408517259265,
      "grad_norm": 1.2393239736557007,
      "learning_rate": 9.658044706479781e-06,
      "loss": 0.7688,
      "step": 5028
    },
    {
      "epoch": 0.1449028986342419,
      "grad_norm": 1.1578788757324219,
      "learning_rate": 9.657875091047574e-06,
      "loss": 0.7706,
      "step": 5029
    },
    {
      "epoch": 0.1449317120958912,
      "grad_norm": 1.1877856254577637,
      "learning_rate": 9.657705435049931e-06,
      "loss": 0.8012,
      "step": 5030
    },
    {
      "epoch": 0.1449605255575405,
      "grad_norm": 1.2781363725662231,
      "learning_rate": 9.65753573848833e-06,
      "loss": 0.7928,
      "step": 5031
    },
    {
      "epoch": 0.14498933901918976,
      "grad_norm": 1.177964687347412,
      "learning_rate": 9.657366001364249e-06,
      "loss": 0.7874,
      "step": 5032
    },
    {
      "epoch": 0.14501815248083905,
      "grad_norm": 1.242102861404419,
      "learning_rate": 9.657196223679165e-06,
      "loss": 0.7913,
      "step": 5033
    },
    {
      "epoch": 0.14504696594248834,
      "grad_norm": 1.2623167037963867,
      "learning_rate": 9.657026405434557e-06,
      "loss": 0.7899,
      "step": 5034
    },
    {
      "epoch": 0.1450757794041376,
      "grad_norm": 1.224433183670044,
      "learning_rate": 9.656856546631906e-06,
      "loss": 0.8055,
      "step": 5035
    },
    {
      "epoch": 0.1451045928657869,
      "grad_norm": 1.2230035066604614,
      "learning_rate": 9.656686647272688e-06,
      "loss": 0.7874,
      "step": 5036
    },
    {
      "epoch": 0.14513340632743618,
      "grad_norm": 1.256390929222107,
      "learning_rate": 9.656516707358385e-06,
      "loss": 0.7718,
      "step": 5037
    },
    {
      "epoch": 0.14516221978908547,
      "grad_norm": 1.267347812652588,
      "learning_rate": 9.656346726890477e-06,
      "loss": 0.7894,
      "step": 5038
    },
    {
      "epoch": 0.14519103325073474,
      "grad_norm": 1.2294261455535889,
      "learning_rate": 9.656176705870443e-06,
      "loss": 0.804,
      "step": 5039
    },
    {
      "epoch": 0.14521984671238403,
      "grad_norm": 1.1906625032424927,
      "learning_rate": 9.656006644299765e-06,
      "loss": 0.7773,
      "step": 5040
    },
    {
      "epoch": 0.14524866017403332,
      "grad_norm": 1.2562490701675415,
      "learning_rate": 9.655836542179921e-06,
      "loss": 0.782,
      "step": 5041
    },
    {
      "epoch": 0.14527747363568258,
      "grad_norm": 1.2506746053695679,
      "learning_rate": 9.655666399512399e-06,
      "loss": 0.8178,
      "step": 5042
    },
    {
      "epoch": 0.14530628709733187,
      "grad_norm": 1.2072360515594482,
      "learning_rate": 9.655496216298674e-06,
      "loss": 0.8029,
      "step": 5043
    },
    {
      "epoch": 0.14533510055898116,
      "grad_norm": 1.2420592308044434,
      "learning_rate": 9.65532599254023e-06,
      "loss": 0.7991,
      "step": 5044
    },
    {
      "epoch": 0.14536391402063042,
      "grad_norm": 1.2084338665008545,
      "learning_rate": 9.65515572823855e-06,
      "loss": 0.8149,
      "step": 5045
    },
    {
      "epoch": 0.14539272748227972,
      "grad_norm": 1.2093805074691772,
      "learning_rate": 9.65498542339512e-06,
      "loss": 0.7798,
      "step": 5046
    },
    {
      "epoch": 0.145421540943929,
      "grad_norm": 1.2796610593795776,
      "learning_rate": 9.654815078011418e-06,
      "loss": 0.804,
      "step": 5047
    },
    {
      "epoch": 0.1454503544055783,
      "grad_norm": 1.200134038925171,
      "learning_rate": 9.65464469208893e-06,
      "loss": 0.8094,
      "step": 5048
    },
    {
      "epoch": 0.14547916786722756,
      "grad_norm": 1.2399064302444458,
      "learning_rate": 9.65447426562914e-06,
      "loss": 0.813,
      "step": 5049
    },
    {
      "epoch": 0.14550798132887685,
      "grad_norm": 1.2232182025909424,
      "learning_rate": 9.654303798633534e-06,
      "loss": 0.8002,
      "step": 5050
    },
    {
      "epoch": 0.14553679479052614,
      "grad_norm": 1.2211941480636597,
      "learning_rate": 9.65413329110359e-06,
      "loss": 0.7866,
      "step": 5051
    },
    {
      "epoch": 0.1455656082521754,
      "grad_norm": 1.2474193572998047,
      "learning_rate": 9.653962743040802e-06,
      "loss": 0.8043,
      "step": 5052
    },
    {
      "epoch": 0.1455944217138247,
      "grad_norm": 1.2212826013565063,
      "learning_rate": 9.653792154446648e-06,
      "loss": 0.8141,
      "step": 5053
    },
    {
      "epoch": 0.145623235175474,
      "grad_norm": 1.1712063550949097,
      "learning_rate": 9.653621525322616e-06,
      "loss": 0.8077,
      "step": 5054
    },
    {
      "epoch": 0.14565204863712328,
      "grad_norm": 1.2086902856826782,
      "learning_rate": 9.653450855670193e-06,
      "loss": 0.8097,
      "step": 5055
    },
    {
      "epoch": 0.14568086209877254,
      "grad_norm": 1.1896848678588867,
      "learning_rate": 9.653280145490866e-06,
      "loss": 0.8107,
      "step": 5056
    },
    {
      "epoch": 0.14570967556042183,
      "grad_norm": 1.1594524383544922,
      "learning_rate": 9.65310939478612e-06,
      "loss": 0.8066,
      "step": 5057
    },
    {
      "epoch": 0.14573848902207112,
      "grad_norm": 1.1736469268798828,
      "learning_rate": 9.652938603557442e-06,
      "loss": 0.7864,
      "step": 5058
    },
    {
      "epoch": 0.14576730248372038,
      "grad_norm": 1.193372130393982,
      "learning_rate": 9.652767771806321e-06,
      "loss": 0.8149,
      "step": 5059
    },
    {
      "epoch": 0.14579611594536968,
      "grad_norm": 1.1750826835632324,
      "learning_rate": 9.652596899534242e-06,
      "loss": 0.7922,
      "step": 5060
    },
    {
      "epoch": 0.14582492940701897,
      "grad_norm": 1.2309819459915161,
      "learning_rate": 9.652425986742696e-06,
      "loss": 0.7992,
      "step": 5061
    },
    {
      "epoch": 0.14585374286866823,
      "grad_norm": 1.2121858596801758,
      "learning_rate": 9.65225503343317e-06,
      "loss": 0.8154,
      "step": 5062
    },
    {
      "epoch": 0.14588255633031752,
      "grad_norm": 1.2574087381362915,
      "learning_rate": 9.652084039607155e-06,
      "loss": 0.8061,
      "step": 5063
    },
    {
      "epoch": 0.1459113697919668,
      "grad_norm": 1.2335612773895264,
      "learning_rate": 9.651913005266136e-06,
      "loss": 0.8189,
      "step": 5064
    },
    {
      "epoch": 0.1459401832536161,
      "grad_norm": 1.2524701356887817,
      "learning_rate": 9.651741930411606e-06,
      "loss": 0.78,
      "step": 5065
    },
    {
      "epoch": 0.14596899671526536,
      "grad_norm": 1.2017669677734375,
      "learning_rate": 9.651570815045055e-06,
      "loss": 0.7866,
      "step": 5066
    },
    {
      "epoch": 0.14599781017691466,
      "grad_norm": 1.2198834419250488,
      "learning_rate": 9.651399659167971e-06,
      "loss": 0.7866,
      "step": 5067
    },
    {
      "epoch": 0.14602662363856395,
      "grad_norm": 1.25254225730896,
      "learning_rate": 9.651228462781845e-06,
      "loss": 0.8284,
      "step": 5068
    },
    {
      "epoch": 0.1460554371002132,
      "grad_norm": 1.2712095975875854,
      "learning_rate": 9.65105722588817e-06,
      "loss": 0.8175,
      "step": 5069
    },
    {
      "epoch": 0.1460842505618625,
      "grad_norm": 1.4624629020690918,
      "learning_rate": 9.650885948488433e-06,
      "loss": 0.8066,
      "step": 5070
    },
    {
      "epoch": 0.1461130640235118,
      "grad_norm": 1.2655723094940186,
      "learning_rate": 9.650714630584132e-06,
      "loss": 0.815,
      "step": 5071
    },
    {
      "epoch": 0.14614187748516105,
      "grad_norm": 1.2053861618041992,
      "learning_rate": 9.650543272176753e-06,
      "loss": 0.8007,
      "step": 5072
    },
    {
      "epoch": 0.14617069094681034,
      "grad_norm": 1.2344131469726562,
      "learning_rate": 9.650371873267793e-06,
      "loss": 0.8083,
      "step": 5073
    },
    {
      "epoch": 0.14619950440845964,
      "grad_norm": 1.2359744310379028,
      "learning_rate": 9.650200433858742e-06,
      "loss": 0.8088,
      "step": 5074
    },
    {
      "epoch": 0.14622831787010893,
      "grad_norm": 1.2809723615646362,
      "learning_rate": 9.650028953951095e-06,
      "loss": 0.8251,
      "step": 5075
    },
    {
      "epoch": 0.1462571313317582,
      "grad_norm": 1.2246167659759521,
      "learning_rate": 9.649857433546341e-06,
      "loss": 0.8075,
      "step": 5076
    },
    {
      "epoch": 0.14628594479340748,
      "grad_norm": 1.2829536199569702,
      "learning_rate": 9.64968587264598e-06,
      "loss": 0.778,
      "step": 5077
    },
    {
      "epoch": 0.14631475825505677,
      "grad_norm": 1.2498271465301514,
      "learning_rate": 9.6495142712515e-06,
      "loss": 0.7745,
      "step": 5078
    },
    {
      "epoch": 0.14634357171670603,
      "grad_norm": 1.1733763217926025,
      "learning_rate": 9.6493426293644e-06,
      "loss": 0.7869,
      "step": 5079
    },
    {
      "epoch": 0.14637238517835532,
      "grad_norm": 1.2551610469818115,
      "learning_rate": 9.649170946986174e-06,
      "loss": 0.8176,
      "step": 5080
    },
    {
      "epoch": 0.14640119864000462,
      "grad_norm": 1.2423537969589233,
      "learning_rate": 9.648999224118317e-06,
      "loss": 0.8331,
      "step": 5081
    },
    {
      "epoch": 0.1464300121016539,
      "grad_norm": 1.1991163492202759,
      "learning_rate": 9.648827460762322e-06,
      "loss": 0.8168,
      "step": 5082
    },
    {
      "epoch": 0.14645882556330317,
      "grad_norm": 1.285208821296692,
      "learning_rate": 9.648655656919687e-06,
      "loss": 0.818,
      "step": 5083
    },
    {
      "epoch": 0.14648763902495246,
      "grad_norm": 1.21217679977417,
      "learning_rate": 9.64848381259191e-06,
      "loss": 0.7883,
      "step": 5084
    },
    {
      "epoch": 0.14651645248660175,
      "grad_norm": 1.2667303085327148,
      "learning_rate": 9.648311927780485e-06,
      "loss": 0.823,
      "step": 5085
    },
    {
      "epoch": 0.146545265948251,
      "grad_norm": 1.269955039024353,
      "learning_rate": 9.64814000248691e-06,
      "loss": 0.8023,
      "step": 5086
    },
    {
      "epoch": 0.1465740794099003,
      "grad_norm": 1.2000532150268555,
      "learning_rate": 9.64796803671268e-06,
      "loss": 0.7879,
      "step": 5087
    },
    {
      "epoch": 0.1466028928715496,
      "grad_norm": 1.2457791566848755,
      "learning_rate": 9.647796030459298e-06,
      "loss": 0.8079,
      "step": 5088
    },
    {
      "epoch": 0.14663170633319886,
      "grad_norm": 1.3060063123703003,
      "learning_rate": 9.647623983728256e-06,
      "loss": 0.8048,
      "step": 5089
    },
    {
      "epoch": 0.14666051979484815,
      "grad_norm": 1.2622045278549194,
      "learning_rate": 9.647451896521057e-06,
      "loss": 0.789,
      "step": 5090
    },
    {
      "epoch": 0.14668933325649744,
      "grad_norm": 1.3019763231277466,
      "learning_rate": 9.647279768839196e-06,
      "loss": 0.7866,
      "step": 5091
    },
    {
      "epoch": 0.14671814671814673,
      "grad_norm": 1.2569646835327148,
      "learning_rate": 9.647107600684174e-06,
      "loss": 0.8329,
      "step": 5092
    },
    {
      "epoch": 0.146746960179796,
      "grad_norm": 1.2566778659820557,
      "learning_rate": 9.646935392057492e-06,
      "loss": 0.8285,
      "step": 5093
    },
    {
      "epoch": 0.14677577364144528,
      "grad_norm": 1.3032020330429077,
      "learning_rate": 9.646763142960646e-06,
      "loss": 0.8181,
      "step": 5094
    },
    {
      "epoch": 0.14680458710309457,
      "grad_norm": 1.1875842809677124,
      "learning_rate": 9.64659085339514e-06,
      "loss": 0.8039,
      "step": 5095
    },
    {
      "epoch": 0.14683340056474384,
      "grad_norm": 1.2346309423446655,
      "learning_rate": 9.64641852336247e-06,
      "loss": 0.7632,
      "step": 5096
    },
    {
      "epoch": 0.14686221402639313,
      "grad_norm": 1.268534779548645,
      "learning_rate": 9.646246152864141e-06,
      "loss": 0.8153,
      "step": 5097
    },
    {
      "epoch": 0.14689102748804242,
      "grad_norm": 1.2111485004425049,
      "learning_rate": 9.646073741901652e-06,
      "loss": 0.8023,
      "step": 5098
    },
    {
      "epoch": 0.14691984094969168,
      "grad_norm": 1.2200446128845215,
      "learning_rate": 9.645901290476507e-06,
      "loss": 0.8097,
      "step": 5099
    },
    {
      "epoch": 0.14694865441134097,
      "grad_norm": 1.2625411748886108,
      "learning_rate": 9.645728798590204e-06,
      "loss": 0.7866,
      "step": 5100
    },
    {
      "epoch": 0.14697746787299026,
      "grad_norm": 1.2533814907073975,
      "learning_rate": 9.645556266244247e-06,
      "loss": 0.7928,
      "step": 5101
    },
    {
      "epoch": 0.14700628133463955,
      "grad_norm": 1.253906011581421,
      "learning_rate": 9.645383693440139e-06,
      "loss": 0.7824,
      "step": 5102
    },
    {
      "epoch": 0.14703509479628882,
      "grad_norm": 1.409687876701355,
      "learning_rate": 9.645211080179384e-06,
      "loss": 0.8212,
      "step": 5103
    },
    {
      "epoch": 0.1470639082579381,
      "grad_norm": 1.2831041812896729,
      "learning_rate": 9.645038426463481e-06,
      "loss": 0.7996,
      "step": 5104
    },
    {
      "epoch": 0.1470927217195874,
      "grad_norm": 1.2056739330291748,
      "learning_rate": 9.64486573229394e-06,
      "loss": 0.7895,
      "step": 5105
    },
    {
      "epoch": 0.14712153518123666,
      "grad_norm": 1.27114999294281,
      "learning_rate": 9.64469299767226e-06,
      "loss": 0.7953,
      "step": 5106
    },
    {
      "epoch": 0.14715034864288595,
      "grad_norm": 1.2024846076965332,
      "learning_rate": 9.644520222599947e-06,
      "loss": 0.7962,
      "step": 5107
    },
    {
      "epoch": 0.14717916210453524,
      "grad_norm": 1.2180715799331665,
      "learning_rate": 9.644347407078507e-06,
      "loss": 0.804,
      "step": 5108
    },
    {
      "epoch": 0.14720797556618453,
      "grad_norm": 1.2423983812332153,
      "learning_rate": 9.644174551109441e-06,
      "loss": 0.8205,
      "step": 5109
    },
    {
      "epoch": 0.1472367890278338,
      "grad_norm": 1.2419688701629639,
      "learning_rate": 9.644001654694261e-06,
      "loss": 0.8013,
      "step": 5110
    },
    {
      "epoch": 0.1472656024894831,
      "grad_norm": 1.202638030052185,
      "learning_rate": 9.643828717834465e-06,
      "loss": 0.8117,
      "step": 5111
    },
    {
      "epoch": 0.14729441595113238,
      "grad_norm": 1.1827267408370972,
      "learning_rate": 9.643655740531565e-06,
      "loss": 0.7914,
      "step": 5112
    },
    {
      "epoch": 0.14732322941278164,
      "grad_norm": 1.2435870170593262,
      "learning_rate": 9.643482722787065e-06,
      "loss": 0.7967,
      "step": 5113
    },
    {
      "epoch": 0.14735204287443093,
      "grad_norm": 1.197752594947815,
      "learning_rate": 9.643309664602473e-06,
      "loss": 0.8012,
      "step": 5114
    },
    {
      "epoch": 0.14738085633608022,
      "grad_norm": 1.1860953569412231,
      "learning_rate": 9.643136565979295e-06,
      "loss": 0.7962,
      "step": 5115
    },
    {
      "epoch": 0.1474096697977295,
      "grad_norm": 1.2782588005065918,
      "learning_rate": 9.642963426919038e-06,
      "loss": 0.8093,
      "step": 5116
    },
    {
      "epoch": 0.14743848325937878,
      "grad_norm": 1.1954394578933716,
      "learning_rate": 9.642790247423213e-06,
      "loss": 0.7677,
      "step": 5117
    },
    {
      "epoch": 0.14746729672102807,
      "grad_norm": 1.245362639427185,
      "learning_rate": 9.642617027493324e-06,
      "loss": 0.8271,
      "step": 5118
    },
    {
      "epoch": 0.14749611018267736,
      "grad_norm": 1.2286792993545532,
      "learning_rate": 9.642443767130883e-06,
      "loss": 0.8136,
      "step": 5119
    },
    {
      "epoch": 0.14752492364432662,
      "grad_norm": 1.2513108253479004,
      "learning_rate": 9.642270466337396e-06,
      "loss": 0.7904,
      "step": 5120
    },
    {
      "epoch": 0.1475537371059759,
      "grad_norm": 1.2536388635635376,
      "learning_rate": 9.642097125114377e-06,
      "loss": 0.8126,
      "step": 5121
    },
    {
      "epoch": 0.1475825505676252,
      "grad_norm": 1.2567318677902222,
      "learning_rate": 9.641923743463328e-06,
      "loss": 0.8117,
      "step": 5122
    },
    {
      "epoch": 0.14761136402927447,
      "grad_norm": 1.1832778453826904,
      "learning_rate": 9.641750321385765e-06,
      "loss": 0.8123,
      "step": 5123
    },
    {
      "epoch": 0.14764017749092376,
      "grad_norm": 1.2226839065551758,
      "learning_rate": 9.641576858883198e-06,
      "loss": 0.8041,
      "step": 5124
    },
    {
      "epoch": 0.14766899095257305,
      "grad_norm": 1.2842339277267456,
      "learning_rate": 9.641403355957136e-06,
      "loss": 0.8192,
      "step": 5125
    },
    {
      "epoch": 0.1476978044142223,
      "grad_norm": 1.2162374258041382,
      "learning_rate": 9.641229812609089e-06,
      "loss": 0.8079,
      "step": 5126
    },
    {
      "epoch": 0.1477266178758716,
      "grad_norm": 1.2825840711593628,
      "learning_rate": 9.641056228840573e-06,
      "loss": 0.7664,
      "step": 5127
    },
    {
      "epoch": 0.1477554313375209,
      "grad_norm": 1.2490571737289429,
      "learning_rate": 9.640882604653094e-06,
      "loss": 0.8158,
      "step": 5128
    },
    {
      "epoch": 0.14778424479917018,
      "grad_norm": 1.2700421810150146,
      "learning_rate": 9.640708940048167e-06,
      "loss": 0.7976,
      "step": 5129
    },
    {
      "epoch": 0.14781305826081945,
      "grad_norm": 1.2956489324569702,
      "learning_rate": 9.640535235027304e-06,
      "loss": 0.8247,
      "step": 5130
    },
    {
      "epoch": 0.14784187172246874,
      "grad_norm": 1.1543707847595215,
      "learning_rate": 9.64036148959202e-06,
      "loss": 0.8109,
      "step": 5131
    },
    {
      "epoch": 0.14787068518411803,
      "grad_norm": 1.1955666542053223,
      "learning_rate": 9.640187703743823e-06,
      "loss": 0.7796,
      "step": 5132
    },
    {
      "epoch": 0.1478994986457673,
      "grad_norm": 1.248342514038086,
      "learning_rate": 9.64001387748423e-06,
      "loss": 0.7982,
      "step": 5133
    },
    {
      "epoch": 0.14792831210741658,
      "grad_norm": 1.21054208278656,
      "learning_rate": 9.639840010814759e-06,
      "loss": 0.8071,
      "step": 5134
    },
    {
      "epoch": 0.14795712556906587,
      "grad_norm": 1.2622313499450684,
      "learning_rate": 9.639666103736915e-06,
      "loss": 0.7613,
      "step": 5135
    },
    {
      "epoch": 0.14798593903071516,
      "grad_norm": 1.227125644683838,
      "learning_rate": 9.639492156252218e-06,
      "loss": 0.7826,
      "step": 5136
    },
    {
      "epoch": 0.14801475249236443,
      "grad_norm": 1.2433078289031982,
      "learning_rate": 9.639318168362183e-06,
      "loss": 0.8105,
      "step": 5137
    },
    {
      "epoch": 0.14804356595401372,
      "grad_norm": 1.1899802684783936,
      "learning_rate": 9.639144140068324e-06,
      "loss": 0.8119,
      "step": 5138
    },
    {
      "epoch": 0.148072379415663,
      "grad_norm": 1.1808754205703735,
      "learning_rate": 9.638970071372155e-06,
      "loss": 0.8184,
      "step": 5139
    },
    {
      "epoch": 0.14810119287731227,
      "grad_norm": 1.2137497663497925,
      "learning_rate": 9.638795962275196e-06,
      "loss": 0.8083,
      "step": 5140
    },
    {
      "epoch": 0.14813000633896156,
      "grad_norm": 1.180148720741272,
      "learning_rate": 9.638621812778962e-06,
      "loss": 0.8281,
      "step": 5141
    },
    {
      "epoch": 0.14815881980061085,
      "grad_norm": 1.1958295106887817,
      "learning_rate": 9.638447622884967e-06,
      "loss": 0.8214,
      "step": 5142
    },
    {
      "epoch": 0.14818763326226012,
      "grad_norm": 1.2160083055496216,
      "learning_rate": 9.638273392594733e-06,
      "loss": 0.8302,
      "step": 5143
    },
    {
      "epoch": 0.1482164467239094,
      "grad_norm": 1.2024180889129639,
      "learning_rate": 9.638099121909772e-06,
      "loss": 0.8018,
      "step": 5144
    },
    {
      "epoch": 0.1482452601855587,
      "grad_norm": 1.227613925933838,
      "learning_rate": 9.637924810831603e-06,
      "loss": 0.8249,
      "step": 5145
    },
    {
      "epoch": 0.148274073647208,
      "grad_norm": 1.1191473007202148,
      "learning_rate": 9.637750459361747e-06,
      "loss": 0.781,
      "step": 5146
    },
    {
      "epoch": 0.14830288710885725,
      "grad_norm": 1.2120262384414673,
      "learning_rate": 9.637576067501721e-06,
      "loss": 0.7936,
      "step": 5147
    },
    {
      "epoch": 0.14833170057050654,
      "grad_norm": 1.2152043581008911,
      "learning_rate": 9.637401635253041e-06,
      "loss": 0.7917,
      "step": 5148
    },
    {
      "epoch": 0.14836051403215583,
      "grad_norm": 1.2067327499389648,
      "learning_rate": 9.637227162617231e-06,
      "loss": 0.8227,
      "step": 5149
    },
    {
      "epoch": 0.1483893274938051,
      "grad_norm": 1.2535269260406494,
      "learning_rate": 9.637052649595806e-06,
      "loss": 0.8119,
      "step": 5150
    },
    {
      "epoch": 0.1484181409554544,
      "grad_norm": 1.178788661956787,
      "learning_rate": 9.63687809619029e-06,
      "loss": 0.8194,
      "step": 5151
    },
    {
      "epoch": 0.14844695441710368,
      "grad_norm": 1.2582027912139893,
      "learning_rate": 9.636703502402199e-06,
      "loss": 0.8097,
      "step": 5152
    },
    {
      "epoch": 0.14847576787875294,
      "grad_norm": 1.293077826499939,
      "learning_rate": 9.636528868233055e-06,
      "loss": 0.8015,
      "step": 5153
    },
    {
      "epoch": 0.14850458134040223,
      "grad_norm": 1.1603758335113525,
      "learning_rate": 9.63635419368438e-06,
      "loss": 0.7963,
      "step": 5154
    },
    {
      "epoch": 0.14853339480205152,
      "grad_norm": 1.2292081117630005,
      "learning_rate": 9.636179478757696e-06,
      "loss": 0.7918,
      "step": 5155
    },
    {
      "epoch": 0.1485622082637008,
      "grad_norm": 1.2306618690490723,
      "learning_rate": 9.636004723454521e-06,
      "loss": 0.8239,
      "step": 5156
    },
    {
      "epoch": 0.14859102172535008,
      "grad_norm": 1.220858097076416,
      "learning_rate": 9.635829927776381e-06,
      "loss": 0.8032,
      "step": 5157
    },
    {
      "epoch": 0.14861983518699937,
      "grad_norm": 1.2389037609100342,
      "learning_rate": 9.635655091724797e-06,
      "loss": 0.8035,
      "step": 5158
    },
    {
      "epoch": 0.14864864864864866,
      "grad_norm": 1.232285499572754,
      "learning_rate": 9.635480215301289e-06,
      "loss": 0.7855,
      "step": 5159
    },
    {
      "epoch": 0.14867746211029792,
      "grad_norm": 1.2203867435455322,
      "learning_rate": 9.635305298507383e-06,
      "loss": 0.798,
      "step": 5160
    },
    {
      "epoch": 0.1487062755719472,
      "grad_norm": 1.220732569694519,
      "learning_rate": 9.635130341344604e-06,
      "loss": 0.8174,
      "step": 5161
    },
    {
      "epoch": 0.1487350890335965,
      "grad_norm": 1.263717532157898,
      "learning_rate": 9.63495534381447e-06,
      "loss": 0.8267,
      "step": 5162
    },
    {
      "epoch": 0.1487639024952458,
      "grad_norm": 1.2575160264968872,
      "learning_rate": 9.634780305918507e-06,
      "loss": 0.8094,
      "step": 5163
    },
    {
      "epoch": 0.14879271595689506,
      "grad_norm": 1.1665260791778564,
      "learning_rate": 9.634605227658244e-06,
      "loss": 0.7962,
      "step": 5164
    },
    {
      "epoch": 0.14882152941854435,
      "grad_norm": 1.232416033744812,
      "learning_rate": 9.6344301090352e-06,
      "loss": 0.7907,
      "step": 5165
    },
    {
      "epoch": 0.14885034288019364,
      "grad_norm": 1.2254104614257812,
      "learning_rate": 9.634254950050905e-06,
      "loss": 0.8192,
      "step": 5166
    },
    {
      "epoch": 0.1488791563418429,
      "grad_norm": 1.2327734231948853,
      "learning_rate": 9.634079750706879e-06,
      "loss": 0.8188,
      "step": 5167
    },
    {
      "epoch": 0.1489079698034922,
      "grad_norm": 1.2488985061645508,
      "learning_rate": 9.633904511004652e-06,
      "loss": 0.7951,
      "step": 5168
    },
    {
      "epoch": 0.14893678326514148,
      "grad_norm": 1.1574499607086182,
      "learning_rate": 9.63372923094575e-06,
      "loss": 0.7789,
      "step": 5169
    },
    {
      "epoch": 0.14896559672679074,
      "grad_norm": 1.2055634260177612,
      "learning_rate": 9.633553910531697e-06,
      "loss": 0.8267,
      "step": 5170
    },
    {
      "epoch": 0.14899441018844004,
      "grad_norm": 1.198471188545227,
      "learning_rate": 9.633378549764022e-06,
      "loss": 0.8114,
      "step": 5171
    },
    {
      "epoch": 0.14902322365008933,
      "grad_norm": 1.1951044797897339,
      "learning_rate": 9.633203148644251e-06,
      "loss": 0.8106,
      "step": 5172
    },
    {
      "epoch": 0.14905203711173862,
      "grad_norm": 1.191399097442627,
      "learning_rate": 9.633027707173913e-06,
      "loss": 0.8107,
      "step": 5173
    },
    {
      "epoch": 0.14908085057338788,
      "grad_norm": 1.202618956565857,
      "learning_rate": 9.632852225354534e-06,
      "loss": 0.8133,
      "step": 5174
    },
    {
      "epoch": 0.14910966403503717,
      "grad_norm": 1.1844849586486816,
      "learning_rate": 9.632676703187643e-06,
      "loss": 0.7878,
      "step": 5175
    },
    {
      "epoch": 0.14913847749668646,
      "grad_norm": 1.1853524446487427,
      "learning_rate": 9.63250114067477e-06,
      "loss": 0.7758,
      "step": 5176
    },
    {
      "epoch": 0.14916729095833572,
      "grad_norm": 1.214170217514038,
      "learning_rate": 9.632325537817442e-06,
      "loss": 0.8203,
      "step": 5177
    },
    {
      "epoch": 0.14919610441998502,
      "grad_norm": 1.1596087217330933,
      "learning_rate": 9.632149894617191e-06,
      "loss": 0.7967,
      "step": 5178
    },
    {
      "epoch": 0.1492249178816343,
      "grad_norm": 1.2643178701400757,
      "learning_rate": 9.631974211075543e-06,
      "loss": 0.7874,
      "step": 5179
    },
    {
      "epoch": 0.14925373134328357,
      "grad_norm": 1.2324482202529907,
      "learning_rate": 9.631798487194031e-06,
      "loss": 0.8271,
      "step": 5180
    },
    {
      "epoch": 0.14928254480493286,
      "grad_norm": 1.2058160305023193,
      "learning_rate": 9.631622722974184e-06,
      "loss": 0.8077,
      "step": 5181
    },
    {
      "epoch": 0.14931135826658215,
      "grad_norm": 1.2292842864990234,
      "learning_rate": 9.631446918417531e-06,
      "loss": 0.788,
      "step": 5182
    },
    {
      "epoch": 0.14934017172823144,
      "grad_norm": 1.274571418762207,
      "learning_rate": 9.631271073525607e-06,
      "loss": 0.8159,
      "step": 5183
    },
    {
      "epoch": 0.1493689851898807,
      "grad_norm": 1.1971813440322876,
      "learning_rate": 9.631095188299942e-06,
      "loss": 0.7885,
      "step": 5184
    },
    {
      "epoch": 0.14939779865153,
      "grad_norm": 1.2278190851211548,
      "learning_rate": 9.630919262742066e-06,
      "loss": 0.8024,
      "step": 5185
    },
    {
      "epoch": 0.14942661211317929,
      "grad_norm": 1.1934974193572998,
      "learning_rate": 9.630743296853513e-06,
      "loss": 0.8021,
      "step": 5186
    },
    {
      "epoch": 0.14945542557482855,
      "grad_norm": 1.2433637380599976,
      "learning_rate": 9.630567290635815e-06,
      "loss": 0.8163,
      "step": 5187
    },
    {
      "epoch": 0.14948423903647784,
      "grad_norm": 1.138289451599121,
      "learning_rate": 9.630391244090505e-06,
      "loss": 0.811,
      "step": 5188
    },
    {
      "epoch": 0.14951305249812713,
      "grad_norm": 1.2182167768478394,
      "learning_rate": 9.630215157219116e-06,
      "loss": 0.8083,
      "step": 5189
    },
    {
      "epoch": 0.14954186595977642,
      "grad_norm": 1.2334411144256592,
      "learning_rate": 9.630039030023182e-06,
      "loss": 0.8189,
      "step": 5190
    },
    {
      "epoch": 0.14957067942142568,
      "grad_norm": 1.2264983654022217,
      "learning_rate": 9.629862862504235e-06,
      "loss": 0.8081,
      "step": 5191
    },
    {
      "epoch": 0.14959949288307497,
      "grad_norm": 1.1415444612503052,
      "learning_rate": 9.62968665466381e-06,
      "loss": 0.7717,
      "step": 5192
    },
    {
      "epoch": 0.14962830634472427,
      "grad_norm": 1.1639511585235596,
      "learning_rate": 9.629510406503445e-06,
      "loss": 0.82,
      "step": 5193
    },
    {
      "epoch": 0.14965711980637353,
      "grad_norm": 1.114988088607788,
      "learning_rate": 9.629334118024669e-06,
      "loss": 0.7796,
      "step": 5194
    },
    {
      "epoch": 0.14968593326802282,
      "grad_norm": 1.139281153678894,
      "learning_rate": 9.629157789229023e-06,
      "loss": 0.784,
      "step": 5195
    },
    {
      "epoch": 0.1497147467296721,
      "grad_norm": 1.1565885543823242,
      "learning_rate": 9.62898142011804e-06,
      "loss": 0.7778,
      "step": 5196
    },
    {
      "epoch": 0.14974356019132137,
      "grad_norm": 1.2096219062805176,
      "learning_rate": 9.628805010693254e-06,
      "loss": 0.7956,
      "step": 5197
    },
    {
      "epoch": 0.14977237365297066,
      "grad_norm": 1.157546877861023,
      "learning_rate": 9.628628560956205e-06,
      "loss": 0.7903,
      "step": 5198
    },
    {
      "epoch": 0.14980118711461995,
      "grad_norm": 1.202745795249939,
      "learning_rate": 9.628452070908428e-06,
      "loss": 0.8038,
      "step": 5199
    },
    {
      "epoch": 0.14983000057626925,
      "grad_norm": 1.1645022630691528,
      "learning_rate": 9.628275540551462e-06,
      "loss": 0.792,
      "step": 5200
    },
    {
      "epoch": 0.1498588140379185,
      "grad_norm": 1.1138570308685303,
      "learning_rate": 9.62809896988684e-06,
      "loss": 0.7769,
      "step": 5201
    },
    {
      "epoch": 0.1498876274995678,
      "grad_norm": 1.1839009523391724,
      "learning_rate": 9.627922358916103e-06,
      "loss": 0.7996,
      "step": 5202
    },
    {
      "epoch": 0.1499164409612171,
      "grad_norm": 1.131245732307434,
      "learning_rate": 9.627745707640788e-06,
      "loss": 0.7832,
      "step": 5203
    },
    {
      "epoch": 0.14994525442286635,
      "grad_norm": 1.1544675827026367,
      "learning_rate": 9.627569016062434e-06,
      "loss": 0.8095,
      "step": 5204
    },
    {
      "epoch": 0.14997406788451564,
      "grad_norm": 1.1997594833374023,
      "learning_rate": 9.627392284182581e-06,
      "loss": 0.7991,
      "step": 5205
    },
    {
      "epoch": 0.15000288134616493,
      "grad_norm": 1.1628601551055908,
      "learning_rate": 9.627215512002768e-06,
      "loss": 0.7702,
      "step": 5206
    },
    {
      "epoch": 0.1500316948078142,
      "grad_norm": 1.2116543054580688,
      "learning_rate": 9.627038699524531e-06,
      "loss": 0.8395,
      "step": 5207
    },
    {
      "epoch": 0.1500605082694635,
      "grad_norm": 1.1937323808670044,
      "learning_rate": 9.626861846749414e-06,
      "loss": 0.8129,
      "step": 5208
    },
    {
      "epoch": 0.15008932173111278,
      "grad_norm": 1.1735563278198242,
      "learning_rate": 9.626684953678952e-06,
      "loss": 0.7998,
      "step": 5209
    },
    {
      "epoch": 0.15011813519276207,
      "grad_norm": 1.1762615442276,
      "learning_rate": 9.626508020314693e-06,
      "loss": 0.8092,
      "step": 5210
    },
    {
      "epoch": 0.15014694865441133,
      "grad_norm": 1.193132758140564,
      "learning_rate": 9.626331046658172e-06,
      "loss": 0.8174,
      "step": 5211
    },
    {
      "epoch": 0.15017576211606062,
      "grad_norm": 1.1907908916473389,
      "learning_rate": 9.626154032710932e-06,
      "loss": 0.7921,
      "step": 5212
    },
    {
      "epoch": 0.15020457557770991,
      "grad_norm": 1.1682995557785034,
      "learning_rate": 9.625976978474515e-06,
      "loss": 0.791,
      "step": 5213
    },
    {
      "epoch": 0.15023338903935918,
      "grad_norm": 1.1880022287368774,
      "learning_rate": 9.625799883950464e-06,
      "loss": 0.7867,
      "step": 5214
    },
    {
      "epoch": 0.15026220250100847,
      "grad_norm": 1.2094608545303345,
      "learning_rate": 9.62562274914032e-06,
      "loss": 0.8178,
      "step": 5215
    },
    {
      "epoch": 0.15029101596265776,
      "grad_norm": 1.210436463356018,
      "learning_rate": 9.625445574045625e-06,
      "loss": 0.8083,
      "step": 5216
    },
    {
      "epoch": 0.15031982942430705,
      "grad_norm": 1.2371340990066528,
      "learning_rate": 9.625268358667923e-06,
      "loss": 0.7793,
      "step": 5217
    },
    {
      "epoch": 0.1503486428859563,
      "grad_norm": 1.158532977104187,
      "learning_rate": 9.625091103008758e-06,
      "loss": 0.8002,
      "step": 5218
    },
    {
      "epoch": 0.1503774563476056,
      "grad_norm": 1.2237268686294556,
      "learning_rate": 9.62491380706967e-06,
      "loss": 0.8112,
      "step": 5219
    },
    {
      "epoch": 0.1504062698092549,
      "grad_norm": 1.2101329565048218,
      "learning_rate": 9.62473647085221e-06,
      "loss": 0.7939,
      "step": 5220
    },
    {
      "epoch": 0.15043508327090416,
      "grad_norm": 1.1712796688079834,
      "learning_rate": 9.624559094357917e-06,
      "loss": 0.7869,
      "step": 5221
    },
    {
      "epoch": 0.15046389673255345,
      "grad_norm": 1.1891151666641235,
      "learning_rate": 9.624381677588336e-06,
      "loss": 0.7901,
      "step": 5222
    },
    {
      "epoch": 0.15049271019420274,
      "grad_norm": 1.171321153640747,
      "learning_rate": 9.624204220545015e-06,
      "loss": 0.8067,
      "step": 5223
    },
    {
      "epoch": 0.150521523655852,
      "grad_norm": 1.1997153759002686,
      "learning_rate": 9.624026723229496e-06,
      "loss": 0.7977,
      "step": 5224
    },
    {
      "epoch": 0.1505503371175013,
      "grad_norm": 1.1850743293762207,
      "learning_rate": 9.623849185643327e-06,
      "loss": 0.8068,
      "step": 5225
    },
    {
      "epoch": 0.15057915057915058,
      "grad_norm": 1.161252498626709,
      "learning_rate": 9.623671607788054e-06,
      "loss": 0.8038,
      "step": 5226
    },
    {
      "epoch": 0.15060796404079987,
      "grad_norm": 1.1752724647521973,
      "learning_rate": 9.623493989665223e-06,
      "loss": 0.8184,
      "step": 5227
    },
    {
      "epoch": 0.15063677750244914,
      "grad_norm": 1.1808674335479736,
      "learning_rate": 9.623316331276382e-06,
      "loss": 0.7945,
      "step": 5228
    },
    {
      "epoch": 0.15066559096409843,
      "grad_norm": 1.1766526699066162,
      "learning_rate": 9.623138632623077e-06,
      "loss": 0.775,
      "step": 5229
    },
    {
      "epoch": 0.15069440442574772,
      "grad_norm": 1.1569260358810425,
      "learning_rate": 9.622960893706854e-06,
      "loss": 0.8019,
      "step": 5230
    },
    {
      "epoch": 0.15072321788739698,
      "grad_norm": 1.182760238647461,
      "learning_rate": 9.622783114529267e-06,
      "loss": 0.774,
      "step": 5231
    },
    {
      "epoch": 0.15075203134904627,
      "grad_norm": 1.2380702495574951,
      "learning_rate": 9.622605295091856e-06,
      "loss": 0.8235,
      "step": 5232
    },
    {
      "epoch": 0.15078084481069556,
      "grad_norm": 1.1574004888534546,
      "learning_rate": 9.622427435396175e-06,
      "loss": 0.7893,
      "step": 5233
    },
    {
      "epoch": 0.15080965827234483,
      "grad_norm": 1.1598396301269531,
      "learning_rate": 9.622249535443773e-06,
      "loss": 0.7954,
      "step": 5234
    },
    {
      "epoch": 0.15083847173399412,
      "grad_norm": 1.1875982284545898,
      "learning_rate": 9.622071595236198e-06,
      "loss": 0.7983,
      "step": 5235
    },
    {
      "epoch": 0.1508672851956434,
      "grad_norm": 1.1846346855163574,
      "learning_rate": 9.621893614775e-06,
      "loss": 0.8178,
      "step": 5236
    },
    {
      "epoch": 0.1508960986572927,
      "grad_norm": 1.2295526266098022,
      "learning_rate": 9.621715594061727e-06,
      "loss": 0.7984,
      "step": 5237
    },
    {
      "epoch": 0.15092491211894196,
      "grad_norm": 1.206970453262329,
      "learning_rate": 9.621537533097933e-06,
      "loss": 0.7735,
      "step": 5238
    },
    {
      "epoch": 0.15095372558059125,
      "grad_norm": 1.2048859596252441,
      "learning_rate": 9.621359431885166e-06,
      "loss": 0.8099,
      "step": 5239
    },
    {
      "epoch": 0.15098253904224054,
      "grad_norm": 1.2616393566131592,
      "learning_rate": 9.621181290424977e-06,
      "loss": 0.7899,
      "step": 5240
    },
    {
      "epoch": 0.1510113525038898,
      "grad_norm": 1.1782515048980713,
      "learning_rate": 9.62100310871892e-06,
      "loss": 0.7563,
      "step": 5241
    },
    {
      "epoch": 0.1510401659655391,
      "grad_norm": 1.2523294687271118,
      "learning_rate": 9.620824886768545e-06,
      "loss": 0.8,
      "step": 5242
    },
    {
      "epoch": 0.1510689794271884,
      "grad_norm": 1.2404613494873047,
      "learning_rate": 9.620646624575405e-06,
      "loss": 0.7891,
      "step": 5243
    },
    {
      "epoch": 0.15109779288883768,
      "grad_norm": 1.2163351774215698,
      "learning_rate": 9.62046832214105e-06,
      "loss": 0.7812,
      "step": 5244
    },
    {
      "epoch": 0.15112660635048694,
      "grad_norm": 1.2078322172164917,
      "learning_rate": 9.620289979467036e-06,
      "loss": 0.7902,
      "step": 5245
    },
    {
      "epoch": 0.15115541981213623,
      "grad_norm": 1.2209627628326416,
      "learning_rate": 9.620111596554915e-06,
      "loss": 0.7909,
      "step": 5246
    },
    {
      "epoch": 0.15118423327378552,
      "grad_norm": 1.2076797485351562,
      "learning_rate": 9.61993317340624e-06,
      "loss": 0.7913,
      "step": 5247
    },
    {
      "epoch": 0.1512130467354348,
      "grad_norm": 1.3850406408309937,
      "learning_rate": 9.619754710022566e-06,
      "loss": 0.8035,
      "step": 5248
    },
    {
      "epoch": 0.15124186019708408,
      "grad_norm": 1.2247568368911743,
      "learning_rate": 9.619576206405446e-06,
      "loss": 0.8083,
      "step": 5249
    },
    {
      "epoch": 0.15127067365873337,
      "grad_norm": 1.3456753492355347,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.8214,
      "step": 5250
    },
    {
      "epoch": 0.15129948712038263,
      "grad_norm": 1.2153713703155518,
      "learning_rate": 9.619219078477087e-06,
      "loss": 0.8155,
      "step": 5251
    },
    {
      "epoch": 0.15132830058203192,
      "grad_norm": 1.189820408821106,
      "learning_rate": 9.61904045416896e-06,
      "loss": 0.8078,
      "step": 5252
    },
    {
      "epoch": 0.1513571140436812,
      "grad_norm": 1.2386094331741333,
      "learning_rate": 9.618861789633607e-06,
      "loss": 0.804,
      "step": 5253
    },
    {
      "epoch": 0.1513859275053305,
      "grad_norm": 1.2013379335403442,
      "learning_rate": 9.618683084872586e-06,
      "loss": 0.7814,
      "step": 5254
    },
    {
      "epoch": 0.15141474096697977,
      "grad_norm": 1.265552043914795,
      "learning_rate": 9.618504339887451e-06,
      "loss": 0.8244,
      "step": 5255
    },
    {
      "epoch": 0.15144355442862906,
      "grad_norm": 1.2592934370040894,
      "learning_rate": 9.618325554679762e-06,
      "loss": 0.7952,
      "step": 5256
    },
    {
      "epoch": 0.15147236789027835,
      "grad_norm": 1.1928021907806396,
      "learning_rate": 9.618146729251072e-06,
      "loss": 0.7727,
      "step": 5257
    },
    {
      "epoch": 0.1515011813519276,
      "grad_norm": 1.2638227939605713,
      "learning_rate": 9.617967863602941e-06,
      "loss": 0.8242,
      "step": 5258
    },
    {
      "epoch": 0.1515299948135769,
      "grad_norm": 1.245074987411499,
      "learning_rate": 9.617788957736928e-06,
      "loss": 0.8093,
      "step": 5259
    },
    {
      "epoch": 0.1515588082752262,
      "grad_norm": 1.2587088346481323,
      "learning_rate": 9.617610011654587e-06,
      "loss": 0.8062,
      "step": 5260
    },
    {
      "epoch": 0.15158762173687546,
      "grad_norm": 1.1947312355041504,
      "learning_rate": 9.61743102535748e-06,
      "loss": 0.7974,
      "step": 5261
    },
    {
      "epoch": 0.15161643519852475,
      "grad_norm": 1.1967923641204834,
      "learning_rate": 9.617251998847163e-06,
      "loss": 0.7879,
      "step": 5262
    },
    {
      "epoch": 0.15164524866017404,
      "grad_norm": 1.1910799741744995,
      "learning_rate": 9.617072932125198e-06,
      "loss": 0.8219,
      "step": 5263
    },
    {
      "epoch": 0.15167406212182333,
      "grad_norm": 1.2168564796447754,
      "learning_rate": 9.616893825193143e-06,
      "loss": 0.8285,
      "step": 5264
    },
    {
      "epoch": 0.1517028755834726,
      "grad_norm": 1.184000015258789,
      "learning_rate": 9.616714678052558e-06,
      "loss": 0.7836,
      "step": 5265
    },
    {
      "epoch": 0.15173168904512188,
      "grad_norm": 1.1758458614349365,
      "learning_rate": 9.616535490705003e-06,
      "loss": 0.7973,
      "step": 5266
    },
    {
      "epoch": 0.15176050250677117,
      "grad_norm": 1.178387999534607,
      "learning_rate": 9.616356263152038e-06,
      "loss": 0.8092,
      "step": 5267
    },
    {
      "epoch": 0.15178931596842044,
      "grad_norm": 1.2492331266403198,
      "learning_rate": 9.616176995395226e-06,
      "loss": 0.8097,
      "step": 5268
    },
    {
      "epoch": 0.15181812943006973,
      "grad_norm": 1.1985076665878296,
      "learning_rate": 9.615997687436127e-06,
      "loss": 0.7858,
      "step": 5269
    },
    {
      "epoch": 0.15184694289171902,
      "grad_norm": 1.2297084331512451,
      "learning_rate": 9.6158183392763e-06,
      "loss": 0.8255,
      "step": 5270
    },
    {
      "epoch": 0.1518757563533683,
      "grad_norm": 1.2000352144241333,
      "learning_rate": 9.615638950917312e-06,
      "loss": 0.81,
      "step": 5271
    },
    {
      "epoch": 0.15190456981501757,
      "grad_norm": 1.107416033744812,
      "learning_rate": 9.615459522360722e-06,
      "loss": 0.791,
      "step": 5272
    },
    {
      "epoch": 0.15193338327666686,
      "grad_norm": 1.1743900775909424,
      "learning_rate": 9.615280053608093e-06,
      "loss": 0.7717,
      "step": 5273
    },
    {
      "epoch": 0.15196219673831615,
      "grad_norm": 1.1773980855941772,
      "learning_rate": 9.615100544660988e-06,
      "loss": 0.8131,
      "step": 5274
    },
    {
      "epoch": 0.15199101019996542,
      "grad_norm": 1.1923917531967163,
      "learning_rate": 9.61492099552097e-06,
      "loss": 0.7945,
      "step": 5275
    },
    {
      "epoch": 0.1520198236616147,
      "grad_norm": 1.1942538022994995,
      "learning_rate": 9.614741406189605e-06,
      "loss": 0.7954,
      "step": 5276
    },
    {
      "epoch": 0.152048637123264,
      "grad_norm": 1.1909065246582031,
      "learning_rate": 9.614561776668454e-06,
      "loss": 0.7938,
      "step": 5277
    },
    {
      "epoch": 0.15207745058491326,
      "grad_norm": 1.149532437324524,
      "learning_rate": 9.614382106959085e-06,
      "loss": 0.795,
      "step": 5278
    },
    {
      "epoch": 0.15210626404656255,
      "grad_norm": 1.241659164428711,
      "learning_rate": 9.61420239706306e-06,
      "loss": 0.8127,
      "step": 5279
    },
    {
      "epoch": 0.15213507750821184,
      "grad_norm": 1.1565239429473877,
      "learning_rate": 9.614022646981942e-06,
      "loss": 0.7864,
      "step": 5280
    },
    {
      "epoch": 0.15216389096986113,
      "grad_norm": 1.1744730472564697,
      "learning_rate": 9.613842856717301e-06,
      "loss": 0.8124,
      "step": 5281
    },
    {
      "epoch": 0.1521927044315104,
      "grad_norm": 1.1870694160461426,
      "learning_rate": 9.613663026270701e-06,
      "loss": 0.7953,
      "step": 5282
    },
    {
      "epoch": 0.15222151789315969,
      "grad_norm": 1.2596453428268433,
      "learning_rate": 9.613483155643707e-06,
      "loss": 0.8016,
      "step": 5283
    },
    {
      "epoch": 0.15225033135480898,
      "grad_norm": 1.17376708984375,
      "learning_rate": 9.613303244837889e-06,
      "loss": 0.8124,
      "step": 5284
    },
    {
      "epoch": 0.15227914481645824,
      "grad_norm": 1.145779013633728,
      "learning_rate": 9.613123293854809e-06,
      "loss": 0.7872,
      "step": 5285
    },
    {
      "epoch": 0.15230795827810753,
      "grad_norm": 1.1810256242752075,
      "learning_rate": 9.612943302696036e-06,
      "loss": 0.7859,
      "step": 5286
    },
    {
      "epoch": 0.15233677173975682,
      "grad_norm": 1.2001758813858032,
      "learning_rate": 9.61276327136314e-06,
      "loss": 0.8029,
      "step": 5287
    },
    {
      "epoch": 0.15236558520140608,
      "grad_norm": 1.1753976345062256,
      "learning_rate": 9.612583199857685e-06,
      "loss": 0.8207,
      "step": 5288
    },
    {
      "epoch": 0.15239439866305537,
      "grad_norm": 1.150674819946289,
      "learning_rate": 9.612403088181244e-06,
      "loss": 0.7911,
      "step": 5289
    },
    {
      "epoch": 0.15242321212470467,
      "grad_norm": 1.1385090351104736,
      "learning_rate": 9.61222293633538e-06,
      "loss": 0.8152,
      "step": 5290
    },
    {
      "epoch": 0.15245202558635396,
      "grad_norm": 1.2038915157318115,
      "learning_rate": 9.612042744321666e-06,
      "loss": 0.7979,
      "step": 5291
    },
    {
      "epoch": 0.15248083904800322,
      "grad_norm": 1.1538867950439453,
      "learning_rate": 9.611862512141671e-06,
      "loss": 0.79,
      "step": 5292
    },
    {
      "epoch": 0.1525096525096525,
      "grad_norm": 1.2482560873031616,
      "learning_rate": 9.611682239796963e-06,
      "loss": 0.7955,
      "step": 5293
    },
    {
      "epoch": 0.1525384659713018,
      "grad_norm": 1.1873975992202759,
      "learning_rate": 9.611501927289111e-06,
      "loss": 0.811,
      "step": 5294
    },
    {
      "epoch": 0.15256727943295106,
      "grad_norm": 1.1985372304916382,
      "learning_rate": 9.61132157461969e-06,
      "loss": 0.8024,
      "step": 5295
    },
    {
      "epoch": 0.15259609289460035,
      "grad_norm": 1.222222924232483,
      "learning_rate": 9.611141181790266e-06,
      "loss": 0.8115,
      "step": 5296
    },
    {
      "epoch": 0.15262490635624965,
      "grad_norm": 1.1727522611618042,
      "learning_rate": 9.610960748802412e-06,
      "loss": 0.78,
      "step": 5297
    },
    {
      "epoch": 0.1526537198178989,
      "grad_norm": 1.1881345510482788,
      "learning_rate": 9.610780275657699e-06,
      "loss": 0.7771,
      "step": 5298
    },
    {
      "epoch": 0.1526825332795482,
      "grad_norm": 1.2107317447662354,
      "learning_rate": 9.6105997623577e-06,
      "loss": 0.8015,
      "step": 5299
    },
    {
      "epoch": 0.1527113467411975,
      "grad_norm": 1.2279753684997559,
      "learning_rate": 9.610419208903984e-06,
      "loss": 0.8438,
      "step": 5300
    },
    {
      "epoch": 0.15274016020284678,
      "grad_norm": 1.1474562883377075,
      "learning_rate": 9.610238615298127e-06,
      "loss": 0.7904,
      "step": 5301
    },
    {
      "epoch": 0.15276897366449604,
      "grad_norm": 1.15253746509552,
      "learning_rate": 9.610057981541698e-06,
      "loss": 0.7937,
      "step": 5302
    },
    {
      "epoch": 0.15279778712614533,
      "grad_norm": 1.1813312768936157,
      "learning_rate": 9.609877307636274e-06,
      "loss": 0.8139,
      "step": 5303
    },
    {
      "epoch": 0.15282660058779463,
      "grad_norm": 1.1657241582870483,
      "learning_rate": 9.609696593583428e-06,
      "loss": 0.7972,
      "step": 5304
    },
    {
      "epoch": 0.1528554140494439,
      "grad_norm": 1.246373176574707,
      "learning_rate": 9.60951583938473e-06,
      "loss": 0.8114,
      "step": 5305
    },
    {
      "epoch": 0.15288422751109318,
      "grad_norm": 1.4988082647323608,
      "learning_rate": 9.609335045041758e-06,
      "loss": 0.8008,
      "step": 5306
    },
    {
      "epoch": 0.15291304097274247,
      "grad_norm": 1.146767258644104,
      "learning_rate": 9.609154210556087e-06,
      "loss": 0.7955,
      "step": 5307
    },
    {
      "epoch": 0.15294185443439176,
      "grad_norm": 1.2436622381210327,
      "learning_rate": 9.60897333592929e-06,
      "loss": 0.7938,
      "step": 5308
    },
    {
      "epoch": 0.15297066789604102,
      "grad_norm": 1.2228065729141235,
      "learning_rate": 9.60879242116294e-06,
      "loss": 0.7931,
      "step": 5309
    },
    {
      "epoch": 0.15299948135769031,
      "grad_norm": 1.1784031391143799,
      "learning_rate": 9.608611466258617e-06,
      "loss": 0.7806,
      "step": 5310
    },
    {
      "epoch": 0.1530282948193396,
      "grad_norm": 1.2112128734588623,
      "learning_rate": 9.608430471217896e-06,
      "loss": 0.8037,
      "step": 5311
    },
    {
      "epoch": 0.15305710828098887,
      "grad_norm": 1.2321648597717285,
      "learning_rate": 9.60824943604235e-06,
      "loss": 0.7913,
      "step": 5312
    },
    {
      "epoch": 0.15308592174263816,
      "grad_norm": 1.264883279800415,
      "learning_rate": 9.60806836073356e-06,
      "loss": 0.805,
      "step": 5313
    },
    {
      "epoch": 0.15311473520428745,
      "grad_norm": 1.2175281047821045,
      "learning_rate": 9.607887245293099e-06,
      "loss": 0.8013,
      "step": 5314
    },
    {
      "epoch": 0.1531435486659367,
      "grad_norm": 1.2370597124099731,
      "learning_rate": 9.607706089722549e-06,
      "loss": 0.7977,
      "step": 5315
    },
    {
      "epoch": 0.153172362127586,
      "grad_norm": 1.1337617635726929,
      "learning_rate": 9.607524894023482e-06,
      "loss": 0.7805,
      "step": 5316
    },
    {
      "epoch": 0.1532011755892353,
      "grad_norm": 1.1863431930541992,
      "learning_rate": 9.607343658197482e-06,
      "loss": 0.8075,
      "step": 5317
    },
    {
      "epoch": 0.15322998905088459,
      "grad_norm": 1.1784998178482056,
      "learning_rate": 9.607162382246121e-06,
      "loss": 0.7885,
      "step": 5318
    },
    {
      "epoch": 0.15325880251253385,
      "grad_norm": 1.1628105640411377,
      "learning_rate": 9.606981066170984e-06,
      "loss": 0.8246,
      "step": 5319
    },
    {
      "epoch": 0.15328761597418314,
      "grad_norm": 1.2108455896377563,
      "learning_rate": 9.606799709973645e-06,
      "loss": 0.8357,
      "step": 5320
    },
    {
      "epoch": 0.15331642943583243,
      "grad_norm": 1.1765917539596558,
      "learning_rate": 9.606618313655687e-06,
      "loss": 0.8076,
      "step": 5321
    },
    {
      "epoch": 0.1533452428974817,
      "grad_norm": 1.1262587308883667,
      "learning_rate": 9.606436877218688e-06,
      "loss": 0.7899,
      "step": 5322
    },
    {
      "epoch": 0.15337405635913098,
      "grad_norm": 1.2494876384735107,
      "learning_rate": 9.606255400664228e-06,
      "loss": 0.8115,
      "step": 5323
    },
    {
      "epoch": 0.15340286982078027,
      "grad_norm": 1.2002208232879639,
      "learning_rate": 9.60607388399389e-06,
      "loss": 0.8072,
      "step": 5324
    },
    {
      "epoch": 0.15343168328242954,
      "grad_norm": 1.1646926403045654,
      "learning_rate": 9.605892327209251e-06,
      "loss": 0.7901,
      "step": 5325
    },
    {
      "epoch": 0.15346049674407883,
      "grad_norm": 1.2152923345565796,
      "learning_rate": 9.605710730311894e-06,
      "loss": 0.7956,
      "step": 5326
    },
    {
      "epoch": 0.15348931020572812,
      "grad_norm": 1.2741550207138062,
      "learning_rate": 9.6055290933034e-06,
      "loss": 0.8388,
      "step": 5327
    },
    {
      "epoch": 0.1535181236673774,
      "grad_norm": 1.157900333404541,
      "learning_rate": 9.605347416185354e-06,
      "loss": 0.7592,
      "step": 5328
    },
    {
      "epoch": 0.15354693712902667,
      "grad_norm": 1.1565629243850708,
      "learning_rate": 9.605165698959335e-06,
      "loss": 0.7693,
      "step": 5329
    },
    {
      "epoch": 0.15357575059067596,
      "grad_norm": 1.259239673614502,
      "learning_rate": 9.604983941626924e-06,
      "loss": 0.8001,
      "step": 5330
    },
    {
      "epoch": 0.15360456405232525,
      "grad_norm": 1.1588432788848877,
      "learning_rate": 9.60480214418971e-06,
      "loss": 0.7703,
      "step": 5331
    },
    {
      "epoch": 0.15363337751397452,
      "grad_norm": 1.2251965999603271,
      "learning_rate": 9.604620306649269e-06,
      "loss": 0.7871,
      "step": 5332
    },
    {
      "epoch": 0.1536621909756238,
      "grad_norm": 1.2478054761886597,
      "learning_rate": 9.604438429007188e-06,
      "loss": 0.8109,
      "step": 5333
    },
    {
      "epoch": 0.1536910044372731,
      "grad_norm": 1.2025872468948364,
      "learning_rate": 9.604256511265054e-06,
      "loss": 0.8314,
      "step": 5334
    },
    {
      "epoch": 0.1537198178989224,
      "grad_norm": 1.179118275642395,
      "learning_rate": 9.604074553424445e-06,
      "loss": 0.7999,
      "step": 5335
    },
    {
      "epoch": 0.15374863136057165,
      "grad_norm": 1.2660322189331055,
      "learning_rate": 9.60389255548695e-06,
      "loss": 0.8111,
      "step": 5336
    },
    {
      "epoch": 0.15377744482222094,
      "grad_norm": 1.1331928968429565,
      "learning_rate": 9.603710517454155e-06,
      "loss": 0.8113,
      "step": 5337
    },
    {
      "epoch": 0.15380625828387023,
      "grad_norm": 1.1647422313690186,
      "learning_rate": 9.603528439327643e-06,
      "loss": 0.8001,
      "step": 5338
    },
    {
      "epoch": 0.1538350717455195,
      "grad_norm": 1.1817766427993774,
      "learning_rate": 9.603346321108999e-06,
      "loss": 0.8208,
      "step": 5339
    },
    {
      "epoch": 0.1538638852071688,
      "grad_norm": 1.229814052581787,
      "learning_rate": 9.60316416279981e-06,
      "loss": 0.8249,
      "step": 5340
    },
    {
      "epoch": 0.15389269866881808,
      "grad_norm": 1.1338549852371216,
      "learning_rate": 9.602981964401663e-06,
      "loss": 0.7727,
      "step": 5341
    },
    {
      "epoch": 0.15392151213046734,
      "grad_norm": 1.2053836584091187,
      "learning_rate": 9.602799725916144e-06,
      "loss": 0.799,
      "step": 5342
    },
    {
      "epoch": 0.15395032559211663,
      "grad_norm": 1.1900607347488403,
      "learning_rate": 9.60261744734484e-06,
      "loss": 0.8092,
      "step": 5343
    },
    {
      "epoch": 0.15397913905376592,
      "grad_norm": 1.1463592052459717,
      "learning_rate": 9.60243512868934e-06,
      "loss": 0.8071,
      "step": 5344
    },
    {
      "epoch": 0.15400795251541521,
      "grad_norm": 1.2358816862106323,
      "learning_rate": 9.602252769951231e-06,
      "loss": 0.799,
      "step": 5345
    },
    {
      "epoch": 0.15403676597706448,
      "grad_norm": 1.2253729104995728,
      "learning_rate": 9.602070371132103e-06,
      "loss": 0.7809,
      "step": 5346
    },
    {
      "epoch": 0.15406557943871377,
      "grad_norm": 1.1750997304916382,
      "learning_rate": 9.601887932233538e-06,
      "loss": 0.7887,
      "step": 5347
    },
    {
      "epoch": 0.15409439290036306,
      "grad_norm": 1.1950196027755737,
      "learning_rate": 9.60170545325713e-06,
      "loss": 0.7683,
      "step": 5348
    },
    {
      "epoch": 0.15412320636201232,
      "grad_norm": 1.202701210975647,
      "learning_rate": 9.60152293420447e-06,
      "loss": 0.7969,
      "step": 5349
    },
    {
      "epoch": 0.1541520198236616,
      "grad_norm": 1.1709113121032715,
      "learning_rate": 9.601340375077145e-06,
      "loss": 0.7698,
      "step": 5350
    },
    {
      "epoch": 0.1541808332853109,
      "grad_norm": 1.1438807249069214,
      "learning_rate": 9.601157775876745e-06,
      "loss": 0.7853,
      "step": 5351
    },
    {
      "epoch": 0.15420964674696017,
      "grad_norm": 1.1639007329940796,
      "learning_rate": 9.600975136604857e-06,
      "loss": 0.8166,
      "step": 5352
    },
    {
      "epoch": 0.15423846020860946,
      "grad_norm": 1.1351492404937744,
      "learning_rate": 9.600792457263077e-06,
      "loss": 0.8,
      "step": 5353
    },
    {
      "epoch": 0.15426727367025875,
      "grad_norm": 1.140006184577942,
      "learning_rate": 9.600609737852994e-06,
      "loss": 0.8146,
      "step": 5354
    },
    {
      "epoch": 0.15429608713190804,
      "grad_norm": 1.2055765390396118,
      "learning_rate": 9.6004269783762e-06,
      "loss": 0.7814,
      "step": 5355
    },
    {
      "epoch": 0.1543249005935573,
      "grad_norm": 1.1274842023849487,
      "learning_rate": 9.600244178834283e-06,
      "loss": 0.7778,
      "step": 5356
    },
    {
      "epoch": 0.1543537140552066,
      "grad_norm": 1.1661584377288818,
      "learning_rate": 9.600061339228841e-06,
      "loss": 0.7872,
      "step": 5357
    },
    {
      "epoch": 0.15438252751685588,
      "grad_norm": 1.156025767326355,
      "learning_rate": 9.599878459561462e-06,
      "loss": 0.7805,
      "step": 5358
    },
    {
      "epoch": 0.15441134097850515,
      "grad_norm": 1.1664408445358276,
      "learning_rate": 9.59969553983374e-06,
      "loss": 0.7872,
      "step": 5359
    },
    {
      "epoch": 0.15444015444015444,
      "grad_norm": 1.1217825412750244,
      "learning_rate": 9.599512580047266e-06,
      "loss": 0.7901,
      "step": 5360
    },
    {
      "epoch": 0.15446896790180373,
      "grad_norm": 1.1658692359924316,
      "learning_rate": 9.599329580203638e-06,
      "loss": 0.8058,
      "step": 5361
    },
    {
      "epoch": 0.15449778136345302,
      "grad_norm": 1.13737154006958,
      "learning_rate": 9.599146540304445e-06,
      "loss": 0.8134,
      "step": 5362
    },
    {
      "epoch": 0.15452659482510228,
      "grad_norm": 1.094435453414917,
      "learning_rate": 9.598963460351283e-06,
      "loss": 0.7749,
      "step": 5363
    },
    {
      "epoch": 0.15455540828675157,
      "grad_norm": 1.1385424137115479,
      "learning_rate": 9.598780340345748e-06,
      "loss": 0.8099,
      "step": 5364
    },
    {
      "epoch": 0.15458422174840086,
      "grad_norm": 1.2169628143310547,
      "learning_rate": 9.598597180289431e-06,
      "loss": 0.8128,
      "step": 5365
    },
    {
      "epoch": 0.15461303521005013,
      "grad_norm": 1.1295175552368164,
      "learning_rate": 9.598413980183931e-06,
      "loss": 0.7921,
      "step": 5366
    },
    {
      "epoch": 0.15464184867169942,
      "grad_norm": 1.1698226928710938,
      "learning_rate": 9.598230740030842e-06,
      "loss": 0.8053,
      "step": 5367
    },
    {
      "epoch": 0.1546706621333487,
      "grad_norm": 1.1916881799697876,
      "learning_rate": 9.598047459831762e-06,
      "loss": 0.7968,
      "step": 5368
    },
    {
      "epoch": 0.15469947559499797,
      "grad_norm": 1.1160000562667847,
      "learning_rate": 9.597864139588282e-06,
      "loss": 0.8028,
      "step": 5369
    },
    {
      "epoch": 0.15472828905664726,
      "grad_norm": 1.1763250827789307,
      "learning_rate": 9.597680779302e-06,
      "loss": 0.8153,
      "step": 5370
    },
    {
      "epoch": 0.15475710251829655,
      "grad_norm": 1.2276667356491089,
      "learning_rate": 9.597497378974519e-06,
      "loss": 0.7882,
      "step": 5371
    },
    {
      "epoch": 0.15478591597994584,
      "grad_norm": 1.1918998956680298,
      "learning_rate": 9.597313938607428e-06,
      "loss": 0.7924,
      "step": 5372
    },
    {
      "epoch": 0.1548147294415951,
      "grad_norm": 1.1551916599273682,
      "learning_rate": 9.597130458202331e-06,
      "loss": 0.7765,
      "step": 5373
    },
    {
      "epoch": 0.1548435429032444,
      "grad_norm": 1.249950885772705,
      "learning_rate": 9.59694693776082e-06,
      "loss": 0.7918,
      "step": 5374
    },
    {
      "epoch": 0.1548723563648937,
      "grad_norm": 1.1746196746826172,
      "learning_rate": 9.596763377284499e-06,
      "loss": 0.7825,
      "step": 5375
    },
    {
      "epoch": 0.15490116982654295,
      "grad_norm": 1.1472511291503906,
      "learning_rate": 9.596579776774964e-06,
      "loss": 0.7988,
      "step": 5376
    },
    {
      "epoch": 0.15492998328819224,
      "grad_norm": 1.2264827489852905,
      "learning_rate": 9.596396136233812e-06,
      "loss": 0.8046,
      "step": 5377
    },
    {
      "epoch": 0.15495879674984153,
      "grad_norm": 1.167279839515686,
      "learning_rate": 9.596212455662645e-06,
      "loss": 0.7938,
      "step": 5378
    },
    {
      "epoch": 0.1549876102114908,
      "grad_norm": 1.1847518682479858,
      "learning_rate": 9.596028735063065e-06,
      "loss": 0.7846,
      "step": 5379
    },
    {
      "epoch": 0.15501642367314009,
      "grad_norm": 1.1431198120117188,
      "learning_rate": 9.595844974436666e-06,
      "loss": 0.7961,
      "step": 5380
    },
    {
      "epoch": 0.15504523713478938,
      "grad_norm": 1.18716299533844,
      "learning_rate": 9.595661173785052e-06,
      "loss": 0.8095,
      "step": 5381
    },
    {
      "epoch": 0.15507405059643867,
      "grad_norm": 1.1817954778671265,
      "learning_rate": 9.595477333109823e-06,
      "loss": 0.7846,
      "step": 5382
    },
    {
      "epoch": 0.15510286405808793,
      "grad_norm": 1.1768242120742798,
      "learning_rate": 9.595293452412581e-06,
      "loss": 0.8158,
      "step": 5383
    },
    {
      "epoch": 0.15513167751973722,
      "grad_norm": 1.133495569229126,
      "learning_rate": 9.595109531694926e-06,
      "loss": 0.7888,
      "step": 5384
    },
    {
      "epoch": 0.1551604909813865,
      "grad_norm": 1.1860241889953613,
      "learning_rate": 9.59492557095846e-06,
      "loss": 0.7878,
      "step": 5385
    },
    {
      "epoch": 0.15518930444303577,
      "grad_norm": 1.1356465816497803,
      "learning_rate": 9.594741570204786e-06,
      "loss": 0.7794,
      "step": 5386
    },
    {
      "epoch": 0.15521811790468507,
      "grad_norm": 1.1946613788604736,
      "learning_rate": 9.594557529435506e-06,
      "loss": 0.806,
      "step": 5387
    },
    {
      "epoch": 0.15524693136633436,
      "grad_norm": 1.2277307510375977,
      "learning_rate": 9.594373448652224e-06,
      "loss": 0.8072,
      "step": 5388
    },
    {
      "epoch": 0.15527574482798365,
      "grad_norm": 1.1343399286270142,
      "learning_rate": 9.59418932785654e-06,
      "loss": 0.794,
      "step": 5389
    },
    {
      "epoch": 0.1553045582896329,
      "grad_norm": 1.20078444480896,
      "learning_rate": 9.594005167050061e-06,
      "loss": 0.8029,
      "step": 5390
    },
    {
      "epoch": 0.1553333717512822,
      "grad_norm": 1.2031408548355103,
      "learning_rate": 9.593820966234389e-06,
      "loss": 0.7993,
      "step": 5391
    },
    {
      "epoch": 0.1553621852129315,
      "grad_norm": 1.1271089315414429,
      "learning_rate": 9.593636725411128e-06,
      "loss": 0.8035,
      "step": 5392
    },
    {
      "epoch": 0.15539099867458075,
      "grad_norm": 1.1636815071105957,
      "learning_rate": 9.593452444581884e-06,
      "loss": 0.7874,
      "step": 5393
    },
    {
      "epoch": 0.15541981213623005,
      "grad_norm": 1.1513465642929077,
      "learning_rate": 9.59326812374826e-06,
      "loss": 0.8116,
      "step": 5394
    },
    {
      "epoch": 0.15544862559787934,
      "grad_norm": 1.1655957698822021,
      "learning_rate": 9.59308376291186e-06,
      "loss": 0.8098,
      "step": 5395
    },
    {
      "epoch": 0.1554774390595286,
      "grad_norm": 1.1624400615692139,
      "learning_rate": 9.592899362074296e-06,
      "loss": 0.7772,
      "step": 5396
    },
    {
      "epoch": 0.1555062525211779,
      "grad_norm": 1.2544206380844116,
      "learning_rate": 9.592714921237169e-06,
      "loss": 0.8005,
      "step": 5397
    },
    {
      "epoch": 0.15553506598282718,
      "grad_norm": 1.189961552619934,
      "learning_rate": 9.592530440402084e-06,
      "loss": 0.8367,
      "step": 5398
    },
    {
      "epoch": 0.15556387944447647,
      "grad_norm": 1.157637119293213,
      "learning_rate": 9.592345919570651e-06,
      "loss": 0.7839,
      "step": 5399
    },
    {
      "epoch": 0.15559269290612573,
      "grad_norm": 1.4263955354690552,
      "learning_rate": 9.592161358744477e-06,
      "loss": 0.7954,
      "step": 5400
    },
    {
      "epoch": 0.15562150636777503,
      "grad_norm": 1.1852341890335083,
      "learning_rate": 9.591976757925166e-06,
      "loss": 0.7763,
      "step": 5401
    },
    {
      "epoch": 0.15565031982942432,
      "grad_norm": 1.1995550394058228,
      "learning_rate": 9.591792117114328e-06,
      "loss": 0.7938,
      "step": 5402
    },
    {
      "epoch": 0.15567913329107358,
      "grad_norm": 1.1739943027496338,
      "learning_rate": 9.59160743631357e-06,
      "loss": 0.7931,
      "step": 5403
    },
    {
      "epoch": 0.15570794675272287,
      "grad_norm": 1.1435600519180298,
      "learning_rate": 9.591422715524504e-06,
      "loss": 0.7761,
      "step": 5404
    },
    {
      "epoch": 0.15573676021437216,
      "grad_norm": 1.1880195140838623,
      "learning_rate": 9.591237954748733e-06,
      "loss": 0.8109,
      "step": 5405
    },
    {
      "epoch": 0.15576557367602142,
      "grad_norm": 1.1621309518814087,
      "learning_rate": 9.591053153987871e-06,
      "loss": 0.7832,
      "step": 5406
    },
    {
      "epoch": 0.15579438713767071,
      "grad_norm": 1.1109001636505127,
      "learning_rate": 9.590868313243525e-06,
      "loss": 0.7766,
      "step": 5407
    },
    {
      "epoch": 0.15582320059932,
      "grad_norm": 1.1329811811447144,
      "learning_rate": 9.590683432517303e-06,
      "loss": 0.7941,
      "step": 5408
    },
    {
      "epoch": 0.1558520140609693,
      "grad_norm": 1.1959854364395142,
      "learning_rate": 9.590498511810819e-06,
      "loss": 0.812,
      "step": 5409
    },
    {
      "epoch": 0.15588082752261856,
      "grad_norm": 1.1600139141082764,
      "learning_rate": 9.590313551125683e-06,
      "loss": 0.7958,
      "step": 5410
    },
    {
      "epoch": 0.15590964098426785,
      "grad_norm": 1.2143656015396118,
      "learning_rate": 9.590128550463502e-06,
      "loss": 0.7976,
      "step": 5411
    },
    {
      "epoch": 0.15593845444591714,
      "grad_norm": 1.2268784046173096,
      "learning_rate": 9.589943509825892e-06,
      "loss": 0.8146,
      "step": 5412
    },
    {
      "epoch": 0.1559672679075664,
      "grad_norm": 1.137006163597107,
      "learning_rate": 9.589758429214462e-06,
      "loss": 0.7698,
      "step": 5413
    },
    {
      "epoch": 0.1559960813692157,
      "grad_norm": 1.175228238105774,
      "learning_rate": 9.589573308630823e-06,
      "loss": 0.8028,
      "step": 5414
    },
    {
      "epoch": 0.15602489483086499,
      "grad_norm": 1.1682707071304321,
      "learning_rate": 9.58938814807659e-06,
      "loss": 0.7723,
      "step": 5415
    },
    {
      "epoch": 0.15605370829251428,
      "grad_norm": 1.104996681213379,
      "learning_rate": 9.589202947553373e-06,
      "loss": 0.7716,
      "step": 5416
    },
    {
      "epoch": 0.15608252175416354,
      "grad_norm": 1.1289805173873901,
      "learning_rate": 9.589017707062785e-06,
      "loss": 0.7893,
      "step": 5417
    },
    {
      "epoch": 0.15611133521581283,
      "grad_norm": 1.1690104007720947,
      "learning_rate": 9.588832426606441e-06,
      "loss": 0.8091,
      "step": 5418
    },
    {
      "epoch": 0.15614014867746212,
      "grad_norm": 1.1518915891647339,
      "learning_rate": 9.588647106185953e-06,
      "loss": 0.7797,
      "step": 5419
    },
    {
      "epoch": 0.15616896213911138,
      "grad_norm": 1.1520026922225952,
      "learning_rate": 9.588461745802937e-06,
      "loss": 0.8184,
      "step": 5420
    },
    {
      "epoch": 0.15619777560076067,
      "grad_norm": 1.1941195726394653,
      "learning_rate": 9.588276345459005e-06,
      "loss": 0.7932,
      "step": 5421
    },
    {
      "epoch": 0.15622658906240997,
      "grad_norm": 1.1914311647415161,
      "learning_rate": 9.588090905155774e-06,
      "loss": 0.7829,
      "step": 5422
    },
    {
      "epoch": 0.15625540252405923,
      "grad_norm": 1.1464176177978516,
      "learning_rate": 9.587905424894856e-06,
      "loss": 0.8076,
      "step": 5423
    },
    {
      "epoch": 0.15628421598570852,
      "grad_norm": 1.1430892944335938,
      "learning_rate": 9.587719904677868e-06,
      "loss": 0.8086,
      "step": 5424
    },
    {
      "epoch": 0.1563130294473578,
      "grad_norm": 1.1355589628219604,
      "learning_rate": 9.587534344506427e-06,
      "loss": 0.8019,
      "step": 5425
    },
    {
      "epoch": 0.1563418429090071,
      "grad_norm": 1.1961030960083008,
      "learning_rate": 9.587348744382146e-06,
      "loss": 0.7771,
      "step": 5426
    },
    {
      "epoch": 0.15637065637065636,
      "grad_norm": 1.1587096452713013,
      "learning_rate": 9.587163104306645e-06,
      "loss": 0.7949,
      "step": 5427
    },
    {
      "epoch": 0.15639946983230565,
      "grad_norm": 1.1401411294937134,
      "learning_rate": 9.586977424281538e-06,
      "loss": 0.7905,
      "step": 5428
    },
    {
      "epoch": 0.15642828329395495,
      "grad_norm": 1.2006391286849976,
      "learning_rate": 9.586791704308443e-06,
      "loss": 0.7927,
      "step": 5429
    },
    {
      "epoch": 0.1564570967556042,
      "grad_norm": 1.1628479957580566,
      "learning_rate": 9.586605944388976e-06,
      "loss": 0.7687,
      "step": 5430
    },
    {
      "epoch": 0.1564859102172535,
      "grad_norm": 1.142641544342041,
      "learning_rate": 9.586420144524758e-06,
      "loss": 0.7905,
      "step": 5431
    },
    {
      "epoch": 0.1565147236789028,
      "grad_norm": 1.2785453796386719,
      "learning_rate": 9.586234304717404e-06,
      "loss": 0.8031,
      "step": 5432
    },
    {
      "epoch": 0.15654353714055205,
      "grad_norm": 1.1990143060684204,
      "learning_rate": 9.586048424968534e-06,
      "loss": 0.7681,
      "step": 5433
    },
    {
      "epoch": 0.15657235060220134,
      "grad_norm": 1.1770411729812622,
      "learning_rate": 9.585862505279766e-06,
      "loss": 0.7775,
      "step": 5434
    },
    {
      "epoch": 0.15660116406385063,
      "grad_norm": 1.2058160305023193,
      "learning_rate": 9.585676545652722e-06,
      "loss": 0.795,
      "step": 5435
    },
    {
      "epoch": 0.15662997752549992,
      "grad_norm": 1.145279884338379,
      "learning_rate": 9.585490546089018e-06,
      "loss": 0.8145,
      "step": 5436
    },
    {
      "epoch": 0.1566587909871492,
      "grad_norm": 1.1530934572219849,
      "learning_rate": 9.585304506590276e-06,
      "loss": 0.7941,
      "step": 5437
    },
    {
      "epoch": 0.15668760444879848,
      "grad_norm": 1.1703956127166748,
      "learning_rate": 9.585118427158114e-06,
      "loss": 0.8116,
      "step": 5438
    },
    {
      "epoch": 0.15671641791044777,
      "grad_norm": 1.157655954360962,
      "learning_rate": 9.584932307794154e-06,
      "loss": 0.7677,
      "step": 5439
    },
    {
      "epoch": 0.15674523137209703,
      "grad_norm": 1.1575181484222412,
      "learning_rate": 9.584746148500016e-06,
      "loss": 0.7976,
      "step": 5440
    },
    {
      "epoch": 0.15677404483374632,
      "grad_norm": 1.1624979972839355,
      "learning_rate": 9.584559949277324e-06,
      "loss": 0.7986,
      "step": 5441
    },
    {
      "epoch": 0.15680285829539561,
      "grad_norm": 1.136459469795227,
      "learning_rate": 9.584373710127698e-06,
      "loss": 0.8012,
      "step": 5442
    },
    {
      "epoch": 0.1568316717570449,
      "grad_norm": 1.193762183189392,
      "learning_rate": 9.584187431052758e-06,
      "loss": 0.8102,
      "step": 5443
    },
    {
      "epoch": 0.15686048521869417,
      "grad_norm": 1.1526399850845337,
      "learning_rate": 9.584001112054129e-06,
      "loss": 0.7732,
      "step": 5444
    },
    {
      "epoch": 0.15688929868034346,
      "grad_norm": 1.1771363019943237,
      "learning_rate": 9.583814753133434e-06,
      "loss": 0.7979,
      "step": 5445
    },
    {
      "epoch": 0.15691811214199275,
      "grad_norm": 1.1406689882278442,
      "learning_rate": 9.583628354292292e-06,
      "loss": 0.793,
      "step": 5446
    },
    {
      "epoch": 0.156946925603642,
      "grad_norm": 1.1318409442901611,
      "learning_rate": 9.58344191553233e-06,
      "loss": 0.7978,
      "step": 5447
    },
    {
      "epoch": 0.1569757390652913,
      "grad_norm": 1.1683964729309082,
      "learning_rate": 9.583255436855171e-06,
      "loss": 0.8015,
      "step": 5448
    },
    {
      "epoch": 0.1570045525269406,
      "grad_norm": 1.1541308164596558,
      "learning_rate": 9.583068918262439e-06,
      "loss": 0.7996,
      "step": 5449
    },
    {
      "epoch": 0.15703336598858986,
      "grad_norm": 1.167843222618103,
      "learning_rate": 9.582882359755757e-06,
      "loss": 0.7864,
      "step": 5450
    },
    {
      "epoch": 0.15706217945023915,
      "grad_norm": 1.1805446147918701,
      "learning_rate": 9.58269576133675e-06,
      "loss": 0.7892,
      "step": 5451
    },
    {
      "epoch": 0.15709099291188844,
      "grad_norm": 1.754154086112976,
      "learning_rate": 9.582509123007046e-06,
      "loss": 0.7826,
      "step": 5452
    },
    {
      "epoch": 0.15711980637353773,
      "grad_norm": 1.2071874141693115,
      "learning_rate": 9.582322444768268e-06,
      "loss": 0.7927,
      "step": 5453
    },
    {
      "epoch": 0.157148619835187,
      "grad_norm": 1.232377052307129,
      "learning_rate": 9.582135726622044e-06,
      "loss": 0.7903,
      "step": 5454
    },
    {
      "epoch": 0.15717743329683628,
      "grad_norm": 1.2109488248825073,
      "learning_rate": 9.581948968569993e-06,
      "loss": 0.8016,
      "step": 5455
    },
    {
      "epoch": 0.15720624675848557,
      "grad_norm": 1.239449381828308,
      "learning_rate": 9.581762170613752e-06,
      "loss": 0.8014,
      "step": 5456
    },
    {
      "epoch": 0.15723506022013484,
      "grad_norm": 1.1870330572128296,
      "learning_rate": 9.58157533275494e-06,
      "loss": 0.8108,
      "step": 5457
    },
    {
      "epoch": 0.15726387368178413,
      "grad_norm": 1.2689924240112305,
      "learning_rate": 9.581388454995188e-06,
      "loss": 0.7974,
      "step": 5458
    },
    {
      "epoch": 0.15729268714343342,
      "grad_norm": 1.202983021736145,
      "learning_rate": 9.581201537336121e-06,
      "loss": 0.8161,
      "step": 5459
    },
    {
      "epoch": 0.15732150060508268,
      "grad_norm": 1.1547127962112427,
      "learning_rate": 9.581014579779368e-06,
      "loss": 0.7856,
      "step": 5460
    },
    {
      "epoch": 0.15735031406673197,
      "grad_norm": 1.1803869009017944,
      "learning_rate": 9.580827582326558e-06,
      "loss": 0.7747,
      "step": 5461
    },
    {
      "epoch": 0.15737912752838126,
      "grad_norm": 1.192209243774414,
      "learning_rate": 9.58064054497932e-06,
      "loss": 0.8074,
      "step": 5462
    },
    {
      "epoch": 0.15740794099003055,
      "grad_norm": 1.1708698272705078,
      "learning_rate": 9.58045346773928e-06,
      "loss": 0.7854,
      "step": 5463
    },
    {
      "epoch": 0.15743675445167982,
      "grad_norm": 1.184123158454895,
      "learning_rate": 9.58026635060807e-06,
      "loss": 0.7996,
      "step": 5464
    },
    {
      "epoch": 0.1574655679133291,
      "grad_norm": 1.2113853693008423,
      "learning_rate": 9.58007919358732e-06,
      "loss": 0.7922,
      "step": 5465
    },
    {
      "epoch": 0.1574943813749784,
      "grad_norm": 1.1559175252914429,
      "learning_rate": 9.579891996678656e-06,
      "loss": 0.7844,
      "step": 5466
    },
    {
      "epoch": 0.15752319483662766,
      "grad_norm": 1.2044254541397095,
      "learning_rate": 9.579704759883713e-06,
      "loss": 0.8407,
      "step": 5467
    },
    {
      "epoch": 0.15755200829827695,
      "grad_norm": 1.1387678384780884,
      "learning_rate": 9.579517483204118e-06,
      "loss": 0.7978,
      "step": 5468
    },
    {
      "epoch": 0.15758082175992624,
      "grad_norm": 1.1772332191467285,
      "learning_rate": 9.579330166641503e-06,
      "loss": 0.7781,
      "step": 5469
    },
    {
      "epoch": 0.15760963522157553,
      "grad_norm": 1.1434166431427002,
      "learning_rate": 9.579142810197501e-06,
      "loss": 0.7876,
      "step": 5470
    },
    {
      "epoch": 0.1576384486832248,
      "grad_norm": 1.15557062625885,
      "learning_rate": 9.578955413873743e-06,
      "loss": 0.7834,
      "step": 5471
    },
    {
      "epoch": 0.1576672621448741,
      "grad_norm": 1.1678215265274048,
      "learning_rate": 9.57876797767186e-06,
      "loss": 0.7908,
      "step": 5472
    },
    {
      "epoch": 0.15769607560652338,
      "grad_norm": 1.1203209161758423,
      "learning_rate": 9.578580501593486e-06,
      "loss": 0.7786,
      "step": 5473
    },
    {
      "epoch": 0.15772488906817264,
      "grad_norm": 1.1715731620788574,
      "learning_rate": 9.578392985640253e-06,
      "loss": 0.7746,
      "step": 5474
    },
    {
      "epoch": 0.15775370252982193,
      "grad_norm": 1.1641252040863037,
      "learning_rate": 9.578205429813792e-06,
      "loss": 0.8202,
      "step": 5475
    },
    {
      "epoch": 0.15778251599147122,
      "grad_norm": 1.1647759675979614,
      "learning_rate": 9.578017834115739e-06,
      "loss": 0.7855,
      "step": 5476
    },
    {
      "epoch": 0.15781132945312049,
      "grad_norm": 1.2233173847198486,
      "learning_rate": 9.577830198547727e-06,
      "loss": 0.8047,
      "step": 5477
    },
    {
      "epoch": 0.15784014291476978,
      "grad_norm": 1.2076302766799927,
      "learning_rate": 9.57764252311139e-06,
      "loss": 0.7852,
      "step": 5478
    },
    {
      "epoch": 0.15786895637641907,
      "grad_norm": 1.1931865215301514,
      "learning_rate": 9.577454807808364e-06,
      "loss": 0.7804,
      "step": 5479
    },
    {
      "epoch": 0.15789776983806836,
      "grad_norm": 1.1359864473342896,
      "learning_rate": 9.57726705264028e-06,
      "loss": 0.7914,
      "step": 5480
    },
    {
      "epoch": 0.15792658329971762,
      "grad_norm": 1.1342055797576904,
      "learning_rate": 9.577079257608777e-06,
      "loss": 0.7835,
      "step": 5481
    },
    {
      "epoch": 0.1579553967613669,
      "grad_norm": 1.1491570472717285,
      "learning_rate": 9.576891422715489e-06,
      "loss": 0.7971,
      "step": 5482
    },
    {
      "epoch": 0.1579842102230162,
      "grad_norm": 1.0934231281280518,
      "learning_rate": 9.576703547962052e-06,
      "loss": 0.7895,
      "step": 5483
    },
    {
      "epoch": 0.15801302368466547,
      "grad_norm": 1.1111295223236084,
      "learning_rate": 9.576515633350102e-06,
      "loss": 0.7877,
      "step": 5484
    },
    {
      "epoch": 0.15804183714631476,
      "grad_norm": 1.1119897365570068,
      "learning_rate": 9.576327678881275e-06,
      "loss": 0.7829,
      "step": 5485
    },
    {
      "epoch": 0.15807065060796405,
      "grad_norm": 1.1031079292297363,
      "learning_rate": 9.57613968455721e-06,
      "loss": 0.7747,
      "step": 5486
    },
    {
      "epoch": 0.1580994640696133,
      "grad_norm": 1.1190547943115234,
      "learning_rate": 9.575951650379543e-06,
      "loss": 0.8078,
      "step": 5487
    },
    {
      "epoch": 0.1581282775312626,
      "grad_norm": 1.1654553413391113,
      "learning_rate": 9.57576357634991e-06,
      "loss": 0.7819,
      "step": 5488
    },
    {
      "epoch": 0.1581570909929119,
      "grad_norm": 1.1619997024536133,
      "learning_rate": 9.57557546246995e-06,
      "loss": 0.7898,
      "step": 5489
    },
    {
      "epoch": 0.15818590445456118,
      "grad_norm": 1.1515706777572632,
      "learning_rate": 9.575387308741301e-06,
      "loss": 0.8155,
      "step": 5490
    },
    {
      "epoch": 0.15821471791621045,
      "grad_norm": 1.1319137811660767,
      "learning_rate": 9.575199115165605e-06,
      "loss": 0.7682,
      "step": 5491
    },
    {
      "epoch": 0.15824353137785974,
      "grad_norm": 1.1513030529022217,
      "learning_rate": 9.575010881744496e-06,
      "loss": 0.7722,
      "step": 5492
    },
    {
      "epoch": 0.15827234483950903,
      "grad_norm": 1.09774649143219,
      "learning_rate": 9.574822608479616e-06,
      "loss": 0.785,
      "step": 5493
    },
    {
      "epoch": 0.1583011583011583,
      "grad_norm": 1.1260135173797607,
      "learning_rate": 9.574634295372606e-06,
      "loss": 0.7821,
      "step": 5494
    },
    {
      "epoch": 0.15832997176280758,
      "grad_norm": 1.1598389148712158,
      "learning_rate": 9.574445942425102e-06,
      "loss": 0.7858,
      "step": 5495
    },
    {
      "epoch": 0.15835878522445687,
      "grad_norm": 1.1653037071228027,
      "learning_rate": 9.574257549638748e-06,
      "loss": 0.7904,
      "step": 5496
    },
    {
      "epoch": 0.15838759868610616,
      "grad_norm": 1.103297233581543,
      "learning_rate": 9.574069117015183e-06,
      "loss": 0.7883,
      "step": 5497
    },
    {
      "epoch": 0.15841641214775543,
      "grad_norm": 1.1781821250915527,
      "learning_rate": 9.573880644556049e-06,
      "loss": 0.7734,
      "step": 5498
    },
    {
      "epoch": 0.15844522560940472,
      "grad_norm": 1.1542946100234985,
      "learning_rate": 9.573692132262984e-06,
      "loss": 0.7971,
      "step": 5499
    },
    {
      "epoch": 0.158474039071054,
      "grad_norm": 1.1239147186279297,
      "learning_rate": 9.573503580137635e-06,
      "loss": 0.8076,
      "step": 5500
    },
    {
      "epoch": 0.15850285253270327,
      "grad_norm": 1.132394552230835,
      "learning_rate": 9.57331498818164e-06,
      "loss": 0.8084,
      "step": 5501
    },
    {
      "epoch": 0.15853166599435256,
      "grad_norm": 1.1944808959960938,
      "learning_rate": 9.573126356396645e-06,
      "loss": 0.8069,
      "step": 5502
    },
    {
      "epoch": 0.15856047945600185,
      "grad_norm": 1.1578940153121948,
      "learning_rate": 9.572937684784292e-06,
      "loss": 0.8098,
      "step": 5503
    },
    {
      "epoch": 0.15858929291765111,
      "grad_norm": 1.2241729497909546,
      "learning_rate": 9.572748973346219e-06,
      "loss": 0.8162,
      "step": 5504
    },
    {
      "epoch": 0.1586181063793004,
      "grad_norm": 1.2495299577713013,
      "learning_rate": 9.572560222084078e-06,
      "loss": 0.8047,
      "step": 5505
    },
    {
      "epoch": 0.1586469198409497,
      "grad_norm": 1.218397617340088,
      "learning_rate": 9.572371430999506e-06,
      "loss": 0.8099,
      "step": 5506
    },
    {
      "epoch": 0.158675733302599,
      "grad_norm": 1.2636632919311523,
      "learning_rate": 9.57218260009415e-06,
      "loss": 0.818,
      "step": 5507
    },
    {
      "epoch": 0.15870454676424825,
      "grad_norm": 1.16177237033844,
      "learning_rate": 9.571993729369652e-06,
      "loss": 0.789,
      "step": 5508
    },
    {
      "epoch": 0.15873336022589754,
      "grad_norm": 1.2189923524856567,
      "learning_rate": 9.571804818827661e-06,
      "loss": 0.7863,
      "step": 5509
    },
    {
      "epoch": 0.15876217368754683,
      "grad_norm": 1.1933424472808838,
      "learning_rate": 9.571615868469819e-06,
      "loss": 0.7733,
      "step": 5510
    },
    {
      "epoch": 0.1587909871491961,
      "grad_norm": 1.1555900573730469,
      "learning_rate": 9.571426878297774e-06,
      "loss": 0.7933,
      "step": 5511
    },
    {
      "epoch": 0.15881980061084539,
      "grad_norm": 1.2319468259811401,
      "learning_rate": 9.57123784831317e-06,
      "loss": 0.8031,
      "step": 5512
    },
    {
      "epoch": 0.15884861407249468,
      "grad_norm": 1.1658295392990112,
      "learning_rate": 9.571048778517655e-06,
      "loss": 0.7773,
      "step": 5513
    },
    {
      "epoch": 0.15887742753414394,
      "grad_norm": 1.1518045663833618,
      "learning_rate": 9.570859668912874e-06,
      "loss": 0.814,
      "step": 5514
    },
    {
      "epoch": 0.15890624099579323,
      "grad_norm": 1.2937183380126953,
      "learning_rate": 9.570670519500473e-06,
      "loss": 0.804,
      "step": 5515
    },
    {
      "epoch": 0.15893505445744252,
      "grad_norm": 1.1376500129699707,
      "learning_rate": 9.570481330282103e-06,
      "loss": 0.7952,
      "step": 5516
    },
    {
      "epoch": 0.1589638679190918,
      "grad_norm": 1.2150975465774536,
      "learning_rate": 9.570292101259406e-06,
      "loss": 0.7987,
      "step": 5517
    },
    {
      "epoch": 0.15899268138074107,
      "grad_norm": 1.1322510242462158,
      "learning_rate": 9.570102832434037e-06,
      "loss": 0.7896,
      "step": 5518
    },
    {
      "epoch": 0.15902149484239037,
      "grad_norm": 1.1066004037857056,
      "learning_rate": 9.56991352380764e-06,
      "loss": 0.7748,
      "step": 5519
    },
    {
      "epoch": 0.15905030830403966,
      "grad_norm": 1.2300235033035278,
      "learning_rate": 9.569724175381863e-06,
      "loss": 0.7794,
      "step": 5520
    },
    {
      "epoch": 0.15907912176568892,
      "grad_norm": 1.1774605512619019,
      "learning_rate": 9.569534787158357e-06,
      "loss": 0.8035,
      "step": 5521
    },
    {
      "epoch": 0.1591079352273382,
      "grad_norm": 1.1904579401016235,
      "learning_rate": 9.569345359138771e-06,
      "loss": 0.8028,
      "step": 5522
    },
    {
      "epoch": 0.1591367486889875,
      "grad_norm": 1.1305499076843262,
      "learning_rate": 9.569155891324756e-06,
      "loss": 0.8017,
      "step": 5523
    },
    {
      "epoch": 0.1591655621506368,
      "grad_norm": 1.1149851083755493,
      "learning_rate": 9.568966383717958e-06,
      "loss": 0.8004,
      "step": 5524
    },
    {
      "epoch": 0.15919437561228605,
      "grad_norm": 1.1929336786270142,
      "learning_rate": 9.568776836320032e-06,
      "loss": 0.7884,
      "step": 5525
    },
    {
      "epoch": 0.15922318907393535,
      "grad_norm": 1.129448413848877,
      "learning_rate": 9.568587249132627e-06,
      "loss": 0.8154,
      "step": 5526
    },
    {
      "epoch": 0.15925200253558464,
      "grad_norm": 1.1485621929168701,
      "learning_rate": 9.568397622157392e-06,
      "loss": 0.8011,
      "step": 5527
    },
    {
      "epoch": 0.1592808159972339,
      "grad_norm": 1.158054232597351,
      "learning_rate": 9.568207955395983e-06,
      "loss": 0.8097,
      "step": 5528
    },
    {
      "epoch": 0.1593096294588832,
      "grad_norm": 1.193120002746582,
      "learning_rate": 9.568018248850048e-06,
      "loss": 0.7893,
      "step": 5529
    },
    {
      "epoch": 0.15933844292053248,
      "grad_norm": 1.1761530637741089,
      "learning_rate": 9.567828502521239e-06,
      "loss": 0.7887,
      "step": 5530
    },
    {
      "epoch": 0.15936725638218174,
      "grad_norm": 1.1831605434417725,
      "learning_rate": 9.567638716411212e-06,
      "loss": 0.8297,
      "step": 5531
    },
    {
      "epoch": 0.15939606984383103,
      "grad_norm": 1.1295583248138428,
      "learning_rate": 9.567448890521618e-06,
      "loss": 0.7969,
      "step": 5532
    },
    {
      "epoch": 0.15942488330548032,
      "grad_norm": 1.2002774477005005,
      "learning_rate": 9.567259024854109e-06,
      "loss": 0.7941,
      "step": 5533
    },
    {
      "epoch": 0.15945369676712962,
      "grad_norm": 1.1095465421676636,
      "learning_rate": 9.56706911941034e-06,
      "loss": 0.7842,
      "step": 5534
    },
    {
      "epoch": 0.15948251022877888,
      "grad_norm": 1.1305674314498901,
      "learning_rate": 9.566879174191964e-06,
      "loss": 0.7673,
      "step": 5535
    },
    {
      "epoch": 0.15951132369042817,
      "grad_norm": 1.1473287343978882,
      "learning_rate": 9.566689189200634e-06,
      "loss": 0.7884,
      "step": 5536
    },
    {
      "epoch": 0.15954013715207746,
      "grad_norm": 1.1559213399887085,
      "learning_rate": 9.566499164438007e-06,
      "loss": 0.7945,
      "step": 5537
    },
    {
      "epoch": 0.15956895061372672,
      "grad_norm": 1.1151925325393677,
      "learning_rate": 9.56630909990574e-06,
      "loss": 0.7883,
      "step": 5538
    },
    {
      "epoch": 0.15959776407537601,
      "grad_norm": 1.092982292175293,
      "learning_rate": 9.566118995605483e-06,
      "loss": 0.7741,
      "step": 5539
    },
    {
      "epoch": 0.1596265775370253,
      "grad_norm": 1.1580097675323486,
      "learning_rate": 9.565928851538892e-06,
      "loss": 0.8107,
      "step": 5540
    },
    {
      "epoch": 0.15965539099867457,
      "grad_norm": 1.1491352319717407,
      "learning_rate": 9.565738667707629e-06,
      "loss": 0.7979,
      "step": 5541
    },
    {
      "epoch": 0.15968420446032386,
      "grad_norm": 1.101741909980774,
      "learning_rate": 9.565548444113343e-06,
      "loss": 0.7932,
      "step": 5542
    },
    {
      "epoch": 0.15971301792197315,
      "grad_norm": 1.111298680305481,
      "learning_rate": 9.565358180757695e-06,
      "loss": 0.8078,
      "step": 5543
    },
    {
      "epoch": 0.15974183138362244,
      "grad_norm": 1.1679391860961914,
      "learning_rate": 9.565167877642342e-06,
      "loss": 0.7716,
      "step": 5544
    },
    {
      "epoch": 0.1597706448452717,
      "grad_norm": 1.155301570892334,
      "learning_rate": 9.56497753476894e-06,
      "loss": 0.7889,
      "step": 5545
    },
    {
      "epoch": 0.159799458306921,
      "grad_norm": 1.115863561630249,
      "learning_rate": 9.564787152139147e-06,
      "loss": 0.798,
      "step": 5546
    },
    {
      "epoch": 0.15982827176857028,
      "grad_norm": 1.201175570487976,
      "learning_rate": 9.564596729754621e-06,
      "loss": 0.8034,
      "step": 5547
    },
    {
      "epoch": 0.15985708523021955,
      "grad_norm": 1.1918283700942993,
      "learning_rate": 9.56440626761702e-06,
      "loss": 0.8162,
      "step": 5548
    },
    {
      "epoch": 0.15988589869186884,
      "grad_norm": 1.1490561962127686,
      "learning_rate": 9.564215765728003e-06,
      "loss": 0.7847,
      "step": 5549
    },
    {
      "epoch": 0.15991471215351813,
      "grad_norm": 1.1256389617919922,
      "learning_rate": 9.564025224089228e-06,
      "loss": 0.7904,
      "step": 5550
    },
    {
      "epoch": 0.15994352561516742,
      "grad_norm": 1.145959496498108,
      "learning_rate": 9.563834642702357e-06,
      "loss": 0.8334,
      "step": 5551
    },
    {
      "epoch": 0.15997233907681668,
      "grad_norm": 1.1074861288070679,
      "learning_rate": 9.56364402156905e-06,
      "loss": 0.7776,
      "step": 5552
    },
    {
      "epoch": 0.16000115253846597,
      "grad_norm": 1.1300268173217773,
      "learning_rate": 9.563453360690963e-06,
      "loss": 0.7917,
      "step": 5553
    },
    {
      "epoch": 0.16002996600011526,
      "grad_norm": 1.195512294769287,
      "learning_rate": 9.56326266006976e-06,
      "loss": 0.8135,
      "step": 5554
    },
    {
      "epoch": 0.16005877946176453,
      "grad_norm": 1.1670632362365723,
      "learning_rate": 9.5630719197071e-06,
      "loss": 0.7978,
      "step": 5555
    },
    {
      "epoch": 0.16008759292341382,
      "grad_norm": 1.1424890756607056,
      "learning_rate": 9.562881139604648e-06,
      "loss": 0.7938,
      "step": 5556
    },
    {
      "epoch": 0.1601164063850631,
      "grad_norm": 1.1800400018692017,
      "learning_rate": 9.562690319764059e-06,
      "loss": 0.7996,
      "step": 5557
    },
    {
      "epoch": 0.16014521984671237,
      "grad_norm": 1.1722251176834106,
      "learning_rate": 9.562499460186999e-06,
      "loss": 0.8055,
      "step": 5558
    },
    {
      "epoch": 0.16017403330836166,
      "grad_norm": 1.1382758617401123,
      "learning_rate": 9.562308560875131e-06,
      "loss": 0.7766,
      "step": 5559
    },
    {
      "epoch": 0.16020284677001095,
      "grad_norm": 1.13948655128479,
      "learning_rate": 9.562117621830113e-06,
      "loss": 0.806,
      "step": 5560
    },
    {
      "epoch": 0.16023166023166024,
      "grad_norm": 1.146347999572754,
      "learning_rate": 9.561926643053614e-06,
      "loss": 0.8049,
      "step": 5561
    },
    {
      "epoch": 0.1602604736933095,
      "grad_norm": 1.1492900848388672,
      "learning_rate": 9.561735624547295e-06,
      "loss": 0.7979,
      "step": 5562
    },
    {
      "epoch": 0.1602892871549588,
      "grad_norm": 1.1760765314102173,
      "learning_rate": 9.561544566312815e-06,
      "loss": 0.7956,
      "step": 5563
    },
    {
      "epoch": 0.1603181006166081,
      "grad_norm": 1.1053290367126465,
      "learning_rate": 9.561353468351845e-06,
      "loss": 0.7998,
      "step": 5564
    },
    {
      "epoch": 0.16034691407825735,
      "grad_norm": 1.1786737442016602,
      "learning_rate": 9.561162330666045e-06,
      "loss": 0.7861,
      "step": 5565
    },
    {
      "epoch": 0.16037572753990664,
      "grad_norm": 1.1273337602615356,
      "learning_rate": 9.56097115325708e-06,
      "loss": 0.7668,
      "step": 5566
    },
    {
      "epoch": 0.16040454100155593,
      "grad_norm": 1.1502556800842285,
      "learning_rate": 9.560779936126616e-06,
      "loss": 0.8173,
      "step": 5567
    },
    {
      "epoch": 0.1604333544632052,
      "grad_norm": 1.1889272928237915,
      "learning_rate": 9.560588679276319e-06,
      "loss": 0.8136,
      "step": 5568
    },
    {
      "epoch": 0.1604621679248545,
      "grad_norm": 1.1292742490768433,
      "learning_rate": 9.560397382707852e-06,
      "loss": 0.782,
      "step": 5569
    },
    {
      "epoch": 0.16049098138650378,
      "grad_norm": 1.1019433736801147,
      "learning_rate": 9.560206046422883e-06,
      "loss": 0.8115,
      "step": 5570
    },
    {
      "epoch": 0.16051979484815307,
      "grad_norm": 1.1799869537353516,
      "learning_rate": 9.560014670423076e-06,
      "loss": 0.8184,
      "step": 5571
    },
    {
      "epoch": 0.16054860830980233,
      "grad_norm": 1.1301038265228271,
      "learning_rate": 9.559823254710103e-06,
      "loss": 0.8308,
      "step": 5572
    },
    {
      "epoch": 0.16057742177145162,
      "grad_norm": 1.1068836450576782,
      "learning_rate": 9.559631799285624e-06,
      "loss": 0.7935,
      "step": 5573
    },
    {
      "epoch": 0.1606062352331009,
      "grad_norm": 1.1687517166137695,
      "learning_rate": 9.559440304151313e-06,
      "loss": 0.8199,
      "step": 5574
    },
    {
      "epoch": 0.16063504869475018,
      "grad_norm": 1.16658616065979,
      "learning_rate": 9.559248769308833e-06,
      "loss": 0.7819,
      "step": 5575
    },
    {
      "epoch": 0.16066386215639947,
      "grad_norm": 1.1432602405548096,
      "learning_rate": 9.559057194759855e-06,
      "loss": 0.793,
      "step": 5576
    },
    {
      "epoch": 0.16069267561804876,
      "grad_norm": 1.1121517419815063,
      "learning_rate": 9.558865580506046e-06,
      "loss": 0.7874,
      "step": 5577
    },
    {
      "epoch": 0.16072148907969802,
      "grad_norm": 1.1093223094940186,
      "learning_rate": 9.558673926549075e-06,
      "loss": 0.7916,
      "step": 5578
    },
    {
      "epoch": 0.1607503025413473,
      "grad_norm": 1.1198362112045288,
      "learning_rate": 9.55848223289061e-06,
      "loss": 0.783,
      "step": 5579
    },
    {
      "epoch": 0.1607791160029966,
      "grad_norm": 1.174147605895996,
      "learning_rate": 9.558290499532323e-06,
      "loss": 0.834,
      "step": 5580
    },
    {
      "epoch": 0.1608079294646459,
      "grad_norm": 1.1011520624160767,
      "learning_rate": 9.558098726475882e-06,
      "loss": 0.8057,
      "step": 5581
    },
    {
      "epoch": 0.16083674292629516,
      "grad_norm": 1.123639702796936,
      "learning_rate": 9.557906913722957e-06,
      "loss": 0.8167,
      "step": 5582
    },
    {
      "epoch": 0.16086555638794445,
      "grad_norm": 1.163598895072937,
      "learning_rate": 9.557715061275218e-06,
      "loss": 0.7994,
      "step": 5583
    },
    {
      "epoch": 0.16089436984959374,
      "grad_norm": 1.1030006408691406,
      "learning_rate": 9.55752316913434e-06,
      "loss": 0.8193,
      "step": 5584
    },
    {
      "epoch": 0.160923183311243,
      "grad_norm": 1.1560558080673218,
      "learning_rate": 9.557331237301988e-06,
      "loss": 0.81,
      "step": 5585
    },
    {
      "epoch": 0.1609519967728923,
      "grad_norm": 1.1241466999053955,
      "learning_rate": 9.557139265779839e-06,
      "loss": 0.797,
      "step": 5586
    },
    {
      "epoch": 0.16098081023454158,
      "grad_norm": 1.1513935327529907,
      "learning_rate": 9.556947254569559e-06,
      "loss": 0.802,
      "step": 5587
    },
    {
      "epoch": 0.16100962369619087,
      "grad_norm": 1.1382195949554443,
      "learning_rate": 9.556755203672828e-06,
      "loss": 0.7877,
      "step": 5588
    },
    {
      "epoch": 0.16103843715784014,
      "grad_norm": 1.1444331407546997,
      "learning_rate": 9.55656311309131e-06,
      "loss": 0.8186,
      "step": 5589
    },
    {
      "epoch": 0.16106725061948943,
      "grad_norm": 1.102851390838623,
      "learning_rate": 9.556370982826686e-06,
      "loss": 0.8051,
      "step": 5590
    },
    {
      "epoch": 0.16109606408113872,
      "grad_norm": 1.1111032962799072,
      "learning_rate": 9.556178812880623e-06,
      "loss": 0.7851,
      "step": 5591
    },
    {
      "epoch": 0.16112487754278798,
      "grad_norm": 1.1625652313232422,
      "learning_rate": 9.555986603254798e-06,
      "loss": 0.7897,
      "step": 5592
    },
    {
      "epoch": 0.16115369100443727,
      "grad_norm": 1.1369144916534424,
      "learning_rate": 9.555794353950885e-06,
      "loss": 0.7908,
      "step": 5593
    },
    {
      "epoch": 0.16118250446608656,
      "grad_norm": 1.1290733814239502,
      "learning_rate": 9.555602064970555e-06,
      "loss": 0.7914,
      "step": 5594
    },
    {
      "epoch": 0.16121131792773583,
      "grad_norm": 1.1219462156295776,
      "learning_rate": 9.555409736315487e-06,
      "loss": 0.8023,
      "step": 5595
    },
    {
      "epoch": 0.16124013138938512,
      "grad_norm": 1.192973256111145,
      "learning_rate": 9.555217367987353e-06,
      "loss": 0.7807,
      "step": 5596
    },
    {
      "epoch": 0.1612689448510344,
      "grad_norm": 1.1119718551635742,
      "learning_rate": 9.555024959987828e-06,
      "loss": 0.759,
      "step": 5597
    },
    {
      "epoch": 0.1612977583126837,
      "grad_norm": 1.178562879562378,
      "learning_rate": 9.554832512318591e-06,
      "loss": 0.8125,
      "step": 5598
    },
    {
      "epoch": 0.16132657177433296,
      "grad_norm": 1.1582162380218506,
      "learning_rate": 9.554640024981315e-06,
      "loss": 0.794,
      "step": 5599
    },
    {
      "epoch": 0.16135538523598225,
      "grad_norm": 1.1683341264724731,
      "learning_rate": 9.554447497977678e-06,
      "loss": 0.7954,
      "step": 5600
    },
    {
      "epoch": 0.16138419869763154,
      "grad_norm": 1.1494781970977783,
      "learning_rate": 9.554254931309354e-06,
      "loss": 0.7886,
      "step": 5601
    },
    {
      "epoch": 0.1614130121592808,
      "grad_norm": 1.0924135446548462,
      "learning_rate": 9.554062324978026e-06,
      "loss": 0.807,
      "step": 5602
    },
    {
      "epoch": 0.1614418256209301,
      "grad_norm": 1.1354790925979614,
      "learning_rate": 9.553869678985365e-06,
      "loss": 0.7969,
      "step": 5603
    },
    {
      "epoch": 0.1614706390825794,
      "grad_norm": 1.1673027276992798,
      "learning_rate": 9.553676993333051e-06,
      "loss": 0.7948,
      "step": 5604
    },
    {
      "epoch": 0.16149945254422865,
      "grad_norm": 1.1372514963150024,
      "learning_rate": 9.553484268022763e-06,
      "loss": 0.8022,
      "step": 5605
    },
    {
      "epoch": 0.16152826600587794,
      "grad_norm": 1.0846235752105713,
      "learning_rate": 9.553291503056178e-06,
      "loss": 0.7963,
      "step": 5606
    },
    {
      "epoch": 0.16155707946752723,
      "grad_norm": 1.0725188255310059,
      "learning_rate": 9.553098698434977e-06,
      "loss": 0.7604,
      "step": 5607
    },
    {
      "epoch": 0.16158589292917652,
      "grad_norm": 1.1360414028167725,
      "learning_rate": 9.55290585416084e-06,
      "loss": 0.7945,
      "step": 5608
    },
    {
      "epoch": 0.16161470639082579,
      "grad_norm": 1.1665211915969849,
      "learning_rate": 9.55271297023544e-06,
      "loss": 0.8163,
      "step": 5609
    },
    {
      "epoch": 0.16164351985247508,
      "grad_norm": 1.1739211082458496,
      "learning_rate": 9.552520046660463e-06,
      "loss": 0.8008,
      "step": 5610
    },
    {
      "epoch": 0.16167233331412437,
      "grad_norm": 1.1464508771896362,
      "learning_rate": 9.552327083437587e-06,
      "loss": 0.7874,
      "step": 5611
    },
    {
      "epoch": 0.16170114677577363,
      "grad_norm": 1.1529592275619507,
      "learning_rate": 9.55213408056849e-06,
      "loss": 0.8026,
      "step": 5612
    },
    {
      "epoch": 0.16172996023742292,
      "grad_norm": 1.1553106307983398,
      "learning_rate": 9.55194103805486e-06,
      "loss": 0.81,
      "step": 5613
    },
    {
      "epoch": 0.1617587736990722,
      "grad_norm": 1.1318318843841553,
      "learning_rate": 9.551747955898372e-06,
      "loss": 0.7869,
      "step": 5614
    },
    {
      "epoch": 0.1617875871607215,
      "grad_norm": 1.2247928380966187,
      "learning_rate": 9.551554834100708e-06,
      "loss": 0.813,
      "step": 5615
    },
    {
      "epoch": 0.16181640062237077,
      "grad_norm": 1.1263271570205688,
      "learning_rate": 9.551361672663554e-06,
      "loss": 0.7913,
      "step": 5616
    },
    {
      "epoch": 0.16184521408402006,
      "grad_norm": 1.1700921058654785,
      "learning_rate": 9.551168471588589e-06,
      "loss": 0.7875,
      "step": 5617
    },
    {
      "epoch": 0.16187402754566935,
      "grad_norm": 1.2018481492996216,
      "learning_rate": 9.550975230877495e-06,
      "loss": 0.8114,
      "step": 5618
    },
    {
      "epoch": 0.1619028410073186,
      "grad_norm": 1.1294039487838745,
      "learning_rate": 9.550781950531957e-06,
      "loss": 0.7978,
      "step": 5619
    },
    {
      "epoch": 0.1619316544689679,
      "grad_norm": 1.1593763828277588,
      "learning_rate": 9.550588630553657e-06,
      "loss": 0.7773,
      "step": 5620
    },
    {
      "epoch": 0.1619604679306172,
      "grad_norm": 1.1667358875274658,
      "learning_rate": 9.550395270944278e-06,
      "loss": 0.8126,
      "step": 5621
    },
    {
      "epoch": 0.16198928139226645,
      "grad_norm": 1.2215619087219238,
      "learning_rate": 9.550201871705507e-06,
      "loss": 0.8185,
      "step": 5622
    },
    {
      "epoch": 0.16201809485391575,
      "grad_norm": 1.1863728761672974,
      "learning_rate": 9.550008432839024e-06,
      "loss": 0.8021,
      "step": 5623
    },
    {
      "epoch": 0.16204690831556504,
      "grad_norm": 1.1718248128890991,
      "learning_rate": 9.549814954346517e-06,
      "loss": 0.7916,
      "step": 5624
    },
    {
      "epoch": 0.16207572177721433,
      "grad_norm": 1.1527348756790161,
      "learning_rate": 9.54962143622967e-06,
      "loss": 0.7849,
      "step": 5625
    },
    {
      "epoch": 0.1621045352388636,
      "grad_norm": 1.176437258720398,
      "learning_rate": 9.549427878490168e-06,
      "loss": 0.7688,
      "step": 5626
    },
    {
      "epoch": 0.16213334870051288,
      "grad_norm": 1.1603785753250122,
      "learning_rate": 9.549234281129696e-06,
      "loss": 0.7809,
      "step": 5627
    },
    {
      "epoch": 0.16216216216216217,
      "grad_norm": 1.1623084545135498,
      "learning_rate": 9.549040644149943e-06,
      "loss": 0.8037,
      "step": 5628
    },
    {
      "epoch": 0.16219097562381143,
      "grad_norm": 1.1290919780731201,
      "learning_rate": 9.548846967552591e-06,
      "loss": 0.7738,
      "step": 5629
    },
    {
      "epoch": 0.16221978908546072,
      "grad_norm": 1.120017170906067,
      "learning_rate": 9.548653251339331e-06,
      "loss": 0.812,
      "step": 5630
    },
    {
      "epoch": 0.16224860254711002,
      "grad_norm": 1.121747612953186,
      "learning_rate": 9.548459495511848e-06,
      "loss": 0.8206,
      "step": 5631
    },
    {
      "epoch": 0.16227741600875928,
      "grad_norm": 1.107342004776001,
      "learning_rate": 9.548265700071829e-06,
      "loss": 0.7794,
      "step": 5632
    },
    {
      "epoch": 0.16230622947040857,
      "grad_norm": 1.152077317237854,
      "learning_rate": 9.548071865020962e-06,
      "loss": 0.8076,
      "step": 5633
    },
    {
      "epoch": 0.16233504293205786,
      "grad_norm": 1.2630243301391602,
      "learning_rate": 9.547877990360936e-06,
      "loss": 0.809,
      "step": 5634
    },
    {
      "epoch": 0.16236385639370715,
      "grad_norm": 1.1849733591079712,
      "learning_rate": 9.547684076093438e-06,
      "loss": 0.7962,
      "step": 5635
    },
    {
      "epoch": 0.16239266985535641,
      "grad_norm": 1.0890196561813354,
      "learning_rate": 9.54749012222016e-06,
      "loss": 0.8116,
      "step": 5636
    },
    {
      "epoch": 0.1624214833170057,
      "grad_norm": 1.1987918615341187,
      "learning_rate": 9.547296128742787e-06,
      "loss": 0.8135,
      "step": 5637
    },
    {
      "epoch": 0.162450296778655,
      "grad_norm": 1.1616076231002808,
      "learning_rate": 9.54710209566301e-06,
      "loss": 0.7996,
      "step": 5638
    },
    {
      "epoch": 0.16247911024030426,
      "grad_norm": 1.1435787677764893,
      "learning_rate": 9.54690802298252e-06,
      "loss": 0.776,
      "step": 5639
    },
    {
      "epoch": 0.16250792370195355,
      "grad_norm": 1.1253111362457275,
      "learning_rate": 9.546713910703006e-06,
      "loss": 0.7733,
      "step": 5640
    },
    {
      "epoch": 0.16253673716360284,
      "grad_norm": 1.2671499252319336,
      "learning_rate": 9.546519758826161e-06,
      "loss": 0.8167,
      "step": 5641
    },
    {
      "epoch": 0.16256555062525213,
      "grad_norm": 1.0984708070755005,
      "learning_rate": 9.546325567353671e-06,
      "loss": 0.7838,
      "step": 5642
    },
    {
      "epoch": 0.1625943640869014,
      "grad_norm": 1.157585620880127,
      "learning_rate": 9.546131336287232e-06,
      "loss": 0.8037,
      "step": 5643
    },
    {
      "epoch": 0.16262317754855068,
      "grad_norm": 1.1821002960205078,
      "learning_rate": 9.54593706562853e-06,
      "loss": 0.78,
      "step": 5644
    },
    {
      "epoch": 0.16265199101019998,
      "grad_norm": 1.1608836650848389,
      "learning_rate": 9.545742755379263e-06,
      "loss": 0.8088,
      "step": 5645
    },
    {
      "epoch": 0.16268080447184924,
      "grad_norm": 1.1513241529464722,
      "learning_rate": 9.545548405541122e-06,
      "loss": 0.8219,
      "step": 5646
    },
    {
      "epoch": 0.16270961793349853,
      "grad_norm": 1.1665140390396118,
      "learning_rate": 9.545354016115797e-06,
      "loss": 0.7911,
      "step": 5647
    },
    {
      "epoch": 0.16273843139514782,
      "grad_norm": 1.1705988645553589,
      "learning_rate": 9.545159587104981e-06,
      "loss": 0.7935,
      "step": 5648
    },
    {
      "epoch": 0.16276724485679708,
      "grad_norm": 1.1190005540847778,
      "learning_rate": 9.54496511851037e-06,
      "loss": 0.7901,
      "step": 5649
    },
    {
      "epoch": 0.16279605831844637,
      "grad_norm": 1.1282883882522583,
      "learning_rate": 9.544770610333656e-06,
      "loss": 0.791,
      "step": 5650
    },
    {
      "epoch": 0.16282487178009566,
      "grad_norm": 1.1634148359298706,
      "learning_rate": 9.544576062576532e-06,
      "loss": 0.7931,
      "step": 5651
    },
    {
      "epoch": 0.16285368524174496,
      "grad_norm": 1.1910063028335571,
      "learning_rate": 9.544381475240694e-06,
      "loss": 0.7859,
      "step": 5652
    },
    {
      "epoch": 0.16288249870339422,
      "grad_norm": 1.1364717483520508,
      "learning_rate": 9.544186848327836e-06,
      "loss": 0.792,
      "step": 5653
    },
    {
      "epoch": 0.1629113121650435,
      "grad_norm": 1.1526975631713867,
      "learning_rate": 9.543992181839654e-06,
      "loss": 0.7936,
      "step": 5654
    },
    {
      "epoch": 0.1629401256266928,
      "grad_norm": 1.2295291423797607,
      "learning_rate": 9.543797475777842e-06,
      "loss": 0.8021,
      "step": 5655
    },
    {
      "epoch": 0.16296893908834206,
      "grad_norm": 1.1429774761199951,
      "learning_rate": 9.543602730144095e-06,
      "loss": 0.7976,
      "step": 5656
    },
    {
      "epoch": 0.16299775254999135,
      "grad_norm": 1.1828755140304565,
      "learning_rate": 9.543407944940112e-06,
      "loss": 0.7996,
      "step": 5657
    },
    {
      "epoch": 0.16302656601164064,
      "grad_norm": 1.2485885620117188,
      "learning_rate": 9.543213120167586e-06,
      "loss": 0.8006,
      "step": 5658
    },
    {
      "epoch": 0.1630553794732899,
      "grad_norm": 1.1191109418869019,
      "learning_rate": 9.543018255828215e-06,
      "loss": 0.797,
      "step": 5659
    },
    {
      "epoch": 0.1630841929349392,
      "grad_norm": 1.2090953588485718,
      "learning_rate": 9.542823351923698e-06,
      "loss": 0.81,
      "step": 5660
    },
    {
      "epoch": 0.1631130063965885,
      "grad_norm": 1.1688917875289917,
      "learning_rate": 9.54262840845573e-06,
      "loss": 0.7966,
      "step": 5661
    },
    {
      "epoch": 0.16314181985823778,
      "grad_norm": 1.173762321472168,
      "learning_rate": 9.54243342542601e-06,
      "loss": 0.7983,
      "step": 5662
    },
    {
      "epoch": 0.16317063331988704,
      "grad_norm": 1.2541265487670898,
      "learning_rate": 9.542238402836234e-06,
      "loss": 0.7926,
      "step": 5663
    },
    {
      "epoch": 0.16319944678153633,
      "grad_norm": 1.1762715578079224,
      "learning_rate": 9.542043340688104e-06,
      "loss": 0.8105,
      "step": 5664
    },
    {
      "epoch": 0.16322826024318562,
      "grad_norm": 1.1851003170013428,
      "learning_rate": 9.541848238983316e-06,
      "loss": 0.8226,
      "step": 5665
    },
    {
      "epoch": 0.1632570737048349,
      "grad_norm": 1.1934934854507446,
      "learning_rate": 9.54165309772357e-06,
      "loss": 0.7851,
      "step": 5666
    },
    {
      "epoch": 0.16328588716648418,
      "grad_norm": 1.1901195049285889,
      "learning_rate": 9.541457916910566e-06,
      "loss": 0.7842,
      "step": 5667
    },
    {
      "epoch": 0.16331470062813347,
      "grad_norm": 1.1523898839950562,
      "learning_rate": 9.541262696546003e-06,
      "loss": 0.7743,
      "step": 5668
    },
    {
      "epoch": 0.16334351408978276,
      "grad_norm": 1.1978734731674194,
      "learning_rate": 9.541067436631582e-06,
      "loss": 0.7791,
      "step": 5669
    },
    {
      "epoch": 0.16337232755143202,
      "grad_norm": 1.1653274297714233,
      "learning_rate": 9.540872137169002e-06,
      "loss": 0.7863,
      "step": 5670
    },
    {
      "epoch": 0.1634011410130813,
      "grad_norm": 1.1365758180618286,
      "learning_rate": 9.540676798159965e-06,
      "loss": 0.7855,
      "step": 5671
    },
    {
      "epoch": 0.1634299544747306,
      "grad_norm": 1.16166090965271,
      "learning_rate": 9.540481419606172e-06,
      "loss": 0.7902,
      "step": 5672
    },
    {
      "epoch": 0.16345876793637987,
      "grad_norm": 1.210842251777649,
      "learning_rate": 9.540286001509325e-06,
      "loss": 0.773,
      "step": 5673
    },
    {
      "epoch": 0.16348758139802916,
      "grad_norm": 1.1385927200317383,
      "learning_rate": 9.540090543871127e-06,
      "loss": 0.7742,
      "step": 5674
    },
    {
      "epoch": 0.16351639485967845,
      "grad_norm": 1.0856188535690308,
      "learning_rate": 9.539895046693277e-06,
      "loss": 0.7996,
      "step": 5675
    },
    {
      "epoch": 0.1635452083213277,
      "grad_norm": 1.1019423007965088,
      "learning_rate": 9.53969950997748e-06,
      "loss": 0.7716,
      "step": 5676
    },
    {
      "epoch": 0.163574021782977,
      "grad_norm": 1.1660045385360718,
      "learning_rate": 9.539503933725438e-06,
      "loss": 0.7624,
      "step": 5677
    },
    {
      "epoch": 0.1636028352446263,
      "grad_norm": 1.142539381980896,
      "learning_rate": 9.539308317938855e-06,
      "loss": 0.7934,
      "step": 5678
    },
    {
      "epoch": 0.16363164870627558,
      "grad_norm": 1.0632177591323853,
      "learning_rate": 9.539112662619434e-06,
      "loss": 0.7979,
      "step": 5679
    },
    {
      "epoch": 0.16366046216792485,
      "grad_norm": 1.1522916555404663,
      "learning_rate": 9.53891696776888e-06,
      "loss": 0.7741,
      "step": 5680
    },
    {
      "epoch": 0.16368927562957414,
      "grad_norm": 1.0884456634521484,
      "learning_rate": 9.538721233388895e-06,
      "loss": 0.7786,
      "step": 5681
    },
    {
      "epoch": 0.16371808909122343,
      "grad_norm": 1.163912057876587,
      "learning_rate": 9.538525459481185e-06,
      "loss": 0.8098,
      "step": 5682
    },
    {
      "epoch": 0.1637469025528727,
      "grad_norm": 1.076608419418335,
      "learning_rate": 9.538329646047456e-06,
      "loss": 0.786,
      "step": 5683
    },
    {
      "epoch": 0.16377571601452198,
      "grad_norm": 1.127402663230896,
      "learning_rate": 9.538133793089413e-06,
      "loss": 0.7729,
      "step": 5684
    },
    {
      "epoch": 0.16380452947617127,
      "grad_norm": 1.1965532302856445,
      "learning_rate": 9.53793790060876e-06,
      "loss": 0.7777,
      "step": 5685
    },
    {
      "epoch": 0.16383334293782054,
      "grad_norm": 1.1220598220825195,
      "learning_rate": 9.537741968607204e-06,
      "loss": 0.7926,
      "step": 5686
    },
    {
      "epoch": 0.16386215639946983,
      "grad_norm": 1.1886106729507446,
      "learning_rate": 9.537545997086453e-06,
      "loss": 0.8077,
      "step": 5687
    },
    {
      "epoch": 0.16389096986111912,
      "grad_norm": 1.1089613437652588,
      "learning_rate": 9.53734998604821e-06,
      "loss": 0.7805,
      "step": 5688
    },
    {
      "epoch": 0.1639197833227684,
      "grad_norm": 1.0814517736434937,
      "learning_rate": 9.537153935494186e-06,
      "loss": 0.787,
      "step": 5689
    },
    {
      "epoch": 0.16394859678441767,
      "grad_norm": 1.0793501138687134,
      "learning_rate": 9.536957845426086e-06,
      "loss": 0.7881,
      "step": 5690
    },
    {
      "epoch": 0.16397741024606696,
      "grad_norm": 1.1300569772720337,
      "learning_rate": 9.53676171584562e-06,
      "loss": 0.7948,
      "step": 5691
    },
    {
      "epoch": 0.16400622370771625,
      "grad_norm": 1.0473397970199585,
      "learning_rate": 9.536565546754493e-06,
      "loss": 0.7708,
      "step": 5692
    },
    {
      "epoch": 0.16403503716936552,
      "grad_norm": 1.1419278383255005,
      "learning_rate": 9.536369338154415e-06,
      "loss": 0.767,
      "step": 5693
    },
    {
      "epoch": 0.1640638506310148,
      "grad_norm": 1.0858125686645508,
      "learning_rate": 9.536173090047097e-06,
      "loss": 0.7891,
      "step": 5694
    },
    {
      "epoch": 0.1640926640926641,
      "grad_norm": 1.126421570777893,
      "learning_rate": 9.535976802434243e-06,
      "loss": 0.7784,
      "step": 5695
    },
    {
      "epoch": 0.1641214775543134,
      "grad_norm": 1.1483800411224365,
      "learning_rate": 9.535780475317566e-06,
      "loss": 0.8072,
      "step": 5696
    },
    {
      "epoch": 0.16415029101596265,
      "grad_norm": 1.0397617816925049,
      "learning_rate": 9.535584108698776e-06,
      "loss": 0.7857,
      "step": 5697
    },
    {
      "epoch": 0.16417910447761194,
      "grad_norm": 1.0886527299880981,
      "learning_rate": 9.535387702579582e-06,
      "loss": 0.7916,
      "step": 5698
    },
    {
      "epoch": 0.16420791793926123,
      "grad_norm": 1.138509750366211,
      "learning_rate": 9.535191256961695e-06,
      "loss": 0.7929,
      "step": 5699
    },
    {
      "epoch": 0.1642367314009105,
      "grad_norm": 1.7932041883468628,
      "learning_rate": 9.534994771846825e-06,
      "loss": 0.8073,
      "step": 5700
    },
    {
      "epoch": 0.1642655448625598,
      "grad_norm": 1.148826241493225,
      "learning_rate": 9.534798247236685e-06,
      "loss": 0.8052,
      "step": 5701
    },
    {
      "epoch": 0.16429435832420908,
      "grad_norm": 1.1769853830337524,
      "learning_rate": 9.534601683132985e-06,
      "loss": 0.7983,
      "step": 5702
    },
    {
      "epoch": 0.16432317178585834,
      "grad_norm": 1.1739912033081055,
      "learning_rate": 9.534405079537437e-06,
      "loss": 0.8182,
      "step": 5703
    },
    {
      "epoch": 0.16435198524750763,
      "grad_norm": 1.0536558628082275,
      "learning_rate": 9.534208436451752e-06,
      "loss": 0.7958,
      "step": 5704
    },
    {
      "epoch": 0.16438079870915692,
      "grad_norm": 1.1507959365844727,
      "learning_rate": 9.534011753877645e-06,
      "loss": 0.8251,
      "step": 5705
    },
    {
      "epoch": 0.1644096121708062,
      "grad_norm": 1.1211669445037842,
      "learning_rate": 9.53381503181683e-06,
      "loss": 0.7999,
      "step": 5706
    },
    {
      "epoch": 0.16443842563245548,
      "grad_norm": 1.0832289457321167,
      "learning_rate": 9.533618270271016e-06,
      "loss": 0.7993,
      "step": 5707
    },
    {
      "epoch": 0.16446723909410477,
      "grad_norm": 1.1381855010986328,
      "learning_rate": 9.533421469241921e-06,
      "loss": 0.8023,
      "step": 5708
    },
    {
      "epoch": 0.16449605255575406,
      "grad_norm": 1.1210784912109375,
      "learning_rate": 9.533224628731255e-06,
      "loss": 0.7858,
      "step": 5709
    },
    {
      "epoch": 0.16452486601740332,
      "grad_norm": 1.0929831266403198,
      "learning_rate": 9.533027748740734e-06,
      "loss": 0.7862,
      "step": 5710
    },
    {
      "epoch": 0.1645536794790526,
      "grad_norm": 1.1329212188720703,
      "learning_rate": 9.532830829272072e-06,
      "loss": 0.798,
      "step": 5711
    },
    {
      "epoch": 0.1645824929407019,
      "grad_norm": 1.0914796590805054,
      "learning_rate": 9.532633870326988e-06,
      "loss": 0.8039,
      "step": 5712
    },
    {
      "epoch": 0.16461130640235117,
      "grad_norm": 1.0915732383728027,
      "learning_rate": 9.532436871907192e-06,
      "loss": 0.7876,
      "step": 5713
    },
    {
      "epoch": 0.16464011986400046,
      "grad_norm": 1.1315449476242065,
      "learning_rate": 9.5322398340144e-06,
      "loss": 0.8172,
      "step": 5714
    },
    {
      "epoch": 0.16466893332564975,
      "grad_norm": 1.0771851539611816,
      "learning_rate": 9.53204275665033e-06,
      "loss": 0.7947,
      "step": 5715
    },
    {
      "epoch": 0.16469774678729904,
      "grad_norm": 1.1471680402755737,
      "learning_rate": 9.5318456398167e-06,
      "loss": 0.8034,
      "step": 5716
    },
    {
      "epoch": 0.1647265602489483,
      "grad_norm": 1.1373038291931152,
      "learning_rate": 9.531648483515224e-06,
      "loss": 0.7781,
      "step": 5717
    },
    {
      "epoch": 0.1647553737105976,
      "grad_norm": 1.123680591583252,
      "learning_rate": 9.53145128774762e-06,
      "loss": 0.7732,
      "step": 5718
    },
    {
      "epoch": 0.16478418717224688,
      "grad_norm": 1.1302528381347656,
      "learning_rate": 9.531254052515605e-06,
      "loss": 0.7942,
      "step": 5719
    },
    {
      "epoch": 0.16481300063389615,
      "grad_norm": 1.1328856945037842,
      "learning_rate": 9.531056777820895e-06,
      "loss": 0.7926,
      "step": 5720
    },
    {
      "epoch": 0.16484181409554544,
      "grad_norm": 1.085838794708252,
      "learning_rate": 9.53085946366521e-06,
      "loss": 0.7905,
      "step": 5721
    },
    {
      "epoch": 0.16487062755719473,
      "grad_norm": 1.119705080986023,
      "learning_rate": 9.53066211005027e-06,
      "loss": 0.7802,
      "step": 5722
    },
    {
      "epoch": 0.16489944101884402,
      "grad_norm": 1.1225818395614624,
      "learning_rate": 9.530464716977792e-06,
      "loss": 0.8049,
      "step": 5723
    },
    {
      "epoch": 0.16492825448049328,
      "grad_norm": 1.3983094692230225,
      "learning_rate": 9.530267284449495e-06,
      "loss": 0.8112,
      "step": 5724
    },
    {
      "epoch": 0.16495706794214257,
      "grad_norm": 1.13595712184906,
      "learning_rate": 9.530069812467097e-06,
      "loss": 0.8116,
      "step": 5725
    },
    {
      "epoch": 0.16498588140379186,
      "grad_norm": 1.1105374097824097,
      "learning_rate": 9.529872301032321e-06,
      "loss": 0.79,
      "step": 5726
    },
    {
      "epoch": 0.16501469486544112,
      "grad_norm": 1.1135438680648804,
      "learning_rate": 9.529674750146885e-06,
      "loss": 0.772,
      "step": 5727
    },
    {
      "epoch": 0.16504350832709042,
      "grad_norm": 1.861522912979126,
      "learning_rate": 9.52947715981251e-06,
      "loss": 0.7897,
      "step": 5728
    },
    {
      "epoch": 0.1650723217887397,
      "grad_norm": 1.1308062076568604,
      "learning_rate": 9.529279530030915e-06,
      "loss": 0.7811,
      "step": 5729
    },
    {
      "epoch": 0.16510113525038897,
      "grad_norm": 1.1127907037734985,
      "learning_rate": 9.529081860803826e-06,
      "loss": 0.7955,
      "step": 5730
    },
    {
      "epoch": 0.16512994871203826,
      "grad_norm": 1.0988585948944092,
      "learning_rate": 9.52888415213296e-06,
      "loss": 0.7837,
      "step": 5731
    },
    {
      "epoch": 0.16515876217368755,
      "grad_norm": 1.2008373737335205,
      "learning_rate": 9.52868640402004e-06,
      "loss": 0.795,
      "step": 5732
    },
    {
      "epoch": 0.16518757563533684,
      "grad_norm": 1.1227569580078125,
      "learning_rate": 9.528488616466789e-06,
      "loss": 0.7937,
      "step": 5733
    },
    {
      "epoch": 0.1652163890969861,
      "grad_norm": 1.134036660194397,
      "learning_rate": 9.528290789474928e-06,
      "loss": 0.7894,
      "step": 5734
    },
    {
      "epoch": 0.1652452025586354,
      "grad_norm": 1.1191235780715942,
      "learning_rate": 9.528092923046181e-06,
      "loss": 0.7855,
      "step": 5735
    },
    {
      "epoch": 0.1652740160202847,
      "grad_norm": 1.1529525518417358,
      "learning_rate": 9.527895017182272e-06,
      "loss": 0.7955,
      "step": 5736
    },
    {
      "epoch": 0.16530282948193395,
      "grad_norm": 1.1595064401626587,
      "learning_rate": 9.527697071884922e-06,
      "loss": 0.7878,
      "step": 5737
    },
    {
      "epoch": 0.16533164294358324,
      "grad_norm": 1.1835383176803589,
      "learning_rate": 9.527499087155859e-06,
      "loss": 0.8043,
      "step": 5738
    },
    {
      "epoch": 0.16536045640523253,
      "grad_norm": 1.1212676763534546,
      "learning_rate": 9.527301062996804e-06,
      "loss": 0.7802,
      "step": 5739
    },
    {
      "epoch": 0.1653892698668818,
      "grad_norm": 1.100937008857727,
      "learning_rate": 9.527102999409481e-06,
      "loss": 0.7857,
      "step": 5740
    },
    {
      "epoch": 0.16541808332853108,
      "grad_norm": 1.1353132724761963,
      "learning_rate": 9.526904896395616e-06,
      "loss": 0.7763,
      "step": 5741
    },
    {
      "epoch": 0.16544689679018038,
      "grad_norm": 1.1135175228118896,
      "learning_rate": 9.526706753956936e-06,
      "loss": 0.8022,
      "step": 5742
    },
    {
      "epoch": 0.16547571025182967,
      "grad_norm": 1.1503369808197021,
      "learning_rate": 9.526508572095166e-06,
      "loss": 0.8113,
      "step": 5743
    },
    {
      "epoch": 0.16550452371347893,
      "grad_norm": 1.1613901853561401,
      "learning_rate": 9.52631035081203e-06,
      "loss": 0.7775,
      "step": 5744
    },
    {
      "epoch": 0.16553333717512822,
      "grad_norm": 1.1251426935195923,
      "learning_rate": 9.526112090109257e-06,
      "loss": 0.7872,
      "step": 5745
    },
    {
      "epoch": 0.1655621506367775,
      "grad_norm": 1.1397968530654907,
      "learning_rate": 9.52591378998857e-06,
      "loss": 0.8089,
      "step": 5746
    },
    {
      "epoch": 0.16559096409842677,
      "grad_norm": 1.1629970073699951,
      "learning_rate": 9.5257154504517e-06,
      "loss": 0.8151,
      "step": 5747
    },
    {
      "epoch": 0.16561977756007606,
      "grad_norm": 1.195797085762024,
      "learning_rate": 9.525517071500369e-06,
      "loss": 0.8041,
      "step": 5748
    },
    {
      "epoch": 0.16564859102172536,
      "grad_norm": 1.102929711341858,
      "learning_rate": 9.525318653136312e-06,
      "loss": 0.7953,
      "step": 5749
    },
    {
      "epoch": 0.16567740448337465,
      "grad_norm": 1.119665503501892,
      "learning_rate": 9.525120195361252e-06,
      "loss": 0.797,
      "step": 5750
    },
    {
      "epoch": 0.1657062179450239,
      "grad_norm": 1.145940899848938,
      "learning_rate": 9.524921698176919e-06,
      "loss": 0.7956,
      "step": 5751
    },
    {
      "epoch": 0.1657350314066732,
      "grad_norm": 1.0980325937271118,
      "learning_rate": 9.524723161585042e-06,
      "loss": 0.7959,
      "step": 5752
    },
    {
      "epoch": 0.1657638448683225,
      "grad_norm": 1.1173237562179565,
      "learning_rate": 9.524524585587347e-06,
      "loss": 0.81,
      "step": 5753
    },
    {
      "epoch": 0.16579265832997175,
      "grad_norm": 1.160927176475525,
      "learning_rate": 9.524325970185566e-06,
      "loss": 0.7823,
      "step": 5754
    },
    {
      "epoch": 0.16582147179162104,
      "grad_norm": 1.118488073348999,
      "learning_rate": 9.52412731538143e-06,
      "loss": 0.8007,
      "step": 5755
    },
    {
      "epoch": 0.16585028525327034,
      "grad_norm": 1.0749311447143555,
      "learning_rate": 9.523928621176666e-06,
      "loss": 0.7912,
      "step": 5756
    },
    {
      "epoch": 0.1658790987149196,
      "grad_norm": 1.1675323247909546,
      "learning_rate": 9.523729887573007e-06,
      "loss": 0.7767,
      "step": 5757
    },
    {
      "epoch": 0.1659079121765689,
      "grad_norm": 1.1422491073608398,
      "learning_rate": 9.523531114572181e-06,
      "loss": 0.7859,
      "step": 5758
    },
    {
      "epoch": 0.16593672563821818,
      "grad_norm": 1.0346943140029907,
      "learning_rate": 9.523332302175924e-06,
      "loss": 0.762,
      "step": 5759
    },
    {
      "epoch": 0.16596553909986747,
      "grad_norm": 1.112925410270691,
      "learning_rate": 9.523133450385962e-06,
      "loss": 0.8,
      "step": 5760
    },
    {
      "epoch": 0.16599435256151673,
      "grad_norm": 1.1345198154449463,
      "learning_rate": 9.522934559204031e-06,
      "loss": 0.7846,
      "step": 5761
    },
    {
      "epoch": 0.16602316602316602,
      "grad_norm": 1.1002485752105713,
      "learning_rate": 9.52273562863186e-06,
      "loss": 0.7767,
      "step": 5762
    },
    {
      "epoch": 0.16605197948481532,
      "grad_norm": 1.108354091644287,
      "learning_rate": 9.522536658671183e-06,
      "loss": 0.7768,
      "step": 5763
    },
    {
      "epoch": 0.16608079294646458,
      "grad_norm": 1.1107550859451294,
      "learning_rate": 9.522337649323733e-06,
      "loss": 0.7858,
      "step": 5764
    },
    {
      "epoch": 0.16610960640811387,
      "grad_norm": 1.1342074871063232,
      "learning_rate": 9.522138600591243e-06,
      "loss": 0.7941,
      "step": 5765
    },
    {
      "epoch": 0.16613841986976316,
      "grad_norm": 1.073162317276001,
      "learning_rate": 9.521939512475444e-06,
      "loss": 0.7956,
      "step": 5766
    },
    {
      "epoch": 0.16616723333141242,
      "grad_norm": 1.1020770072937012,
      "learning_rate": 9.521740384978074e-06,
      "loss": 0.77,
      "step": 5767
    },
    {
      "epoch": 0.1661960467930617,
      "grad_norm": 1.1538364887237549,
      "learning_rate": 9.521541218100864e-06,
      "loss": 0.8056,
      "step": 5768
    },
    {
      "epoch": 0.166224860254711,
      "grad_norm": 1.1419596672058105,
      "learning_rate": 9.52134201184555e-06,
      "loss": 0.8281,
      "step": 5769
    },
    {
      "epoch": 0.1662536737163603,
      "grad_norm": 1.1567561626434326,
      "learning_rate": 9.521142766213869e-06,
      "loss": 0.8016,
      "step": 5770
    },
    {
      "epoch": 0.16628248717800956,
      "grad_norm": 1.1148990392684937,
      "learning_rate": 9.520943481207552e-06,
      "loss": 0.7805,
      "step": 5771
    },
    {
      "epoch": 0.16631130063965885,
      "grad_norm": 1.1466270685195923,
      "learning_rate": 9.520744156828337e-06,
      "loss": 0.7928,
      "step": 5772
    },
    {
      "epoch": 0.16634011410130814,
      "grad_norm": 1.1703834533691406,
      "learning_rate": 9.52054479307796e-06,
      "loss": 0.7826,
      "step": 5773
    },
    {
      "epoch": 0.1663689275629574,
      "grad_norm": 1.0641072988510132,
      "learning_rate": 9.520345389958157e-06,
      "loss": 0.8048,
      "step": 5774
    },
    {
      "epoch": 0.1663977410246067,
      "grad_norm": 1.1535828113555908,
      "learning_rate": 9.520145947470662e-06,
      "loss": 0.7901,
      "step": 5775
    },
    {
      "epoch": 0.16642655448625598,
      "grad_norm": 1.1417316198349,
      "learning_rate": 9.519946465617217e-06,
      "loss": 0.7882,
      "step": 5776
    },
    {
      "epoch": 0.16645536794790528,
      "grad_norm": 1.1390228271484375,
      "learning_rate": 9.519746944399556e-06,
      "loss": 0.8013,
      "step": 5777
    },
    {
      "epoch": 0.16648418140955454,
      "grad_norm": 1.1681710481643677,
      "learning_rate": 9.519547383819417e-06,
      "loss": 0.7899,
      "step": 5778
    },
    {
      "epoch": 0.16651299487120383,
      "grad_norm": 1.078402042388916,
      "learning_rate": 9.519347783878539e-06,
      "loss": 0.799,
      "step": 5779
    },
    {
      "epoch": 0.16654180833285312,
      "grad_norm": 1.1502617597579956,
      "learning_rate": 9.519148144578658e-06,
      "loss": 0.7825,
      "step": 5780
    },
    {
      "epoch": 0.16657062179450238,
      "grad_norm": 1.0913571119308472,
      "learning_rate": 9.518948465921514e-06,
      "loss": 0.8017,
      "step": 5781
    },
    {
      "epoch": 0.16659943525615167,
      "grad_norm": 1.1169979572296143,
      "learning_rate": 9.518748747908848e-06,
      "loss": 0.791,
      "step": 5782
    },
    {
      "epoch": 0.16662824871780096,
      "grad_norm": 1.130846619606018,
      "learning_rate": 9.518548990542395e-06,
      "loss": 0.8036,
      "step": 5783
    },
    {
      "epoch": 0.16665706217945023,
      "grad_norm": 1.117244839668274,
      "learning_rate": 9.518349193823899e-06,
      "loss": 0.7635,
      "step": 5784
    },
    {
      "epoch": 0.16668587564109952,
      "grad_norm": 1.0891883373260498,
      "learning_rate": 9.518149357755097e-06,
      "loss": 0.7976,
      "step": 5785
    },
    {
      "epoch": 0.1667146891027488,
      "grad_norm": 1.1862374544143677,
      "learning_rate": 9.517949482337731e-06,
      "loss": 0.8119,
      "step": 5786
    },
    {
      "epoch": 0.1667435025643981,
      "grad_norm": 1.0671627521514893,
      "learning_rate": 9.517749567573542e-06,
      "loss": 0.8115,
      "step": 5787
    },
    {
      "epoch": 0.16677231602604736,
      "grad_norm": 1.200798749923706,
      "learning_rate": 9.51754961346427e-06,
      "loss": 0.7891,
      "step": 5788
    },
    {
      "epoch": 0.16680112948769665,
      "grad_norm": 1.1641258001327515,
      "learning_rate": 9.517349620011658e-06,
      "loss": 0.7814,
      "step": 5789
    },
    {
      "epoch": 0.16682994294934594,
      "grad_norm": 1.0741147994995117,
      "learning_rate": 9.517149587217444e-06,
      "loss": 0.7744,
      "step": 5790
    },
    {
      "epoch": 0.1668587564109952,
      "grad_norm": 1.1633946895599365,
      "learning_rate": 9.516949515083374e-06,
      "loss": 0.8151,
      "step": 5791
    },
    {
      "epoch": 0.1668875698726445,
      "grad_norm": 1.1921170949935913,
      "learning_rate": 9.51674940361119e-06,
      "loss": 0.8114,
      "step": 5792
    },
    {
      "epoch": 0.1669163833342938,
      "grad_norm": 1.0848066806793213,
      "learning_rate": 9.516549252802631e-06,
      "loss": 0.7891,
      "step": 5793
    },
    {
      "epoch": 0.16694519679594305,
      "grad_norm": 1.0493602752685547,
      "learning_rate": 9.516349062659445e-06,
      "loss": 0.773,
      "step": 5794
    },
    {
      "epoch": 0.16697401025759234,
      "grad_norm": 1.0884370803833008,
      "learning_rate": 9.516148833183372e-06,
      "loss": 0.7877,
      "step": 5795
    },
    {
      "epoch": 0.16700282371924163,
      "grad_norm": 1.1353216171264648,
      "learning_rate": 9.515948564376158e-06,
      "loss": 0.7886,
      "step": 5796
    },
    {
      "epoch": 0.16703163718089092,
      "grad_norm": 1.0149048566818237,
      "learning_rate": 9.515748256239548e-06,
      "loss": 0.7886,
      "step": 5797
    },
    {
      "epoch": 0.1670604506425402,
      "grad_norm": 1.1040219068527222,
      "learning_rate": 9.515547908775282e-06,
      "loss": 0.7902,
      "step": 5798
    },
    {
      "epoch": 0.16708926410418948,
      "grad_norm": 1.0955814123153687,
      "learning_rate": 9.51534752198511e-06,
      "loss": 0.7927,
      "step": 5799
    },
    {
      "epoch": 0.16711807756583877,
      "grad_norm": 1.064710259437561,
      "learning_rate": 9.515147095870774e-06,
      "loss": 0.7955,
      "step": 5800
    },
    {
      "epoch": 0.16714689102748803,
      "grad_norm": 1.0887113809585571,
      "learning_rate": 9.514946630434019e-06,
      "loss": 0.7937,
      "step": 5801
    },
    {
      "epoch": 0.16717570448913732,
      "grad_norm": 1.118260383605957,
      "learning_rate": 9.514746125676592e-06,
      "loss": 0.787,
      "step": 5802
    },
    {
      "epoch": 0.1672045179507866,
      "grad_norm": 1.0608820915222168,
      "learning_rate": 9.51454558160024e-06,
      "loss": 0.793,
      "step": 5803
    },
    {
      "epoch": 0.1672333314124359,
      "grad_norm": 1.0926570892333984,
      "learning_rate": 9.51434499820671e-06,
      "loss": 0.7753,
      "step": 5804
    },
    {
      "epoch": 0.16726214487408517,
      "grad_norm": 1.136027455329895,
      "learning_rate": 9.514144375497747e-06,
      "loss": 0.8104,
      "step": 5805
    },
    {
      "epoch": 0.16729095833573446,
      "grad_norm": 1.0758503675460815,
      "learning_rate": 9.5139437134751e-06,
      "loss": 0.7583,
      "step": 5806
    },
    {
      "epoch": 0.16731977179738375,
      "grad_norm": 1.1229292154312134,
      "learning_rate": 9.513743012140514e-06,
      "loss": 0.771,
      "step": 5807
    },
    {
      "epoch": 0.167348585259033,
      "grad_norm": 1.1068216562271118,
      "learning_rate": 9.51354227149574e-06,
      "loss": 0.779,
      "step": 5808
    },
    {
      "epoch": 0.1673773987206823,
      "grad_norm": 1.0936179161071777,
      "learning_rate": 9.513341491542523e-06,
      "loss": 0.7746,
      "step": 5809
    },
    {
      "epoch": 0.1674062121823316,
      "grad_norm": 1.1032503843307495,
      "learning_rate": 9.513140672282613e-06,
      "loss": 0.7728,
      "step": 5810
    },
    {
      "epoch": 0.16743502564398086,
      "grad_norm": 1.110113263130188,
      "learning_rate": 9.51293981371776e-06,
      "loss": 0.7829,
      "step": 5811
    },
    {
      "epoch": 0.16746383910563015,
      "grad_norm": 1.085000991821289,
      "learning_rate": 9.512738915849712e-06,
      "loss": 0.8159,
      "step": 5812
    },
    {
      "epoch": 0.16749265256727944,
      "grad_norm": 1.0775972604751587,
      "learning_rate": 9.512537978680221e-06,
      "loss": 0.8208,
      "step": 5813
    },
    {
      "epoch": 0.16752146602892873,
      "grad_norm": 1.0870354175567627,
      "learning_rate": 9.512337002211034e-06,
      "loss": 0.7978,
      "step": 5814
    },
    {
      "epoch": 0.167550279490578,
      "grad_norm": 1.0784862041473389,
      "learning_rate": 9.512135986443902e-06,
      "loss": 0.7825,
      "step": 5815
    },
    {
      "epoch": 0.16757909295222728,
      "grad_norm": 1.0632976293563843,
      "learning_rate": 9.511934931380577e-06,
      "loss": 0.7943,
      "step": 5816
    },
    {
      "epoch": 0.16760790641387657,
      "grad_norm": 1.1406033039093018,
      "learning_rate": 9.511733837022808e-06,
      "loss": 0.7648,
      "step": 5817
    },
    {
      "epoch": 0.16763671987552584,
      "grad_norm": 1.1365842819213867,
      "learning_rate": 9.511532703372349e-06,
      "loss": 0.822,
      "step": 5818
    },
    {
      "epoch": 0.16766553333717513,
      "grad_norm": 1.1126751899719238,
      "learning_rate": 9.511331530430947e-06,
      "loss": 0.7987,
      "step": 5819
    },
    {
      "epoch": 0.16769434679882442,
      "grad_norm": 1.0834826231002808,
      "learning_rate": 9.51113031820036e-06,
      "loss": 0.803,
      "step": 5820
    },
    {
      "epoch": 0.16772316026047368,
      "grad_norm": 1.0969481468200684,
      "learning_rate": 9.510929066682335e-06,
      "loss": 0.8072,
      "step": 5821
    },
    {
      "epoch": 0.16775197372212297,
      "grad_norm": 1.1180232763290405,
      "learning_rate": 9.510727775878628e-06,
      "loss": 0.7896,
      "step": 5822
    },
    {
      "epoch": 0.16778078718377226,
      "grad_norm": 1.1493194103240967,
      "learning_rate": 9.510526445790993e-06,
      "loss": 0.8091,
      "step": 5823
    },
    {
      "epoch": 0.16780960064542155,
      "grad_norm": 1.0869050025939941,
      "learning_rate": 9.510325076421179e-06,
      "loss": 0.7982,
      "step": 5824
    },
    {
      "epoch": 0.16783841410707082,
      "grad_norm": 1.129452109336853,
      "learning_rate": 9.510123667770943e-06,
      "loss": 0.7984,
      "step": 5825
    },
    {
      "epoch": 0.1678672275687201,
      "grad_norm": 1.0724173784255981,
      "learning_rate": 9.50992221984204e-06,
      "loss": 0.7861,
      "step": 5826
    },
    {
      "epoch": 0.1678960410303694,
      "grad_norm": 1.1226699352264404,
      "learning_rate": 9.509720732636221e-06,
      "loss": 0.801,
      "step": 5827
    },
    {
      "epoch": 0.16792485449201866,
      "grad_norm": 1.077942132949829,
      "learning_rate": 9.509519206155242e-06,
      "loss": 0.7518,
      "step": 5828
    },
    {
      "epoch": 0.16795366795366795,
      "grad_norm": 1.052773356437683,
      "learning_rate": 9.50931764040086e-06,
      "loss": 0.7892,
      "step": 5829
    },
    {
      "epoch": 0.16798248141531724,
      "grad_norm": 1.14486825466156,
      "learning_rate": 9.509116035374827e-06,
      "loss": 0.7818,
      "step": 5830
    },
    {
      "epoch": 0.1680112948769665,
      "grad_norm": 1.13209867477417,
      "learning_rate": 9.508914391078903e-06,
      "loss": 0.7871,
      "step": 5831
    },
    {
      "epoch": 0.1680401083386158,
      "grad_norm": 1.0818114280700684,
      "learning_rate": 9.508712707514842e-06,
      "loss": 0.801,
      "step": 5832
    },
    {
      "epoch": 0.1680689218002651,
      "grad_norm": 1.1349148750305176,
      "learning_rate": 9.5085109846844e-06,
      "loss": 0.7864,
      "step": 5833
    },
    {
      "epoch": 0.16809773526191438,
      "grad_norm": 1.1463074684143066,
      "learning_rate": 9.508309222589333e-06,
      "loss": 0.8069,
      "step": 5834
    },
    {
      "epoch": 0.16812654872356364,
      "grad_norm": 1.1523492336273193,
      "learning_rate": 9.508107421231401e-06,
      "loss": 0.819,
      "step": 5835
    },
    {
      "epoch": 0.16815536218521293,
      "grad_norm": 6.398066520690918,
      "learning_rate": 9.507905580612357e-06,
      "loss": 0.7836,
      "step": 5836
    },
    {
      "epoch": 0.16818417564686222,
      "grad_norm": 1.111319661140442,
      "learning_rate": 9.507703700733964e-06,
      "loss": 0.7769,
      "step": 5837
    },
    {
      "epoch": 0.16821298910851148,
      "grad_norm": 1.2093312740325928,
      "learning_rate": 9.507501781597978e-06,
      "loss": 0.7761,
      "step": 5838
    },
    {
      "epoch": 0.16824180257016078,
      "grad_norm": 1.1880412101745605,
      "learning_rate": 9.507299823206156e-06,
      "loss": 0.7758,
      "step": 5839
    },
    {
      "epoch": 0.16827061603181007,
      "grad_norm": 1.2784970998764038,
      "learning_rate": 9.507097825560258e-06,
      "loss": 0.7753,
      "step": 5840
    },
    {
      "epoch": 0.16829942949345936,
      "grad_norm": 1.196534514427185,
      "learning_rate": 9.506895788662045e-06,
      "loss": 0.7877,
      "step": 5841
    },
    {
      "epoch": 0.16832824295510862,
      "grad_norm": 1.2190510034561157,
      "learning_rate": 9.506693712513276e-06,
      "loss": 0.8203,
      "step": 5842
    },
    {
      "epoch": 0.1683570564167579,
      "grad_norm": 1.207650899887085,
      "learning_rate": 9.506491597115707e-06,
      "loss": 0.8011,
      "step": 5843
    },
    {
      "epoch": 0.1683858698784072,
      "grad_norm": 1.1964874267578125,
      "learning_rate": 9.506289442471101e-06,
      "loss": 0.7804,
      "step": 5844
    },
    {
      "epoch": 0.16841468334005646,
      "grad_norm": 1.210526466369629,
      "learning_rate": 9.50608724858122e-06,
      "loss": 0.7609,
      "step": 5845
    },
    {
      "epoch": 0.16844349680170576,
      "grad_norm": 1.1834956407546997,
      "learning_rate": 9.505885015447825e-06,
      "loss": 0.8043,
      "step": 5846
    },
    {
      "epoch": 0.16847231026335505,
      "grad_norm": 1.1488529443740845,
      "learning_rate": 9.505682743072672e-06,
      "loss": 0.7937,
      "step": 5847
    },
    {
      "epoch": 0.1685011237250043,
      "grad_norm": 1.2011579275131226,
      "learning_rate": 9.505480431457527e-06,
      "loss": 0.8001,
      "step": 5848
    },
    {
      "epoch": 0.1685299371866536,
      "grad_norm": 1.2399260997772217,
      "learning_rate": 9.505278080604155e-06,
      "loss": 0.785,
      "step": 5849
    },
    {
      "epoch": 0.1685587506483029,
      "grad_norm": 1.1077483892440796,
      "learning_rate": 9.505075690514311e-06,
      "loss": 0.7864,
      "step": 5850
    },
    {
      "epoch": 0.16858756410995218,
      "grad_norm": 1.0822802782058716,
      "learning_rate": 9.504873261189762e-06,
      "loss": 0.773,
      "step": 5851
    },
    {
      "epoch": 0.16861637757160144,
      "grad_norm": 1.1558921337127686,
      "learning_rate": 9.504670792632272e-06,
      "loss": 0.807,
      "step": 5852
    },
    {
      "epoch": 0.16864519103325074,
      "grad_norm": 1.1035884618759155,
      "learning_rate": 9.5044682848436e-06,
      "loss": 0.7706,
      "step": 5853
    },
    {
      "epoch": 0.16867400449490003,
      "grad_norm": 1.174235463142395,
      "learning_rate": 9.504265737825512e-06,
      "loss": 0.782,
      "step": 5854
    },
    {
      "epoch": 0.1687028179565493,
      "grad_norm": 1.0749680995941162,
      "learning_rate": 9.504063151579773e-06,
      "loss": 0.7993,
      "step": 5855
    },
    {
      "epoch": 0.16873163141819858,
      "grad_norm": 1.1402302980422974,
      "learning_rate": 9.503860526108147e-06,
      "loss": 0.802,
      "step": 5856
    },
    {
      "epoch": 0.16876044487984787,
      "grad_norm": 1.2469806671142578,
      "learning_rate": 9.503657861412396e-06,
      "loss": 0.7808,
      "step": 5857
    },
    {
      "epoch": 0.16878925834149713,
      "grad_norm": 1.1221235990524292,
      "learning_rate": 9.503455157494289e-06,
      "loss": 0.799,
      "step": 5858
    },
    {
      "epoch": 0.16881807180314642,
      "grad_norm": 1.2115187644958496,
      "learning_rate": 9.503252414355588e-06,
      "loss": 0.8109,
      "step": 5859
    },
    {
      "epoch": 0.16884688526479572,
      "grad_norm": 1.0895003080368042,
      "learning_rate": 9.503049631998061e-06,
      "loss": 0.812,
      "step": 5860
    },
    {
      "epoch": 0.168875698726445,
      "grad_norm": 1.0975476503372192,
      "learning_rate": 9.502846810423474e-06,
      "loss": 0.7965,
      "step": 5861
    },
    {
      "epoch": 0.16890451218809427,
      "grad_norm": 1.153579592704773,
      "learning_rate": 9.50264394963359e-06,
      "loss": 0.7632,
      "step": 5862
    },
    {
      "epoch": 0.16893332564974356,
      "grad_norm": 1.0931141376495361,
      "learning_rate": 9.50244104963018e-06,
      "loss": 0.7932,
      "step": 5863
    },
    {
      "epoch": 0.16896213911139285,
      "grad_norm": 1.1154526472091675,
      "learning_rate": 9.502238110415008e-06,
      "loss": 0.7837,
      "step": 5864
    },
    {
      "epoch": 0.1689909525730421,
      "grad_norm": 1.1517298221588135,
      "learning_rate": 9.502035131989845e-06,
      "loss": 0.7907,
      "step": 5865
    },
    {
      "epoch": 0.1690197660346914,
      "grad_norm": 1.0894068479537964,
      "learning_rate": 9.501832114356455e-06,
      "loss": 0.7927,
      "step": 5866
    },
    {
      "epoch": 0.1690485794963407,
      "grad_norm": 1.1684637069702148,
      "learning_rate": 9.501629057516607e-06,
      "loss": 0.8065,
      "step": 5867
    },
    {
      "epoch": 0.16907739295798999,
      "grad_norm": 1.096903920173645,
      "learning_rate": 9.501425961472069e-06,
      "loss": 0.8182,
      "step": 5868
    },
    {
      "epoch": 0.16910620641963925,
      "grad_norm": 1.1294437646865845,
      "learning_rate": 9.501222826224612e-06,
      "loss": 0.7938,
      "step": 5869
    },
    {
      "epoch": 0.16913501988128854,
      "grad_norm": 1.1089847087860107,
      "learning_rate": 9.501019651776005e-06,
      "loss": 0.7962,
      "step": 5870
    },
    {
      "epoch": 0.16916383334293783,
      "grad_norm": 1.0687495470046997,
      "learning_rate": 9.500816438128013e-06,
      "loss": 0.7737,
      "step": 5871
    },
    {
      "epoch": 0.1691926468045871,
      "grad_norm": 1.145897626876831,
      "learning_rate": 9.500613185282412e-06,
      "loss": 0.8074,
      "step": 5872
    },
    {
      "epoch": 0.16922146026623638,
      "grad_norm": 1.1740626096725464,
      "learning_rate": 9.500409893240967e-06,
      "loss": 0.7951,
      "step": 5873
    },
    {
      "epoch": 0.16925027372788568,
      "grad_norm": 1.0862809419631958,
      "learning_rate": 9.500206562005451e-06,
      "loss": 0.7927,
      "step": 5874
    },
    {
      "epoch": 0.16927908718953494,
      "grad_norm": 1.1635411977767944,
      "learning_rate": 9.500003191577637e-06,
      "loss": 0.7976,
      "step": 5875
    },
    {
      "epoch": 0.16930790065118423,
      "grad_norm": 1.6343822479248047,
      "learning_rate": 9.49979978195929e-06,
      "loss": 0.7699,
      "step": 5876
    },
    {
      "epoch": 0.16933671411283352,
      "grad_norm": 1.168592095375061,
      "learning_rate": 9.499596333152187e-06,
      "loss": 0.8288,
      "step": 5877
    },
    {
      "epoch": 0.1693655275744828,
      "grad_norm": 1.1489375829696655,
      "learning_rate": 9.499392845158098e-06,
      "loss": 0.7886,
      "step": 5878
    },
    {
      "epoch": 0.16939434103613207,
      "grad_norm": 1.0705000162124634,
      "learning_rate": 9.499189317978795e-06,
      "loss": 0.7897,
      "step": 5879
    },
    {
      "epoch": 0.16942315449778136,
      "grad_norm": 1.1547476053237915,
      "learning_rate": 9.498985751616048e-06,
      "loss": 0.8071,
      "step": 5880
    },
    {
      "epoch": 0.16945196795943065,
      "grad_norm": 1.0878130197525024,
      "learning_rate": 9.498782146071635e-06,
      "loss": 0.7847,
      "step": 5881
    },
    {
      "epoch": 0.16948078142107992,
      "grad_norm": 1.1245695352554321,
      "learning_rate": 9.498578501347327e-06,
      "loss": 0.8165,
      "step": 5882
    },
    {
      "epoch": 0.1695095948827292,
      "grad_norm": 1.0921112298965454,
      "learning_rate": 9.498374817444895e-06,
      "loss": 0.7985,
      "step": 5883
    },
    {
      "epoch": 0.1695384083443785,
      "grad_norm": 1.1242390871047974,
      "learning_rate": 9.498171094366117e-06,
      "loss": 0.7952,
      "step": 5884
    },
    {
      "epoch": 0.16956722180602776,
      "grad_norm": 1.114686369895935,
      "learning_rate": 9.497967332112765e-06,
      "loss": 0.8076,
      "step": 5885
    },
    {
      "epoch": 0.16959603526767705,
      "grad_norm": 1.1035112142562866,
      "learning_rate": 9.497763530686612e-06,
      "loss": 0.809,
      "step": 5886
    },
    {
      "epoch": 0.16962484872932634,
      "grad_norm": 1.128991961479187,
      "learning_rate": 9.497559690089438e-06,
      "loss": 0.7661,
      "step": 5887
    },
    {
      "epoch": 0.16965366219097563,
      "grad_norm": 1.1245249509811401,
      "learning_rate": 9.497355810323012e-06,
      "loss": 0.8147,
      "step": 5888
    },
    {
      "epoch": 0.1696824756526249,
      "grad_norm": 1.0832573175430298,
      "learning_rate": 9.497151891389115e-06,
      "loss": 0.7827,
      "step": 5889
    },
    {
      "epoch": 0.1697112891142742,
      "grad_norm": 1.1796537637710571,
      "learning_rate": 9.49694793328952e-06,
      "loss": 0.808,
      "step": 5890
    },
    {
      "epoch": 0.16974010257592348,
      "grad_norm": 1.1093593835830688,
      "learning_rate": 9.496743936026002e-06,
      "loss": 0.8018,
      "step": 5891
    },
    {
      "epoch": 0.16976891603757274,
      "grad_norm": 1.080178141593933,
      "learning_rate": 9.496539899600342e-06,
      "loss": 0.7891,
      "step": 5892
    },
    {
      "epoch": 0.16979772949922203,
      "grad_norm": 1.0839940309524536,
      "learning_rate": 9.496335824014312e-06,
      "loss": 0.7847,
      "step": 5893
    },
    {
      "epoch": 0.16982654296087132,
      "grad_norm": 1.1291993856430054,
      "learning_rate": 9.496131709269693e-06,
      "loss": 0.7845,
      "step": 5894
    },
    {
      "epoch": 0.16985535642252061,
      "grad_norm": 1.0723246335983276,
      "learning_rate": 9.49592755536826e-06,
      "loss": 0.8108,
      "step": 5895
    },
    {
      "epoch": 0.16988416988416988,
      "grad_norm": 1.108231782913208,
      "learning_rate": 9.495723362311795e-06,
      "loss": 0.7898,
      "step": 5896
    },
    {
      "epoch": 0.16991298334581917,
      "grad_norm": 1.0882623195648193,
      "learning_rate": 9.495519130102071e-06,
      "loss": 0.8062,
      "step": 5897
    },
    {
      "epoch": 0.16994179680746846,
      "grad_norm": 1.0948301553726196,
      "learning_rate": 9.495314858740869e-06,
      "loss": 0.7783,
      "step": 5898
    },
    {
      "epoch": 0.16997061026911772,
      "grad_norm": 1.1168473958969116,
      "learning_rate": 9.495110548229972e-06,
      "loss": 0.7793,
      "step": 5899
    },
    {
      "epoch": 0.169999423730767,
      "grad_norm": 1.0951887369155884,
      "learning_rate": 9.494906198571153e-06,
      "loss": 0.795,
      "step": 5900
    },
    {
      "epoch": 0.1700282371924163,
      "grad_norm": 1.3512187004089355,
      "learning_rate": 9.494701809766193e-06,
      "loss": 0.7596,
      "step": 5901
    },
    {
      "epoch": 0.17005705065406557,
      "grad_norm": 1.0913985967636108,
      "learning_rate": 9.494497381816874e-06,
      "loss": 0.7831,
      "step": 5902
    },
    {
      "epoch": 0.17008586411571486,
      "grad_norm": 1.0602021217346191,
      "learning_rate": 9.494292914724976e-06,
      "loss": 0.78,
      "step": 5903
    },
    {
      "epoch": 0.17011467757736415,
      "grad_norm": 1.157479166984558,
      "learning_rate": 9.494088408492281e-06,
      "loss": 0.8111,
      "step": 5904
    },
    {
      "epoch": 0.17014349103901344,
      "grad_norm": 1.0898925065994263,
      "learning_rate": 9.493883863120565e-06,
      "loss": 0.797,
      "step": 5905
    },
    {
      "epoch": 0.1701723045006627,
      "grad_norm": 1.0831750631332397,
      "learning_rate": 9.493679278611616e-06,
      "loss": 0.7753,
      "step": 5906
    },
    {
      "epoch": 0.170201117962312,
      "grad_norm": 1.0800360441207886,
      "learning_rate": 9.49347465496721e-06,
      "loss": 0.7782,
      "step": 5907
    },
    {
      "epoch": 0.17022993142396128,
      "grad_norm": 1.089375376701355,
      "learning_rate": 9.493269992189134e-06,
      "loss": 0.792,
      "step": 5908
    },
    {
      "epoch": 0.17025874488561055,
      "grad_norm": 1.0527299642562866,
      "learning_rate": 9.493065290279167e-06,
      "loss": 0.7752,
      "step": 5909
    },
    {
      "epoch": 0.17028755834725984,
      "grad_norm": 1.0966938734054565,
      "learning_rate": 9.492860549239093e-06,
      "loss": 0.796,
      "step": 5910
    },
    {
      "epoch": 0.17031637180890913,
      "grad_norm": 1.0282665491104126,
      "learning_rate": 9.492655769070693e-06,
      "loss": 0.7903,
      "step": 5911
    },
    {
      "epoch": 0.1703451852705584,
      "grad_norm": 1.559958577156067,
      "learning_rate": 9.492450949775754e-06,
      "loss": 0.782,
      "step": 5912
    },
    {
      "epoch": 0.17037399873220768,
      "grad_norm": 1.0736429691314697,
      "learning_rate": 9.492246091356059e-06,
      "loss": 0.7756,
      "step": 5913
    },
    {
      "epoch": 0.17040281219385697,
      "grad_norm": 1.055188775062561,
      "learning_rate": 9.492041193813389e-06,
      "loss": 0.8028,
      "step": 5914
    },
    {
      "epoch": 0.17043162565550626,
      "grad_norm": 1.1111481189727783,
      "learning_rate": 9.491836257149531e-06,
      "loss": 0.7882,
      "step": 5915
    },
    {
      "epoch": 0.17046043911715553,
      "grad_norm": 1.0855711698532104,
      "learning_rate": 9.49163128136627e-06,
      "loss": 0.8038,
      "step": 5916
    },
    {
      "epoch": 0.17048925257880482,
      "grad_norm": 1.0650608539581299,
      "learning_rate": 9.491426266465391e-06,
      "loss": 0.8077,
      "step": 5917
    },
    {
      "epoch": 0.1705180660404541,
      "grad_norm": 1.1664133071899414,
      "learning_rate": 9.491221212448678e-06,
      "loss": 0.7669,
      "step": 5918
    },
    {
      "epoch": 0.17054687950210337,
      "grad_norm": 1.0479836463928223,
      "learning_rate": 9.491016119317918e-06,
      "loss": 0.7705,
      "step": 5919
    },
    {
      "epoch": 0.17057569296375266,
      "grad_norm": 1.0765585899353027,
      "learning_rate": 9.490810987074898e-06,
      "loss": 0.8241,
      "step": 5920
    },
    {
      "epoch": 0.17060450642540195,
      "grad_norm": 1.1454476118087769,
      "learning_rate": 9.490605815721403e-06,
      "loss": 0.8047,
      "step": 5921
    },
    {
      "epoch": 0.17063331988705124,
      "grad_norm": 1.0987759828567505,
      "learning_rate": 9.490400605259219e-06,
      "loss": 0.8077,
      "step": 5922
    },
    {
      "epoch": 0.1706621333487005,
      "grad_norm": 1.054229736328125,
      "learning_rate": 9.490195355690135e-06,
      "loss": 0.7841,
      "step": 5923
    },
    {
      "epoch": 0.1706909468103498,
      "grad_norm": 1.0271941423416138,
      "learning_rate": 9.489990067015939e-06,
      "loss": 0.7879,
      "step": 5924
    },
    {
      "epoch": 0.1707197602719991,
      "grad_norm": 1.0766680240631104,
      "learning_rate": 9.489784739238416e-06,
      "loss": 0.7817,
      "step": 5925
    },
    {
      "epoch": 0.17074857373364835,
      "grad_norm": 1.1000183820724487,
      "learning_rate": 9.489579372359359e-06,
      "loss": 0.791,
      "step": 5926
    },
    {
      "epoch": 0.17077738719529764,
      "grad_norm": 1.1095725297927856,
      "learning_rate": 9.48937396638055e-06,
      "loss": 0.7642,
      "step": 5927
    },
    {
      "epoch": 0.17080620065694693,
      "grad_norm": 1.038596510887146,
      "learning_rate": 9.489168521303784e-06,
      "loss": 0.7792,
      "step": 5928
    },
    {
      "epoch": 0.1708350141185962,
      "grad_norm": 1.1347466707229614,
      "learning_rate": 9.488963037130849e-06,
      "loss": 0.7795,
      "step": 5929
    },
    {
      "epoch": 0.1708638275802455,
      "grad_norm": 1.0435161590576172,
      "learning_rate": 9.48875751386353e-06,
      "loss": 0.7754,
      "step": 5930
    },
    {
      "epoch": 0.17089264104189478,
      "grad_norm": 1.1032472848892212,
      "learning_rate": 9.488551951503621e-06,
      "loss": 0.7834,
      "step": 5931
    },
    {
      "epoch": 0.17092145450354407,
      "grad_norm": 1.1089026927947998,
      "learning_rate": 9.488346350052913e-06,
      "loss": 0.8121,
      "step": 5932
    },
    {
      "epoch": 0.17095026796519333,
      "grad_norm": 1.0666728019714355,
      "learning_rate": 9.488140709513194e-06,
      "loss": 0.7703,
      "step": 5933
    },
    {
      "epoch": 0.17097908142684262,
      "grad_norm": 1.0820980072021484,
      "learning_rate": 9.487935029886257e-06,
      "loss": 0.795,
      "step": 5934
    },
    {
      "epoch": 0.1710078948884919,
      "grad_norm": 1.1064027547836304,
      "learning_rate": 9.48772931117389e-06,
      "loss": 0.8143,
      "step": 5935
    },
    {
      "epoch": 0.17103670835014118,
      "grad_norm": 1.1023870706558228,
      "learning_rate": 9.487523553377888e-06,
      "loss": 0.7891,
      "step": 5936
    },
    {
      "epoch": 0.17106552181179047,
      "grad_norm": 1.0640695095062256,
      "learning_rate": 9.487317756500041e-06,
      "loss": 0.7991,
      "step": 5937
    },
    {
      "epoch": 0.17109433527343976,
      "grad_norm": 1.0745800733566284,
      "learning_rate": 9.487111920542144e-06,
      "loss": 0.7947,
      "step": 5938
    },
    {
      "epoch": 0.17112314873508902,
      "grad_norm": 1.0911544561386108,
      "learning_rate": 9.486906045505987e-06,
      "loss": 0.7759,
      "step": 5939
    },
    {
      "epoch": 0.1711519621967383,
      "grad_norm": 1.0551693439483643,
      "learning_rate": 9.486700131393362e-06,
      "loss": 0.7838,
      "step": 5940
    },
    {
      "epoch": 0.1711807756583876,
      "grad_norm": 1.1024309396743774,
      "learning_rate": 9.486494178206064e-06,
      "loss": 0.8099,
      "step": 5941
    },
    {
      "epoch": 0.1712095891200369,
      "grad_norm": 1.0850685834884644,
      "learning_rate": 9.486288185945888e-06,
      "loss": 0.7863,
      "step": 5942
    },
    {
      "epoch": 0.17123840258168616,
      "grad_norm": 1.045215368270874,
      "learning_rate": 9.486082154614626e-06,
      "loss": 0.7954,
      "step": 5943
    },
    {
      "epoch": 0.17126721604333545,
      "grad_norm": 1.0567103624343872,
      "learning_rate": 9.485876084214073e-06,
      "loss": 0.7854,
      "step": 5944
    },
    {
      "epoch": 0.17129602950498474,
      "grad_norm": 1.0807653665542603,
      "learning_rate": 9.485669974746024e-06,
      "loss": 0.7836,
      "step": 5945
    },
    {
      "epoch": 0.171324842966634,
      "grad_norm": 1.0753109455108643,
      "learning_rate": 9.48546382621227e-06,
      "loss": 0.7723,
      "step": 5946
    },
    {
      "epoch": 0.1713536564282833,
      "grad_norm": 1.1375510692596436,
      "learning_rate": 9.485257638614615e-06,
      "loss": 0.7835,
      "step": 5947
    },
    {
      "epoch": 0.17138246988993258,
      "grad_norm": 1.1009043455123901,
      "learning_rate": 9.485051411954846e-06,
      "loss": 0.7991,
      "step": 5948
    },
    {
      "epoch": 0.17141128335158187,
      "grad_norm": 1.2508286237716675,
      "learning_rate": 9.484845146234764e-06,
      "loss": 0.7967,
      "step": 5949
    },
    {
      "epoch": 0.17144009681323114,
      "grad_norm": 1.1502845287322998,
      "learning_rate": 9.484638841456165e-06,
      "loss": 0.7876,
      "step": 5950
    },
    {
      "epoch": 0.17146891027488043,
      "grad_norm": 1.1032835245132446,
      "learning_rate": 9.484432497620843e-06,
      "loss": 0.797,
      "step": 5951
    },
    {
      "epoch": 0.17149772373652972,
      "grad_norm": 1.182495355606079,
      "learning_rate": 9.484226114730598e-06,
      "loss": 0.798,
      "step": 5952
    },
    {
      "epoch": 0.17152653719817898,
      "grad_norm": 1.0925605297088623,
      "learning_rate": 9.484019692787226e-06,
      "loss": 0.7752,
      "step": 5953
    },
    {
      "epoch": 0.17155535065982827,
      "grad_norm": 1.1529293060302734,
      "learning_rate": 9.483813231792524e-06,
      "loss": 0.7892,
      "step": 5954
    },
    {
      "epoch": 0.17158416412147756,
      "grad_norm": 1.1327271461486816,
      "learning_rate": 9.483606731748292e-06,
      "loss": 0.7875,
      "step": 5955
    },
    {
      "epoch": 0.17161297758312682,
      "grad_norm": 1.1448168754577637,
      "learning_rate": 9.483400192656326e-06,
      "loss": 0.7935,
      "step": 5956
    },
    {
      "epoch": 0.17164179104477612,
      "grad_norm": 1.1220835447311401,
      "learning_rate": 9.483193614518427e-06,
      "loss": 0.7751,
      "step": 5957
    },
    {
      "epoch": 0.1716706045064254,
      "grad_norm": 1.2095311880111694,
      "learning_rate": 9.482986997336391e-06,
      "loss": 0.7842,
      "step": 5958
    },
    {
      "epoch": 0.1716994179680747,
      "grad_norm": 1.1170607805252075,
      "learning_rate": 9.482780341112023e-06,
      "loss": 0.7908,
      "step": 5959
    },
    {
      "epoch": 0.17172823142972396,
      "grad_norm": 1.1169602870941162,
      "learning_rate": 9.482573645847117e-06,
      "loss": 0.7923,
      "step": 5960
    },
    {
      "epoch": 0.17175704489137325,
      "grad_norm": 1.1027066707611084,
      "learning_rate": 9.482366911543477e-06,
      "loss": 0.7832,
      "step": 5961
    },
    {
      "epoch": 0.17178585835302254,
      "grad_norm": 1.1757005453109741,
      "learning_rate": 9.482160138202901e-06,
      "loss": 0.7669,
      "step": 5962
    },
    {
      "epoch": 0.1718146718146718,
      "grad_norm": 1.1377876996994019,
      "learning_rate": 9.481953325827191e-06,
      "loss": 0.7888,
      "step": 5963
    },
    {
      "epoch": 0.1718434852763211,
      "grad_norm": 1.1335783004760742,
      "learning_rate": 9.481746474418148e-06,
      "loss": 0.7819,
      "step": 5964
    },
    {
      "epoch": 0.17187229873797039,
      "grad_norm": 1.153099775314331,
      "learning_rate": 9.481539583977572e-06,
      "loss": 0.7993,
      "step": 5965
    },
    {
      "epoch": 0.17190111219961965,
      "grad_norm": 1.1219465732574463,
      "learning_rate": 9.481332654507269e-06,
      "loss": 0.7989,
      "step": 5966
    },
    {
      "epoch": 0.17192992566126894,
      "grad_norm": 1.1012738943099976,
      "learning_rate": 9.481125686009036e-06,
      "loss": 0.8188,
      "step": 5967
    },
    {
      "epoch": 0.17195873912291823,
      "grad_norm": 1.146794319152832,
      "learning_rate": 9.480918678484679e-06,
      "loss": 0.7955,
      "step": 5968
    },
    {
      "epoch": 0.17198755258456752,
      "grad_norm": 1.025987982749939,
      "learning_rate": 9.480711631935999e-06,
      "loss": 0.793,
      "step": 5969
    },
    {
      "epoch": 0.17201636604621678,
      "grad_norm": 1.0661687850952148,
      "learning_rate": 9.480504546364799e-06,
      "loss": 0.759,
      "step": 5970
    },
    {
      "epoch": 0.17204517950786608,
      "grad_norm": 1.1517667770385742,
      "learning_rate": 9.480297421772883e-06,
      "loss": 0.7775,
      "step": 5971
    },
    {
      "epoch": 0.17207399296951537,
      "grad_norm": 1.1134207248687744,
      "learning_rate": 9.480090258162056e-06,
      "loss": 0.7681,
      "step": 5972
    },
    {
      "epoch": 0.17210280643116463,
      "grad_norm": 1.072679877281189,
      "learning_rate": 9.479883055534123e-06,
      "loss": 0.8007,
      "step": 5973
    },
    {
      "epoch": 0.17213161989281392,
      "grad_norm": 1.120145559310913,
      "learning_rate": 9.479675813890884e-06,
      "loss": 0.8002,
      "step": 5974
    },
    {
      "epoch": 0.1721604333544632,
      "grad_norm": 1.1184089183807373,
      "learning_rate": 9.47946853323415e-06,
      "loss": 0.7964,
      "step": 5975
    },
    {
      "epoch": 0.1721892468161125,
      "grad_norm": 1.1426715850830078,
      "learning_rate": 9.479261213565718e-06,
      "loss": 0.7949,
      "step": 5976
    },
    {
      "epoch": 0.17221806027776176,
      "grad_norm": 1.0593804121017456,
      "learning_rate": 9.479053854887401e-06,
      "loss": 0.7818,
      "step": 5977
    },
    {
      "epoch": 0.17224687373941105,
      "grad_norm": 1.0959736108779907,
      "learning_rate": 9.478846457201003e-06,
      "loss": 0.775,
      "step": 5978
    },
    {
      "epoch": 0.17227568720106035,
      "grad_norm": 1.133568525314331,
      "learning_rate": 9.478639020508329e-06,
      "loss": 0.8109,
      "step": 5979
    },
    {
      "epoch": 0.1723045006627096,
      "grad_norm": 1.053706169128418,
      "learning_rate": 9.478431544811185e-06,
      "loss": 0.7555,
      "step": 5980
    },
    {
      "epoch": 0.1723333141243589,
      "grad_norm": 1.129678726196289,
      "learning_rate": 9.478224030111383e-06,
      "loss": 0.7862,
      "step": 5981
    },
    {
      "epoch": 0.1723621275860082,
      "grad_norm": 1.0685750246047974,
      "learning_rate": 9.478016476410724e-06,
      "loss": 0.8046,
      "step": 5982
    },
    {
      "epoch": 0.17239094104765745,
      "grad_norm": 1.0379197597503662,
      "learning_rate": 9.477808883711017e-06,
      "loss": 0.7668,
      "step": 5983
    },
    {
      "epoch": 0.17241975450930674,
      "grad_norm": 1.1186503171920776,
      "learning_rate": 9.47760125201407e-06,
      "loss": 0.8143,
      "step": 5984
    },
    {
      "epoch": 0.17244856797095603,
      "grad_norm": 1.0369458198547363,
      "learning_rate": 9.477393581321693e-06,
      "loss": 0.8018,
      "step": 5985
    },
    {
      "epoch": 0.17247738143260533,
      "grad_norm": 1.0540233850479126,
      "learning_rate": 9.477185871635694e-06,
      "loss": 0.8032,
      "step": 5986
    },
    {
      "epoch": 0.1725061948942546,
      "grad_norm": 1.0789504051208496,
      "learning_rate": 9.476978122957882e-06,
      "loss": 0.7877,
      "step": 5987
    },
    {
      "epoch": 0.17253500835590388,
      "grad_norm": 1.125075340270996,
      "learning_rate": 9.476770335290066e-06,
      "loss": 0.8092,
      "step": 5988
    },
    {
      "epoch": 0.17256382181755317,
      "grad_norm": 1.0696967840194702,
      "learning_rate": 9.476562508634055e-06,
      "loss": 0.8037,
      "step": 5989
    },
    {
      "epoch": 0.17259263527920243,
      "grad_norm": 1.0886772871017456,
      "learning_rate": 9.476354642991659e-06,
      "loss": 0.791,
      "step": 5990
    },
    {
      "epoch": 0.17262144874085172,
      "grad_norm": 1.1340007781982422,
      "learning_rate": 9.476146738364688e-06,
      "loss": 0.7887,
      "step": 5991
    },
    {
      "epoch": 0.17265026220250101,
      "grad_norm": 1.1008979082107544,
      "learning_rate": 9.475938794754955e-06,
      "loss": 0.7953,
      "step": 5992
    },
    {
      "epoch": 0.17267907566415028,
      "grad_norm": 1.102623701095581,
      "learning_rate": 9.47573081216427e-06,
      "loss": 0.7841,
      "step": 5993
    },
    {
      "epoch": 0.17270788912579957,
      "grad_norm": 1.0931015014648438,
      "learning_rate": 9.475522790594442e-06,
      "loss": 0.7651,
      "step": 5994
    },
    {
      "epoch": 0.17273670258744886,
      "grad_norm": 1.0482325553894043,
      "learning_rate": 9.475314730047284e-06,
      "loss": 0.793,
      "step": 5995
    },
    {
      "epoch": 0.17276551604909815,
      "grad_norm": 1.0405213832855225,
      "learning_rate": 9.47510663052461e-06,
      "loss": 0.778,
      "step": 5996
    },
    {
      "epoch": 0.1727943295107474,
      "grad_norm": 1.0613847970962524,
      "learning_rate": 9.474898492028231e-06,
      "loss": 0.7961,
      "step": 5997
    },
    {
      "epoch": 0.1728231429723967,
      "grad_norm": 1.1714800596237183,
      "learning_rate": 9.47469031455996e-06,
      "loss": 0.8065,
      "step": 5998
    },
    {
      "epoch": 0.172851956434046,
      "grad_norm": 1.0271903276443481,
      "learning_rate": 9.474482098121607e-06,
      "loss": 0.7695,
      "step": 5999
    },
    {
      "epoch": 0.17288076989569526,
      "grad_norm": 1.0168853998184204,
      "learning_rate": 9.474273842714988e-06,
      "loss": 0.7912,
      "step": 6000
    },
    {
      "epoch": 0.17290958335734455,
      "grad_norm": 1.108441948890686,
      "learning_rate": 9.474065548341917e-06,
      "loss": 0.8001,
      "step": 6001
    },
    {
      "epoch": 0.17293839681899384,
      "grad_norm": 1.0750640630722046,
      "learning_rate": 9.47385721500421e-06,
      "loss": 0.7689,
      "step": 6002
    },
    {
      "epoch": 0.17296721028064313,
      "grad_norm": 1.0938822031021118,
      "learning_rate": 9.473648842703675e-06,
      "loss": 0.7756,
      "step": 6003
    },
    {
      "epoch": 0.1729960237422924,
      "grad_norm": 1.1211774349212646,
      "learning_rate": 9.473440431442133e-06,
      "loss": 0.7901,
      "step": 6004
    },
    {
      "epoch": 0.17302483720394168,
      "grad_norm": 1.054848313331604,
      "learning_rate": 9.473231981221395e-06,
      "loss": 0.8,
      "step": 6005
    },
    {
      "epoch": 0.17305365066559097,
      "grad_norm": 1.0593162775039673,
      "learning_rate": 9.473023492043279e-06,
      "loss": 0.7912,
      "step": 6006
    },
    {
      "epoch": 0.17308246412724024,
      "grad_norm": 1.1103047132492065,
      "learning_rate": 9.4728149639096e-06,
      "loss": 0.8027,
      "step": 6007
    },
    {
      "epoch": 0.17311127758888953,
      "grad_norm": 1.1586291790008545,
      "learning_rate": 9.472606396822172e-06,
      "loss": 0.8021,
      "step": 6008
    },
    {
      "epoch": 0.17314009105053882,
      "grad_norm": 1.1474802494049072,
      "learning_rate": 9.472397790782815e-06,
      "loss": 0.7976,
      "step": 6009
    },
    {
      "epoch": 0.17316890451218808,
      "grad_norm": 1.0791125297546387,
      "learning_rate": 9.472189145793344e-06,
      "loss": 0.8089,
      "step": 6010
    },
    {
      "epoch": 0.17319771797383737,
      "grad_norm": 1.1259653568267822,
      "learning_rate": 9.471980461855577e-06,
      "loss": 0.7795,
      "step": 6011
    },
    {
      "epoch": 0.17322653143548666,
      "grad_norm": 1.1623824834823608,
      "learning_rate": 9.471771738971329e-06,
      "loss": 0.7805,
      "step": 6012
    },
    {
      "epoch": 0.17325534489713595,
      "grad_norm": 1.1763384342193604,
      "learning_rate": 9.47156297714242e-06,
      "loss": 0.7992,
      "step": 6013
    },
    {
      "epoch": 0.17328415835878522,
      "grad_norm": 1.120438814163208,
      "learning_rate": 9.471354176370667e-06,
      "loss": 0.7923,
      "step": 6014
    },
    {
      "epoch": 0.1733129718204345,
      "grad_norm": 1.099208116531372,
      "learning_rate": 9.471145336657888e-06,
      "loss": 0.7718,
      "step": 6015
    },
    {
      "epoch": 0.1733417852820838,
      "grad_norm": 1.212499737739563,
      "learning_rate": 9.470936458005903e-06,
      "loss": 0.7696,
      "step": 6016
    },
    {
      "epoch": 0.17337059874373306,
      "grad_norm": 1.0895497798919678,
      "learning_rate": 9.470727540416532e-06,
      "loss": 0.7836,
      "step": 6017
    },
    {
      "epoch": 0.17339941220538235,
      "grad_norm": 1.0558035373687744,
      "learning_rate": 9.470518583891592e-06,
      "loss": 0.7641,
      "step": 6018
    },
    {
      "epoch": 0.17342822566703164,
      "grad_norm": 1.1108886003494263,
      "learning_rate": 9.470309588432905e-06,
      "loss": 0.7929,
      "step": 6019
    },
    {
      "epoch": 0.1734570391286809,
      "grad_norm": 1.0931421518325806,
      "learning_rate": 9.47010055404229e-06,
      "loss": 0.7926,
      "step": 6020
    },
    {
      "epoch": 0.1734858525903302,
      "grad_norm": 1.05841863155365,
      "learning_rate": 9.469891480721566e-06,
      "loss": 0.792,
      "step": 6021
    },
    {
      "epoch": 0.1735146660519795,
      "grad_norm": 1.1147396564483643,
      "learning_rate": 9.469682368472559e-06,
      "loss": 0.7609,
      "step": 6022
    },
    {
      "epoch": 0.17354347951362878,
      "grad_norm": 1.11758291721344,
      "learning_rate": 9.469473217297084e-06,
      "loss": 0.7792,
      "step": 6023
    },
    {
      "epoch": 0.17357229297527804,
      "grad_norm": 1.0808972120285034,
      "learning_rate": 9.469264027196965e-06,
      "loss": 0.8053,
      "step": 6024
    },
    {
      "epoch": 0.17360110643692733,
      "grad_norm": 1.0932104587554932,
      "learning_rate": 9.469054798174025e-06,
      "loss": 0.7906,
      "step": 6025
    },
    {
      "epoch": 0.17362991989857662,
      "grad_norm": 1.0951547622680664,
      "learning_rate": 9.468845530230083e-06,
      "loss": 0.817,
      "step": 6026
    },
    {
      "epoch": 0.1736587333602259,
      "grad_norm": 1.0759708881378174,
      "learning_rate": 9.468636223366967e-06,
      "loss": 0.7942,
      "step": 6027
    },
    {
      "epoch": 0.17368754682187518,
      "grad_norm": 1.0436458587646484,
      "learning_rate": 9.468426877586495e-06,
      "loss": 0.7992,
      "step": 6028
    },
    {
      "epoch": 0.17371636028352447,
      "grad_norm": 1.0446155071258545,
      "learning_rate": 9.46821749289049e-06,
      "loss": 0.7715,
      "step": 6029
    },
    {
      "epoch": 0.17374517374517376,
      "grad_norm": 1.0321681499481201,
      "learning_rate": 9.468008069280778e-06,
      "loss": 0.7581,
      "step": 6030
    },
    {
      "epoch": 0.17377398720682302,
      "grad_norm": 1.0652835369110107,
      "learning_rate": 9.467798606759184e-06,
      "loss": 0.7698,
      "step": 6031
    },
    {
      "epoch": 0.1738028006684723,
      "grad_norm": 1.0427950620651245,
      "learning_rate": 9.46758910532753e-06,
      "loss": 0.7717,
      "step": 6032
    },
    {
      "epoch": 0.1738316141301216,
      "grad_norm": 1.0925880670547485,
      "learning_rate": 9.467379564987638e-06,
      "loss": 0.8003,
      "step": 6033
    },
    {
      "epoch": 0.17386042759177087,
      "grad_norm": 1.1419904232025146,
      "learning_rate": 9.467169985741338e-06,
      "loss": 0.7792,
      "step": 6034
    },
    {
      "epoch": 0.17388924105342016,
      "grad_norm": 1.1344468593597412,
      "learning_rate": 9.466960367590452e-06,
      "loss": 0.7968,
      "step": 6035
    },
    {
      "epoch": 0.17391805451506945,
      "grad_norm": 1.0703924894332886,
      "learning_rate": 9.466750710536809e-06,
      "loss": 0.7923,
      "step": 6036
    },
    {
      "epoch": 0.1739468679767187,
      "grad_norm": 1.0437482595443726,
      "learning_rate": 9.46654101458223e-06,
      "loss": 0.7748,
      "step": 6037
    },
    {
      "epoch": 0.173975681438368,
      "grad_norm": 1.1427299976348877,
      "learning_rate": 9.466331279728544e-06,
      "loss": 0.7891,
      "step": 6038
    },
    {
      "epoch": 0.1740044949000173,
      "grad_norm": 1.1839981079101562,
      "learning_rate": 9.466121505977577e-06,
      "loss": 0.7845,
      "step": 6039
    },
    {
      "epoch": 0.17403330836166658,
      "grad_norm": 1.1742528676986694,
      "learning_rate": 9.465911693331157e-06,
      "loss": 0.8297,
      "step": 6040
    },
    {
      "epoch": 0.17406212182331585,
      "grad_norm": 1.1200510263442993,
      "learning_rate": 9.46570184179111e-06,
      "loss": 0.793,
      "step": 6041
    },
    {
      "epoch": 0.17409093528496514,
      "grad_norm": 1.1154958009719849,
      "learning_rate": 9.465491951359264e-06,
      "loss": 0.8075,
      "step": 6042
    },
    {
      "epoch": 0.17411974874661443,
      "grad_norm": 1.0686004161834717,
      "learning_rate": 9.465282022037447e-06,
      "loss": 0.7806,
      "step": 6043
    },
    {
      "epoch": 0.1741485622082637,
      "grad_norm": 1.0652896165847778,
      "learning_rate": 9.465072053827487e-06,
      "loss": 0.7631,
      "step": 6044
    },
    {
      "epoch": 0.17417737566991298,
      "grad_norm": 1.0676746368408203,
      "learning_rate": 9.464862046731213e-06,
      "loss": 0.7765,
      "step": 6045
    },
    {
      "epoch": 0.17420618913156227,
      "grad_norm": 1.0966360569000244,
      "learning_rate": 9.464652000750454e-06,
      "loss": 0.7856,
      "step": 6046
    },
    {
      "epoch": 0.17423500259321154,
      "grad_norm": 1.116274356842041,
      "learning_rate": 9.46444191588704e-06,
      "loss": 0.7807,
      "step": 6047
    },
    {
      "epoch": 0.17426381605486083,
      "grad_norm": 1.0601457357406616,
      "learning_rate": 9.464231792142798e-06,
      "loss": 0.7809,
      "step": 6048
    },
    {
      "epoch": 0.17429262951651012,
      "grad_norm": 1.1102347373962402,
      "learning_rate": 9.46402162951956e-06,
      "loss": 0.8061,
      "step": 6049
    },
    {
      "epoch": 0.1743214429781594,
      "grad_norm": 1.1407661437988281,
      "learning_rate": 9.463811428019156e-06,
      "loss": 0.7823,
      "step": 6050
    },
    {
      "epoch": 0.17435025643980867,
      "grad_norm": 1.0508710145950317,
      "learning_rate": 9.463601187643419e-06,
      "loss": 0.7923,
      "step": 6051
    },
    {
      "epoch": 0.17437906990145796,
      "grad_norm": 1.1032273769378662,
      "learning_rate": 9.463390908394175e-06,
      "loss": 0.766,
      "step": 6052
    },
    {
      "epoch": 0.17440788336310725,
      "grad_norm": 1.0915706157684326,
      "learning_rate": 9.463180590273257e-06,
      "loss": 0.7841,
      "step": 6053
    },
    {
      "epoch": 0.17443669682475652,
      "grad_norm": 1.097604513168335,
      "learning_rate": 9.4629702332825e-06,
      "loss": 0.7744,
      "step": 6054
    },
    {
      "epoch": 0.1744655102864058,
      "grad_norm": 1.0564099550247192,
      "learning_rate": 9.462759837423734e-06,
      "loss": 0.8029,
      "step": 6055
    },
    {
      "epoch": 0.1744943237480551,
      "grad_norm": 1.0747452974319458,
      "learning_rate": 9.46254940269879e-06,
      "loss": 0.8027,
      "step": 6056
    },
    {
      "epoch": 0.1745231372097044,
      "grad_norm": 1.0688260793685913,
      "learning_rate": 9.4623389291095e-06,
      "loss": 0.7786,
      "step": 6057
    },
    {
      "epoch": 0.17455195067135365,
      "grad_norm": 1.060671091079712,
      "learning_rate": 9.462128416657701e-06,
      "loss": 0.7764,
      "step": 6058
    },
    {
      "epoch": 0.17458076413300294,
      "grad_norm": 1.0819419622421265,
      "learning_rate": 9.461917865345223e-06,
      "loss": 0.8152,
      "step": 6059
    },
    {
      "epoch": 0.17460957759465223,
      "grad_norm": 1.1105537414550781,
      "learning_rate": 9.4617072751739e-06,
      "loss": 0.8012,
      "step": 6060
    },
    {
      "epoch": 0.1746383910563015,
      "grad_norm": 1.1816898584365845,
      "learning_rate": 9.461496646145567e-06,
      "loss": 0.801,
      "step": 6061
    },
    {
      "epoch": 0.17466720451795079,
      "grad_norm": 1.045345425605774,
      "learning_rate": 9.461285978262059e-06,
      "loss": 0.7937,
      "step": 6062
    },
    {
      "epoch": 0.17469601797960008,
      "grad_norm": 1.0870630741119385,
      "learning_rate": 9.46107527152521e-06,
      "loss": 0.7897,
      "step": 6063
    },
    {
      "epoch": 0.17472483144124934,
      "grad_norm": 1.116686463356018,
      "learning_rate": 9.460864525936854e-06,
      "loss": 0.8029,
      "step": 6064
    },
    {
      "epoch": 0.17475364490289863,
      "grad_norm": 1.1284570693969727,
      "learning_rate": 9.460653741498826e-06,
      "loss": 0.7732,
      "step": 6065
    },
    {
      "epoch": 0.17478245836454792,
      "grad_norm": 1.0156826972961426,
      "learning_rate": 9.460442918212965e-06,
      "loss": 0.7877,
      "step": 6066
    },
    {
      "epoch": 0.1748112718261972,
      "grad_norm": 1.0837178230285645,
      "learning_rate": 9.460232056081105e-06,
      "loss": 0.8012,
      "step": 6067
    },
    {
      "epoch": 0.17484008528784648,
      "grad_norm": 1.0836946964263916,
      "learning_rate": 9.460021155105082e-06,
      "loss": 0.7807,
      "step": 6068
    },
    {
      "epoch": 0.17486889874949577,
      "grad_norm": 1.109951138496399,
      "learning_rate": 9.459810215286732e-06,
      "loss": 0.7756,
      "step": 6069
    },
    {
      "epoch": 0.17489771221114506,
      "grad_norm": 1.1280285120010376,
      "learning_rate": 9.459599236627894e-06,
      "loss": 0.7896,
      "step": 6070
    },
    {
      "epoch": 0.17492652567279432,
      "grad_norm": 1.1153966188430786,
      "learning_rate": 9.459388219130405e-06,
      "loss": 0.7827,
      "step": 6071
    },
    {
      "epoch": 0.1749553391344436,
      "grad_norm": 1.0800540447235107,
      "learning_rate": 9.4591771627961e-06,
      "loss": 0.7891,
      "step": 6072
    },
    {
      "epoch": 0.1749841525960929,
      "grad_norm": 1.130998134613037,
      "learning_rate": 9.458966067626823e-06,
      "loss": 0.7892,
      "step": 6073
    },
    {
      "epoch": 0.17501296605774216,
      "grad_norm": 1.1065795421600342,
      "learning_rate": 9.458754933624406e-06,
      "loss": 0.7945,
      "step": 6074
    },
    {
      "epoch": 0.17504177951939145,
      "grad_norm": 1.0871708393096924,
      "learning_rate": 9.458543760790692e-06,
      "loss": 0.7815,
      "step": 6075
    },
    {
      "epoch": 0.17507059298104075,
      "grad_norm": 1.138223648071289,
      "learning_rate": 9.45833254912752e-06,
      "loss": 0.777,
      "step": 6076
    },
    {
      "epoch": 0.17509940644269004,
      "grad_norm": 1.137470006942749,
      "learning_rate": 9.458121298636727e-06,
      "loss": 0.8095,
      "step": 6077
    },
    {
      "epoch": 0.1751282199043393,
      "grad_norm": 1.087449550628662,
      "learning_rate": 9.457910009320153e-06,
      "loss": 0.8066,
      "step": 6078
    },
    {
      "epoch": 0.1751570333659886,
      "grad_norm": 1.1743956804275513,
      "learning_rate": 9.45769868117964e-06,
      "loss": 0.8263,
      "step": 6079
    },
    {
      "epoch": 0.17518584682763788,
      "grad_norm": 1.0728144645690918,
      "learning_rate": 9.457487314217028e-06,
      "loss": 0.8065,
      "step": 6080
    },
    {
      "epoch": 0.17521466028928714,
      "grad_norm": 1.1200189590454102,
      "learning_rate": 9.457275908434156e-06,
      "loss": 0.7919,
      "step": 6081
    },
    {
      "epoch": 0.17524347375093643,
      "grad_norm": 1.0828943252563477,
      "learning_rate": 9.457064463832868e-06,
      "loss": 0.7815,
      "step": 6082
    },
    {
      "epoch": 0.17527228721258573,
      "grad_norm": 1.1197021007537842,
      "learning_rate": 9.456852980415003e-06,
      "loss": 0.7994,
      "step": 6083
    },
    {
      "epoch": 0.17530110067423502,
      "grad_norm": 1.0809742212295532,
      "learning_rate": 9.456641458182404e-06,
      "loss": 0.7857,
      "step": 6084
    },
    {
      "epoch": 0.17532991413588428,
      "grad_norm": 1.09108304977417,
      "learning_rate": 9.456429897136914e-06,
      "loss": 0.7822,
      "step": 6085
    },
    {
      "epoch": 0.17535872759753357,
      "grad_norm": 1.0894100666046143,
      "learning_rate": 9.456218297280372e-06,
      "loss": 0.7757,
      "step": 6086
    },
    {
      "epoch": 0.17538754105918286,
      "grad_norm": 1.0434257984161377,
      "learning_rate": 9.456006658614624e-06,
      "loss": 0.785,
      "step": 6087
    },
    {
      "epoch": 0.17541635452083212,
      "grad_norm": 1.152799129486084,
      "learning_rate": 9.455794981141514e-06,
      "loss": 0.7957,
      "step": 6088
    },
    {
      "epoch": 0.17544516798248141,
      "grad_norm": 1.0161447525024414,
      "learning_rate": 9.455583264862883e-06,
      "loss": 0.8072,
      "step": 6089
    },
    {
      "epoch": 0.1754739814441307,
      "grad_norm": 1.0918864011764526,
      "learning_rate": 9.455371509780576e-06,
      "loss": 0.7939,
      "step": 6090
    },
    {
      "epoch": 0.17550279490577997,
      "grad_norm": 1.1250053644180298,
      "learning_rate": 9.455159715896436e-06,
      "loss": 0.8106,
      "step": 6091
    },
    {
      "epoch": 0.17553160836742926,
      "grad_norm": 1.075483798980713,
      "learning_rate": 9.454947883212308e-06,
      "loss": 0.7656,
      "step": 6092
    },
    {
      "epoch": 0.17556042182907855,
      "grad_norm": 1.1087489128112793,
      "learning_rate": 9.454736011730039e-06,
      "loss": 0.8072,
      "step": 6093
    },
    {
      "epoch": 0.17558923529072784,
      "grad_norm": 1.0798988342285156,
      "learning_rate": 9.454524101451472e-06,
      "loss": 0.7632,
      "step": 6094
    },
    {
      "epoch": 0.1756180487523771,
      "grad_norm": 1.0963306427001953,
      "learning_rate": 9.454312152378452e-06,
      "loss": 0.7851,
      "step": 6095
    },
    {
      "epoch": 0.1756468622140264,
      "grad_norm": 1.0384293794631958,
      "learning_rate": 9.454100164512828e-06,
      "loss": 0.7891,
      "step": 6096
    },
    {
      "epoch": 0.17567567567567569,
      "grad_norm": 1.0667353868484497,
      "learning_rate": 9.45388813785644e-06,
      "loss": 0.7781,
      "step": 6097
    },
    {
      "epoch": 0.17570448913732495,
      "grad_norm": 1.1202199459075928,
      "learning_rate": 9.453676072411142e-06,
      "loss": 0.7897,
      "step": 6098
    },
    {
      "epoch": 0.17573330259897424,
      "grad_norm": 1.058722734451294,
      "learning_rate": 9.453463968178775e-06,
      "loss": 0.787,
      "step": 6099
    },
    {
      "epoch": 0.17576211606062353,
      "grad_norm": 1.103666067123413,
      "learning_rate": 9.453251825161192e-06,
      "loss": 0.7569,
      "step": 6100
    },
    {
      "epoch": 0.1757909295222728,
      "grad_norm": 1.0712997913360596,
      "learning_rate": 9.453039643360234e-06,
      "loss": 0.7935,
      "step": 6101
    },
    {
      "epoch": 0.17581974298392208,
      "grad_norm": 1.0943279266357422,
      "learning_rate": 9.452827422777753e-06,
      "loss": 0.7701,
      "step": 6102
    },
    {
      "epoch": 0.17584855644557137,
      "grad_norm": 1.0840338468551636,
      "learning_rate": 9.452615163415598e-06,
      "loss": 0.7753,
      "step": 6103
    },
    {
      "epoch": 0.17587736990722067,
      "grad_norm": 1.1151598691940308,
      "learning_rate": 9.452402865275614e-06,
      "loss": 0.7909,
      "step": 6104
    },
    {
      "epoch": 0.17590618336886993,
      "grad_norm": 1.074678897857666,
      "learning_rate": 9.452190528359652e-06,
      "loss": 0.7748,
      "step": 6105
    },
    {
      "epoch": 0.17593499683051922,
      "grad_norm": 1.1040749549865723,
      "learning_rate": 9.451978152669562e-06,
      "loss": 0.81,
      "step": 6106
    },
    {
      "epoch": 0.1759638102921685,
      "grad_norm": 1.0702193975448608,
      "learning_rate": 9.451765738207192e-06,
      "loss": 0.784,
      "step": 6107
    },
    {
      "epoch": 0.17599262375381777,
      "grad_norm": 1.0779120922088623,
      "learning_rate": 9.451553284974393e-06,
      "loss": 0.7944,
      "step": 6108
    },
    {
      "epoch": 0.17602143721546706,
      "grad_norm": 1.0892815589904785,
      "learning_rate": 9.451340792973016e-06,
      "loss": 0.7676,
      "step": 6109
    },
    {
      "epoch": 0.17605025067711635,
      "grad_norm": 1.0848158597946167,
      "learning_rate": 9.451128262204909e-06,
      "loss": 0.7952,
      "step": 6110
    },
    {
      "epoch": 0.17607906413876562,
      "grad_norm": 1.0840245485305786,
      "learning_rate": 9.450915692671923e-06,
      "loss": 0.7864,
      "step": 6111
    },
    {
      "epoch": 0.1761078776004149,
      "grad_norm": 1.0520724058151245,
      "learning_rate": 9.450703084375912e-06,
      "loss": 0.7721,
      "step": 6112
    },
    {
      "epoch": 0.1761366910620642,
      "grad_norm": 1.1172295808792114,
      "learning_rate": 9.450490437318726e-06,
      "loss": 0.8069,
      "step": 6113
    },
    {
      "epoch": 0.1761655045237135,
      "grad_norm": 1.0568792819976807,
      "learning_rate": 9.450277751502219e-06,
      "loss": 0.8141,
      "step": 6114
    },
    {
      "epoch": 0.17619431798536275,
      "grad_norm": 1.0549262762069702,
      "learning_rate": 9.450065026928238e-06,
      "loss": 0.7761,
      "step": 6115
    },
    {
      "epoch": 0.17622313144701204,
      "grad_norm": 1.0855106115341187,
      "learning_rate": 9.449852263598641e-06,
      "loss": 0.8052,
      "step": 6116
    },
    {
      "epoch": 0.17625194490866133,
      "grad_norm": 1.1077320575714111,
      "learning_rate": 9.449639461515278e-06,
      "loss": 0.8113,
      "step": 6117
    },
    {
      "epoch": 0.1762807583703106,
      "grad_norm": 1.070317029953003,
      "learning_rate": 9.449426620680005e-06,
      "loss": 0.7874,
      "step": 6118
    },
    {
      "epoch": 0.1763095718319599,
      "grad_norm": 1.1166067123413086,
      "learning_rate": 9.449213741094674e-06,
      "loss": 0.8268,
      "step": 6119
    },
    {
      "epoch": 0.17633838529360918,
      "grad_norm": 1.1097090244293213,
      "learning_rate": 9.449000822761137e-06,
      "loss": 0.7672,
      "step": 6120
    },
    {
      "epoch": 0.17636719875525847,
      "grad_norm": 1.0583306550979614,
      "learning_rate": 9.448787865681251e-06,
      "loss": 0.7965,
      "step": 6121
    },
    {
      "epoch": 0.17639601221690773,
      "grad_norm": 1.0765541791915894,
      "learning_rate": 9.44857486985687e-06,
      "loss": 0.7891,
      "step": 6122
    },
    {
      "epoch": 0.17642482567855702,
      "grad_norm": 1.051464557647705,
      "learning_rate": 9.44836183528985e-06,
      "loss": 0.7891,
      "step": 6123
    },
    {
      "epoch": 0.17645363914020631,
      "grad_norm": 1.0694005489349365,
      "learning_rate": 9.448148761982044e-06,
      "loss": 0.7721,
      "step": 6124
    },
    {
      "epoch": 0.17648245260185558,
      "grad_norm": 1.0648694038391113,
      "learning_rate": 9.447935649935308e-06,
      "loss": 0.7847,
      "step": 6125
    },
    {
      "epoch": 0.17651126606350487,
      "grad_norm": 1.0503761768341064,
      "learning_rate": 9.4477224991515e-06,
      "loss": 0.79,
      "step": 6126
    },
    {
      "epoch": 0.17654007952515416,
      "grad_norm": 1.050690770149231,
      "learning_rate": 9.447509309632474e-06,
      "loss": 0.7919,
      "step": 6127
    },
    {
      "epoch": 0.17656889298680342,
      "grad_norm": 1.0658048391342163,
      "learning_rate": 9.447296081380089e-06,
      "loss": 0.7962,
      "step": 6128
    },
    {
      "epoch": 0.1765977064484527,
      "grad_norm": 1.0781259536743164,
      "learning_rate": 9.4470828143962e-06,
      "loss": 0.8025,
      "step": 6129
    },
    {
      "epoch": 0.176626519910102,
      "grad_norm": 1.0585323572158813,
      "learning_rate": 9.446869508682666e-06,
      "loss": 0.7888,
      "step": 6130
    },
    {
      "epoch": 0.1766553333717513,
      "grad_norm": 1.0801719427108765,
      "learning_rate": 9.446656164241342e-06,
      "loss": 0.7976,
      "step": 6131
    },
    {
      "epoch": 0.17668414683340056,
      "grad_norm": 1.1198171377182007,
      "learning_rate": 9.44644278107409e-06,
      "loss": 0.8046,
      "step": 6132
    },
    {
      "epoch": 0.17671296029504985,
      "grad_norm": 1.0631133317947388,
      "learning_rate": 9.446229359182767e-06,
      "loss": 0.787,
      "step": 6133
    },
    {
      "epoch": 0.17674177375669914,
      "grad_norm": 1.094710111618042,
      "learning_rate": 9.446015898569227e-06,
      "loss": 0.8054,
      "step": 6134
    },
    {
      "epoch": 0.1767705872183484,
      "grad_norm": 1.1233781576156616,
      "learning_rate": 9.445802399235335e-06,
      "loss": 0.812,
      "step": 6135
    },
    {
      "epoch": 0.1767994006799977,
      "grad_norm": 1.0806821584701538,
      "learning_rate": 9.445588861182948e-06,
      "loss": 0.7892,
      "step": 6136
    },
    {
      "epoch": 0.17682821414164698,
      "grad_norm": 1.0935310125350952,
      "learning_rate": 9.445375284413927e-06,
      "loss": 0.8125,
      "step": 6137
    },
    {
      "epoch": 0.17685702760329625,
      "grad_norm": 1.0635114908218384,
      "learning_rate": 9.44516166893013e-06,
      "loss": 0.7767,
      "step": 6138
    },
    {
      "epoch": 0.17688584106494554,
      "grad_norm": 1.060213327407837,
      "learning_rate": 9.44494801473342e-06,
      "loss": 0.8047,
      "step": 6139
    },
    {
      "epoch": 0.17691465452659483,
      "grad_norm": 1.1008938550949097,
      "learning_rate": 9.444734321825655e-06,
      "loss": 0.7868,
      "step": 6140
    },
    {
      "epoch": 0.17694346798824412,
      "grad_norm": 1.0925016403198242,
      "learning_rate": 9.444520590208697e-06,
      "loss": 0.7991,
      "step": 6141
    },
    {
      "epoch": 0.17697228144989338,
      "grad_norm": 1.066679835319519,
      "learning_rate": 9.444306819884406e-06,
      "loss": 0.8104,
      "step": 6142
    },
    {
      "epoch": 0.17700109491154267,
      "grad_norm": 1.0322378873825073,
      "learning_rate": 9.444093010854647e-06,
      "loss": 0.7637,
      "step": 6143
    },
    {
      "epoch": 0.17702990837319196,
      "grad_norm": 1.1444021463394165,
      "learning_rate": 9.443879163121282e-06,
      "loss": 0.8085,
      "step": 6144
    },
    {
      "epoch": 0.17705872183484123,
      "grad_norm": 1.062294363975525,
      "learning_rate": 9.443665276686168e-06,
      "loss": 0.7689,
      "step": 6145
    },
    {
      "epoch": 0.17708753529649052,
      "grad_norm": 1.080492377281189,
      "learning_rate": 9.443451351551173e-06,
      "loss": 0.7888,
      "step": 6146
    },
    {
      "epoch": 0.1771163487581398,
      "grad_norm": 1.110275387763977,
      "learning_rate": 9.44323738771816e-06,
      "loss": 0.7822,
      "step": 6147
    },
    {
      "epoch": 0.1771451622197891,
      "grad_norm": 1.093043565750122,
      "learning_rate": 9.44302338518899e-06,
      "loss": 0.8013,
      "step": 6148
    },
    {
      "epoch": 0.17717397568143836,
      "grad_norm": 1.0418394804000854,
      "learning_rate": 9.442809343965528e-06,
      "loss": 0.7947,
      "step": 6149
    },
    {
      "epoch": 0.17720278914308765,
      "grad_norm": 1.032556414604187,
      "learning_rate": 9.442595264049636e-06,
      "loss": 0.7987,
      "step": 6150
    },
    {
      "epoch": 0.17723160260473694,
      "grad_norm": 1.1046680212020874,
      "learning_rate": 9.442381145443183e-06,
      "loss": 0.791,
      "step": 6151
    },
    {
      "epoch": 0.1772604160663862,
      "grad_norm": 1.0103238821029663,
      "learning_rate": 9.44216698814803e-06,
      "loss": 0.8031,
      "step": 6152
    },
    {
      "epoch": 0.1772892295280355,
      "grad_norm": 1.0844379663467407,
      "learning_rate": 9.441952792166043e-06,
      "loss": 0.7871,
      "step": 6153
    },
    {
      "epoch": 0.1773180429896848,
      "grad_norm": 1.0619453191757202,
      "learning_rate": 9.441738557499087e-06,
      "loss": 0.7939,
      "step": 6154
    },
    {
      "epoch": 0.17734685645133405,
      "grad_norm": 2.7094995975494385,
      "learning_rate": 9.441524284149028e-06,
      "loss": 0.8184,
      "step": 6155
    },
    {
      "epoch": 0.17737566991298334,
      "grad_norm": 1.0469310283660889,
      "learning_rate": 9.441309972117733e-06,
      "loss": 0.7722,
      "step": 6156
    },
    {
      "epoch": 0.17740448337463263,
      "grad_norm": 1.1009856462478638,
      "learning_rate": 9.441095621407068e-06,
      "loss": 0.787,
      "step": 6157
    },
    {
      "epoch": 0.17743329683628192,
      "grad_norm": 1.1260360479354858,
      "learning_rate": 9.440881232018898e-06,
      "loss": 0.7823,
      "step": 6158
    },
    {
      "epoch": 0.17746211029793119,
      "grad_norm": 1.0711835622787476,
      "learning_rate": 9.440666803955094e-06,
      "loss": 0.7913,
      "step": 6159
    },
    {
      "epoch": 0.17749092375958048,
      "grad_norm": 1.0907140970230103,
      "learning_rate": 9.44045233721752e-06,
      "loss": 0.7774,
      "step": 6160
    },
    {
      "epoch": 0.17751973722122977,
      "grad_norm": 1.1045455932617188,
      "learning_rate": 9.440237831808046e-06,
      "loss": 0.8029,
      "step": 6161
    },
    {
      "epoch": 0.17754855068287903,
      "grad_norm": 1.1148203611373901,
      "learning_rate": 9.440023287728536e-06,
      "loss": 0.7957,
      "step": 6162
    },
    {
      "epoch": 0.17757736414452832,
      "grad_norm": 1.087451457977295,
      "learning_rate": 9.439808704980866e-06,
      "loss": 0.7804,
      "step": 6163
    },
    {
      "epoch": 0.1776061776061776,
      "grad_norm": 1.1503227949142456,
      "learning_rate": 9.439594083566897e-06,
      "loss": 0.7743,
      "step": 6164
    },
    {
      "epoch": 0.17763499106782688,
      "grad_norm": 1.1418490409851074,
      "learning_rate": 9.439379423488504e-06,
      "loss": 0.7937,
      "step": 6165
    },
    {
      "epoch": 0.17766380452947617,
      "grad_norm": 1.0686554908752441,
      "learning_rate": 9.439164724747552e-06,
      "loss": 0.7847,
      "step": 6166
    },
    {
      "epoch": 0.17769261799112546,
      "grad_norm": 1.1140589714050293,
      "learning_rate": 9.438949987345913e-06,
      "loss": 0.802,
      "step": 6167
    },
    {
      "epoch": 0.17772143145277475,
      "grad_norm": 1.0982204675674438,
      "learning_rate": 9.438735211285458e-06,
      "loss": 0.8138,
      "step": 6168
    },
    {
      "epoch": 0.177750244914424,
      "grad_norm": 1.0349704027175903,
      "learning_rate": 9.438520396568055e-06,
      "loss": 0.7877,
      "step": 6169
    },
    {
      "epoch": 0.1777790583760733,
      "grad_norm": 1.05805504322052,
      "learning_rate": 9.438305543195578e-06,
      "loss": 0.8321,
      "step": 6170
    },
    {
      "epoch": 0.1778078718377226,
      "grad_norm": 1.1458051204681396,
      "learning_rate": 9.438090651169895e-06,
      "loss": 0.8008,
      "step": 6171
    },
    {
      "epoch": 0.17783668529937185,
      "grad_norm": 1.1048423051834106,
      "learning_rate": 9.437875720492879e-06,
      "loss": 0.7917,
      "step": 6172
    },
    {
      "epoch": 0.17786549876102115,
      "grad_norm": 1.0361980199813843,
      "learning_rate": 9.437660751166404e-06,
      "loss": 0.7987,
      "step": 6173
    },
    {
      "epoch": 0.17789431222267044,
      "grad_norm": 1.077184796333313,
      "learning_rate": 9.437445743192338e-06,
      "loss": 0.7748,
      "step": 6174
    },
    {
      "epoch": 0.17792312568431973,
      "grad_norm": 1.150989055633545,
      "learning_rate": 9.437230696572554e-06,
      "loss": 0.7714,
      "step": 6175
    },
    {
      "epoch": 0.177951939145969,
      "grad_norm": 1.0735467672348022,
      "learning_rate": 9.437015611308929e-06,
      "loss": 0.8116,
      "step": 6176
    },
    {
      "epoch": 0.17798075260761828,
      "grad_norm": 1.0682640075683594,
      "learning_rate": 9.436800487403331e-06,
      "loss": 0.8176,
      "step": 6177
    },
    {
      "epoch": 0.17800956606926757,
      "grad_norm": 1.0708879232406616,
      "learning_rate": 9.436585324857636e-06,
      "loss": 0.7742,
      "step": 6178
    },
    {
      "epoch": 0.17803837953091683,
      "grad_norm": 1.0643696784973145,
      "learning_rate": 9.43637012367372e-06,
      "loss": 0.7886,
      "step": 6179
    },
    {
      "epoch": 0.17806719299256613,
      "grad_norm": 1.0825564861297607,
      "learning_rate": 9.436154883853455e-06,
      "loss": 0.7897,
      "step": 6180
    },
    {
      "epoch": 0.17809600645421542,
      "grad_norm": 1.0353213548660278,
      "learning_rate": 9.435939605398712e-06,
      "loss": 0.7973,
      "step": 6181
    },
    {
      "epoch": 0.17812481991586468,
      "grad_norm": 1.0473846197128296,
      "learning_rate": 9.435724288311371e-06,
      "loss": 0.8082,
      "step": 6182
    },
    {
      "epoch": 0.17815363337751397,
      "grad_norm": 1.0037070512771606,
      "learning_rate": 9.435508932593309e-06,
      "loss": 0.7728,
      "step": 6183
    },
    {
      "epoch": 0.17818244683916326,
      "grad_norm": 1.0453205108642578,
      "learning_rate": 9.435293538246394e-06,
      "loss": 0.7776,
      "step": 6184
    },
    {
      "epoch": 0.17821126030081255,
      "grad_norm": 1.0684247016906738,
      "learning_rate": 9.435078105272508e-06,
      "loss": 0.7946,
      "step": 6185
    },
    {
      "epoch": 0.17824007376246181,
      "grad_norm": 1.0741198062896729,
      "learning_rate": 9.434862633673525e-06,
      "loss": 0.7873,
      "step": 6186
    },
    {
      "epoch": 0.1782688872241111,
      "grad_norm": 1.0524089336395264,
      "learning_rate": 9.43464712345132e-06,
      "loss": 0.7873,
      "step": 6187
    },
    {
      "epoch": 0.1782977006857604,
      "grad_norm": 1.0368729829788208,
      "learning_rate": 9.434431574607772e-06,
      "loss": 0.7965,
      "step": 6188
    },
    {
      "epoch": 0.17832651414740966,
      "grad_norm": 1.06803560256958,
      "learning_rate": 9.43421598714476e-06,
      "loss": 0.7801,
      "step": 6189
    },
    {
      "epoch": 0.17835532760905895,
      "grad_norm": 1.0414460897445679,
      "learning_rate": 9.434000361064157e-06,
      "loss": 0.7818,
      "step": 6190
    },
    {
      "epoch": 0.17838414107070824,
      "grad_norm": 1.0188729763031006,
      "learning_rate": 9.433784696367843e-06,
      "loss": 0.7827,
      "step": 6191
    },
    {
      "epoch": 0.1784129545323575,
      "grad_norm": 1.0363603830337524,
      "learning_rate": 9.433568993057698e-06,
      "loss": 0.7868,
      "step": 6192
    },
    {
      "epoch": 0.1784417679940068,
      "grad_norm": 1.044907808303833,
      "learning_rate": 9.433353251135599e-06,
      "loss": 0.7681,
      "step": 6193
    },
    {
      "epoch": 0.17847058145565609,
      "grad_norm": 1.0533732175827026,
      "learning_rate": 9.433137470603424e-06,
      "loss": 0.7944,
      "step": 6194
    },
    {
      "epoch": 0.17849939491730538,
      "grad_norm": 1.06887686252594,
      "learning_rate": 9.432921651463053e-06,
      "loss": 0.808,
      "step": 6195
    },
    {
      "epoch": 0.17852820837895464,
      "grad_norm": 1.0066012144088745,
      "learning_rate": 9.432705793716368e-06,
      "loss": 0.7615,
      "step": 6196
    },
    {
      "epoch": 0.17855702184060393,
      "grad_norm": 1.0489208698272705,
      "learning_rate": 9.432489897365244e-06,
      "loss": 0.7793,
      "step": 6197
    },
    {
      "epoch": 0.17858583530225322,
      "grad_norm": 1.0289232730865479,
      "learning_rate": 9.432273962411565e-06,
      "loss": 0.7922,
      "step": 6198
    },
    {
      "epoch": 0.17861464876390248,
      "grad_norm": 1.0640100240707397,
      "learning_rate": 9.43205798885721e-06,
      "loss": 0.7779,
      "step": 6199
    },
    {
      "epoch": 0.17864346222555177,
      "grad_norm": 1.065335750579834,
      "learning_rate": 9.43184197670406e-06,
      "loss": 0.7889,
      "step": 6200
    },
    {
      "epoch": 0.17867227568720107,
      "grad_norm": 1.0640102624893188,
      "learning_rate": 9.431625925953998e-06,
      "loss": 0.7711,
      "step": 6201
    },
    {
      "epoch": 0.17870108914885036,
      "grad_norm": 1.0918241739273071,
      "learning_rate": 9.431409836608903e-06,
      "loss": 0.7693,
      "step": 6202
    },
    {
      "epoch": 0.17872990261049962,
      "grad_norm": 1.0723768472671509,
      "learning_rate": 9.431193708670659e-06,
      "loss": 0.7855,
      "step": 6203
    },
    {
      "epoch": 0.1787587160721489,
      "grad_norm": 1.0232640504837036,
      "learning_rate": 9.430977542141146e-06,
      "loss": 0.7776,
      "step": 6204
    },
    {
      "epoch": 0.1787875295337982,
      "grad_norm": 1.0144637823104858,
      "learning_rate": 9.430761337022248e-06,
      "loss": 0.7738,
      "step": 6205
    },
    {
      "epoch": 0.17881634299544746,
      "grad_norm": 1.0929114818572998,
      "learning_rate": 9.43054509331585e-06,
      "loss": 0.7982,
      "step": 6206
    },
    {
      "epoch": 0.17884515645709675,
      "grad_norm": 1.0230704545974731,
      "learning_rate": 9.430328811023832e-06,
      "loss": 0.7855,
      "step": 6207
    },
    {
      "epoch": 0.17887396991874605,
      "grad_norm": 1.0739437341690063,
      "learning_rate": 9.430112490148077e-06,
      "loss": 0.798,
      "step": 6208
    },
    {
      "epoch": 0.1789027833803953,
      "grad_norm": 1.069976568222046,
      "learning_rate": 9.429896130690472e-06,
      "loss": 0.8095,
      "step": 6209
    },
    {
      "epoch": 0.1789315968420446,
      "grad_norm": 1.0460186004638672,
      "learning_rate": 9.429679732652901e-06,
      "loss": 0.7867,
      "step": 6210
    },
    {
      "epoch": 0.1789604103036939,
      "grad_norm": 1.0478414297103882,
      "learning_rate": 9.429463296037247e-06,
      "loss": 0.7742,
      "step": 6211
    },
    {
      "epoch": 0.17898922376534318,
      "grad_norm": 1.1047484874725342,
      "learning_rate": 9.429246820845394e-06,
      "loss": 0.8088,
      "step": 6212
    },
    {
      "epoch": 0.17901803722699244,
      "grad_norm": 1.0856608152389526,
      "learning_rate": 9.429030307079233e-06,
      "loss": 0.7881,
      "step": 6213
    },
    {
      "epoch": 0.17904685068864173,
      "grad_norm": 1.6038726568222046,
      "learning_rate": 9.428813754740641e-06,
      "loss": 0.7713,
      "step": 6214
    },
    {
      "epoch": 0.17907566415029103,
      "grad_norm": 1.017137050628662,
      "learning_rate": 9.42859716383151e-06,
      "loss": 0.7833,
      "step": 6215
    },
    {
      "epoch": 0.1791044776119403,
      "grad_norm": 1.1652281284332275,
      "learning_rate": 9.428380534353724e-06,
      "loss": 0.7938,
      "step": 6216
    },
    {
      "epoch": 0.17913329107358958,
      "grad_norm": 1.0756536722183228,
      "learning_rate": 9.428163866309172e-06,
      "loss": 0.7976,
      "step": 6217
    },
    {
      "epoch": 0.17916210453523887,
      "grad_norm": 1.1286383867263794,
      "learning_rate": 9.427947159699738e-06,
      "loss": 0.8118,
      "step": 6218
    },
    {
      "epoch": 0.17919091799688813,
      "grad_norm": 1.2186334133148193,
      "learning_rate": 9.427730414527313e-06,
      "loss": 0.7856,
      "step": 6219
    },
    {
      "epoch": 0.17921973145853742,
      "grad_norm": 1.0788383483886719,
      "learning_rate": 9.427513630793779e-06,
      "loss": 0.7899,
      "step": 6220
    },
    {
      "epoch": 0.17924854492018671,
      "grad_norm": 1.1188427209854126,
      "learning_rate": 9.427296808501028e-06,
      "loss": 0.7824,
      "step": 6221
    },
    {
      "epoch": 0.179277358381836,
      "grad_norm": 1.285854697227478,
      "learning_rate": 9.427079947650948e-06,
      "loss": 0.8019,
      "step": 6222
    },
    {
      "epoch": 0.17930617184348527,
      "grad_norm": 1.0454434156417847,
      "learning_rate": 9.426863048245426e-06,
      "loss": 0.7928,
      "step": 6223
    },
    {
      "epoch": 0.17933498530513456,
      "grad_norm": 1.2050888538360596,
      "learning_rate": 9.426646110286352e-06,
      "loss": 0.7731,
      "step": 6224
    },
    {
      "epoch": 0.17936379876678385,
      "grad_norm": 1.1467245817184448,
      "learning_rate": 9.426429133775616e-06,
      "loss": 0.8146,
      "step": 6225
    },
    {
      "epoch": 0.1793926122284331,
      "grad_norm": 1.0952599048614502,
      "learning_rate": 9.426212118715107e-06,
      "loss": 0.7933,
      "step": 6226
    },
    {
      "epoch": 0.1794214256900824,
      "grad_norm": 1.0592230558395386,
      "learning_rate": 9.425995065106717e-06,
      "loss": 0.7722,
      "step": 6227
    },
    {
      "epoch": 0.1794502391517317,
      "grad_norm": 1.1016347408294678,
      "learning_rate": 9.425777972952332e-06,
      "loss": 0.8157,
      "step": 6228
    },
    {
      "epoch": 0.17947905261338098,
      "grad_norm": 1.0584185123443604,
      "learning_rate": 9.425560842253847e-06,
      "loss": 0.79,
      "step": 6229
    },
    {
      "epoch": 0.17950786607503025,
      "grad_norm": 1.0810492038726807,
      "learning_rate": 9.425343673013149e-06,
      "loss": 0.8052,
      "step": 6230
    },
    {
      "epoch": 0.17953667953667954,
      "grad_norm": 1.0935767889022827,
      "learning_rate": 9.425126465232133e-06,
      "loss": 0.7901,
      "step": 6231
    },
    {
      "epoch": 0.17956549299832883,
      "grad_norm": 1.064342737197876,
      "learning_rate": 9.424909218912689e-06,
      "loss": 0.7998,
      "step": 6232
    },
    {
      "epoch": 0.1795943064599781,
      "grad_norm": 1.2297438383102417,
      "learning_rate": 9.424691934056707e-06,
      "loss": 0.8131,
      "step": 6233
    },
    {
      "epoch": 0.17962311992162738,
      "grad_norm": 1.0852391719818115,
      "learning_rate": 9.424474610666082e-06,
      "loss": 0.7805,
      "step": 6234
    },
    {
      "epoch": 0.17965193338327667,
      "grad_norm": 1.0503441095352173,
      "learning_rate": 9.424257248742707e-06,
      "loss": 0.7728,
      "step": 6235
    },
    {
      "epoch": 0.17968074684492594,
      "grad_norm": 1.081674337387085,
      "learning_rate": 9.424039848288474e-06,
      "loss": 0.7886,
      "step": 6236
    },
    {
      "epoch": 0.17970956030657523,
      "grad_norm": 1.0886952877044678,
      "learning_rate": 9.423822409305275e-06,
      "loss": 0.7868,
      "step": 6237
    },
    {
      "epoch": 0.17973837376822452,
      "grad_norm": 1.1296747922897339,
      "learning_rate": 9.423604931795006e-06,
      "loss": 0.7771,
      "step": 6238
    },
    {
      "epoch": 0.1797671872298738,
      "grad_norm": 1.0398882627487183,
      "learning_rate": 9.423387415759559e-06,
      "loss": 0.7764,
      "step": 6239
    },
    {
      "epoch": 0.17979600069152307,
      "grad_norm": 1.078147053718567,
      "learning_rate": 9.42316986120083e-06,
      "loss": 0.7948,
      "step": 6240
    },
    {
      "epoch": 0.17982481415317236,
      "grad_norm": 1.0229475498199463,
      "learning_rate": 9.422952268120714e-06,
      "loss": 0.8016,
      "step": 6241
    },
    {
      "epoch": 0.17985362761482165,
      "grad_norm": 1.0645698308944702,
      "learning_rate": 9.422734636521104e-06,
      "loss": 0.7699,
      "step": 6242
    },
    {
      "epoch": 0.17988244107647092,
      "grad_norm": 1.0536038875579834,
      "learning_rate": 9.422516966403897e-06,
      "loss": 0.773,
      "step": 6243
    },
    {
      "epoch": 0.1799112545381202,
      "grad_norm": 1.049214482307434,
      "learning_rate": 9.422299257770988e-06,
      "loss": 0.8026,
      "step": 6244
    },
    {
      "epoch": 0.1799400679997695,
      "grad_norm": 1.1156378984451294,
      "learning_rate": 9.422081510624272e-06,
      "loss": 0.791,
      "step": 6245
    },
    {
      "epoch": 0.17996888146141876,
      "grad_norm": 1.0458875894546509,
      "learning_rate": 9.421863724965648e-06,
      "loss": 0.7665,
      "step": 6246
    },
    {
      "epoch": 0.17999769492306805,
      "grad_norm": 1.0208933353424072,
      "learning_rate": 9.421645900797011e-06,
      "loss": 0.8058,
      "step": 6247
    },
    {
      "epoch": 0.18002650838471734,
      "grad_norm": 1.0655394792556763,
      "learning_rate": 9.421428038120258e-06,
      "loss": 0.7875,
      "step": 6248
    },
    {
      "epoch": 0.18005532184636663,
      "grad_norm": 1.0658105611801147,
      "learning_rate": 9.421210136937287e-06,
      "loss": 0.7766,
      "step": 6249
    },
    {
      "epoch": 0.1800841353080159,
      "grad_norm": 1.0120022296905518,
      "learning_rate": 9.420992197249994e-06,
      "loss": 0.774,
      "step": 6250
    },
    {
      "epoch": 0.1801129487696652,
      "grad_norm": 1.0481489896774292,
      "learning_rate": 9.420774219060281e-06,
      "loss": 0.777,
      "step": 6251
    },
    {
      "epoch": 0.18014176223131448,
      "grad_norm": 1.0129306316375732,
      "learning_rate": 9.420556202370041e-06,
      "loss": 0.7951,
      "step": 6252
    },
    {
      "epoch": 0.18017057569296374,
      "grad_norm": 1.041831135749817,
      "learning_rate": 9.420338147181176e-06,
      "loss": 0.7654,
      "step": 6253
    },
    {
      "epoch": 0.18019938915461303,
      "grad_norm": 1.0801242589950562,
      "learning_rate": 9.420120053495585e-06,
      "loss": 0.7831,
      "step": 6254
    },
    {
      "epoch": 0.18022820261626232,
      "grad_norm": 1.0690510272979736,
      "learning_rate": 9.419901921315167e-06,
      "loss": 0.8041,
      "step": 6255
    },
    {
      "epoch": 0.1802570160779116,
      "grad_norm": 2.2268569469451904,
      "learning_rate": 9.41968375064182e-06,
      "loss": 0.7948,
      "step": 6256
    },
    {
      "epoch": 0.18028582953956088,
      "grad_norm": 1.0238938331604004,
      "learning_rate": 9.419465541477448e-06,
      "loss": 0.7513,
      "step": 6257
    },
    {
      "epoch": 0.18031464300121017,
      "grad_norm": 1.0791715383529663,
      "learning_rate": 9.419247293823947e-06,
      "loss": 0.7699,
      "step": 6258
    },
    {
      "epoch": 0.18034345646285946,
      "grad_norm": 1.026982069015503,
      "learning_rate": 9.41902900768322e-06,
      "loss": 0.799,
      "step": 6259
    },
    {
      "epoch": 0.18037226992450872,
      "grad_norm": 1.084758996963501,
      "learning_rate": 9.41881068305717e-06,
      "loss": 0.8226,
      "step": 6260
    },
    {
      "epoch": 0.180401083386158,
      "grad_norm": 1.0258837938308716,
      "learning_rate": 9.418592319947693e-06,
      "loss": 0.7611,
      "step": 6261
    },
    {
      "epoch": 0.1804298968478073,
      "grad_norm": 1.036199927330017,
      "learning_rate": 9.418373918356697e-06,
      "loss": 0.7726,
      "step": 6262
    },
    {
      "epoch": 0.18045871030945657,
      "grad_norm": 1.057078242301941,
      "learning_rate": 9.418155478286077e-06,
      "loss": 0.7901,
      "step": 6263
    },
    {
      "epoch": 0.18048752377110586,
      "grad_norm": 1.0567225217819214,
      "learning_rate": 9.417936999737743e-06,
      "loss": 0.7964,
      "step": 6264
    },
    {
      "epoch": 0.18051633723275515,
      "grad_norm": 1.002243161201477,
      "learning_rate": 9.417718482713591e-06,
      "loss": 0.7715,
      "step": 6265
    },
    {
      "epoch": 0.18054515069440444,
      "grad_norm": 1.0945895910263062,
      "learning_rate": 9.417499927215529e-06,
      "loss": 0.7886,
      "step": 6266
    },
    {
      "epoch": 0.1805739641560537,
      "grad_norm": 1.0242187976837158,
      "learning_rate": 9.417281333245459e-06,
      "loss": 0.7693,
      "step": 6267
    },
    {
      "epoch": 0.180602777617703,
      "grad_norm": 1.0820671319961548,
      "learning_rate": 9.417062700805282e-06,
      "loss": 0.7732,
      "step": 6268
    },
    {
      "epoch": 0.18063159107935228,
      "grad_norm": 1.071204423904419,
      "learning_rate": 9.416844029896907e-06,
      "loss": 0.7936,
      "step": 6269
    },
    {
      "epoch": 0.18066040454100155,
      "grad_norm": 1.0467309951782227,
      "learning_rate": 9.416625320522233e-06,
      "loss": 0.7539,
      "step": 6270
    },
    {
      "epoch": 0.18068921800265084,
      "grad_norm": 1.0688174962997437,
      "learning_rate": 9.41640657268317e-06,
      "loss": 0.8072,
      "step": 6271
    },
    {
      "epoch": 0.18071803146430013,
      "grad_norm": 1.102998971939087,
      "learning_rate": 9.416187786381618e-06,
      "loss": 0.8183,
      "step": 6272
    },
    {
      "epoch": 0.1807468449259494,
      "grad_norm": 1.087449550628662,
      "learning_rate": 9.415968961619488e-06,
      "loss": 0.7953,
      "step": 6273
    },
    {
      "epoch": 0.18077565838759868,
      "grad_norm": 1.1095775365829468,
      "learning_rate": 9.415750098398681e-06,
      "loss": 0.7795,
      "step": 6274
    },
    {
      "epoch": 0.18080447184924797,
      "grad_norm": 1.0419842004776,
      "learning_rate": 9.415531196721105e-06,
      "loss": 0.7897,
      "step": 6275
    },
    {
      "epoch": 0.18083328531089726,
      "grad_norm": 1.0347003936767578,
      "learning_rate": 9.415312256588666e-06,
      "loss": 0.8085,
      "step": 6276
    },
    {
      "epoch": 0.18086209877254653,
      "grad_norm": 1.0856674909591675,
      "learning_rate": 9.41509327800327e-06,
      "loss": 0.8093,
      "step": 6277
    },
    {
      "epoch": 0.18089091223419582,
      "grad_norm": 1.0890167951583862,
      "learning_rate": 9.414874260966827e-06,
      "loss": 0.7875,
      "step": 6278
    },
    {
      "epoch": 0.1809197256958451,
      "grad_norm": 1.1113102436065674,
      "learning_rate": 9.414655205481242e-06,
      "loss": 0.7701,
      "step": 6279
    },
    {
      "epoch": 0.18094853915749437,
      "grad_norm": 1.0353227853775024,
      "learning_rate": 9.414436111548421e-06,
      "loss": 0.8199,
      "step": 6280
    },
    {
      "epoch": 0.18097735261914366,
      "grad_norm": 1.0947988033294678,
      "learning_rate": 9.414216979170276e-06,
      "loss": 0.7989,
      "step": 6281
    },
    {
      "epoch": 0.18100616608079295,
      "grad_norm": 1.0832244157791138,
      "learning_rate": 9.413997808348716e-06,
      "loss": 0.7919,
      "step": 6282
    },
    {
      "epoch": 0.18103497954244224,
      "grad_norm": 1.022678017616272,
      "learning_rate": 9.413778599085644e-06,
      "loss": 0.7773,
      "step": 6283
    },
    {
      "epoch": 0.1810637930040915,
      "grad_norm": 1.098064661026001,
      "learning_rate": 9.413559351382974e-06,
      "loss": 0.7596,
      "step": 6284
    },
    {
      "epoch": 0.1810926064657408,
      "grad_norm": 1.049273133277893,
      "learning_rate": 9.413340065242614e-06,
      "loss": 0.7819,
      "step": 6285
    },
    {
      "epoch": 0.1811214199273901,
      "grad_norm": 1.012290120124817,
      "learning_rate": 9.413120740666473e-06,
      "loss": 0.7716,
      "step": 6286
    },
    {
      "epoch": 0.18115023338903935,
      "grad_norm": 1.0403586626052856,
      "learning_rate": 9.412901377656462e-06,
      "loss": 0.7569,
      "step": 6287
    },
    {
      "epoch": 0.18117904685068864,
      "grad_norm": 1.0559687614440918,
      "learning_rate": 9.412681976214493e-06,
      "loss": 0.7776,
      "step": 6288
    },
    {
      "epoch": 0.18120786031233793,
      "grad_norm": 1.03476083278656,
      "learning_rate": 9.412462536342473e-06,
      "loss": 0.7953,
      "step": 6289
    },
    {
      "epoch": 0.1812366737739872,
      "grad_norm": 1.039061188697815,
      "learning_rate": 9.412243058042316e-06,
      "loss": 0.8044,
      "step": 6290
    },
    {
      "epoch": 0.18126548723563649,
      "grad_norm": 1.0340808629989624,
      "learning_rate": 9.412023541315933e-06,
      "loss": 0.797,
      "step": 6291
    },
    {
      "epoch": 0.18129430069728578,
      "grad_norm": 1.0346059799194336,
      "learning_rate": 9.411803986165233e-06,
      "loss": 0.8171,
      "step": 6292
    },
    {
      "epoch": 0.18132311415893507,
      "grad_norm": 1.005402684211731,
      "learning_rate": 9.411584392592132e-06,
      "loss": 0.7905,
      "step": 6293
    },
    {
      "epoch": 0.18135192762058433,
      "grad_norm": 1.0849523544311523,
      "learning_rate": 9.411364760598542e-06,
      "loss": 0.8054,
      "step": 6294
    },
    {
      "epoch": 0.18138074108223362,
      "grad_norm": 1.0758005380630493,
      "learning_rate": 9.411145090186375e-06,
      "loss": 0.7927,
      "step": 6295
    },
    {
      "epoch": 0.1814095545438829,
      "grad_norm": 1.068381905555725,
      "learning_rate": 9.410925381357542e-06,
      "loss": 0.8123,
      "step": 6296
    },
    {
      "epoch": 0.18143836800553217,
      "grad_norm": 1.0655488967895508,
      "learning_rate": 9.410705634113958e-06,
      "loss": 0.7971,
      "step": 6297
    },
    {
      "epoch": 0.18146718146718147,
      "grad_norm": 0.9918321371078491,
      "learning_rate": 9.410485848457538e-06,
      "loss": 0.7822,
      "step": 6298
    },
    {
      "epoch": 0.18149599492883076,
      "grad_norm": 1.064164161682129,
      "learning_rate": 9.410266024390196e-06,
      "loss": 0.7715,
      "step": 6299
    },
    {
      "epoch": 0.18152480839048002,
      "grad_norm": 1.0504826307296753,
      "learning_rate": 9.410046161913843e-06,
      "loss": 0.8079,
      "step": 6300
    },
    {
      "epoch": 0.1815536218521293,
      "grad_norm": 1.03999924659729,
      "learning_rate": 9.4098262610304e-06,
      "loss": 0.7785,
      "step": 6301
    },
    {
      "epoch": 0.1815824353137786,
      "grad_norm": 1.053736686706543,
      "learning_rate": 9.409606321741776e-06,
      "loss": 0.795,
      "step": 6302
    },
    {
      "epoch": 0.1816112487754279,
      "grad_norm": 1.0114258527755737,
      "learning_rate": 9.409386344049889e-06,
      "loss": 0.7772,
      "step": 6303
    },
    {
      "epoch": 0.18164006223707715,
      "grad_norm": 1.0665918588638306,
      "learning_rate": 9.409166327956656e-06,
      "loss": 0.8145,
      "step": 6304
    },
    {
      "epoch": 0.18166887569872645,
      "grad_norm": 1.1061553955078125,
      "learning_rate": 9.40894627346399e-06,
      "loss": 0.7871,
      "step": 6305
    },
    {
      "epoch": 0.18169768916037574,
      "grad_norm": 1.1263091564178467,
      "learning_rate": 9.40872618057381e-06,
      "loss": 0.8075,
      "step": 6306
    },
    {
      "epoch": 0.181726502622025,
      "grad_norm": 1.0984148979187012,
      "learning_rate": 9.408506049288034e-06,
      "loss": 0.7749,
      "step": 6307
    },
    {
      "epoch": 0.1817553160836743,
      "grad_norm": 1.134050965309143,
      "learning_rate": 9.408285879608574e-06,
      "loss": 0.8027,
      "step": 6308
    },
    {
      "epoch": 0.18178412954532358,
      "grad_norm": 1.1047465801239014,
      "learning_rate": 9.408065671537353e-06,
      "loss": 0.7851,
      "step": 6309
    },
    {
      "epoch": 0.18181294300697287,
      "grad_norm": 1.1436268091201782,
      "learning_rate": 9.407845425076285e-06,
      "loss": 0.7777,
      "step": 6310
    },
    {
      "epoch": 0.18184175646862213,
      "grad_norm": 1.1183863878250122,
      "learning_rate": 9.40762514022729e-06,
      "loss": 0.7908,
      "step": 6311
    },
    {
      "epoch": 0.18187056993027143,
      "grad_norm": 1.0606762170791626,
      "learning_rate": 9.407404816992287e-06,
      "loss": 0.7795,
      "step": 6312
    },
    {
      "epoch": 0.18189938339192072,
      "grad_norm": 1.0688399076461792,
      "learning_rate": 9.407184455373193e-06,
      "loss": 0.7946,
      "step": 6313
    },
    {
      "epoch": 0.18192819685356998,
      "grad_norm": 1.1039351224899292,
      "learning_rate": 9.406964055371928e-06,
      "loss": 0.7681,
      "step": 6314
    },
    {
      "epoch": 0.18195701031521927,
      "grad_norm": 1.049788475036621,
      "learning_rate": 9.406743616990412e-06,
      "loss": 0.7598,
      "step": 6315
    },
    {
      "epoch": 0.18198582377686856,
      "grad_norm": 1.046199083328247,
      "learning_rate": 9.406523140230563e-06,
      "loss": 0.7952,
      "step": 6316
    },
    {
      "epoch": 0.18201463723851782,
      "grad_norm": 1.0659230947494507,
      "learning_rate": 9.406302625094303e-06,
      "loss": 0.7893,
      "step": 6317
    },
    {
      "epoch": 0.18204345070016711,
      "grad_norm": 1.0645908117294312,
      "learning_rate": 9.406082071583553e-06,
      "loss": 0.7837,
      "step": 6318
    },
    {
      "epoch": 0.1820722641618164,
      "grad_norm": 0.987424910068512,
      "learning_rate": 9.405861479700231e-06,
      "loss": 0.78,
      "step": 6319
    },
    {
      "epoch": 0.1821010776234657,
      "grad_norm": 1.0749083757400513,
      "learning_rate": 9.40564084944626e-06,
      "loss": 0.7689,
      "step": 6320
    },
    {
      "epoch": 0.18212989108511496,
      "grad_norm": 1.053194522857666,
      "learning_rate": 9.405420180823561e-06,
      "loss": 0.7799,
      "step": 6321
    },
    {
      "epoch": 0.18215870454676425,
      "grad_norm": 1.045609951019287,
      "learning_rate": 9.405199473834058e-06,
      "loss": 0.7634,
      "step": 6322
    },
    {
      "epoch": 0.18218751800841354,
      "grad_norm": 1.073846459388733,
      "learning_rate": 9.40497872847967e-06,
      "loss": 0.766,
      "step": 6323
    },
    {
      "epoch": 0.1822163314700628,
      "grad_norm": 1.077048659324646,
      "learning_rate": 9.404757944762319e-06,
      "loss": 0.8028,
      "step": 6324
    },
    {
      "epoch": 0.1822451449317121,
      "grad_norm": 1.0597560405731201,
      "learning_rate": 9.404537122683933e-06,
      "loss": 0.794,
      "step": 6325
    },
    {
      "epoch": 0.18227395839336138,
      "grad_norm": 1.0512492656707764,
      "learning_rate": 9.404316262246428e-06,
      "loss": 0.7625,
      "step": 6326
    },
    {
      "epoch": 0.18230277185501065,
      "grad_norm": 1.240513563156128,
      "learning_rate": 9.404095363451733e-06,
      "loss": 0.793,
      "step": 6327
    },
    {
      "epoch": 0.18233158531665994,
      "grad_norm": 1.0464503765106201,
      "learning_rate": 9.403874426301769e-06,
      "loss": 0.7708,
      "step": 6328
    },
    {
      "epoch": 0.18236039877830923,
      "grad_norm": 1.027215838432312,
      "learning_rate": 9.40365345079846e-06,
      "loss": 0.8026,
      "step": 6329
    },
    {
      "epoch": 0.18238921223995852,
      "grad_norm": 1.0617588758468628,
      "learning_rate": 9.403432436943733e-06,
      "loss": 0.7871,
      "step": 6330
    },
    {
      "epoch": 0.18241802570160778,
      "grad_norm": 1.3009974956512451,
      "learning_rate": 9.403211384739512e-06,
      "loss": 0.7692,
      "step": 6331
    },
    {
      "epoch": 0.18244683916325707,
      "grad_norm": 1.0619324445724487,
      "learning_rate": 9.402990294187721e-06,
      "loss": 0.7905,
      "step": 6332
    },
    {
      "epoch": 0.18247565262490636,
      "grad_norm": 1.0279985666275024,
      "learning_rate": 9.402769165290285e-06,
      "loss": 0.7825,
      "step": 6333
    },
    {
      "epoch": 0.18250446608655563,
      "grad_norm": 1.0284959077835083,
      "learning_rate": 9.402547998049132e-06,
      "loss": 0.775,
      "step": 6334
    },
    {
      "epoch": 0.18253327954820492,
      "grad_norm": 1.0795069932937622,
      "learning_rate": 9.402326792466186e-06,
      "loss": 0.7725,
      "step": 6335
    },
    {
      "epoch": 0.1825620930098542,
      "grad_norm": 1.0670777559280396,
      "learning_rate": 9.402105548543375e-06,
      "loss": 0.7929,
      "step": 6336
    },
    {
      "epoch": 0.1825909064715035,
      "grad_norm": 1.0937987565994263,
      "learning_rate": 9.401884266282624e-06,
      "loss": 0.7859,
      "step": 6337
    },
    {
      "epoch": 0.18261971993315276,
      "grad_norm": 1.0946558713912964,
      "learning_rate": 9.401662945685861e-06,
      "loss": 0.7907,
      "step": 6338
    },
    {
      "epoch": 0.18264853339480205,
      "grad_norm": 1.0879570245742798,
      "learning_rate": 9.401441586755017e-06,
      "loss": 0.7912,
      "step": 6339
    },
    {
      "epoch": 0.18267734685645134,
      "grad_norm": 1.0632693767547607,
      "learning_rate": 9.401220189492013e-06,
      "loss": 0.7734,
      "step": 6340
    },
    {
      "epoch": 0.1827061603181006,
      "grad_norm": 1.1125478744506836,
      "learning_rate": 9.400998753898783e-06,
      "loss": 0.8001,
      "step": 6341
    },
    {
      "epoch": 0.1827349737797499,
      "grad_norm": 1.0534993410110474,
      "learning_rate": 9.40077727997725e-06,
      "loss": 0.7938,
      "step": 6342
    },
    {
      "epoch": 0.1827637872413992,
      "grad_norm": 1.064589500427246,
      "learning_rate": 9.400555767729348e-06,
      "loss": 0.7777,
      "step": 6343
    },
    {
      "epoch": 0.18279260070304845,
      "grad_norm": 1.0342082977294922,
      "learning_rate": 9.400334217157006e-06,
      "loss": 0.7788,
      "step": 6344
    },
    {
      "epoch": 0.18282141416469774,
      "grad_norm": 1.0587412118911743,
      "learning_rate": 9.40011262826215e-06,
      "loss": 0.791,
      "step": 6345
    },
    {
      "epoch": 0.18285022762634703,
      "grad_norm": 1.0979803800582886,
      "learning_rate": 9.399891001046712e-06,
      "loss": 0.7702,
      "step": 6346
    },
    {
      "epoch": 0.18287904108799632,
      "grad_norm": 1.0256763696670532,
      "learning_rate": 9.399669335512623e-06,
      "loss": 0.7797,
      "step": 6347
    },
    {
      "epoch": 0.1829078545496456,
      "grad_norm": 1.0919139385223389,
      "learning_rate": 9.399447631661809e-06,
      "loss": 0.7516,
      "step": 6348
    },
    {
      "epoch": 0.18293666801129488,
      "grad_norm": 1.043773889541626,
      "learning_rate": 9.399225889496206e-06,
      "loss": 0.7879,
      "step": 6349
    },
    {
      "epoch": 0.18296548147294417,
      "grad_norm": 1.0511548519134521,
      "learning_rate": 9.399004109017742e-06,
      "loss": 0.7899,
      "step": 6350
    },
    {
      "epoch": 0.18299429493459343,
      "grad_norm": 1.1043907403945923,
      "learning_rate": 9.398782290228351e-06,
      "loss": 0.7978,
      "step": 6351
    },
    {
      "epoch": 0.18302310839624272,
      "grad_norm": 1.0194671154022217,
      "learning_rate": 9.398560433129961e-06,
      "loss": 0.8178,
      "step": 6352
    },
    {
      "epoch": 0.183051921857892,
      "grad_norm": 1.0986183881759644,
      "learning_rate": 9.39833853772451e-06,
      "loss": 0.796,
      "step": 6353
    },
    {
      "epoch": 0.18308073531954128,
      "grad_norm": 0.9839873909950256,
      "learning_rate": 9.398116604013926e-06,
      "loss": 0.8021,
      "step": 6354
    },
    {
      "epoch": 0.18310954878119057,
      "grad_norm": 1.1003239154815674,
      "learning_rate": 9.397894632000143e-06,
      "loss": 0.801,
      "step": 6355
    },
    {
      "epoch": 0.18313836224283986,
      "grad_norm": 1.0240702629089355,
      "learning_rate": 9.397672621685092e-06,
      "loss": 0.7717,
      "step": 6356
    },
    {
      "epoch": 0.18316717570448915,
      "grad_norm": 1.0545744895935059,
      "learning_rate": 9.39745057307071e-06,
      "loss": 0.7905,
      "step": 6357
    },
    {
      "epoch": 0.1831959891661384,
      "grad_norm": 1.0045641660690308,
      "learning_rate": 9.39722848615893e-06,
      "loss": 0.7967,
      "step": 6358
    },
    {
      "epoch": 0.1832248026277877,
      "grad_norm": 1.063974380493164,
      "learning_rate": 9.397006360951684e-06,
      "loss": 0.8064,
      "step": 6359
    },
    {
      "epoch": 0.183253616089437,
      "grad_norm": 1.0708528757095337,
      "learning_rate": 9.396784197450908e-06,
      "loss": 0.8004,
      "step": 6360
    },
    {
      "epoch": 0.18328242955108626,
      "grad_norm": 1.0578445196151733,
      "learning_rate": 9.396561995658538e-06,
      "loss": 0.7958,
      "step": 6361
    },
    {
      "epoch": 0.18331124301273555,
      "grad_norm": 1.0276862382888794,
      "learning_rate": 9.396339755576506e-06,
      "loss": 0.7786,
      "step": 6362
    },
    {
      "epoch": 0.18334005647438484,
      "grad_norm": 1.0368703603744507,
      "learning_rate": 9.396117477206753e-06,
      "loss": 0.7809,
      "step": 6363
    },
    {
      "epoch": 0.18336886993603413,
      "grad_norm": 0.9874593019485474,
      "learning_rate": 9.395895160551208e-06,
      "loss": 0.781,
      "step": 6364
    },
    {
      "epoch": 0.1833976833976834,
      "grad_norm": 1.052291750907898,
      "learning_rate": 9.395672805611814e-06,
      "loss": 0.7859,
      "step": 6365
    },
    {
      "epoch": 0.18342649685933268,
      "grad_norm": 0.9978642463684082,
      "learning_rate": 9.3954504123905e-06,
      "loss": 0.794,
      "step": 6366
    },
    {
      "epoch": 0.18345531032098197,
      "grad_norm": 1.0085138082504272,
      "learning_rate": 9.395227980889208e-06,
      "loss": 0.7786,
      "step": 6367
    },
    {
      "epoch": 0.18348412378263124,
      "grad_norm": 1.0456149578094482,
      "learning_rate": 9.395005511109873e-06,
      "loss": 0.7827,
      "step": 6368
    },
    {
      "epoch": 0.18351293724428053,
      "grad_norm": 1.0445739030838013,
      "learning_rate": 9.394783003054435e-06,
      "loss": 0.789,
      "step": 6369
    },
    {
      "epoch": 0.18354175070592982,
      "grad_norm": 1.016610026359558,
      "learning_rate": 9.394560456724831e-06,
      "loss": 0.765,
      "step": 6370
    },
    {
      "epoch": 0.18357056416757908,
      "grad_norm": 1.032198429107666,
      "learning_rate": 9.394337872122997e-06,
      "loss": 0.7664,
      "step": 6371
    },
    {
      "epoch": 0.18359937762922837,
      "grad_norm": 1.059191107749939,
      "learning_rate": 9.394115249250873e-06,
      "loss": 0.7592,
      "step": 6372
    },
    {
      "epoch": 0.18362819109087766,
      "grad_norm": 1.0623395442962646,
      "learning_rate": 9.393892588110398e-06,
      "loss": 0.7646,
      "step": 6373
    },
    {
      "epoch": 0.18365700455252695,
      "grad_norm": 1.0294822454452515,
      "learning_rate": 9.39366988870351e-06,
      "loss": 0.8001,
      "step": 6374
    },
    {
      "epoch": 0.18368581801417622,
      "grad_norm": 1.1358163356781006,
      "learning_rate": 9.393447151032152e-06,
      "loss": 0.799,
      "step": 6375
    },
    {
      "epoch": 0.1837146314758255,
      "grad_norm": 1.0257445573806763,
      "learning_rate": 9.393224375098258e-06,
      "loss": 0.7685,
      "step": 6376
    },
    {
      "epoch": 0.1837434449374748,
      "grad_norm": 0.9917792081832886,
      "learning_rate": 9.393001560903772e-06,
      "loss": 0.7676,
      "step": 6377
    },
    {
      "epoch": 0.18377225839912406,
      "grad_norm": 1.0725840330123901,
      "learning_rate": 9.392778708450635e-06,
      "loss": 0.7804,
      "step": 6378
    },
    {
      "epoch": 0.18380107186077335,
      "grad_norm": 1.1909273862838745,
      "learning_rate": 9.392555817740785e-06,
      "loss": 0.7769,
      "step": 6379
    },
    {
      "epoch": 0.18382988532242264,
      "grad_norm": 1.0431838035583496,
      "learning_rate": 9.392332888776165e-06,
      "loss": 0.7717,
      "step": 6380
    },
    {
      "epoch": 0.1838586987840719,
      "grad_norm": 1.076706051826477,
      "learning_rate": 9.392109921558718e-06,
      "loss": 0.7546,
      "step": 6381
    },
    {
      "epoch": 0.1838875122457212,
      "grad_norm": 1.1281847953796387,
      "learning_rate": 9.391886916090382e-06,
      "loss": 0.7662,
      "step": 6382
    },
    {
      "epoch": 0.1839163257073705,
      "grad_norm": 1.1057960987091064,
      "learning_rate": 9.391663872373102e-06,
      "loss": 0.7591,
      "step": 6383
    },
    {
      "epoch": 0.18394513916901978,
      "grad_norm": 1.0552630424499512,
      "learning_rate": 9.39144079040882e-06,
      "loss": 0.7814,
      "step": 6384
    },
    {
      "epoch": 0.18397395263066904,
      "grad_norm": 1.052689790725708,
      "learning_rate": 9.391217670199479e-06,
      "loss": 0.7832,
      "step": 6385
    },
    {
      "epoch": 0.18400276609231833,
      "grad_norm": 1.1137421131134033,
      "learning_rate": 9.390994511747021e-06,
      "loss": 0.7955,
      "step": 6386
    },
    {
      "epoch": 0.18403157955396762,
      "grad_norm": 1.057292103767395,
      "learning_rate": 9.39077131505339e-06,
      "loss": 0.7745,
      "step": 6387
    },
    {
      "epoch": 0.18406039301561689,
      "grad_norm": 1.0856821537017822,
      "learning_rate": 9.390548080120532e-06,
      "loss": 0.816,
      "step": 6388
    },
    {
      "epoch": 0.18408920647726618,
      "grad_norm": 1.083932638168335,
      "learning_rate": 9.390324806950385e-06,
      "loss": 0.7779,
      "step": 6389
    },
    {
      "epoch": 0.18411801993891547,
      "grad_norm": 1.0049529075622559,
      "learning_rate": 9.3901014955449e-06,
      "loss": 0.7808,
      "step": 6390
    },
    {
      "epoch": 0.18414683340056473,
      "grad_norm": 1.1623103618621826,
      "learning_rate": 9.389878145906022e-06,
      "loss": 0.7842,
      "step": 6391
    },
    {
      "epoch": 0.18417564686221402,
      "grad_norm": 1.035592794418335,
      "learning_rate": 9.389654758035692e-06,
      "loss": 0.7855,
      "step": 6392
    },
    {
      "epoch": 0.1842044603238633,
      "grad_norm": 1.071687936782837,
      "learning_rate": 9.389431331935854e-06,
      "loss": 0.8098,
      "step": 6393
    },
    {
      "epoch": 0.1842332737855126,
      "grad_norm": 1.0982054471969604,
      "learning_rate": 9.38920786760846e-06,
      "loss": 0.7753,
      "step": 6394
    },
    {
      "epoch": 0.18426208724716187,
      "grad_norm": 1.1021724939346313,
      "learning_rate": 9.388984365055453e-06,
      "loss": 0.8033,
      "step": 6395
    },
    {
      "epoch": 0.18429090070881116,
      "grad_norm": 1.0351117849349976,
      "learning_rate": 9.388760824278779e-06,
      "loss": 0.7783,
      "step": 6396
    },
    {
      "epoch": 0.18431971417046045,
      "grad_norm": 1.1043446063995361,
      "learning_rate": 9.388537245280385e-06,
      "loss": 0.761,
      "step": 6397
    },
    {
      "epoch": 0.1843485276321097,
      "grad_norm": 1.0144003629684448,
      "learning_rate": 9.388313628062218e-06,
      "loss": 0.7661,
      "step": 6398
    },
    {
      "epoch": 0.184377341093759,
      "grad_norm": 1.050315022468567,
      "learning_rate": 9.388089972626228e-06,
      "loss": 0.7941,
      "step": 6399
    },
    {
      "epoch": 0.1844061545554083,
      "grad_norm": 1.025739312171936,
      "learning_rate": 9.387866278974361e-06,
      "loss": 0.7975,
      "step": 6400
    },
    {
      "epoch": 0.18443496801705758,
      "grad_norm": 1.0051761865615845,
      "learning_rate": 9.387642547108562e-06,
      "loss": 0.7721,
      "step": 6401
    },
    {
      "epoch": 0.18446378147870685,
      "grad_norm": 1.0537952184677124,
      "learning_rate": 9.387418777030785e-06,
      "loss": 0.7878,
      "step": 6402
    },
    {
      "epoch": 0.18449259494035614,
      "grad_norm": 1.010382890701294,
      "learning_rate": 9.387194968742975e-06,
      "loss": 0.7896,
      "step": 6403
    },
    {
      "epoch": 0.18452140840200543,
      "grad_norm": 1.0401619672775269,
      "learning_rate": 9.386971122247082e-06,
      "loss": 0.7908,
      "step": 6404
    },
    {
      "epoch": 0.1845502218636547,
      "grad_norm": 0.9969756603240967,
      "learning_rate": 9.386747237545057e-06,
      "loss": 0.7542,
      "step": 6405
    },
    {
      "epoch": 0.18457903532530398,
      "grad_norm": 1.1009879112243652,
      "learning_rate": 9.386523314638849e-06,
      "loss": 0.7777,
      "step": 6406
    },
    {
      "epoch": 0.18460784878695327,
      "grad_norm": 1.0807487964630127,
      "learning_rate": 9.386299353530406e-06,
      "loss": 0.7909,
      "step": 6407
    },
    {
      "epoch": 0.18463666224860253,
      "grad_norm": 1.0589131116867065,
      "learning_rate": 9.386075354221682e-06,
      "loss": 0.7908,
      "step": 6408
    },
    {
      "epoch": 0.18466547571025183,
      "grad_norm": 1.0347048044204712,
      "learning_rate": 9.385851316714627e-06,
      "loss": 0.7648,
      "step": 6409
    },
    {
      "epoch": 0.18469428917190112,
      "grad_norm": 1.1412752866744995,
      "learning_rate": 9.385627241011188e-06,
      "loss": 0.8126,
      "step": 6410
    },
    {
      "epoch": 0.1847231026335504,
      "grad_norm": 1.0034455060958862,
      "learning_rate": 9.385403127113323e-06,
      "loss": 0.778,
      "step": 6411
    },
    {
      "epoch": 0.18475191609519967,
      "grad_norm": 1.0243829488754272,
      "learning_rate": 9.38517897502298e-06,
      "loss": 0.7965,
      "step": 6412
    },
    {
      "epoch": 0.18478072955684896,
      "grad_norm": 1.0747230052947998,
      "learning_rate": 9.38495478474211e-06,
      "loss": 0.7635,
      "step": 6413
    },
    {
      "epoch": 0.18480954301849825,
      "grad_norm": 1.1078232526779175,
      "learning_rate": 9.384730556272668e-06,
      "loss": 0.7791,
      "step": 6414
    },
    {
      "epoch": 0.18483835648014751,
      "grad_norm": 1.0667227506637573,
      "learning_rate": 9.384506289616607e-06,
      "loss": 0.7971,
      "step": 6415
    },
    {
      "epoch": 0.1848671699417968,
      "grad_norm": 1.0469430685043335,
      "learning_rate": 9.384281984775879e-06,
      "loss": 0.7767,
      "step": 6416
    },
    {
      "epoch": 0.1848959834034461,
      "grad_norm": 1.0961858034133911,
      "learning_rate": 9.384057641752435e-06,
      "loss": 0.7775,
      "step": 6417
    },
    {
      "epoch": 0.18492479686509536,
      "grad_norm": 0.9880399107933044,
      "learning_rate": 9.383833260548234e-06,
      "loss": 0.7684,
      "step": 6418
    },
    {
      "epoch": 0.18495361032674465,
      "grad_norm": 1.0449517965316772,
      "learning_rate": 9.383608841165226e-06,
      "loss": 0.7736,
      "step": 6419
    },
    {
      "epoch": 0.18498242378839394,
      "grad_norm": 1.0520634651184082,
      "learning_rate": 9.383384383605369e-06,
      "loss": 0.7641,
      "step": 6420
    },
    {
      "epoch": 0.18501123725004323,
      "grad_norm": 1.0761451721191406,
      "learning_rate": 9.383159887870614e-06,
      "loss": 0.7823,
      "step": 6421
    },
    {
      "epoch": 0.1850400507116925,
      "grad_norm": 1.0246264934539795,
      "learning_rate": 9.382935353962919e-06,
      "loss": 0.7997,
      "step": 6422
    },
    {
      "epoch": 0.18506886417334178,
      "grad_norm": 1.0497987270355225,
      "learning_rate": 9.382710781884236e-06,
      "loss": 0.7911,
      "step": 6423
    },
    {
      "epoch": 0.18509767763499108,
      "grad_norm": 1.1085284948349,
      "learning_rate": 9.382486171636526e-06,
      "loss": 0.7835,
      "step": 6424
    },
    {
      "epoch": 0.18512649109664034,
      "grad_norm": 1.0138521194458008,
      "learning_rate": 9.38226152322174e-06,
      "loss": 0.7892,
      "step": 6425
    },
    {
      "epoch": 0.18515530455828963,
      "grad_norm": 1.0396991968154907,
      "learning_rate": 9.38203683664184e-06,
      "loss": 0.7881,
      "step": 6426
    },
    {
      "epoch": 0.18518411801993892,
      "grad_norm": 1.0146931409835815,
      "learning_rate": 9.381812111898778e-06,
      "loss": 0.7605,
      "step": 6427
    },
    {
      "epoch": 0.1852129314815882,
      "grad_norm": 1.0429881811141968,
      "learning_rate": 9.381587348994513e-06,
      "loss": 0.7848,
      "step": 6428
    },
    {
      "epoch": 0.18524174494323747,
      "grad_norm": 1.0289781093597412,
      "learning_rate": 9.381362547931001e-06,
      "loss": 0.8089,
      "step": 6429
    },
    {
      "epoch": 0.18527055840488676,
      "grad_norm": 1.0933406352996826,
      "learning_rate": 9.381137708710202e-06,
      "loss": 0.8002,
      "step": 6430
    },
    {
      "epoch": 0.18529937186653606,
      "grad_norm": 0.9985471963882446,
      "learning_rate": 9.380912831334072e-06,
      "loss": 0.7935,
      "step": 6431
    },
    {
      "epoch": 0.18532818532818532,
      "grad_norm": 1.0979788303375244,
      "learning_rate": 9.380687915804571e-06,
      "loss": 0.7573,
      "step": 6432
    },
    {
      "epoch": 0.1853569987898346,
      "grad_norm": 1.081431269645691,
      "learning_rate": 9.380462962123658e-06,
      "loss": 0.7871,
      "step": 6433
    },
    {
      "epoch": 0.1853858122514839,
      "grad_norm": 1.0246838331222534,
      "learning_rate": 9.380237970293292e-06,
      "loss": 0.7434,
      "step": 6434
    },
    {
      "epoch": 0.18541462571313316,
      "grad_norm": 1.0364406108856201,
      "learning_rate": 9.38001294031543e-06,
      "loss": 0.7754,
      "step": 6435
    },
    {
      "epoch": 0.18544343917478245,
      "grad_norm": 1.0747995376586914,
      "learning_rate": 9.379787872192034e-06,
      "loss": 0.8,
      "step": 6436
    },
    {
      "epoch": 0.18547225263643174,
      "grad_norm": 1.0465744733810425,
      "learning_rate": 9.379562765925065e-06,
      "loss": 0.7552,
      "step": 6437
    },
    {
      "epoch": 0.18550106609808104,
      "grad_norm": 1.0906858444213867,
      "learning_rate": 9.379337621516483e-06,
      "loss": 0.7692,
      "step": 6438
    },
    {
      "epoch": 0.1855298795597303,
      "grad_norm": 1.1341462135314941,
      "learning_rate": 9.379112438968246e-06,
      "loss": 0.7714,
      "step": 6439
    },
    {
      "epoch": 0.1855586930213796,
      "grad_norm": 1.0527632236480713,
      "learning_rate": 9.378887218282317e-06,
      "loss": 0.7762,
      "step": 6440
    },
    {
      "epoch": 0.18558750648302888,
      "grad_norm": 1.093814492225647,
      "learning_rate": 9.378661959460659e-06,
      "loss": 0.7934,
      "step": 6441
    },
    {
      "epoch": 0.18561631994467814,
      "grad_norm": 1.0478464365005493,
      "learning_rate": 9.378436662505232e-06,
      "loss": 0.7854,
      "step": 6442
    },
    {
      "epoch": 0.18564513340632743,
      "grad_norm": 1.041983723640442,
      "learning_rate": 9.378211327417998e-06,
      "loss": 0.7899,
      "step": 6443
    },
    {
      "epoch": 0.18567394686797672,
      "grad_norm": 1.0169621706008911,
      "learning_rate": 9.37798595420092e-06,
      "loss": 0.7868,
      "step": 6444
    },
    {
      "epoch": 0.185702760329626,
      "grad_norm": 1.0064603090286255,
      "learning_rate": 9.377760542855963e-06,
      "loss": 0.756,
      "step": 6445
    },
    {
      "epoch": 0.18573157379127528,
      "grad_norm": 1.0184844732284546,
      "learning_rate": 9.377535093385086e-06,
      "loss": 0.7945,
      "step": 6446
    },
    {
      "epoch": 0.18576038725292457,
      "grad_norm": 1.0307257175445557,
      "learning_rate": 9.377309605790255e-06,
      "loss": 0.777,
      "step": 6447
    },
    {
      "epoch": 0.18578920071457386,
      "grad_norm": 1.0228192806243896,
      "learning_rate": 9.377084080073431e-06,
      "loss": 0.7675,
      "step": 6448
    },
    {
      "epoch": 0.18581801417622312,
      "grad_norm": 1.0510013103485107,
      "learning_rate": 9.376858516236583e-06,
      "loss": 0.7903,
      "step": 6449
    },
    {
      "epoch": 0.1858468276378724,
      "grad_norm": 1.0116446018218994,
      "learning_rate": 9.37663291428167e-06,
      "loss": 0.7744,
      "step": 6450
    },
    {
      "epoch": 0.1858756410995217,
      "grad_norm": 1.035029411315918,
      "learning_rate": 9.37640727421066e-06,
      "loss": 0.7825,
      "step": 6451
    },
    {
      "epoch": 0.18590445456117097,
      "grad_norm": 1.0104080438613892,
      "learning_rate": 9.37618159602552e-06,
      "loss": 0.7785,
      "step": 6452
    },
    {
      "epoch": 0.18593326802282026,
      "grad_norm": 1.0477720499038696,
      "learning_rate": 9.375955879728209e-06,
      "loss": 0.8009,
      "step": 6453
    },
    {
      "epoch": 0.18596208148446955,
      "grad_norm": 1.0591366291046143,
      "learning_rate": 9.375730125320699e-06,
      "loss": 0.7912,
      "step": 6454
    },
    {
      "epoch": 0.18599089494611884,
      "grad_norm": 1.0210968255996704,
      "learning_rate": 9.375504332804952e-06,
      "loss": 0.7956,
      "step": 6455
    },
    {
      "epoch": 0.1860197084077681,
      "grad_norm": 0.9945021271705627,
      "learning_rate": 9.375278502182939e-06,
      "loss": 0.7721,
      "step": 6456
    },
    {
      "epoch": 0.1860485218694174,
      "grad_norm": 1.1490898132324219,
      "learning_rate": 9.37505263345662e-06,
      "loss": 0.7643,
      "step": 6457
    },
    {
      "epoch": 0.18607733533106668,
      "grad_norm": 1.0357997417449951,
      "learning_rate": 9.37482672662797e-06,
      "loss": 0.7795,
      "step": 6458
    },
    {
      "epoch": 0.18610614879271595,
      "grad_norm": 1.0485671758651733,
      "learning_rate": 9.37460078169895e-06,
      "loss": 0.7818,
      "step": 6459
    },
    {
      "epoch": 0.18613496225436524,
      "grad_norm": 1.0626742839813232,
      "learning_rate": 9.37437479867153e-06,
      "loss": 0.7715,
      "step": 6460
    },
    {
      "epoch": 0.18616377571601453,
      "grad_norm": 1.0293364524841309,
      "learning_rate": 9.37414877754768e-06,
      "loss": 0.7718,
      "step": 6461
    },
    {
      "epoch": 0.1861925891776638,
      "grad_norm": 1.0650615692138672,
      "learning_rate": 9.373922718329366e-06,
      "loss": 0.8071,
      "step": 6462
    },
    {
      "epoch": 0.18622140263931308,
      "grad_norm": 1.1080249547958374,
      "learning_rate": 9.373696621018557e-06,
      "loss": 0.7781,
      "step": 6463
    },
    {
      "epoch": 0.18625021610096237,
      "grad_norm": 1.0887668132781982,
      "learning_rate": 9.373470485617222e-06,
      "loss": 0.7969,
      "step": 6464
    },
    {
      "epoch": 0.18627902956261166,
      "grad_norm": 1.0445061922073364,
      "learning_rate": 9.373244312127332e-06,
      "loss": 0.8011,
      "step": 6465
    },
    {
      "epoch": 0.18630784302426093,
      "grad_norm": 1.0767738819122314,
      "learning_rate": 9.373018100550855e-06,
      "loss": 0.7926,
      "step": 6466
    },
    {
      "epoch": 0.18633665648591022,
      "grad_norm": 1.1122467517852783,
      "learning_rate": 9.372791850889763e-06,
      "loss": 0.7804,
      "step": 6467
    },
    {
      "epoch": 0.1863654699475595,
      "grad_norm": 1.0028338432312012,
      "learning_rate": 9.372565563146025e-06,
      "loss": 0.7663,
      "step": 6468
    },
    {
      "epoch": 0.18639428340920877,
      "grad_norm": 1.168790340423584,
      "learning_rate": 9.37233923732161e-06,
      "loss": 0.8237,
      "step": 6469
    },
    {
      "epoch": 0.18642309687085806,
      "grad_norm": 1.0784672498703003,
      "learning_rate": 9.372112873418494e-06,
      "loss": 0.8058,
      "step": 6470
    },
    {
      "epoch": 0.18645191033250735,
      "grad_norm": 1.0452719926834106,
      "learning_rate": 9.371886471438643e-06,
      "loss": 0.7937,
      "step": 6471
    },
    {
      "epoch": 0.18648072379415662,
      "grad_norm": 1.123640775680542,
      "learning_rate": 9.37166003138403e-06,
      "loss": 0.7839,
      "step": 6472
    },
    {
      "epoch": 0.1865095372558059,
      "grad_norm": 1.042283535003662,
      "learning_rate": 9.371433553256632e-06,
      "loss": 0.7834,
      "step": 6473
    },
    {
      "epoch": 0.1865383507174552,
      "grad_norm": 1.0370956659317017,
      "learning_rate": 9.371207037058415e-06,
      "loss": 0.7689,
      "step": 6474
    },
    {
      "epoch": 0.1865671641791045,
      "grad_norm": 1.0685040950775146,
      "learning_rate": 9.370980482791354e-06,
      "loss": 0.7891,
      "step": 6475
    },
    {
      "epoch": 0.18659597764075375,
      "grad_norm": 0.9959927797317505,
      "learning_rate": 9.370753890457424e-06,
      "loss": 0.7597,
      "step": 6476
    },
    {
      "epoch": 0.18662479110240304,
      "grad_norm": 1.1029064655303955,
      "learning_rate": 9.370527260058595e-06,
      "loss": 0.7862,
      "step": 6477
    },
    {
      "epoch": 0.18665360456405233,
      "grad_norm": 1.0230650901794434,
      "learning_rate": 9.370300591596844e-06,
      "loss": 0.8052,
      "step": 6478
    },
    {
      "epoch": 0.1866824180257016,
      "grad_norm": 1.0437020063400269,
      "learning_rate": 9.370073885074144e-06,
      "loss": 0.7799,
      "step": 6479
    },
    {
      "epoch": 0.1867112314873509,
      "grad_norm": 0.9949324131011963,
      "learning_rate": 9.369847140492467e-06,
      "loss": 0.7663,
      "step": 6480
    },
    {
      "epoch": 0.18674004494900018,
      "grad_norm": 1.0531476736068726,
      "learning_rate": 9.36962035785379e-06,
      "loss": 0.7962,
      "step": 6481
    },
    {
      "epoch": 0.18676885841064947,
      "grad_norm": 1.0367622375488281,
      "learning_rate": 9.36939353716009e-06,
      "loss": 0.7636,
      "step": 6482
    },
    {
      "epoch": 0.18679767187229873,
      "grad_norm": 1.0291471481323242,
      "learning_rate": 9.369166678413336e-06,
      "loss": 0.7894,
      "step": 6483
    },
    {
      "epoch": 0.18682648533394802,
      "grad_norm": 1.0138760805130005,
      "learning_rate": 9.368939781615511e-06,
      "loss": 0.7954,
      "step": 6484
    },
    {
      "epoch": 0.1868552987955973,
      "grad_norm": 0.9808580279350281,
      "learning_rate": 9.368712846768588e-06,
      "loss": 0.7891,
      "step": 6485
    },
    {
      "epoch": 0.18688411225724658,
      "grad_norm": 1.0641093254089355,
      "learning_rate": 9.368485873874543e-06,
      "loss": 0.7572,
      "step": 6486
    },
    {
      "epoch": 0.18691292571889587,
      "grad_norm": 1.0530366897583008,
      "learning_rate": 9.36825886293535e-06,
      "loss": 0.7869,
      "step": 6487
    },
    {
      "epoch": 0.18694173918054516,
      "grad_norm": 0.9973617196083069,
      "learning_rate": 9.368031813952994e-06,
      "loss": 0.7838,
      "step": 6488
    },
    {
      "epoch": 0.18697055264219442,
      "grad_norm": 1.0213165283203125,
      "learning_rate": 9.367804726929443e-06,
      "loss": 0.7725,
      "step": 6489
    },
    {
      "epoch": 0.1869993661038437,
      "grad_norm": 1.0326157808303833,
      "learning_rate": 9.367577601866681e-06,
      "loss": 0.7694,
      "step": 6490
    },
    {
      "epoch": 0.187028179565493,
      "grad_norm": 0.9897106289863586,
      "learning_rate": 9.367350438766683e-06,
      "loss": 0.7922,
      "step": 6491
    },
    {
      "epoch": 0.1870569930271423,
      "grad_norm": 1.0111478567123413,
      "learning_rate": 9.36712323763143e-06,
      "loss": 0.7958,
      "step": 6492
    },
    {
      "epoch": 0.18708580648879156,
      "grad_norm": 1.0233994722366333,
      "learning_rate": 9.366895998462897e-06,
      "loss": 0.7879,
      "step": 6493
    },
    {
      "epoch": 0.18711461995044085,
      "grad_norm": 1.051548957824707,
      "learning_rate": 9.366668721263066e-06,
      "loss": 0.7924,
      "step": 6494
    },
    {
      "epoch": 0.18714343341209014,
      "grad_norm": 1.0240278244018555,
      "learning_rate": 9.366441406033916e-06,
      "loss": 0.7625,
      "step": 6495
    },
    {
      "epoch": 0.1871722468737394,
      "grad_norm": 1.0231600999832153,
      "learning_rate": 9.366214052777427e-06,
      "loss": 0.7865,
      "step": 6496
    },
    {
      "epoch": 0.1872010603353887,
      "grad_norm": 1.102810263633728,
      "learning_rate": 9.365986661495576e-06,
      "loss": 0.7892,
      "step": 6497
    },
    {
      "epoch": 0.18722987379703798,
      "grad_norm": 0.9744277000427246,
      "learning_rate": 9.365759232190347e-06,
      "loss": 0.7815,
      "step": 6498
    },
    {
      "epoch": 0.18725868725868725,
      "grad_norm": 1.0344213247299194,
      "learning_rate": 9.36553176486372e-06,
      "loss": 0.797,
      "step": 6499
    },
    {
      "epoch": 0.18728750072033654,
      "grad_norm": 1.1495659351348877,
      "learning_rate": 9.365304259517674e-06,
      "loss": 0.7873,
      "step": 6500
    },
    {
      "epoch": 0.18731631418198583,
      "grad_norm": 1.1119401454925537,
      "learning_rate": 9.36507671615419e-06,
      "loss": 0.7765,
      "step": 6501
    },
    {
      "epoch": 0.18734512764363512,
      "grad_norm": 1.0526429414749146,
      "learning_rate": 9.364849134775256e-06,
      "loss": 0.7723,
      "step": 6502
    },
    {
      "epoch": 0.18737394110528438,
      "grad_norm": 1.0698561668395996,
      "learning_rate": 9.364621515382846e-06,
      "loss": 0.7698,
      "step": 6503
    },
    {
      "epoch": 0.18740275456693367,
      "grad_norm": 1.0565555095672607,
      "learning_rate": 9.364393857978946e-06,
      "loss": 0.7775,
      "step": 6504
    },
    {
      "epoch": 0.18743156802858296,
      "grad_norm": 0.9766888618469238,
      "learning_rate": 9.364166162565538e-06,
      "loss": 0.7753,
      "step": 6505
    },
    {
      "epoch": 0.18746038149023223,
      "grad_norm": 1.0828479528427124,
      "learning_rate": 9.363938429144605e-06,
      "loss": 0.8069,
      "step": 6506
    },
    {
      "epoch": 0.18748919495188152,
      "grad_norm": 1.047277569770813,
      "learning_rate": 9.363710657718133e-06,
      "loss": 0.7631,
      "step": 6507
    },
    {
      "epoch": 0.1875180084135308,
      "grad_norm": 1.1190721988677979,
      "learning_rate": 9.363482848288101e-06,
      "loss": 0.7897,
      "step": 6508
    },
    {
      "epoch": 0.1875468218751801,
      "grad_norm": 1.0246487855911255,
      "learning_rate": 9.363255000856495e-06,
      "loss": 0.7698,
      "step": 6509
    },
    {
      "epoch": 0.18757563533682936,
      "grad_norm": 1.1449565887451172,
      "learning_rate": 9.3630271154253e-06,
      "loss": 0.7689,
      "step": 6510
    },
    {
      "epoch": 0.18760444879847865,
      "grad_norm": 1.0702859163284302,
      "learning_rate": 9.362799191996501e-06,
      "loss": 0.7878,
      "step": 6511
    },
    {
      "epoch": 0.18763326226012794,
      "grad_norm": 1.0127664804458618,
      "learning_rate": 9.362571230572083e-06,
      "loss": 0.7747,
      "step": 6512
    },
    {
      "epoch": 0.1876620757217772,
      "grad_norm": 1.0431112051010132,
      "learning_rate": 9.36234323115403e-06,
      "loss": 0.7941,
      "step": 6513
    },
    {
      "epoch": 0.1876908891834265,
      "grad_norm": 1.0449278354644775,
      "learning_rate": 9.362115193744329e-06,
      "loss": 0.7692,
      "step": 6514
    },
    {
      "epoch": 0.1877197026450758,
      "grad_norm": 1.0202423334121704,
      "learning_rate": 9.361887118344962e-06,
      "loss": 0.7736,
      "step": 6515
    },
    {
      "epoch": 0.18774851610672505,
      "grad_norm": 1.0476891994476318,
      "learning_rate": 9.361659004957921e-06,
      "loss": 0.7794,
      "step": 6516
    },
    {
      "epoch": 0.18777732956837434,
      "grad_norm": 1.0275459289550781,
      "learning_rate": 9.36143085358519e-06,
      "loss": 0.7891,
      "step": 6517
    },
    {
      "epoch": 0.18780614303002363,
      "grad_norm": 1.0028504133224487,
      "learning_rate": 9.361202664228755e-06,
      "loss": 0.7887,
      "step": 6518
    },
    {
      "epoch": 0.18783495649167292,
      "grad_norm": 1.0263649225234985,
      "learning_rate": 9.360974436890603e-06,
      "loss": 0.7842,
      "step": 6519
    },
    {
      "epoch": 0.18786376995332218,
      "grad_norm": 1.031217098236084,
      "learning_rate": 9.360746171572726e-06,
      "loss": 0.778,
      "step": 6520
    },
    {
      "epoch": 0.18789258341497148,
      "grad_norm": 1.0238285064697266,
      "learning_rate": 9.360517868277109e-06,
      "loss": 0.7888,
      "step": 6521
    },
    {
      "epoch": 0.18792139687662077,
      "grad_norm": 1.0077773332595825,
      "learning_rate": 9.360289527005738e-06,
      "loss": 0.7584,
      "step": 6522
    },
    {
      "epoch": 0.18795021033827003,
      "grad_norm": 1.017087697982788,
      "learning_rate": 9.360061147760607e-06,
      "loss": 0.8007,
      "step": 6523
    },
    {
      "epoch": 0.18797902379991932,
      "grad_norm": 1.0637446641921997,
      "learning_rate": 9.3598327305437e-06,
      "loss": 0.7827,
      "step": 6524
    },
    {
      "epoch": 0.1880078372615686,
      "grad_norm": 1.0541648864746094,
      "learning_rate": 9.359604275357008e-06,
      "loss": 0.7879,
      "step": 6525
    },
    {
      "epoch": 0.18803665072321787,
      "grad_norm": 1.0447559356689453,
      "learning_rate": 9.359375782202521e-06,
      "loss": 0.774,
      "step": 6526
    },
    {
      "epoch": 0.18806546418486716,
      "grad_norm": 1.0099793672561646,
      "learning_rate": 9.35914725108223e-06,
      "loss": 0.7892,
      "step": 6527
    },
    {
      "epoch": 0.18809427764651646,
      "grad_norm": 1.067922830581665,
      "learning_rate": 9.358918681998122e-06,
      "loss": 0.8089,
      "step": 6528
    },
    {
      "epoch": 0.18812309110816575,
      "grad_norm": 0.9885058999061584,
      "learning_rate": 9.35869007495219e-06,
      "loss": 0.7821,
      "step": 6529
    },
    {
      "epoch": 0.188151904569815,
      "grad_norm": 0.9987288117408752,
      "learning_rate": 9.358461429946425e-06,
      "loss": 0.7757,
      "step": 6530
    },
    {
      "epoch": 0.1881807180314643,
      "grad_norm": 1.0167242288589478,
      "learning_rate": 9.358232746982819e-06,
      "loss": 0.8001,
      "step": 6531
    },
    {
      "epoch": 0.1882095314931136,
      "grad_norm": 1.0069594383239746,
      "learning_rate": 9.35800402606336e-06,
      "loss": 0.7668,
      "step": 6532
    },
    {
      "epoch": 0.18823834495476285,
      "grad_norm": 1.080145001411438,
      "learning_rate": 9.357775267190043e-06,
      "loss": 0.7955,
      "step": 6533
    },
    {
      "epoch": 0.18826715841641214,
      "grad_norm": 1.0157155990600586,
      "learning_rate": 9.35754647036486e-06,
      "loss": 0.7593,
      "step": 6534
    },
    {
      "epoch": 0.18829597187806144,
      "grad_norm": 1.0119088888168335,
      "learning_rate": 9.357317635589804e-06,
      "loss": 0.8015,
      "step": 6535
    },
    {
      "epoch": 0.18832478533971073,
      "grad_norm": 1.0423866510391235,
      "learning_rate": 9.357088762866866e-06,
      "loss": 0.7541,
      "step": 6536
    },
    {
      "epoch": 0.18835359880136,
      "grad_norm": 1.0164397954940796,
      "learning_rate": 9.356859852198042e-06,
      "loss": 0.7871,
      "step": 6537
    },
    {
      "epoch": 0.18838241226300928,
      "grad_norm": 1.0142600536346436,
      "learning_rate": 9.356630903585322e-06,
      "loss": 0.7873,
      "step": 6538
    },
    {
      "epoch": 0.18841122572465857,
      "grad_norm": 1.0659655332565308,
      "learning_rate": 9.356401917030702e-06,
      "loss": 0.7912,
      "step": 6539
    },
    {
      "epoch": 0.18844003918630783,
      "grad_norm": 1.009081482887268,
      "learning_rate": 9.356172892536178e-06,
      "loss": 0.7727,
      "step": 6540
    },
    {
      "epoch": 0.18846885264795712,
      "grad_norm": 1.0477148294448853,
      "learning_rate": 9.355943830103739e-06,
      "loss": 0.7912,
      "step": 6541
    },
    {
      "epoch": 0.18849766610960642,
      "grad_norm": 1.041299819946289,
      "learning_rate": 9.355714729735389e-06,
      "loss": 0.7929,
      "step": 6542
    },
    {
      "epoch": 0.18852647957125568,
      "grad_norm": 1.030004858970642,
      "learning_rate": 9.355485591433113e-06,
      "loss": 0.7951,
      "step": 6543
    },
    {
      "epoch": 0.18855529303290497,
      "grad_norm": 1.0762927532196045,
      "learning_rate": 9.355256415198913e-06,
      "loss": 0.7635,
      "step": 6544
    },
    {
      "epoch": 0.18858410649455426,
      "grad_norm": 1.060791254043579,
      "learning_rate": 9.355027201034783e-06,
      "loss": 0.7958,
      "step": 6545
    },
    {
      "epoch": 0.18861291995620355,
      "grad_norm": 1.0507349967956543,
      "learning_rate": 9.35479794894272e-06,
      "loss": 0.7967,
      "step": 6546
    },
    {
      "epoch": 0.1886417334178528,
      "grad_norm": 1.0530487298965454,
      "learning_rate": 9.35456865892472e-06,
      "loss": 0.7927,
      "step": 6547
    },
    {
      "epoch": 0.1886705468795021,
      "grad_norm": 1.1257191896438599,
      "learning_rate": 9.35433933098278e-06,
      "loss": 0.7834,
      "step": 6548
    },
    {
      "epoch": 0.1886993603411514,
      "grad_norm": 1.0448508262634277,
      "learning_rate": 9.354109965118898e-06,
      "loss": 0.7954,
      "step": 6549
    },
    {
      "epoch": 0.18872817380280066,
      "grad_norm": 1.0560302734375,
      "learning_rate": 9.35388056133507e-06,
      "loss": 0.778,
      "step": 6550
    },
    {
      "epoch": 0.18875698726444995,
      "grad_norm": 1.0235568284988403,
      "learning_rate": 9.353651119633294e-06,
      "loss": 0.7766,
      "step": 6551
    },
    {
      "epoch": 0.18878580072609924,
      "grad_norm": 1.1759763956069946,
      "learning_rate": 9.353421640015569e-06,
      "loss": 0.7813,
      "step": 6552
    },
    {
      "epoch": 0.1888146141877485,
      "grad_norm": 1.0953449010849,
      "learning_rate": 9.353192122483893e-06,
      "loss": 0.7745,
      "step": 6553
    },
    {
      "epoch": 0.1888434276493978,
      "grad_norm": 1.0266574621200562,
      "learning_rate": 9.352962567040267e-06,
      "loss": 0.7669,
      "step": 6554
    },
    {
      "epoch": 0.18887224111104708,
      "grad_norm": 1.2112891674041748,
      "learning_rate": 9.352732973686687e-06,
      "loss": 0.8065,
      "step": 6555
    },
    {
      "epoch": 0.18890105457269638,
      "grad_norm": 1.115950345993042,
      "learning_rate": 9.352503342425155e-06,
      "loss": 0.806,
      "step": 6556
    },
    {
      "epoch": 0.18892986803434564,
      "grad_norm": 1.020521640777588,
      "learning_rate": 9.35227367325767e-06,
      "loss": 0.7534,
      "step": 6557
    },
    {
      "epoch": 0.18895868149599493,
      "grad_norm": 1.1129533052444458,
      "learning_rate": 9.352043966186231e-06,
      "loss": 0.7869,
      "step": 6558
    },
    {
      "epoch": 0.18898749495764422,
      "grad_norm": 1.056946873664856,
      "learning_rate": 9.351814221212838e-06,
      "loss": 0.763,
      "step": 6559
    },
    {
      "epoch": 0.18901630841929348,
      "grad_norm": 1.1139243841171265,
      "learning_rate": 9.351584438339496e-06,
      "loss": 0.7892,
      "step": 6560
    },
    {
      "epoch": 0.18904512188094277,
      "grad_norm": 1.05824613571167,
      "learning_rate": 9.351354617568202e-06,
      "loss": 0.7792,
      "step": 6561
    },
    {
      "epoch": 0.18907393534259206,
      "grad_norm": 1.1023213863372803,
      "learning_rate": 9.35112475890096e-06,
      "loss": 0.793,
      "step": 6562
    },
    {
      "epoch": 0.18910274880424136,
      "grad_norm": 1.077202320098877,
      "learning_rate": 9.35089486233977e-06,
      "loss": 0.7761,
      "step": 6563
    },
    {
      "epoch": 0.18913156226589062,
      "grad_norm": 1.0194040536880493,
      "learning_rate": 9.350664927886635e-06,
      "loss": 0.7861,
      "step": 6564
    },
    {
      "epoch": 0.1891603757275399,
      "grad_norm": 1.0346871614456177,
      "learning_rate": 9.350434955543557e-06,
      "loss": 0.7753,
      "step": 6565
    },
    {
      "epoch": 0.1891891891891892,
      "grad_norm": 1.035666584968567,
      "learning_rate": 9.350204945312542e-06,
      "loss": 0.7606,
      "step": 6566
    },
    {
      "epoch": 0.18921800265083846,
      "grad_norm": 1.0286749601364136,
      "learning_rate": 9.349974897195588e-06,
      "loss": 0.7914,
      "step": 6567
    },
    {
      "epoch": 0.18924681611248775,
      "grad_norm": 1.1139949560165405,
      "learning_rate": 9.349744811194702e-06,
      "loss": 0.7535,
      "step": 6568
    },
    {
      "epoch": 0.18927562957413704,
      "grad_norm": 1.045194387435913,
      "learning_rate": 9.349514687311887e-06,
      "loss": 0.769,
      "step": 6569
    },
    {
      "epoch": 0.1893044430357863,
      "grad_norm": 0.990865170955658,
      "learning_rate": 9.349284525549148e-06,
      "loss": 0.7632,
      "step": 6570
    },
    {
      "epoch": 0.1893332564974356,
      "grad_norm": 1.0370105504989624,
      "learning_rate": 9.349054325908487e-06,
      "loss": 0.7937,
      "step": 6571
    },
    {
      "epoch": 0.1893620699590849,
      "grad_norm": 1.0049680471420288,
      "learning_rate": 9.34882408839191e-06,
      "loss": 0.7778,
      "step": 6572
    },
    {
      "epoch": 0.18939088342073418,
      "grad_norm": 0.9733625650405884,
      "learning_rate": 9.348593813001421e-06,
      "loss": 0.7683,
      "step": 6573
    },
    {
      "epoch": 0.18941969688238344,
      "grad_norm": 1.0233228206634521,
      "learning_rate": 9.348363499739031e-06,
      "loss": 0.7703,
      "step": 6574
    },
    {
      "epoch": 0.18944851034403273,
      "grad_norm": 1.052627682685852,
      "learning_rate": 9.348133148606738e-06,
      "loss": 0.7649,
      "step": 6575
    },
    {
      "epoch": 0.18947732380568202,
      "grad_norm": 1.069382905960083,
      "learning_rate": 9.347902759606553e-06,
      "loss": 0.787,
      "step": 6576
    },
    {
      "epoch": 0.1895061372673313,
      "grad_norm": 1.0803143978118896,
      "learning_rate": 9.347672332740483e-06,
      "loss": 0.7771,
      "step": 6577
    },
    {
      "epoch": 0.18953495072898058,
      "grad_norm": 1.052715539932251,
      "learning_rate": 9.347441868010531e-06,
      "loss": 0.772,
      "step": 6578
    },
    {
      "epoch": 0.18956376419062987,
      "grad_norm": 1.0696682929992676,
      "learning_rate": 9.347211365418707e-06,
      "loss": 0.7701,
      "step": 6579
    },
    {
      "epoch": 0.18959257765227913,
      "grad_norm": 1.0181779861450195,
      "learning_rate": 9.346980824967018e-06,
      "loss": 0.8032,
      "step": 6580
    },
    {
      "epoch": 0.18962139111392842,
      "grad_norm": 1.0494250059127808,
      "learning_rate": 9.34675024665747e-06,
      "loss": 0.7642,
      "step": 6581
    },
    {
      "epoch": 0.1896502045755777,
      "grad_norm": 1.0211434364318848,
      "learning_rate": 9.346519630492074e-06,
      "loss": 0.7515,
      "step": 6582
    },
    {
      "epoch": 0.189679018037227,
      "grad_norm": 1.0657743215560913,
      "learning_rate": 9.346288976472835e-06,
      "loss": 0.7703,
      "step": 6583
    },
    {
      "epoch": 0.18970783149887627,
      "grad_norm": 1.0771076679229736,
      "learning_rate": 9.346058284601766e-06,
      "loss": 0.7769,
      "step": 6584
    },
    {
      "epoch": 0.18973664496052556,
      "grad_norm": 1.0104531049728394,
      "learning_rate": 9.345827554880874e-06,
      "loss": 0.8017,
      "step": 6585
    },
    {
      "epoch": 0.18976545842217485,
      "grad_norm": 1.0011075735092163,
      "learning_rate": 9.345596787312166e-06,
      "loss": 0.7883,
      "step": 6586
    },
    {
      "epoch": 0.1897942718838241,
      "grad_norm": 1.0471402406692505,
      "learning_rate": 9.345365981897655e-06,
      "loss": 0.7943,
      "step": 6587
    },
    {
      "epoch": 0.1898230853454734,
      "grad_norm": 1.013954520225525,
      "learning_rate": 9.34513513863935e-06,
      "loss": 0.7793,
      "step": 6588
    },
    {
      "epoch": 0.1898518988071227,
      "grad_norm": 1.0391048192977905,
      "learning_rate": 9.344904257539262e-06,
      "loss": 0.7838,
      "step": 6589
    },
    {
      "epoch": 0.18988071226877198,
      "grad_norm": 1.0519721508026123,
      "learning_rate": 9.344673338599401e-06,
      "loss": 0.7776,
      "step": 6590
    },
    {
      "epoch": 0.18990952573042125,
      "grad_norm": 0.997622549533844,
      "learning_rate": 9.344442381821778e-06,
      "loss": 0.7877,
      "step": 6591
    },
    {
      "epoch": 0.18993833919207054,
      "grad_norm": 1.066552996635437,
      "learning_rate": 9.344211387208404e-06,
      "loss": 0.795,
      "step": 6592
    },
    {
      "epoch": 0.18996715265371983,
      "grad_norm": 1.0747239589691162,
      "learning_rate": 9.343980354761292e-06,
      "loss": 0.7684,
      "step": 6593
    },
    {
      "epoch": 0.1899959661153691,
      "grad_norm": 1.0348894596099854,
      "learning_rate": 9.343749284482455e-06,
      "loss": 0.7643,
      "step": 6594
    },
    {
      "epoch": 0.19002477957701838,
      "grad_norm": 1.1367603540420532,
      "learning_rate": 9.343518176373902e-06,
      "loss": 0.798,
      "step": 6595
    },
    {
      "epoch": 0.19005359303866767,
      "grad_norm": 1.0668269395828247,
      "learning_rate": 9.343287030437648e-06,
      "loss": 0.759,
      "step": 6596
    },
    {
      "epoch": 0.19008240650031694,
      "grad_norm": 1.0164954662322998,
      "learning_rate": 9.343055846675707e-06,
      "loss": 0.7528,
      "step": 6597
    },
    {
      "epoch": 0.19011121996196623,
      "grad_norm": 1.0802464485168457,
      "learning_rate": 9.34282462509009e-06,
      "loss": 0.7549,
      "step": 6598
    },
    {
      "epoch": 0.19014003342361552,
      "grad_norm": 1.119675636291504,
      "learning_rate": 9.342593365682812e-06,
      "loss": 0.7826,
      "step": 6599
    },
    {
      "epoch": 0.1901688468852648,
      "grad_norm": 1.0393003225326538,
      "learning_rate": 9.342362068455886e-06,
      "loss": 0.7793,
      "step": 6600
    },
    {
      "epoch": 0.19019766034691407,
      "grad_norm": 1.0387332439422607,
      "learning_rate": 9.342130733411328e-06,
      "loss": 0.7882,
      "step": 6601
    },
    {
      "epoch": 0.19022647380856336,
      "grad_norm": 1.0273369550704956,
      "learning_rate": 9.341899360551151e-06,
      "loss": 0.7709,
      "step": 6602
    },
    {
      "epoch": 0.19025528727021265,
      "grad_norm": 1.0263837575912476,
      "learning_rate": 9.34166794987737e-06,
      "loss": 0.7917,
      "step": 6603
    },
    {
      "epoch": 0.19028410073186192,
      "grad_norm": 1.325995683670044,
      "learning_rate": 9.341436501392005e-06,
      "loss": 0.7809,
      "step": 6604
    },
    {
      "epoch": 0.1903129141935112,
      "grad_norm": 1.0106624364852905,
      "learning_rate": 9.341205015097066e-06,
      "loss": 0.7793,
      "step": 6605
    },
    {
      "epoch": 0.1903417276551605,
      "grad_norm": 1.0097907781600952,
      "learning_rate": 9.34097349099457e-06,
      "loss": 0.7792,
      "step": 6606
    },
    {
      "epoch": 0.19037054111680976,
      "grad_norm": 1.050304651260376,
      "learning_rate": 9.340741929086534e-06,
      "loss": 0.7608,
      "step": 6607
    },
    {
      "epoch": 0.19039935457845905,
      "grad_norm": 0.9944027662277222,
      "learning_rate": 9.340510329374976e-06,
      "loss": 0.7867,
      "step": 6608
    },
    {
      "epoch": 0.19042816804010834,
      "grad_norm": 1.0051510334014893,
      "learning_rate": 9.340278691861913e-06,
      "loss": 0.7884,
      "step": 6609
    },
    {
      "epoch": 0.19045698150175763,
      "grad_norm": 1.016221523284912,
      "learning_rate": 9.340047016549359e-06,
      "loss": 0.7946,
      "step": 6610
    },
    {
      "epoch": 0.1904857949634069,
      "grad_norm": 1.0243263244628906,
      "learning_rate": 9.339815303439334e-06,
      "loss": 0.743,
      "step": 6611
    },
    {
      "epoch": 0.1905146084250562,
      "grad_norm": 1.023639440536499,
      "learning_rate": 9.339583552533857e-06,
      "loss": 0.7672,
      "step": 6612
    },
    {
      "epoch": 0.19054342188670548,
      "grad_norm": 1.0310741662979126,
      "learning_rate": 9.339351763834947e-06,
      "loss": 0.7866,
      "step": 6613
    },
    {
      "epoch": 0.19057223534835474,
      "grad_norm": 1.0059832334518433,
      "learning_rate": 9.339119937344619e-06,
      "loss": 0.78,
      "step": 6614
    },
    {
      "epoch": 0.19060104881000403,
      "grad_norm": 1.0297218561172485,
      "learning_rate": 9.338888073064893e-06,
      "loss": 0.7698,
      "step": 6615
    },
    {
      "epoch": 0.19062986227165332,
      "grad_norm": 1.0032402276992798,
      "learning_rate": 9.33865617099779e-06,
      "loss": 0.8012,
      "step": 6616
    },
    {
      "epoch": 0.1906586757333026,
      "grad_norm": 1.0049468278884888,
      "learning_rate": 9.33842423114533e-06,
      "loss": 0.7634,
      "step": 6617
    },
    {
      "epoch": 0.19068748919495188,
      "grad_norm": 1.0224863290786743,
      "learning_rate": 9.33819225350953e-06,
      "loss": 0.7721,
      "step": 6618
    },
    {
      "epoch": 0.19071630265660117,
      "grad_norm": 1.041917085647583,
      "learning_rate": 9.337960238092413e-06,
      "loss": 0.7719,
      "step": 6619
    },
    {
      "epoch": 0.19074511611825046,
      "grad_norm": 1.00808584690094,
      "learning_rate": 9.337728184895998e-06,
      "loss": 0.7762,
      "step": 6620
    },
    {
      "epoch": 0.19077392957989972,
      "grad_norm": 0.9853479266166687,
      "learning_rate": 9.337496093922307e-06,
      "loss": 0.7767,
      "step": 6621
    },
    {
      "epoch": 0.190802743041549,
      "grad_norm": 1.0328965187072754,
      "learning_rate": 9.337263965173362e-06,
      "loss": 0.7992,
      "step": 6622
    },
    {
      "epoch": 0.1908315565031983,
      "grad_norm": 1.0070496797561646,
      "learning_rate": 9.337031798651181e-06,
      "loss": 0.7953,
      "step": 6623
    },
    {
      "epoch": 0.19086036996484756,
      "grad_norm": 0.9891191124916077,
      "learning_rate": 9.33679959435779e-06,
      "loss": 0.776,
      "step": 6624
    },
    {
      "epoch": 0.19088918342649686,
      "grad_norm": 1.0466142892837524,
      "learning_rate": 9.336567352295212e-06,
      "loss": 0.7464,
      "step": 6625
    },
    {
      "epoch": 0.19091799688814615,
      "grad_norm": 1.0202231407165527,
      "learning_rate": 9.336335072465464e-06,
      "loss": 0.7831,
      "step": 6626
    },
    {
      "epoch": 0.19094681034979544,
      "grad_norm": 1.007951259613037,
      "learning_rate": 9.336102754870573e-06,
      "loss": 0.7755,
      "step": 6627
    },
    {
      "epoch": 0.1909756238114447,
      "grad_norm": 1.0448826551437378,
      "learning_rate": 9.335870399512564e-06,
      "loss": 0.7767,
      "step": 6628
    },
    {
      "epoch": 0.191004437273094,
      "grad_norm": 1.103750228881836,
      "learning_rate": 9.335638006393455e-06,
      "loss": 0.7919,
      "step": 6629
    },
    {
      "epoch": 0.19103325073474328,
      "grad_norm": 1.0327167510986328,
      "learning_rate": 9.335405575515275e-06,
      "loss": 0.7922,
      "step": 6630
    },
    {
      "epoch": 0.19106206419639254,
      "grad_norm": 1.0406473875045776,
      "learning_rate": 9.335173106880045e-06,
      "loss": 0.7779,
      "step": 6631
    },
    {
      "epoch": 0.19109087765804184,
      "grad_norm": 1.1049526929855347,
      "learning_rate": 9.33494060048979e-06,
      "loss": 0.7906,
      "step": 6632
    },
    {
      "epoch": 0.19111969111969113,
      "grad_norm": 1.0315338373184204,
      "learning_rate": 9.334708056346538e-06,
      "loss": 0.8105,
      "step": 6633
    },
    {
      "epoch": 0.1911485045813404,
      "grad_norm": 1.0757042169570923,
      "learning_rate": 9.334475474452309e-06,
      "loss": 0.765,
      "step": 6634
    },
    {
      "epoch": 0.19117731804298968,
      "grad_norm": 1.0722227096557617,
      "learning_rate": 9.334242854809134e-06,
      "loss": 0.7801,
      "step": 6635
    },
    {
      "epoch": 0.19120613150463897,
      "grad_norm": 1.0231120586395264,
      "learning_rate": 9.334010197419036e-06,
      "loss": 0.7854,
      "step": 6636
    },
    {
      "epoch": 0.19123494496628826,
      "grad_norm": 1.06557297706604,
      "learning_rate": 9.333777502284042e-06,
      "loss": 0.799,
      "step": 6637
    },
    {
      "epoch": 0.19126375842793752,
      "grad_norm": 1.0046799182891846,
      "learning_rate": 9.333544769406177e-06,
      "loss": 0.7771,
      "step": 6638
    },
    {
      "epoch": 0.19129257188958682,
      "grad_norm": 1.037186622619629,
      "learning_rate": 9.33331199878747e-06,
      "loss": 0.8034,
      "step": 6639
    },
    {
      "epoch": 0.1913213853512361,
      "grad_norm": 1.0504341125488281,
      "learning_rate": 9.333079190429945e-06,
      "loss": 0.7829,
      "step": 6640
    },
    {
      "epoch": 0.19135019881288537,
      "grad_norm": 1.0763351917266846,
      "learning_rate": 9.332846344335636e-06,
      "loss": 0.7694,
      "step": 6641
    },
    {
      "epoch": 0.19137901227453466,
      "grad_norm": 1.0121766328811646,
      "learning_rate": 9.332613460506563e-06,
      "loss": 0.7613,
      "step": 6642
    },
    {
      "epoch": 0.19140782573618395,
      "grad_norm": 1.0675591230392456,
      "learning_rate": 9.332380538944758e-06,
      "loss": 0.788,
      "step": 6643
    },
    {
      "epoch": 0.1914366391978332,
      "grad_norm": 1.0833932161331177,
      "learning_rate": 9.33214757965225e-06,
      "loss": 0.7856,
      "step": 6644
    },
    {
      "epoch": 0.1914654526594825,
      "grad_norm": 1.0012600421905518,
      "learning_rate": 9.331914582631067e-06,
      "loss": 0.7833,
      "step": 6645
    },
    {
      "epoch": 0.1914942661211318,
      "grad_norm": 0.9891976118087769,
      "learning_rate": 9.33168154788324e-06,
      "loss": 0.7853,
      "step": 6646
    },
    {
      "epoch": 0.1915230795827811,
      "grad_norm": 1.0417959690093994,
      "learning_rate": 9.331448475410796e-06,
      "loss": 0.8132,
      "step": 6647
    },
    {
      "epoch": 0.19155189304443035,
      "grad_norm": 1.0175673961639404,
      "learning_rate": 9.331215365215764e-06,
      "loss": 0.7642,
      "step": 6648
    },
    {
      "epoch": 0.19158070650607964,
      "grad_norm": 0.9927467107772827,
      "learning_rate": 9.330982217300177e-06,
      "loss": 0.7667,
      "step": 6649
    },
    {
      "epoch": 0.19160951996772893,
      "grad_norm": 1.0384453535079956,
      "learning_rate": 9.330749031666063e-06,
      "loss": 0.757,
      "step": 6650
    },
    {
      "epoch": 0.1916383334293782,
      "grad_norm": 1.050832986831665,
      "learning_rate": 9.330515808315457e-06,
      "loss": 0.7789,
      "step": 6651
    },
    {
      "epoch": 0.19166714689102748,
      "grad_norm": 1.0837305784225464,
      "learning_rate": 9.330282547250384e-06,
      "loss": 0.7889,
      "step": 6652
    },
    {
      "epoch": 0.19169596035267678,
      "grad_norm": 1.0174877643585205,
      "learning_rate": 9.330049248472881e-06,
      "loss": 0.7791,
      "step": 6653
    },
    {
      "epoch": 0.19172477381432607,
      "grad_norm": 1.0389817953109741,
      "learning_rate": 9.329815911984976e-06,
      "loss": 0.7976,
      "step": 6654
    },
    {
      "epoch": 0.19175358727597533,
      "grad_norm": 1.0539089441299438,
      "learning_rate": 9.329582537788705e-06,
      "loss": 0.7435,
      "step": 6655
    },
    {
      "epoch": 0.19178240073762462,
      "grad_norm": 1.058527946472168,
      "learning_rate": 9.329349125886096e-06,
      "loss": 0.7927,
      "step": 6656
    },
    {
      "epoch": 0.1918112141992739,
      "grad_norm": 1.0633214712142944,
      "learning_rate": 9.329115676279184e-06,
      "loss": 0.7929,
      "step": 6657
    },
    {
      "epoch": 0.19184002766092317,
      "grad_norm": 1.0260237455368042,
      "learning_rate": 9.328882188970003e-06,
      "loss": 0.777,
      "step": 6658
    },
    {
      "epoch": 0.19186884112257246,
      "grad_norm": 1.097530722618103,
      "learning_rate": 9.328648663960584e-06,
      "loss": 0.7721,
      "step": 6659
    },
    {
      "epoch": 0.19189765458422176,
      "grad_norm": 1.0196858644485474,
      "learning_rate": 9.328415101252963e-06,
      "loss": 0.7577,
      "step": 6660
    },
    {
      "epoch": 0.19192646804587102,
      "grad_norm": 1.0435508489608765,
      "learning_rate": 9.328181500849174e-06,
      "loss": 0.7896,
      "step": 6661
    },
    {
      "epoch": 0.1919552815075203,
      "grad_norm": 1.0788180828094482,
      "learning_rate": 9.327947862751249e-06,
      "loss": 0.8024,
      "step": 6662
    },
    {
      "epoch": 0.1919840949691696,
      "grad_norm": 1.0230351686477661,
      "learning_rate": 9.327714186961226e-06,
      "loss": 0.767,
      "step": 6663
    },
    {
      "epoch": 0.1920129084308189,
      "grad_norm": 1.0398776531219482,
      "learning_rate": 9.327480473481138e-06,
      "loss": 0.7891,
      "step": 6664
    },
    {
      "epoch": 0.19204172189246815,
      "grad_norm": 1.1668095588684082,
      "learning_rate": 9.327246722313021e-06,
      "loss": 0.7817,
      "step": 6665
    },
    {
      "epoch": 0.19207053535411744,
      "grad_norm": 1.0078387260437012,
      "learning_rate": 9.32701293345891e-06,
      "loss": 0.7755,
      "step": 6666
    },
    {
      "epoch": 0.19209934881576674,
      "grad_norm": 1.0195467472076416,
      "learning_rate": 9.326779106920843e-06,
      "loss": 0.7782,
      "step": 6667
    },
    {
      "epoch": 0.192128162277416,
      "grad_norm": 1.0729228258132935,
      "learning_rate": 9.326545242700855e-06,
      "loss": 0.7816,
      "step": 6668
    },
    {
      "epoch": 0.1921569757390653,
      "grad_norm": 1.0289840698242188,
      "learning_rate": 9.326311340800981e-06,
      "loss": 0.8115,
      "step": 6669
    },
    {
      "epoch": 0.19218578920071458,
      "grad_norm": 1.0668522119522095,
      "learning_rate": 9.326077401223264e-06,
      "loss": 0.7887,
      "step": 6670
    },
    {
      "epoch": 0.19221460266236384,
      "grad_norm": 1.0214474201202393,
      "learning_rate": 9.325843423969734e-06,
      "loss": 0.756,
      "step": 6671
    },
    {
      "epoch": 0.19224341612401313,
      "grad_norm": 1.0214265584945679,
      "learning_rate": 9.325609409042433e-06,
      "loss": 0.7785,
      "step": 6672
    },
    {
      "epoch": 0.19227222958566242,
      "grad_norm": 1.1363805532455444,
      "learning_rate": 9.325375356443398e-06,
      "loss": 0.7984,
      "step": 6673
    },
    {
      "epoch": 0.19230104304731171,
      "grad_norm": 1.0467593669891357,
      "learning_rate": 9.325141266174666e-06,
      "loss": 0.8157,
      "step": 6674
    },
    {
      "epoch": 0.19232985650896098,
      "grad_norm": 0.9808245301246643,
      "learning_rate": 9.32490713823828e-06,
      "loss": 0.7942,
      "step": 6675
    },
    {
      "epoch": 0.19235866997061027,
      "grad_norm": 1.102705717086792,
      "learning_rate": 9.324672972636276e-06,
      "loss": 0.811,
      "step": 6676
    },
    {
      "epoch": 0.19238748343225956,
      "grad_norm": 1.0158518552780151,
      "learning_rate": 9.32443876937069e-06,
      "loss": 0.798,
      "step": 6677
    },
    {
      "epoch": 0.19241629689390882,
      "grad_norm": 1.0533908605575562,
      "learning_rate": 9.324204528443568e-06,
      "loss": 0.7729,
      "step": 6678
    },
    {
      "epoch": 0.1924451103555581,
      "grad_norm": 1.0322903394699097,
      "learning_rate": 9.323970249856946e-06,
      "loss": 0.7891,
      "step": 6679
    },
    {
      "epoch": 0.1924739238172074,
      "grad_norm": 0.9974323511123657,
      "learning_rate": 9.323735933612866e-06,
      "loss": 0.7816,
      "step": 6680
    },
    {
      "epoch": 0.1925027372788567,
      "grad_norm": 1.0038913488388062,
      "learning_rate": 9.323501579713368e-06,
      "loss": 0.7582,
      "step": 6681
    },
    {
      "epoch": 0.19253155074050596,
      "grad_norm": 1.0121049880981445,
      "learning_rate": 9.323267188160494e-06,
      "loss": 0.7696,
      "step": 6682
    },
    {
      "epoch": 0.19256036420215525,
      "grad_norm": 1.0317490100860596,
      "learning_rate": 9.323032758956284e-06,
      "loss": 0.769,
      "step": 6683
    },
    {
      "epoch": 0.19258917766380454,
      "grad_norm": 1.0411421060562134,
      "learning_rate": 9.322798292102779e-06,
      "loss": 0.7757,
      "step": 6684
    },
    {
      "epoch": 0.1926179911254538,
      "grad_norm": 0.9950242638587952,
      "learning_rate": 9.322563787602024e-06,
      "loss": 0.8209,
      "step": 6685
    },
    {
      "epoch": 0.1926468045871031,
      "grad_norm": 1.029089331626892,
      "learning_rate": 9.322329245456058e-06,
      "loss": 0.7797,
      "step": 6686
    },
    {
      "epoch": 0.19267561804875238,
      "grad_norm": 0.9990517497062683,
      "learning_rate": 9.322094665666925e-06,
      "loss": 0.8004,
      "step": 6687
    },
    {
      "epoch": 0.19270443151040165,
      "grad_norm": 0.9945635199546814,
      "learning_rate": 9.32186004823667e-06,
      "loss": 0.7864,
      "step": 6688
    },
    {
      "epoch": 0.19273324497205094,
      "grad_norm": 1.0424150228500366,
      "learning_rate": 9.321625393167333e-06,
      "loss": 0.7938,
      "step": 6689
    },
    {
      "epoch": 0.19276205843370023,
      "grad_norm": 1.0436007976531982,
      "learning_rate": 9.321390700460958e-06,
      "loss": 0.7971,
      "step": 6690
    },
    {
      "epoch": 0.19279087189534952,
      "grad_norm": 1.027342677116394,
      "learning_rate": 9.321155970119589e-06,
      "loss": 0.7722,
      "step": 6691
    },
    {
      "epoch": 0.19281968535699878,
      "grad_norm": 0.9759502410888672,
      "learning_rate": 9.320921202145273e-06,
      "loss": 0.7856,
      "step": 6692
    },
    {
      "epoch": 0.19284849881864807,
      "grad_norm": 1.0588932037353516,
      "learning_rate": 9.320686396540054e-06,
      "loss": 0.7853,
      "step": 6693
    },
    {
      "epoch": 0.19287731228029736,
      "grad_norm": 1.000502586364746,
      "learning_rate": 9.320451553305973e-06,
      "loss": 0.791,
      "step": 6694
    },
    {
      "epoch": 0.19290612574194663,
      "grad_norm": 1.015244483947754,
      "learning_rate": 9.320216672445079e-06,
      "loss": 0.7574,
      "step": 6695
    },
    {
      "epoch": 0.19293493920359592,
      "grad_norm": 1.0124953985214233,
      "learning_rate": 9.319981753959415e-06,
      "loss": 0.7804,
      "step": 6696
    },
    {
      "epoch": 0.1929637526652452,
      "grad_norm": 0.9901260733604431,
      "learning_rate": 9.319746797851029e-06,
      "loss": 0.7718,
      "step": 6697
    },
    {
      "epoch": 0.19299256612689447,
      "grad_norm": 1.018173336982727,
      "learning_rate": 9.319511804121966e-06,
      "loss": 0.7767,
      "step": 6698
    },
    {
      "epoch": 0.19302137958854376,
      "grad_norm": 0.9957154989242554,
      "learning_rate": 9.319276772774275e-06,
      "loss": 0.7606,
      "step": 6699
    },
    {
      "epoch": 0.19305019305019305,
      "grad_norm": 0.9693863987922668,
      "learning_rate": 9.31904170381e-06,
      "loss": 0.7688,
      "step": 6700
    },
    {
      "epoch": 0.19307900651184234,
      "grad_norm": 0.9814503788948059,
      "learning_rate": 9.318806597231189e-06,
      "loss": 0.7746,
      "step": 6701
    },
    {
      "epoch": 0.1931078199734916,
      "grad_norm": 1.0281647443771362,
      "learning_rate": 9.318571453039892e-06,
      "loss": 0.7872,
      "step": 6702
    },
    {
      "epoch": 0.1931366334351409,
      "grad_norm": 1.0265990495681763,
      "learning_rate": 9.318336271238151e-06,
      "loss": 0.7741,
      "step": 6703
    },
    {
      "epoch": 0.1931654468967902,
      "grad_norm": 0.9784792065620422,
      "learning_rate": 9.318101051828019e-06,
      "loss": 0.7674,
      "step": 6704
    },
    {
      "epoch": 0.19319426035843945,
      "grad_norm": 0.9879542589187622,
      "learning_rate": 9.317865794811545e-06,
      "loss": 0.7622,
      "step": 6705
    },
    {
      "epoch": 0.19322307382008874,
      "grad_norm": 1.081153154373169,
      "learning_rate": 9.317630500190774e-06,
      "loss": 0.7699,
      "step": 6706
    },
    {
      "epoch": 0.19325188728173803,
      "grad_norm": 1.01522696018219,
      "learning_rate": 9.317395167967758e-06,
      "loss": 0.775,
      "step": 6707
    },
    {
      "epoch": 0.19328070074338732,
      "grad_norm": 0.9826676249504089,
      "learning_rate": 9.317159798144546e-06,
      "loss": 0.7601,
      "step": 6708
    },
    {
      "epoch": 0.1933095142050366,
      "grad_norm": 1.0730206966400146,
      "learning_rate": 9.31692439072319e-06,
      "loss": 0.7703,
      "step": 6709
    },
    {
      "epoch": 0.19333832766668588,
      "grad_norm": 1.0773061513900757,
      "learning_rate": 9.316688945705734e-06,
      "loss": 0.7748,
      "step": 6710
    },
    {
      "epoch": 0.19336714112833517,
      "grad_norm": 1.0161490440368652,
      "learning_rate": 9.316453463094235e-06,
      "loss": 0.7924,
      "step": 6711
    },
    {
      "epoch": 0.19339595458998443,
      "grad_norm": 0.9888389706611633,
      "learning_rate": 9.316217942890741e-06,
      "loss": 0.7587,
      "step": 6712
    },
    {
      "epoch": 0.19342476805163372,
      "grad_norm": 0.9690656661987305,
      "learning_rate": 9.315982385097301e-06,
      "loss": 0.7771,
      "step": 6713
    },
    {
      "epoch": 0.193453581513283,
      "grad_norm": 0.9729353785514832,
      "learning_rate": 9.315746789715973e-06,
      "loss": 0.7691,
      "step": 6714
    },
    {
      "epoch": 0.19348239497493228,
      "grad_norm": 1.0124804973602295,
      "learning_rate": 9.315511156748803e-06,
      "loss": 0.7867,
      "step": 6715
    },
    {
      "epoch": 0.19351120843658157,
      "grad_norm": 1.0587403774261475,
      "learning_rate": 9.315275486197844e-06,
      "loss": 0.8027,
      "step": 6716
    },
    {
      "epoch": 0.19354002189823086,
      "grad_norm": 1.0362979173660278,
      "learning_rate": 9.315039778065148e-06,
      "loss": 0.8076,
      "step": 6717
    },
    {
      "epoch": 0.19356883535988015,
      "grad_norm": 1.0545618534088135,
      "learning_rate": 9.314804032352771e-06,
      "loss": 0.7831,
      "step": 6718
    },
    {
      "epoch": 0.1935976488215294,
      "grad_norm": 0.9525043964385986,
      "learning_rate": 9.314568249062764e-06,
      "loss": 0.7708,
      "step": 6719
    },
    {
      "epoch": 0.1936264622831787,
      "grad_norm": 1.0951688289642334,
      "learning_rate": 9.314332428197178e-06,
      "loss": 0.775,
      "step": 6720
    },
    {
      "epoch": 0.193655275744828,
      "grad_norm": 0.9942865371704102,
      "learning_rate": 9.314096569758073e-06,
      "loss": 0.7927,
      "step": 6721
    },
    {
      "epoch": 0.19368408920647726,
      "grad_norm": 1.0914424657821655,
      "learning_rate": 9.313860673747497e-06,
      "loss": 0.7824,
      "step": 6722
    },
    {
      "epoch": 0.19371290266812655,
      "grad_norm": 1.011439323425293,
      "learning_rate": 9.313624740167507e-06,
      "loss": 0.8174,
      "step": 6723
    },
    {
      "epoch": 0.19374171612977584,
      "grad_norm": 1.0250808000564575,
      "learning_rate": 9.31338876902016e-06,
      "loss": 0.7513,
      "step": 6724
    },
    {
      "epoch": 0.1937705295914251,
      "grad_norm": 0.9929583668708801,
      "learning_rate": 9.313152760307505e-06,
      "loss": 0.7698,
      "step": 6725
    },
    {
      "epoch": 0.1937993430530744,
      "grad_norm": 1.0874782800674438,
      "learning_rate": 9.312916714031604e-06,
      "loss": 0.7719,
      "step": 6726
    },
    {
      "epoch": 0.19382815651472368,
      "grad_norm": 1.0141892433166504,
      "learning_rate": 9.312680630194508e-06,
      "loss": 0.8044,
      "step": 6727
    },
    {
      "epoch": 0.19385696997637297,
      "grad_norm": 1.0842756032943726,
      "learning_rate": 9.312444508798274e-06,
      "loss": 0.7783,
      "step": 6728
    },
    {
      "epoch": 0.19388578343802224,
      "grad_norm": 1.0281708240509033,
      "learning_rate": 9.31220834984496e-06,
      "loss": 0.7778,
      "step": 6729
    },
    {
      "epoch": 0.19391459689967153,
      "grad_norm": 1.1177655458450317,
      "learning_rate": 9.311972153336623e-06,
      "loss": 0.7966,
      "step": 6730
    },
    {
      "epoch": 0.19394341036132082,
      "grad_norm": 1.0680125951766968,
      "learning_rate": 9.311735919275318e-06,
      "loss": 0.787,
      "step": 6731
    },
    {
      "epoch": 0.19397222382297008,
      "grad_norm": 1.0099191665649414,
      "learning_rate": 9.311499647663105e-06,
      "loss": 0.7819,
      "step": 6732
    },
    {
      "epoch": 0.19400103728461937,
      "grad_norm": 1.014583706855774,
      "learning_rate": 9.311263338502039e-06,
      "loss": 0.7413,
      "step": 6733
    },
    {
      "epoch": 0.19402985074626866,
      "grad_norm": 1.0017211437225342,
      "learning_rate": 9.311026991794177e-06,
      "loss": 0.774,
      "step": 6734
    },
    {
      "epoch": 0.19405866420791795,
      "grad_norm": 1.0080784559249878,
      "learning_rate": 9.310790607541583e-06,
      "loss": 0.8,
      "step": 6735
    },
    {
      "epoch": 0.19408747766956722,
      "grad_norm": 1.0122281312942505,
      "learning_rate": 9.310554185746308e-06,
      "loss": 0.7903,
      "step": 6736
    },
    {
      "epoch": 0.1941162911312165,
      "grad_norm": 1.027989387512207,
      "learning_rate": 9.310317726410419e-06,
      "loss": 0.7726,
      "step": 6737
    },
    {
      "epoch": 0.1941451045928658,
      "grad_norm": 0.9698618054389954,
      "learning_rate": 9.310081229535968e-06,
      "loss": 0.7647,
      "step": 6738
    },
    {
      "epoch": 0.19417391805451506,
      "grad_norm": 0.9823084473609924,
      "learning_rate": 9.30984469512502e-06,
      "loss": 0.7894,
      "step": 6739
    },
    {
      "epoch": 0.19420273151616435,
      "grad_norm": 1.0398718118667603,
      "learning_rate": 9.309608123179633e-06,
      "loss": 0.793,
      "step": 6740
    },
    {
      "epoch": 0.19423154497781364,
      "grad_norm": 1.0235373973846436,
      "learning_rate": 9.309371513701866e-06,
      "loss": 0.7778,
      "step": 6741
    },
    {
      "epoch": 0.1942603584394629,
      "grad_norm": 0.9730142951011658,
      "learning_rate": 9.309134866693781e-06,
      "loss": 0.7727,
      "step": 6742
    },
    {
      "epoch": 0.1942891719011122,
      "grad_norm": 1.024721622467041,
      "learning_rate": 9.30889818215744e-06,
      "loss": 0.775,
      "step": 6743
    },
    {
      "epoch": 0.1943179853627615,
      "grad_norm": 1.0383909940719604,
      "learning_rate": 9.308661460094901e-06,
      "loss": 0.7847,
      "step": 6744
    },
    {
      "epoch": 0.19434679882441078,
      "grad_norm": 0.9680806994438171,
      "learning_rate": 9.308424700508229e-06,
      "loss": 0.7733,
      "step": 6745
    },
    {
      "epoch": 0.19437561228606004,
      "grad_norm": 1.0045424699783325,
      "learning_rate": 9.308187903399486e-06,
      "loss": 0.7923,
      "step": 6746
    },
    {
      "epoch": 0.19440442574770933,
      "grad_norm": 1.0058152675628662,
      "learning_rate": 9.307951068770733e-06,
      "loss": 0.7741,
      "step": 6747
    },
    {
      "epoch": 0.19443323920935862,
      "grad_norm": 1.0113530158996582,
      "learning_rate": 9.307714196624032e-06,
      "loss": 0.7788,
      "step": 6748
    },
    {
      "epoch": 0.19446205267100788,
      "grad_norm": 1.0451607704162598,
      "learning_rate": 9.307477286961444e-06,
      "loss": 0.7762,
      "step": 6749
    },
    {
      "epoch": 0.19449086613265718,
      "grad_norm": 1.0419310331344604,
      "learning_rate": 9.307240339785038e-06,
      "loss": 0.8041,
      "step": 6750
    },
    {
      "epoch": 0.19451967959430647,
      "grad_norm": 0.9600046873092651,
      "learning_rate": 9.307003355096873e-06,
      "loss": 0.8066,
      "step": 6751
    },
    {
      "epoch": 0.19454849305595573,
      "grad_norm": 0.9945117831230164,
      "learning_rate": 9.306766332899013e-06,
      "loss": 0.7603,
      "step": 6752
    },
    {
      "epoch": 0.19457730651760502,
      "grad_norm": 1.0234782695770264,
      "learning_rate": 9.306529273193525e-06,
      "loss": 0.7668,
      "step": 6753
    },
    {
      "epoch": 0.1946061199792543,
      "grad_norm": 0.985267162322998,
      "learning_rate": 9.306292175982471e-06,
      "loss": 0.7544,
      "step": 6754
    },
    {
      "epoch": 0.1946349334409036,
      "grad_norm": 1.0136786699295044,
      "learning_rate": 9.306055041267919e-06,
      "loss": 0.7591,
      "step": 6755
    },
    {
      "epoch": 0.19466374690255286,
      "grad_norm": 1.0527071952819824,
      "learning_rate": 9.305817869051931e-06,
      "loss": 0.7794,
      "step": 6756
    },
    {
      "epoch": 0.19469256036420216,
      "grad_norm": 0.993053138256073,
      "learning_rate": 9.305580659336573e-06,
      "loss": 0.7725,
      "step": 6757
    },
    {
      "epoch": 0.19472137382585145,
      "grad_norm": 0.987563967704773,
      "learning_rate": 9.305343412123912e-06,
      "loss": 0.7358,
      "step": 6758
    },
    {
      "epoch": 0.1947501872875007,
      "grad_norm": 0.9893617033958435,
      "learning_rate": 9.305106127416013e-06,
      "loss": 0.764,
      "step": 6759
    },
    {
      "epoch": 0.19477900074915,
      "grad_norm": 1.008057713508606,
      "learning_rate": 9.304868805214945e-06,
      "loss": 0.7974,
      "step": 6760
    },
    {
      "epoch": 0.1948078142107993,
      "grad_norm": 0.9540310502052307,
      "learning_rate": 9.30463144552277e-06,
      "loss": 0.7506,
      "step": 6761
    },
    {
      "epoch": 0.19483662767244858,
      "grad_norm": 1.0879414081573486,
      "learning_rate": 9.30439404834156e-06,
      "loss": 0.8049,
      "step": 6762
    },
    {
      "epoch": 0.19486544113409784,
      "grad_norm": 0.9751487374305725,
      "learning_rate": 9.30415661367338e-06,
      "loss": 0.7705,
      "step": 6763
    },
    {
      "epoch": 0.19489425459574714,
      "grad_norm": 1.1045947074890137,
      "learning_rate": 9.3039191415203e-06,
      "loss": 0.7981,
      "step": 6764
    },
    {
      "epoch": 0.19492306805739643,
      "grad_norm": 1.0307326316833496,
      "learning_rate": 9.303681631884385e-06,
      "loss": 0.7745,
      "step": 6765
    },
    {
      "epoch": 0.1949518815190457,
      "grad_norm": 0.9865830540657043,
      "learning_rate": 9.303444084767705e-06,
      "loss": 0.7912,
      "step": 6766
    },
    {
      "epoch": 0.19498069498069498,
      "grad_norm": 1.055005431175232,
      "learning_rate": 9.303206500172328e-06,
      "loss": 0.8297,
      "step": 6767
    },
    {
      "epoch": 0.19500950844234427,
      "grad_norm": 1.0074551105499268,
      "learning_rate": 9.302968878100326e-06,
      "loss": 0.778,
      "step": 6768
    },
    {
      "epoch": 0.19503832190399353,
      "grad_norm": 2.2622265815734863,
      "learning_rate": 9.302731218553767e-06,
      "loss": 0.7707,
      "step": 6769
    },
    {
      "epoch": 0.19506713536564282,
      "grad_norm": 0.9833028316497803,
      "learning_rate": 9.302493521534718e-06,
      "loss": 0.7768,
      "step": 6770
    },
    {
      "epoch": 0.19509594882729211,
      "grad_norm": 1.037789225578308,
      "learning_rate": 9.302255787045253e-06,
      "loss": 0.7725,
      "step": 6771
    },
    {
      "epoch": 0.1951247622889414,
      "grad_norm": 1.0326652526855469,
      "learning_rate": 9.30201801508744e-06,
      "loss": 0.7745,
      "step": 6772
    },
    {
      "epoch": 0.19515357575059067,
      "grad_norm": 1.0249861478805542,
      "learning_rate": 9.301780205663352e-06,
      "loss": 0.7879,
      "step": 6773
    },
    {
      "epoch": 0.19518238921223996,
      "grad_norm": 1.0160549879074097,
      "learning_rate": 9.301542358775056e-06,
      "loss": 0.796,
      "step": 6774
    },
    {
      "epoch": 0.19521120267388925,
      "grad_norm": 0.970416247844696,
      "learning_rate": 9.301304474424629e-06,
      "loss": 0.7792,
      "step": 6775
    },
    {
      "epoch": 0.1952400161355385,
      "grad_norm": 1.0579270124435425,
      "learning_rate": 9.301066552614137e-06,
      "loss": 0.7699,
      "step": 6776
    },
    {
      "epoch": 0.1952688295971878,
      "grad_norm": 1.0200719833374023,
      "learning_rate": 9.300828593345657e-06,
      "loss": 0.7854,
      "step": 6777
    },
    {
      "epoch": 0.1952976430588371,
      "grad_norm": 1.0032436847686768,
      "learning_rate": 9.300590596621258e-06,
      "loss": 0.7918,
      "step": 6778
    },
    {
      "epoch": 0.19532645652048636,
      "grad_norm": 0.9859365224838257,
      "learning_rate": 9.300352562443013e-06,
      "loss": 0.7708,
      "step": 6779
    },
    {
      "epoch": 0.19535526998213565,
      "grad_norm": 1.0587997436523438,
      "learning_rate": 9.300114490812997e-06,
      "loss": 0.7547,
      "step": 6780
    },
    {
      "epoch": 0.19538408344378494,
      "grad_norm": 1.06272554397583,
      "learning_rate": 9.299876381733283e-06,
      "loss": 0.7991,
      "step": 6781
    },
    {
      "epoch": 0.19541289690543423,
      "grad_norm": 0.9932327270507812,
      "learning_rate": 9.299638235205944e-06,
      "loss": 0.7457,
      "step": 6782
    },
    {
      "epoch": 0.1954417103670835,
      "grad_norm": 0.9754451513290405,
      "learning_rate": 9.299400051233053e-06,
      "loss": 0.7692,
      "step": 6783
    },
    {
      "epoch": 0.19547052382873278,
      "grad_norm": 1.0425752401351929,
      "learning_rate": 9.299161829816685e-06,
      "loss": 0.7478,
      "step": 6784
    },
    {
      "epoch": 0.19549933729038207,
      "grad_norm": 1.0217317342758179,
      "learning_rate": 9.298923570958917e-06,
      "loss": 0.7808,
      "step": 6785
    },
    {
      "epoch": 0.19552815075203134,
      "grad_norm": 0.9851099848747253,
      "learning_rate": 9.298685274661821e-06,
      "loss": 0.7774,
      "step": 6786
    },
    {
      "epoch": 0.19555696421368063,
      "grad_norm": 1.0276371240615845,
      "learning_rate": 9.298446940927472e-06,
      "loss": 0.7657,
      "step": 6787
    },
    {
      "epoch": 0.19558577767532992,
      "grad_norm": 1.0695948600769043,
      "learning_rate": 9.298208569757949e-06,
      "loss": 0.8051,
      "step": 6788
    },
    {
      "epoch": 0.1956145911369792,
      "grad_norm": 0.9796585440635681,
      "learning_rate": 9.297970161155326e-06,
      "loss": 0.7698,
      "step": 6789
    },
    {
      "epoch": 0.19564340459862847,
      "grad_norm": 1.036357045173645,
      "learning_rate": 9.297731715121678e-06,
      "loss": 0.758,
      "step": 6790
    },
    {
      "epoch": 0.19567221806027776,
      "grad_norm": 1.002484679222107,
      "learning_rate": 9.297493231659084e-06,
      "loss": 0.7729,
      "step": 6791
    },
    {
      "epoch": 0.19570103152192705,
      "grad_norm": 1.0059365034103394,
      "learning_rate": 9.29725471076962e-06,
      "loss": 0.7584,
      "step": 6792
    },
    {
      "epoch": 0.19572984498357632,
      "grad_norm": 1.070482611656189,
      "learning_rate": 9.297016152455363e-06,
      "loss": 0.7988,
      "step": 6793
    },
    {
      "epoch": 0.1957586584452256,
      "grad_norm": 1.0614595413208008,
      "learning_rate": 9.296777556718389e-06,
      "loss": 0.7726,
      "step": 6794
    },
    {
      "epoch": 0.1957874719068749,
      "grad_norm": 0.9309858083724976,
      "learning_rate": 9.296538923560779e-06,
      "loss": 0.7768,
      "step": 6795
    },
    {
      "epoch": 0.19581628536852416,
      "grad_norm": 1.004480004310608,
      "learning_rate": 9.29630025298461e-06,
      "loss": 0.7879,
      "step": 6796
    },
    {
      "epoch": 0.19584509883017345,
      "grad_norm": 1.0528987646102905,
      "learning_rate": 9.296061544991961e-06,
      "loss": 0.786,
      "step": 6797
    },
    {
      "epoch": 0.19587391229182274,
      "grad_norm": 0.9856438040733337,
      "learning_rate": 9.295822799584911e-06,
      "loss": 0.7943,
      "step": 6798
    },
    {
      "epoch": 0.19590272575347203,
      "grad_norm": 1.0346795320510864,
      "learning_rate": 9.295584016765536e-06,
      "loss": 0.8087,
      "step": 6799
    },
    {
      "epoch": 0.1959315392151213,
      "grad_norm": 1.0248314142227173,
      "learning_rate": 9.29534519653592e-06,
      "loss": 0.7973,
      "step": 6800
    },
    {
      "epoch": 0.1959603526767706,
      "grad_norm": 0.9620211124420166,
      "learning_rate": 9.295106338898141e-06,
      "loss": 0.7611,
      "step": 6801
    },
    {
      "epoch": 0.19598916613841988,
      "grad_norm": 0.9600381255149841,
      "learning_rate": 9.294867443854278e-06,
      "loss": 0.7557,
      "step": 6802
    },
    {
      "epoch": 0.19601797960006914,
      "grad_norm": 1.0094536542892456,
      "learning_rate": 9.294628511406415e-06,
      "loss": 0.7879,
      "step": 6803
    },
    {
      "epoch": 0.19604679306171843,
      "grad_norm": 1.0317448377609253,
      "learning_rate": 9.29438954155663e-06,
      "loss": 0.8181,
      "step": 6804
    },
    {
      "epoch": 0.19607560652336772,
      "grad_norm": 0.9826138615608215,
      "learning_rate": 9.294150534307006e-06,
      "loss": 0.7858,
      "step": 6805
    },
    {
      "epoch": 0.196104419985017,
      "grad_norm": 0.9751235842704773,
      "learning_rate": 9.293911489659622e-06,
      "loss": 0.7859,
      "step": 6806
    },
    {
      "epoch": 0.19613323344666628,
      "grad_norm": 0.9483069777488708,
      "learning_rate": 9.293672407616561e-06,
      "loss": 0.774,
      "step": 6807
    },
    {
      "epoch": 0.19616204690831557,
      "grad_norm": 1.0034503936767578,
      "learning_rate": 9.293433288179904e-06,
      "loss": 0.7883,
      "step": 6808
    },
    {
      "epoch": 0.19619086036996486,
      "grad_norm": 1.0183855295181274,
      "learning_rate": 9.293194131351738e-06,
      "loss": 0.772,
      "step": 6809
    },
    {
      "epoch": 0.19621967383161412,
      "grad_norm": 0.9957045912742615,
      "learning_rate": 9.292954937134142e-06,
      "loss": 0.7858,
      "step": 6810
    },
    {
      "epoch": 0.1962484872932634,
      "grad_norm": 0.9838064908981323,
      "learning_rate": 9.292715705529198e-06,
      "loss": 0.8056,
      "step": 6811
    },
    {
      "epoch": 0.1962773007549127,
      "grad_norm": 1.0199836492538452,
      "learning_rate": 9.292476436538993e-06,
      "loss": 0.7718,
      "step": 6812
    },
    {
      "epoch": 0.19630611421656197,
      "grad_norm": 0.9848812222480774,
      "learning_rate": 9.29223713016561e-06,
      "loss": 0.7826,
      "step": 6813
    },
    {
      "epoch": 0.19633492767821126,
      "grad_norm": 1.0055203437805176,
      "learning_rate": 9.291997786411132e-06,
      "loss": 0.7893,
      "step": 6814
    },
    {
      "epoch": 0.19636374113986055,
      "grad_norm": 1.0611149072647095,
      "learning_rate": 9.291758405277642e-06,
      "loss": 0.8047,
      "step": 6815
    },
    {
      "epoch": 0.19639255460150984,
      "grad_norm": 0.9635359048843384,
      "learning_rate": 9.291518986767227e-06,
      "loss": 0.7762,
      "step": 6816
    },
    {
      "epoch": 0.1964213680631591,
      "grad_norm": 0.9964767098426819,
      "learning_rate": 9.291279530881972e-06,
      "loss": 0.7883,
      "step": 6817
    },
    {
      "epoch": 0.1964501815248084,
      "grad_norm": 0.9890223145484924,
      "learning_rate": 9.291040037623963e-06,
      "loss": 0.781,
      "step": 6818
    },
    {
      "epoch": 0.19647899498645768,
      "grad_norm": 0.9939032793045044,
      "learning_rate": 9.290800506995281e-06,
      "loss": 0.7767,
      "step": 6819
    },
    {
      "epoch": 0.19650780844810695,
      "grad_norm": 1.0001331567764282,
      "learning_rate": 9.29056093899802e-06,
      "loss": 0.7945,
      "step": 6820
    },
    {
      "epoch": 0.19653662190975624,
      "grad_norm": 1.0308358669281006,
      "learning_rate": 9.29032133363426e-06,
      "loss": 0.7929,
      "step": 6821
    },
    {
      "epoch": 0.19656543537140553,
      "grad_norm": 1.0418202877044678,
      "learning_rate": 9.290081690906092e-06,
      "loss": 0.7757,
      "step": 6822
    },
    {
      "epoch": 0.1965942488330548,
      "grad_norm": 0.9987673759460449,
      "learning_rate": 9.2898420108156e-06,
      "loss": 0.7819,
      "step": 6823
    },
    {
      "epoch": 0.19662306229470408,
      "grad_norm": 1.0223052501678467,
      "learning_rate": 9.28960229336487e-06,
      "loss": 0.7832,
      "step": 6824
    },
    {
      "epoch": 0.19665187575635337,
      "grad_norm": 0.9786959886550903,
      "learning_rate": 9.289362538555995e-06,
      "loss": 0.7772,
      "step": 6825
    },
    {
      "epoch": 0.19668068921800266,
      "grad_norm": 1.0136299133300781,
      "learning_rate": 9.28912274639106e-06,
      "loss": 0.8051,
      "step": 6826
    },
    {
      "epoch": 0.19670950267965193,
      "grad_norm": 1.0143462419509888,
      "learning_rate": 9.288882916872151e-06,
      "loss": 0.7787,
      "step": 6827
    },
    {
      "epoch": 0.19673831614130122,
      "grad_norm": 0.9597243070602417,
      "learning_rate": 9.288643050001362e-06,
      "loss": 0.7741,
      "step": 6828
    },
    {
      "epoch": 0.1967671296029505,
      "grad_norm": 1.032854437828064,
      "learning_rate": 9.288403145780778e-06,
      "loss": 0.7881,
      "step": 6829
    },
    {
      "epoch": 0.19679594306459977,
      "grad_norm": 1.0571303367614746,
      "learning_rate": 9.288163204212489e-06,
      "loss": 0.752,
      "step": 6830
    },
    {
      "epoch": 0.19682475652624906,
      "grad_norm": 1.0501834154129028,
      "learning_rate": 9.287923225298585e-06,
      "loss": 0.7985,
      "step": 6831
    },
    {
      "epoch": 0.19685356998789835,
      "grad_norm": 1.0263737440109253,
      "learning_rate": 9.287683209041156e-06,
      "loss": 0.7627,
      "step": 6832
    },
    {
      "epoch": 0.19688238344954762,
      "grad_norm": 1.0366448163986206,
      "learning_rate": 9.287443155442292e-06,
      "loss": 0.7716,
      "step": 6833
    },
    {
      "epoch": 0.1969111969111969,
      "grad_norm": 1.0242729187011719,
      "learning_rate": 9.287203064504084e-06,
      "loss": 0.7668,
      "step": 6834
    },
    {
      "epoch": 0.1969400103728462,
      "grad_norm": 0.9748502373695374,
      "learning_rate": 9.286962936228626e-06,
      "loss": 0.7779,
      "step": 6835
    },
    {
      "epoch": 0.1969688238344955,
      "grad_norm": 1.0204728841781616,
      "learning_rate": 9.286722770618003e-06,
      "loss": 0.7823,
      "step": 6836
    },
    {
      "epoch": 0.19699763729614475,
      "grad_norm": 1.030202865600586,
      "learning_rate": 9.286482567674311e-06,
      "loss": 0.7635,
      "step": 6837
    },
    {
      "epoch": 0.19702645075779404,
      "grad_norm": 0.9731404185295105,
      "learning_rate": 9.28624232739964e-06,
      "loss": 0.7748,
      "step": 6838
    },
    {
      "epoch": 0.19705526421944333,
      "grad_norm": 1.0362534523010254,
      "learning_rate": 9.286002049796084e-06,
      "loss": 0.7894,
      "step": 6839
    },
    {
      "epoch": 0.1970840776810926,
      "grad_norm": 0.999468982219696,
      "learning_rate": 9.285761734865732e-06,
      "loss": 0.762,
      "step": 6840
    },
    {
      "epoch": 0.1971128911427419,
      "grad_norm": 1.0329140424728394,
      "learning_rate": 9.285521382610683e-06,
      "loss": 0.7879,
      "step": 6841
    },
    {
      "epoch": 0.19714170460439118,
      "grad_norm": 1.0160491466522217,
      "learning_rate": 9.285280993033024e-06,
      "loss": 0.785,
      "step": 6842
    },
    {
      "epoch": 0.19717051806604047,
      "grad_norm": 1.0408605337142944,
      "learning_rate": 9.285040566134853e-06,
      "loss": 0.758,
      "step": 6843
    },
    {
      "epoch": 0.19719933152768973,
      "grad_norm": 1.0520198345184326,
      "learning_rate": 9.284800101918262e-06,
      "loss": 0.8003,
      "step": 6844
    },
    {
      "epoch": 0.19722814498933902,
      "grad_norm": 0.9896054863929749,
      "learning_rate": 9.284559600385346e-06,
      "loss": 0.778,
      "step": 6845
    },
    {
      "epoch": 0.1972569584509883,
      "grad_norm": 1.1030406951904297,
      "learning_rate": 9.284319061538196e-06,
      "loss": 0.7624,
      "step": 6846
    },
    {
      "epoch": 0.19728577191263758,
      "grad_norm": 1.0229947566986084,
      "learning_rate": 9.284078485378913e-06,
      "loss": 0.7576,
      "step": 6847
    },
    {
      "epoch": 0.19731458537428687,
      "grad_norm": 0.9554161429405212,
      "learning_rate": 9.283837871909587e-06,
      "loss": 0.8089,
      "step": 6848
    },
    {
      "epoch": 0.19734339883593616,
      "grad_norm": 1.6234904527664185,
      "learning_rate": 9.283597221132315e-06,
      "loss": 0.7701,
      "step": 6849
    },
    {
      "epoch": 0.19737221229758542,
      "grad_norm": 1.0253795385360718,
      "learning_rate": 9.283356533049195e-06,
      "loss": 0.803,
      "step": 6850
    },
    {
      "epoch": 0.1974010257592347,
      "grad_norm": 0.96783047914505,
      "learning_rate": 9.28311580766232e-06,
      "loss": 0.773,
      "step": 6851
    },
    {
      "epoch": 0.197429839220884,
      "grad_norm": 0.957921028137207,
      "learning_rate": 9.282875044973789e-06,
      "loss": 0.7779,
      "step": 6852
    },
    {
      "epoch": 0.1974586526825333,
      "grad_norm": 1.0252575874328613,
      "learning_rate": 9.282634244985697e-06,
      "loss": 0.7834,
      "step": 6853
    },
    {
      "epoch": 0.19748746614418256,
      "grad_norm": 0.9943679571151733,
      "learning_rate": 9.282393407700141e-06,
      "loss": 0.7894,
      "step": 6854
    },
    {
      "epoch": 0.19751627960583185,
      "grad_norm": 1.0040146112442017,
      "learning_rate": 9.28215253311922e-06,
      "loss": 0.7627,
      "step": 6855
    },
    {
      "epoch": 0.19754509306748114,
      "grad_norm": 1.0093978643417358,
      "learning_rate": 9.281911621245033e-06,
      "loss": 0.7533,
      "step": 6856
    },
    {
      "epoch": 0.1975739065291304,
      "grad_norm": 1.1049151420593262,
      "learning_rate": 9.281670672079672e-06,
      "loss": 0.7773,
      "step": 6857
    },
    {
      "epoch": 0.1976027199907797,
      "grad_norm": 1.0597599744796753,
      "learning_rate": 9.281429685625242e-06,
      "loss": 0.7826,
      "step": 6858
    },
    {
      "epoch": 0.19763153345242898,
      "grad_norm": 1.0139434337615967,
      "learning_rate": 9.28118866188384e-06,
      "loss": 0.8073,
      "step": 6859
    },
    {
      "epoch": 0.19766034691407824,
      "grad_norm": 1.1033989191055298,
      "learning_rate": 9.280947600857563e-06,
      "loss": 0.8055,
      "step": 6860
    },
    {
      "epoch": 0.19768916037572754,
      "grad_norm": 1.0694146156311035,
      "learning_rate": 9.280706502548512e-06,
      "loss": 0.7736,
      "step": 6861
    },
    {
      "epoch": 0.19771797383737683,
      "grad_norm": 0.9996665120124817,
      "learning_rate": 9.280465366958786e-06,
      "loss": 0.7842,
      "step": 6862
    },
    {
      "epoch": 0.19774678729902612,
      "grad_norm": 1.0245579481124878,
      "learning_rate": 9.280224194090487e-06,
      "loss": 0.7639,
      "step": 6863
    },
    {
      "epoch": 0.19777560076067538,
      "grad_norm": 1.0716898441314697,
      "learning_rate": 9.279982983945713e-06,
      "loss": 0.8095,
      "step": 6864
    },
    {
      "epoch": 0.19780441422232467,
      "grad_norm": 1.061044454574585,
      "learning_rate": 9.279741736526567e-06,
      "loss": 0.7987,
      "step": 6865
    },
    {
      "epoch": 0.19783322768397396,
      "grad_norm": 1.0297940969467163,
      "learning_rate": 9.279500451835146e-06,
      "loss": 0.7898,
      "step": 6866
    },
    {
      "epoch": 0.19786204114562322,
      "grad_norm": 0.9968045949935913,
      "learning_rate": 9.279259129873555e-06,
      "loss": 0.8003,
      "step": 6867
    },
    {
      "epoch": 0.19789085460727251,
      "grad_norm": 0.9964064955711365,
      "learning_rate": 9.279017770643896e-06,
      "loss": 0.7677,
      "step": 6868
    },
    {
      "epoch": 0.1979196680689218,
      "grad_norm": 0.9992887377738953,
      "learning_rate": 9.278776374148269e-06,
      "loss": 0.7787,
      "step": 6869
    },
    {
      "epoch": 0.1979484815305711,
      "grad_norm": 1.0078747272491455,
      "learning_rate": 9.278534940388777e-06,
      "loss": 0.785,
      "step": 6870
    },
    {
      "epoch": 0.19797729499222036,
      "grad_norm": 1.0084418058395386,
      "learning_rate": 9.278293469367522e-06,
      "loss": 0.7877,
      "step": 6871
    },
    {
      "epoch": 0.19800610845386965,
      "grad_norm": 0.9632179737091064,
      "learning_rate": 9.278051961086607e-06,
      "loss": 0.7588,
      "step": 6872
    },
    {
      "epoch": 0.19803492191551894,
      "grad_norm": 1.0320603847503662,
      "learning_rate": 9.277810415548138e-06,
      "loss": 0.7742,
      "step": 6873
    },
    {
      "epoch": 0.1980637353771682,
      "grad_norm": 1.0387623310089111,
      "learning_rate": 9.277568832754216e-06,
      "loss": 0.7991,
      "step": 6874
    },
    {
      "epoch": 0.1980925488388175,
      "grad_norm": 1.0213490724563599,
      "learning_rate": 9.277327212706943e-06,
      "loss": 0.7717,
      "step": 6875
    },
    {
      "epoch": 0.19812136230046679,
      "grad_norm": 1.0788532495498657,
      "learning_rate": 9.27708555540843e-06,
      "loss": 0.7866,
      "step": 6876
    },
    {
      "epoch": 0.19815017576211605,
      "grad_norm": 1.0226274728775024,
      "learning_rate": 9.276843860860774e-06,
      "loss": 0.793,
      "step": 6877
    },
    {
      "epoch": 0.19817898922376534,
      "grad_norm": 0.9806534647941589,
      "learning_rate": 9.276602129066084e-06,
      "loss": 0.7731,
      "step": 6878
    },
    {
      "epoch": 0.19820780268541463,
      "grad_norm": 0.9976248741149902,
      "learning_rate": 9.276360360026466e-06,
      "loss": 0.7658,
      "step": 6879
    },
    {
      "epoch": 0.19823661614706392,
      "grad_norm": 1.0255053043365479,
      "learning_rate": 9.276118553744022e-06,
      "loss": 0.787,
      "step": 6880
    },
    {
      "epoch": 0.19826542960871318,
      "grad_norm": 1.0655906200408936,
      "learning_rate": 9.275876710220861e-06,
      "loss": 0.815,
      "step": 6881
    },
    {
      "epoch": 0.19829424307036247,
      "grad_norm": 1.0686651468276978,
      "learning_rate": 9.275634829459088e-06,
      "loss": 0.7591,
      "step": 6882
    },
    {
      "epoch": 0.19832305653201177,
      "grad_norm": 1.0129334926605225,
      "learning_rate": 9.27539291146081e-06,
      "loss": 0.7892,
      "step": 6883
    },
    {
      "epoch": 0.19835186999366103,
      "grad_norm": 0.9861426949501038,
      "learning_rate": 9.275150956228132e-06,
      "loss": 0.777,
      "step": 6884
    },
    {
      "epoch": 0.19838068345531032,
      "grad_norm": 1.028806209564209,
      "learning_rate": 9.274908963763164e-06,
      "loss": 0.7826,
      "step": 6885
    },
    {
      "epoch": 0.1984094969169596,
      "grad_norm": 0.9730656147003174,
      "learning_rate": 9.274666934068013e-06,
      "loss": 0.7751,
      "step": 6886
    },
    {
      "epoch": 0.19843831037860887,
      "grad_norm": 1.076638102531433,
      "learning_rate": 9.274424867144785e-06,
      "loss": 0.8125,
      "step": 6887
    },
    {
      "epoch": 0.19846712384025816,
      "grad_norm": 0.9907661080360413,
      "learning_rate": 9.274182762995589e-06,
      "loss": 0.7816,
      "step": 6888
    },
    {
      "epoch": 0.19849593730190745,
      "grad_norm": 0.9910725951194763,
      "learning_rate": 9.273940621622535e-06,
      "loss": 0.7856,
      "step": 6889
    },
    {
      "epoch": 0.19852475076355675,
      "grad_norm": 0.9868327975273132,
      "learning_rate": 9.27369844302773e-06,
      "loss": 0.7857,
      "step": 6890
    },
    {
      "epoch": 0.198553564225206,
      "grad_norm": 0.9707314968109131,
      "learning_rate": 9.273456227213283e-06,
      "loss": 0.7804,
      "step": 6891
    },
    {
      "epoch": 0.1985823776868553,
      "grad_norm": 0.9777941107749939,
      "learning_rate": 9.273213974181305e-06,
      "loss": 0.7741,
      "step": 6892
    },
    {
      "epoch": 0.1986111911485046,
      "grad_norm": 0.9891629815101624,
      "learning_rate": 9.272971683933905e-06,
      "loss": 0.777,
      "step": 6893
    },
    {
      "epoch": 0.19864000461015385,
      "grad_norm": 0.9600427746772766,
      "learning_rate": 9.272729356473194e-06,
      "loss": 0.7679,
      "step": 6894
    },
    {
      "epoch": 0.19866881807180314,
      "grad_norm": 0.9758826494216919,
      "learning_rate": 9.27248699180128e-06,
      "loss": 0.8007,
      "step": 6895
    },
    {
      "epoch": 0.19869763153345243,
      "grad_norm": 0.9882736802101135,
      "learning_rate": 9.272244589920275e-06,
      "loss": 0.772,
      "step": 6896
    },
    {
      "epoch": 0.19872644499510173,
      "grad_norm": 1.0360684394836426,
      "learning_rate": 9.27200215083229e-06,
      "loss": 0.7709,
      "step": 6897
    },
    {
      "epoch": 0.198755258456751,
      "grad_norm": 1.0196260213851929,
      "learning_rate": 9.271759674539437e-06,
      "loss": 0.801,
      "step": 6898
    },
    {
      "epoch": 0.19878407191840028,
      "grad_norm": 1.0198780298233032,
      "learning_rate": 9.271517161043829e-06,
      "loss": 0.7531,
      "step": 6899
    },
    {
      "epoch": 0.19881288538004957,
      "grad_norm": 1.0402559041976929,
      "learning_rate": 9.271274610347575e-06,
      "loss": 0.7599,
      "step": 6900
    },
    {
      "epoch": 0.19884169884169883,
      "grad_norm": 0.9981577396392822,
      "learning_rate": 9.27103202245279e-06,
      "loss": 0.7561,
      "step": 6901
    },
    {
      "epoch": 0.19887051230334812,
      "grad_norm": 1.0832008123397827,
      "learning_rate": 9.270789397361583e-06,
      "loss": 0.7941,
      "step": 6902
    },
    {
      "epoch": 0.19889932576499741,
      "grad_norm": 1.0225610733032227,
      "learning_rate": 9.270546735076072e-06,
      "loss": 0.7711,
      "step": 6903
    },
    {
      "epoch": 0.19892813922664668,
      "grad_norm": 1.0416793823242188,
      "learning_rate": 9.270304035598368e-06,
      "loss": 0.7859,
      "step": 6904
    },
    {
      "epoch": 0.19895695268829597,
      "grad_norm": 1.0728614330291748,
      "learning_rate": 9.270061298930583e-06,
      "loss": 0.7819,
      "step": 6905
    },
    {
      "epoch": 0.19898576614994526,
      "grad_norm": 0.9973767995834351,
      "learning_rate": 9.269818525074834e-06,
      "loss": 0.7921,
      "step": 6906
    },
    {
      "epoch": 0.19901457961159455,
      "grad_norm": 1.028794765472412,
      "learning_rate": 9.269575714033233e-06,
      "loss": 0.7736,
      "step": 6907
    },
    {
      "epoch": 0.1990433930732438,
      "grad_norm": 1.0112462043762207,
      "learning_rate": 9.269332865807897e-06,
      "loss": 0.7805,
      "step": 6908
    },
    {
      "epoch": 0.1990722065348931,
      "grad_norm": 1.0066667795181274,
      "learning_rate": 9.269089980400938e-06,
      "loss": 0.7954,
      "step": 6909
    },
    {
      "epoch": 0.1991010199965424,
      "grad_norm": 0.99421226978302,
      "learning_rate": 9.268847057814473e-06,
      "loss": 0.7643,
      "step": 6910
    },
    {
      "epoch": 0.19912983345819166,
      "grad_norm": 0.946097195148468,
      "learning_rate": 9.268604098050617e-06,
      "loss": 0.7874,
      "step": 6911
    },
    {
      "epoch": 0.19915864691984095,
      "grad_norm": 1.0379146337509155,
      "learning_rate": 9.268361101111488e-06,
      "loss": 0.7776,
      "step": 6912
    },
    {
      "epoch": 0.19918746038149024,
      "grad_norm": 1.010028600692749,
      "learning_rate": 9.2681180669992e-06,
      "loss": 0.7967,
      "step": 6913
    },
    {
      "epoch": 0.1992162738431395,
      "grad_norm": 1.024869680404663,
      "learning_rate": 9.267874995715868e-06,
      "loss": 0.7821,
      "step": 6914
    },
    {
      "epoch": 0.1992450873047888,
      "grad_norm": 0.9833165407180786,
      "learning_rate": 9.267631887263615e-06,
      "loss": 0.774,
      "step": 6915
    },
    {
      "epoch": 0.19927390076643808,
      "grad_norm": 0.9934638738632202,
      "learning_rate": 9.267388741644552e-06,
      "loss": 0.8007,
      "step": 6916
    },
    {
      "epoch": 0.19930271422808737,
      "grad_norm": 0.9960405826568604,
      "learning_rate": 9.267145558860796e-06,
      "loss": 0.7975,
      "step": 6917
    },
    {
      "epoch": 0.19933152768973664,
      "grad_norm": 0.9845392107963562,
      "learning_rate": 9.266902338914472e-06,
      "loss": 0.7767,
      "step": 6918
    },
    {
      "epoch": 0.19936034115138593,
      "grad_norm": 0.9947654604911804,
      "learning_rate": 9.266659081807694e-06,
      "loss": 0.7719,
      "step": 6919
    },
    {
      "epoch": 0.19938915461303522,
      "grad_norm": 0.9457389712333679,
      "learning_rate": 9.266415787542579e-06,
      "loss": 0.7741,
      "step": 6920
    },
    {
      "epoch": 0.19941796807468448,
      "grad_norm": 1.0589147806167603,
      "learning_rate": 9.266172456121248e-06,
      "loss": 0.7816,
      "step": 6921
    },
    {
      "epoch": 0.19944678153633377,
      "grad_norm": 0.9892321228981018,
      "learning_rate": 9.26592908754582e-06,
      "loss": 0.773,
      "step": 6922
    },
    {
      "epoch": 0.19947559499798306,
      "grad_norm": 0.9812189936637878,
      "learning_rate": 9.265685681818413e-06,
      "loss": 0.7726,
      "step": 6923
    },
    {
      "epoch": 0.19950440845963233,
      "grad_norm": 0.9343408942222595,
      "learning_rate": 9.265442238941146e-06,
      "loss": 0.7565,
      "step": 6924
    },
    {
      "epoch": 0.19953322192128162,
      "grad_norm": 0.9628275632858276,
      "learning_rate": 9.265198758916144e-06,
      "loss": 0.7661,
      "step": 6925
    },
    {
      "epoch": 0.1995620353829309,
      "grad_norm": 0.9995818138122559,
      "learning_rate": 9.264955241745523e-06,
      "loss": 0.7723,
      "step": 6926
    },
    {
      "epoch": 0.1995908488445802,
      "grad_norm": 0.9539040923118591,
      "learning_rate": 9.264711687431408e-06,
      "loss": 0.7888,
      "step": 6927
    },
    {
      "epoch": 0.19961966230622946,
      "grad_norm": 1.0356502532958984,
      "learning_rate": 9.264468095975914e-06,
      "loss": 0.7905,
      "step": 6928
    },
    {
      "epoch": 0.19964847576787875,
      "grad_norm": 1.0103017091751099,
      "learning_rate": 9.264224467381167e-06,
      "loss": 0.7685,
      "step": 6929
    },
    {
      "epoch": 0.19967728922952804,
      "grad_norm": 1.0153205394744873,
      "learning_rate": 9.263980801649286e-06,
      "loss": 0.8002,
      "step": 6930
    },
    {
      "epoch": 0.1997061026911773,
      "grad_norm": 0.9513406753540039,
      "learning_rate": 9.263737098782397e-06,
      "loss": 0.7573,
      "step": 6931
    },
    {
      "epoch": 0.1997349161528266,
      "grad_norm": 0.993701696395874,
      "learning_rate": 9.263493358782619e-06,
      "loss": 0.7727,
      "step": 6932
    },
    {
      "epoch": 0.1997637296144759,
      "grad_norm": 1.0002018213272095,
      "learning_rate": 9.263249581652075e-06,
      "loss": 0.7857,
      "step": 6933
    },
    {
      "epoch": 0.19979254307612518,
      "grad_norm": 0.9448992013931274,
      "learning_rate": 9.26300576739289e-06,
      "loss": 0.7566,
      "step": 6934
    },
    {
      "epoch": 0.19982135653777444,
      "grad_norm": 1.0112930536270142,
      "learning_rate": 9.262761916007186e-06,
      "loss": 0.7855,
      "step": 6935
    },
    {
      "epoch": 0.19985016999942373,
      "grad_norm": 1.0535857677459717,
      "learning_rate": 9.262518027497087e-06,
      "loss": 0.785,
      "step": 6936
    },
    {
      "epoch": 0.19987898346107302,
      "grad_norm": 1.007742166519165,
      "learning_rate": 9.262274101864716e-06,
      "loss": 0.779,
      "step": 6937
    },
    {
      "epoch": 0.1999077969227223,
      "grad_norm": 0.9657357335090637,
      "learning_rate": 9.2620301391122e-06,
      "loss": 0.7807,
      "step": 6938
    },
    {
      "epoch": 0.19993661038437158,
      "grad_norm": 1.0427207946777344,
      "learning_rate": 9.26178613924166e-06,
      "loss": 0.7955,
      "step": 6939
    },
    {
      "epoch": 0.19996542384602087,
      "grad_norm": 0.9725935459136963,
      "learning_rate": 9.261542102255226e-06,
      "loss": 0.7536,
      "step": 6940
    },
    {
      "epoch": 0.19999423730767013,
      "grad_norm": 0.9943670034408569,
      "learning_rate": 9.261298028155016e-06,
      "loss": 0.7976,
      "step": 6941
    },
    {
      "epoch": 0.20002305076931942,
      "grad_norm": 0.9543824195861816,
      "learning_rate": 9.261053916943164e-06,
      "loss": 0.7918,
      "step": 6942
    },
    {
      "epoch": 0.2000518642309687,
      "grad_norm": 0.9497678875923157,
      "learning_rate": 9.26080976862179e-06,
      "loss": 0.7739,
      "step": 6943
    },
    {
      "epoch": 0.200080677692618,
      "grad_norm": 0.9968233108520508,
      "learning_rate": 9.260565583193021e-06,
      "loss": 0.7751,
      "step": 6944
    },
    {
      "epoch": 0.20010949115426727,
      "grad_norm": 0.982767641544342,
      "learning_rate": 9.260321360658986e-06,
      "loss": 0.8003,
      "step": 6945
    },
    {
      "epoch": 0.20013830461591656,
      "grad_norm": 0.9702419638633728,
      "learning_rate": 9.260077101021811e-06,
      "loss": 0.7592,
      "step": 6946
    },
    {
      "epoch": 0.20016711807756585,
      "grad_norm": 1.0095362663269043,
      "learning_rate": 9.259832804283623e-06,
      "loss": 0.8089,
      "step": 6947
    },
    {
      "epoch": 0.2001959315392151,
      "grad_norm": 1.0002762079238892,
      "learning_rate": 9.259588470446548e-06,
      "loss": 0.7953,
      "step": 6948
    },
    {
      "epoch": 0.2002247450008644,
      "grad_norm": 0.953418493270874,
      "learning_rate": 9.259344099512717e-06,
      "loss": 0.7431,
      "step": 6949
    },
    {
      "epoch": 0.2002535584625137,
      "grad_norm": 1.067872166633606,
      "learning_rate": 9.259099691484257e-06,
      "loss": 0.7918,
      "step": 6950
    },
    {
      "epoch": 0.20028237192416296,
      "grad_norm": 0.9610307812690735,
      "learning_rate": 9.258855246363296e-06,
      "loss": 0.7824,
      "step": 6951
    },
    {
      "epoch": 0.20031118538581225,
      "grad_norm": 1.018384575843811,
      "learning_rate": 9.258610764151963e-06,
      "loss": 0.79,
      "step": 6952
    },
    {
      "epoch": 0.20033999884746154,
      "grad_norm": 1.0119788646697998,
      "learning_rate": 9.258366244852386e-06,
      "loss": 0.7723,
      "step": 6953
    },
    {
      "epoch": 0.20036881230911083,
      "grad_norm": 1.0305699110031128,
      "learning_rate": 9.258121688466697e-06,
      "loss": 0.7782,
      "step": 6954
    },
    {
      "epoch": 0.2003976257707601,
      "grad_norm": 1.03227961063385,
      "learning_rate": 9.257877094997024e-06,
      "loss": 0.8002,
      "step": 6955
    },
    {
      "epoch": 0.20042643923240938,
      "grad_norm": 1.0525938272476196,
      "learning_rate": 9.257632464445499e-06,
      "loss": 0.7891,
      "step": 6956
    },
    {
      "epoch": 0.20045525269405867,
      "grad_norm": 1.04208505153656,
      "learning_rate": 9.25738779681425e-06,
      "loss": 0.8093,
      "step": 6957
    },
    {
      "epoch": 0.20048406615570794,
      "grad_norm": 0.9752280712127686,
      "learning_rate": 9.25714309210541e-06,
      "loss": 0.7717,
      "step": 6958
    },
    {
      "epoch": 0.20051287961735723,
      "grad_norm": 1.0380076169967651,
      "learning_rate": 9.256898350321108e-06,
      "loss": 0.7605,
      "step": 6959
    },
    {
      "epoch": 0.20054169307900652,
      "grad_norm": 1.0279678106307983,
      "learning_rate": 9.256653571463479e-06,
      "loss": 0.7992,
      "step": 6960
    },
    {
      "epoch": 0.2005705065406558,
      "grad_norm": 0.9863850474357605,
      "learning_rate": 9.256408755534651e-06,
      "loss": 0.7835,
      "step": 6961
    },
    {
      "epoch": 0.20059932000230507,
      "grad_norm": 0.9958418607711792,
      "learning_rate": 9.256163902536756e-06,
      "loss": 0.7583,
      "step": 6962
    },
    {
      "epoch": 0.20062813346395436,
      "grad_norm": 1.0317442417144775,
      "learning_rate": 9.255919012471931e-06,
      "loss": 0.7662,
      "step": 6963
    },
    {
      "epoch": 0.20065694692560365,
      "grad_norm": 1.0209280252456665,
      "learning_rate": 9.255674085342304e-06,
      "loss": 0.7889,
      "step": 6964
    },
    {
      "epoch": 0.20068576038725291,
      "grad_norm": 0.9875801801681519,
      "learning_rate": 9.25542912115001e-06,
      "loss": 0.7778,
      "step": 6965
    },
    {
      "epoch": 0.2007145738489022,
      "grad_norm": 1.0261430740356445,
      "learning_rate": 9.255184119897182e-06,
      "loss": 0.7893,
      "step": 6966
    },
    {
      "epoch": 0.2007433873105515,
      "grad_norm": 0.974876880645752,
      "learning_rate": 9.254939081585956e-06,
      "loss": 0.7611,
      "step": 6967
    },
    {
      "epoch": 0.20077220077220076,
      "grad_norm": 1.0137771368026733,
      "learning_rate": 9.254694006218462e-06,
      "loss": 0.7953,
      "step": 6968
    },
    {
      "epoch": 0.20080101423385005,
      "grad_norm": 0.9790269732475281,
      "learning_rate": 9.254448893796837e-06,
      "loss": 0.8197,
      "step": 6969
    },
    {
      "epoch": 0.20082982769549934,
      "grad_norm": 1.011383295059204,
      "learning_rate": 9.254203744323216e-06,
      "loss": 0.7866,
      "step": 6970
    },
    {
      "epoch": 0.20085864115714863,
      "grad_norm": 1.0127254724502563,
      "learning_rate": 9.253958557799732e-06,
      "loss": 0.7889,
      "step": 6971
    },
    {
      "epoch": 0.2008874546187979,
      "grad_norm": 0.9608984589576721,
      "learning_rate": 9.25371333422852e-06,
      "loss": 0.7841,
      "step": 6972
    },
    {
      "epoch": 0.20091626808044719,
      "grad_norm": 0.9700909852981567,
      "learning_rate": 9.253468073611718e-06,
      "loss": 0.7854,
      "step": 6973
    },
    {
      "epoch": 0.20094508154209648,
      "grad_norm": 1.025579810142517,
      "learning_rate": 9.253222775951462e-06,
      "loss": 0.7893,
      "step": 6974
    },
    {
      "epoch": 0.20097389500374574,
      "grad_norm": 1.01553213596344,
      "learning_rate": 9.252977441249886e-06,
      "loss": 0.7778,
      "step": 6975
    },
    {
      "epoch": 0.20100270846539503,
      "grad_norm": 0.972838282585144,
      "learning_rate": 9.25273206950913e-06,
      "loss": 0.7807,
      "step": 6976
    },
    {
      "epoch": 0.20103152192704432,
      "grad_norm": 0.9703779220581055,
      "learning_rate": 9.252486660731325e-06,
      "loss": 0.7893,
      "step": 6977
    },
    {
      "epoch": 0.20106033538869358,
      "grad_norm": 1.0076346397399902,
      "learning_rate": 9.252241214918616e-06,
      "loss": 0.7962,
      "step": 6978
    },
    {
      "epoch": 0.20108914885034287,
      "grad_norm": 1.0628384351730347,
      "learning_rate": 9.251995732073134e-06,
      "loss": 0.7546,
      "step": 6979
    },
    {
      "epoch": 0.20111796231199217,
      "grad_norm": 0.9838549494743347,
      "learning_rate": 9.251750212197022e-06,
      "loss": 0.7711,
      "step": 6980
    },
    {
      "epoch": 0.20114677577364146,
      "grad_norm": 1.0400651693344116,
      "learning_rate": 9.251504655292413e-06,
      "loss": 0.7819,
      "step": 6981
    },
    {
      "epoch": 0.20117558923529072,
      "grad_norm": 1.0235328674316406,
      "learning_rate": 9.25125906136145e-06,
      "loss": 0.78,
      "step": 6982
    },
    {
      "epoch": 0.20120440269694,
      "grad_norm": 0.9520480632781982,
      "learning_rate": 9.25101343040627e-06,
      "loss": 0.7803,
      "step": 6983
    },
    {
      "epoch": 0.2012332161585893,
      "grad_norm": 1.0012086629867554,
      "learning_rate": 9.250767762429014e-06,
      "loss": 0.7805,
      "step": 6984
    },
    {
      "epoch": 0.20126202962023856,
      "grad_norm": 1.0492922067642212,
      "learning_rate": 9.25052205743182e-06,
      "loss": 0.7882,
      "step": 6985
    },
    {
      "epoch": 0.20129084308188785,
      "grad_norm": 1.0192618370056152,
      "learning_rate": 9.250276315416826e-06,
      "loss": 0.7885,
      "step": 6986
    },
    {
      "epoch": 0.20131965654353715,
      "grad_norm": 1.0255683660507202,
      "learning_rate": 9.250030536386174e-06,
      "loss": 0.776,
      "step": 6987
    },
    {
      "epoch": 0.20134847000518644,
      "grad_norm": 1.0645225048065186,
      "learning_rate": 9.249784720342006e-06,
      "loss": 0.7705,
      "step": 6988
    },
    {
      "epoch": 0.2013772834668357,
      "grad_norm": 0.9522028565406799,
      "learning_rate": 9.24953886728646e-06,
      "loss": 0.7721,
      "step": 6989
    },
    {
      "epoch": 0.201406096928485,
      "grad_norm": 0.9825756549835205,
      "learning_rate": 9.249292977221677e-06,
      "loss": 0.7607,
      "step": 6990
    },
    {
      "epoch": 0.20143491039013428,
      "grad_norm": 1.0507688522338867,
      "learning_rate": 9.249047050149803e-06,
      "loss": 0.7621,
      "step": 6991
    },
    {
      "epoch": 0.20146372385178354,
      "grad_norm": 0.9692363142967224,
      "learning_rate": 9.248801086072975e-06,
      "loss": 0.7557,
      "step": 6992
    },
    {
      "epoch": 0.20149253731343283,
      "grad_norm": 0.9897730946540833,
      "learning_rate": 9.248555084993338e-06,
      "loss": 0.7893,
      "step": 6993
    },
    {
      "epoch": 0.20152135077508213,
      "grad_norm": 0.9894453287124634,
      "learning_rate": 9.248309046913032e-06,
      "loss": 0.7779,
      "step": 6994
    },
    {
      "epoch": 0.2015501642367314,
      "grad_norm": 1.013403058052063,
      "learning_rate": 9.248062971834202e-06,
      "loss": 0.7666,
      "step": 6995
    },
    {
      "epoch": 0.20157897769838068,
      "grad_norm": 0.9376949667930603,
      "learning_rate": 9.247816859758989e-06,
      "loss": 0.7905,
      "step": 6996
    },
    {
      "epoch": 0.20160779116002997,
      "grad_norm": 0.962307870388031,
      "learning_rate": 9.247570710689537e-06,
      "loss": 0.7785,
      "step": 6997
    },
    {
      "epoch": 0.20163660462167926,
      "grad_norm": 1.0493626594543457,
      "learning_rate": 9.247324524627992e-06,
      "loss": 0.7973,
      "step": 6998
    },
    {
      "epoch": 0.20166541808332852,
      "grad_norm": 1.0105623006820679,
      "learning_rate": 9.247078301576496e-06,
      "loss": 0.8245,
      "step": 6999
    },
    {
      "epoch": 0.20169423154497781,
      "grad_norm": 0.9397611021995544,
      "learning_rate": 9.246832041537193e-06,
      "loss": 0.7807,
      "step": 7000
    },
    {
      "epoch": 0.2017230450066271,
      "grad_norm": 0.9865788817405701,
      "learning_rate": 9.246585744512226e-06,
      "loss": 0.7927,
      "step": 7001
    },
    {
      "epoch": 0.20175185846827637,
      "grad_norm": 0.9719786643981934,
      "learning_rate": 9.246339410503745e-06,
      "loss": 0.7982,
      "step": 7002
    },
    {
      "epoch": 0.20178067192992566,
      "grad_norm": 0.9569240212440491,
      "learning_rate": 9.24609303951389e-06,
      "loss": 0.7843,
      "step": 7003
    },
    {
      "epoch": 0.20180948539157495,
      "grad_norm": 0.9581137895584106,
      "learning_rate": 9.245846631544812e-06,
      "loss": 0.7869,
      "step": 7004
    },
    {
      "epoch": 0.2018382988532242,
      "grad_norm": 0.9596526026725769,
      "learning_rate": 9.245600186598653e-06,
      "loss": 0.7653,
      "step": 7005
    },
    {
      "epoch": 0.2018671123148735,
      "grad_norm": 0.9879555106163025,
      "learning_rate": 9.24535370467756e-06,
      "loss": 0.7832,
      "step": 7006
    },
    {
      "epoch": 0.2018959257765228,
      "grad_norm": 0.9964126944541931,
      "learning_rate": 9.24510718578368e-06,
      "loss": 0.7654,
      "step": 7007
    },
    {
      "epoch": 0.20192473923817209,
      "grad_norm": 0.9518215656280518,
      "learning_rate": 9.24486062991916e-06,
      "loss": 0.7797,
      "step": 7008
    },
    {
      "epoch": 0.20195355269982135,
      "grad_norm": 0.9632174968719482,
      "learning_rate": 9.244614037086148e-06,
      "loss": 0.7892,
      "step": 7009
    },
    {
      "epoch": 0.20198236616147064,
      "grad_norm": 0.9615598320960999,
      "learning_rate": 9.244367407286791e-06,
      "loss": 0.7815,
      "step": 7010
    },
    {
      "epoch": 0.20201117962311993,
      "grad_norm": 0.9594351649284363,
      "learning_rate": 9.244120740523236e-06,
      "loss": 0.7667,
      "step": 7011
    },
    {
      "epoch": 0.2020399930847692,
      "grad_norm": 0.9698454141616821,
      "learning_rate": 9.24387403679763e-06,
      "loss": 0.7843,
      "step": 7012
    },
    {
      "epoch": 0.20206880654641848,
      "grad_norm": 0.9695853590965271,
      "learning_rate": 9.243627296112128e-06,
      "loss": 0.7912,
      "step": 7013
    },
    {
      "epoch": 0.20209762000806777,
      "grad_norm": 0.9638190865516663,
      "learning_rate": 9.243380518468873e-06,
      "loss": 0.7725,
      "step": 7014
    },
    {
      "epoch": 0.20212643346971707,
      "grad_norm": 0.9811990261077881,
      "learning_rate": 9.243133703870012e-06,
      "loss": 0.7505,
      "step": 7015
    },
    {
      "epoch": 0.20215524693136633,
      "grad_norm": 1.0156673192977905,
      "learning_rate": 9.2428868523177e-06,
      "loss": 0.7819,
      "step": 7016
    },
    {
      "epoch": 0.20218406039301562,
      "grad_norm": 0.9922874569892883,
      "learning_rate": 9.242639963814086e-06,
      "loss": 0.7944,
      "step": 7017
    },
    {
      "epoch": 0.2022128738546649,
      "grad_norm": 0.9570931196212769,
      "learning_rate": 9.242393038361316e-06,
      "loss": 0.7765,
      "step": 7018
    },
    {
      "epoch": 0.20224168731631417,
      "grad_norm": 0.9979106187820435,
      "learning_rate": 9.242146075961547e-06,
      "loss": 0.7726,
      "step": 7019
    },
    {
      "epoch": 0.20227050077796346,
      "grad_norm": 0.9838657379150391,
      "learning_rate": 9.241899076616925e-06,
      "loss": 0.7729,
      "step": 7020
    },
    {
      "epoch": 0.20229931423961275,
      "grad_norm": 0.9645748734474182,
      "learning_rate": 9.241652040329602e-06,
      "loss": 0.7852,
      "step": 7021
    },
    {
      "epoch": 0.20232812770126202,
      "grad_norm": 0.9849443435668945,
      "learning_rate": 9.24140496710173e-06,
      "loss": 0.7913,
      "step": 7022
    },
    {
      "epoch": 0.2023569411629113,
      "grad_norm": 1.0149435997009277,
      "learning_rate": 9.241157856935462e-06,
      "loss": 0.7902,
      "step": 7023
    },
    {
      "epoch": 0.2023857546245606,
      "grad_norm": 0.9550697207450867,
      "learning_rate": 9.240910709832946e-06,
      "loss": 0.7678,
      "step": 7024
    },
    {
      "epoch": 0.2024145680862099,
      "grad_norm": 0.9679924845695496,
      "learning_rate": 9.240663525796339e-06,
      "loss": 0.7779,
      "step": 7025
    },
    {
      "epoch": 0.20244338154785915,
      "grad_norm": 0.9981127977371216,
      "learning_rate": 9.24041630482779e-06,
      "loss": 0.7818,
      "step": 7026
    },
    {
      "epoch": 0.20247219500950844,
      "grad_norm": 1.0400450229644775,
      "learning_rate": 9.240169046929456e-06,
      "loss": 0.7861,
      "step": 7027
    },
    {
      "epoch": 0.20250100847115773,
      "grad_norm": 0.9980404376983643,
      "learning_rate": 9.239921752103487e-06,
      "loss": 0.7459,
      "step": 7028
    },
    {
      "epoch": 0.202529821932807,
      "grad_norm": 0.9856496453285217,
      "learning_rate": 9.239674420352038e-06,
      "loss": 0.7631,
      "step": 7029
    },
    {
      "epoch": 0.2025586353944563,
      "grad_norm": 0.9782190918922424,
      "learning_rate": 9.239427051677264e-06,
      "loss": 0.7684,
      "step": 7030
    },
    {
      "epoch": 0.20258744885610558,
      "grad_norm": 1.0648008584976196,
      "learning_rate": 9.239179646081318e-06,
      "loss": 0.7745,
      "step": 7031
    },
    {
      "epoch": 0.20261626231775484,
      "grad_norm": 1.0285378694534302,
      "learning_rate": 9.238932203566354e-06,
      "loss": 0.7789,
      "step": 7032
    },
    {
      "epoch": 0.20264507577940413,
      "grad_norm": 0.9965730905532837,
      "learning_rate": 9.238684724134528e-06,
      "loss": 0.7691,
      "step": 7033
    },
    {
      "epoch": 0.20267388924105342,
      "grad_norm": 0.9296735525131226,
      "learning_rate": 9.238437207787994e-06,
      "loss": 0.7851,
      "step": 7034
    },
    {
      "epoch": 0.20270270270270271,
      "grad_norm": 1.0082608461380005,
      "learning_rate": 9.23818965452891e-06,
      "loss": 0.7868,
      "step": 7035
    },
    {
      "epoch": 0.20273151616435198,
      "grad_norm": 1.0092192888259888,
      "learning_rate": 9.23794206435943e-06,
      "loss": 0.7877,
      "step": 7036
    },
    {
      "epoch": 0.20276032962600127,
      "grad_norm": 0.9953212738037109,
      "learning_rate": 9.237694437281713e-06,
      "loss": 0.7789,
      "step": 7037
    },
    {
      "epoch": 0.20278914308765056,
      "grad_norm": 0.9629499316215515,
      "learning_rate": 9.237446773297914e-06,
      "loss": 0.7689,
      "step": 7038
    },
    {
      "epoch": 0.20281795654929982,
      "grad_norm": 0.9272571206092834,
      "learning_rate": 9.237199072410188e-06,
      "loss": 0.7665,
      "step": 7039
    },
    {
      "epoch": 0.2028467700109491,
      "grad_norm": 0.9894314408302307,
      "learning_rate": 9.236951334620694e-06,
      "loss": 0.7985,
      "step": 7040
    },
    {
      "epoch": 0.2028755834725984,
      "grad_norm": 0.954206109046936,
      "learning_rate": 9.236703559931588e-06,
      "loss": 0.7488,
      "step": 7041
    },
    {
      "epoch": 0.2029043969342477,
      "grad_norm": 0.972051203250885,
      "learning_rate": 9.23645574834503e-06,
      "loss": 0.7617,
      "step": 7042
    },
    {
      "epoch": 0.20293321039589696,
      "grad_norm": 0.9855139255523682,
      "learning_rate": 9.236207899863177e-06,
      "loss": 0.7854,
      "step": 7043
    },
    {
      "epoch": 0.20296202385754625,
      "grad_norm": 0.9739845991134644,
      "learning_rate": 9.235960014488188e-06,
      "loss": 0.7763,
      "step": 7044
    },
    {
      "epoch": 0.20299083731919554,
      "grad_norm": 0.9997241497039795,
      "learning_rate": 9.235712092222223e-06,
      "loss": 0.7637,
      "step": 7045
    },
    {
      "epoch": 0.2030196507808448,
      "grad_norm": 0.962570309638977,
      "learning_rate": 9.23546413306744e-06,
      "loss": 0.7918,
      "step": 7046
    },
    {
      "epoch": 0.2030484642424941,
      "grad_norm": 0.9670509099960327,
      "learning_rate": 9.235216137025997e-06,
      "loss": 0.7821,
      "step": 7047
    },
    {
      "epoch": 0.20307727770414338,
      "grad_norm": 0.9843698143959045,
      "learning_rate": 9.234968104100056e-06,
      "loss": 0.7593,
      "step": 7048
    },
    {
      "epoch": 0.20310609116579265,
      "grad_norm": 1.0017060041427612,
      "learning_rate": 9.234720034291774e-06,
      "loss": 0.7726,
      "step": 7049
    },
    {
      "epoch": 0.20313490462744194,
      "grad_norm": 0.991644024848938,
      "learning_rate": 9.234471927603314e-06,
      "loss": 0.7857,
      "step": 7050
    },
    {
      "epoch": 0.20316371808909123,
      "grad_norm": 0.9888482093811035,
      "learning_rate": 9.234223784036839e-06,
      "loss": 0.7762,
      "step": 7051
    },
    {
      "epoch": 0.20319253155074052,
      "grad_norm": 0.9668787121772766,
      "learning_rate": 9.233975603594506e-06,
      "loss": 0.7413,
      "step": 7052
    },
    {
      "epoch": 0.20322134501238978,
      "grad_norm": 1.010190486907959,
      "learning_rate": 9.233727386278478e-06,
      "loss": 0.7932,
      "step": 7053
    },
    {
      "epoch": 0.20325015847403907,
      "grad_norm": 0.9425225257873535,
      "learning_rate": 9.233479132090917e-06,
      "loss": 0.761,
      "step": 7054
    },
    {
      "epoch": 0.20327897193568836,
      "grad_norm": 1.0231515169143677,
      "learning_rate": 9.233230841033984e-06,
      "loss": 0.7632,
      "step": 7055
    },
    {
      "epoch": 0.20330778539733763,
      "grad_norm": 0.971539318561554,
      "learning_rate": 9.232982513109842e-06,
      "loss": 0.7749,
      "step": 7056
    },
    {
      "epoch": 0.20333659885898692,
      "grad_norm": 0.9782581329345703,
      "learning_rate": 9.232734148320655e-06,
      "loss": 0.7862,
      "step": 7057
    },
    {
      "epoch": 0.2033654123206362,
      "grad_norm": 1.0076411962509155,
      "learning_rate": 9.232485746668583e-06,
      "loss": 0.7569,
      "step": 7058
    },
    {
      "epoch": 0.20339422578228547,
      "grad_norm": 1.0178258419036865,
      "learning_rate": 9.232237308155794e-06,
      "loss": 0.7836,
      "step": 7059
    },
    {
      "epoch": 0.20342303924393476,
      "grad_norm": 0.9822106957435608,
      "learning_rate": 9.231988832784445e-06,
      "loss": 0.7856,
      "step": 7060
    },
    {
      "epoch": 0.20345185270558405,
      "grad_norm": 0.9822602868080139,
      "learning_rate": 9.231740320556706e-06,
      "loss": 0.7831,
      "step": 7061
    },
    {
      "epoch": 0.20348066616723334,
      "grad_norm": 1.0288355350494385,
      "learning_rate": 9.231491771474739e-06,
      "loss": 0.7826,
      "step": 7062
    },
    {
      "epoch": 0.2035094796288826,
      "grad_norm": 1.0278611183166504,
      "learning_rate": 9.231243185540709e-06,
      "loss": 0.7825,
      "step": 7063
    },
    {
      "epoch": 0.2035382930905319,
      "grad_norm": 0.9682940244674683,
      "learning_rate": 9.230994562756778e-06,
      "loss": 0.7845,
      "step": 7064
    },
    {
      "epoch": 0.2035671065521812,
      "grad_norm": 0.9584320187568665,
      "learning_rate": 9.230745903125115e-06,
      "loss": 0.7917,
      "step": 7065
    },
    {
      "epoch": 0.20359592001383045,
      "grad_norm": 0.9689512252807617,
      "learning_rate": 9.230497206647886e-06,
      "loss": 0.7814,
      "step": 7066
    },
    {
      "epoch": 0.20362473347547974,
      "grad_norm": 1.0019969940185547,
      "learning_rate": 9.230248473327254e-06,
      "loss": 0.7906,
      "step": 7067
    },
    {
      "epoch": 0.20365354693712903,
      "grad_norm": 1.0012357234954834,
      "learning_rate": 9.229999703165386e-06,
      "loss": 0.767,
      "step": 7068
    },
    {
      "epoch": 0.20368236039877832,
      "grad_norm": 0.9771924614906311,
      "learning_rate": 9.22975089616445e-06,
      "loss": 0.7803,
      "step": 7069
    },
    {
      "epoch": 0.20371117386042759,
      "grad_norm": 1.0072780847549438,
      "learning_rate": 9.229502052326611e-06,
      "loss": 0.7624,
      "step": 7070
    },
    {
      "epoch": 0.20373998732207688,
      "grad_norm": 0.9789117574691772,
      "learning_rate": 9.229253171654039e-06,
      "loss": 0.7867,
      "step": 7071
    },
    {
      "epoch": 0.20376880078372617,
      "grad_norm": 0.9886066317558289,
      "learning_rate": 9.229004254148897e-06,
      "loss": 0.7528,
      "step": 7072
    },
    {
      "epoch": 0.20379761424537543,
      "grad_norm": 1.0214406251907349,
      "learning_rate": 9.228755299813357e-06,
      "loss": 0.8062,
      "step": 7073
    },
    {
      "epoch": 0.20382642770702472,
      "grad_norm": 0.9488570690155029,
      "learning_rate": 9.228506308649585e-06,
      "loss": 0.7707,
      "step": 7074
    },
    {
      "epoch": 0.203855241168674,
      "grad_norm": 0.9714361429214478,
      "learning_rate": 9.22825728065975e-06,
      "loss": 0.7758,
      "step": 7075
    },
    {
      "epoch": 0.20388405463032327,
      "grad_norm": 1.0244619846343994,
      "learning_rate": 9.22800821584602e-06,
      "loss": 0.7852,
      "step": 7076
    },
    {
      "epoch": 0.20391286809197257,
      "grad_norm": 1.0044833421707153,
      "learning_rate": 9.227759114210568e-06,
      "loss": 0.7828,
      "step": 7077
    },
    {
      "epoch": 0.20394168155362186,
      "grad_norm": 0.9740687608718872,
      "learning_rate": 9.227509975755557e-06,
      "loss": 0.7738,
      "step": 7078
    },
    {
      "epoch": 0.20397049501527115,
      "grad_norm": 0.9637404680252075,
      "learning_rate": 9.227260800483162e-06,
      "loss": 0.7961,
      "step": 7079
    },
    {
      "epoch": 0.2039993084769204,
      "grad_norm": 0.9436156153678894,
      "learning_rate": 9.22701158839555e-06,
      "loss": 0.7634,
      "step": 7080
    },
    {
      "epoch": 0.2040281219385697,
      "grad_norm": 1.0614814758300781,
      "learning_rate": 9.226762339494892e-06,
      "loss": 0.8025,
      "step": 7081
    },
    {
      "epoch": 0.204056935400219,
      "grad_norm": 0.9670392870903015,
      "learning_rate": 9.226513053783361e-06,
      "loss": 0.7743,
      "step": 7082
    },
    {
      "epoch": 0.20408574886186825,
      "grad_norm": 1.01778244972229,
      "learning_rate": 9.226263731263126e-06,
      "loss": 0.7679,
      "step": 7083
    },
    {
      "epoch": 0.20411456232351755,
      "grad_norm": 1.0252264738082886,
      "learning_rate": 9.226014371936358e-06,
      "loss": 0.7797,
      "step": 7084
    },
    {
      "epoch": 0.20414337578516684,
      "grad_norm": 1.0164945125579834,
      "learning_rate": 9.225764975805229e-06,
      "loss": 0.778,
      "step": 7085
    },
    {
      "epoch": 0.2041721892468161,
      "grad_norm": 1.0134038925170898,
      "learning_rate": 9.225515542871913e-06,
      "loss": 0.7844,
      "step": 7086
    },
    {
      "epoch": 0.2042010027084654,
      "grad_norm": 1.0225403308868408,
      "learning_rate": 9.225266073138578e-06,
      "loss": 0.7646,
      "step": 7087
    },
    {
      "epoch": 0.20422981617011468,
      "grad_norm": 1.0089513063430786,
      "learning_rate": 9.2250165666074e-06,
      "loss": 0.7638,
      "step": 7088
    },
    {
      "epoch": 0.20425862963176397,
      "grad_norm": 1.0055062770843506,
      "learning_rate": 9.224767023280551e-06,
      "loss": 0.7979,
      "step": 7089
    },
    {
      "epoch": 0.20428744309341323,
      "grad_norm": 0.9963708519935608,
      "learning_rate": 9.224517443160205e-06,
      "loss": 0.7771,
      "step": 7090
    },
    {
      "epoch": 0.20431625655506253,
      "grad_norm": 0.9977015256881714,
      "learning_rate": 9.224267826248536e-06,
      "loss": 0.7701,
      "step": 7091
    },
    {
      "epoch": 0.20434507001671182,
      "grad_norm": 1.0146106481552124,
      "learning_rate": 9.224018172547716e-06,
      "loss": 0.7696,
      "step": 7092
    },
    {
      "epoch": 0.20437388347836108,
      "grad_norm": 0.9833593964576721,
      "learning_rate": 9.22376848205992e-06,
      "loss": 0.7799,
      "step": 7093
    },
    {
      "epoch": 0.20440269694001037,
      "grad_norm": 1.0452032089233398,
      "learning_rate": 9.223518754787324e-06,
      "loss": 0.747,
      "step": 7094
    },
    {
      "epoch": 0.20443151040165966,
      "grad_norm": 1.0388189554214478,
      "learning_rate": 9.2232689907321e-06,
      "loss": 0.763,
      "step": 7095
    },
    {
      "epoch": 0.20446032386330895,
      "grad_norm": 1.0008351802825928,
      "learning_rate": 9.223019189896425e-06,
      "loss": 0.7824,
      "step": 7096
    },
    {
      "epoch": 0.20448913732495821,
      "grad_norm": 0.9626652598381042,
      "learning_rate": 9.222769352282474e-06,
      "loss": 0.7747,
      "step": 7097
    },
    {
      "epoch": 0.2045179507866075,
      "grad_norm": 1.0002752542495728,
      "learning_rate": 9.222519477892425e-06,
      "loss": 0.7868,
      "step": 7098
    },
    {
      "epoch": 0.2045467642482568,
      "grad_norm": 1.0201013088226318,
      "learning_rate": 9.22226956672845e-06,
      "loss": 0.7779,
      "step": 7099
    },
    {
      "epoch": 0.20457557770990606,
      "grad_norm": 0.9848450422286987,
      "learning_rate": 9.22201961879273e-06,
      "loss": 0.7738,
      "step": 7100
    },
    {
      "epoch": 0.20460439117155535,
      "grad_norm": 1.0115165710449219,
      "learning_rate": 9.22176963408744e-06,
      "loss": 0.7534,
      "step": 7101
    },
    {
      "epoch": 0.20463320463320464,
      "grad_norm": 1.0474876165390015,
      "learning_rate": 9.221519612614755e-06,
      "loss": 0.785,
      "step": 7102
    },
    {
      "epoch": 0.2046620180948539,
      "grad_norm": 0.9823766946792603,
      "learning_rate": 9.221269554376855e-06,
      "loss": 0.7657,
      "step": 7103
    },
    {
      "epoch": 0.2046908315565032,
      "grad_norm": 0.965924859046936,
      "learning_rate": 9.221019459375916e-06,
      "loss": 0.8168,
      "step": 7104
    },
    {
      "epoch": 0.20471964501815249,
      "grad_norm": 1.007049798965454,
      "learning_rate": 9.220769327614116e-06,
      "loss": 0.7732,
      "step": 7105
    },
    {
      "epoch": 0.20474845847980178,
      "grad_norm": 0.9960541725158691,
      "learning_rate": 9.220519159093638e-06,
      "loss": 0.7626,
      "step": 7106
    },
    {
      "epoch": 0.20477727194145104,
      "grad_norm": 1.010288953781128,
      "learning_rate": 9.220268953816654e-06,
      "loss": 0.8007,
      "step": 7107
    },
    {
      "epoch": 0.20480608540310033,
      "grad_norm": 0.973725438117981,
      "learning_rate": 9.220018711785346e-06,
      "loss": 0.7695,
      "step": 7108
    },
    {
      "epoch": 0.20483489886474962,
      "grad_norm": 1.0145524740219116,
      "learning_rate": 9.219768433001894e-06,
      "loss": 0.7874,
      "step": 7109
    },
    {
      "epoch": 0.20486371232639888,
      "grad_norm": 0.9994033575057983,
      "learning_rate": 9.219518117468479e-06,
      "loss": 0.7707,
      "step": 7110
    },
    {
      "epoch": 0.20489252578804817,
      "grad_norm": 0.9767507314682007,
      "learning_rate": 9.219267765187277e-06,
      "loss": 0.7795,
      "step": 7111
    },
    {
      "epoch": 0.20492133924969747,
      "grad_norm": 1.0129131078720093,
      "learning_rate": 9.219017376160473e-06,
      "loss": 0.7638,
      "step": 7112
    },
    {
      "epoch": 0.20495015271134673,
      "grad_norm": 0.9801667332649231,
      "learning_rate": 9.218766950390243e-06,
      "loss": 0.817,
      "step": 7113
    },
    {
      "epoch": 0.20497896617299602,
      "grad_norm": 0.9918999075889587,
      "learning_rate": 9.21851648787877e-06,
      "loss": 0.7616,
      "step": 7114
    },
    {
      "epoch": 0.2050077796346453,
      "grad_norm": 0.9425663352012634,
      "learning_rate": 9.218265988628237e-06,
      "loss": 0.7713,
      "step": 7115
    },
    {
      "epoch": 0.2050365930962946,
      "grad_norm": 1.0085877180099487,
      "learning_rate": 9.218015452640822e-06,
      "loss": 0.7802,
      "step": 7116
    },
    {
      "epoch": 0.20506540655794386,
      "grad_norm": 0.9835847020149231,
      "learning_rate": 9.21776487991871e-06,
      "loss": 0.7593,
      "step": 7117
    },
    {
      "epoch": 0.20509422001959315,
      "grad_norm": 0.9488531947135925,
      "learning_rate": 9.21751427046408e-06,
      "loss": 0.7681,
      "step": 7118
    },
    {
      "epoch": 0.20512303348124244,
      "grad_norm": 0.9391463994979858,
      "learning_rate": 9.217263624279121e-06,
      "loss": 0.7862,
      "step": 7119
    },
    {
      "epoch": 0.2051518469428917,
      "grad_norm": 1.015067219734192,
      "learning_rate": 9.217012941366008e-06,
      "loss": 0.7552,
      "step": 7120
    },
    {
      "epoch": 0.205180660404541,
      "grad_norm": 0.9915921688079834,
      "learning_rate": 9.21676222172693e-06,
      "loss": 0.7951,
      "step": 7121
    },
    {
      "epoch": 0.2052094738661903,
      "grad_norm": 0.9770553708076477,
      "learning_rate": 9.216511465364066e-06,
      "loss": 0.7802,
      "step": 7122
    },
    {
      "epoch": 0.20523828732783958,
      "grad_norm": 0.981048583984375,
      "learning_rate": 9.216260672279604e-06,
      "loss": 0.7804,
      "step": 7123
    },
    {
      "epoch": 0.20526710078948884,
      "grad_norm": 0.9900677800178528,
      "learning_rate": 9.216009842475724e-06,
      "loss": 0.792,
      "step": 7124
    },
    {
      "epoch": 0.20529591425113813,
      "grad_norm": 0.9942789077758789,
      "learning_rate": 9.215758975954615e-06,
      "loss": 0.7707,
      "step": 7125
    },
    {
      "epoch": 0.20532472771278742,
      "grad_norm": 0.990339457988739,
      "learning_rate": 9.215508072718458e-06,
      "loss": 0.7612,
      "step": 7126
    },
    {
      "epoch": 0.2053535411744367,
      "grad_norm": 0.9490044116973877,
      "learning_rate": 9.215257132769442e-06,
      "loss": 0.7797,
      "step": 7127
    },
    {
      "epoch": 0.20538235463608598,
      "grad_norm": 0.9599798917770386,
      "learning_rate": 9.215006156109746e-06,
      "loss": 0.7723,
      "step": 7128
    },
    {
      "epoch": 0.20541116809773527,
      "grad_norm": 0.9670884609222412,
      "learning_rate": 9.214755142741564e-06,
      "loss": 0.7629,
      "step": 7129
    },
    {
      "epoch": 0.20543998155938453,
      "grad_norm": 0.980168342590332,
      "learning_rate": 9.214504092667075e-06,
      "loss": 0.8,
      "step": 7130
    },
    {
      "epoch": 0.20546879502103382,
      "grad_norm": 0.9135002493858337,
      "learning_rate": 9.214253005888472e-06,
      "loss": 0.7634,
      "step": 7131
    },
    {
      "epoch": 0.20549760848268311,
      "grad_norm": 0.9584575891494751,
      "learning_rate": 9.214001882407935e-06,
      "loss": 0.7485,
      "step": 7132
    },
    {
      "epoch": 0.2055264219443324,
      "grad_norm": 0.980688750743866,
      "learning_rate": 9.213750722227655e-06,
      "loss": 0.7951,
      "step": 7133
    },
    {
      "epoch": 0.20555523540598167,
      "grad_norm": 0.962536096572876,
      "learning_rate": 9.213499525349817e-06,
      "loss": 0.7722,
      "step": 7134
    },
    {
      "epoch": 0.20558404886763096,
      "grad_norm": 0.9580338001251221,
      "learning_rate": 9.21324829177661e-06,
      "loss": 0.7743,
      "step": 7135
    },
    {
      "epoch": 0.20561286232928025,
      "grad_norm": 0.9559460878372192,
      "learning_rate": 9.212997021510225e-06,
      "loss": 0.7728,
      "step": 7136
    },
    {
      "epoch": 0.2056416757909295,
      "grad_norm": 0.9263920187950134,
      "learning_rate": 9.212745714552846e-06,
      "loss": 0.7747,
      "step": 7137
    },
    {
      "epoch": 0.2056704892525788,
      "grad_norm": 0.9586453437805176,
      "learning_rate": 9.212494370906662e-06,
      "loss": 0.7733,
      "step": 7138
    },
    {
      "epoch": 0.2056993027142281,
      "grad_norm": 0.9682595729827881,
      "learning_rate": 9.212242990573864e-06,
      "loss": 0.7784,
      "step": 7139
    },
    {
      "epoch": 0.20572811617587736,
      "grad_norm": 0.9819762110710144,
      "learning_rate": 9.21199157355664e-06,
      "loss": 0.796,
      "step": 7140
    },
    {
      "epoch": 0.20575692963752665,
      "grad_norm": 0.9610283374786377,
      "learning_rate": 9.21174011985718e-06,
      "loss": 0.7695,
      "step": 7141
    },
    {
      "epoch": 0.20578574309917594,
      "grad_norm": 0.9607659578323364,
      "learning_rate": 9.211488629477673e-06,
      "loss": 0.7967,
      "step": 7142
    },
    {
      "epoch": 0.20581455656082523,
      "grad_norm": 0.9983656406402588,
      "learning_rate": 9.21123710242031e-06,
      "loss": 0.7758,
      "step": 7143
    },
    {
      "epoch": 0.2058433700224745,
      "grad_norm": 1.0181723833084106,
      "learning_rate": 9.210985538687282e-06,
      "loss": 0.7772,
      "step": 7144
    },
    {
      "epoch": 0.20587218348412378,
      "grad_norm": 0.955666184425354,
      "learning_rate": 9.210733938280781e-06,
      "loss": 0.7688,
      "step": 7145
    },
    {
      "epoch": 0.20590099694577307,
      "grad_norm": 0.9935047030448914,
      "learning_rate": 9.210482301202995e-06,
      "loss": 0.7774,
      "step": 7146
    },
    {
      "epoch": 0.20592981040742234,
      "grad_norm": 0.9574035406112671,
      "learning_rate": 9.210230627456116e-06,
      "loss": 0.7857,
      "step": 7147
    },
    {
      "epoch": 0.20595862386907163,
      "grad_norm": 0.9651152491569519,
      "learning_rate": 9.20997891704234e-06,
      "loss": 0.7539,
      "step": 7148
    },
    {
      "epoch": 0.20598743733072092,
      "grad_norm": 0.9766582250595093,
      "learning_rate": 9.209727169963854e-06,
      "loss": 0.7799,
      "step": 7149
    },
    {
      "epoch": 0.2060162507923702,
      "grad_norm": 0.9113829731941223,
      "learning_rate": 9.209475386222852e-06,
      "loss": 0.7629,
      "step": 7150
    },
    {
      "epoch": 0.20604506425401947,
      "grad_norm": 0.9600811004638672,
      "learning_rate": 9.209223565821528e-06,
      "loss": 0.7805,
      "step": 7151
    },
    {
      "epoch": 0.20607387771566876,
      "grad_norm": 0.942451000213623,
      "learning_rate": 9.208971708762073e-06,
      "loss": 0.7729,
      "step": 7152
    },
    {
      "epoch": 0.20610269117731805,
      "grad_norm": 0.9887769818305969,
      "learning_rate": 9.208719815046683e-06,
      "loss": 0.7699,
      "step": 7153
    },
    {
      "epoch": 0.20613150463896732,
      "grad_norm": 0.954520583152771,
      "learning_rate": 9.208467884677553e-06,
      "loss": 0.7439,
      "step": 7154
    },
    {
      "epoch": 0.2061603181006166,
      "grad_norm": 0.9450900554656982,
      "learning_rate": 9.208215917656871e-06,
      "loss": 0.7563,
      "step": 7155
    },
    {
      "epoch": 0.2061891315622659,
      "grad_norm": 1.0827356576919556,
      "learning_rate": 9.207963913986837e-06,
      "loss": 0.7855,
      "step": 7156
    },
    {
      "epoch": 0.20621794502391516,
      "grad_norm": 1.0043110847473145,
      "learning_rate": 9.207711873669643e-06,
      "loss": 0.7805,
      "step": 7157
    },
    {
      "epoch": 0.20624675848556445,
      "grad_norm": 0.9963001608848572,
      "learning_rate": 9.207459796707485e-06,
      "loss": 0.7826,
      "step": 7158
    },
    {
      "epoch": 0.20627557194721374,
      "grad_norm": 1.0062379837036133,
      "learning_rate": 9.20720768310256e-06,
      "loss": 0.8087,
      "step": 7159
    },
    {
      "epoch": 0.20630438540886303,
      "grad_norm": 1.0174212455749512,
      "learning_rate": 9.206955532857059e-06,
      "loss": 0.8078,
      "step": 7160
    },
    {
      "epoch": 0.2063331988705123,
      "grad_norm": 0.9863080978393555,
      "learning_rate": 9.206703345973182e-06,
      "loss": 0.7847,
      "step": 7161
    },
    {
      "epoch": 0.2063620123321616,
      "grad_norm": 0.9679970145225525,
      "learning_rate": 9.206451122453122e-06,
      "loss": 0.778,
      "step": 7162
    },
    {
      "epoch": 0.20639082579381088,
      "grad_norm": 0.9896146655082703,
      "learning_rate": 9.206198862299081e-06,
      "loss": 0.8065,
      "step": 7163
    },
    {
      "epoch": 0.20641963925546014,
      "grad_norm": 0.9608865976333618,
      "learning_rate": 9.20594656551325e-06,
      "loss": 0.798,
      "step": 7164
    },
    {
      "epoch": 0.20644845271710943,
      "grad_norm": 0.9789994359016418,
      "learning_rate": 9.20569423209783e-06,
      "loss": 0.7676,
      "step": 7165
    },
    {
      "epoch": 0.20647726617875872,
      "grad_norm": 0.9680419564247131,
      "learning_rate": 9.205441862055016e-06,
      "loss": 0.7717,
      "step": 7166
    },
    {
      "epoch": 0.20650607964040799,
      "grad_norm": 0.94805508852005,
      "learning_rate": 9.20518945538701e-06,
      "loss": 0.7725,
      "step": 7167
    },
    {
      "epoch": 0.20653489310205728,
      "grad_norm": 0.989666223526001,
      "learning_rate": 9.204937012096005e-06,
      "loss": 0.7682,
      "step": 7168
    },
    {
      "epoch": 0.20656370656370657,
      "grad_norm": 0.9840209484100342,
      "learning_rate": 9.204684532184202e-06,
      "loss": 0.7736,
      "step": 7169
    },
    {
      "epoch": 0.20659252002535586,
      "grad_norm": 1.0119376182556152,
      "learning_rate": 9.204432015653801e-06,
      "loss": 0.7834,
      "step": 7170
    },
    {
      "epoch": 0.20662133348700512,
      "grad_norm": 0.9821696281433105,
      "learning_rate": 9.204179462507e-06,
      "loss": 0.7818,
      "step": 7171
    },
    {
      "epoch": 0.2066501469486544,
      "grad_norm": 0.9800329208374023,
      "learning_rate": 9.203926872745999e-06,
      "loss": 0.7771,
      "step": 7172
    },
    {
      "epoch": 0.2066789604103037,
      "grad_norm": 0.9450981616973877,
      "learning_rate": 9.203674246372996e-06,
      "loss": 0.7569,
      "step": 7173
    },
    {
      "epoch": 0.20670777387195297,
      "grad_norm": 0.9888426661491394,
      "learning_rate": 9.203421583390193e-06,
      "loss": 0.7699,
      "step": 7174
    },
    {
      "epoch": 0.20673658733360226,
      "grad_norm": 0.9946346879005432,
      "learning_rate": 9.203168883799789e-06,
      "loss": 0.7651,
      "step": 7175
    },
    {
      "epoch": 0.20676540079525155,
      "grad_norm": 0.9481627941131592,
      "learning_rate": 9.202916147603987e-06,
      "loss": 0.7907,
      "step": 7176
    },
    {
      "epoch": 0.20679421425690084,
      "grad_norm": 1.0031291246414185,
      "learning_rate": 9.202663374804986e-06,
      "loss": 0.7868,
      "step": 7177
    },
    {
      "epoch": 0.2068230277185501,
      "grad_norm": 0.9435850381851196,
      "learning_rate": 9.20241056540499e-06,
      "loss": 0.7775,
      "step": 7178
    },
    {
      "epoch": 0.2068518411801994,
      "grad_norm": 0.9657526016235352,
      "learning_rate": 9.202157719406197e-06,
      "loss": 0.7664,
      "step": 7179
    },
    {
      "epoch": 0.20688065464184868,
      "grad_norm": 0.9452857375144958,
      "learning_rate": 9.201904836810811e-06,
      "loss": 0.771,
      "step": 7180
    },
    {
      "epoch": 0.20690946810349795,
      "grad_norm": 0.9574840664863586,
      "learning_rate": 9.201651917621036e-06,
      "loss": 0.7698,
      "step": 7181
    },
    {
      "epoch": 0.20693828156514724,
      "grad_norm": 0.9638925194740295,
      "learning_rate": 9.20139896183907e-06,
      "loss": 0.7877,
      "step": 7182
    },
    {
      "epoch": 0.20696709502679653,
      "grad_norm": 0.9903046488761902,
      "learning_rate": 9.20114596946712e-06,
      "loss": 0.7856,
      "step": 7183
    },
    {
      "epoch": 0.2069959084884458,
      "grad_norm": 1.4650253057479858,
      "learning_rate": 9.200892940507388e-06,
      "loss": 0.751,
      "step": 7184
    },
    {
      "epoch": 0.20702472195009508,
      "grad_norm": 0.9466618895530701,
      "learning_rate": 9.200639874962078e-06,
      "loss": 0.8066,
      "step": 7185
    },
    {
      "epoch": 0.20705353541174437,
      "grad_norm": 0.9654495716094971,
      "learning_rate": 9.200386772833394e-06,
      "loss": 0.7997,
      "step": 7186
    },
    {
      "epoch": 0.20708234887339366,
      "grad_norm": 1.00584876537323,
      "learning_rate": 9.20013363412354e-06,
      "loss": 0.8134,
      "step": 7187
    },
    {
      "epoch": 0.20711116233504293,
      "grad_norm": 0.9319947957992554,
      "learning_rate": 9.199880458834722e-06,
      "loss": 0.7691,
      "step": 7188
    },
    {
      "epoch": 0.20713997579669222,
      "grad_norm": 0.9317789077758789,
      "learning_rate": 9.199627246969142e-06,
      "loss": 0.7642,
      "step": 7189
    },
    {
      "epoch": 0.2071687892583415,
      "grad_norm": 0.9083206653594971,
      "learning_rate": 9.199373998529006e-06,
      "loss": 0.7455,
      "step": 7190
    },
    {
      "epoch": 0.20719760271999077,
      "grad_norm": 0.9885913729667664,
      "learning_rate": 9.199120713516522e-06,
      "loss": 0.7893,
      "step": 7191
    },
    {
      "epoch": 0.20722641618164006,
      "grad_norm": 0.9550609588623047,
      "learning_rate": 9.198867391933894e-06,
      "loss": 0.791,
      "step": 7192
    },
    {
      "epoch": 0.20725522964328935,
      "grad_norm": 0.9219788908958435,
      "learning_rate": 9.198614033783328e-06,
      "loss": 0.7754,
      "step": 7193
    },
    {
      "epoch": 0.20728404310493861,
      "grad_norm": 0.9634948968887329,
      "learning_rate": 9.198360639067032e-06,
      "loss": 0.7683,
      "step": 7194
    },
    {
      "epoch": 0.2073128565665879,
      "grad_norm": 0.9487155079841614,
      "learning_rate": 9.19810720778721e-06,
      "loss": 0.7868,
      "step": 7195
    },
    {
      "epoch": 0.2073416700282372,
      "grad_norm": 0.9258156418800354,
      "learning_rate": 9.19785373994607e-06,
      "loss": 0.7781,
      "step": 7196
    },
    {
      "epoch": 0.2073704834898865,
      "grad_norm": 0.974228024482727,
      "learning_rate": 9.197600235545822e-06,
      "loss": 0.7881,
      "step": 7197
    },
    {
      "epoch": 0.20739929695153575,
      "grad_norm": 0.9478340744972229,
      "learning_rate": 9.197346694588672e-06,
      "loss": 0.7616,
      "step": 7198
    },
    {
      "epoch": 0.20742811041318504,
      "grad_norm": 0.9422029852867126,
      "learning_rate": 9.197093117076827e-06,
      "loss": 0.7663,
      "step": 7199
    },
    {
      "epoch": 0.20745692387483433,
      "grad_norm": 1.0009835958480835,
      "learning_rate": 9.196839503012498e-06,
      "loss": 0.7695,
      "step": 7200
    },
    {
      "epoch": 0.2074857373364836,
      "grad_norm": 0.9345175623893738,
      "learning_rate": 9.19658585239789e-06,
      "loss": 0.7673,
      "step": 7201
    },
    {
      "epoch": 0.20751455079813289,
      "grad_norm": 0.9513053894042969,
      "learning_rate": 9.196332165235216e-06,
      "loss": 0.7834,
      "step": 7202
    },
    {
      "epoch": 0.20754336425978218,
      "grad_norm": 0.9561120271682739,
      "learning_rate": 9.196078441526684e-06,
      "loss": 0.7713,
      "step": 7203
    },
    {
      "epoch": 0.20757217772143144,
      "grad_norm": 0.9511185884475708,
      "learning_rate": 9.195824681274502e-06,
      "loss": 0.7849,
      "step": 7204
    },
    {
      "epoch": 0.20760099118308073,
      "grad_norm": 0.9768335223197937,
      "learning_rate": 9.195570884480882e-06,
      "loss": 0.798,
      "step": 7205
    },
    {
      "epoch": 0.20762980464473002,
      "grad_norm": 0.9692129492759705,
      "learning_rate": 9.195317051148033e-06,
      "loss": 0.801,
      "step": 7206
    },
    {
      "epoch": 0.2076586181063793,
      "grad_norm": 1.1776189804077148,
      "learning_rate": 9.195063181278168e-06,
      "loss": 0.7904,
      "step": 7207
    },
    {
      "epoch": 0.20768743156802857,
      "grad_norm": 1.032802939414978,
      "learning_rate": 9.194809274873494e-06,
      "loss": 0.7897,
      "step": 7208
    },
    {
      "epoch": 0.20771624502967787,
      "grad_norm": 0.9827466607093811,
      "learning_rate": 9.194555331936225e-06,
      "loss": 0.7745,
      "step": 7209
    },
    {
      "epoch": 0.20774505849132716,
      "grad_norm": 1.0241830348968506,
      "learning_rate": 9.194301352468574e-06,
      "loss": 0.7709,
      "step": 7210
    },
    {
      "epoch": 0.20777387195297642,
      "grad_norm": 0.9998372793197632,
      "learning_rate": 9.19404733647275e-06,
      "loss": 0.7838,
      "step": 7211
    },
    {
      "epoch": 0.2078026854146257,
      "grad_norm": 1.046509861946106,
      "learning_rate": 9.193793283950964e-06,
      "loss": 0.7812,
      "step": 7212
    },
    {
      "epoch": 0.207831498876275,
      "grad_norm": 1.0134987831115723,
      "learning_rate": 9.193539194905433e-06,
      "loss": 0.7756,
      "step": 7213
    },
    {
      "epoch": 0.2078603123379243,
      "grad_norm": 0.9497000575065613,
      "learning_rate": 9.193285069338367e-06,
      "loss": 0.7434,
      "step": 7214
    },
    {
      "epoch": 0.20788912579957355,
      "grad_norm": 1.0524460077285767,
      "learning_rate": 9.19303090725198e-06,
      "loss": 0.775,
      "step": 7215
    },
    {
      "epoch": 0.20791793926122284,
      "grad_norm": 0.9641541838645935,
      "learning_rate": 9.192776708648486e-06,
      "loss": 0.7599,
      "step": 7216
    },
    {
      "epoch": 0.20794675272287214,
      "grad_norm": 0.9789540767669678,
      "learning_rate": 9.192522473530097e-06,
      "loss": 0.7894,
      "step": 7217
    },
    {
      "epoch": 0.2079755661845214,
      "grad_norm": 0.9721737504005432,
      "learning_rate": 9.192268201899028e-06,
      "loss": 0.7777,
      "step": 7218
    },
    {
      "epoch": 0.2080043796461707,
      "grad_norm": 0.9638831615447998,
      "learning_rate": 9.192013893757494e-06,
      "loss": 0.7619,
      "step": 7219
    },
    {
      "epoch": 0.20803319310781998,
      "grad_norm": 1.0054199695587158,
      "learning_rate": 9.19175954910771e-06,
      "loss": 0.8017,
      "step": 7220
    },
    {
      "epoch": 0.20806200656946924,
      "grad_norm": 0.9985860586166382,
      "learning_rate": 9.19150516795189e-06,
      "loss": 0.7744,
      "step": 7221
    },
    {
      "epoch": 0.20809082003111853,
      "grad_norm": 0.9654388427734375,
      "learning_rate": 9.191250750292252e-06,
      "loss": 0.7933,
      "step": 7222
    },
    {
      "epoch": 0.20811963349276782,
      "grad_norm": 0.9631644487380981,
      "learning_rate": 9.190996296131007e-06,
      "loss": 0.761,
      "step": 7223
    },
    {
      "epoch": 0.20814844695441712,
      "grad_norm": 0.9431234002113342,
      "learning_rate": 9.190741805470374e-06,
      "loss": 0.7803,
      "step": 7224
    },
    {
      "epoch": 0.20817726041606638,
      "grad_norm": 0.9858876466751099,
      "learning_rate": 9.19048727831257e-06,
      "loss": 0.7838,
      "step": 7225
    },
    {
      "epoch": 0.20820607387771567,
      "grad_norm": 1.0045394897460938,
      "learning_rate": 9.19023271465981e-06,
      "loss": 0.805,
      "step": 7226
    },
    {
      "epoch": 0.20823488733936496,
      "grad_norm": 1.0014944076538086,
      "learning_rate": 9.189978114514313e-06,
      "loss": 0.8045,
      "step": 7227
    },
    {
      "epoch": 0.20826370080101422,
      "grad_norm": 0.9981149435043335,
      "learning_rate": 9.189723477878293e-06,
      "loss": 0.771,
      "step": 7228
    },
    {
      "epoch": 0.20829251426266351,
      "grad_norm": 0.9701336622238159,
      "learning_rate": 9.189468804753973e-06,
      "loss": 0.7525,
      "step": 7229
    },
    {
      "epoch": 0.2083213277243128,
      "grad_norm": 0.9495701193809509,
      "learning_rate": 9.189214095143565e-06,
      "loss": 0.7733,
      "step": 7230
    },
    {
      "epoch": 0.20835014118596207,
      "grad_norm": 0.9877557158470154,
      "learning_rate": 9.188959349049292e-06,
      "loss": 0.7947,
      "step": 7231
    },
    {
      "epoch": 0.20837895464761136,
      "grad_norm": 1.029370665550232,
      "learning_rate": 9.188704566473368e-06,
      "loss": 0.8085,
      "step": 7232
    },
    {
      "epoch": 0.20840776810926065,
      "grad_norm": 0.9839236736297607,
      "learning_rate": 9.188449747418016e-06,
      "loss": 0.7663,
      "step": 7233
    },
    {
      "epoch": 0.20843658157090994,
      "grad_norm": 0.9589337110519409,
      "learning_rate": 9.188194891885452e-06,
      "loss": 0.7833,
      "step": 7234
    },
    {
      "epoch": 0.2084653950325592,
      "grad_norm": 0.9622363448143005,
      "learning_rate": 9.1879399998779e-06,
      "loss": 0.7618,
      "step": 7235
    },
    {
      "epoch": 0.2084942084942085,
      "grad_norm": 0.9145703315734863,
      "learning_rate": 9.187685071397573e-06,
      "loss": 0.7625,
      "step": 7236
    },
    {
      "epoch": 0.20852302195585778,
      "grad_norm": 0.9503191113471985,
      "learning_rate": 9.187430106446698e-06,
      "loss": 0.7761,
      "step": 7237
    },
    {
      "epoch": 0.20855183541750705,
      "grad_norm": 1.0166932344436646,
      "learning_rate": 9.187175105027492e-06,
      "loss": 0.7753,
      "step": 7238
    },
    {
      "epoch": 0.20858064887915634,
      "grad_norm": 0.9811162948608398,
      "learning_rate": 9.186920067142177e-06,
      "loss": 0.7733,
      "step": 7239
    },
    {
      "epoch": 0.20860946234080563,
      "grad_norm": 0.9501125812530518,
      "learning_rate": 9.186664992792973e-06,
      "loss": 0.7727,
      "step": 7240
    },
    {
      "epoch": 0.20863827580245492,
      "grad_norm": 0.9938365817070007,
      "learning_rate": 9.186409881982103e-06,
      "loss": 0.7857,
      "step": 7241
    },
    {
      "epoch": 0.20866708926410418,
      "grad_norm": 1.0185335874557495,
      "learning_rate": 9.186154734711787e-06,
      "loss": 0.8083,
      "step": 7242
    },
    {
      "epoch": 0.20869590272575347,
      "grad_norm": 0.9744595885276794,
      "learning_rate": 9.185899550984248e-06,
      "loss": 0.7991,
      "step": 7243
    },
    {
      "epoch": 0.20872471618740276,
      "grad_norm": 0.9627944827079773,
      "learning_rate": 9.185644330801708e-06,
      "loss": 0.7353,
      "step": 7244
    },
    {
      "epoch": 0.20875352964905203,
      "grad_norm": 1.005194902420044,
      "learning_rate": 9.18538907416639e-06,
      "loss": 0.7687,
      "step": 7245
    },
    {
      "epoch": 0.20878234311070132,
      "grad_norm": 1.0074998140335083,
      "learning_rate": 9.185133781080517e-06,
      "loss": 0.7798,
      "step": 7246
    },
    {
      "epoch": 0.2088111565723506,
      "grad_norm": 1.0327184200286865,
      "learning_rate": 9.184878451546312e-06,
      "loss": 0.8125,
      "step": 7247
    },
    {
      "epoch": 0.20883997003399987,
      "grad_norm": 0.9832820892333984,
      "learning_rate": 9.184623085566001e-06,
      "loss": 0.7815,
      "step": 7248
    },
    {
      "epoch": 0.20886878349564916,
      "grad_norm": 1.0694166421890259,
      "learning_rate": 9.184367683141804e-06,
      "loss": 0.783,
      "step": 7249
    },
    {
      "epoch": 0.20889759695729845,
      "grad_norm": 0.9583245515823364,
      "learning_rate": 9.184112244275949e-06,
      "loss": 0.7803,
      "step": 7250
    },
    {
      "epoch": 0.20892641041894774,
      "grad_norm": 0.974116861820221,
      "learning_rate": 9.183856768970657e-06,
      "loss": 0.7665,
      "step": 7251
    },
    {
      "epoch": 0.208955223880597,
      "grad_norm": 0.9699560403823853,
      "learning_rate": 9.183601257228157e-06,
      "loss": 0.7666,
      "step": 7252
    },
    {
      "epoch": 0.2089840373422463,
      "grad_norm": 0.9521128535270691,
      "learning_rate": 9.183345709050669e-06,
      "loss": 0.7948,
      "step": 7253
    },
    {
      "epoch": 0.2090128508038956,
      "grad_norm": 0.9636477828025818,
      "learning_rate": 9.183090124440424e-06,
      "loss": 0.7383,
      "step": 7254
    },
    {
      "epoch": 0.20904166426554485,
      "grad_norm": 0.9949929118156433,
      "learning_rate": 9.182834503399645e-06,
      "loss": 0.7799,
      "step": 7255
    },
    {
      "epoch": 0.20907047772719414,
      "grad_norm": 1.0191410779953003,
      "learning_rate": 9.182578845930558e-06,
      "loss": 0.7826,
      "step": 7256
    },
    {
      "epoch": 0.20909929118884343,
      "grad_norm": 0.9905584454536438,
      "learning_rate": 9.182323152035394e-06,
      "loss": 0.773,
      "step": 7257
    },
    {
      "epoch": 0.2091281046504927,
      "grad_norm": 0.9374607801437378,
      "learning_rate": 9.182067421716372e-06,
      "loss": 0.7619,
      "step": 7258
    },
    {
      "epoch": 0.209156918112142,
      "grad_norm": 0.9318328499794006,
      "learning_rate": 9.181811654975725e-06,
      "loss": 0.7811,
      "step": 7259
    },
    {
      "epoch": 0.20918573157379128,
      "grad_norm": 0.9862211346626282,
      "learning_rate": 9.181555851815677e-06,
      "loss": 0.7774,
      "step": 7260
    },
    {
      "epoch": 0.20921454503544057,
      "grad_norm": 1.0133540630340576,
      "learning_rate": 9.18130001223846e-06,
      "loss": 0.8159,
      "step": 7261
    },
    {
      "epoch": 0.20924335849708983,
      "grad_norm": 0.9783549308776855,
      "learning_rate": 9.1810441362463e-06,
      "loss": 0.7792,
      "step": 7262
    },
    {
      "epoch": 0.20927217195873912,
      "grad_norm": 0.9768669009208679,
      "learning_rate": 9.180788223841422e-06,
      "loss": 0.7832,
      "step": 7263
    },
    {
      "epoch": 0.2093009854203884,
      "grad_norm": 1.0078577995300293,
      "learning_rate": 9.180532275026059e-06,
      "loss": 0.8224,
      "step": 7264
    },
    {
      "epoch": 0.20932979888203768,
      "grad_norm": 0.9874541163444519,
      "learning_rate": 9.180276289802442e-06,
      "loss": 0.782,
      "step": 7265
    },
    {
      "epoch": 0.20935861234368697,
      "grad_norm": 0.9635449051856995,
      "learning_rate": 9.180020268172793e-06,
      "loss": 0.7736,
      "step": 7266
    },
    {
      "epoch": 0.20938742580533626,
      "grad_norm": 0.974700391292572,
      "learning_rate": 9.179764210139348e-06,
      "loss": 0.7681,
      "step": 7267
    },
    {
      "epoch": 0.20941623926698555,
      "grad_norm": 0.9916765093803406,
      "learning_rate": 9.179508115704334e-06,
      "loss": 0.7681,
      "step": 7268
    },
    {
      "epoch": 0.2094450527286348,
      "grad_norm": 0.9582535624504089,
      "learning_rate": 9.179251984869983e-06,
      "loss": 0.7836,
      "step": 7269
    },
    {
      "epoch": 0.2094738661902841,
      "grad_norm": 0.9914268851280212,
      "learning_rate": 9.178995817638525e-06,
      "loss": 0.8116,
      "step": 7270
    },
    {
      "epoch": 0.2095026796519334,
      "grad_norm": 0.9328097105026245,
      "learning_rate": 9.17873961401219e-06,
      "loss": 0.7429,
      "step": 7271
    },
    {
      "epoch": 0.20953149311358266,
      "grad_norm": 1.0208995342254639,
      "learning_rate": 9.178483373993212e-06,
      "loss": 0.7544,
      "step": 7272
    },
    {
      "epoch": 0.20956030657523195,
      "grad_norm": 0.9741523861885071,
      "learning_rate": 9.17822709758382e-06,
      "loss": 0.7726,
      "step": 7273
    },
    {
      "epoch": 0.20958912003688124,
      "grad_norm": 0.9763630628585815,
      "learning_rate": 9.177970784786245e-06,
      "loss": 0.7607,
      "step": 7274
    },
    {
      "epoch": 0.2096179334985305,
      "grad_norm": 0.9358840584754944,
      "learning_rate": 9.177714435602723e-06,
      "loss": 0.7809,
      "step": 7275
    },
    {
      "epoch": 0.2096467469601798,
      "grad_norm": 0.9956486225128174,
      "learning_rate": 9.177458050035483e-06,
      "loss": 0.8153,
      "step": 7276
    },
    {
      "epoch": 0.20967556042182908,
      "grad_norm": 0.9749994874000549,
      "learning_rate": 9.17720162808676e-06,
      "loss": 0.7833,
      "step": 7277
    },
    {
      "epoch": 0.20970437388347837,
      "grad_norm": 1.0055124759674072,
      "learning_rate": 9.176945169758787e-06,
      "loss": 0.7813,
      "step": 7278
    },
    {
      "epoch": 0.20973318734512764,
      "grad_norm": 0.9436851143836975,
      "learning_rate": 9.176688675053797e-06,
      "loss": 0.7551,
      "step": 7279
    },
    {
      "epoch": 0.20976200080677693,
      "grad_norm": 0.9761061072349548,
      "learning_rate": 9.176432143974024e-06,
      "loss": 0.7894,
      "step": 7280
    },
    {
      "epoch": 0.20979081426842622,
      "grad_norm": 0.983971118927002,
      "learning_rate": 9.1761755765217e-06,
      "loss": 0.7544,
      "step": 7281
    },
    {
      "epoch": 0.20981962773007548,
      "grad_norm": 0.9688578844070435,
      "learning_rate": 9.175918972699063e-06,
      "loss": 0.793,
      "step": 7282
    },
    {
      "epoch": 0.20984844119172477,
      "grad_norm": 0.9934625625610352,
      "learning_rate": 9.175662332508345e-06,
      "loss": 0.8045,
      "step": 7283
    },
    {
      "epoch": 0.20987725465337406,
      "grad_norm": 0.9913925528526306,
      "learning_rate": 9.175405655951783e-06,
      "loss": 0.792,
      "step": 7284
    },
    {
      "epoch": 0.20990606811502333,
      "grad_norm": 0.956840991973877,
      "learning_rate": 9.175148943031614e-06,
      "loss": 0.756,
      "step": 7285
    },
    {
      "epoch": 0.20993488157667262,
      "grad_norm": 0.9707804322242737,
      "learning_rate": 9.174892193750068e-06,
      "loss": 0.7826,
      "step": 7286
    },
    {
      "epoch": 0.2099636950383219,
      "grad_norm": 0.9920419454574585,
      "learning_rate": 9.174635408109384e-06,
      "loss": 0.7608,
      "step": 7287
    },
    {
      "epoch": 0.2099925084999712,
      "grad_norm": 1.0201640129089355,
      "learning_rate": 9.174378586111801e-06,
      "loss": 0.7783,
      "step": 7288
    },
    {
      "epoch": 0.21002132196162046,
      "grad_norm": 0.9553521275520325,
      "learning_rate": 9.174121727759553e-06,
      "loss": 0.8077,
      "step": 7289
    },
    {
      "epoch": 0.21005013542326975,
      "grad_norm": 0.9648612141609192,
      "learning_rate": 9.173864833054874e-06,
      "loss": 0.7714,
      "step": 7290
    },
    {
      "epoch": 0.21007894888491904,
      "grad_norm": 1.0140244960784912,
      "learning_rate": 9.173607902000009e-06,
      "loss": 0.7751,
      "step": 7291
    },
    {
      "epoch": 0.2101077623465683,
      "grad_norm": 0.938203752040863,
      "learning_rate": 9.17335093459719e-06,
      "loss": 0.7614,
      "step": 7292
    },
    {
      "epoch": 0.2101365758082176,
      "grad_norm": 1.0511358976364136,
      "learning_rate": 9.173093930848656e-06,
      "loss": 0.7844,
      "step": 7293
    },
    {
      "epoch": 0.2101653892698669,
      "grad_norm": 1.7950795888900757,
      "learning_rate": 9.172836890756644e-06,
      "loss": 0.7803,
      "step": 7294
    },
    {
      "epoch": 0.21019420273151618,
      "grad_norm": 1.0147473812103271,
      "learning_rate": 9.172579814323396e-06,
      "loss": 0.8,
      "step": 7295
    },
    {
      "epoch": 0.21022301619316544,
      "grad_norm": 0.9551165103912354,
      "learning_rate": 9.172322701551148e-06,
      "loss": 0.7776,
      "step": 7296
    },
    {
      "epoch": 0.21025182965481473,
      "grad_norm": 1.083709955215454,
      "learning_rate": 9.17206555244214e-06,
      "loss": 0.7842,
      "step": 7297
    },
    {
      "epoch": 0.21028064311646402,
      "grad_norm": 0.9846024513244629,
      "learning_rate": 9.171808366998611e-06,
      "loss": 0.7487,
      "step": 7298
    },
    {
      "epoch": 0.21030945657811329,
      "grad_norm": 1.012995719909668,
      "learning_rate": 9.171551145222804e-06,
      "loss": 0.7728,
      "step": 7299
    },
    {
      "epoch": 0.21033827003976258,
      "grad_norm": 1.0471036434173584,
      "learning_rate": 9.171293887116954e-06,
      "loss": 0.7925,
      "step": 7300
    },
    {
      "epoch": 0.21036708350141187,
      "grad_norm": 0.9438174962997437,
      "learning_rate": 9.171036592683305e-06,
      "loss": 0.7666,
      "step": 7301
    },
    {
      "epoch": 0.21039589696306113,
      "grad_norm": 1.0989645719528198,
      "learning_rate": 9.170779261924096e-06,
      "loss": 0.7876,
      "step": 7302
    },
    {
      "epoch": 0.21042471042471042,
      "grad_norm": 1.0406285524368286,
      "learning_rate": 9.17052189484157e-06,
      "loss": 0.8173,
      "step": 7303
    },
    {
      "epoch": 0.2104535238863597,
      "grad_norm": 1.0141689777374268,
      "learning_rate": 9.170264491437966e-06,
      "loss": 0.7808,
      "step": 7304
    },
    {
      "epoch": 0.210482337348009,
      "grad_norm": 1.0230281352996826,
      "learning_rate": 9.170007051715528e-06,
      "loss": 0.7732,
      "step": 7305
    },
    {
      "epoch": 0.21051115080965827,
      "grad_norm": 0.9843199253082275,
      "learning_rate": 9.169749575676497e-06,
      "loss": 0.7775,
      "step": 7306
    },
    {
      "epoch": 0.21053996427130756,
      "grad_norm": 0.9455735683441162,
      "learning_rate": 9.169492063323115e-06,
      "loss": 0.7427,
      "step": 7307
    },
    {
      "epoch": 0.21056877773295685,
      "grad_norm": 1.0079219341278076,
      "learning_rate": 9.169234514657625e-06,
      "loss": 0.7754,
      "step": 7308
    },
    {
      "epoch": 0.2105975911946061,
      "grad_norm": 0.9512419700622559,
      "learning_rate": 9.16897692968227e-06,
      "loss": 0.7578,
      "step": 7309
    },
    {
      "epoch": 0.2106264046562554,
      "grad_norm": 1.011177659034729,
      "learning_rate": 9.168719308399295e-06,
      "loss": 0.7923,
      "step": 7310
    },
    {
      "epoch": 0.2106552181179047,
      "grad_norm": 0.947700560092926,
      "learning_rate": 9.16846165081094e-06,
      "loss": 0.7802,
      "step": 7311
    },
    {
      "epoch": 0.21068403157955395,
      "grad_norm": 0.945365309715271,
      "learning_rate": 9.168203956919451e-06,
      "loss": 0.778,
      "step": 7312
    },
    {
      "epoch": 0.21071284504120324,
      "grad_norm": 0.9479764699935913,
      "learning_rate": 9.167946226727074e-06,
      "loss": 0.7933,
      "step": 7313
    },
    {
      "epoch": 0.21074165850285254,
      "grad_norm": 0.9334822297096252,
      "learning_rate": 9.167688460236049e-06,
      "loss": 0.7827,
      "step": 7314
    },
    {
      "epoch": 0.21077047196450183,
      "grad_norm": 0.9931308627128601,
      "learning_rate": 9.167430657448625e-06,
      "loss": 0.7824,
      "step": 7315
    },
    {
      "epoch": 0.2107992854261511,
      "grad_norm": 0.9560772180557251,
      "learning_rate": 9.167172818367048e-06,
      "loss": 0.7822,
      "step": 7316
    },
    {
      "epoch": 0.21082809888780038,
      "grad_norm": 0.9360358715057373,
      "learning_rate": 9.166914942993559e-06,
      "loss": 0.7589,
      "step": 7317
    },
    {
      "epoch": 0.21085691234944967,
      "grad_norm": 0.932789146900177,
      "learning_rate": 9.166657031330408e-06,
      "loss": 0.7701,
      "step": 7318
    },
    {
      "epoch": 0.21088572581109893,
      "grad_norm": 0.9515885710716248,
      "learning_rate": 9.166399083379838e-06,
      "loss": 0.7795,
      "step": 7319
    },
    {
      "epoch": 0.21091453927274822,
      "grad_norm": 0.9384995698928833,
      "learning_rate": 9.166141099144098e-06,
      "loss": 0.7738,
      "step": 7320
    },
    {
      "epoch": 0.21094335273439752,
      "grad_norm": 0.9477676749229431,
      "learning_rate": 9.16588307862543e-06,
      "loss": 0.7694,
      "step": 7321
    },
    {
      "epoch": 0.2109721661960468,
      "grad_norm": 0.975013017654419,
      "learning_rate": 9.165625021826088e-06,
      "loss": 0.7806,
      "step": 7322
    },
    {
      "epoch": 0.21100097965769607,
      "grad_norm": 1.006722092628479,
      "learning_rate": 9.165366928748315e-06,
      "loss": 0.7781,
      "step": 7323
    },
    {
      "epoch": 0.21102979311934536,
      "grad_norm": 1.0244115591049194,
      "learning_rate": 9.165108799394362e-06,
      "loss": 0.7723,
      "step": 7324
    },
    {
      "epoch": 0.21105860658099465,
      "grad_norm": 0.9347310662269592,
      "learning_rate": 9.164850633766473e-06,
      "loss": 0.7692,
      "step": 7325
    },
    {
      "epoch": 0.21108742004264391,
      "grad_norm": 0.9872741103172302,
      "learning_rate": 9.164592431866899e-06,
      "loss": 0.7675,
      "step": 7326
    },
    {
      "epoch": 0.2111162335042932,
      "grad_norm": 0.9604657888412476,
      "learning_rate": 9.164334193697888e-06,
      "loss": 0.7783,
      "step": 7327
    },
    {
      "epoch": 0.2111450469659425,
      "grad_norm": 0.9433544278144836,
      "learning_rate": 9.164075919261689e-06,
      "loss": 0.7683,
      "step": 7328
    },
    {
      "epoch": 0.21117386042759176,
      "grad_norm": 0.9723798632621765,
      "learning_rate": 9.16381760856055e-06,
      "loss": 0.7954,
      "step": 7329
    },
    {
      "epoch": 0.21120267388924105,
      "grad_norm": 0.9012947678565979,
      "learning_rate": 9.163559261596725e-06,
      "loss": 0.7741,
      "step": 7330
    },
    {
      "epoch": 0.21123148735089034,
      "grad_norm": 0.9182496070861816,
      "learning_rate": 9.163300878372459e-06,
      "loss": 0.7731,
      "step": 7331
    },
    {
      "epoch": 0.21126030081253963,
      "grad_norm": 0.9642087817192078,
      "learning_rate": 9.163042458890003e-06,
      "loss": 0.7681,
      "step": 7332
    },
    {
      "epoch": 0.2112891142741889,
      "grad_norm": 0.9661329388618469,
      "learning_rate": 9.16278400315161e-06,
      "loss": 0.7759,
      "step": 7333
    },
    {
      "epoch": 0.21131792773583818,
      "grad_norm": 0.9398237466812134,
      "learning_rate": 9.16252551115953e-06,
      "loss": 0.7925,
      "step": 7334
    },
    {
      "epoch": 0.21134674119748748,
      "grad_norm": 0.9352608919143677,
      "learning_rate": 9.162266982916011e-06,
      "loss": 0.7705,
      "step": 7335
    },
    {
      "epoch": 0.21137555465913674,
      "grad_norm": 0.9402210712432861,
      "learning_rate": 9.16200841842331e-06,
      "loss": 0.7612,
      "step": 7336
    },
    {
      "epoch": 0.21140436812078603,
      "grad_norm": 1.0022306442260742,
      "learning_rate": 9.161749817683676e-06,
      "loss": 0.7783,
      "step": 7337
    },
    {
      "epoch": 0.21143318158243532,
      "grad_norm": 0.9318708181381226,
      "learning_rate": 9.16149118069936e-06,
      "loss": 0.7528,
      "step": 7338
    },
    {
      "epoch": 0.21146199504408458,
      "grad_norm": 0.9853746891021729,
      "learning_rate": 9.161232507472616e-06,
      "loss": 0.7529,
      "step": 7339
    },
    {
      "epoch": 0.21149080850573387,
      "grad_norm": 0.967944324016571,
      "learning_rate": 9.1609737980057e-06,
      "loss": 0.7843,
      "step": 7340
    },
    {
      "epoch": 0.21151962196738316,
      "grad_norm": 1.0168482065200806,
      "learning_rate": 9.160715052300857e-06,
      "loss": 0.7632,
      "step": 7341
    },
    {
      "epoch": 0.21154843542903246,
      "grad_norm": 0.9818502068519592,
      "learning_rate": 9.160456270360348e-06,
      "loss": 0.7789,
      "step": 7342
    },
    {
      "epoch": 0.21157724889068172,
      "grad_norm": 0.9589918255805969,
      "learning_rate": 9.160197452186424e-06,
      "loss": 0.7891,
      "step": 7343
    },
    {
      "epoch": 0.211606062352331,
      "grad_norm": 1.0223585367202759,
      "learning_rate": 9.159938597781337e-06,
      "loss": 0.7748,
      "step": 7344
    },
    {
      "epoch": 0.2116348758139803,
      "grad_norm": 1.0422860383987427,
      "learning_rate": 9.159679707147344e-06,
      "loss": 0.7686,
      "step": 7345
    },
    {
      "epoch": 0.21166368927562956,
      "grad_norm": 0.9901424646377563,
      "learning_rate": 9.159420780286699e-06,
      "loss": 0.7629,
      "step": 7346
    },
    {
      "epoch": 0.21169250273727885,
      "grad_norm": 1.0104924440383911,
      "learning_rate": 9.159161817201657e-06,
      "loss": 0.7521,
      "step": 7347
    },
    {
      "epoch": 0.21172131619892814,
      "grad_norm": 0.9576760530471802,
      "learning_rate": 9.158902817894472e-06,
      "loss": 0.7977,
      "step": 7348
    },
    {
      "epoch": 0.21175012966057744,
      "grad_norm": 0.995055079460144,
      "learning_rate": 9.158643782367401e-06,
      "loss": 0.7682,
      "step": 7349
    },
    {
      "epoch": 0.2117789431222267,
      "grad_norm": 0.991173267364502,
      "learning_rate": 9.1583847106227e-06,
      "loss": 0.7762,
      "step": 7350
    },
    {
      "epoch": 0.211807756583876,
      "grad_norm": 0.9131861925125122,
      "learning_rate": 9.158125602662627e-06,
      "loss": 0.7906,
      "step": 7351
    },
    {
      "epoch": 0.21183657004552528,
      "grad_norm": 0.9412030577659607,
      "learning_rate": 9.157866458489435e-06,
      "loss": 0.7842,
      "step": 7352
    },
    {
      "epoch": 0.21186538350717454,
      "grad_norm": 1.006188154220581,
      "learning_rate": 9.157607278105383e-06,
      "loss": 0.7555,
      "step": 7353
    },
    {
      "epoch": 0.21189419696882383,
      "grad_norm": 0.9482517838478088,
      "learning_rate": 9.157348061512728e-06,
      "loss": 0.7894,
      "step": 7354
    },
    {
      "epoch": 0.21192301043047312,
      "grad_norm": 0.9425077438354492,
      "learning_rate": 9.157088808713725e-06,
      "loss": 0.7703,
      "step": 7355
    },
    {
      "epoch": 0.2119518238921224,
      "grad_norm": 0.9766605496406555,
      "learning_rate": 9.156829519710635e-06,
      "loss": 0.765,
      "step": 7356
    },
    {
      "epoch": 0.21198063735377168,
      "grad_norm": 0.9816906452178955,
      "learning_rate": 9.156570194505717e-06,
      "loss": 0.7925,
      "step": 7357
    },
    {
      "epoch": 0.21200945081542097,
      "grad_norm": 0.8915678262710571,
      "learning_rate": 9.156310833101227e-06,
      "loss": 0.7509,
      "step": 7358
    },
    {
      "epoch": 0.21203826427707026,
      "grad_norm": 0.9667198061943054,
      "learning_rate": 9.156051435499423e-06,
      "loss": 0.7782,
      "step": 7359
    },
    {
      "epoch": 0.21206707773871952,
      "grad_norm": 0.9697993397712708,
      "learning_rate": 9.155792001702566e-06,
      "loss": 0.7873,
      "step": 7360
    },
    {
      "epoch": 0.2120958912003688,
      "grad_norm": 0.9685846567153931,
      "learning_rate": 9.155532531712917e-06,
      "loss": 0.7755,
      "step": 7361
    },
    {
      "epoch": 0.2121247046620181,
      "grad_norm": 0.9625496864318848,
      "learning_rate": 9.155273025532732e-06,
      "loss": 0.7522,
      "step": 7362
    },
    {
      "epoch": 0.21215351812366737,
      "grad_norm": 0.9704114198684692,
      "learning_rate": 9.155013483164273e-06,
      "loss": 0.7466,
      "step": 7363
    },
    {
      "epoch": 0.21218233158531666,
      "grad_norm": 0.9893775582313538,
      "learning_rate": 9.154753904609799e-06,
      "loss": 0.784,
      "step": 7364
    },
    {
      "epoch": 0.21221114504696595,
      "grad_norm": 1.0069397687911987,
      "learning_rate": 9.154494289871572e-06,
      "loss": 0.7749,
      "step": 7365
    },
    {
      "epoch": 0.2122399585086152,
      "grad_norm": 0.9762619733810425,
      "learning_rate": 9.154234638951852e-06,
      "loss": 0.7905,
      "step": 7366
    },
    {
      "epoch": 0.2122687719702645,
      "grad_norm": 0.9850831031799316,
      "learning_rate": 9.153974951852905e-06,
      "loss": 0.7907,
      "step": 7367
    },
    {
      "epoch": 0.2122975854319138,
      "grad_norm": 1.0080920457839966,
      "learning_rate": 9.153715228576985e-06,
      "loss": 0.7649,
      "step": 7368
    },
    {
      "epoch": 0.21232639889356308,
      "grad_norm": 1.0174250602722168,
      "learning_rate": 9.153455469126359e-06,
      "loss": 0.7787,
      "step": 7369
    },
    {
      "epoch": 0.21235521235521235,
      "grad_norm": 1.0159811973571777,
      "learning_rate": 9.153195673503286e-06,
      "loss": 0.7843,
      "step": 7370
    },
    {
      "epoch": 0.21238402581686164,
      "grad_norm": 1.0453698635101318,
      "learning_rate": 9.15293584171003e-06,
      "loss": 0.794,
      "step": 7371
    },
    {
      "epoch": 0.21241283927851093,
      "grad_norm": 0.9761572480201721,
      "learning_rate": 9.152675973748857e-06,
      "loss": 0.7942,
      "step": 7372
    },
    {
      "epoch": 0.2124416527401602,
      "grad_norm": 0.9678346514701843,
      "learning_rate": 9.152416069622028e-06,
      "loss": 0.7721,
      "step": 7373
    },
    {
      "epoch": 0.21247046620180948,
      "grad_norm": 1.0040833950042725,
      "learning_rate": 9.152156129331804e-06,
      "loss": 0.7406,
      "step": 7374
    },
    {
      "epoch": 0.21249927966345877,
      "grad_norm": 0.9991849064826965,
      "learning_rate": 9.15189615288045e-06,
      "loss": 0.7832,
      "step": 7375
    },
    {
      "epoch": 0.21252809312510806,
      "grad_norm": 0.9195847511291504,
      "learning_rate": 9.151636140270232e-06,
      "loss": 0.7748,
      "step": 7376
    },
    {
      "epoch": 0.21255690658675733,
      "grad_norm": 0.9506336450576782,
      "learning_rate": 9.151376091503414e-06,
      "loss": 0.7714,
      "step": 7377
    },
    {
      "epoch": 0.21258572004840662,
      "grad_norm": 0.9864799976348877,
      "learning_rate": 9.151116006582259e-06,
      "loss": 0.7486,
      "step": 7378
    },
    {
      "epoch": 0.2126145335100559,
      "grad_norm": 0.9960100650787354,
      "learning_rate": 9.150855885509034e-06,
      "loss": 0.7746,
      "step": 7379
    },
    {
      "epoch": 0.21264334697170517,
      "grad_norm": 0.9670732617378235,
      "learning_rate": 9.150595728286002e-06,
      "loss": 0.7728,
      "step": 7380
    },
    {
      "epoch": 0.21267216043335446,
      "grad_norm": 0.9847320318222046,
      "learning_rate": 9.150335534915432e-06,
      "loss": 0.7687,
      "step": 7381
    },
    {
      "epoch": 0.21270097389500375,
      "grad_norm": 0.9515901803970337,
      "learning_rate": 9.150075305399586e-06,
      "loss": 0.764,
      "step": 7382
    },
    {
      "epoch": 0.21272978735665302,
      "grad_norm": 1.0442315340042114,
      "learning_rate": 9.149815039740735e-06,
      "loss": 0.7442,
      "step": 7383
    },
    {
      "epoch": 0.2127586008183023,
      "grad_norm": 1.0006803274154663,
      "learning_rate": 9.149554737941142e-06,
      "loss": 0.7729,
      "step": 7384
    },
    {
      "epoch": 0.2127874142799516,
      "grad_norm": 0.9284325242042542,
      "learning_rate": 9.149294400003077e-06,
      "loss": 0.7659,
      "step": 7385
    },
    {
      "epoch": 0.2128162277416009,
      "grad_norm": 0.9375273585319519,
      "learning_rate": 9.149034025928803e-06,
      "loss": 0.7952,
      "step": 7386
    },
    {
      "epoch": 0.21284504120325015,
      "grad_norm": 0.9488041996955872,
      "learning_rate": 9.148773615720591e-06,
      "loss": 0.765,
      "step": 7387
    },
    {
      "epoch": 0.21287385466489944,
      "grad_norm": 0.9724968671798706,
      "learning_rate": 9.14851316938071e-06,
      "loss": 0.7712,
      "step": 7388
    },
    {
      "epoch": 0.21290266812654873,
      "grad_norm": 0.925595223903656,
      "learning_rate": 9.148252686911425e-06,
      "loss": 0.771,
      "step": 7389
    },
    {
      "epoch": 0.212931481588198,
      "grad_norm": 0.9601022005081177,
      "learning_rate": 9.147992168315006e-06,
      "loss": 0.7986,
      "step": 7390
    },
    {
      "epoch": 0.2129602950498473,
      "grad_norm": 0.9559935927391052,
      "learning_rate": 9.147731613593722e-06,
      "loss": 0.7647,
      "step": 7391
    },
    {
      "epoch": 0.21298910851149658,
      "grad_norm": 0.9558907151222229,
      "learning_rate": 9.14747102274984e-06,
      "loss": 0.7824,
      "step": 7392
    },
    {
      "epoch": 0.21301792197314584,
      "grad_norm": 0.9785637259483337,
      "learning_rate": 9.147210395785633e-06,
      "loss": 0.7757,
      "step": 7393
    },
    {
      "epoch": 0.21304673543479513,
      "grad_norm": 3.9612433910369873,
      "learning_rate": 9.146949732703369e-06,
      "loss": 0.7598,
      "step": 7394
    },
    {
      "epoch": 0.21307554889644442,
      "grad_norm": 0.9325993061065674,
      "learning_rate": 9.146689033505318e-06,
      "loss": 0.7281,
      "step": 7395
    },
    {
      "epoch": 0.2131043623580937,
      "grad_norm": 0.9733716249465942,
      "learning_rate": 9.14642829819375e-06,
      "loss": 0.7801,
      "step": 7396
    },
    {
      "epoch": 0.21313317581974298,
      "grad_norm": 1.0796431303024292,
      "learning_rate": 9.146167526770937e-06,
      "loss": 0.7876,
      "step": 7397
    },
    {
      "epoch": 0.21316198928139227,
      "grad_norm": 1.0024964809417725,
      "learning_rate": 9.14590671923915e-06,
      "loss": 0.7803,
      "step": 7398
    },
    {
      "epoch": 0.21319080274304156,
      "grad_norm": 0.9978121519088745,
      "learning_rate": 9.145645875600659e-06,
      "loss": 0.7709,
      "step": 7399
    },
    {
      "epoch": 0.21321961620469082,
      "grad_norm": 1.027029275894165,
      "learning_rate": 9.145384995857737e-06,
      "loss": 0.77,
      "step": 7400
    },
    {
      "epoch": 0.2132484296663401,
      "grad_norm": 0.9854108095169067,
      "learning_rate": 9.145124080012656e-06,
      "loss": 0.7593,
      "step": 7401
    },
    {
      "epoch": 0.2132772431279894,
      "grad_norm": 1.0940994024276733,
      "learning_rate": 9.144863128067687e-06,
      "loss": 0.8009,
      "step": 7402
    },
    {
      "epoch": 0.2133060565896387,
      "grad_norm": 1.0008198022842407,
      "learning_rate": 9.144602140025105e-06,
      "loss": 0.7769,
      "step": 7403
    },
    {
      "epoch": 0.21333487005128796,
      "grad_norm": 0.9762819409370422,
      "learning_rate": 9.144341115887179e-06,
      "loss": 0.7973,
      "step": 7404
    },
    {
      "epoch": 0.21336368351293725,
      "grad_norm": 1.0067641735076904,
      "learning_rate": 9.144080055656187e-06,
      "loss": 0.7862,
      "step": 7405
    },
    {
      "epoch": 0.21339249697458654,
      "grad_norm": 0.9886564016342163,
      "learning_rate": 9.1438189593344e-06,
      "loss": 0.7758,
      "step": 7406
    },
    {
      "epoch": 0.2134213104362358,
      "grad_norm": 0.9480663537979126,
      "learning_rate": 9.143557826924091e-06,
      "loss": 0.778,
      "step": 7407
    },
    {
      "epoch": 0.2134501238978851,
      "grad_norm": 0.9578189253807068,
      "learning_rate": 9.143296658427536e-06,
      "loss": 0.7877,
      "step": 7408
    },
    {
      "epoch": 0.21347893735953438,
      "grad_norm": 0.9552061557769775,
      "learning_rate": 9.143035453847009e-06,
      "loss": 0.7698,
      "step": 7409
    },
    {
      "epoch": 0.21350775082118365,
      "grad_norm": 0.9709367156028748,
      "learning_rate": 9.142774213184785e-06,
      "loss": 0.7752,
      "step": 7410
    },
    {
      "epoch": 0.21353656428283294,
      "grad_norm": 0.9711406230926514,
      "learning_rate": 9.142512936443137e-06,
      "loss": 0.7911,
      "step": 7411
    },
    {
      "epoch": 0.21356537774448223,
      "grad_norm": 0.9115456342697144,
      "learning_rate": 9.142251623624345e-06,
      "loss": 0.7518,
      "step": 7412
    },
    {
      "epoch": 0.21359419120613152,
      "grad_norm": 0.9874223470687866,
      "learning_rate": 9.14199027473068e-06,
      "loss": 0.772,
      "step": 7413
    },
    {
      "epoch": 0.21362300466778078,
      "grad_norm": 1.010267734527588,
      "learning_rate": 9.14172888976442e-06,
      "loss": 0.7832,
      "step": 7414
    },
    {
      "epoch": 0.21365181812943007,
      "grad_norm": 0.9981455206871033,
      "learning_rate": 9.141467468727843e-06,
      "loss": 0.8015,
      "step": 7415
    },
    {
      "epoch": 0.21368063159107936,
      "grad_norm": 0.9365958571434021,
      "learning_rate": 9.141206011623225e-06,
      "loss": 0.7658,
      "step": 7416
    },
    {
      "epoch": 0.21370944505272862,
      "grad_norm": 0.9726446270942688,
      "learning_rate": 9.14094451845284e-06,
      "loss": 0.8023,
      "step": 7417
    },
    {
      "epoch": 0.21373825851437792,
      "grad_norm": 0.9866440296173096,
      "learning_rate": 9.14068298921897e-06,
      "loss": 0.7845,
      "step": 7418
    },
    {
      "epoch": 0.2137670719760272,
      "grad_norm": 0.9976290464401245,
      "learning_rate": 9.140421423923889e-06,
      "loss": 0.7619,
      "step": 7419
    },
    {
      "epoch": 0.21379588543767647,
      "grad_norm": 0.951773464679718,
      "learning_rate": 9.140159822569876e-06,
      "loss": 0.7613,
      "step": 7420
    },
    {
      "epoch": 0.21382469889932576,
      "grad_norm": 1.3715027570724487,
      "learning_rate": 9.13989818515921e-06,
      "loss": 0.7638,
      "step": 7421
    },
    {
      "epoch": 0.21385351236097505,
      "grad_norm": 0.9545172452926636,
      "learning_rate": 9.13963651169417e-06,
      "loss": 0.7875,
      "step": 7422
    },
    {
      "epoch": 0.21388232582262434,
      "grad_norm": 0.9941548705101013,
      "learning_rate": 9.139374802177032e-06,
      "loss": 0.7805,
      "step": 7423
    },
    {
      "epoch": 0.2139111392842736,
      "grad_norm": 0.9723294973373413,
      "learning_rate": 9.13911305661008e-06,
      "loss": 0.7684,
      "step": 7424
    },
    {
      "epoch": 0.2139399527459229,
      "grad_norm": 0.9978263974189758,
      "learning_rate": 9.13885127499559e-06,
      "loss": 0.792,
      "step": 7425
    },
    {
      "epoch": 0.2139687662075722,
      "grad_norm": 0.9973260760307312,
      "learning_rate": 9.138589457335842e-06,
      "loss": 0.7659,
      "step": 7426
    },
    {
      "epoch": 0.21399757966922145,
      "grad_norm": 0.9432398080825806,
      "learning_rate": 9.138327603633116e-06,
      "loss": 0.7613,
      "step": 7427
    },
    {
      "epoch": 0.21402639313087074,
      "grad_norm": 0.9544690847396851,
      "learning_rate": 9.138065713889696e-06,
      "loss": 0.7901,
      "step": 7428
    },
    {
      "epoch": 0.21405520659252003,
      "grad_norm": 1.0135382413864136,
      "learning_rate": 9.137803788107858e-06,
      "loss": 0.768,
      "step": 7429
    },
    {
      "epoch": 0.21408402005416932,
      "grad_norm": 0.9655538201332092,
      "learning_rate": 9.137541826289887e-06,
      "loss": 0.781,
      "step": 7430
    },
    {
      "epoch": 0.21411283351581858,
      "grad_norm": 0.9728496670722961,
      "learning_rate": 9.13727982843806e-06,
      "loss": 0.7573,
      "step": 7431
    },
    {
      "epoch": 0.21414164697746788,
      "grad_norm": 0.9882880449295044,
      "learning_rate": 9.137017794554664e-06,
      "loss": 0.7756,
      "step": 7432
    },
    {
      "epoch": 0.21417046043911717,
      "grad_norm": 0.9738170504570007,
      "learning_rate": 9.13675572464198e-06,
      "loss": 0.794,
      "step": 7433
    },
    {
      "epoch": 0.21419927390076643,
      "grad_norm": 0.9983313679695129,
      "learning_rate": 9.136493618702284e-06,
      "loss": 0.7769,
      "step": 7434
    },
    {
      "epoch": 0.21422808736241572,
      "grad_norm": 1.0146119594573975,
      "learning_rate": 9.136231476737867e-06,
      "loss": 0.7757,
      "step": 7435
    },
    {
      "epoch": 0.214256900824065,
      "grad_norm": 0.9624285697937012,
      "learning_rate": 9.135969298751008e-06,
      "loss": 0.794,
      "step": 7436
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 0.9423530101776123,
      "learning_rate": 9.13570708474399e-06,
      "loss": 0.7861,
      "step": 7437
    },
    {
      "epoch": 0.21431452774736356,
      "grad_norm": 0.9752873778343201,
      "learning_rate": 9.135444834719098e-06,
      "loss": 0.8007,
      "step": 7438
    },
    {
      "epoch": 0.21434334120901286,
      "grad_norm": 1.0010948181152344,
      "learning_rate": 9.135182548678617e-06,
      "loss": 0.7741,
      "step": 7439
    },
    {
      "epoch": 0.21437215467066215,
      "grad_norm": 0.9301812648773193,
      "learning_rate": 9.134920226624828e-06,
      "loss": 0.7881,
      "step": 7440
    },
    {
      "epoch": 0.2144009681323114,
      "grad_norm": 0.9602518677711487,
      "learning_rate": 9.134657868560017e-06,
      "loss": 0.7749,
      "step": 7441
    },
    {
      "epoch": 0.2144297815939607,
      "grad_norm": 1.015600562095642,
      "learning_rate": 9.134395474486471e-06,
      "loss": 0.775,
      "step": 7442
    },
    {
      "epoch": 0.21445859505561,
      "grad_norm": 0.9597721695899963,
      "learning_rate": 9.13413304440647e-06,
      "loss": 0.7959,
      "step": 7443
    },
    {
      "epoch": 0.21448740851725925,
      "grad_norm": 1.0142062902450562,
      "learning_rate": 9.133870578322306e-06,
      "loss": 0.7907,
      "step": 7444
    },
    {
      "epoch": 0.21451622197890854,
      "grad_norm": 0.9303926825523376,
      "learning_rate": 9.13360807623626e-06,
      "loss": 0.7864,
      "step": 7445
    },
    {
      "epoch": 0.21454503544055784,
      "grad_norm": 0.9459275007247925,
      "learning_rate": 9.13334553815062e-06,
      "loss": 0.7858,
      "step": 7446
    },
    {
      "epoch": 0.2145738489022071,
      "grad_norm": 0.9186997413635254,
      "learning_rate": 9.133082964067672e-06,
      "loss": 0.7686,
      "step": 7447
    },
    {
      "epoch": 0.2146026623638564,
      "grad_norm": 0.9717205762863159,
      "learning_rate": 9.132820353989703e-06,
      "loss": 0.7935,
      "step": 7448
    },
    {
      "epoch": 0.21463147582550568,
      "grad_norm": 0.9791024327278137,
      "learning_rate": 9.132557707919e-06,
      "loss": 0.8049,
      "step": 7449
    },
    {
      "epoch": 0.21466028928715497,
      "grad_norm": 0.9141120314598083,
      "learning_rate": 9.132295025857851e-06,
      "loss": 0.7687,
      "step": 7450
    },
    {
      "epoch": 0.21468910274880423,
      "grad_norm": 0.9292391538619995,
      "learning_rate": 9.132032307808543e-06,
      "loss": 0.7849,
      "step": 7451
    },
    {
      "epoch": 0.21471791621045352,
      "grad_norm": 0.9909691214561462,
      "learning_rate": 9.131769553773364e-06,
      "loss": 0.7831,
      "step": 7452
    },
    {
      "epoch": 0.21474672967210282,
      "grad_norm": 0.932248055934906,
      "learning_rate": 9.131506763754601e-06,
      "loss": 0.7751,
      "step": 7453
    },
    {
      "epoch": 0.21477554313375208,
      "grad_norm": 0.9943211674690247,
      "learning_rate": 9.131243937754546e-06,
      "loss": 0.7617,
      "step": 7454
    },
    {
      "epoch": 0.21480435659540137,
      "grad_norm": 4.172848224639893,
      "learning_rate": 9.130981075775485e-06,
      "loss": 0.7855,
      "step": 7455
    },
    {
      "epoch": 0.21483317005705066,
      "grad_norm": 0.9178249835968018,
      "learning_rate": 9.13071817781971e-06,
      "loss": 0.7799,
      "step": 7456
    },
    {
      "epoch": 0.21486198351869992,
      "grad_norm": 0.9629192352294922,
      "learning_rate": 9.130455243889507e-06,
      "loss": 0.7608,
      "step": 7457
    },
    {
      "epoch": 0.2148907969803492,
      "grad_norm": 0.9639421105384827,
      "learning_rate": 9.130192273987168e-06,
      "loss": 0.7782,
      "step": 7458
    },
    {
      "epoch": 0.2149196104419985,
      "grad_norm": 0.9722594022750854,
      "learning_rate": 9.129929268114983e-06,
      "loss": 0.7649,
      "step": 7459
    },
    {
      "epoch": 0.2149484239036478,
      "grad_norm": 0.9774595499038696,
      "learning_rate": 9.129666226275243e-06,
      "loss": 0.8083,
      "step": 7460
    },
    {
      "epoch": 0.21497723736529706,
      "grad_norm": 0.9515091776847839,
      "learning_rate": 9.129403148470238e-06,
      "loss": 0.7661,
      "step": 7461
    },
    {
      "epoch": 0.21500605082694635,
      "grad_norm": 1.000789999961853,
      "learning_rate": 9.12914003470226e-06,
      "loss": 0.7931,
      "step": 7462
    },
    {
      "epoch": 0.21503486428859564,
      "grad_norm": 0.9595852494239807,
      "learning_rate": 9.1288768849736e-06,
      "loss": 0.7963,
      "step": 7463
    },
    {
      "epoch": 0.2150636777502449,
      "grad_norm": 0.9706538319587708,
      "learning_rate": 9.128613699286549e-06,
      "loss": 0.7729,
      "step": 7464
    },
    {
      "epoch": 0.2150924912118942,
      "grad_norm": 1.1666302680969238,
      "learning_rate": 9.128350477643399e-06,
      "loss": 0.7823,
      "step": 7465
    },
    {
      "epoch": 0.21512130467354348,
      "grad_norm": 0.9570233821868896,
      "learning_rate": 9.128087220046445e-06,
      "loss": 0.7831,
      "step": 7466
    },
    {
      "epoch": 0.21515011813519277,
      "grad_norm": 0.9892816543579102,
      "learning_rate": 9.127823926497977e-06,
      "loss": 0.7758,
      "step": 7467
    },
    {
      "epoch": 0.21517893159684204,
      "grad_norm": 1.0041307210922241,
      "learning_rate": 9.12756059700029e-06,
      "loss": 0.774,
      "step": 7468
    },
    {
      "epoch": 0.21520774505849133,
      "grad_norm": 1.0368938446044922,
      "learning_rate": 9.127297231555674e-06,
      "loss": 0.7911,
      "step": 7469
    },
    {
      "epoch": 0.21523655852014062,
      "grad_norm": 0.9277751445770264,
      "learning_rate": 9.127033830166427e-06,
      "loss": 0.7706,
      "step": 7470
    },
    {
      "epoch": 0.21526537198178988,
      "grad_norm": 1.0200780630111694,
      "learning_rate": 9.126770392834841e-06,
      "loss": 0.7708,
      "step": 7471
    },
    {
      "epoch": 0.21529418544343917,
      "grad_norm": 1.027438998222351,
      "learning_rate": 9.126506919563209e-06,
      "loss": 0.7701,
      "step": 7472
    },
    {
      "epoch": 0.21532299890508846,
      "grad_norm": 0.9715165495872498,
      "learning_rate": 9.126243410353826e-06,
      "loss": 0.8011,
      "step": 7473
    },
    {
      "epoch": 0.21535181236673773,
      "grad_norm": 0.9985926151275635,
      "learning_rate": 9.12597986520899e-06,
      "loss": 0.7632,
      "step": 7474
    },
    {
      "epoch": 0.21538062582838702,
      "grad_norm": 1.1048774719238281,
      "learning_rate": 9.12571628413099e-06,
      "loss": 0.7669,
      "step": 7475
    },
    {
      "epoch": 0.2154094392900363,
      "grad_norm": 0.9525818228721619,
      "learning_rate": 9.125452667122129e-06,
      "loss": 0.7746,
      "step": 7476
    },
    {
      "epoch": 0.2154382527516856,
      "grad_norm": 1.03402578830719,
      "learning_rate": 9.125189014184697e-06,
      "loss": 0.7926,
      "step": 7477
    },
    {
      "epoch": 0.21546706621333486,
      "grad_norm": 0.9662851691246033,
      "learning_rate": 9.124925325320994e-06,
      "loss": 0.7812,
      "step": 7478
    },
    {
      "epoch": 0.21549587967498415,
      "grad_norm": 0.9329919219017029,
      "learning_rate": 9.124661600533313e-06,
      "loss": 0.7606,
      "step": 7479
    },
    {
      "epoch": 0.21552469313663344,
      "grad_norm": 0.9897555708885193,
      "learning_rate": 9.124397839823955e-06,
      "loss": 0.762,
      "step": 7480
    },
    {
      "epoch": 0.2155535065982827,
      "grad_norm": 0.9675180315971375,
      "learning_rate": 9.124134043195214e-06,
      "loss": 0.7519,
      "step": 7481
    },
    {
      "epoch": 0.215582320059932,
      "grad_norm": 0.9562046527862549,
      "learning_rate": 9.123870210649386e-06,
      "loss": 0.7819,
      "step": 7482
    },
    {
      "epoch": 0.2156111335215813,
      "grad_norm": 0.9906163215637207,
      "learning_rate": 9.123606342188773e-06,
      "loss": 0.763,
      "step": 7483
    },
    {
      "epoch": 0.21563994698323055,
      "grad_norm": 0.9618735909461975,
      "learning_rate": 9.123342437815672e-06,
      "loss": 0.7668,
      "step": 7484
    },
    {
      "epoch": 0.21566876044487984,
      "grad_norm": 0.9484066963195801,
      "learning_rate": 9.123078497532376e-06,
      "loss": 0.768,
      "step": 7485
    },
    {
      "epoch": 0.21569757390652913,
      "grad_norm": 0.9846232533454895,
      "learning_rate": 9.122814521341193e-06,
      "loss": 0.7563,
      "step": 7486
    },
    {
      "epoch": 0.21572638736817842,
      "grad_norm": 0.9258737564086914,
      "learning_rate": 9.122550509244412e-06,
      "loss": 0.7725,
      "step": 7487
    },
    {
      "epoch": 0.2157552008298277,
      "grad_norm": 0.9533830285072327,
      "learning_rate": 9.12228646124434e-06,
      "loss": 0.78,
      "step": 7488
    },
    {
      "epoch": 0.21578401429147698,
      "grad_norm": 1.0045957565307617,
      "learning_rate": 9.122022377343273e-06,
      "loss": 0.758,
      "step": 7489
    },
    {
      "epoch": 0.21581282775312627,
      "grad_norm": 0.9744799733161926,
      "learning_rate": 9.121758257543512e-06,
      "loss": 0.7623,
      "step": 7490
    },
    {
      "epoch": 0.21584164121477553,
      "grad_norm": 0.9563214778900146,
      "learning_rate": 9.121494101847356e-06,
      "loss": 0.7572,
      "step": 7491
    },
    {
      "epoch": 0.21587045467642482,
      "grad_norm": 0.9528480172157288,
      "learning_rate": 9.121229910257107e-06,
      "loss": 0.7901,
      "step": 7492
    },
    {
      "epoch": 0.2158992681380741,
      "grad_norm": 1.015578269958496,
      "learning_rate": 9.120965682775065e-06,
      "loss": 0.7914,
      "step": 7493
    },
    {
      "epoch": 0.2159280815997234,
      "grad_norm": 0.9420563578605652,
      "learning_rate": 9.12070141940353e-06,
      "loss": 0.7768,
      "step": 7494
    },
    {
      "epoch": 0.21595689506137267,
      "grad_norm": 0.9492080211639404,
      "learning_rate": 9.120437120144808e-06,
      "loss": 0.7919,
      "step": 7495
    },
    {
      "epoch": 0.21598570852302196,
      "grad_norm": 0.9719055891036987,
      "learning_rate": 9.120172785001194e-06,
      "loss": 0.7608,
      "step": 7496
    },
    {
      "epoch": 0.21601452198467125,
      "grad_norm": 0.9364324808120728,
      "learning_rate": 9.119908413974996e-06,
      "loss": 0.7848,
      "step": 7497
    },
    {
      "epoch": 0.2160433354463205,
      "grad_norm": 0.9565799832344055,
      "learning_rate": 9.119644007068514e-06,
      "loss": 0.7642,
      "step": 7498
    },
    {
      "epoch": 0.2160721489079698,
      "grad_norm": 0.97169029712677,
      "learning_rate": 9.11937956428405e-06,
      "loss": 0.7586,
      "step": 7499
    },
    {
      "epoch": 0.2161009623696191,
      "grad_norm": 0.9499953985214233,
      "learning_rate": 9.11911508562391e-06,
      "loss": 0.7591,
      "step": 7500
    },
    {
      "epoch": 0.21612977583126836,
      "grad_norm": 0.9637508392333984,
      "learning_rate": 9.118850571090392e-06,
      "loss": 0.7868,
      "step": 7501
    },
    {
      "epoch": 0.21615858929291765,
      "grad_norm": 0.9709029793739319,
      "learning_rate": 9.118586020685804e-06,
      "loss": 0.7785,
      "step": 7502
    },
    {
      "epoch": 0.21618740275456694,
      "grad_norm": 0.9672399163246155,
      "learning_rate": 9.11832143441245e-06,
      "loss": 0.7593,
      "step": 7503
    },
    {
      "epoch": 0.21621621621621623,
      "grad_norm": 0.9966880083084106,
      "learning_rate": 9.118056812272632e-06,
      "loss": 0.756,
      "step": 7504
    },
    {
      "epoch": 0.2162450296778655,
      "grad_norm": 0.9355714321136475,
      "learning_rate": 9.117792154268656e-06,
      "loss": 0.7962,
      "step": 7505
    },
    {
      "epoch": 0.21627384313951478,
      "grad_norm": 0.9840354919433594,
      "learning_rate": 9.117527460402826e-06,
      "loss": 0.7764,
      "step": 7506
    },
    {
      "epoch": 0.21630265660116407,
      "grad_norm": 0.9497789144515991,
      "learning_rate": 9.11726273067745e-06,
      "loss": 0.761,
      "step": 7507
    },
    {
      "epoch": 0.21633147006281334,
      "grad_norm": 0.9464945197105408,
      "learning_rate": 9.11699796509483e-06,
      "loss": 0.7879,
      "step": 7508
    },
    {
      "epoch": 0.21636028352446263,
      "grad_norm": 0.9810227155685425,
      "learning_rate": 9.116733163657273e-06,
      "loss": 0.7854,
      "step": 7509
    },
    {
      "epoch": 0.21638909698611192,
      "grad_norm": 0.9063573479652405,
      "learning_rate": 9.116468326367086e-06,
      "loss": 0.7625,
      "step": 7510
    },
    {
      "epoch": 0.21641791044776118,
      "grad_norm": 0.9421701431274414,
      "learning_rate": 9.116203453226573e-06,
      "loss": 0.7762,
      "step": 7511
    },
    {
      "epoch": 0.21644672390941047,
      "grad_norm": 0.9743598103523254,
      "learning_rate": 9.115938544238043e-06,
      "loss": 0.7659,
      "step": 7512
    },
    {
      "epoch": 0.21647553737105976,
      "grad_norm": 0.9804569482803345,
      "learning_rate": 9.115673599403805e-06,
      "loss": 0.7778,
      "step": 7513
    },
    {
      "epoch": 0.21650435083270905,
      "grad_norm": 0.9578443169593811,
      "learning_rate": 9.115408618726163e-06,
      "loss": 0.7766,
      "step": 7514
    },
    {
      "epoch": 0.21653316429435832,
      "grad_norm": 0.9610117077827454,
      "learning_rate": 9.115143602207425e-06,
      "loss": 0.7591,
      "step": 7515
    },
    {
      "epoch": 0.2165619777560076,
      "grad_norm": 0.9241695404052734,
      "learning_rate": 9.114878549849899e-06,
      "loss": 0.7536,
      "step": 7516
    },
    {
      "epoch": 0.2165907912176569,
      "grad_norm": 0.976828396320343,
      "learning_rate": 9.114613461655897e-06,
      "loss": 0.8019,
      "step": 7517
    },
    {
      "epoch": 0.21661960467930616,
      "grad_norm": 0.9074607491493225,
      "learning_rate": 9.114348337627724e-06,
      "loss": 0.7858,
      "step": 7518
    },
    {
      "epoch": 0.21664841814095545,
      "grad_norm": 1.0250790119171143,
      "learning_rate": 9.114083177767689e-06,
      "loss": 0.7888,
      "step": 7519
    },
    {
      "epoch": 0.21667723160260474,
      "grad_norm": 0.9909679293632507,
      "learning_rate": 9.1138179820781e-06,
      "loss": 0.7782,
      "step": 7520
    },
    {
      "epoch": 0.21670604506425403,
      "grad_norm": 0.9531581997871399,
      "learning_rate": 9.113552750561274e-06,
      "loss": 0.7698,
      "step": 7521
    },
    {
      "epoch": 0.2167348585259033,
      "grad_norm": 0.9723525047302246,
      "learning_rate": 9.113287483219512e-06,
      "loss": 0.7914,
      "step": 7522
    },
    {
      "epoch": 0.2167636719875526,
      "grad_norm": 0.9195886850357056,
      "learning_rate": 9.11302218005513e-06,
      "loss": 0.7529,
      "step": 7523
    },
    {
      "epoch": 0.21679248544920188,
      "grad_norm": 0.9088631272315979,
      "learning_rate": 9.112756841070433e-06,
      "loss": 0.7658,
      "step": 7524
    },
    {
      "epoch": 0.21682129891085114,
      "grad_norm": 0.9357715249061584,
      "learning_rate": 9.112491466267736e-06,
      "loss": 0.7697,
      "step": 7525
    },
    {
      "epoch": 0.21685011237250043,
      "grad_norm": 0.9527407288551331,
      "learning_rate": 9.112226055649351e-06,
      "loss": 0.7577,
      "step": 7526
    },
    {
      "epoch": 0.21687892583414972,
      "grad_norm": 0.9538980722427368,
      "learning_rate": 9.111960609217587e-06,
      "loss": 0.7711,
      "step": 7527
    },
    {
      "epoch": 0.21690773929579898,
      "grad_norm": 0.9322242140769958,
      "learning_rate": 9.111695126974758e-06,
      "loss": 0.7786,
      "step": 7528
    },
    {
      "epoch": 0.21693655275744828,
      "grad_norm": 0.9148539900779724,
      "learning_rate": 9.111429608923173e-06,
      "loss": 0.7971,
      "step": 7529
    },
    {
      "epoch": 0.21696536621909757,
      "grad_norm": 0.9220195412635803,
      "learning_rate": 9.111164055065146e-06,
      "loss": 0.7647,
      "step": 7530
    },
    {
      "epoch": 0.21699417968074686,
      "grad_norm": 0.959697425365448,
      "learning_rate": 9.11089846540299e-06,
      "loss": 0.7992,
      "step": 7531
    },
    {
      "epoch": 0.21702299314239612,
      "grad_norm": 0.9405474066734314,
      "learning_rate": 9.110632839939017e-06,
      "loss": 0.7785,
      "step": 7532
    },
    {
      "epoch": 0.2170518066040454,
      "grad_norm": 0.887181282043457,
      "learning_rate": 9.110367178675544e-06,
      "loss": 0.7437,
      "step": 7533
    },
    {
      "epoch": 0.2170806200656947,
      "grad_norm": 0.956131100654602,
      "learning_rate": 9.110101481614879e-06,
      "loss": 0.7894,
      "step": 7534
    },
    {
      "epoch": 0.21710943352734396,
      "grad_norm": 0.941312313079834,
      "learning_rate": 9.109835748759339e-06,
      "loss": 0.7772,
      "step": 7535
    },
    {
      "epoch": 0.21713824698899326,
      "grad_norm": 0.96578049659729,
      "learning_rate": 9.109569980111237e-06,
      "loss": 0.764,
      "step": 7536
    },
    {
      "epoch": 0.21716706045064255,
      "grad_norm": 0.9224735498428345,
      "learning_rate": 9.10930417567289e-06,
      "loss": 0.7833,
      "step": 7537
    },
    {
      "epoch": 0.2171958739122918,
      "grad_norm": 0.9455657005310059,
      "learning_rate": 9.109038335446612e-06,
      "loss": 0.7585,
      "step": 7538
    },
    {
      "epoch": 0.2172246873739411,
      "grad_norm": 0.9489198327064514,
      "learning_rate": 9.108772459434718e-06,
      "loss": 0.7525,
      "step": 7539
    },
    {
      "epoch": 0.2172535008355904,
      "grad_norm": 0.9775040149688721,
      "learning_rate": 9.108506547639522e-06,
      "loss": 0.7931,
      "step": 7540
    },
    {
      "epoch": 0.21728231429723968,
      "grad_norm": 0.9248535633087158,
      "learning_rate": 9.10824060006334e-06,
      "loss": 0.7799,
      "step": 7541
    },
    {
      "epoch": 0.21731112775888894,
      "grad_norm": 0.9169818162918091,
      "learning_rate": 9.10797461670849e-06,
      "loss": 0.8101,
      "step": 7542
    },
    {
      "epoch": 0.21733994122053824,
      "grad_norm": 0.9329153895378113,
      "learning_rate": 9.107708597577287e-06,
      "loss": 0.7707,
      "step": 7543
    },
    {
      "epoch": 0.21736875468218753,
      "grad_norm": 0.962818443775177,
      "learning_rate": 9.10744254267205e-06,
      "loss": 0.7774,
      "step": 7544
    },
    {
      "epoch": 0.2173975681438368,
      "grad_norm": 0.9429304599761963,
      "learning_rate": 9.107176451995093e-06,
      "loss": 0.764,
      "step": 7545
    },
    {
      "epoch": 0.21742638160548608,
      "grad_norm": 0.958349883556366,
      "learning_rate": 9.106910325548734e-06,
      "loss": 0.7717,
      "step": 7546
    },
    {
      "epoch": 0.21745519506713537,
      "grad_norm": 0.9804650545120239,
      "learning_rate": 9.106644163335295e-06,
      "loss": 0.7876,
      "step": 7547
    },
    {
      "epoch": 0.21748400852878466,
      "grad_norm": 0.9797863364219666,
      "learning_rate": 9.106377965357087e-06,
      "loss": 0.7792,
      "step": 7548
    },
    {
      "epoch": 0.21751282199043392,
      "grad_norm": 0.9135181903839111,
      "learning_rate": 9.106111731616434e-06,
      "loss": 0.7569,
      "step": 7549
    },
    {
      "epoch": 0.21754163545208322,
      "grad_norm": 0.953057587146759,
      "learning_rate": 9.10584546211565e-06,
      "loss": 0.7845,
      "step": 7550
    },
    {
      "epoch": 0.2175704489137325,
      "grad_norm": 1.055051326751709,
      "learning_rate": 9.105579156857058e-06,
      "loss": 0.7987,
      "step": 7551
    },
    {
      "epoch": 0.21759926237538177,
      "grad_norm": 1.0061421394348145,
      "learning_rate": 9.105312815842977e-06,
      "loss": 0.7846,
      "step": 7552
    },
    {
      "epoch": 0.21762807583703106,
      "grad_norm": 0.914214015007019,
      "learning_rate": 9.105046439075723e-06,
      "loss": 0.7727,
      "step": 7553
    },
    {
      "epoch": 0.21765688929868035,
      "grad_norm": 0.9645978808403015,
      "learning_rate": 9.104780026557619e-06,
      "loss": 0.7719,
      "step": 7554
    },
    {
      "epoch": 0.2176857027603296,
      "grad_norm": 0.9630320072174072,
      "learning_rate": 9.104513578290984e-06,
      "loss": 0.7681,
      "step": 7555
    },
    {
      "epoch": 0.2177145162219789,
      "grad_norm": 0.9297203421592712,
      "learning_rate": 9.104247094278137e-06,
      "loss": 0.7411,
      "step": 7556
    },
    {
      "epoch": 0.2177433296836282,
      "grad_norm": 0.9441240429878235,
      "learning_rate": 9.103980574521403e-06,
      "loss": 0.7713,
      "step": 7557
    },
    {
      "epoch": 0.21777214314527749,
      "grad_norm": 0.9271679520606995,
      "learning_rate": 9.1037140190231e-06,
      "loss": 0.7687,
      "step": 7558
    },
    {
      "epoch": 0.21780095660692675,
      "grad_norm": 0.9592112898826599,
      "learning_rate": 9.103447427785549e-06,
      "loss": 0.7852,
      "step": 7559
    },
    {
      "epoch": 0.21782977006857604,
      "grad_norm": 0.9268908500671387,
      "learning_rate": 9.103180800811074e-06,
      "loss": 0.796,
      "step": 7560
    },
    {
      "epoch": 0.21785858353022533,
      "grad_norm": 0.9169591069221497,
      "learning_rate": 9.102914138101995e-06,
      "loss": 0.7669,
      "step": 7561
    },
    {
      "epoch": 0.2178873969918746,
      "grad_norm": 0.9457791447639465,
      "learning_rate": 9.102647439660637e-06,
      "loss": 0.78,
      "step": 7562
    },
    {
      "epoch": 0.21791621045352388,
      "grad_norm": 0.943655788898468,
      "learning_rate": 9.10238070548932e-06,
      "loss": 0.7916,
      "step": 7563
    },
    {
      "epoch": 0.21794502391517317,
      "grad_norm": 0.9408902525901794,
      "learning_rate": 9.102113935590367e-06,
      "loss": 0.743,
      "step": 7564
    },
    {
      "epoch": 0.21797383737682244,
      "grad_norm": 0.975633978843689,
      "learning_rate": 9.101847129966103e-06,
      "loss": 0.7882,
      "step": 7565
    },
    {
      "epoch": 0.21800265083847173,
      "grad_norm": 0.9497560262680054,
      "learning_rate": 9.101580288618851e-06,
      "loss": 0.7461,
      "step": 7566
    },
    {
      "epoch": 0.21803146430012102,
      "grad_norm": 0.9734731316566467,
      "learning_rate": 9.101313411550935e-06,
      "loss": 0.7748,
      "step": 7567
    },
    {
      "epoch": 0.2180602777617703,
      "grad_norm": 0.9414304494857788,
      "learning_rate": 9.101046498764679e-06,
      "loss": 0.7599,
      "step": 7568
    },
    {
      "epoch": 0.21808909122341957,
      "grad_norm": 0.9746713638305664,
      "learning_rate": 9.100779550262406e-06,
      "loss": 0.8119,
      "step": 7569
    },
    {
      "epoch": 0.21811790468506886,
      "grad_norm": 0.9539658427238464,
      "learning_rate": 9.100512566046445e-06,
      "loss": 0.7726,
      "step": 7570
    },
    {
      "epoch": 0.21814671814671815,
      "grad_norm": 0.9319527745246887,
      "learning_rate": 9.100245546119117e-06,
      "loss": 0.7793,
      "step": 7571
    },
    {
      "epoch": 0.21817553160836742,
      "grad_norm": 0.9315453767776489,
      "learning_rate": 9.099978490482748e-06,
      "loss": 0.7989,
      "step": 7572
    },
    {
      "epoch": 0.2182043450700167,
      "grad_norm": 0.9458534121513367,
      "learning_rate": 9.099711399139667e-06,
      "loss": 0.7743,
      "step": 7573
    },
    {
      "epoch": 0.218233158531666,
      "grad_norm": 0.915545642375946,
      "learning_rate": 9.099444272092197e-06,
      "loss": 0.767,
      "step": 7574
    },
    {
      "epoch": 0.2182619719933153,
      "grad_norm": 0.9469794034957886,
      "learning_rate": 9.099177109342665e-06,
      "loss": 0.7586,
      "step": 7575
    },
    {
      "epoch": 0.21829078545496455,
      "grad_norm": 0.9379966855049133,
      "learning_rate": 9.098909910893399e-06,
      "loss": 0.7841,
      "step": 7576
    },
    {
      "epoch": 0.21831959891661384,
      "grad_norm": 0.9140955805778503,
      "learning_rate": 9.098642676746722e-06,
      "loss": 0.7755,
      "step": 7577
    },
    {
      "epoch": 0.21834841237826313,
      "grad_norm": 0.9862255454063416,
      "learning_rate": 9.098375406904969e-06,
      "loss": 0.7763,
      "step": 7578
    },
    {
      "epoch": 0.2183772258399124,
      "grad_norm": 0.9580332636833191,
      "learning_rate": 9.098108101370459e-06,
      "loss": 0.7635,
      "step": 7579
    },
    {
      "epoch": 0.2184060393015617,
      "grad_norm": 0.8873940706253052,
      "learning_rate": 9.097840760145526e-06,
      "loss": 0.7608,
      "step": 7580
    },
    {
      "epoch": 0.21843485276321098,
      "grad_norm": 0.9434971809387207,
      "learning_rate": 9.097573383232498e-06,
      "loss": 0.7626,
      "step": 7581
    },
    {
      "epoch": 0.21846366622486024,
      "grad_norm": 0.9669071435928345,
      "learning_rate": 9.097305970633701e-06,
      "loss": 0.7526,
      "step": 7582
    },
    {
      "epoch": 0.21849247968650953,
      "grad_norm": 0.9802746772766113,
      "learning_rate": 9.097038522351463e-06,
      "loss": 0.7835,
      "step": 7583
    },
    {
      "epoch": 0.21852129314815882,
      "grad_norm": 0.991065263748169,
      "learning_rate": 9.096771038388117e-06,
      "loss": 0.783,
      "step": 7584
    },
    {
      "epoch": 0.21855010660980811,
      "grad_norm": 0.912720263004303,
      "learning_rate": 9.096503518745989e-06,
      "loss": 0.7535,
      "step": 7585
    },
    {
      "epoch": 0.21857892007145738,
      "grad_norm": 0.952648401260376,
      "learning_rate": 9.09623596342741e-06,
      "loss": 0.7884,
      "step": 7586
    },
    {
      "epoch": 0.21860773353310667,
      "grad_norm": 0.9848323464393616,
      "learning_rate": 9.095968372434712e-06,
      "loss": 0.7978,
      "step": 7587
    },
    {
      "epoch": 0.21863654699475596,
      "grad_norm": 0.958860456943512,
      "learning_rate": 9.095700745770224e-06,
      "loss": 0.7957,
      "step": 7588
    },
    {
      "epoch": 0.21866536045640522,
      "grad_norm": 0.9380216598510742,
      "learning_rate": 9.095433083436275e-06,
      "loss": 0.7694,
      "step": 7589
    },
    {
      "epoch": 0.2186941739180545,
      "grad_norm": 0.9451403021812439,
      "learning_rate": 9.095165385435199e-06,
      "loss": 0.7602,
      "step": 7590
    },
    {
      "epoch": 0.2187229873797038,
      "grad_norm": 0.9181925058364868,
      "learning_rate": 9.094897651769325e-06,
      "loss": 0.757,
      "step": 7591
    },
    {
      "epoch": 0.21875180084135307,
      "grad_norm": 0.9685989022254944,
      "learning_rate": 9.094629882440988e-06,
      "loss": 0.7734,
      "step": 7592
    },
    {
      "epoch": 0.21878061430300236,
      "grad_norm": 0.9478105306625366,
      "learning_rate": 9.094362077452516e-06,
      "loss": 0.7689,
      "step": 7593
    },
    {
      "epoch": 0.21880942776465165,
      "grad_norm": 0.9506344199180603,
      "learning_rate": 9.094094236806245e-06,
      "loss": 0.7811,
      "step": 7594
    },
    {
      "epoch": 0.21883824122630094,
      "grad_norm": 0.9562602043151855,
      "learning_rate": 9.093826360504504e-06,
      "loss": 0.8015,
      "step": 7595
    },
    {
      "epoch": 0.2188670546879502,
      "grad_norm": 0.9550659656524658,
      "learning_rate": 9.093558448549628e-06,
      "loss": 0.7751,
      "step": 7596
    },
    {
      "epoch": 0.2188958681495995,
      "grad_norm": 0.911163866519928,
      "learning_rate": 9.093290500943951e-06,
      "loss": 0.7583,
      "step": 7597
    },
    {
      "epoch": 0.21892468161124878,
      "grad_norm": 1.0107895135879517,
      "learning_rate": 9.093022517689804e-06,
      "loss": 0.7706,
      "step": 7598
    },
    {
      "epoch": 0.21895349507289805,
      "grad_norm": 0.9412890076637268,
      "learning_rate": 9.092754498789525e-06,
      "loss": 0.7615,
      "step": 7599
    },
    {
      "epoch": 0.21898230853454734,
      "grad_norm": 0.9402620792388916,
      "learning_rate": 9.092486444245443e-06,
      "loss": 0.7668,
      "step": 7600
    },
    {
      "epoch": 0.21901112199619663,
      "grad_norm": 0.9754842519760132,
      "learning_rate": 9.092218354059896e-06,
      "loss": 0.7702,
      "step": 7601
    },
    {
      "epoch": 0.21903993545784592,
      "grad_norm": 1.0500317811965942,
      "learning_rate": 9.091950228235218e-06,
      "loss": 0.7688,
      "step": 7602
    },
    {
      "epoch": 0.21906874891949518,
      "grad_norm": 0.944958508014679,
      "learning_rate": 9.091682066773744e-06,
      "loss": 0.787,
      "step": 7603
    },
    {
      "epoch": 0.21909756238114447,
      "grad_norm": 0.963010847568512,
      "learning_rate": 9.091413869677809e-06,
      "loss": 0.773,
      "step": 7604
    },
    {
      "epoch": 0.21912637584279376,
      "grad_norm": 0.9670841693878174,
      "learning_rate": 9.091145636949748e-06,
      "loss": 0.7886,
      "step": 7605
    },
    {
      "epoch": 0.21915518930444303,
      "grad_norm": 0.9488402605056763,
      "learning_rate": 9.090877368591901e-06,
      "loss": 0.8052,
      "step": 7606
    },
    {
      "epoch": 0.21918400276609232,
      "grad_norm": 0.9694336652755737,
      "learning_rate": 9.0906090646066e-06,
      "loss": 0.7707,
      "step": 7607
    },
    {
      "epoch": 0.2192128162277416,
      "grad_norm": 0.9538979530334473,
      "learning_rate": 9.090340724996184e-06,
      "loss": 0.7775,
      "step": 7608
    },
    {
      "epoch": 0.21924162968939087,
      "grad_norm": 0.9101896286010742,
      "learning_rate": 9.090072349762987e-06,
      "loss": 0.7754,
      "step": 7609
    },
    {
      "epoch": 0.21927044315104016,
      "grad_norm": 1.0022951364517212,
      "learning_rate": 9.08980393890935e-06,
      "loss": 0.7691,
      "step": 7610
    },
    {
      "epoch": 0.21929925661268945,
      "grad_norm": 0.9685573577880859,
      "learning_rate": 9.089535492437607e-06,
      "loss": 0.7761,
      "step": 7611
    },
    {
      "epoch": 0.21932807007433874,
      "grad_norm": 1.1534510850906372,
      "learning_rate": 9.0892670103501e-06,
      "loss": 0.7845,
      "step": 7612
    },
    {
      "epoch": 0.219356883535988,
      "grad_norm": 0.9702924489974976,
      "learning_rate": 9.088998492649165e-06,
      "loss": 0.7684,
      "step": 7613
    },
    {
      "epoch": 0.2193856969976373,
      "grad_norm": 1.001717209815979,
      "learning_rate": 9.08872993933714e-06,
      "loss": 0.7843,
      "step": 7614
    },
    {
      "epoch": 0.2194145104592866,
      "grad_norm": 0.917343258857727,
      "learning_rate": 9.088461350416364e-06,
      "loss": 0.7632,
      "step": 7615
    },
    {
      "epoch": 0.21944332392093585,
      "grad_norm": 0.9207226037979126,
      "learning_rate": 9.088192725889177e-06,
      "loss": 0.7743,
      "step": 7616
    },
    {
      "epoch": 0.21947213738258514,
      "grad_norm": 0.9616498351097107,
      "learning_rate": 9.08792406575792e-06,
      "loss": 0.7681,
      "step": 7617
    },
    {
      "epoch": 0.21950095084423443,
      "grad_norm": 0.9514581561088562,
      "learning_rate": 9.087655370024928e-06,
      "loss": 0.7709,
      "step": 7618
    },
    {
      "epoch": 0.2195297643058837,
      "grad_norm": 0.9728921055793762,
      "learning_rate": 9.087386638692546e-06,
      "loss": 0.7706,
      "step": 7619
    },
    {
      "epoch": 0.219558577767533,
      "grad_norm": 0.9178126454353333,
      "learning_rate": 9.087117871763112e-06,
      "loss": 0.7703,
      "step": 7620
    },
    {
      "epoch": 0.21958739122918228,
      "grad_norm": 0.941576361656189,
      "learning_rate": 9.086849069238965e-06,
      "loss": 0.7708,
      "step": 7621
    },
    {
      "epoch": 0.21961620469083157,
      "grad_norm": 0.9441902041435242,
      "learning_rate": 9.086580231122449e-06,
      "loss": 0.7522,
      "step": 7622
    },
    {
      "epoch": 0.21964501815248083,
      "grad_norm": 0.9642331600189209,
      "learning_rate": 9.086311357415904e-06,
      "loss": 0.79,
      "step": 7623
    },
    {
      "epoch": 0.21967383161413012,
      "grad_norm": 0.9383315443992615,
      "learning_rate": 9.086042448121672e-06,
      "loss": 0.7532,
      "step": 7624
    },
    {
      "epoch": 0.2197026450757794,
      "grad_norm": 0.9068524837493896,
      "learning_rate": 9.085773503242096e-06,
      "loss": 0.7582,
      "step": 7625
    },
    {
      "epoch": 0.21973145853742868,
      "grad_norm": 0.9616331458091736,
      "learning_rate": 9.085504522779517e-06,
      "loss": 0.8097,
      "step": 7626
    },
    {
      "epoch": 0.21976027199907797,
      "grad_norm": 0.897132933139801,
      "learning_rate": 9.085235506736276e-06,
      "loss": 0.7631,
      "step": 7627
    },
    {
      "epoch": 0.21978908546072726,
      "grad_norm": 0.9658293724060059,
      "learning_rate": 9.084966455114718e-06,
      "loss": 0.7691,
      "step": 7628
    },
    {
      "epoch": 0.21981789892237655,
      "grad_norm": 0.9293500185012817,
      "learning_rate": 9.084697367917188e-06,
      "loss": 0.7491,
      "step": 7629
    },
    {
      "epoch": 0.2198467123840258,
      "grad_norm": 0.926398754119873,
      "learning_rate": 9.084428245146025e-06,
      "loss": 0.7743,
      "step": 7630
    },
    {
      "epoch": 0.2198755258456751,
      "grad_norm": 1.0118812322616577,
      "learning_rate": 9.084159086803575e-06,
      "loss": 0.7797,
      "step": 7631
    },
    {
      "epoch": 0.2199043393073244,
      "grad_norm": 0.960796058177948,
      "learning_rate": 9.083889892892181e-06,
      "loss": 0.79,
      "step": 7632
    },
    {
      "epoch": 0.21993315276897366,
      "grad_norm": 0.9591547846794128,
      "learning_rate": 9.08362066341419e-06,
      "loss": 0.7945,
      "step": 7633
    },
    {
      "epoch": 0.21996196623062295,
      "grad_norm": 0.9595354199409485,
      "learning_rate": 9.083351398371944e-06,
      "loss": 0.761,
      "step": 7634
    },
    {
      "epoch": 0.21999077969227224,
      "grad_norm": 0.9467610716819763,
      "learning_rate": 9.08308209776779e-06,
      "loss": 0.7501,
      "step": 7635
    },
    {
      "epoch": 0.2200195931539215,
      "grad_norm": 0.9284171462059021,
      "learning_rate": 9.082812761604072e-06,
      "loss": 0.7643,
      "step": 7636
    },
    {
      "epoch": 0.2200484066155708,
      "grad_norm": 0.9753338098526001,
      "learning_rate": 9.082543389883137e-06,
      "loss": 0.7676,
      "step": 7637
    },
    {
      "epoch": 0.22007722007722008,
      "grad_norm": 0.9606574177742004,
      "learning_rate": 9.08227398260733e-06,
      "loss": 0.7702,
      "step": 7638
    },
    {
      "epoch": 0.22010603353886937,
      "grad_norm": 0.8985231518745422,
      "learning_rate": 9.082004539778998e-06,
      "loss": 0.7614,
      "step": 7639
    },
    {
      "epoch": 0.22013484700051864,
      "grad_norm": 0.9475724101066589,
      "learning_rate": 9.081735061400485e-06,
      "loss": 0.7791,
      "step": 7640
    },
    {
      "epoch": 0.22016366046216793,
      "grad_norm": 1.0059894323349,
      "learning_rate": 9.081465547474142e-06,
      "loss": 0.7974,
      "step": 7641
    },
    {
      "epoch": 0.22019247392381722,
      "grad_norm": 1.017066478729248,
      "learning_rate": 9.081195998002313e-06,
      "loss": 0.7792,
      "step": 7642
    },
    {
      "epoch": 0.22022128738546648,
      "grad_norm": 0.9440346956253052,
      "learning_rate": 9.080926412987345e-06,
      "loss": 0.7567,
      "step": 7643
    },
    {
      "epoch": 0.22025010084711577,
      "grad_norm": 0.9807847738265991,
      "learning_rate": 9.08065679243159e-06,
      "loss": 0.7754,
      "step": 7644
    },
    {
      "epoch": 0.22027891430876506,
      "grad_norm": 0.9827441573143005,
      "learning_rate": 9.080387136337391e-06,
      "loss": 0.7823,
      "step": 7645
    },
    {
      "epoch": 0.22030772777041432,
      "grad_norm": 0.9461979269981384,
      "learning_rate": 9.080117444707101e-06,
      "loss": 0.7767,
      "step": 7646
    },
    {
      "epoch": 0.22033654123206362,
      "grad_norm": 0.930889368057251,
      "learning_rate": 9.079847717543066e-06,
      "loss": 0.7539,
      "step": 7647
    },
    {
      "epoch": 0.2203653546937129,
      "grad_norm": 0.9695230722427368,
      "learning_rate": 9.079577954847637e-06,
      "loss": 0.7756,
      "step": 7648
    },
    {
      "epoch": 0.2203941681553622,
      "grad_norm": 0.9008944034576416,
      "learning_rate": 9.07930815662316e-06,
      "loss": 0.7702,
      "step": 7649
    },
    {
      "epoch": 0.22042298161701146,
      "grad_norm": 0.9756471514701843,
      "learning_rate": 9.079038322871986e-06,
      "loss": 0.7692,
      "step": 7650
    },
    {
      "epoch": 0.22045179507866075,
      "grad_norm": 0.90998774766922,
      "learning_rate": 9.07876845359647e-06,
      "loss": 0.7822,
      "step": 7651
    },
    {
      "epoch": 0.22048060854031004,
      "grad_norm": 0.9175862669944763,
      "learning_rate": 9.078498548798954e-06,
      "loss": 0.7909,
      "step": 7652
    },
    {
      "epoch": 0.2205094220019593,
      "grad_norm": 0.9544772505760193,
      "learning_rate": 9.078228608481793e-06,
      "loss": 0.7919,
      "step": 7653
    },
    {
      "epoch": 0.2205382354636086,
      "grad_norm": 0.9829593896865845,
      "learning_rate": 9.077958632647338e-06,
      "loss": 0.7623,
      "step": 7654
    },
    {
      "epoch": 0.22056704892525789,
      "grad_norm": 0.9439228177070618,
      "learning_rate": 9.077688621297942e-06,
      "loss": 0.8073,
      "step": 7655
    },
    {
      "epoch": 0.22059586238690718,
      "grad_norm": 0.9340195655822754,
      "learning_rate": 9.077418574435952e-06,
      "loss": 0.7781,
      "step": 7656
    },
    {
      "epoch": 0.22062467584855644,
      "grad_norm": 0.9771760702133179,
      "learning_rate": 9.077148492063724e-06,
      "loss": 0.788,
      "step": 7657
    },
    {
      "epoch": 0.22065348931020573,
      "grad_norm": 0.9581988453865051,
      "learning_rate": 9.076878374183607e-06,
      "loss": 0.7808,
      "step": 7658
    },
    {
      "epoch": 0.22068230277185502,
      "grad_norm": 1.0251948833465576,
      "learning_rate": 9.076608220797954e-06,
      "loss": 0.7765,
      "step": 7659
    },
    {
      "epoch": 0.22071111623350428,
      "grad_norm": 0.9402991533279419,
      "learning_rate": 9.07633803190912e-06,
      "loss": 0.7708,
      "step": 7660
    },
    {
      "epoch": 0.22073992969515357,
      "grad_norm": 0.9279049634933472,
      "learning_rate": 9.076067807519456e-06,
      "loss": 0.7719,
      "step": 7661
    },
    {
      "epoch": 0.22076874315680287,
      "grad_norm": 0.912466824054718,
      "learning_rate": 9.075797547631316e-06,
      "loss": 0.7447,
      "step": 7662
    },
    {
      "epoch": 0.22079755661845213,
      "grad_norm": 0.9228683114051819,
      "learning_rate": 9.075527252247054e-06,
      "loss": 0.7638,
      "step": 7663
    },
    {
      "epoch": 0.22082637008010142,
      "grad_norm": 0.9218493700027466,
      "learning_rate": 9.075256921369022e-06,
      "loss": 0.7757,
      "step": 7664
    },
    {
      "epoch": 0.2208551835417507,
      "grad_norm": 0.9247820377349854,
      "learning_rate": 9.074986554999579e-06,
      "loss": 0.7594,
      "step": 7665
    },
    {
      "epoch": 0.2208839970034,
      "grad_norm": 0.9372572302818298,
      "learning_rate": 9.074716153141074e-06,
      "loss": 0.7748,
      "step": 7666
    },
    {
      "epoch": 0.22091281046504926,
      "grad_norm": 0.9793933033943176,
      "learning_rate": 9.074445715795865e-06,
      "loss": 0.7971,
      "step": 7667
    },
    {
      "epoch": 0.22094162392669855,
      "grad_norm": 0.9353010058403015,
      "learning_rate": 9.074175242966307e-06,
      "loss": 0.7616,
      "step": 7668
    },
    {
      "epoch": 0.22097043738834785,
      "grad_norm": 0.9321249723434448,
      "learning_rate": 9.073904734654756e-06,
      "loss": 0.7905,
      "step": 7669
    },
    {
      "epoch": 0.2209992508499971,
      "grad_norm": 0.9380775690078735,
      "learning_rate": 9.073634190863567e-06,
      "loss": 0.758,
      "step": 7670
    },
    {
      "epoch": 0.2210280643116464,
      "grad_norm": 0.910155713558197,
      "learning_rate": 9.073363611595097e-06,
      "loss": 0.7406,
      "step": 7671
    },
    {
      "epoch": 0.2210568777732957,
      "grad_norm": 0.9075837731361389,
      "learning_rate": 9.0730929968517e-06,
      "loss": 0.7475,
      "step": 7672
    },
    {
      "epoch": 0.22108569123494495,
      "grad_norm": 0.9652786254882812,
      "learning_rate": 9.072822346635734e-06,
      "loss": 0.7957,
      "step": 7673
    },
    {
      "epoch": 0.22111450469659424,
      "grad_norm": 0.8921946883201599,
      "learning_rate": 9.072551660949558e-06,
      "loss": 0.7615,
      "step": 7674
    },
    {
      "epoch": 0.22114331815824353,
      "grad_norm": 0.9107450246810913,
      "learning_rate": 9.07228093979553e-06,
      "loss": 0.768,
      "step": 7675
    },
    {
      "epoch": 0.22117213161989283,
      "grad_norm": 0.9477900266647339,
      "learning_rate": 9.072010183176004e-06,
      "loss": 0.7713,
      "step": 7676
    },
    {
      "epoch": 0.2212009450815421,
      "grad_norm": 0.9512028694152832,
      "learning_rate": 9.07173939109334e-06,
      "loss": 0.7734,
      "step": 7677
    },
    {
      "epoch": 0.22122975854319138,
      "grad_norm": 0.9874090552330017,
      "learning_rate": 9.071468563549894e-06,
      "loss": 0.7946,
      "step": 7678
    },
    {
      "epoch": 0.22125857200484067,
      "grad_norm": 0.9822656512260437,
      "learning_rate": 9.071197700548029e-06,
      "loss": 0.7788,
      "step": 7679
    },
    {
      "epoch": 0.22128738546648993,
      "grad_norm": 0.9150013327598572,
      "learning_rate": 9.070926802090102e-06,
      "loss": 0.7717,
      "step": 7680
    },
    {
      "epoch": 0.22131619892813922,
      "grad_norm": 0.9461588859558105,
      "learning_rate": 9.070655868178472e-06,
      "loss": 0.7792,
      "step": 7681
    },
    {
      "epoch": 0.22134501238978851,
      "grad_norm": 0.9766154885292053,
      "learning_rate": 9.070384898815497e-06,
      "loss": 0.8037,
      "step": 7682
    },
    {
      "epoch": 0.2213738258514378,
      "grad_norm": 0.9432986378669739,
      "learning_rate": 9.07011389400354e-06,
      "loss": 0.7894,
      "step": 7683
    },
    {
      "epoch": 0.22140263931308707,
      "grad_norm": 0.8903989791870117,
      "learning_rate": 9.069842853744959e-06,
      "loss": 0.7568,
      "step": 7684
    },
    {
      "epoch": 0.22143145277473636,
      "grad_norm": 0.9420040845870972,
      "learning_rate": 9.069571778042115e-06,
      "loss": 0.7514,
      "step": 7685
    },
    {
      "epoch": 0.22146026623638565,
      "grad_norm": 0.9662144184112549,
      "learning_rate": 9.069300666897369e-06,
      "loss": 0.7839,
      "step": 7686
    },
    {
      "epoch": 0.2214890796980349,
      "grad_norm": 0.9818941354751587,
      "learning_rate": 9.069029520313081e-06,
      "loss": 0.7865,
      "step": 7687
    },
    {
      "epoch": 0.2215178931596842,
      "grad_norm": 0.9194151163101196,
      "learning_rate": 9.068758338291617e-06,
      "loss": 0.7626,
      "step": 7688
    },
    {
      "epoch": 0.2215467066213335,
      "grad_norm": 0.9730367064476013,
      "learning_rate": 9.068487120835332e-06,
      "loss": 0.7987,
      "step": 7689
    },
    {
      "epoch": 0.22157552008298276,
      "grad_norm": 0.9288508892059326,
      "learning_rate": 9.068215867946592e-06,
      "loss": 0.7655,
      "step": 7690
    },
    {
      "epoch": 0.22160433354463205,
      "grad_norm": 0.9267982244491577,
      "learning_rate": 9.067944579627758e-06,
      "loss": 0.7907,
      "step": 7691
    },
    {
      "epoch": 0.22163314700628134,
      "grad_norm": 0.894680380821228,
      "learning_rate": 9.067673255881193e-06,
      "loss": 0.7479,
      "step": 7692
    },
    {
      "epoch": 0.22166196046793063,
      "grad_norm": 0.9266699552536011,
      "learning_rate": 9.06740189670926e-06,
      "loss": 0.7762,
      "step": 7693
    },
    {
      "epoch": 0.2216907739295799,
      "grad_norm": 0.9244378805160522,
      "learning_rate": 9.067130502114323e-06,
      "loss": 0.7789,
      "step": 7694
    },
    {
      "epoch": 0.22171958739122918,
      "grad_norm": 0.9545866847038269,
      "learning_rate": 9.066859072098746e-06,
      "loss": 0.7874,
      "step": 7695
    },
    {
      "epoch": 0.22174840085287847,
      "grad_norm": 0.9170622229576111,
      "learning_rate": 9.066587606664892e-06,
      "loss": 0.7696,
      "step": 7696
    },
    {
      "epoch": 0.22177721431452774,
      "grad_norm": 0.9314597845077515,
      "learning_rate": 9.066316105815122e-06,
      "loss": 0.7734,
      "step": 7697
    },
    {
      "epoch": 0.22180602777617703,
      "grad_norm": 0.9102855324745178,
      "learning_rate": 9.066044569551806e-06,
      "loss": 0.7698,
      "step": 7698
    },
    {
      "epoch": 0.22183484123782632,
      "grad_norm": 0.9212512969970703,
      "learning_rate": 9.065772997877307e-06,
      "loss": 0.7802,
      "step": 7699
    },
    {
      "epoch": 0.22186365469947558,
      "grad_norm": 0.9712033271789551,
      "learning_rate": 9.065501390793988e-06,
      "loss": 0.7793,
      "step": 7700
    },
    {
      "epoch": 0.22189246816112487,
      "grad_norm": 0.9391728043556213,
      "learning_rate": 9.065229748304217e-06,
      "loss": 0.8046,
      "step": 7701
    },
    {
      "epoch": 0.22192128162277416,
      "grad_norm": 0.9579712152481079,
      "learning_rate": 9.064958070410356e-06,
      "loss": 0.7544,
      "step": 7702
    },
    {
      "epoch": 0.22195009508442345,
      "grad_norm": 0.9553279876708984,
      "learning_rate": 9.064686357114776e-06,
      "loss": 0.7644,
      "step": 7703
    },
    {
      "epoch": 0.22197890854607272,
      "grad_norm": 0.9638612270355225,
      "learning_rate": 9.06441460841984e-06,
      "loss": 0.787,
      "step": 7704
    },
    {
      "epoch": 0.222007722007722,
      "grad_norm": 0.9499921202659607,
      "learning_rate": 9.064142824327916e-06,
      "loss": 0.7486,
      "step": 7705
    },
    {
      "epoch": 0.2220365354693713,
      "grad_norm": 0.9637762308120728,
      "learning_rate": 9.063871004841371e-06,
      "loss": 0.795,
      "step": 7706
    },
    {
      "epoch": 0.22206534893102056,
      "grad_norm": 0.9180085062980652,
      "learning_rate": 9.063599149962572e-06,
      "loss": 0.7476,
      "step": 7707
    },
    {
      "epoch": 0.22209416239266985,
      "grad_norm": 0.908666729927063,
      "learning_rate": 9.063327259693885e-06,
      "loss": 0.7495,
      "step": 7708
    },
    {
      "epoch": 0.22212297585431914,
      "grad_norm": 0.9121173620223999,
      "learning_rate": 9.063055334037683e-06,
      "loss": 0.7829,
      "step": 7709
    },
    {
      "epoch": 0.22215178931596843,
      "grad_norm": 0.9503437876701355,
      "learning_rate": 9.062783372996326e-06,
      "loss": 0.7719,
      "step": 7710
    },
    {
      "epoch": 0.2221806027776177,
      "grad_norm": 0.9444248080253601,
      "learning_rate": 9.062511376572192e-06,
      "loss": 0.7756,
      "step": 7711
    },
    {
      "epoch": 0.222209416239267,
      "grad_norm": 0.9484056234359741,
      "learning_rate": 9.06223934476764e-06,
      "loss": 0.7791,
      "step": 7712
    },
    {
      "epoch": 0.22223822970091628,
      "grad_norm": 0.947919487953186,
      "learning_rate": 9.061967277585047e-06,
      "loss": 0.7528,
      "step": 7713
    },
    {
      "epoch": 0.22226704316256554,
      "grad_norm": 0.9998189806938171,
      "learning_rate": 9.06169517502678e-06,
      "loss": 0.8053,
      "step": 7714
    },
    {
      "epoch": 0.22229585662421483,
      "grad_norm": 0.8946278691291809,
      "learning_rate": 9.061423037095207e-06,
      "loss": 0.7783,
      "step": 7715
    },
    {
      "epoch": 0.22232467008586412,
      "grad_norm": 0.9701440334320068,
      "learning_rate": 9.061150863792699e-06,
      "loss": 0.7647,
      "step": 7716
    },
    {
      "epoch": 0.2223534835475134,
      "grad_norm": 1.0825749635696411,
      "learning_rate": 9.060878655121627e-06,
      "loss": 0.7495,
      "step": 7717
    },
    {
      "epoch": 0.22238229700916268,
      "grad_norm": 0.9183415174484253,
      "learning_rate": 9.060606411084362e-06,
      "loss": 0.7653,
      "step": 7718
    },
    {
      "epoch": 0.22241111047081197,
      "grad_norm": 0.9679294228553772,
      "learning_rate": 9.060334131683272e-06,
      "loss": 0.7932,
      "step": 7719
    },
    {
      "epoch": 0.22243992393246126,
      "grad_norm": 0.9908983111381531,
      "learning_rate": 9.060061816920733e-06,
      "loss": 0.7783,
      "step": 7720
    },
    {
      "epoch": 0.22246873739411052,
      "grad_norm": 0.9498832821846008,
      "learning_rate": 9.059789466799114e-06,
      "loss": 0.7425,
      "step": 7721
    },
    {
      "epoch": 0.2224975508557598,
      "grad_norm": 0.9397662281990051,
      "learning_rate": 9.059517081320787e-06,
      "loss": 0.7516,
      "step": 7722
    },
    {
      "epoch": 0.2225263643174091,
      "grad_norm": 0.9337829947471619,
      "learning_rate": 9.059244660488122e-06,
      "loss": 0.7671,
      "step": 7723
    },
    {
      "epoch": 0.22255517777905837,
      "grad_norm": 0.9410919547080994,
      "learning_rate": 9.058972204303496e-06,
      "loss": 0.7638,
      "step": 7724
    },
    {
      "epoch": 0.22258399124070766,
      "grad_norm": 0.9074461460113525,
      "learning_rate": 9.05869971276928e-06,
      "loss": 0.7626,
      "step": 7725
    },
    {
      "epoch": 0.22261280470235695,
      "grad_norm": 0.9422440528869629,
      "learning_rate": 9.058427185887847e-06,
      "loss": 0.7658,
      "step": 7726
    },
    {
      "epoch": 0.2226416181640062,
      "grad_norm": 1.0002014636993408,
      "learning_rate": 9.058154623661567e-06,
      "loss": 0.7775,
      "step": 7727
    },
    {
      "epoch": 0.2226704316256555,
      "grad_norm": 0.9688292145729065,
      "learning_rate": 9.057882026092819e-06,
      "loss": 0.7519,
      "step": 7728
    },
    {
      "epoch": 0.2226992450873048,
      "grad_norm": 0.9569203853607178,
      "learning_rate": 9.057609393183976e-06,
      "loss": 0.7567,
      "step": 7729
    },
    {
      "epoch": 0.22272805854895408,
      "grad_norm": 0.9884748458862305,
      "learning_rate": 9.057336724937409e-06,
      "loss": 0.7651,
      "step": 7730
    },
    {
      "epoch": 0.22275687201060335,
      "grad_norm": 0.9799972772598267,
      "learning_rate": 9.057064021355496e-06,
      "loss": 0.7885,
      "step": 7731
    },
    {
      "epoch": 0.22278568547225264,
      "grad_norm": 0.9489534497261047,
      "learning_rate": 9.056791282440612e-06,
      "loss": 0.7666,
      "step": 7732
    },
    {
      "epoch": 0.22281449893390193,
      "grad_norm": 0.9523993730545044,
      "learning_rate": 9.05651850819513e-06,
      "loss": 0.7847,
      "step": 7733
    },
    {
      "epoch": 0.2228433123955512,
      "grad_norm": 1.0221513509750366,
      "learning_rate": 9.056245698621426e-06,
      "loss": 0.7873,
      "step": 7734
    },
    {
      "epoch": 0.22287212585720048,
      "grad_norm": 0.951108455657959,
      "learning_rate": 9.055972853721878e-06,
      "loss": 0.7659,
      "step": 7735
    },
    {
      "epoch": 0.22290093931884977,
      "grad_norm": 0.9595503807067871,
      "learning_rate": 9.05569997349886e-06,
      "loss": 0.7525,
      "step": 7736
    },
    {
      "epoch": 0.22292975278049904,
      "grad_norm": 1.0056589841842651,
      "learning_rate": 9.055427057954752e-06,
      "loss": 0.7701,
      "step": 7737
    },
    {
      "epoch": 0.22295856624214833,
      "grad_norm": 1.038292407989502,
      "learning_rate": 9.055154107091926e-06,
      "loss": 0.7937,
      "step": 7738
    },
    {
      "epoch": 0.22298737970379762,
      "grad_norm": 0.9221970438957214,
      "learning_rate": 9.054881120912762e-06,
      "loss": 0.7719,
      "step": 7739
    },
    {
      "epoch": 0.2230161931654469,
      "grad_norm": 0.9304283261299133,
      "learning_rate": 9.054608099419635e-06,
      "loss": 0.7611,
      "step": 7740
    },
    {
      "epoch": 0.22304500662709617,
      "grad_norm": 1.0494370460510254,
      "learning_rate": 9.054335042614929e-06,
      "loss": 0.7741,
      "step": 7741
    },
    {
      "epoch": 0.22307382008874546,
      "grad_norm": 0.9387048482894897,
      "learning_rate": 9.054061950501014e-06,
      "loss": 0.7857,
      "step": 7742
    },
    {
      "epoch": 0.22310263355039475,
      "grad_norm": 0.9529058933258057,
      "learning_rate": 9.053788823080272e-06,
      "loss": 0.7704,
      "step": 7743
    },
    {
      "epoch": 0.22313144701204402,
      "grad_norm": 0.9485874772071838,
      "learning_rate": 9.053515660355082e-06,
      "loss": 0.767,
      "step": 7744
    },
    {
      "epoch": 0.2231602604736933,
      "grad_norm": 0.9579219818115234,
      "learning_rate": 9.053242462327823e-06,
      "loss": 0.789,
      "step": 7745
    },
    {
      "epoch": 0.2231890739353426,
      "grad_norm": 0.929553210735321,
      "learning_rate": 9.052969229000875e-06,
      "loss": 0.773,
      "step": 7746
    },
    {
      "epoch": 0.2232178873969919,
      "grad_norm": 0.9488157033920288,
      "learning_rate": 9.052695960376615e-06,
      "loss": 0.7439,
      "step": 7747
    },
    {
      "epoch": 0.22324670085864115,
      "grad_norm": 0.967621386051178,
      "learning_rate": 9.052422656457424e-06,
      "loss": 0.8009,
      "step": 7748
    },
    {
      "epoch": 0.22327551432029044,
      "grad_norm": 1.0233373641967773,
      "learning_rate": 9.052149317245684e-06,
      "loss": 0.7944,
      "step": 7749
    },
    {
      "epoch": 0.22330432778193973,
      "grad_norm": 0.9448553323745728,
      "learning_rate": 9.051875942743774e-06,
      "loss": 0.7912,
      "step": 7750
    },
    {
      "epoch": 0.223333141243589,
      "grad_norm": 0.9743431210517883,
      "learning_rate": 9.051602532954074e-06,
      "loss": 0.7793,
      "step": 7751
    },
    {
      "epoch": 0.22336195470523829,
      "grad_norm": 0.9664852619171143,
      "learning_rate": 9.051329087878967e-06,
      "loss": 0.7857,
      "step": 7752
    },
    {
      "epoch": 0.22339076816688758,
      "grad_norm": 0.9832719564437866,
      "learning_rate": 9.051055607520833e-06,
      "loss": 0.7775,
      "step": 7753
    },
    {
      "epoch": 0.22341958162853684,
      "grad_norm": 0.9740322232246399,
      "learning_rate": 9.050782091882055e-06,
      "loss": 0.7643,
      "step": 7754
    },
    {
      "epoch": 0.22344839509018613,
      "grad_norm": 0.9762954711914062,
      "learning_rate": 9.050508540965015e-06,
      "loss": 0.7609,
      "step": 7755
    },
    {
      "epoch": 0.22347720855183542,
      "grad_norm": 0.9386816620826721,
      "learning_rate": 9.050234954772092e-06,
      "loss": 0.783,
      "step": 7756
    },
    {
      "epoch": 0.2235060220134847,
      "grad_norm": 0.9856491088867188,
      "learning_rate": 9.049961333305672e-06,
      "loss": 0.7814,
      "step": 7757
    },
    {
      "epoch": 0.22353483547513398,
      "grad_norm": 0.9216517210006714,
      "learning_rate": 9.049687676568138e-06,
      "loss": 0.7765,
      "step": 7758
    },
    {
      "epoch": 0.22356364893678327,
      "grad_norm": 0.8696157336235046,
      "learning_rate": 9.049413984561873e-06,
      "loss": 0.746,
      "step": 7759
    },
    {
      "epoch": 0.22359246239843256,
      "grad_norm": 0.9219188094139099,
      "learning_rate": 9.04914025728926e-06,
      "loss": 0.7709,
      "step": 7760
    },
    {
      "epoch": 0.22362127586008182,
      "grad_norm": 0.9558112025260925,
      "learning_rate": 9.048866494752682e-06,
      "loss": 0.745,
      "step": 7761
    },
    {
      "epoch": 0.2236500893217311,
      "grad_norm": 0.9025953412055969,
      "learning_rate": 9.048592696954525e-06,
      "loss": 0.7797,
      "step": 7762
    },
    {
      "epoch": 0.2236789027833804,
      "grad_norm": 0.94898521900177,
      "learning_rate": 9.048318863897172e-06,
      "loss": 0.7799,
      "step": 7763
    },
    {
      "epoch": 0.22370771624502966,
      "grad_norm": 0.9403063654899597,
      "learning_rate": 9.04804499558301e-06,
      "loss": 0.7607,
      "step": 7764
    },
    {
      "epoch": 0.22373652970667895,
      "grad_norm": 0.9221758842468262,
      "learning_rate": 9.047771092014421e-06,
      "loss": 0.7732,
      "step": 7765
    },
    {
      "epoch": 0.22376534316832825,
      "grad_norm": 0.9498958587646484,
      "learning_rate": 9.047497153193792e-06,
      "loss": 0.7954,
      "step": 7766
    },
    {
      "epoch": 0.22379415662997754,
      "grad_norm": 1.0211431980133057,
      "learning_rate": 9.047223179123509e-06,
      "loss": 0.7408,
      "step": 7767
    },
    {
      "epoch": 0.2238229700916268,
      "grad_norm": 0.9130333065986633,
      "learning_rate": 9.04694916980596e-06,
      "loss": 0.7645,
      "step": 7768
    },
    {
      "epoch": 0.2238517835532761,
      "grad_norm": 0.9713237881660461,
      "learning_rate": 9.046675125243527e-06,
      "loss": 0.7478,
      "step": 7769
    },
    {
      "epoch": 0.22388059701492538,
      "grad_norm": 0.9658136963844299,
      "learning_rate": 9.0464010454386e-06,
      "loss": 0.7745,
      "step": 7770
    },
    {
      "epoch": 0.22390941047657464,
      "grad_norm": 0.9232286810874939,
      "learning_rate": 9.046126930393563e-06,
      "loss": 0.7785,
      "step": 7771
    },
    {
      "epoch": 0.22393822393822393,
      "grad_norm": 0.9361299872398376,
      "learning_rate": 9.045852780110808e-06,
      "loss": 0.7524,
      "step": 7772
    },
    {
      "epoch": 0.22396703739987323,
      "grad_norm": 0.9524012804031372,
      "learning_rate": 9.045578594592718e-06,
      "loss": 0.7578,
      "step": 7773
    },
    {
      "epoch": 0.22399585086152252,
      "grad_norm": 0.9007628560066223,
      "learning_rate": 9.045304373841682e-06,
      "loss": 0.7697,
      "step": 7774
    },
    {
      "epoch": 0.22402466432317178,
      "grad_norm": 0.9892755150794983,
      "learning_rate": 9.04503011786009e-06,
      "loss": 0.7763,
      "step": 7775
    },
    {
      "epoch": 0.22405347778482107,
      "grad_norm": 0.9156814813613892,
      "learning_rate": 9.044755826650328e-06,
      "loss": 0.7687,
      "step": 7776
    },
    {
      "epoch": 0.22408229124647036,
      "grad_norm": 0.9274227023124695,
      "learning_rate": 9.044481500214786e-06,
      "loss": 0.7859,
      "step": 7777
    },
    {
      "epoch": 0.22411110470811962,
      "grad_norm": 0.8942031264305115,
      "learning_rate": 9.044207138555854e-06,
      "loss": 0.7474,
      "step": 7778
    },
    {
      "epoch": 0.22413991816976891,
      "grad_norm": 0.9157674312591553,
      "learning_rate": 9.04393274167592e-06,
      "loss": 0.7707,
      "step": 7779
    },
    {
      "epoch": 0.2241687316314182,
      "grad_norm": 0.9288368225097656,
      "learning_rate": 9.043658309577374e-06,
      "loss": 0.7664,
      "step": 7780
    },
    {
      "epoch": 0.22419754509306747,
      "grad_norm": 0.9599548578262329,
      "learning_rate": 9.043383842262607e-06,
      "loss": 0.7717,
      "step": 7781
    },
    {
      "epoch": 0.22422635855471676,
      "grad_norm": 0.898490309715271,
      "learning_rate": 9.043109339734008e-06,
      "loss": 0.7648,
      "step": 7782
    },
    {
      "epoch": 0.22425517201636605,
      "grad_norm": 0.9381580352783203,
      "learning_rate": 9.04283480199397e-06,
      "loss": 0.7611,
      "step": 7783
    },
    {
      "epoch": 0.22428398547801534,
      "grad_norm": 0.9490538835525513,
      "learning_rate": 9.042560229044881e-06,
      "loss": 0.7695,
      "step": 7784
    },
    {
      "epoch": 0.2243127989396646,
      "grad_norm": 0.9813317656517029,
      "learning_rate": 9.042285620889134e-06,
      "loss": 0.7692,
      "step": 7785
    },
    {
      "epoch": 0.2243416124013139,
      "grad_norm": 0.9163299798965454,
      "learning_rate": 9.042010977529118e-06,
      "loss": 0.7799,
      "step": 7786
    },
    {
      "epoch": 0.22437042586296319,
      "grad_norm": 0.9186716079711914,
      "learning_rate": 9.04173629896723e-06,
      "loss": 0.7744,
      "step": 7787
    },
    {
      "epoch": 0.22439923932461245,
      "grad_norm": 0.9312511682510376,
      "learning_rate": 9.041461585205857e-06,
      "loss": 0.7637,
      "step": 7788
    },
    {
      "epoch": 0.22442805278626174,
      "grad_norm": 0.9092571139335632,
      "learning_rate": 9.041186836247393e-06,
      "loss": 0.7758,
      "step": 7789
    },
    {
      "epoch": 0.22445686624791103,
      "grad_norm": 0.9803320169448853,
      "learning_rate": 9.040912052094233e-06,
      "loss": 0.7577,
      "step": 7790
    },
    {
      "epoch": 0.2244856797095603,
      "grad_norm": 0.8703638315200806,
      "learning_rate": 9.040637232748768e-06,
      "loss": 0.7462,
      "step": 7791
    },
    {
      "epoch": 0.22451449317120958,
      "grad_norm": 0.9680024981498718,
      "learning_rate": 9.040362378213393e-06,
      "loss": 0.7936,
      "step": 7792
    },
    {
      "epoch": 0.22454330663285887,
      "grad_norm": 0.9450812935829163,
      "learning_rate": 9.040087488490499e-06,
      "loss": 0.772,
      "step": 7793
    },
    {
      "epoch": 0.22457212009450817,
      "grad_norm": 0.916492223739624,
      "learning_rate": 9.039812563582481e-06,
      "loss": 0.7743,
      "step": 7794
    },
    {
      "epoch": 0.22460093355615743,
      "grad_norm": 0.9305776357650757,
      "learning_rate": 9.039537603491735e-06,
      "loss": 0.7828,
      "step": 7795
    },
    {
      "epoch": 0.22462974701780672,
      "grad_norm": 1.013521432876587,
      "learning_rate": 9.039262608220654e-06,
      "loss": 0.7978,
      "step": 7796
    },
    {
      "epoch": 0.224658560479456,
      "grad_norm": 0.9324504137039185,
      "learning_rate": 9.038987577771635e-06,
      "loss": 0.7673,
      "step": 7797
    },
    {
      "epoch": 0.22468737394110527,
      "grad_norm": 0.9371281266212463,
      "learning_rate": 9.03871251214707e-06,
      "loss": 0.7767,
      "step": 7798
    },
    {
      "epoch": 0.22471618740275456,
      "grad_norm": 1.0123947858810425,
      "learning_rate": 9.038437411349356e-06,
      "loss": 0.7743,
      "step": 7799
    },
    {
      "epoch": 0.22474500086440385,
      "grad_norm": 0.9995591044425964,
      "learning_rate": 9.038162275380891e-06,
      "loss": 0.7809,
      "step": 7800
    },
    {
      "epoch": 0.22477381432605315,
      "grad_norm": 0.9206264615058899,
      "learning_rate": 9.037887104244067e-06,
      "loss": 0.7949,
      "step": 7801
    },
    {
      "epoch": 0.2248026277877024,
      "grad_norm": 0.9729763269424438,
      "learning_rate": 9.037611897941284e-06,
      "loss": 0.7773,
      "step": 7802
    },
    {
      "epoch": 0.2248314412493517,
      "grad_norm": 0.9031559824943542,
      "learning_rate": 9.037336656474938e-06,
      "loss": 0.7688,
      "step": 7803
    },
    {
      "epoch": 0.224860254711001,
      "grad_norm": 0.9010152220726013,
      "learning_rate": 9.037061379847424e-06,
      "loss": 0.7539,
      "step": 7804
    },
    {
      "epoch": 0.22488906817265025,
      "grad_norm": 1.0030419826507568,
      "learning_rate": 9.036786068061143e-06,
      "loss": 0.7926,
      "step": 7805
    },
    {
      "epoch": 0.22491788163429954,
      "grad_norm": 0.9318112134933472,
      "learning_rate": 9.03651072111849e-06,
      "loss": 0.7876,
      "step": 7806
    },
    {
      "epoch": 0.22494669509594883,
      "grad_norm": 0.9158919453620911,
      "learning_rate": 9.036235339021863e-06,
      "loss": 0.7569,
      "step": 7807
    },
    {
      "epoch": 0.2249755085575981,
      "grad_norm": 0.9675710201263428,
      "learning_rate": 9.03595992177366e-06,
      "loss": 0.7805,
      "step": 7808
    },
    {
      "epoch": 0.2250043220192474,
      "grad_norm": 0.9543309211730957,
      "learning_rate": 9.035684469376283e-06,
      "loss": 0.7902,
      "step": 7809
    },
    {
      "epoch": 0.22503313548089668,
      "grad_norm": 0.9307059645652771,
      "learning_rate": 9.035408981832127e-06,
      "loss": 0.7793,
      "step": 7810
    },
    {
      "epoch": 0.22506194894254597,
      "grad_norm": 0.9674443006515503,
      "learning_rate": 9.035133459143592e-06,
      "loss": 0.756,
      "step": 7811
    },
    {
      "epoch": 0.22509076240419523,
      "grad_norm": 0.9550279974937439,
      "learning_rate": 9.03485790131308e-06,
      "loss": 0.7673,
      "step": 7812
    },
    {
      "epoch": 0.22511957586584452,
      "grad_norm": 0.9472737908363342,
      "learning_rate": 9.034582308342987e-06,
      "loss": 0.8066,
      "step": 7813
    },
    {
      "epoch": 0.22514838932749381,
      "grad_norm": 0.9161575436592102,
      "learning_rate": 9.034306680235718e-06,
      "loss": 0.7816,
      "step": 7814
    },
    {
      "epoch": 0.22517720278914308,
      "grad_norm": 0.9830116629600525,
      "learning_rate": 9.03403101699367e-06,
      "loss": 0.7437,
      "step": 7815
    },
    {
      "epoch": 0.22520601625079237,
      "grad_norm": 0.9985262155532837,
      "learning_rate": 9.033755318619241e-06,
      "loss": 0.8045,
      "step": 7816
    },
    {
      "epoch": 0.22523482971244166,
      "grad_norm": 0.9139218926429749,
      "learning_rate": 9.033479585114839e-06,
      "loss": 0.7712,
      "step": 7817
    },
    {
      "epoch": 0.22526364317409092,
      "grad_norm": 6.010842800140381,
      "learning_rate": 9.033203816482861e-06,
      "loss": 0.7656,
      "step": 7818
    },
    {
      "epoch": 0.2252924566357402,
      "grad_norm": 0.8908787369728088,
      "learning_rate": 9.032928012725709e-06,
      "loss": 0.7606,
      "step": 7819
    },
    {
      "epoch": 0.2253212700973895,
      "grad_norm": 0.9108152985572815,
      "learning_rate": 9.032652173845787e-06,
      "loss": 0.7514,
      "step": 7820
    },
    {
      "epoch": 0.2253500835590388,
      "grad_norm": 0.9721323847770691,
      "learning_rate": 9.032376299845494e-06,
      "loss": 0.7883,
      "step": 7821
    },
    {
      "epoch": 0.22537889702068806,
      "grad_norm": 0.9514904022216797,
      "learning_rate": 9.032100390727235e-06,
      "loss": 0.7837,
      "step": 7822
    },
    {
      "epoch": 0.22540771048233735,
      "grad_norm": 0.9631690979003906,
      "learning_rate": 9.03182444649341e-06,
      "loss": 0.7475,
      "step": 7823
    },
    {
      "epoch": 0.22543652394398664,
      "grad_norm": 0.9387834072113037,
      "learning_rate": 9.031548467146427e-06,
      "loss": 0.7845,
      "step": 7824
    },
    {
      "epoch": 0.2254653374056359,
      "grad_norm": 0.9385156631469727,
      "learning_rate": 9.031272452688686e-06,
      "loss": 0.7771,
      "step": 7825
    },
    {
      "epoch": 0.2254941508672852,
      "grad_norm": 0.9902865290641785,
      "learning_rate": 9.030996403122593e-06,
      "loss": 0.7803,
      "step": 7826
    },
    {
      "epoch": 0.22552296432893448,
      "grad_norm": 0.9971489310264587,
      "learning_rate": 9.03072031845055e-06,
      "loss": 0.7809,
      "step": 7827
    },
    {
      "epoch": 0.22555177779058377,
      "grad_norm": 0.9992773532867432,
      "learning_rate": 9.030444198674962e-06,
      "loss": 0.768,
      "step": 7828
    },
    {
      "epoch": 0.22558059125223304,
      "grad_norm": 1.0073606967926025,
      "learning_rate": 9.030168043798233e-06,
      "loss": 0.7908,
      "step": 7829
    },
    {
      "epoch": 0.22560940471388233,
      "grad_norm": 0.9926844835281372,
      "learning_rate": 9.029891853822772e-06,
      "loss": 0.7765,
      "step": 7830
    },
    {
      "epoch": 0.22563821817553162,
      "grad_norm": 0.9619130492210388,
      "learning_rate": 9.029615628750978e-06,
      "loss": 0.7824,
      "step": 7831
    },
    {
      "epoch": 0.22566703163718088,
      "grad_norm": 0.931812047958374,
      "learning_rate": 9.029339368585262e-06,
      "loss": 0.7806,
      "step": 7832
    },
    {
      "epoch": 0.22569584509883017,
      "grad_norm": 0.9453991651535034,
      "learning_rate": 9.029063073328029e-06,
      "loss": 0.787,
      "step": 7833
    },
    {
      "epoch": 0.22572465856047946,
      "grad_norm": 0.9615216255187988,
      "learning_rate": 9.028786742981681e-06,
      "loss": 0.783,
      "step": 7834
    },
    {
      "epoch": 0.22575347202212873,
      "grad_norm": 0.9163594841957092,
      "learning_rate": 9.02851037754863e-06,
      "loss": 0.771,
      "step": 7835
    },
    {
      "epoch": 0.22578228548377802,
      "grad_norm": 0.9378842115402222,
      "learning_rate": 9.028233977031281e-06,
      "loss": 0.7675,
      "step": 7836
    },
    {
      "epoch": 0.2258110989454273,
      "grad_norm": 0.9016227126121521,
      "learning_rate": 9.02795754143204e-06,
      "loss": 0.7597,
      "step": 7837
    },
    {
      "epoch": 0.2258399124070766,
      "grad_norm": 0.9200945496559143,
      "learning_rate": 9.027681070753316e-06,
      "loss": 0.7713,
      "step": 7838
    },
    {
      "epoch": 0.22586872586872586,
      "grad_norm": 0.9353227019309998,
      "learning_rate": 9.027404564997515e-06,
      "loss": 0.7952,
      "step": 7839
    },
    {
      "epoch": 0.22589753933037515,
      "grad_norm": 0.9013590216636658,
      "learning_rate": 9.027128024167046e-06,
      "loss": 0.7572,
      "step": 7840
    },
    {
      "epoch": 0.22592635279202444,
      "grad_norm": 0.9441902041435242,
      "learning_rate": 9.026851448264317e-06,
      "loss": 0.7843,
      "step": 7841
    },
    {
      "epoch": 0.2259551662536737,
      "grad_norm": 0.9411628246307373,
      "learning_rate": 9.02657483729174e-06,
      "loss": 0.7585,
      "step": 7842
    },
    {
      "epoch": 0.225983979715323,
      "grad_norm": 0.9237834215164185,
      "learning_rate": 9.02629819125172e-06,
      "loss": 0.7746,
      "step": 7843
    },
    {
      "epoch": 0.2260127931769723,
      "grad_norm": 0.932144045829773,
      "learning_rate": 9.026021510146667e-06,
      "loss": 0.7608,
      "step": 7844
    },
    {
      "epoch": 0.22604160663862155,
      "grad_norm": 0.9217021465301514,
      "learning_rate": 9.025744793978992e-06,
      "loss": 0.7988,
      "step": 7845
    },
    {
      "epoch": 0.22607042010027084,
      "grad_norm": 0.906206488609314,
      "learning_rate": 9.025468042751102e-06,
      "loss": 0.7603,
      "step": 7846
    },
    {
      "epoch": 0.22609923356192013,
      "grad_norm": 0.9283467531204224,
      "learning_rate": 9.025191256465412e-06,
      "loss": 0.7943,
      "step": 7847
    },
    {
      "epoch": 0.22612804702356942,
      "grad_norm": 1.4773906469345093,
      "learning_rate": 9.024914435124329e-06,
      "loss": 0.7586,
      "step": 7848
    },
    {
      "epoch": 0.22615686048521869,
      "grad_norm": 0.918525218963623,
      "learning_rate": 9.024637578730265e-06,
      "loss": 0.7511,
      "step": 7849
    },
    {
      "epoch": 0.22618567394686798,
      "grad_norm": 0.9849377870559692,
      "learning_rate": 9.024360687285629e-06,
      "loss": 0.7899,
      "step": 7850
    },
    {
      "epoch": 0.22621448740851727,
      "grad_norm": 0.9135457873344421,
      "learning_rate": 9.024083760792837e-06,
      "loss": 0.7326,
      "step": 7851
    },
    {
      "epoch": 0.22624330087016653,
      "grad_norm": 0.9315159916877747,
      "learning_rate": 9.023806799254295e-06,
      "loss": 0.7568,
      "step": 7852
    },
    {
      "epoch": 0.22627211433181582,
      "grad_norm": 0.9323741793632507,
      "learning_rate": 9.023529802672421e-06,
      "loss": 0.7786,
      "step": 7853
    },
    {
      "epoch": 0.2263009277934651,
      "grad_norm": 0.9600539803504944,
      "learning_rate": 9.023252771049624e-06,
      "loss": 0.777,
      "step": 7854
    },
    {
      "epoch": 0.2263297412551144,
      "grad_norm": 0.895795464515686,
      "learning_rate": 9.022975704388314e-06,
      "loss": 0.7693,
      "step": 7855
    },
    {
      "epoch": 0.22635855471676367,
      "grad_norm": 0.9384704232215881,
      "learning_rate": 9.022698602690912e-06,
      "loss": 0.7891,
      "step": 7856
    },
    {
      "epoch": 0.22638736817841296,
      "grad_norm": 0.9798151850700378,
      "learning_rate": 9.022421465959823e-06,
      "loss": 0.7884,
      "step": 7857
    },
    {
      "epoch": 0.22641618164006225,
      "grad_norm": 1.1469186544418335,
      "learning_rate": 9.022144294197465e-06,
      "loss": 0.7677,
      "step": 7858
    },
    {
      "epoch": 0.2264449951017115,
      "grad_norm": 0.974747896194458,
      "learning_rate": 9.02186708740625e-06,
      "loss": 0.7667,
      "step": 7859
    },
    {
      "epoch": 0.2264738085633608,
      "grad_norm": 0.8947806358337402,
      "learning_rate": 9.021589845588594e-06,
      "loss": 0.7576,
      "step": 7860
    },
    {
      "epoch": 0.2265026220250101,
      "grad_norm": 0.9792234897613525,
      "learning_rate": 9.021312568746912e-06,
      "loss": 0.7936,
      "step": 7861
    },
    {
      "epoch": 0.22653143548665935,
      "grad_norm": 1.07177734375,
      "learning_rate": 9.021035256883615e-06,
      "loss": 0.7629,
      "step": 7862
    },
    {
      "epoch": 0.22656024894830865,
      "grad_norm": 0.902983546257019,
      "learning_rate": 9.020757910001122e-06,
      "loss": 0.756,
      "step": 7863
    },
    {
      "epoch": 0.22658906240995794,
      "grad_norm": 0.9240947961807251,
      "learning_rate": 9.020480528101846e-06,
      "loss": 0.7759,
      "step": 7864
    },
    {
      "epoch": 0.22661787587160723,
      "grad_norm": 0.9862973690032959,
      "learning_rate": 9.020203111188204e-06,
      "loss": 0.7892,
      "step": 7865
    },
    {
      "epoch": 0.2266466893332565,
      "grad_norm": 0.888302743434906,
      "learning_rate": 9.019925659262612e-06,
      "loss": 0.7588,
      "step": 7866
    },
    {
      "epoch": 0.22667550279490578,
      "grad_norm": 0.9263126850128174,
      "learning_rate": 9.019648172327485e-06,
      "loss": 0.763,
      "step": 7867
    },
    {
      "epoch": 0.22670431625655507,
      "grad_norm": 0.9773089289665222,
      "learning_rate": 9.01937065038524e-06,
      "loss": 0.768,
      "step": 7868
    },
    {
      "epoch": 0.22673312971820433,
      "grad_norm": 0.949500560760498,
      "learning_rate": 9.019093093438296e-06,
      "loss": 0.7651,
      "step": 7869
    },
    {
      "epoch": 0.22676194317985363,
      "grad_norm": 0.895952045917511,
      "learning_rate": 9.018815501489069e-06,
      "loss": 0.7737,
      "step": 7870
    },
    {
      "epoch": 0.22679075664150292,
      "grad_norm": 0.954818844795227,
      "learning_rate": 9.018537874539974e-06,
      "loss": 0.7873,
      "step": 7871
    },
    {
      "epoch": 0.22681957010315218,
      "grad_norm": 1.0281486511230469,
      "learning_rate": 9.018260212593432e-06,
      "loss": 0.7769,
      "step": 7872
    },
    {
      "epoch": 0.22684838356480147,
      "grad_norm": 0.9537624716758728,
      "learning_rate": 9.017982515651862e-06,
      "loss": 0.8008,
      "step": 7873
    },
    {
      "epoch": 0.22687719702645076,
      "grad_norm": 0.9062821865081787,
      "learning_rate": 9.01770478371768e-06,
      "loss": 0.7456,
      "step": 7874
    },
    {
      "epoch": 0.22690601048810005,
      "grad_norm": 0.9754772186279297,
      "learning_rate": 9.017427016793306e-06,
      "loss": 0.7812,
      "step": 7875
    },
    {
      "epoch": 0.22693482394974931,
      "grad_norm": 0.9822823405265808,
      "learning_rate": 9.017149214881159e-06,
      "loss": 0.7959,
      "step": 7876
    },
    {
      "epoch": 0.2269636374113986,
      "grad_norm": 0.8875630497932434,
      "learning_rate": 9.016871377983656e-06,
      "loss": 0.7912,
      "step": 7877
    },
    {
      "epoch": 0.2269924508730479,
      "grad_norm": 1.0019049644470215,
      "learning_rate": 9.016593506103221e-06,
      "loss": 0.7865,
      "step": 7878
    },
    {
      "epoch": 0.22702126433469716,
      "grad_norm": 0.8871150612831116,
      "learning_rate": 9.016315599242269e-06,
      "loss": 0.7464,
      "step": 7879
    },
    {
      "epoch": 0.22705007779634645,
      "grad_norm": 0.9471032619476318,
      "learning_rate": 9.016037657403225e-06,
      "loss": 0.7909,
      "step": 7880
    },
    {
      "epoch": 0.22707889125799574,
      "grad_norm": 0.9789057970046997,
      "learning_rate": 9.015759680588507e-06,
      "loss": 0.7842,
      "step": 7881
    },
    {
      "epoch": 0.22710770471964503,
      "grad_norm": 0.9378454685211182,
      "learning_rate": 9.015481668800537e-06,
      "loss": 0.7886,
      "step": 7882
    },
    {
      "epoch": 0.2271365181812943,
      "grad_norm": 0.9244633913040161,
      "learning_rate": 9.015203622041734e-06,
      "loss": 0.7329,
      "step": 7883
    },
    {
      "epoch": 0.22716533164294359,
      "grad_norm": 0.9595905542373657,
      "learning_rate": 9.014925540314523e-06,
      "loss": 0.7631,
      "step": 7884
    },
    {
      "epoch": 0.22719414510459288,
      "grad_norm": 0.9035458564758301,
      "learning_rate": 9.014647423621323e-06,
      "loss": 0.7816,
      "step": 7885
    },
    {
      "epoch": 0.22722295856624214,
      "grad_norm": 0.8977305293083191,
      "learning_rate": 9.014369271964558e-06,
      "loss": 0.7657,
      "step": 7886
    },
    {
      "epoch": 0.22725177202789143,
      "grad_norm": 0.9772021174430847,
      "learning_rate": 9.01409108534665e-06,
      "loss": 0.7523,
      "step": 7887
    },
    {
      "epoch": 0.22728058548954072,
      "grad_norm": 0.908798336982727,
      "learning_rate": 9.01381286377002e-06,
      "loss": 0.7595,
      "step": 7888
    },
    {
      "epoch": 0.22730939895118998,
      "grad_norm": 0.9400907158851624,
      "learning_rate": 9.013534607237094e-06,
      "loss": 0.8033,
      "step": 7889
    },
    {
      "epoch": 0.22733821241283927,
      "grad_norm": 0.8947681188583374,
      "learning_rate": 9.013256315750291e-06,
      "loss": 0.7745,
      "step": 7890
    },
    {
      "epoch": 0.22736702587448857,
      "grad_norm": 0.8841879963874817,
      "learning_rate": 9.012977989312038e-06,
      "loss": 0.8002,
      "step": 7891
    },
    {
      "epoch": 0.22739583933613786,
      "grad_norm": 0.936237633228302,
      "learning_rate": 9.01269962792476e-06,
      "loss": 0.748,
      "step": 7892
    },
    {
      "epoch": 0.22742465279778712,
      "grad_norm": 0.9206035733222961,
      "learning_rate": 9.012421231590877e-06,
      "loss": 0.7637,
      "step": 7893
    },
    {
      "epoch": 0.2274534662594364,
      "grad_norm": 0.9005721807479858,
      "learning_rate": 9.012142800312817e-06,
      "loss": 0.7686,
      "step": 7894
    },
    {
      "epoch": 0.2274822797210857,
      "grad_norm": 0.8968114256858826,
      "learning_rate": 9.011864334093006e-06,
      "loss": 0.7975,
      "step": 7895
    },
    {
      "epoch": 0.22751109318273496,
      "grad_norm": 0.8901791572570801,
      "learning_rate": 9.011585832933865e-06,
      "loss": 0.7565,
      "step": 7896
    },
    {
      "epoch": 0.22753990664438425,
      "grad_norm": 0.9258314967155457,
      "learning_rate": 9.011307296837821e-06,
      "loss": 0.7466,
      "step": 7897
    },
    {
      "epoch": 0.22756872010603355,
      "grad_norm": 0.9124671220779419,
      "learning_rate": 9.011028725807302e-06,
      "loss": 0.7645,
      "step": 7898
    },
    {
      "epoch": 0.2275975335676828,
      "grad_norm": 0.9465133547782898,
      "learning_rate": 9.010750119844732e-06,
      "loss": 0.7949,
      "step": 7899
    },
    {
      "epoch": 0.2276263470293321,
      "grad_norm": 1.642926573753357,
      "learning_rate": 9.010471478952536e-06,
      "loss": 0.7601,
      "step": 7900
    },
    {
      "epoch": 0.2276551604909814,
      "grad_norm": 0.9048080444335938,
      "learning_rate": 9.010192803133144e-06,
      "loss": 0.7893,
      "step": 7901
    },
    {
      "epoch": 0.22768397395263068,
      "grad_norm": 0.9152262806892395,
      "learning_rate": 9.00991409238898e-06,
      "loss": 0.7573,
      "step": 7902
    },
    {
      "epoch": 0.22771278741427994,
      "grad_norm": 0.9384397268295288,
      "learning_rate": 9.009635346722474e-06,
      "loss": 0.7577,
      "step": 7903
    },
    {
      "epoch": 0.22774160087592923,
      "grad_norm": 0.918423056602478,
      "learning_rate": 9.009356566136051e-06,
      "loss": 0.7796,
      "step": 7904
    },
    {
      "epoch": 0.22777041433757853,
      "grad_norm": 0.9073989987373352,
      "learning_rate": 9.009077750632143e-06,
      "loss": 0.7599,
      "step": 7905
    },
    {
      "epoch": 0.2277992277992278,
      "grad_norm": 0.9077218174934387,
      "learning_rate": 9.008798900213174e-06,
      "loss": 0.7795,
      "step": 7906
    },
    {
      "epoch": 0.22782804126087708,
      "grad_norm": 0.9305132031440735,
      "learning_rate": 9.008520014881573e-06,
      "loss": 0.7708,
      "step": 7907
    },
    {
      "epoch": 0.22785685472252637,
      "grad_norm": 0.9587551951408386,
      "learning_rate": 9.008241094639769e-06,
      "loss": 0.7549,
      "step": 7908
    },
    {
      "epoch": 0.22788566818417566,
      "grad_norm": 0.9680165648460388,
      "learning_rate": 9.007962139490194e-06,
      "loss": 0.7948,
      "step": 7909
    },
    {
      "epoch": 0.22791448164582492,
      "grad_norm": 0.917276918888092,
      "learning_rate": 9.007683149435274e-06,
      "loss": 0.7801,
      "step": 7910
    },
    {
      "epoch": 0.22794329510747421,
      "grad_norm": 0.9231658577919006,
      "learning_rate": 9.00740412447744e-06,
      "loss": 0.7714,
      "step": 7911
    },
    {
      "epoch": 0.2279721085691235,
      "grad_norm": 0.9656897187232971,
      "learning_rate": 9.007125064619123e-06,
      "loss": 0.7537,
      "step": 7912
    },
    {
      "epoch": 0.22800092203077277,
      "grad_norm": 0.8976654410362244,
      "learning_rate": 9.006845969862751e-06,
      "loss": 0.7733,
      "step": 7913
    },
    {
      "epoch": 0.22802973549242206,
      "grad_norm": 0.9548470973968506,
      "learning_rate": 9.006566840210757e-06,
      "loss": 0.7708,
      "step": 7914
    },
    {
      "epoch": 0.22805854895407135,
      "grad_norm": 0.9937644600868225,
      "learning_rate": 9.00628767566557e-06,
      "loss": 0.7791,
      "step": 7915
    },
    {
      "epoch": 0.2280873624157206,
      "grad_norm": 0.9025373458862305,
      "learning_rate": 9.006008476229624e-06,
      "loss": 0.7704,
      "step": 7916
    },
    {
      "epoch": 0.2281161758773699,
      "grad_norm": 0.9868463277816772,
      "learning_rate": 9.005729241905346e-06,
      "loss": 0.7702,
      "step": 7917
    },
    {
      "epoch": 0.2281449893390192,
      "grad_norm": 0.9312574863433838,
      "learning_rate": 9.005449972695173e-06,
      "loss": 0.7498,
      "step": 7918
    },
    {
      "epoch": 0.22817380280066848,
      "grad_norm": 0.9710062742233276,
      "learning_rate": 9.005170668601533e-06,
      "loss": 0.7665,
      "step": 7919
    },
    {
      "epoch": 0.22820261626231775,
      "grad_norm": 0.9312459230422974,
      "learning_rate": 9.004891329626862e-06,
      "loss": 0.7607,
      "step": 7920
    },
    {
      "epoch": 0.22823142972396704,
      "grad_norm": 0.896003782749176,
      "learning_rate": 9.004611955773589e-06,
      "loss": 0.7738,
      "step": 7921
    },
    {
      "epoch": 0.22826024318561633,
      "grad_norm": 0.9348578453063965,
      "learning_rate": 9.00433254704415e-06,
      "loss": 0.7675,
      "step": 7922
    },
    {
      "epoch": 0.2282890566472656,
      "grad_norm": 0.9477894902229309,
      "learning_rate": 9.004053103440977e-06,
      "loss": 0.7734,
      "step": 7923
    },
    {
      "epoch": 0.22831787010891488,
      "grad_norm": 0.9393057823181152,
      "learning_rate": 9.003773624966504e-06,
      "loss": 0.7683,
      "step": 7924
    },
    {
      "epoch": 0.22834668357056417,
      "grad_norm": 0.9161579012870789,
      "learning_rate": 9.003494111623165e-06,
      "loss": 0.7944,
      "step": 7925
    },
    {
      "epoch": 0.22837549703221344,
      "grad_norm": 0.9841199517250061,
      "learning_rate": 9.003214563413396e-06,
      "loss": 0.7806,
      "step": 7926
    },
    {
      "epoch": 0.22840431049386273,
      "grad_norm": 0.9132088422775269,
      "learning_rate": 9.002934980339629e-06,
      "loss": 0.7841,
      "step": 7927
    },
    {
      "epoch": 0.22843312395551202,
      "grad_norm": 0.9148603081703186,
      "learning_rate": 9.002655362404298e-06,
      "loss": 0.7803,
      "step": 7928
    },
    {
      "epoch": 0.2284619374171613,
      "grad_norm": 0.9597634673118591,
      "learning_rate": 9.002375709609841e-06,
      "loss": 0.7742,
      "step": 7929
    },
    {
      "epoch": 0.22849075087881057,
      "grad_norm": 0.8875986933708191,
      "learning_rate": 9.002096021958693e-06,
      "loss": 0.7933,
      "step": 7930
    },
    {
      "epoch": 0.22851956434045986,
      "grad_norm": 0.9331899285316467,
      "learning_rate": 9.00181629945329e-06,
      "loss": 0.7812,
      "step": 7931
    },
    {
      "epoch": 0.22854837780210915,
      "grad_norm": 0.9174995422363281,
      "learning_rate": 9.001536542096065e-06,
      "loss": 0.7712,
      "step": 7932
    },
    {
      "epoch": 0.22857719126375842,
      "grad_norm": 0.8840512037277222,
      "learning_rate": 9.00125674988946e-06,
      "loss": 0.7792,
      "step": 7933
    },
    {
      "epoch": 0.2286060047254077,
      "grad_norm": 0.8863388299942017,
      "learning_rate": 9.000976922835906e-06,
      "loss": 0.7647,
      "step": 7934
    },
    {
      "epoch": 0.228634818187057,
      "grad_norm": 0.9129393696784973,
      "learning_rate": 9.000697060937844e-06,
      "loss": 0.7812,
      "step": 7935
    },
    {
      "epoch": 0.2286636316487063,
      "grad_norm": 0.931024432182312,
      "learning_rate": 9.000417164197707e-06,
      "loss": 0.7902,
      "step": 7936
    },
    {
      "epoch": 0.22869244511035555,
      "grad_norm": 0.9169915318489075,
      "learning_rate": 9.00013723261794e-06,
      "loss": 0.754,
      "step": 7937
    },
    {
      "epoch": 0.22872125857200484,
      "grad_norm": 0.9102880358695984,
      "learning_rate": 8.999857266200974e-06,
      "loss": 0.7633,
      "step": 7938
    },
    {
      "epoch": 0.22875007203365413,
      "grad_norm": 3.722555160522461,
      "learning_rate": 8.99957726494925e-06,
      "loss": 0.8046,
      "step": 7939
    },
    {
      "epoch": 0.2287788854953034,
      "grad_norm": 0.9210695624351501,
      "learning_rate": 8.999297228865204e-06,
      "loss": 0.788,
      "step": 7940
    },
    {
      "epoch": 0.2288076989569527,
      "grad_norm": 0.896091103553772,
      "learning_rate": 8.99901715795128e-06,
      "loss": 0.7556,
      "step": 7941
    },
    {
      "epoch": 0.22883651241860198,
      "grad_norm": 0.8995839357376099,
      "learning_rate": 8.998737052209912e-06,
      "loss": 0.7743,
      "step": 7942
    },
    {
      "epoch": 0.22886532588025124,
      "grad_norm": 0.9388149976730347,
      "learning_rate": 8.998456911643543e-06,
      "loss": 0.7938,
      "step": 7943
    },
    {
      "epoch": 0.22889413934190053,
      "grad_norm": 0.9177637696266174,
      "learning_rate": 8.99817673625461e-06,
      "loss": 0.7698,
      "step": 7944
    },
    {
      "epoch": 0.22892295280354982,
      "grad_norm": 0.9441898465156555,
      "learning_rate": 8.997896526045556e-06,
      "loss": 0.7577,
      "step": 7945
    },
    {
      "epoch": 0.2289517662651991,
      "grad_norm": 0.92855304479599,
      "learning_rate": 8.997616281018817e-06,
      "loss": 0.7764,
      "step": 7946
    },
    {
      "epoch": 0.22898057972684838,
      "grad_norm": 0.9186990261077881,
      "learning_rate": 8.997336001176838e-06,
      "loss": 0.7609,
      "step": 7947
    },
    {
      "epoch": 0.22900939318849767,
      "grad_norm": 0.9504530429840088,
      "learning_rate": 8.997055686522058e-06,
      "loss": 0.783,
      "step": 7948
    },
    {
      "epoch": 0.22903820665014696,
      "grad_norm": 0.9556378722190857,
      "learning_rate": 8.99677533705692e-06,
      "loss": 0.7575,
      "step": 7949
    },
    {
      "epoch": 0.22906702011179622,
      "grad_norm": 0.9474332332611084,
      "learning_rate": 8.996494952783863e-06,
      "loss": 0.7824,
      "step": 7950
    },
    {
      "epoch": 0.2290958335734455,
      "grad_norm": 0.9399465322494507,
      "learning_rate": 8.99621453370533e-06,
      "loss": 0.7819,
      "step": 7951
    },
    {
      "epoch": 0.2291246470350948,
      "grad_norm": 0.9295064210891724,
      "learning_rate": 8.995934079823763e-06,
      "loss": 0.7777,
      "step": 7952
    },
    {
      "epoch": 0.22915346049674407,
      "grad_norm": 0.9627864956855774,
      "learning_rate": 8.995653591141604e-06,
      "loss": 0.7648,
      "step": 7953
    },
    {
      "epoch": 0.22918227395839336,
      "grad_norm": 0.9038231372833252,
      "learning_rate": 8.995373067661297e-06,
      "loss": 0.7427,
      "step": 7954
    },
    {
      "epoch": 0.22921108742004265,
      "grad_norm": 0.9079709053039551,
      "learning_rate": 8.995092509385285e-06,
      "loss": 0.7753,
      "step": 7955
    },
    {
      "epoch": 0.22923990088169194,
      "grad_norm": 0.9222890734672546,
      "learning_rate": 8.994811916316011e-06,
      "loss": 0.7817,
      "step": 7956
    },
    {
      "epoch": 0.2292687143433412,
      "grad_norm": 0.8880943655967712,
      "learning_rate": 8.994531288455918e-06,
      "loss": 0.7757,
      "step": 7957
    },
    {
      "epoch": 0.2292975278049905,
      "grad_norm": 0.9665322303771973,
      "learning_rate": 8.99425062580745e-06,
      "loss": 0.7773,
      "step": 7958
    },
    {
      "epoch": 0.22932634126663978,
      "grad_norm": 0.9330721497535706,
      "learning_rate": 8.993969928373053e-06,
      "loss": 0.749,
      "step": 7959
    },
    {
      "epoch": 0.22935515472828905,
      "grad_norm": 0.8939927220344543,
      "learning_rate": 8.99368919615517e-06,
      "loss": 0.774,
      "step": 7960
    },
    {
      "epoch": 0.22938396818993834,
      "grad_norm": 0.9238367676734924,
      "learning_rate": 8.993408429156246e-06,
      "loss": 0.7717,
      "step": 7961
    },
    {
      "epoch": 0.22941278165158763,
      "grad_norm": 0.9344233274459839,
      "learning_rate": 8.993127627378728e-06,
      "loss": 0.7701,
      "step": 7962
    },
    {
      "epoch": 0.22944159511323692,
      "grad_norm": 0.9399198293685913,
      "learning_rate": 8.992846790825058e-06,
      "loss": 0.7754,
      "step": 7963
    },
    {
      "epoch": 0.22947040857488618,
      "grad_norm": 0.8981267809867859,
      "learning_rate": 8.992565919497685e-06,
      "loss": 0.8027,
      "step": 7964
    },
    {
      "epoch": 0.22949922203653547,
      "grad_norm": 0.9528573155403137,
      "learning_rate": 8.992285013399054e-06,
      "loss": 0.7885,
      "step": 7965
    },
    {
      "epoch": 0.22952803549818476,
      "grad_norm": 0.9615246653556824,
      "learning_rate": 8.99200407253161e-06,
      "loss": 0.7746,
      "step": 7966
    },
    {
      "epoch": 0.22955684895983403,
      "grad_norm": 0.9043678045272827,
      "learning_rate": 8.991723096897801e-06,
      "loss": 0.7471,
      "step": 7967
    },
    {
      "epoch": 0.22958566242148332,
      "grad_norm": 0.9560022354125977,
      "learning_rate": 8.991442086500077e-06,
      "loss": 0.7407,
      "step": 7968
    },
    {
      "epoch": 0.2296144758831326,
      "grad_norm": 0.9127838015556335,
      "learning_rate": 8.991161041340879e-06,
      "loss": 0.787,
      "step": 7969
    },
    {
      "epoch": 0.22964328934478187,
      "grad_norm": 0.9538013935089111,
      "learning_rate": 8.990879961422661e-06,
      "loss": 0.7672,
      "step": 7970
    },
    {
      "epoch": 0.22967210280643116,
      "grad_norm": 0.8916583061218262,
      "learning_rate": 8.990598846747866e-06,
      "loss": 0.7797,
      "step": 7971
    },
    {
      "epoch": 0.22970091626808045,
      "grad_norm": 0.9212860465049744,
      "learning_rate": 8.990317697318945e-06,
      "loss": 0.7757,
      "step": 7972
    },
    {
      "epoch": 0.22972972972972974,
      "grad_norm": 0.9540539383888245,
      "learning_rate": 8.990036513138345e-06,
      "loss": 0.7984,
      "step": 7973
    },
    {
      "epoch": 0.229758543191379,
      "grad_norm": 0.9054445624351501,
      "learning_rate": 8.989755294208517e-06,
      "loss": 0.7602,
      "step": 7974
    },
    {
      "epoch": 0.2297873566530283,
      "grad_norm": 0.9118028879165649,
      "learning_rate": 8.989474040531907e-06,
      "loss": 0.7535,
      "step": 7975
    },
    {
      "epoch": 0.2298161701146776,
      "grad_norm": 0.9365689754486084,
      "learning_rate": 8.989192752110967e-06,
      "loss": 0.7859,
      "step": 7976
    },
    {
      "epoch": 0.22984498357632685,
      "grad_norm": 1.0136102437973022,
      "learning_rate": 8.988911428948147e-06,
      "loss": 0.8045,
      "step": 7977
    },
    {
      "epoch": 0.22987379703797614,
      "grad_norm": 0.9030413627624512,
      "learning_rate": 8.988630071045892e-06,
      "loss": 0.7617,
      "step": 7978
    },
    {
      "epoch": 0.22990261049962543,
      "grad_norm": 0.8952848315238953,
      "learning_rate": 8.98834867840666e-06,
      "loss": 0.7534,
      "step": 7979
    },
    {
      "epoch": 0.2299314239612747,
      "grad_norm": 1.6611706018447876,
      "learning_rate": 8.988067251032897e-06,
      "loss": 0.7698,
      "step": 7980
    },
    {
      "epoch": 0.22996023742292399,
      "grad_norm": 0.9787076711654663,
      "learning_rate": 8.987785788927054e-06,
      "loss": 0.7634,
      "step": 7981
    },
    {
      "epoch": 0.22998905088457328,
      "grad_norm": 0.9218622446060181,
      "learning_rate": 8.987504292091583e-06,
      "loss": 0.7598,
      "step": 7982
    },
    {
      "epoch": 0.23001786434622257,
      "grad_norm": 0.9705833792686462,
      "learning_rate": 8.987222760528938e-06,
      "loss": 0.7911,
      "step": 7983
    },
    {
      "epoch": 0.23004667780787183,
      "grad_norm": 0.8858551383018494,
      "learning_rate": 8.986941194241566e-06,
      "loss": 0.7789,
      "step": 7984
    },
    {
      "epoch": 0.23007549126952112,
      "grad_norm": 0.9207566976547241,
      "learning_rate": 8.986659593231923e-06,
      "loss": 0.7785,
      "step": 7985
    },
    {
      "epoch": 0.2301043047311704,
      "grad_norm": 0.9409076571464539,
      "learning_rate": 8.986377957502459e-06,
      "loss": 0.7773,
      "step": 7986
    },
    {
      "epoch": 0.23013311819281967,
      "grad_norm": 0.9515045285224915,
      "learning_rate": 8.986096287055628e-06,
      "loss": 0.7783,
      "step": 7987
    },
    {
      "epoch": 0.23016193165446897,
      "grad_norm": 0.9725731015205383,
      "learning_rate": 8.985814581893884e-06,
      "loss": 0.7882,
      "step": 7988
    },
    {
      "epoch": 0.23019074511611826,
      "grad_norm": 0.8759579658508301,
      "learning_rate": 8.98553284201968e-06,
      "loss": 0.7536,
      "step": 7989
    },
    {
      "epoch": 0.23021955857776755,
      "grad_norm": 0.956906795501709,
      "learning_rate": 8.985251067435467e-06,
      "loss": 0.7644,
      "step": 7990
    },
    {
      "epoch": 0.2302483720394168,
      "grad_norm": 0.9806979298591614,
      "learning_rate": 8.9849692581437e-06,
      "loss": 0.7978,
      "step": 7991
    },
    {
      "epoch": 0.2302771855010661,
      "grad_norm": 0.9172583222389221,
      "learning_rate": 8.984687414146837e-06,
      "loss": 0.7629,
      "step": 7992
    },
    {
      "epoch": 0.2303059989627154,
      "grad_norm": 0.9556735754013062,
      "learning_rate": 8.98440553544733e-06,
      "loss": 0.7881,
      "step": 7993
    },
    {
      "epoch": 0.23033481242436465,
      "grad_norm": 0.9232205152511597,
      "learning_rate": 8.984123622047632e-06,
      "loss": 0.774,
      "step": 7994
    },
    {
      "epoch": 0.23036362588601395,
      "grad_norm": 0.8568091988563538,
      "learning_rate": 8.983841673950202e-06,
      "loss": 0.7499,
      "step": 7995
    },
    {
      "epoch": 0.23039243934766324,
      "grad_norm": 0.9093139171600342,
      "learning_rate": 8.98355969115749e-06,
      "loss": 0.7774,
      "step": 7996
    },
    {
      "epoch": 0.2304212528093125,
      "grad_norm": 0.9285745024681091,
      "learning_rate": 8.983277673671959e-06,
      "loss": 0.7645,
      "step": 7997
    },
    {
      "epoch": 0.2304500662709618,
      "grad_norm": 0.8977891802787781,
      "learning_rate": 8.98299562149606e-06,
      "loss": 0.7345,
      "step": 7998
    },
    {
      "epoch": 0.23047887973261108,
      "grad_norm": 0.8943876624107361,
      "learning_rate": 8.98271353463225e-06,
      "loss": 0.7517,
      "step": 7999
    },
    {
      "epoch": 0.23050769319426037,
      "grad_norm": 0.9610622525215149,
      "learning_rate": 8.982431413082986e-06,
      "loss": 0.8018,
      "step": 8000
    },
    {
      "epoch": 0.23053650665590963,
      "grad_norm": 0.9313956499099731,
      "learning_rate": 8.982149256850726e-06,
      "loss": 0.7873,
      "step": 8001
    },
    {
      "epoch": 0.23056532011755893,
      "grad_norm": 0.933961808681488,
      "learning_rate": 8.981867065937926e-06,
      "loss": 0.7657,
      "step": 8002
    },
    {
      "epoch": 0.23059413357920822,
      "grad_norm": 0.9221268892288208,
      "learning_rate": 8.981584840347047e-06,
      "loss": 0.7883,
      "step": 8003
    },
    {
      "epoch": 0.23062294704085748,
      "grad_norm": 0.9056411981582642,
      "learning_rate": 8.981302580080541e-06,
      "loss": 0.7553,
      "step": 8004
    },
    {
      "epoch": 0.23065176050250677,
      "grad_norm": 0.9102274179458618,
      "learning_rate": 8.98102028514087e-06,
      "loss": 0.762,
      "step": 8005
    },
    {
      "epoch": 0.23068057396415606,
      "grad_norm": 0.9518870711326599,
      "learning_rate": 8.980737955530492e-06,
      "loss": 0.7994,
      "step": 8006
    },
    {
      "epoch": 0.23070938742580532,
      "grad_norm": 0.9593898057937622,
      "learning_rate": 8.980455591251867e-06,
      "loss": 0.7723,
      "step": 8007
    },
    {
      "epoch": 0.23073820088745461,
      "grad_norm": 0.9225140810012817,
      "learning_rate": 8.98017319230745e-06,
      "loss": 0.7856,
      "step": 8008
    },
    {
      "epoch": 0.2307670143491039,
      "grad_norm": 1.0021859407424927,
      "learning_rate": 8.979890758699706e-06,
      "loss": 0.7593,
      "step": 8009
    },
    {
      "epoch": 0.2307958278107532,
      "grad_norm": 0.9084057807922363,
      "learning_rate": 8.97960829043109e-06,
      "loss": 0.7747,
      "step": 8010
    },
    {
      "epoch": 0.23082464127240246,
      "grad_norm": 0.9443413019180298,
      "learning_rate": 8.979325787504065e-06,
      "loss": 0.7819,
      "step": 8011
    },
    {
      "epoch": 0.23085345473405175,
      "grad_norm": 0.9824950098991394,
      "learning_rate": 8.97904324992109e-06,
      "loss": 0.7696,
      "step": 8012
    },
    {
      "epoch": 0.23088226819570104,
      "grad_norm": 0.9289856553077698,
      "learning_rate": 8.978760677684627e-06,
      "loss": 0.7516,
      "step": 8013
    },
    {
      "epoch": 0.2309110816573503,
      "grad_norm": 0.9373820424079895,
      "learning_rate": 8.978478070797133e-06,
      "loss": 0.7656,
      "step": 8014
    },
    {
      "epoch": 0.2309398951189996,
      "grad_norm": 0.9037526845932007,
      "learning_rate": 8.978195429261076e-06,
      "loss": 0.7656,
      "step": 8015
    },
    {
      "epoch": 0.23096870858064888,
      "grad_norm": 0.9740498661994934,
      "learning_rate": 8.977912753078911e-06,
      "loss": 0.773,
      "step": 8016
    },
    {
      "epoch": 0.23099752204229815,
      "grad_norm": 0.9278327226638794,
      "learning_rate": 8.977630042253102e-06,
      "loss": 0.7759,
      "step": 8017
    },
    {
      "epoch": 0.23102633550394744,
      "grad_norm": 0.9216468334197998,
      "learning_rate": 8.977347296786113e-06,
      "loss": 0.7853,
      "step": 8018
    },
    {
      "epoch": 0.23105514896559673,
      "grad_norm": 0.9315834045410156,
      "learning_rate": 8.977064516680405e-06,
      "loss": 0.7732,
      "step": 8019
    },
    {
      "epoch": 0.23108396242724602,
      "grad_norm": 0.914943277835846,
      "learning_rate": 8.97678170193844e-06,
      "loss": 0.7936,
      "step": 8020
    },
    {
      "epoch": 0.23111277588889528,
      "grad_norm": 0.9459089636802673,
      "learning_rate": 8.97649885256268e-06,
      "loss": 0.7539,
      "step": 8021
    },
    {
      "epoch": 0.23114158935054457,
      "grad_norm": 0.9555687308311462,
      "learning_rate": 8.976215968555594e-06,
      "loss": 0.7872,
      "step": 8022
    },
    {
      "epoch": 0.23117040281219386,
      "grad_norm": 0.9315853714942932,
      "learning_rate": 8.975933049919638e-06,
      "loss": 0.7489,
      "step": 8023
    },
    {
      "epoch": 0.23119921627384313,
      "grad_norm": 0.9821494221687317,
      "learning_rate": 8.975650096657282e-06,
      "loss": 0.7596,
      "step": 8024
    },
    {
      "epoch": 0.23122802973549242,
      "grad_norm": 0.9087659120559692,
      "learning_rate": 8.975367108770986e-06,
      "loss": 0.763,
      "step": 8025
    },
    {
      "epoch": 0.2312568431971417,
      "grad_norm": 0.9963256120681763,
      "learning_rate": 8.975084086263217e-06,
      "loss": 0.7735,
      "step": 8026
    },
    {
      "epoch": 0.231285656658791,
      "grad_norm": 1.0476491451263428,
      "learning_rate": 8.97480102913644e-06,
      "loss": 0.7988,
      "step": 8027
    },
    {
      "epoch": 0.23131447012044026,
      "grad_norm": 0.9701735377311707,
      "learning_rate": 8.974517937393118e-06,
      "loss": 0.7524,
      "step": 8028
    },
    {
      "epoch": 0.23134328358208955,
      "grad_norm": 0.9782135486602783,
      "learning_rate": 8.974234811035718e-06,
      "loss": 0.8065,
      "step": 8029
    },
    {
      "epoch": 0.23137209704373884,
      "grad_norm": 1.0055099725723267,
      "learning_rate": 8.973951650066706e-06,
      "loss": 0.7818,
      "step": 8030
    },
    {
      "epoch": 0.2314009105053881,
      "grad_norm": 0.9300271272659302,
      "learning_rate": 8.97366845448855e-06,
      "loss": 0.7684,
      "step": 8031
    },
    {
      "epoch": 0.2314297239670374,
      "grad_norm": 0.9099296927452087,
      "learning_rate": 8.97338522430371e-06,
      "loss": 0.774,
      "step": 8032
    },
    {
      "epoch": 0.2314585374286867,
      "grad_norm": 0.9644425511360168,
      "learning_rate": 8.973101959514659e-06,
      "loss": 0.7985,
      "step": 8033
    },
    {
      "epoch": 0.23148735089033595,
      "grad_norm": 0.98565673828125,
      "learning_rate": 8.972818660123861e-06,
      "loss": 0.7639,
      "step": 8034
    },
    {
      "epoch": 0.23151616435198524,
      "grad_norm": 0.9163992404937744,
      "learning_rate": 8.972535326133785e-06,
      "loss": 0.7576,
      "step": 8035
    },
    {
      "epoch": 0.23154497781363453,
      "grad_norm": 0.9419833421707153,
      "learning_rate": 8.972251957546896e-06,
      "loss": 0.7867,
      "step": 8036
    },
    {
      "epoch": 0.23157379127528382,
      "grad_norm": 0.9472910165786743,
      "learning_rate": 8.971968554365661e-06,
      "loss": 0.7823,
      "step": 8037
    },
    {
      "epoch": 0.2316026047369331,
      "grad_norm": 0.9079146385192871,
      "learning_rate": 8.971685116592555e-06,
      "loss": 0.7734,
      "step": 8038
    },
    {
      "epoch": 0.23163141819858238,
      "grad_norm": 0.9528481364250183,
      "learning_rate": 8.97140164423004e-06,
      "loss": 0.7885,
      "step": 8039
    },
    {
      "epoch": 0.23166023166023167,
      "grad_norm": 0.9327308535575867,
      "learning_rate": 8.971118137280585e-06,
      "loss": 0.7728,
      "step": 8040
    },
    {
      "epoch": 0.23168904512188093,
      "grad_norm": 0.9135410189628601,
      "learning_rate": 8.970834595746663e-06,
      "loss": 0.7735,
      "step": 8041
    },
    {
      "epoch": 0.23171785858353022,
      "grad_norm": 0.933705747127533,
      "learning_rate": 8.97055101963074e-06,
      "loss": 0.7743,
      "step": 8042
    },
    {
      "epoch": 0.2317466720451795,
      "grad_norm": 0.9329653382301331,
      "learning_rate": 8.970267408935287e-06,
      "loss": 0.7783,
      "step": 8043
    },
    {
      "epoch": 0.23177548550682878,
      "grad_norm": 0.9179990887641907,
      "learning_rate": 8.969983763662773e-06,
      "loss": 0.7795,
      "step": 8044
    },
    {
      "epoch": 0.23180429896847807,
      "grad_norm": 0.9058977365493774,
      "learning_rate": 8.96970008381567e-06,
      "loss": 0.7652,
      "step": 8045
    },
    {
      "epoch": 0.23183311243012736,
      "grad_norm": 0.997596025466919,
      "learning_rate": 8.969416369396447e-06,
      "loss": 0.7666,
      "step": 8046
    },
    {
      "epoch": 0.23186192589177665,
      "grad_norm": 0.9110532402992249,
      "learning_rate": 8.969132620407574e-06,
      "loss": 0.7626,
      "step": 8047
    },
    {
      "epoch": 0.2318907393534259,
      "grad_norm": 0.8780282735824585,
      "learning_rate": 8.968848836851526e-06,
      "loss": 0.7571,
      "step": 8048
    },
    {
      "epoch": 0.2319195528150752,
      "grad_norm": 0.9455767273902893,
      "learning_rate": 8.96856501873077e-06,
      "loss": 0.76,
      "step": 8049
    },
    {
      "epoch": 0.2319483662767245,
      "grad_norm": 0.9212369322776794,
      "learning_rate": 8.968281166047782e-06,
      "loss": 0.7753,
      "step": 8050
    },
    {
      "epoch": 0.23197717973837376,
      "grad_norm": 0.9436533451080322,
      "learning_rate": 8.967997278805031e-06,
      "loss": 0.7593,
      "step": 8051
    },
    {
      "epoch": 0.23200599320002305,
      "grad_norm": 0.9223506450653076,
      "learning_rate": 8.96771335700499e-06,
      "loss": 0.7801,
      "step": 8052
    },
    {
      "epoch": 0.23203480666167234,
      "grad_norm": 0.9245640635490417,
      "learning_rate": 8.967429400650132e-06,
      "loss": 0.7728,
      "step": 8053
    },
    {
      "epoch": 0.23206362012332163,
      "grad_norm": 0.9029539823532104,
      "learning_rate": 8.96714540974293e-06,
      "loss": 0.7541,
      "step": 8054
    },
    {
      "epoch": 0.2320924335849709,
      "grad_norm": 0.9106793999671936,
      "learning_rate": 8.966861384285857e-06,
      "loss": 0.7644,
      "step": 8055
    },
    {
      "epoch": 0.23212124704662018,
      "grad_norm": 0.9656795859336853,
      "learning_rate": 8.966577324281385e-06,
      "loss": 0.7598,
      "step": 8056
    },
    {
      "epoch": 0.23215006050826947,
      "grad_norm": 0.9418941736221313,
      "learning_rate": 8.966293229731991e-06,
      "loss": 0.7664,
      "step": 8057
    },
    {
      "epoch": 0.23217887396991874,
      "grad_norm": 0.9527870416641235,
      "learning_rate": 8.966009100640148e-06,
      "loss": 0.7881,
      "step": 8058
    },
    {
      "epoch": 0.23220768743156803,
      "grad_norm": 0.955247163772583,
      "learning_rate": 8.965724937008331e-06,
      "loss": 0.7554,
      "step": 8059
    },
    {
      "epoch": 0.23223650089321732,
      "grad_norm": 0.9767088890075684,
      "learning_rate": 8.965440738839013e-06,
      "loss": 0.7461,
      "step": 8060
    },
    {
      "epoch": 0.23226531435486658,
      "grad_norm": 0.8989159464836121,
      "learning_rate": 8.965156506134671e-06,
      "loss": 0.7759,
      "step": 8061
    },
    {
      "epoch": 0.23229412781651587,
      "grad_norm": 0.9345100522041321,
      "learning_rate": 8.96487223889778e-06,
      "loss": 0.7706,
      "step": 8062
    },
    {
      "epoch": 0.23232294127816516,
      "grad_norm": 0.9669122695922852,
      "learning_rate": 8.964587937130814e-06,
      "loss": 0.787,
      "step": 8063
    },
    {
      "epoch": 0.23235175473981445,
      "grad_norm": 0.9375719428062439,
      "learning_rate": 8.964303600836251e-06,
      "loss": 0.7633,
      "step": 8064
    },
    {
      "epoch": 0.23238056820146372,
      "grad_norm": 0.8908656239509583,
      "learning_rate": 8.964019230016565e-06,
      "loss": 0.7574,
      "step": 8065
    },
    {
      "epoch": 0.232409381663113,
      "grad_norm": 0.9662517309188843,
      "learning_rate": 8.963734824674235e-06,
      "loss": 0.7815,
      "step": 8066
    },
    {
      "epoch": 0.2324381951247623,
      "grad_norm": 0.8958591222763062,
      "learning_rate": 8.963450384811738e-06,
      "loss": 0.7529,
      "step": 8067
    },
    {
      "epoch": 0.23246700858641156,
      "grad_norm": 0.8915424942970276,
      "learning_rate": 8.963165910431549e-06,
      "loss": 0.7762,
      "step": 8068
    },
    {
      "epoch": 0.23249582204806085,
      "grad_norm": 0.8771575093269348,
      "learning_rate": 8.962881401536146e-06,
      "loss": 0.7687,
      "step": 8069
    },
    {
      "epoch": 0.23252463550971014,
      "grad_norm": 0.8695765137672424,
      "learning_rate": 8.962596858128008e-06,
      "loss": 0.7633,
      "step": 8070
    },
    {
      "epoch": 0.2325534489713594,
      "grad_norm": 0.9041262269020081,
      "learning_rate": 8.962312280209614e-06,
      "loss": 0.7902,
      "step": 8071
    },
    {
      "epoch": 0.2325822624330087,
      "grad_norm": 0.8918474316596985,
      "learning_rate": 8.962027667783439e-06,
      "loss": 0.7832,
      "step": 8072
    },
    {
      "epoch": 0.232611075894658,
      "grad_norm": 0.9773657321929932,
      "learning_rate": 8.961743020851966e-06,
      "loss": 0.7609,
      "step": 8073
    },
    {
      "epoch": 0.23263988935630728,
      "grad_norm": 0.9279929399490356,
      "learning_rate": 8.96145833941767e-06,
      "loss": 0.8105,
      "step": 8074
    },
    {
      "epoch": 0.23266870281795654,
      "grad_norm": 0.8998839855194092,
      "learning_rate": 8.96117362348303e-06,
      "loss": 0.7824,
      "step": 8075
    },
    {
      "epoch": 0.23269751627960583,
      "grad_norm": 0.8828535676002502,
      "learning_rate": 8.960888873050531e-06,
      "loss": 0.7617,
      "step": 8076
    },
    {
      "epoch": 0.23272632974125512,
      "grad_norm": 0.918982744216919,
      "learning_rate": 8.960604088122646e-06,
      "loss": 0.7448,
      "step": 8077
    },
    {
      "epoch": 0.23275514320290439,
      "grad_norm": 0.878985583782196,
      "learning_rate": 8.96031926870186e-06,
      "loss": 0.7686,
      "step": 8078
    },
    {
      "epoch": 0.23278395666455368,
      "grad_norm": 0.8747260570526123,
      "learning_rate": 8.960034414790652e-06,
      "loss": 0.7812,
      "step": 8079
    },
    {
      "epoch": 0.23281277012620297,
      "grad_norm": 0.8844226598739624,
      "learning_rate": 8.959749526391504e-06,
      "loss": 0.7523,
      "step": 8080
    },
    {
      "epoch": 0.23284158358785226,
      "grad_norm": 0.904434084892273,
      "learning_rate": 8.959464603506895e-06,
      "loss": 0.7336,
      "step": 8081
    },
    {
      "epoch": 0.23287039704950152,
      "grad_norm": 0.9362193942070007,
      "learning_rate": 8.959179646139307e-06,
      "loss": 0.7772,
      "step": 8082
    },
    {
      "epoch": 0.2328992105111508,
      "grad_norm": 0.909112274646759,
      "learning_rate": 8.958894654291222e-06,
      "loss": 0.7732,
      "step": 8083
    },
    {
      "epoch": 0.2329280239728001,
      "grad_norm": 0.9181168079376221,
      "learning_rate": 8.958609627965123e-06,
      "loss": 0.749,
      "step": 8084
    },
    {
      "epoch": 0.23295683743444937,
      "grad_norm": 0.9316290616989136,
      "learning_rate": 8.95832456716349e-06,
      "loss": 0.7451,
      "step": 8085
    },
    {
      "epoch": 0.23298565089609866,
      "grad_norm": 0.8852164149284363,
      "learning_rate": 8.958039471888807e-06,
      "loss": 0.7754,
      "step": 8086
    },
    {
      "epoch": 0.23301446435774795,
      "grad_norm": 0.9024086594581604,
      "learning_rate": 8.957754342143558e-06,
      "loss": 0.7528,
      "step": 8087
    },
    {
      "epoch": 0.2330432778193972,
      "grad_norm": 0.8917440176010132,
      "learning_rate": 8.957469177930223e-06,
      "loss": 0.771,
      "step": 8088
    },
    {
      "epoch": 0.2330720912810465,
      "grad_norm": 0.9304479956626892,
      "learning_rate": 8.95718397925129e-06,
      "loss": 0.7761,
      "step": 8089
    },
    {
      "epoch": 0.2331009047426958,
      "grad_norm": 0.9043879508972168,
      "learning_rate": 8.95689874610924e-06,
      "loss": 0.79,
      "step": 8090
    },
    {
      "epoch": 0.23312971820434508,
      "grad_norm": 0.8794907331466675,
      "learning_rate": 8.956613478506554e-06,
      "loss": 0.765,
      "step": 8091
    },
    {
      "epoch": 0.23315853166599435,
      "grad_norm": 0.9105620384216309,
      "learning_rate": 8.956328176445724e-06,
      "loss": 0.7517,
      "step": 8092
    },
    {
      "epoch": 0.23318734512764364,
      "grad_norm": 0.9215894341468811,
      "learning_rate": 8.956042839929228e-06,
      "loss": 0.7677,
      "step": 8093
    },
    {
      "epoch": 0.23321615858929293,
      "grad_norm": 0.9035854339599609,
      "learning_rate": 8.955757468959555e-06,
      "loss": 0.7772,
      "step": 8094
    },
    {
      "epoch": 0.2332449720509422,
      "grad_norm": 0.9059041738510132,
      "learning_rate": 8.955472063539186e-06,
      "loss": 0.7579,
      "step": 8095
    },
    {
      "epoch": 0.23327378551259148,
      "grad_norm": 0.8746873736381531,
      "learning_rate": 8.95518662367061e-06,
      "loss": 0.7582,
      "step": 8096
    },
    {
      "epoch": 0.23330259897424077,
      "grad_norm": 0.8980125784873962,
      "learning_rate": 8.954901149356316e-06,
      "loss": 0.7708,
      "step": 8097
    },
    {
      "epoch": 0.23333141243589003,
      "grad_norm": 0.8906350135803223,
      "learning_rate": 8.954615640598783e-06,
      "loss": 0.7752,
      "step": 8098
    },
    {
      "epoch": 0.23336022589753933,
      "grad_norm": 0.900871217250824,
      "learning_rate": 8.9543300974005e-06,
      "loss": 0.76,
      "step": 8099
    },
    {
      "epoch": 0.23338903935918862,
      "grad_norm": 0.9083274006843567,
      "learning_rate": 8.954044519763958e-06,
      "loss": 0.7695,
      "step": 8100
    },
    {
      "epoch": 0.2334178528208379,
      "grad_norm": 0.9424760341644287,
      "learning_rate": 8.95375890769164e-06,
      "loss": 0.7681,
      "step": 8101
    },
    {
      "epoch": 0.23344666628248717,
      "grad_norm": 0.9154739379882812,
      "learning_rate": 8.953473261186032e-06,
      "loss": 0.7464,
      "step": 8102
    },
    {
      "epoch": 0.23347547974413646,
      "grad_norm": 0.9177629947662354,
      "learning_rate": 8.953187580249628e-06,
      "loss": 0.7514,
      "step": 8103
    },
    {
      "epoch": 0.23350429320578575,
      "grad_norm": 0.892141580581665,
      "learning_rate": 8.952901864884908e-06,
      "loss": 0.778,
      "step": 8104
    },
    {
      "epoch": 0.23353310666743501,
      "grad_norm": 0.9098677039146423,
      "learning_rate": 8.952616115094366e-06,
      "loss": 0.7898,
      "step": 8105
    },
    {
      "epoch": 0.2335619201290843,
      "grad_norm": 0.9146847128868103,
      "learning_rate": 8.952330330880488e-06,
      "loss": 0.7662,
      "step": 8106
    },
    {
      "epoch": 0.2335907335907336,
      "grad_norm": 0.9169938564300537,
      "learning_rate": 8.952044512245764e-06,
      "loss": 0.7711,
      "step": 8107
    },
    {
      "epoch": 0.2336195470523829,
      "grad_norm": 0.9061707258224487,
      "learning_rate": 8.951758659192684e-06,
      "loss": 0.7687,
      "step": 8108
    },
    {
      "epoch": 0.23364836051403215,
      "grad_norm": 0.887080729007721,
      "learning_rate": 8.951472771723735e-06,
      "loss": 0.7567,
      "step": 8109
    },
    {
      "epoch": 0.23367717397568144,
      "grad_norm": 0.9266502261161804,
      "learning_rate": 8.951186849841409e-06,
      "loss": 0.786,
      "step": 8110
    },
    {
      "epoch": 0.23370598743733073,
      "grad_norm": 0.9713443517684937,
      "learning_rate": 8.950900893548195e-06,
      "loss": 0.7673,
      "step": 8111
    },
    {
      "epoch": 0.23373480089898,
      "grad_norm": 0.8810907602310181,
      "learning_rate": 8.950614902846583e-06,
      "loss": 0.7538,
      "step": 8112
    },
    {
      "epoch": 0.23376361436062928,
      "grad_norm": 0.906099259853363,
      "learning_rate": 8.950328877739065e-06,
      "loss": 0.7442,
      "step": 8113
    },
    {
      "epoch": 0.23379242782227858,
      "grad_norm": 0.9102564454078674,
      "learning_rate": 8.95004281822813e-06,
      "loss": 0.7612,
      "step": 8114
    },
    {
      "epoch": 0.23382124128392784,
      "grad_norm": 0.8982797861099243,
      "learning_rate": 8.949756724316272e-06,
      "loss": 0.764,
      "step": 8115
    },
    {
      "epoch": 0.23385005474557713,
      "grad_norm": 0.8577885031700134,
      "learning_rate": 8.949470596005983e-06,
      "loss": 0.7741,
      "step": 8116
    },
    {
      "epoch": 0.23387886820722642,
      "grad_norm": 0.93494713306427,
      "learning_rate": 8.949184433299752e-06,
      "loss": 0.7862,
      "step": 8117
    },
    {
      "epoch": 0.2339076816688757,
      "grad_norm": 0.9015427231788635,
      "learning_rate": 8.94889823620007e-06,
      "loss": 0.7629,
      "step": 8118
    },
    {
      "epoch": 0.23393649513052497,
      "grad_norm": 0.895954430103302,
      "learning_rate": 8.948612004709434e-06,
      "loss": 0.767,
      "step": 8119
    },
    {
      "epoch": 0.23396530859217426,
      "grad_norm": 0.8623643517494202,
      "learning_rate": 8.948325738830335e-06,
      "loss": 0.7556,
      "step": 8120
    },
    {
      "epoch": 0.23399412205382356,
      "grad_norm": 0.9438585042953491,
      "learning_rate": 8.948039438565264e-06,
      "loss": 0.7906,
      "step": 8121
    },
    {
      "epoch": 0.23402293551547282,
      "grad_norm": 0.9370380640029907,
      "learning_rate": 8.947753103916717e-06,
      "loss": 0.7986,
      "step": 8122
    },
    {
      "epoch": 0.2340517489771221,
      "grad_norm": 0.9338687062263489,
      "learning_rate": 8.947466734887187e-06,
      "loss": 0.7569,
      "step": 8123
    },
    {
      "epoch": 0.2340805624387714,
      "grad_norm": 0.9805107116699219,
      "learning_rate": 8.947180331479166e-06,
      "loss": 0.7504,
      "step": 8124
    },
    {
      "epoch": 0.23410937590042066,
      "grad_norm": 0.9222347736358643,
      "learning_rate": 8.946893893695152e-06,
      "loss": 0.7674,
      "step": 8125
    },
    {
      "epoch": 0.23413818936206995,
      "grad_norm": 0.9649714827537537,
      "learning_rate": 8.946607421537636e-06,
      "loss": 0.7951,
      "step": 8126
    },
    {
      "epoch": 0.23416700282371924,
      "grad_norm": 0.9776954650878906,
      "learning_rate": 8.946320915009117e-06,
      "loss": 0.7717,
      "step": 8127
    },
    {
      "epoch": 0.23419581628536854,
      "grad_norm": 0.9846818447113037,
      "learning_rate": 8.946034374112086e-06,
      "loss": 0.7951,
      "step": 8128
    },
    {
      "epoch": 0.2342246297470178,
      "grad_norm": 0.8853676319122314,
      "learning_rate": 8.945747798849041e-06,
      "loss": 0.768,
      "step": 8129
    },
    {
      "epoch": 0.2342534432086671,
      "grad_norm": 0.962159276008606,
      "learning_rate": 8.945461189222476e-06,
      "loss": 0.7726,
      "step": 8130
    },
    {
      "epoch": 0.23428225667031638,
      "grad_norm": 0.9557548761367798,
      "learning_rate": 8.945174545234889e-06,
      "loss": 0.7779,
      "step": 8131
    },
    {
      "epoch": 0.23431107013196564,
      "grad_norm": 0.9236361384391785,
      "learning_rate": 8.944887866888774e-06,
      "loss": 0.7764,
      "step": 8132
    },
    {
      "epoch": 0.23433988359361493,
      "grad_norm": 0.9243695139884949,
      "learning_rate": 8.94460115418663e-06,
      "loss": 0.7735,
      "step": 8133
    },
    {
      "epoch": 0.23436869705526422,
      "grad_norm": 0.9569180607795715,
      "learning_rate": 8.944314407130953e-06,
      "loss": 0.7789,
      "step": 8134
    },
    {
      "epoch": 0.23439751051691352,
      "grad_norm": 0.971420168876648,
      "learning_rate": 8.944027625724241e-06,
      "loss": 0.7777,
      "step": 8135
    },
    {
      "epoch": 0.23442632397856278,
      "grad_norm": 0.918533980846405,
      "learning_rate": 8.943740809968991e-06,
      "loss": 0.7566,
      "step": 8136
    },
    {
      "epoch": 0.23445513744021207,
      "grad_norm": 0.9280393719673157,
      "learning_rate": 8.9434539598677e-06,
      "loss": 0.7715,
      "step": 8137
    },
    {
      "epoch": 0.23448395090186136,
      "grad_norm": 0.9039229154586792,
      "learning_rate": 8.943167075422867e-06,
      "loss": 0.7652,
      "step": 8138
    },
    {
      "epoch": 0.23451276436351062,
      "grad_norm": 0.9130387306213379,
      "learning_rate": 8.94288015663699e-06,
      "loss": 0.7633,
      "step": 8139
    },
    {
      "epoch": 0.2345415778251599,
      "grad_norm": 0.9420956969261169,
      "learning_rate": 8.94259320351257e-06,
      "loss": 0.7892,
      "step": 8140
    },
    {
      "epoch": 0.2345703912868092,
      "grad_norm": 0.9018370509147644,
      "learning_rate": 8.942306216052105e-06,
      "loss": 0.7707,
      "step": 8141
    },
    {
      "epoch": 0.23459920474845847,
      "grad_norm": 0.907518744468689,
      "learning_rate": 8.942019194258091e-06,
      "loss": 0.7497,
      "step": 8142
    },
    {
      "epoch": 0.23462801821010776,
      "grad_norm": 0.9263885021209717,
      "learning_rate": 8.941732138133032e-06,
      "loss": 0.7856,
      "step": 8143
    },
    {
      "epoch": 0.23465683167175705,
      "grad_norm": 0.9367966055870056,
      "learning_rate": 8.941445047679427e-06,
      "loss": 0.7746,
      "step": 8144
    },
    {
      "epoch": 0.23468564513340634,
      "grad_norm": 0.9226201772689819,
      "learning_rate": 8.941157922899773e-06,
      "loss": 0.8015,
      "step": 8145
    },
    {
      "epoch": 0.2347144585950556,
      "grad_norm": 0.9281942844390869,
      "learning_rate": 8.940870763796575e-06,
      "loss": 0.7646,
      "step": 8146
    },
    {
      "epoch": 0.2347432720567049,
      "grad_norm": 0.9427006244659424,
      "learning_rate": 8.940583570372333e-06,
      "loss": 0.7937,
      "step": 8147
    },
    {
      "epoch": 0.23477208551835418,
      "grad_norm": 0.8609273433685303,
      "learning_rate": 8.940296342629546e-06,
      "loss": 0.7544,
      "step": 8148
    },
    {
      "epoch": 0.23480089898000345,
      "grad_norm": 0.9359025955200195,
      "learning_rate": 8.940009080570718e-06,
      "loss": 0.7412,
      "step": 8149
    },
    {
      "epoch": 0.23482971244165274,
      "grad_norm": 0.9132676124572754,
      "learning_rate": 8.939721784198347e-06,
      "loss": 0.7446,
      "step": 8150
    },
    {
      "epoch": 0.23485852590330203,
      "grad_norm": 0.916872501373291,
      "learning_rate": 8.939434453514939e-06,
      "loss": 0.7817,
      "step": 8151
    },
    {
      "epoch": 0.2348873393649513,
      "grad_norm": 0.9298280477523804,
      "learning_rate": 8.939147088522994e-06,
      "loss": 0.764,
      "step": 8152
    },
    {
      "epoch": 0.23491615282660058,
      "grad_norm": 0.9117103815078735,
      "learning_rate": 8.938859689225017e-06,
      "loss": 0.7875,
      "step": 8153
    },
    {
      "epoch": 0.23494496628824987,
      "grad_norm": 0.8751398921012878,
      "learning_rate": 8.93857225562351e-06,
      "loss": 0.754,
      "step": 8154
    },
    {
      "epoch": 0.23497377974989916,
      "grad_norm": 0.9004303216934204,
      "learning_rate": 8.938284787720975e-06,
      "loss": 0.788,
      "step": 8155
    },
    {
      "epoch": 0.23500259321154843,
      "grad_norm": 0.8965709805488586,
      "learning_rate": 8.937997285519916e-06,
      "loss": 0.7524,
      "step": 8156
    },
    {
      "epoch": 0.23503140667319772,
      "grad_norm": 0.8946303725242615,
      "learning_rate": 8.937709749022837e-06,
      "loss": 0.7801,
      "step": 8157
    },
    {
      "epoch": 0.235060220134847,
      "grad_norm": 0.9467129707336426,
      "learning_rate": 8.937422178232241e-06,
      "loss": 0.7661,
      "step": 8158
    },
    {
      "epoch": 0.23508903359649627,
      "grad_norm": 0.9135291576385498,
      "learning_rate": 8.937134573150637e-06,
      "loss": 0.7461,
      "step": 8159
    },
    {
      "epoch": 0.23511784705814556,
      "grad_norm": 0.8871086835861206,
      "learning_rate": 8.936846933780524e-06,
      "loss": 0.7553,
      "step": 8160
    },
    {
      "epoch": 0.23514666051979485,
      "grad_norm": 0.9605872631072998,
      "learning_rate": 8.93655926012441e-06,
      "loss": 0.7666,
      "step": 8161
    },
    {
      "epoch": 0.23517547398144414,
      "grad_norm": 0.9388047456741333,
      "learning_rate": 8.9362715521848e-06,
      "loss": 0.7554,
      "step": 8162
    },
    {
      "epoch": 0.2352042874430934,
      "grad_norm": 0.8842254877090454,
      "learning_rate": 8.935983809964201e-06,
      "loss": 0.7467,
      "step": 8163
    },
    {
      "epoch": 0.2352331009047427,
      "grad_norm": 0.9049485921859741,
      "learning_rate": 8.935696033465117e-06,
      "loss": 0.7799,
      "step": 8164
    },
    {
      "epoch": 0.235261914366392,
      "grad_norm": 0.8978177309036255,
      "learning_rate": 8.935408222690054e-06,
      "loss": 0.7977,
      "step": 8165
    },
    {
      "epoch": 0.23529072782804125,
      "grad_norm": 0.9361450672149658,
      "learning_rate": 8.93512037764152e-06,
      "loss": 0.7684,
      "step": 8166
    },
    {
      "epoch": 0.23531954128969054,
      "grad_norm": 1.008880853652954,
      "learning_rate": 8.934832498322021e-06,
      "loss": 0.7701,
      "step": 8167
    },
    {
      "epoch": 0.23534835475133983,
      "grad_norm": 0.8959435224533081,
      "learning_rate": 8.934544584734065e-06,
      "loss": 0.7466,
      "step": 8168
    },
    {
      "epoch": 0.2353771682129891,
      "grad_norm": 0.943524956703186,
      "learning_rate": 8.934256636880158e-06,
      "loss": 0.7552,
      "step": 8169
    },
    {
      "epoch": 0.2354059816746384,
      "grad_norm": 0.9721916317939758,
      "learning_rate": 8.933968654762808e-06,
      "loss": 0.795,
      "step": 8170
    },
    {
      "epoch": 0.23543479513628768,
      "grad_norm": 0.8979207873344421,
      "learning_rate": 8.933680638384525e-06,
      "loss": 0.7699,
      "step": 8171
    },
    {
      "epoch": 0.23546360859793697,
      "grad_norm": 0.9120725393295288,
      "learning_rate": 8.933392587747815e-06,
      "loss": 0.7805,
      "step": 8172
    },
    {
      "epoch": 0.23549242205958623,
      "grad_norm": 0.9343242645263672,
      "learning_rate": 8.933104502855187e-06,
      "loss": 0.7698,
      "step": 8173
    },
    {
      "epoch": 0.23552123552123552,
      "grad_norm": 0.9430821537971497,
      "learning_rate": 8.932816383709151e-06,
      "loss": 0.7549,
      "step": 8174
    },
    {
      "epoch": 0.2355500489828848,
      "grad_norm": 0.9945819973945618,
      "learning_rate": 8.932528230312217e-06,
      "loss": 0.7769,
      "step": 8175
    },
    {
      "epoch": 0.23557886244453408,
      "grad_norm": 0.9922563433647156,
      "learning_rate": 8.932240042666891e-06,
      "loss": 0.7727,
      "step": 8176
    },
    {
      "epoch": 0.23560767590618337,
      "grad_norm": 0.9479523301124573,
      "learning_rate": 8.931951820775686e-06,
      "loss": 0.7947,
      "step": 8177
    },
    {
      "epoch": 0.23563648936783266,
      "grad_norm": 0.9119746088981628,
      "learning_rate": 8.93166356464111e-06,
      "loss": 0.7664,
      "step": 8178
    },
    {
      "epoch": 0.23566530282948192,
      "grad_norm": 0.9559798836708069,
      "learning_rate": 8.931375274265677e-06,
      "loss": 0.7582,
      "step": 8179
    },
    {
      "epoch": 0.2356941162911312,
      "grad_norm": 0.9644976258277893,
      "learning_rate": 8.931086949651893e-06,
      "loss": 0.7644,
      "step": 8180
    },
    {
      "epoch": 0.2357229297527805,
      "grad_norm": 0.9343159198760986,
      "learning_rate": 8.930798590802273e-06,
      "loss": 0.7735,
      "step": 8181
    },
    {
      "epoch": 0.2357517432144298,
      "grad_norm": 0.9285393953323364,
      "learning_rate": 8.930510197719326e-06,
      "loss": 0.7684,
      "step": 8182
    },
    {
      "epoch": 0.23578055667607906,
      "grad_norm": 0.9287545680999756,
      "learning_rate": 8.930221770405565e-06,
      "loss": 0.7709,
      "step": 8183
    },
    {
      "epoch": 0.23580937013772835,
      "grad_norm": 0.9053083062171936,
      "learning_rate": 8.9299333088635e-06,
      "loss": 0.7797,
      "step": 8184
    },
    {
      "epoch": 0.23583818359937764,
      "grad_norm": 0.9353432059288025,
      "learning_rate": 8.929644813095644e-06,
      "loss": 0.785,
      "step": 8185
    },
    {
      "epoch": 0.2358669970610269,
      "grad_norm": 0.9382407069206238,
      "learning_rate": 8.92935628310451e-06,
      "loss": 0.7758,
      "step": 8186
    },
    {
      "epoch": 0.2358958105226762,
      "grad_norm": 0.8942466378211975,
      "learning_rate": 8.929067718892612e-06,
      "loss": 0.7879,
      "step": 8187
    },
    {
      "epoch": 0.23592462398432548,
      "grad_norm": 0.8626745343208313,
      "learning_rate": 8.92877912046246e-06,
      "loss": 0.7624,
      "step": 8188
    },
    {
      "epoch": 0.23595343744597477,
      "grad_norm": 0.9452343583106995,
      "learning_rate": 8.928490487816573e-06,
      "loss": 0.7663,
      "step": 8189
    },
    {
      "epoch": 0.23598225090762404,
      "grad_norm": 0.9174634218215942,
      "learning_rate": 8.928201820957456e-06,
      "loss": 0.762,
      "step": 8190
    },
    {
      "epoch": 0.23601106436927333,
      "grad_norm": 0.8897039890289307,
      "learning_rate": 8.92791311988763e-06,
      "loss": 0.7544,
      "step": 8191
    },
    {
      "epoch": 0.23603987783092262,
      "grad_norm": 0.9329196214675903,
      "learning_rate": 8.927624384609607e-06,
      "loss": 0.7444,
      "step": 8192
    },
    {
      "epoch": 0.23606869129257188,
      "grad_norm": 0.9984038472175598,
      "learning_rate": 8.927335615125903e-06,
      "loss": 0.7569,
      "step": 8193
    },
    {
      "epoch": 0.23609750475422117,
      "grad_norm": 0.9568217396736145,
      "learning_rate": 8.92704681143903e-06,
      "loss": 0.7712,
      "step": 8194
    },
    {
      "epoch": 0.23612631821587046,
      "grad_norm": 0.9353212118148804,
      "learning_rate": 8.926757973551507e-06,
      "loss": 0.742,
      "step": 8195
    },
    {
      "epoch": 0.23615513167751973,
      "grad_norm": 0.9082231521606445,
      "learning_rate": 8.926469101465845e-06,
      "loss": 0.7654,
      "step": 8196
    },
    {
      "epoch": 0.23618394513916902,
      "grad_norm": 0.9333398342132568,
      "learning_rate": 8.926180195184565e-06,
      "loss": 0.796,
      "step": 8197
    },
    {
      "epoch": 0.2362127586008183,
      "grad_norm": 0.8759164214134216,
      "learning_rate": 8.925891254710177e-06,
      "loss": 0.7676,
      "step": 8198
    },
    {
      "epoch": 0.2362415720624676,
      "grad_norm": 0.9073238968849182,
      "learning_rate": 8.925602280045203e-06,
      "loss": 0.768,
      "step": 8199
    },
    {
      "epoch": 0.23627038552411686,
      "grad_norm": 0.8902122378349304,
      "learning_rate": 8.925313271192155e-06,
      "loss": 0.7672,
      "step": 8200
    },
    {
      "epoch": 0.23629919898576615,
      "grad_norm": 0.93109530210495,
      "learning_rate": 8.925024228153555e-06,
      "loss": 0.7532,
      "step": 8201
    },
    {
      "epoch": 0.23632801244741544,
      "grad_norm": 0.9335991740226746,
      "learning_rate": 8.924735150931915e-06,
      "loss": 0.7798,
      "step": 8202
    },
    {
      "epoch": 0.2363568259090647,
      "grad_norm": 0.8500086665153503,
      "learning_rate": 8.924446039529757e-06,
      "loss": 0.7278,
      "step": 8203
    },
    {
      "epoch": 0.236385639370714,
      "grad_norm": 0.9245460629463196,
      "learning_rate": 8.924156893949595e-06,
      "loss": 0.7581,
      "step": 8204
    },
    {
      "epoch": 0.2364144528323633,
      "grad_norm": 0.9424853324890137,
      "learning_rate": 8.923867714193952e-06,
      "loss": 0.7705,
      "step": 8205
    },
    {
      "epoch": 0.23644326629401255,
      "grad_norm": 0.9441462159156799,
      "learning_rate": 8.923578500265341e-06,
      "loss": 0.7842,
      "step": 8206
    },
    {
      "epoch": 0.23647207975566184,
      "grad_norm": 0.9478377103805542,
      "learning_rate": 8.923289252166285e-06,
      "loss": 0.7546,
      "step": 8207
    },
    {
      "epoch": 0.23650089321731113,
      "grad_norm": 0.9143943190574646,
      "learning_rate": 8.922999969899303e-06,
      "loss": 0.7936,
      "step": 8208
    },
    {
      "epoch": 0.23652970667896042,
      "grad_norm": 0.9420090317726135,
      "learning_rate": 8.92271065346691e-06,
      "loss": 0.7771,
      "step": 8209
    },
    {
      "epoch": 0.23655852014060968,
      "grad_norm": 0.8695710301399231,
      "learning_rate": 8.922421302871628e-06,
      "loss": 0.761,
      "step": 8210
    },
    {
      "epoch": 0.23658733360225898,
      "grad_norm": 0.8982723951339722,
      "learning_rate": 8.92213191811598e-06,
      "loss": 0.7835,
      "step": 8211
    },
    {
      "epoch": 0.23661614706390827,
      "grad_norm": 0.9277316927909851,
      "learning_rate": 8.921842499202482e-06,
      "loss": 0.7593,
      "step": 8212
    },
    {
      "epoch": 0.23664496052555753,
      "grad_norm": 0.9202002286911011,
      "learning_rate": 8.921553046133657e-06,
      "loss": 0.7621,
      "step": 8213
    },
    {
      "epoch": 0.23667377398720682,
      "grad_norm": 0.8984789848327637,
      "learning_rate": 8.921263558912026e-06,
      "loss": 0.7579,
      "step": 8214
    },
    {
      "epoch": 0.2367025874488561,
      "grad_norm": 0.9129093885421753,
      "learning_rate": 8.920974037540106e-06,
      "loss": 0.7856,
      "step": 8215
    },
    {
      "epoch": 0.2367314009105054,
      "grad_norm": 0.9430802464485168,
      "learning_rate": 8.920684482020425e-06,
      "loss": 0.7819,
      "step": 8216
    },
    {
      "epoch": 0.23676021437215466,
      "grad_norm": 0.9384976625442505,
      "learning_rate": 8.920394892355499e-06,
      "loss": 0.7757,
      "step": 8217
    },
    {
      "epoch": 0.23678902783380396,
      "grad_norm": 0.944148600101471,
      "learning_rate": 8.920105268547854e-06,
      "loss": 0.7799,
      "step": 8218
    },
    {
      "epoch": 0.23681784129545325,
      "grad_norm": 0.8957644104957581,
      "learning_rate": 8.91981561060001e-06,
      "loss": 0.7424,
      "step": 8219
    },
    {
      "epoch": 0.2368466547571025,
      "grad_norm": 0.9358124732971191,
      "learning_rate": 8.919525918514491e-06,
      "loss": 0.7628,
      "step": 8220
    },
    {
      "epoch": 0.2368754682187518,
      "grad_norm": 0.9422397613525391,
      "learning_rate": 8.91923619229382e-06,
      "loss": 0.7729,
      "step": 8221
    },
    {
      "epoch": 0.2369042816804011,
      "grad_norm": 0.8892631530761719,
      "learning_rate": 8.918946431940519e-06,
      "loss": 0.753,
      "step": 8222
    },
    {
      "epoch": 0.23693309514205035,
      "grad_norm": 0.9131280183792114,
      "learning_rate": 8.918656637457112e-06,
      "loss": 0.791,
      "step": 8223
    },
    {
      "epoch": 0.23696190860369964,
      "grad_norm": 0.9098237752914429,
      "learning_rate": 8.918366808846121e-06,
      "loss": 0.7763,
      "step": 8224
    },
    {
      "epoch": 0.23699072206534894,
      "grad_norm": 0.9734644889831543,
      "learning_rate": 8.918076946110074e-06,
      "loss": 0.7599,
      "step": 8225
    },
    {
      "epoch": 0.23701953552699823,
      "grad_norm": 0.878837525844574,
      "learning_rate": 8.917787049251494e-06,
      "loss": 0.7649,
      "step": 8226
    },
    {
      "epoch": 0.2370483489886475,
      "grad_norm": 0.888934314250946,
      "learning_rate": 8.917497118272904e-06,
      "loss": 0.7731,
      "step": 8227
    },
    {
      "epoch": 0.23707716245029678,
      "grad_norm": 0.9499020576477051,
      "learning_rate": 8.91720715317683e-06,
      "loss": 0.7666,
      "step": 8228
    },
    {
      "epoch": 0.23710597591194607,
      "grad_norm": 0.8738627433776855,
      "learning_rate": 8.916917153965798e-06,
      "loss": 0.7483,
      "step": 8229
    },
    {
      "epoch": 0.23713478937359533,
      "grad_norm": 0.90941321849823,
      "learning_rate": 8.916627120642334e-06,
      "loss": 0.7578,
      "step": 8230
    },
    {
      "epoch": 0.23716360283524462,
      "grad_norm": 0.948249340057373,
      "learning_rate": 8.91633705320896e-06,
      "loss": 0.7858,
      "step": 8231
    },
    {
      "epoch": 0.23719241629689392,
      "grad_norm": 0.9343569278717041,
      "learning_rate": 8.91604695166821e-06,
      "loss": 0.7513,
      "step": 8232
    },
    {
      "epoch": 0.23722122975854318,
      "grad_norm": 0.9017964601516724,
      "learning_rate": 8.915756816022602e-06,
      "loss": 0.7803,
      "step": 8233
    },
    {
      "epoch": 0.23725004322019247,
      "grad_norm": 0.8792852759361267,
      "learning_rate": 8.915466646274666e-06,
      "loss": 0.7667,
      "step": 8234
    },
    {
      "epoch": 0.23727885668184176,
      "grad_norm": 0.9732091426849365,
      "learning_rate": 8.915176442426931e-06,
      "loss": 0.7903,
      "step": 8235
    },
    {
      "epoch": 0.23730767014349105,
      "grad_norm": 0.902025580406189,
      "learning_rate": 8.914886204481922e-06,
      "loss": 0.7772,
      "step": 8236
    },
    {
      "epoch": 0.2373364836051403,
      "grad_norm": 0.8730404376983643,
      "learning_rate": 8.91459593244217e-06,
      "loss": 0.7589,
      "step": 8237
    },
    {
      "epoch": 0.2373652970667896,
      "grad_norm": 0.9456363916397095,
      "learning_rate": 8.914305626310197e-06,
      "loss": 0.7655,
      "step": 8238
    },
    {
      "epoch": 0.2373941105284389,
      "grad_norm": 0.9368256330490112,
      "learning_rate": 8.914015286088538e-06,
      "loss": 0.7607,
      "step": 8239
    },
    {
      "epoch": 0.23742292399008816,
      "grad_norm": 0.9294854402542114,
      "learning_rate": 8.913724911779717e-06,
      "loss": 0.7722,
      "step": 8240
    },
    {
      "epoch": 0.23745173745173745,
      "grad_norm": 0.9154859185218811,
      "learning_rate": 8.913434503386266e-06,
      "loss": 0.7745,
      "step": 8241
    },
    {
      "epoch": 0.23748055091338674,
      "grad_norm": 0.9231518507003784,
      "learning_rate": 8.91314406091071e-06,
      "loss": 0.7673,
      "step": 8242
    },
    {
      "epoch": 0.23750936437503603,
      "grad_norm": 0.9148849844932556,
      "learning_rate": 8.912853584355583e-06,
      "loss": 0.7857,
      "step": 8243
    },
    {
      "epoch": 0.2375381778366853,
      "grad_norm": 0.9406912922859192,
      "learning_rate": 8.912563073723412e-06,
      "loss": 0.7957,
      "step": 8244
    },
    {
      "epoch": 0.23756699129833458,
      "grad_norm": 0.9353243708610535,
      "learning_rate": 8.912272529016727e-06,
      "loss": 0.7734,
      "step": 8245
    },
    {
      "epoch": 0.23759580475998388,
      "grad_norm": 0.9031746983528137,
      "learning_rate": 8.91198195023806e-06,
      "loss": 0.7691,
      "step": 8246
    },
    {
      "epoch": 0.23762461822163314,
      "grad_norm": 0.9070513844490051,
      "learning_rate": 8.91169133738994e-06,
      "loss": 0.7675,
      "step": 8247
    },
    {
      "epoch": 0.23765343168328243,
      "grad_norm": 0.9729399681091309,
      "learning_rate": 8.911400690474898e-06,
      "loss": 0.7968,
      "step": 8248
    },
    {
      "epoch": 0.23768224514493172,
      "grad_norm": 0.9187281131744385,
      "learning_rate": 8.911110009495467e-06,
      "loss": 0.7816,
      "step": 8249
    },
    {
      "epoch": 0.23771105860658098,
      "grad_norm": 0.9251505136489868,
      "learning_rate": 8.910819294454178e-06,
      "loss": 0.7552,
      "step": 8250
    },
    {
      "epoch": 0.23773987206823027,
      "grad_norm": 0.8992893695831299,
      "learning_rate": 8.910528545353563e-06,
      "loss": 0.7738,
      "step": 8251
    },
    {
      "epoch": 0.23776868552987956,
      "grad_norm": 0.8763574957847595,
      "learning_rate": 8.910237762196153e-06,
      "loss": 0.7684,
      "step": 8252
    },
    {
      "epoch": 0.23779749899152886,
      "grad_norm": 0.8958055377006531,
      "learning_rate": 8.90994694498448e-06,
      "loss": 0.7692,
      "step": 8253
    },
    {
      "epoch": 0.23782631245317812,
      "grad_norm": 0.9074973464012146,
      "learning_rate": 8.909656093721077e-06,
      "loss": 0.753,
      "step": 8254
    },
    {
      "epoch": 0.2378551259148274,
      "grad_norm": 0.8822408318519592,
      "learning_rate": 8.909365208408478e-06,
      "loss": 0.7346,
      "step": 8255
    },
    {
      "epoch": 0.2378839393764767,
      "grad_norm": 0.8512738943099976,
      "learning_rate": 8.909074289049215e-06,
      "loss": 0.7495,
      "step": 8256
    },
    {
      "epoch": 0.23791275283812596,
      "grad_norm": 0.8874322772026062,
      "learning_rate": 8.908783335645825e-06,
      "loss": 0.7778,
      "step": 8257
    },
    {
      "epoch": 0.23794156629977525,
      "grad_norm": 0.910564124584198,
      "learning_rate": 8.90849234820084e-06,
      "loss": 0.7647,
      "step": 8258
    },
    {
      "epoch": 0.23797037976142454,
      "grad_norm": 0.9181308150291443,
      "learning_rate": 8.908201326716791e-06,
      "loss": 0.7442,
      "step": 8259
    },
    {
      "epoch": 0.2379991932230738,
      "grad_norm": 0.8841615915298462,
      "learning_rate": 8.907910271196216e-06,
      "loss": 0.7701,
      "step": 8260
    },
    {
      "epoch": 0.2380280066847231,
      "grad_norm": 0.8863446712493896,
      "learning_rate": 8.90761918164165e-06,
      "loss": 0.7679,
      "step": 8261
    },
    {
      "epoch": 0.2380568201463724,
      "grad_norm": 0.8860533237457275,
      "learning_rate": 8.907328058055627e-06,
      "loss": 0.7796,
      "step": 8262
    },
    {
      "epoch": 0.23808563360802168,
      "grad_norm": 0.8790935277938843,
      "learning_rate": 8.907036900440682e-06,
      "loss": 0.7592,
      "step": 8263
    },
    {
      "epoch": 0.23811444706967094,
      "grad_norm": 0.8940017819404602,
      "learning_rate": 8.906745708799351e-06,
      "loss": 0.7546,
      "step": 8264
    },
    {
      "epoch": 0.23814326053132023,
      "grad_norm": 0.9016168713569641,
      "learning_rate": 8.90645448313417e-06,
      "loss": 0.7572,
      "step": 8265
    },
    {
      "epoch": 0.23817207399296952,
      "grad_norm": 0.8625037670135498,
      "learning_rate": 8.906163223447678e-06,
      "loss": 0.76,
      "step": 8266
    },
    {
      "epoch": 0.2382008874546188,
      "grad_norm": 0.9072645306587219,
      "learning_rate": 8.905871929742406e-06,
      "loss": 0.7713,
      "step": 8267
    },
    {
      "epoch": 0.23822970091626808,
      "grad_norm": 0.8736093640327454,
      "learning_rate": 8.905580602020895e-06,
      "loss": 0.7574,
      "step": 8268
    },
    {
      "epoch": 0.23825851437791737,
      "grad_norm": 0.8849208354949951,
      "learning_rate": 8.905289240285682e-06,
      "loss": 0.7718,
      "step": 8269
    },
    {
      "epoch": 0.23828732783956663,
      "grad_norm": 0.9024584293365479,
      "learning_rate": 8.904997844539301e-06,
      "loss": 0.7742,
      "step": 8270
    },
    {
      "epoch": 0.23831614130121592,
      "grad_norm": 0.8650311827659607,
      "learning_rate": 8.904706414784295e-06,
      "loss": 0.7779,
      "step": 8271
    },
    {
      "epoch": 0.2383449547628652,
      "grad_norm": 0.8788854479789734,
      "learning_rate": 8.904414951023198e-06,
      "loss": 0.7571,
      "step": 8272
    },
    {
      "epoch": 0.2383737682245145,
      "grad_norm": 0.9444255232810974,
      "learning_rate": 8.904123453258551e-06,
      "loss": 0.782,
      "step": 8273
    },
    {
      "epoch": 0.23840258168616377,
      "grad_norm": 0.8918011784553528,
      "learning_rate": 8.903831921492889e-06,
      "loss": 0.7488,
      "step": 8274
    },
    {
      "epoch": 0.23843139514781306,
      "grad_norm": 0.9298084378242493,
      "learning_rate": 8.903540355728755e-06,
      "loss": 0.764,
      "step": 8275
    },
    {
      "epoch": 0.23846020860946235,
      "grad_norm": 0.9087092876434326,
      "learning_rate": 8.903248755968687e-06,
      "loss": 0.751,
      "step": 8276
    },
    {
      "epoch": 0.2384890220711116,
      "grad_norm": 0.8957142233848572,
      "learning_rate": 8.902957122215223e-06,
      "loss": 0.7599,
      "step": 8277
    },
    {
      "epoch": 0.2385178355327609,
      "grad_norm": 0.9135828018188477,
      "learning_rate": 8.902665454470906e-06,
      "loss": 0.7774,
      "step": 8278
    },
    {
      "epoch": 0.2385466489944102,
      "grad_norm": 0.9266155362129211,
      "learning_rate": 8.902373752738272e-06,
      "loss": 0.7831,
      "step": 8279
    },
    {
      "epoch": 0.23857546245605948,
      "grad_norm": 0.9280433058738708,
      "learning_rate": 8.902082017019865e-06,
      "loss": 0.7828,
      "step": 8280
    },
    {
      "epoch": 0.23860427591770875,
      "grad_norm": 0.8978965878486633,
      "learning_rate": 8.901790247318224e-06,
      "loss": 0.7808,
      "step": 8281
    },
    {
      "epoch": 0.23863308937935804,
      "grad_norm": 0.995424211025238,
      "learning_rate": 8.90149844363589e-06,
      "loss": 0.8143,
      "step": 8282
    },
    {
      "epoch": 0.23866190284100733,
      "grad_norm": 0.9654719233512878,
      "learning_rate": 8.901206605975404e-06,
      "loss": 0.7821,
      "step": 8283
    },
    {
      "epoch": 0.2386907163026566,
      "grad_norm": 0.8636229038238525,
      "learning_rate": 8.90091473433931e-06,
      "loss": 0.7802,
      "step": 8284
    },
    {
      "epoch": 0.23871952976430588,
      "grad_norm": 0.8900563716888428,
      "learning_rate": 8.900622828730145e-06,
      "loss": 0.7508,
      "step": 8285
    },
    {
      "epoch": 0.23874834322595517,
      "grad_norm": 1.012597680091858,
      "learning_rate": 8.900330889150457e-06,
      "loss": 0.757,
      "step": 8286
    },
    {
      "epoch": 0.23877715668760444,
      "grad_norm": 0.9096269011497498,
      "learning_rate": 8.900038915602787e-06,
      "loss": 0.7892,
      "step": 8287
    },
    {
      "epoch": 0.23880597014925373,
      "grad_norm": 0.9103176593780518,
      "learning_rate": 8.899746908089674e-06,
      "loss": 0.786,
      "step": 8288
    },
    {
      "epoch": 0.23883478361090302,
      "grad_norm": 0.9317898154258728,
      "learning_rate": 8.899454866613663e-06,
      "loss": 0.7639,
      "step": 8289
    },
    {
      "epoch": 0.2388635970725523,
      "grad_norm": 0.8825844526290894,
      "learning_rate": 8.899162791177301e-06,
      "loss": 0.7752,
      "step": 8290
    },
    {
      "epoch": 0.23889241053420157,
      "grad_norm": 0.9009700417518616,
      "learning_rate": 8.898870681783128e-06,
      "loss": 0.7702,
      "step": 8291
    },
    {
      "epoch": 0.23892122399585086,
      "grad_norm": 0.9248087406158447,
      "learning_rate": 8.898578538433689e-06,
      "loss": 0.7971,
      "step": 8292
    },
    {
      "epoch": 0.23895003745750015,
      "grad_norm": 0.9023637175559998,
      "learning_rate": 8.898286361131527e-06,
      "loss": 0.768,
      "step": 8293
    },
    {
      "epoch": 0.23897885091914942,
      "grad_norm": 0.8849248886108398,
      "learning_rate": 8.897994149879187e-06,
      "loss": 0.7531,
      "step": 8294
    },
    {
      "epoch": 0.2390076643807987,
      "grad_norm": 0.8875941634178162,
      "learning_rate": 8.897701904679217e-06,
      "loss": 0.7735,
      "step": 8295
    },
    {
      "epoch": 0.239036477842448,
      "grad_norm": 0.9070354104042053,
      "learning_rate": 8.897409625534159e-06,
      "loss": 0.7549,
      "step": 8296
    },
    {
      "epoch": 0.23906529130409726,
      "grad_norm": 0.929035484790802,
      "learning_rate": 8.897117312446557e-06,
      "loss": 0.7885,
      "step": 8297
    },
    {
      "epoch": 0.23909410476574655,
      "grad_norm": 0.8737877011299133,
      "learning_rate": 8.896824965418961e-06,
      "loss": 0.7598,
      "step": 8298
    },
    {
      "epoch": 0.23912291822739584,
      "grad_norm": 0.8889357447624207,
      "learning_rate": 8.896532584453914e-06,
      "loss": 0.7784,
      "step": 8299
    },
    {
      "epoch": 0.23915173168904513,
      "grad_norm": 0.8730996251106262,
      "learning_rate": 8.896240169553965e-06,
      "loss": 0.7644,
      "step": 8300
    },
    {
      "epoch": 0.2391805451506944,
      "grad_norm": 0.9248403906822205,
      "learning_rate": 8.895947720721655e-06,
      "loss": 0.776,
      "step": 8301
    },
    {
      "epoch": 0.2392093586123437,
      "grad_norm": 0.8585414886474609,
      "learning_rate": 8.89565523795954e-06,
      "loss": 0.7778,
      "step": 8302
    },
    {
      "epoch": 0.23923817207399298,
      "grad_norm": 0.8679609894752502,
      "learning_rate": 8.895362721270158e-06,
      "loss": 0.7716,
      "step": 8303
    },
    {
      "epoch": 0.23926698553564224,
      "grad_norm": 0.869002640247345,
      "learning_rate": 8.895070170656063e-06,
      "loss": 0.7867,
      "step": 8304
    },
    {
      "epoch": 0.23929579899729153,
      "grad_norm": 0.9371008276939392,
      "learning_rate": 8.8947775861198e-06,
      "loss": 0.7742,
      "step": 8305
    },
    {
      "epoch": 0.23932461245894082,
      "grad_norm": 0.8969863057136536,
      "learning_rate": 8.894484967663917e-06,
      "loss": 0.7456,
      "step": 8306
    },
    {
      "epoch": 0.2393534259205901,
      "grad_norm": 0.8802213072776794,
      "learning_rate": 8.894192315290962e-06,
      "loss": 0.7573,
      "step": 8307
    },
    {
      "epoch": 0.23938223938223938,
      "grad_norm": 0.9029304385185242,
      "learning_rate": 8.893899629003485e-06,
      "loss": 0.772,
      "step": 8308
    },
    {
      "epoch": 0.23941105284388867,
      "grad_norm": 0.9002759456634521,
      "learning_rate": 8.893606908804038e-06,
      "loss": 0.7745,
      "step": 8309
    },
    {
      "epoch": 0.23943986630553796,
      "grad_norm": 0.8682321906089783,
      "learning_rate": 8.893314154695164e-06,
      "loss": 0.7483,
      "step": 8310
    },
    {
      "epoch": 0.23946867976718722,
      "grad_norm": 0.8659100532531738,
      "learning_rate": 8.893021366679418e-06,
      "loss": 0.768,
      "step": 8311
    },
    {
      "epoch": 0.2394974932288365,
      "grad_norm": 0.9314107894897461,
      "learning_rate": 8.892728544759344e-06,
      "loss": 0.7825,
      "step": 8312
    },
    {
      "epoch": 0.2395263066904858,
      "grad_norm": 0.8788018226623535,
      "learning_rate": 8.892435688937497e-06,
      "loss": 0.7554,
      "step": 8313
    },
    {
      "epoch": 0.23955512015213506,
      "grad_norm": 0.8776196241378784,
      "learning_rate": 8.892142799216427e-06,
      "loss": 0.7565,
      "step": 8314
    },
    {
      "epoch": 0.23958393361378436,
      "grad_norm": 0.8780533671379089,
      "learning_rate": 8.891849875598684e-06,
      "loss": 0.7828,
      "step": 8315
    },
    {
      "epoch": 0.23961274707543365,
      "grad_norm": 0.8583004474639893,
      "learning_rate": 8.891556918086819e-06,
      "loss": 0.7621,
      "step": 8316
    },
    {
      "epoch": 0.23964156053708294,
      "grad_norm": 0.893537163734436,
      "learning_rate": 8.891263926683382e-06,
      "loss": 0.7857,
      "step": 8317
    },
    {
      "epoch": 0.2396703739987322,
      "grad_norm": 0.9093045592308044,
      "learning_rate": 8.890970901390927e-06,
      "loss": 0.7738,
      "step": 8318
    },
    {
      "epoch": 0.2396991874603815,
      "grad_norm": 0.874118983745575,
      "learning_rate": 8.890677842212005e-06,
      "loss": 0.7897,
      "step": 8319
    },
    {
      "epoch": 0.23972800092203078,
      "grad_norm": 0.9092429280281067,
      "learning_rate": 8.890384749149168e-06,
      "loss": 0.7615,
      "step": 8320
    },
    {
      "epoch": 0.23975681438368004,
      "grad_norm": 0.8888942003250122,
      "learning_rate": 8.890091622204969e-06,
      "loss": 0.7651,
      "step": 8321
    },
    {
      "epoch": 0.23978562784532934,
      "grad_norm": 0.8934257626533508,
      "learning_rate": 8.88979846138196e-06,
      "loss": 0.7817,
      "step": 8322
    },
    {
      "epoch": 0.23981444130697863,
      "grad_norm": 0.8944334387779236,
      "learning_rate": 8.889505266682696e-06,
      "loss": 0.7685,
      "step": 8323
    },
    {
      "epoch": 0.2398432547686279,
      "grad_norm": 0.9113098978996277,
      "learning_rate": 8.889212038109729e-06,
      "loss": 0.7543,
      "step": 8324
    },
    {
      "epoch": 0.23987206823027718,
      "grad_norm": 0.9078351855278015,
      "learning_rate": 8.888918775665612e-06,
      "loss": 0.7741,
      "step": 8325
    },
    {
      "epoch": 0.23990088169192647,
      "grad_norm": 0.9262397289276123,
      "learning_rate": 8.8886254793529e-06,
      "loss": 0.7607,
      "step": 8326
    },
    {
      "epoch": 0.23992969515357576,
      "grad_norm": 0.9181671738624573,
      "learning_rate": 8.888332149174149e-06,
      "loss": 0.7687,
      "step": 8327
    },
    {
      "epoch": 0.23995850861522502,
      "grad_norm": 0.884796679019928,
      "learning_rate": 8.88803878513191e-06,
      "loss": 0.7773,
      "step": 8328
    },
    {
      "epoch": 0.23998732207687432,
      "grad_norm": 0.8765462636947632,
      "learning_rate": 8.887745387228741e-06,
      "loss": 0.7514,
      "step": 8329
    },
    {
      "epoch": 0.2400161355385236,
      "grad_norm": 0.9163505434989929,
      "learning_rate": 8.887451955467195e-06,
      "loss": 0.7669,
      "step": 8330
    },
    {
      "epoch": 0.24004494900017287,
      "grad_norm": 0.9164249300956726,
      "learning_rate": 8.88715848984983e-06,
      "loss": 0.7777,
      "step": 8331
    },
    {
      "epoch": 0.24007376246182216,
      "grad_norm": 0.8729758262634277,
      "learning_rate": 8.886864990379199e-06,
      "loss": 0.7715,
      "step": 8332
    },
    {
      "epoch": 0.24010257592347145,
      "grad_norm": 0.9044448733329773,
      "learning_rate": 8.88657145705786e-06,
      "loss": 0.7572,
      "step": 8333
    },
    {
      "epoch": 0.24013138938512074,
      "grad_norm": 0.9186785817146301,
      "learning_rate": 8.886277889888368e-06,
      "loss": 0.775,
      "step": 8334
    },
    {
      "epoch": 0.24016020284677,
      "grad_norm": 0.8624826669692993,
      "learning_rate": 8.88598428887328e-06,
      "loss": 0.7541,
      "step": 8335
    },
    {
      "epoch": 0.2401890163084193,
      "grad_norm": 0.9104354977607727,
      "learning_rate": 8.885690654015155e-06,
      "loss": 0.7821,
      "step": 8336
    },
    {
      "epoch": 0.24021782977006859,
      "grad_norm": 0.9353364109992981,
      "learning_rate": 8.885396985316548e-06,
      "loss": 0.7503,
      "step": 8337
    },
    {
      "epoch": 0.24024664323171785,
      "grad_norm": 0.9043651819229126,
      "learning_rate": 8.885103282780017e-06,
      "loss": 0.7716,
      "step": 8338
    },
    {
      "epoch": 0.24027545669336714,
      "grad_norm": 0.9479696750640869,
      "learning_rate": 8.88480954640812e-06,
      "loss": 0.7986,
      "step": 8339
    },
    {
      "epoch": 0.24030427015501643,
      "grad_norm": 0.9273532032966614,
      "learning_rate": 8.884515776203416e-06,
      "loss": 0.7602,
      "step": 8340
    },
    {
      "epoch": 0.2403330836166657,
      "grad_norm": 0.9110715985298157,
      "learning_rate": 8.884221972168461e-06,
      "loss": 0.7576,
      "step": 8341
    },
    {
      "epoch": 0.24036189707831498,
      "grad_norm": 0.9641653299331665,
      "learning_rate": 8.883928134305817e-06,
      "loss": 0.7588,
      "step": 8342
    },
    {
      "epoch": 0.24039071053996428,
      "grad_norm": 0.9298924803733826,
      "learning_rate": 8.883634262618041e-06,
      "loss": 0.7642,
      "step": 8343
    },
    {
      "epoch": 0.24041952400161357,
      "grad_norm": 0.9241784811019897,
      "learning_rate": 8.883340357107693e-06,
      "loss": 0.7763,
      "step": 8344
    },
    {
      "epoch": 0.24044833746326283,
      "grad_norm": 0.9134669303894043,
      "learning_rate": 8.883046417777333e-06,
      "loss": 0.7516,
      "step": 8345
    },
    {
      "epoch": 0.24047715092491212,
      "grad_norm": 0.891338050365448,
      "learning_rate": 8.882752444629518e-06,
      "loss": 0.7793,
      "step": 8346
    },
    {
      "epoch": 0.2405059643865614,
      "grad_norm": 0.9007371068000793,
      "learning_rate": 8.882458437666813e-06,
      "loss": 0.7998,
      "step": 8347
    },
    {
      "epoch": 0.24053477784821067,
      "grad_norm": 0.9159932136535645,
      "learning_rate": 8.882164396891774e-06,
      "loss": 0.7901,
      "step": 8348
    },
    {
      "epoch": 0.24056359130985996,
      "grad_norm": 0.8624531030654907,
      "learning_rate": 8.881870322306967e-06,
      "loss": 0.764,
      "step": 8349
    },
    {
      "epoch": 0.24059240477150926,
      "grad_norm": 0.9373149871826172,
      "learning_rate": 8.881576213914948e-06,
      "loss": 0.7902,
      "step": 8350
    },
    {
      "epoch": 0.24062121823315852,
      "grad_norm": 0.8805795907974243,
      "learning_rate": 8.88128207171828e-06,
      "loss": 0.7581,
      "step": 8351
    },
    {
      "epoch": 0.2406500316948078,
      "grad_norm": 0.9249997138977051,
      "learning_rate": 8.880987895719525e-06,
      "loss": 0.7816,
      "step": 8352
    },
    {
      "epoch": 0.2406788451564571,
      "grad_norm": 0.9076758623123169,
      "learning_rate": 8.880693685921245e-06,
      "loss": 0.7616,
      "step": 8353
    },
    {
      "epoch": 0.2407076586181064,
      "grad_norm": 0.9529888033866882,
      "learning_rate": 8.880399442326002e-06,
      "loss": 0.7825,
      "step": 8354
    },
    {
      "epoch": 0.24073647207975565,
      "grad_norm": 0.9288215637207031,
      "learning_rate": 8.88010516493636e-06,
      "loss": 0.7958,
      "step": 8355
    },
    {
      "epoch": 0.24076528554140494,
      "grad_norm": 0.9230200052261353,
      "learning_rate": 8.87981085375488e-06,
      "loss": 0.7796,
      "step": 8356
    },
    {
      "epoch": 0.24079409900305423,
      "grad_norm": 0.9490646719932556,
      "learning_rate": 8.879516508784125e-06,
      "loss": 0.7904,
      "step": 8357
    },
    {
      "epoch": 0.2408229124647035,
      "grad_norm": 0.9575337767601013,
      "learning_rate": 8.87922213002666e-06,
      "loss": 0.793,
      "step": 8358
    },
    {
      "epoch": 0.2408517259263528,
      "grad_norm": 0.9095544815063477,
      "learning_rate": 8.87892771748505e-06,
      "loss": 0.7826,
      "step": 8359
    },
    {
      "epoch": 0.24088053938800208,
      "grad_norm": 0.8980303406715393,
      "learning_rate": 8.878633271161855e-06,
      "loss": 0.7712,
      "step": 8360
    },
    {
      "epoch": 0.24090935284965137,
      "grad_norm": 0.8609490394592285,
      "learning_rate": 8.878338791059641e-06,
      "loss": 0.7707,
      "step": 8361
    },
    {
      "epoch": 0.24093816631130063,
      "grad_norm": 0.9151630401611328,
      "learning_rate": 8.878044277180975e-06,
      "loss": 0.7723,
      "step": 8362
    },
    {
      "epoch": 0.24096697977294992,
      "grad_norm": 0.8884391188621521,
      "learning_rate": 8.877749729528417e-06,
      "loss": 0.7497,
      "step": 8363
    },
    {
      "epoch": 0.24099579323459921,
      "grad_norm": 0.9098986387252808,
      "learning_rate": 8.877455148104539e-06,
      "loss": 0.7717,
      "step": 8364
    },
    {
      "epoch": 0.24102460669624848,
      "grad_norm": 0.9565921425819397,
      "learning_rate": 8.877160532911901e-06,
      "loss": 0.7835,
      "step": 8365
    },
    {
      "epoch": 0.24105342015789777,
      "grad_norm": 0.9595949649810791,
      "learning_rate": 8.87686588395307e-06,
      "loss": 0.7837,
      "step": 8366
    },
    {
      "epoch": 0.24108223361954706,
      "grad_norm": 0.9197466373443604,
      "learning_rate": 8.876571201230613e-06,
      "loss": 0.779,
      "step": 8367
    },
    {
      "epoch": 0.24111104708119632,
      "grad_norm": 0.8970045447349548,
      "learning_rate": 8.876276484747096e-06,
      "loss": 0.7705,
      "step": 8368
    },
    {
      "epoch": 0.2411398605428456,
      "grad_norm": 0.9335793256759644,
      "learning_rate": 8.875981734505083e-06,
      "loss": 0.7551,
      "step": 8369
    },
    {
      "epoch": 0.2411686740044949,
      "grad_norm": 0.9009377956390381,
      "learning_rate": 8.875686950507148e-06,
      "loss": 0.7665,
      "step": 8370
    },
    {
      "epoch": 0.2411974874661442,
      "grad_norm": 0.9235032796859741,
      "learning_rate": 8.875392132755851e-06,
      "loss": 0.7826,
      "step": 8371
    },
    {
      "epoch": 0.24122630092779346,
      "grad_norm": 0.938936173915863,
      "learning_rate": 8.875097281253764e-06,
      "loss": 0.7828,
      "step": 8372
    },
    {
      "epoch": 0.24125511438944275,
      "grad_norm": 0.9343746900558472,
      "learning_rate": 8.874802396003452e-06,
      "loss": 0.7708,
      "step": 8373
    },
    {
      "epoch": 0.24128392785109204,
      "grad_norm": 0.868697464466095,
      "learning_rate": 8.874507477007487e-06,
      "loss": 0.7627,
      "step": 8374
    },
    {
      "epoch": 0.2413127413127413,
      "grad_norm": 0.9660394787788391,
      "learning_rate": 8.874212524268432e-06,
      "loss": 0.7686,
      "step": 8375
    },
    {
      "epoch": 0.2413415547743906,
      "grad_norm": 0.9344559907913208,
      "learning_rate": 8.873917537788862e-06,
      "loss": 0.7865,
      "step": 8376
    },
    {
      "epoch": 0.24137036823603988,
      "grad_norm": 0.890532910823822,
      "learning_rate": 8.87362251757134e-06,
      "loss": 0.7658,
      "step": 8377
    },
    {
      "epoch": 0.24139918169768915,
      "grad_norm": 0.9445977210998535,
      "learning_rate": 8.873327463618438e-06,
      "loss": 0.7593,
      "step": 8378
    },
    {
      "epoch": 0.24142799515933844,
      "grad_norm": 0.897707462310791,
      "learning_rate": 8.873032375932728e-06,
      "loss": 0.7619,
      "step": 8379
    },
    {
      "epoch": 0.24145680862098773,
      "grad_norm": 0.8697946071624756,
      "learning_rate": 8.872737254516777e-06,
      "loss": 0.7757,
      "step": 8380
    },
    {
      "epoch": 0.24148562208263702,
      "grad_norm": 0.8503908514976501,
      "learning_rate": 8.872442099373155e-06,
      "loss": 0.7553,
      "step": 8381
    },
    {
      "epoch": 0.24151443554428628,
      "grad_norm": 0.8906696438789368,
      "learning_rate": 8.872146910504433e-06,
      "loss": 0.7742,
      "step": 8382
    },
    {
      "epoch": 0.24154324900593557,
      "grad_norm": 0.8708488345146179,
      "learning_rate": 8.871851687913181e-06,
      "loss": 0.7455,
      "step": 8383
    },
    {
      "epoch": 0.24157206246758486,
      "grad_norm": 0.9126395583152771,
      "learning_rate": 8.871556431601974e-06,
      "loss": 0.7736,
      "step": 8384
    },
    {
      "epoch": 0.24160087592923413,
      "grad_norm": 0.8930805921554565,
      "learning_rate": 8.871261141573379e-06,
      "loss": 0.7759,
      "step": 8385
    },
    {
      "epoch": 0.24162968939088342,
      "grad_norm": 0.8624604940414429,
      "learning_rate": 8.870965817829971e-06,
      "loss": 0.7706,
      "step": 8386
    },
    {
      "epoch": 0.2416585028525327,
      "grad_norm": 0.8634228706359863,
      "learning_rate": 8.870670460374317e-06,
      "loss": 0.7631,
      "step": 8387
    },
    {
      "epoch": 0.241687316314182,
      "grad_norm": 0.8630176782608032,
      "learning_rate": 8.870375069208996e-06,
      "loss": 0.7604,
      "step": 8388
    },
    {
      "epoch": 0.24171612977583126,
      "grad_norm": 0.9025815725326538,
      "learning_rate": 8.870079644336575e-06,
      "loss": 0.7792,
      "step": 8389
    },
    {
      "epoch": 0.24174494323748055,
      "grad_norm": 0.8774878978729248,
      "learning_rate": 8.869784185759628e-06,
      "loss": 0.7597,
      "step": 8390
    },
    {
      "epoch": 0.24177375669912984,
      "grad_norm": 0.8772242665290833,
      "learning_rate": 8.869488693480732e-06,
      "loss": 0.7819,
      "step": 8391
    },
    {
      "epoch": 0.2418025701607791,
      "grad_norm": 0.875603973865509,
      "learning_rate": 8.869193167502457e-06,
      "loss": 0.7629,
      "step": 8392
    },
    {
      "epoch": 0.2418313836224284,
      "grad_norm": 0.8825379014015198,
      "learning_rate": 8.868897607827376e-06,
      "loss": 0.7615,
      "step": 8393
    },
    {
      "epoch": 0.2418601970840777,
      "grad_norm": 0.9194895029067993,
      "learning_rate": 8.868602014458065e-06,
      "loss": 0.7738,
      "step": 8394
    },
    {
      "epoch": 0.24188901054572695,
      "grad_norm": 0.9114146828651428,
      "learning_rate": 8.868306387397097e-06,
      "loss": 0.7981,
      "step": 8395
    },
    {
      "epoch": 0.24191782400737624,
      "grad_norm": 0.8656824231147766,
      "learning_rate": 8.868010726647049e-06,
      "loss": 0.7555,
      "step": 8396
    },
    {
      "epoch": 0.24194663746902553,
      "grad_norm": 0.964066743850708,
      "learning_rate": 8.867715032210492e-06,
      "loss": 0.7601,
      "step": 8397
    },
    {
      "epoch": 0.24197545093067482,
      "grad_norm": 0.894767701625824,
      "learning_rate": 8.867419304090005e-06,
      "loss": 0.7708,
      "step": 8398
    },
    {
      "epoch": 0.2420042643923241,
      "grad_norm": 0.9316755533218384,
      "learning_rate": 8.86712354228816e-06,
      "loss": 0.7706,
      "step": 8399
    },
    {
      "epoch": 0.24203307785397338,
      "grad_norm": 0.9282196164131165,
      "learning_rate": 8.866827746807537e-06,
      "loss": 0.7922,
      "step": 8400
    },
    {
      "epoch": 0.24206189131562267,
      "grad_norm": 0.9125849008560181,
      "learning_rate": 8.866531917650708e-06,
      "loss": 0.7917,
      "step": 8401
    },
    {
      "epoch": 0.24209070477727193,
      "grad_norm": 0.8647530674934387,
      "learning_rate": 8.866236054820252e-06,
      "loss": 0.7437,
      "step": 8402
    },
    {
      "epoch": 0.24211951823892122,
      "grad_norm": 0.9087411165237427,
      "learning_rate": 8.865940158318744e-06,
      "loss": 0.7708,
      "step": 8403
    },
    {
      "epoch": 0.2421483317005705,
      "grad_norm": 0.938468337059021,
      "learning_rate": 8.865644228148762e-06,
      "loss": 0.7835,
      "step": 8404
    },
    {
      "epoch": 0.24217714516221978,
      "grad_norm": 4.37237024307251,
      "learning_rate": 8.865348264312883e-06,
      "loss": 0.7609,
      "step": 8405
    },
    {
      "epoch": 0.24220595862386907,
      "grad_norm": 0.9242699146270752,
      "learning_rate": 8.865052266813686e-06,
      "loss": 0.7618,
      "step": 8406
    },
    {
      "epoch": 0.24223477208551836,
      "grad_norm": 0.9040055871009827,
      "learning_rate": 8.864756235653746e-06,
      "loss": 0.7657,
      "step": 8407
    },
    {
      "epoch": 0.24226358554716765,
      "grad_norm": 0.9399797916412354,
      "learning_rate": 8.864460170835643e-06,
      "loss": 0.7931,
      "step": 8408
    },
    {
      "epoch": 0.2422923990088169,
      "grad_norm": 0.909410834312439,
      "learning_rate": 8.864164072361954e-06,
      "loss": 0.7651,
      "step": 8409
    },
    {
      "epoch": 0.2423212124704662,
      "grad_norm": 1.2953722476959229,
      "learning_rate": 8.86386794023526e-06,
      "loss": 0.7724,
      "step": 8410
    },
    {
      "epoch": 0.2423500259321155,
      "grad_norm": 0.9226974248886108,
      "learning_rate": 8.863571774458138e-06,
      "loss": 0.7829,
      "step": 8411
    },
    {
      "epoch": 0.24237883939376476,
      "grad_norm": 0.9569065570831299,
      "learning_rate": 8.863275575033168e-06,
      "loss": 0.7853,
      "step": 8412
    },
    {
      "epoch": 0.24240765285541405,
      "grad_norm": 0.8862513899803162,
      "learning_rate": 8.862979341962929e-06,
      "loss": 0.7409,
      "step": 8413
    },
    {
      "epoch": 0.24243646631706334,
      "grad_norm": 0.8811537623405457,
      "learning_rate": 8.862683075250003e-06,
      "loss": 0.7815,
      "step": 8414
    },
    {
      "epoch": 0.24246527977871263,
      "grad_norm": 0.92403244972229,
      "learning_rate": 8.862386774896967e-06,
      "loss": 0.7459,
      "step": 8415
    },
    {
      "epoch": 0.2424940932403619,
      "grad_norm": 0.9284659028053284,
      "learning_rate": 8.862090440906403e-06,
      "loss": 0.758,
      "step": 8416
    },
    {
      "epoch": 0.24252290670201118,
      "grad_norm": 0.8955702185630798,
      "learning_rate": 8.861794073280893e-06,
      "loss": 0.7449,
      "step": 8417
    },
    {
      "epoch": 0.24255172016366047,
      "grad_norm": 0.905312716960907,
      "learning_rate": 8.861497672023018e-06,
      "loss": 0.7625,
      "step": 8418
    },
    {
      "epoch": 0.24258053362530974,
      "grad_norm": 1.0255581140518188,
      "learning_rate": 8.861201237135358e-06,
      "loss": 0.7929,
      "step": 8419
    },
    {
      "epoch": 0.24260934708695903,
      "grad_norm": 0.9076933264732361,
      "learning_rate": 8.860904768620493e-06,
      "loss": 0.7644,
      "step": 8420
    },
    {
      "epoch": 0.24263816054860832,
      "grad_norm": 0.908427357673645,
      "learning_rate": 8.860608266481008e-06,
      "loss": 0.7668,
      "step": 8421
    },
    {
      "epoch": 0.24266697401025758,
      "grad_norm": 0.9318297505378723,
      "learning_rate": 8.860311730719485e-06,
      "loss": 0.7969,
      "step": 8422
    },
    {
      "epoch": 0.24269578747190687,
      "grad_norm": 0.8693996667861938,
      "learning_rate": 8.860015161338505e-06,
      "loss": 0.759,
      "step": 8423
    },
    {
      "epoch": 0.24272460093355616,
      "grad_norm": 1.0721896886825562,
      "learning_rate": 8.859718558340653e-06,
      "loss": 0.7794,
      "step": 8424
    },
    {
      "epoch": 0.24275341439520545,
      "grad_norm": 0.9086847305297852,
      "learning_rate": 8.859421921728508e-06,
      "loss": 0.7685,
      "step": 8425
    },
    {
      "epoch": 0.24278222785685472,
      "grad_norm": 0.8786080479621887,
      "learning_rate": 8.859125251504658e-06,
      "loss": 0.7434,
      "step": 8426
    },
    {
      "epoch": 0.242811041318504,
      "grad_norm": 0.8892868161201477,
      "learning_rate": 8.858828547671684e-06,
      "loss": 0.7681,
      "step": 8427
    },
    {
      "epoch": 0.2428398547801533,
      "grad_norm": 0.8684629797935486,
      "learning_rate": 8.85853181023217e-06,
      "loss": 0.7629,
      "step": 8428
    },
    {
      "epoch": 0.24286866824180256,
      "grad_norm": 0.8662788271903992,
      "learning_rate": 8.858235039188703e-06,
      "loss": 0.7929,
      "step": 8429
    },
    {
      "epoch": 0.24289748170345185,
      "grad_norm": 0.8985726237297058,
      "learning_rate": 8.857938234543866e-06,
      "loss": 0.7814,
      "step": 8430
    },
    {
      "epoch": 0.24292629516510114,
      "grad_norm": 0.9002996683120728,
      "learning_rate": 8.857641396300241e-06,
      "loss": 0.7977,
      "step": 8431
    },
    {
      "epoch": 0.2429551086267504,
      "grad_norm": 0.8516778945922852,
      "learning_rate": 8.857344524460417e-06,
      "loss": 0.7763,
      "step": 8432
    },
    {
      "epoch": 0.2429839220883997,
      "grad_norm": 0.8989262580871582,
      "learning_rate": 8.857047619026977e-06,
      "loss": 0.7724,
      "step": 8433
    },
    {
      "epoch": 0.24301273555004899,
      "grad_norm": 0.8665566444396973,
      "learning_rate": 8.85675068000251e-06,
      "loss": 0.7791,
      "step": 8434
    },
    {
      "epoch": 0.24304154901169828,
      "grad_norm": 0.9047931432723999,
      "learning_rate": 8.856453707389598e-06,
      "loss": 0.7539,
      "step": 8435
    },
    {
      "epoch": 0.24307036247334754,
      "grad_norm": 0.889629065990448,
      "learning_rate": 8.85615670119083e-06,
      "loss": 0.765,
      "step": 8436
    },
    {
      "epoch": 0.24309917593499683,
      "grad_norm": 0.8863012194633484,
      "learning_rate": 8.855859661408793e-06,
      "loss": 0.7577,
      "step": 8437
    },
    {
      "epoch": 0.24312798939664612,
      "grad_norm": 0.9157271981239319,
      "learning_rate": 8.855562588046072e-06,
      "loss": 0.7563,
      "step": 8438
    },
    {
      "epoch": 0.24315680285829538,
      "grad_norm": 0.8729559183120728,
      "learning_rate": 8.855265481105253e-06,
      "loss": 0.75,
      "step": 8439
    },
    {
      "epoch": 0.24318561631994468,
      "grad_norm": 0.90968918800354,
      "learning_rate": 8.854968340588927e-06,
      "loss": 0.7579,
      "step": 8440
    },
    {
      "epoch": 0.24321442978159397,
      "grad_norm": 0.8912925720214844,
      "learning_rate": 8.854671166499682e-06,
      "loss": 0.7848,
      "step": 8441
    },
    {
      "epoch": 0.24324324324324326,
      "grad_norm": 0.880148708820343,
      "learning_rate": 8.854373958840103e-06,
      "loss": 0.7861,
      "step": 8442
    },
    {
      "epoch": 0.24327205670489252,
      "grad_norm": 0.8848557472229004,
      "learning_rate": 8.854076717612779e-06,
      "loss": 0.754,
      "step": 8443
    },
    {
      "epoch": 0.2433008701665418,
      "grad_norm": 0.8607282042503357,
      "learning_rate": 8.853779442820301e-06,
      "loss": 0.7645,
      "step": 8444
    },
    {
      "epoch": 0.2433296836281911,
      "grad_norm": 0.8697222471237183,
      "learning_rate": 8.853482134465254e-06,
      "loss": 0.7331,
      "step": 8445
    },
    {
      "epoch": 0.24335849708984036,
      "grad_norm": 0.8537964224815369,
      "learning_rate": 8.853184792550233e-06,
      "loss": 0.7307,
      "step": 8446
    },
    {
      "epoch": 0.24338731055148966,
      "grad_norm": 0.9238186478614807,
      "learning_rate": 8.852887417077823e-06,
      "loss": 0.7723,
      "step": 8447
    },
    {
      "epoch": 0.24341612401313895,
      "grad_norm": 0.8872237801551819,
      "learning_rate": 8.852590008050613e-06,
      "loss": 0.7619,
      "step": 8448
    },
    {
      "epoch": 0.2434449374747882,
      "grad_norm": 0.9143684506416321,
      "learning_rate": 8.852292565471196e-06,
      "loss": 0.7905,
      "step": 8449
    },
    {
      "epoch": 0.2434737509364375,
      "grad_norm": 0.8988944292068481,
      "learning_rate": 8.851995089342163e-06,
      "loss": 0.7778,
      "step": 8450
    },
    {
      "epoch": 0.2435025643980868,
      "grad_norm": 0.9034966230392456,
      "learning_rate": 8.851697579666103e-06,
      "loss": 0.7663,
      "step": 8451
    },
    {
      "epoch": 0.24353137785973608,
      "grad_norm": 0.8955353498458862,
      "learning_rate": 8.851400036445608e-06,
      "loss": 0.7918,
      "step": 8452
    },
    {
      "epoch": 0.24356019132138534,
      "grad_norm": 0.8828951120376587,
      "learning_rate": 8.851102459683267e-06,
      "loss": 0.7718,
      "step": 8453
    },
    {
      "epoch": 0.24358900478303463,
      "grad_norm": 0.8656241297721863,
      "learning_rate": 8.850804849381674e-06,
      "loss": 0.7732,
      "step": 8454
    },
    {
      "epoch": 0.24361781824468393,
      "grad_norm": 0.8698406219482422,
      "learning_rate": 8.85050720554342e-06,
      "loss": 0.7511,
      "step": 8455
    },
    {
      "epoch": 0.2436466317063332,
      "grad_norm": 0.9016057252883911,
      "learning_rate": 8.8502095281711e-06,
      "loss": 0.7689,
      "step": 8456
    },
    {
      "epoch": 0.24367544516798248,
      "grad_norm": 0.8866444826126099,
      "learning_rate": 8.8499118172673e-06,
      "loss": 0.7788,
      "step": 8457
    },
    {
      "epoch": 0.24370425862963177,
      "grad_norm": 0.9166905283927917,
      "learning_rate": 8.849614072834617e-06,
      "loss": 0.7567,
      "step": 8458
    },
    {
      "epoch": 0.24373307209128103,
      "grad_norm": 0.8952310681343079,
      "learning_rate": 8.849316294875646e-06,
      "loss": 0.7629,
      "step": 8459
    },
    {
      "epoch": 0.24376188555293032,
      "grad_norm": 0.8999898433685303,
      "learning_rate": 8.849018483392976e-06,
      "loss": 0.7573,
      "step": 8460
    },
    {
      "epoch": 0.24379069901457961,
      "grad_norm": 0.9378122687339783,
      "learning_rate": 8.848720638389203e-06,
      "loss": 0.7747,
      "step": 8461
    },
    {
      "epoch": 0.2438195124762289,
      "grad_norm": 0.905975341796875,
      "learning_rate": 8.84842275986692e-06,
      "loss": 0.7761,
      "step": 8462
    },
    {
      "epoch": 0.24384832593787817,
      "grad_norm": 0.9100976586341858,
      "learning_rate": 8.848124847828723e-06,
      "loss": 0.782,
      "step": 8463
    },
    {
      "epoch": 0.24387713939952746,
      "grad_norm": 0.9071996212005615,
      "learning_rate": 8.847826902277205e-06,
      "loss": 0.7831,
      "step": 8464
    },
    {
      "epoch": 0.24390595286117675,
      "grad_norm": 0.9152718782424927,
      "learning_rate": 8.847528923214961e-06,
      "loss": 0.76,
      "step": 8465
    },
    {
      "epoch": 0.243934766322826,
      "grad_norm": 0.9040888547897339,
      "learning_rate": 8.847230910644586e-06,
      "loss": 0.7663,
      "step": 8466
    },
    {
      "epoch": 0.2439635797844753,
      "grad_norm": 0.9089508056640625,
      "learning_rate": 8.846932864568676e-06,
      "loss": 0.7588,
      "step": 8467
    },
    {
      "epoch": 0.2439923932461246,
      "grad_norm": 0.8758925795555115,
      "learning_rate": 8.846634784989826e-06,
      "loss": 0.753,
      "step": 8468
    },
    {
      "epoch": 0.24402120670777389,
      "grad_norm": 0.9192883372306824,
      "learning_rate": 8.846336671910632e-06,
      "loss": 0.7909,
      "step": 8469
    },
    {
      "epoch": 0.24405002016942315,
      "grad_norm": 0.9128542542457581,
      "learning_rate": 8.846038525333692e-06,
      "loss": 0.7495,
      "step": 8470
    },
    {
      "epoch": 0.24407883363107244,
      "grad_norm": 0.8783330917358398,
      "learning_rate": 8.845740345261598e-06,
      "loss": 0.7538,
      "step": 8471
    },
    {
      "epoch": 0.24410764709272173,
      "grad_norm": 0.9001603722572327,
      "learning_rate": 8.845442131696954e-06,
      "loss": 0.7507,
      "step": 8472
    },
    {
      "epoch": 0.244136460554371,
      "grad_norm": 0.8789632320404053,
      "learning_rate": 8.845143884642349e-06,
      "loss": 0.7747,
      "step": 8473
    },
    {
      "epoch": 0.24416527401602028,
      "grad_norm": 1.0297908782958984,
      "learning_rate": 8.844845604100386e-06,
      "loss": 0.7745,
      "step": 8474
    },
    {
      "epoch": 0.24419408747766957,
      "grad_norm": 0.9366621971130371,
      "learning_rate": 8.844547290073661e-06,
      "loss": 0.7639,
      "step": 8475
    },
    {
      "epoch": 0.24422290093931884,
      "grad_norm": 0.9193832278251648,
      "learning_rate": 8.844248942564774e-06,
      "loss": 0.7689,
      "step": 8476
    },
    {
      "epoch": 0.24425171440096813,
      "grad_norm": 0.900123655796051,
      "learning_rate": 8.84395056157632e-06,
      "loss": 0.7371,
      "step": 8477
    },
    {
      "epoch": 0.24428052786261742,
      "grad_norm": 0.8719863295555115,
      "learning_rate": 8.8436521471109e-06,
      "loss": 0.7502,
      "step": 8478
    },
    {
      "epoch": 0.2443093413242667,
      "grad_norm": 0.9089786410331726,
      "learning_rate": 8.84335369917111e-06,
      "loss": 0.7963,
      "step": 8479
    },
    {
      "epoch": 0.24433815478591597,
      "grad_norm": 0.8680189251899719,
      "learning_rate": 8.843055217759554e-06,
      "loss": 0.7678,
      "step": 8480
    },
    {
      "epoch": 0.24436696824756526,
      "grad_norm": 0.8980516195297241,
      "learning_rate": 8.842756702878828e-06,
      "loss": 0.7495,
      "step": 8481
    },
    {
      "epoch": 0.24439578170921455,
      "grad_norm": 0.8928244709968567,
      "learning_rate": 8.842458154531533e-06,
      "loss": 0.7822,
      "step": 8482
    },
    {
      "epoch": 0.24442459517086382,
      "grad_norm": 0.8777498006820679,
      "learning_rate": 8.842159572720266e-06,
      "loss": 0.7563,
      "step": 8483
    },
    {
      "epoch": 0.2444534086325131,
      "grad_norm": 0.9388493299484253,
      "learning_rate": 8.841860957447632e-06,
      "loss": 0.7704,
      "step": 8484
    },
    {
      "epoch": 0.2444822220941624,
      "grad_norm": 0.8953596949577332,
      "learning_rate": 8.841562308716228e-06,
      "loss": 0.7711,
      "step": 8485
    },
    {
      "epoch": 0.24451103555581166,
      "grad_norm": 0.9776483774185181,
      "learning_rate": 8.841263626528658e-06,
      "loss": 0.7808,
      "step": 8486
    },
    {
      "epoch": 0.24453984901746095,
      "grad_norm": 0.9051138758659363,
      "learning_rate": 8.84096491088752e-06,
      "loss": 0.7551,
      "step": 8487
    },
    {
      "epoch": 0.24456866247911024,
      "grad_norm": 0.9062541723251343,
      "learning_rate": 8.84066616179542e-06,
      "loss": 0.7673,
      "step": 8488
    },
    {
      "epoch": 0.24459747594075953,
      "grad_norm": 0.892219066619873,
      "learning_rate": 8.840367379254956e-06,
      "loss": 0.7749,
      "step": 8489
    },
    {
      "epoch": 0.2446262894024088,
      "grad_norm": 0.9281882643699646,
      "learning_rate": 8.840068563268729e-06,
      "loss": 0.7578,
      "step": 8490
    },
    {
      "epoch": 0.2446551028640581,
      "grad_norm": 0.9352681636810303,
      "learning_rate": 8.839769713839344e-06,
      "loss": 0.7821,
      "step": 8491
    },
    {
      "epoch": 0.24468391632570738,
      "grad_norm": 0.8953990340232849,
      "learning_rate": 8.839470830969403e-06,
      "loss": 0.7751,
      "step": 8492
    },
    {
      "epoch": 0.24471272978735664,
      "grad_norm": 0.902164101600647,
      "learning_rate": 8.83917191466151e-06,
      "loss": 0.75,
      "step": 8493
    },
    {
      "epoch": 0.24474154324900593,
      "grad_norm": 0.8747314810752869,
      "learning_rate": 8.838872964918268e-06,
      "loss": 0.7687,
      "step": 8494
    },
    {
      "epoch": 0.24477035671065522,
      "grad_norm": 0.8669959306716919,
      "learning_rate": 8.83857398174228e-06,
      "loss": 0.767,
      "step": 8495
    },
    {
      "epoch": 0.24479917017230451,
      "grad_norm": 0.9172487854957581,
      "learning_rate": 8.838274965136149e-06,
      "loss": 0.7782,
      "step": 8496
    },
    {
      "epoch": 0.24482798363395378,
      "grad_norm": 0.9500084519386292,
      "learning_rate": 8.83797591510248e-06,
      "loss": 0.7712,
      "step": 8497
    },
    {
      "epoch": 0.24485679709560307,
      "grad_norm": 0.8637178540229797,
      "learning_rate": 8.837676831643878e-06,
      "loss": 0.768,
      "step": 8498
    },
    {
      "epoch": 0.24488561055725236,
      "grad_norm": 0.8625984191894531,
      "learning_rate": 8.837377714762945e-06,
      "loss": 0.7535,
      "step": 8499
    },
    {
      "epoch": 0.24491442401890162,
      "grad_norm": 0.8786967992782593,
      "learning_rate": 8.83707856446229e-06,
      "loss": 0.7625,
      "step": 8500
    },
    {
      "epoch": 0.2449432374805509,
      "grad_norm": 0.8545451760292053,
      "learning_rate": 8.836779380744516e-06,
      "loss": 0.7613,
      "step": 8501
    },
    {
      "epoch": 0.2449720509422002,
      "grad_norm": 0.8636351823806763,
      "learning_rate": 8.83648016361223e-06,
      "loss": 0.7567,
      "step": 8502
    },
    {
      "epoch": 0.24500086440384947,
      "grad_norm": 0.8940028548240662,
      "learning_rate": 8.836180913068036e-06,
      "loss": 0.7648,
      "step": 8503
    },
    {
      "epoch": 0.24502967786549876,
      "grad_norm": 0.8863698840141296,
      "learning_rate": 8.83588162911454e-06,
      "loss": 0.8084,
      "step": 8504
    },
    {
      "epoch": 0.24505849132714805,
      "grad_norm": 0.8859768509864807,
      "learning_rate": 8.835582311754349e-06,
      "loss": 0.754,
      "step": 8505
    },
    {
      "epoch": 0.24508730478879734,
      "grad_norm": 0.8791670203208923,
      "learning_rate": 8.835282960990072e-06,
      "loss": 0.7759,
      "step": 8506
    },
    {
      "epoch": 0.2451161182504466,
      "grad_norm": 0.8497312664985657,
      "learning_rate": 8.834983576824314e-06,
      "loss": 0.7538,
      "step": 8507
    },
    {
      "epoch": 0.2451449317120959,
      "grad_norm": 0.9257352948188782,
      "learning_rate": 8.834684159259683e-06,
      "loss": 0.7786,
      "step": 8508
    },
    {
      "epoch": 0.24517374517374518,
      "grad_norm": 0.8983699679374695,
      "learning_rate": 8.834384708298785e-06,
      "loss": 0.7787,
      "step": 8509
    },
    {
      "epoch": 0.24520255863539445,
      "grad_norm": 0.9004577994346619,
      "learning_rate": 8.83408522394423e-06,
      "loss": 0.7702,
      "step": 8510
    },
    {
      "epoch": 0.24523137209704374,
      "grad_norm": 0.8745080232620239,
      "learning_rate": 8.833785706198625e-06,
      "loss": 0.7624,
      "step": 8511
    },
    {
      "epoch": 0.24526018555869303,
      "grad_norm": 0.8614724278450012,
      "learning_rate": 8.83348615506458e-06,
      "loss": 0.7565,
      "step": 8512
    },
    {
      "epoch": 0.2452889990203423,
      "grad_norm": 0.8768879175186157,
      "learning_rate": 8.8331865705447e-06,
      "loss": 0.7778,
      "step": 8513
    },
    {
      "epoch": 0.24531781248199158,
      "grad_norm": 0.9217856526374817,
      "learning_rate": 8.8328869526416e-06,
      "loss": 0.7613,
      "step": 8514
    },
    {
      "epoch": 0.24534662594364087,
      "grad_norm": 0.8851386308670044,
      "learning_rate": 8.832587301357884e-06,
      "loss": 0.7829,
      "step": 8515
    },
    {
      "epoch": 0.24537543940529016,
      "grad_norm": 0.9074423909187317,
      "learning_rate": 8.832287616696165e-06,
      "loss": 0.7719,
      "step": 8516
    },
    {
      "epoch": 0.24540425286693943,
      "grad_norm": 0.898711621761322,
      "learning_rate": 8.831987898659052e-06,
      "loss": 0.7741,
      "step": 8517
    },
    {
      "epoch": 0.24543306632858872,
      "grad_norm": 0.928379237651825,
      "learning_rate": 8.831688147249154e-06,
      "loss": 0.7888,
      "step": 8518
    },
    {
      "epoch": 0.245461879790238,
      "grad_norm": 0.8907781839370728,
      "learning_rate": 8.831388362469082e-06,
      "loss": 0.7841,
      "step": 8519
    },
    {
      "epoch": 0.24549069325188727,
      "grad_norm": 0.9465113282203674,
      "learning_rate": 8.831088544321449e-06,
      "loss": 0.7589,
      "step": 8520
    },
    {
      "epoch": 0.24551950671353656,
      "grad_norm": 0.8778803944587708,
      "learning_rate": 8.830788692808863e-06,
      "loss": 0.7554,
      "step": 8521
    },
    {
      "epoch": 0.24554832017518585,
      "grad_norm": 0.8867486715316772,
      "learning_rate": 8.830488807933937e-06,
      "loss": 0.7596,
      "step": 8522
    },
    {
      "epoch": 0.24557713363683514,
      "grad_norm": 0.8926783204078674,
      "learning_rate": 8.830188889699282e-06,
      "loss": 0.7931,
      "step": 8523
    },
    {
      "epoch": 0.2456059470984844,
      "grad_norm": 0.8951761722564697,
      "learning_rate": 8.829888938107514e-06,
      "loss": 0.7566,
      "step": 8524
    },
    {
      "epoch": 0.2456347605601337,
      "grad_norm": 0.894331157207489,
      "learning_rate": 8.829588953161239e-06,
      "loss": 0.7503,
      "step": 8525
    },
    {
      "epoch": 0.245663574021783,
      "grad_norm": 0.8797167539596558,
      "learning_rate": 8.829288934863072e-06,
      "loss": 0.7536,
      "step": 8526
    },
    {
      "epoch": 0.24569238748343225,
      "grad_norm": 0.8777350187301636,
      "learning_rate": 8.828988883215628e-06,
      "loss": 0.7763,
      "step": 8527
    },
    {
      "epoch": 0.24572120094508154,
      "grad_norm": 0.9077826142311096,
      "learning_rate": 8.828688798221516e-06,
      "loss": 0.7694,
      "step": 8528
    },
    {
      "epoch": 0.24575001440673083,
      "grad_norm": 0.8733097314834595,
      "learning_rate": 8.828388679883352e-06,
      "loss": 0.7911,
      "step": 8529
    },
    {
      "epoch": 0.2457788278683801,
      "grad_norm": 0.8581761717796326,
      "learning_rate": 8.828088528203754e-06,
      "loss": 0.7566,
      "step": 8530
    },
    {
      "epoch": 0.24580764133002939,
      "grad_norm": 0.9122238159179688,
      "learning_rate": 8.827788343185327e-06,
      "loss": 0.7781,
      "step": 8531
    },
    {
      "epoch": 0.24583645479167868,
      "grad_norm": 0.9004707336425781,
      "learning_rate": 8.827488124830691e-06,
      "loss": 0.7657,
      "step": 8532
    },
    {
      "epoch": 0.24586526825332797,
      "grad_norm": 0.8356572985649109,
      "learning_rate": 8.82718787314246e-06,
      "loss": 0.7675,
      "step": 8533
    },
    {
      "epoch": 0.24589408171497723,
      "grad_norm": 0.9138839840888977,
      "learning_rate": 8.826887588123249e-06,
      "loss": 0.7975,
      "step": 8534
    },
    {
      "epoch": 0.24592289517662652,
      "grad_norm": 0.9014933109283447,
      "learning_rate": 8.826587269775673e-06,
      "loss": 0.7905,
      "step": 8535
    },
    {
      "epoch": 0.2459517086382758,
      "grad_norm": 0.8993445038795471,
      "learning_rate": 8.826286918102344e-06,
      "loss": 0.7873,
      "step": 8536
    },
    {
      "epoch": 0.24598052209992508,
      "grad_norm": 0.8957653045654297,
      "learning_rate": 8.825986533105884e-06,
      "loss": 0.746,
      "step": 8537
    },
    {
      "epoch": 0.24600933556157437,
      "grad_norm": 0.8818634748458862,
      "learning_rate": 8.825686114788905e-06,
      "loss": 0.77,
      "step": 8538
    },
    {
      "epoch": 0.24603814902322366,
      "grad_norm": 0.878963053226471,
      "learning_rate": 8.825385663154023e-06,
      "loss": 0.7619,
      "step": 8539
    },
    {
      "epoch": 0.24606696248487292,
      "grad_norm": 0.8952950835227966,
      "learning_rate": 8.825085178203857e-06,
      "loss": 0.7578,
      "step": 8540
    },
    {
      "epoch": 0.2460957759465222,
      "grad_norm": 0.9240154027938843,
      "learning_rate": 8.824784659941023e-06,
      "loss": 0.8067,
      "step": 8541
    },
    {
      "epoch": 0.2461245894081715,
      "grad_norm": 0.8609698414802551,
      "learning_rate": 8.824484108368136e-06,
      "loss": 0.7858,
      "step": 8542
    },
    {
      "epoch": 0.2461534028698208,
      "grad_norm": 0.8756932616233826,
      "learning_rate": 8.82418352348782e-06,
      "loss": 0.7327,
      "step": 8543
    },
    {
      "epoch": 0.24618221633147006,
      "grad_norm": 0.8656761646270752,
      "learning_rate": 8.823882905302684e-06,
      "loss": 0.781,
      "step": 8544
    },
    {
      "epoch": 0.24621102979311935,
      "grad_norm": 0.8829008340835571,
      "learning_rate": 8.823582253815351e-06,
      "loss": 0.7476,
      "step": 8545
    },
    {
      "epoch": 0.24623984325476864,
      "grad_norm": 0.9154717326164246,
      "learning_rate": 8.82328156902844e-06,
      "loss": 0.7634,
      "step": 8546
    },
    {
      "epoch": 0.2462686567164179,
      "grad_norm": 0.8728806972503662,
      "learning_rate": 8.82298085094457e-06,
      "loss": 0.7552,
      "step": 8547
    },
    {
      "epoch": 0.2462974701780672,
      "grad_norm": 0.9128577709197998,
      "learning_rate": 8.822680099566354e-06,
      "loss": 0.7816,
      "step": 8548
    },
    {
      "epoch": 0.24632628363971648,
      "grad_norm": 0.8694197535514832,
      "learning_rate": 8.822379314896419e-06,
      "loss": 0.7558,
      "step": 8549
    },
    {
      "epoch": 0.24635509710136574,
      "grad_norm": 0.8615185618400574,
      "learning_rate": 8.822078496937379e-06,
      "loss": 0.7435,
      "step": 8550
    },
    {
      "epoch": 0.24638391056301504,
      "grad_norm": 0.8855615258216858,
      "learning_rate": 8.821777645691858e-06,
      "loss": 0.7578,
      "step": 8551
    },
    {
      "epoch": 0.24641272402466433,
      "grad_norm": 0.8968737721443176,
      "learning_rate": 8.821476761162473e-06,
      "loss": 0.7645,
      "step": 8552
    },
    {
      "epoch": 0.24644153748631362,
      "grad_norm": 0.8799864053726196,
      "learning_rate": 8.821175843351847e-06,
      "loss": 0.7716,
      "step": 8553
    },
    {
      "epoch": 0.24647035094796288,
      "grad_norm": 0.9038013219833374,
      "learning_rate": 8.820874892262597e-06,
      "loss": 0.7923,
      "step": 8554
    },
    {
      "epoch": 0.24649916440961217,
      "grad_norm": 0.8833948969841003,
      "learning_rate": 8.820573907897346e-06,
      "loss": 0.7986,
      "step": 8555
    },
    {
      "epoch": 0.24652797787126146,
      "grad_norm": 0.8893906474113464,
      "learning_rate": 8.820272890258717e-06,
      "loss": 0.7503,
      "step": 8556
    },
    {
      "epoch": 0.24655679133291072,
      "grad_norm": 0.9541937112808228,
      "learning_rate": 8.819971839349328e-06,
      "loss": 0.7673,
      "step": 8557
    },
    {
      "epoch": 0.24658560479456001,
      "grad_norm": 0.867643415927887,
      "learning_rate": 8.819670755171805e-06,
      "loss": 0.7637,
      "step": 8558
    },
    {
      "epoch": 0.2466144182562093,
      "grad_norm": 0.9204844832420349,
      "learning_rate": 8.819369637728767e-06,
      "loss": 0.786,
      "step": 8559
    },
    {
      "epoch": 0.2466432317178586,
      "grad_norm": 0.8934987783432007,
      "learning_rate": 8.819068487022837e-06,
      "loss": 0.7777,
      "step": 8560
    },
    {
      "epoch": 0.24667204517950786,
      "grad_norm": 0.9688340425491333,
      "learning_rate": 8.818767303056641e-06,
      "loss": 0.767,
      "step": 8561
    },
    {
      "epoch": 0.24670085864115715,
      "grad_norm": 0.9076507091522217,
      "learning_rate": 8.818466085832797e-06,
      "loss": 0.7498,
      "step": 8562
    },
    {
      "epoch": 0.24672967210280644,
      "grad_norm": 0.9343093037605286,
      "learning_rate": 8.81816483535393e-06,
      "loss": 0.7577,
      "step": 8563
    },
    {
      "epoch": 0.2467584855644557,
      "grad_norm": 0.8885083794593811,
      "learning_rate": 8.817863551622664e-06,
      "loss": 0.7632,
      "step": 8564
    },
    {
      "epoch": 0.246787299026105,
      "grad_norm": 0.8943465352058411,
      "learning_rate": 8.817562234641624e-06,
      "loss": 0.7741,
      "step": 8565
    },
    {
      "epoch": 0.24681611248775429,
      "grad_norm": 0.8711923956871033,
      "learning_rate": 8.817260884413433e-06,
      "loss": 0.7717,
      "step": 8566
    },
    {
      "epoch": 0.24684492594940355,
      "grad_norm": 0.9435797333717346,
      "learning_rate": 8.816959500940714e-06,
      "loss": 0.7624,
      "step": 8567
    },
    {
      "epoch": 0.24687373941105284,
      "grad_norm": 0.9423201680183411,
      "learning_rate": 8.816658084226095e-06,
      "loss": 0.7615,
      "step": 8568
    },
    {
      "epoch": 0.24690255287270213,
      "grad_norm": 0.8710261583328247,
      "learning_rate": 8.816356634272198e-06,
      "loss": 0.7667,
      "step": 8569
    },
    {
      "epoch": 0.24693136633435142,
      "grad_norm": 0.8897258639335632,
      "learning_rate": 8.81605515108165e-06,
      "loss": 0.7766,
      "step": 8570
    },
    {
      "epoch": 0.24696017979600068,
      "grad_norm": 0.9093735814094543,
      "learning_rate": 8.815753634657079e-06,
      "loss": 0.7823,
      "step": 8571
    },
    {
      "epoch": 0.24698899325764997,
      "grad_norm": 0.8791581392288208,
      "learning_rate": 8.815452085001104e-06,
      "loss": 0.7554,
      "step": 8572
    },
    {
      "epoch": 0.24701780671929927,
      "grad_norm": 0.9037892818450928,
      "learning_rate": 8.815150502116357e-06,
      "loss": 0.7674,
      "step": 8573
    },
    {
      "epoch": 0.24704662018094853,
      "grad_norm": 0.9124715328216553,
      "learning_rate": 8.814848886005465e-06,
      "loss": 0.7474,
      "step": 8574
    },
    {
      "epoch": 0.24707543364259782,
      "grad_norm": 0.8795797824859619,
      "learning_rate": 8.81454723667105e-06,
      "loss": 0.7524,
      "step": 8575
    },
    {
      "epoch": 0.2471042471042471,
      "grad_norm": 0.8985702991485596,
      "learning_rate": 8.814245554115743e-06,
      "loss": 0.7569,
      "step": 8576
    },
    {
      "epoch": 0.24713306056589637,
      "grad_norm": 0.8552834391593933,
      "learning_rate": 8.813943838342169e-06,
      "loss": 0.7676,
      "step": 8577
    },
    {
      "epoch": 0.24716187402754566,
      "grad_norm": 0.8721147775650024,
      "learning_rate": 8.813642089352958e-06,
      "loss": 0.7584,
      "step": 8578
    },
    {
      "epoch": 0.24719068748919495,
      "grad_norm": 0.9195780754089355,
      "learning_rate": 8.813340307150735e-06,
      "loss": 0.7997,
      "step": 8579
    },
    {
      "epoch": 0.24721950095084425,
      "grad_norm": 0.8563046455383301,
      "learning_rate": 8.813038491738131e-06,
      "loss": 0.7823,
      "step": 8580
    },
    {
      "epoch": 0.2472483144124935,
      "grad_norm": 0.856037974357605,
      "learning_rate": 8.812736643117774e-06,
      "loss": 0.7549,
      "step": 8581
    },
    {
      "epoch": 0.2472771278741428,
      "grad_norm": 0.8668912649154663,
      "learning_rate": 8.812434761292291e-06,
      "loss": 0.7717,
      "step": 8582
    },
    {
      "epoch": 0.2473059413357921,
      "grad_norm": 0.8730886578559875,
      "learning_rate": 8.812132846264313e-06,
      "loss": 0.7776,
      "step": 8583
    },
    {
      "epoch": 0.24733475479744135,
      "grad_norm": 0.8858415484428406,
      "learning_rate": 8.811830898036468e-06,
      "loss": 0.7768,
      "step": 8584
    },
    {
      "epoch": 0.24736356825909064,
      "grad_norm": 0.8583654761314392,
      "learning_rate": 8.811528916611386e-06,
      "loss": 0.7632,
      "step": 8585
    },
    {
      "epoch": 0.24739238172073993,
      "grad_norm": 0.8519843220710754,
      "learning_rate": 8.811226901991698e-06,
      "loss": 0.7797,
      "step": 8586
    },
    {
      "epoch": 0.24742119518238923,
      "grad_norm": 0.8942779302597046,
      "learning_rate": 8.810924854180035e-06,
      "loss": 0.7456,
      "step": 8587
    },
    {
      "epoch": 0.2474500086440385,
      "grad_norm": 0.8619985580444336,
      "learning_rate": 8.810622773179023e-06,
      "loss": 0.7675,
      "step": 8588
    },
    {
      "epoch": 0.24747882210568778,
      "grad_norm": 0.913804292678833,
      "learning_rate": 8.810320658991298e-06,
      "loss": 0.7634,
      "step": 8589
    },
    {
      "epoch": 0.24750763556733707,
      "grad_norm": 0.8500623106956482,
      "learning_rate": 8.810018511619489e-06,
      "loss": 0.7681,
      "step": 8590
    },
    {
      "epoch": 0.24753644902898633,
      "grad_norm": 0.8753992319107056,
      "learning_rate": 8.809716331066228e-06,
      "loss": 0.7873,
      "step": 8591
    },
    {
      "epoch": 0.24756526249063562,
      "grad_norm": 0.8764601945877075,
      "learning_rate": 8.809414117334144e-06,
      "loss": 0.7685,
      "step": 8592
    },
    {
      "epoch": 0.24759407595228491,
      "grad_norm": 0.8776435256004333,
      "learning_rate": 8.80911187042587e-06,
      "loss": 0.7557,
      "step": 8593
    },
    {
      "epoch": 0.24762288941393418,
      "grad_norm": 0.8713236451148987,
      "learning_rate": 8.808809590344043e-06,
      "loss": 0.7496,
      "step": 8594
    },
    {
      "epoch": 0.24765170287558347,
      "grad_norm": 0.8658658266067505,
      "learning_rate": 8.808507277091289e-06,
      "loss": 0.751,
      "step": 8595
    },
    {
      "epoch": 0.24768051633723276,
      "grad_norm": 0.906910240650177,
      "learning_rate": 8.808204930670245e-06,
      "loss": 0.7759,
      "step": 8596
    },
    {
      "epoch": 0.24770932979888205,
      "grad_norm": 0.8928353786468506,
      "learning_rate": 8.807902551083543e-06,
      "loss": 0.7437,
      "step": 8597
    },
    {
      "epoch": 0.2477381432605313,
      "grad_norm": 0.9334588050842285,
      "learning_rate": 8.807600138333816e-06,
      "loss": 0.7548,
      "step": 8598
    },
    {
      "epoch": 0.2477669567221806,
      "grad_norm": 0.8689054846763611,
      "learning_rate": 8.807297692423698e-06,
      "loss": 0.7734,
      "step": 8599
    },
    {
      "epoch": 0.2477957701838299,
      "grad_norm": 0.9154523015022278,
      "learning_rate": 8.806995213355822e-06,
      "loss": 0.7508,
      "step": 8600
    },
    {
      "epoch": 0.24782458364547916,
      "grad_norm": 0.8711967468261719,
      "learning_rate": 8.806692701132824e-06,
      "loss": 0.7362,
      "step": 8601
    },
    {
      "epoch": 0.24785339710712845,
      "grad_norm": 0.9154866933822632,
      "learning_rate": 8.80639015575734e-06,
      "loss": 0.7808,
      "step": 8602
    },
    {
      "epoch": 0.24788221056877774,
      "grad_norm": 0.9002479910850525,
      "learning_rate": 8.806087577232001e-06,
      "loss": 0.759,
      "step": 8603
    },
    {
      "epoch": 0.247911024030427,
      "grad_norm": 0.8836525082588196,
      "learning_rate": 8.805784965559442e-06,
      "loss": 0.7734,
      "step": 8604
    },
    {
      "epoch": 0.2479398374920763,
      "grad_norm": 0.8836202621459961,
      "learning_rate": 8.805482320742303e-06,
      "loss": 0.7541,
      "step": 8605
    },
    {
      "epoch": 0.24796865095372558,
      "grad_norm": 0.9022014141082764,
      "learning_rate": 8.805179642783216e-06,
      "loss": 0.7592,
      "step": 8606
    },
    {
      "epoch": 0.24799746441537487,
      "grad_norm": 0.8710538744926453,
      "learning_rate": 8.804876931684819e-06,
      "loss": 0.7675,
      "step": 8607
    },
    {
      "epoch": 0.24802627787702414,
      "grad_norm": 0.897912323474884,
      "learning_rate": 8.804574187449747e-06,
      "loss": 0.7592,
      "step": 8608
    },
    {
      "epoch": 0.24805509133867343,
      "grad_norm": 0.89226895570755,
      "learning_rate": 8.804271410080636e-06,
      "loss": 0.784,
      "step": 8609
    },
    {
      "epoch": 0.24808390480032272,
      "grad_norm": 0.9053977727890015,
      "learning_rate": 8.803968599580125e-06,
      "loss": 0.7759,
      "step": 8610
    },
    {
      "epoch": 0.24811271826197198,
      "grad_norm": 0.9137889742851257,
      "learning_rate": 8.80366575595085e-06,
      "loss": 0.7679,
      "step": 8611
    },
    {
      "epoch": 0.24814153172362127,
      "grad_norm": 0.8637253642082214,
      "learning_rate": 8.803362879195448e-06,
      "loss": 0.7451,
      "step": 8612
    },
    {
      "epoch": 0.24817034518527056,
      "grad_norm": 0.8833580613136292,
      "learning_rate": 8.803059969316558e-06,
      "loss": 0.743,
      "step": 8613
    },
    {
      "epoch": 0.24819915864691985,
      "grad_norm": 0.9293470978736877,
      "learning_rate": 8.802757026316818e-06,
      "loss": 0.7721,
      "step": 8614
    },
    {
      "epoch": 0.24822797210856912,
      "grad_norm": 0.9165955781936646,
      "learning_rate": 8.802454050198864e-06,
      "loss": 0.758,
      "step": 8615
    },
    {
      "epoch": 0.2482567855702184,
      "grad_norm": 0.8896413445472717,
      "learning_rate": 8.802151040965337e-06,
      "loss": 0.7511,
      "step": 8616
    },
    {
      "epoch": 0.2482855990318677,
      "grad_norm": 0.9506435990333557,
      "learning_rate": 8.801847998618873e-06,
      "loss": 0.7562,
      "step": 8617
    },
    {
      "epoch": 0.24831441249351696,
      "grad_norm": 1.0128408670425415,
      "learning_rate": 8.801544923162116e-06,
      "loss": 0.75,
      "step": 8618
    },
    {
      "epoch": 0.24834322595516625,
      "grad_norm": 0.9328815937042236,
      "learning_rate": 8.801241814597703e-06,
      "loss": 0.7677,
      "step": 8619
    },
    {
      "epoch": 0.24837203941681554,
      "grad_norm": 0.9430286288261414,
      "learning_rate": 8.800938672928272e-06,
      "loss": 0.7872,
      "step": 8620
    },
    {
      "epoch": 0.2484008528784648,
      "grad_norm": 0.9758137464523315,
      "learning_rate": 8.800635498156467e-06,
      "loss": 0.7816,
      "step": 8621
    },
    {
      "epoch": 0.2484296663401141,
      "grad_norm": 0.9614037275314331,
      "learning_rate": 8.800332290284924e-06,
      "loss": 0.7673,
      "step": 8622
    },
    {
      "epoch": 0.2484584798017634,
      "grad_norm": 0.881768524646759,
      "learning_rate": 8.800029049316286e-06,
      "loss": 0.7944,
      "step": 8623
    },
    {
      "epoch": 0.24848729326341268,
      "grad_norm": 0.961810290813446,
      "learning_rate": 8.799725775253193e-06,
      "loss": 0.7606,
      "step": 8624
    },
    {
      "epoch": 0.24851610672506194,
      "grad_norm": 0.9990534782409668,
      "learning_rate": 8.799422468098289e-06,
      "loss": 0.7844,
      "step": 8625
    },
    {
      "epoch": 0.24854492018671123,
      "grad_norm": 0.8843140602111816,
      "learning_rate": 8.799119127854211e-06,
      "loss": 0.7564,
      "step": 8626
    },
    {
      "epoch": 0.24857373364836052,
      "grad_norm": 0.9345785975456238,
      "learning_rate": 8.798815754523606e-06,
      "loss": 0.7795,
      "step": 8627
    },
    {
      "epoch": 0.24860254711000979,
      "grad_norm": 0.9410912990570068,
      "learning_rate": 8.798512348109111e-06,
      "loss": 0.7526,
      "step": 8628
    },
    {
      "epoch": 0.24863136057165908,
      "grad_norm": 0.9397479891777039,
      "learning_rate": 8.798208908613371e-06,
      "loss": 0.782,
      "step": 8629
    },
    {
      "epoch": 0.24866017403330837,
      "grad_norm": 0.8707739114761353,
      "learning_rate": 8.797905436039029e-06,
      "loss": 0.7879,
      "step": 8630
    },
    {
      "epoch": 0.24868898749495763,
      "grad_norm": 0.8619935512542725,
      "learning_rate": 8.797601930388726e-06,
      "loss": 0.7843,
      "step": 8631
    },
    {
      "epoch": 0.24871780095660692,
      "grad_norm": 0.885040819644928,
      "learning_rate": 8.797298391665107e-06,
      "loss": 0.7479,
      "step": 8632
    },
    {
      "epoch": 0.2487466144182562,
      "grad_norm": 0.9621379375457764,
      "learning_rate": 8.796994819870814e-06,
      "loss": 0.7709,
      "step": 8633
    },
    {
      "epoch": 0.2487754278799055,
      "grad_norm": 0.8575378060340881,
      "learning_rate": 8.796691215008492e-06,
      "loss": 0.7942,
      "step": 8634
    },
    {
      "epoch": 0.24880424134155477,
      "grad_norm": 0.8679373264312744,
      "learning_rate": 8.796387577080788e-06,
      "loss": 0.7646,
      "step": 8635
    },
    {
      "epoch": 0.24883305480320406,
      "grad_norm": 0.9584304690361023,
      "learning_rate": 8.79608390609034e-06,
      "loss": 0.7707,
      "step": 8636
    },
    {
      "epoch": 0.24886186826485335,
      "grad_norm": 0.8933956623077393,
      "learning_rate": 8.795780202039796e-06,
      "loss": 0.7701,
      "step": 8637
    },
    {
      "epoch": 0.2488906817265026,
      "grad_norm": 0.904315173625946,
      "learning_rate": 8.7954764649318e-06,
      "loss": 0.7763,
      "step": 8638
    },
    {
      "epoch": 0.2489194951881519,
      "grad_norm": 0.858981192111969,
      "learning_rate": 8.795172694769e-06,
      "loss": 0.7692,
      "step": 8639
    },
    {
      "epoch": 0.2489483086498012,
      "grad_norm": 0.8423818945884705,
      "learning_rate": 8.79486889155404e-06,
      "loss": 0.7471,
      "step": 8640
    },
    {
      "epoch": 0.24897712211145048,
      "grad_norm": 0.8692365288734436,
      "learning_rate": 8.794565055289564e-06,
      "loss": 0.7749,
      "step": 8641
    },
    {
      "epoch": 0.24900593557309975,
      "grad_norm": 0.8650571703910828,
      "learning_rate": 8.79426118597822e-06,
      "loss": 0.7916,
      "step": 8642
    },
    {
      "epoch": 0.24903474903474904,
      "grad_norm": 0.8511591553688049,
      "learning_rate": 8.793957283622653e-06,
      "loss": 0.7572,
      "step": 8643
    },
    {
      "epoch": 0.24906356249639833,
      "grad_norm": 0.8835509419441223,
      "learning_rate": 8.793653348225512e-06,
      "loss": 0.771,
      "step": 8644
    },
    {
      "epoch": 0.2490923759580476,
      "grad_norm": 0.8856927752494812,
      "learning_rate": 8.793349379789441e-06,
      "loss": 0.775,
      "step": 8645
    },
    {
      "epoch": 0.24912118941969688,
      "grad_norm": 0.894391655921936,
      "learning_rate": 8.793045378317091e-06,
      "loss": 0.7656,
      "step": 8646
    },
    {
      "epoch": 0.24915000288134617,
      "grad_norm": 0.8685413002967834,
      "learning_rate": 8.792741343811107e-06,
      "loss": 0.7696,
      "step": 8647
    },
    {
      "epoch": 0.24917881634299544,
      "grad_norm": 0.9140568375587463,
      "learning_rate": 8.792437276274135e-06,
      "loss": 0.7682,
      "step": 8648
    },
    {
      "epoch": 0.24920762980464473,
      "grad_norm": 0.8580152988433838,
      "learning_rate": 8.792133175708828e-06,
      "loss": 0.7683,
      "step": 8649
    },
    {
      "epoch": 0.24923644326629402,
      "grad_norm": 0.8472973704338074,
      "learning_rate": 8.79182904211783e-06,
      "loss": 0.7668,
      "step": 8650
    },
    {
      "epoch": 0.2492652567279433,
      "grad_norm": 0.8427506685256958,
      "learning_rate": 8.791524875503792e-06,
      "loss": 0.7697,
      "step": 8651
    },
    {
      "epoch": 0.24929407018959257,
      "grad_norm": 0.8936774730682373,
      "learning_rate": 8.791220675869364e-06,
      "loss": 0.7595,
      "step": 8652
    },
    {
      "epoch": 0.24932288365124186,
      "grad_norm": 0.8567338585853577,
      "learning_rate": 8.790916443217192e-06,
      "loss": 0.7599,
      "step": 8653
    },
    {
      "epoch": 0.24935169711289115,
      "grad_norm": 0.875346302986145,
      "learning_rate": 8.790612177549928e-06,
      "loss": 0.7727,
      "step": 8654
    },
    {
      "epoch": 0.24938051057454041,
      "grad_norm": 0.8964062929153442,
      "learning_rate": 8.79030787887022e-06,
      "loss": 0.7788,
      "step": 8655
    },
    {
      "epoch": 0.2494093240361897,
      "grad_norm": 0.9101236462593079,
      "learning_rate": 8.790003547180721e-06,
      "loss": 0.7674,
      "step": 8656
    },
    {
      "epoch": 0.249438137497839,
      "grad_norm": 0.9088935256004333,
      "learning_rate": 8.789699182484079e-06,
      "loss": 0.7638,
      "step": 8657
    },
    {
      "epoch": 0.24946695095948826,
      "grad_norm": 0.9147329926490784,
      "learning_rate": 8.789394784782945e-06,
      "loss": 0.7527,
      "step": 8658
    },
    {
      "epoch": 0.24949576442113755,
      "grad_norm": 0.8784633874893188,
      "learning_rate": 8.789090354079972e-06,
      "loss": 0.7844,
      "step": 8659
    },
    {
      "epoch": 0.24952457788278684,
      "grad_norm": 0.8997082710266113,
      "learning_rate": 8.788785890377807e-06,
      "loss": 0.7534,
      "step": 8660
    },
    {
      "epoch": 0.24955339134443613,
      "grad_norm": 0.882392942905426,
      "learning_rate": 8.788481393679107e-06,
      "loss": 0.7463,
      "step": 8661
    },
    {
      "epoch": 0.2495822048060854,
      "grad_norm": 0.8698379993438721,
      "learning_rate": 8.78817686398652e-06,
      "loss": 0.7553,
      "step": 8662
    },
    {
      "epoch": 0.24961101826773469,
      "grad_norm": 0.9535166025161743,
      "learning_rate": 8.787872301302699e-06,
      "loss": 0.7962,
      "step": 8663
    },
    {
      "epoch": 0.24963983172938398,
      "grad_norm": 0.904970109462738,
      "learning_rate": 8.787567705630297e-06,
      "loss": 0.7794,
      "step": 8664
    },
    {
      "epoch": 0.24966864519103324,
      "grad_norm": 0.8723390102386475,
      "learning_rate": 8.787263076971966e-06,
      "loss": 0.7681,
      "step": 8665
    },
    {
      "epoch": 0.24969745865268253,
      "grad_norm": 0.86879563331604,
      "learning_rate": 8.78695841533036e-06,
      "loss": 0.7514,
      "step": 8666
    },
    {
      "epoch": 0.24972627211433182,
      "grad_norm": 0.8957692384719849,
      "learning_rate": 8.786653720708132e-06,
      "loss": 0.7599,
      "step": 8667
    },
    {
      "epoch": 0.2497550855759811,
      "grad_norm": 0.8656340837478638,
      "learning_rate": 8.786348993107935e-06,
      "loss": 0.7609,
      "step": 8668
    },
    {
      "epoch": 0.24978389903763037,
      "grad_norm": 0.9064173698425293,
      "learning_rate": 8.786044232532423e-06,
      "loss": 0.7854,
      "step": 8669
    },
    {
      "epoch": 0.24981271249927967,
      "grad_norm": 0.8424815535545349,
      "learning_rate": 8.78573943898425e-06,
      "loss": 0.7362,
      "step": 8670
    },
    {
      "epoch": 0.24984152596092896,
      "grad_norm": 0.8871890902519226,
      "learning_rate": 8.785434612466073e-06,
      "loss": 0.7674,
      "step": 8671
    },
    {
      "epoch": 0.24987033942257822,
      "grad_norm": 0.91908860206604,
      "learning_rate": 8.785129752980544e-06,
      "loss": 0.7574,
      "step": 8672
    },
    {
      "epoch": 0.2498991528842275,
      "grad_norm": 0.8501632809638977,
      "learning_rate": 8.784824860530317e-06,
      "loss": 0.7479,
      "step": 8673
    },
    {
      "epoch": 0.2499279663458768,
      "grad_norm": 0.8395729064941406,
      "learning_rate": 8.78451993511805e-06,
      "loss": 0.7316,
      "step": 8674
    },
    {
      "epoch": 0.24995677980752606,
      "grad_norm": 0.9222275018692017,
      "learning_rate": 8.784214976746397e-06,
      "loss": 0.7742,
      "step": 8675
    },
    {
      "epoch": 0.24998559326917535,
      "grad_norm": 0.8376320600509644,
      "learning_rate": 8.783909985418014e-06,
      "loss": 0.7633,
      "step": 8676
    },
    {
      "epoch": 0.25001440673082465,
      "grad_norm": 0.8970139026641846,
      "learning_rate": 8.783604961135559e-06,
      "loss": 0.8054,
      "step": 8677
    },
    {
      "epoch": 0.25004322019247394,
      "grad_norm": 0.8734245300292969,
      "learning_rate": 8.783299903901686e-06,
      "loss": 0.7705,
      "step": 8678
    },
    {
      "epoch": 0.2500720336541232,
      "grad_norm": 0.8767783045768738,
      "learning_rate": 8.782994813719052e-06,
      "loss": 0.7631,
      "step": 8679
    },
    {
      "epoch": 0.25010084711577246,
      "grad_norm": 0.8956614136695862,
      "learning_rate": 8.782689690590315e-06,
      "loss": 0.7704,
      "step": 8680
    },
    {
      "epoch": 0.25012966057742175,
      "grad_norm": 0.8741223812103271,
      "learning_rate": 8.782384534518134e-06,
      "loss": 0.7523,
      "step": 8681
    },
    {
      "epoch": 0.25015847403907104,
      "grad_norm": 0.9282186627388,
      "learning_rate": 8.782079345505164e-06,
      "loss": 0.7702,
      "step": 8682
    },
    {
      "epoch": 0.25018728750072033,
      "grad_norm": 0.8748956918716431,
      "learning_rate": 8.781774123554062e-06,
      "loss": 0.7642,
      "step": 8683
    },
    {
      "epoch": 0.2502161009623696,
      "grad_norm": 0.9198837876319885,
      "learning_rate": 8.781468868667488e-06,
      "loss": 0.7916,
      "step": 8684
    },
    {
      "epoch": 0.2502449144240189,
      "grad_norm": 0.8828309774398804,
      "learning_rate": 8.781163580848102e-06,
      "loss": 0.7638,
      "step": 8685
    },
    {
      "epoch": 0.2502737278856682,
      "grad_norm": 0.8995965123176575,
      "learning_rate": 8.78085826009856e-06,
      "loss": 0.7813,
      "step": 8686
    },
    {
      "epoch": 0.25030254134731744,
      "grad_norm": 0.9435730576515198,
      "learning_rate": 8.780552906421523e-06,
      "loss": 0.7334,
      "step": 8687
    },
    {
      "epoch": 0.25033135480896673,
      "grad_norm": 0.9086421132087708,
      "learning_rate": 8.780247519819647e-06,
      "loss": 0.7655,
      "step": 8688
    },
    {
      "epoch": 0.250360168270616,
      "grad_norm": 0.8807031512260437,
      "learning_rate": 8.779942100295596e-06,
      "loss": 0.7595,
      "step": 8689
    },
    {
      "epoch": 0.2503889817322653,
      "grad_norm": 0.9320023059844971,
      "learning_rate": 8.779636647852028e-06,
      "loss": 0.7609,
      "step": 8690
    },
    {
      "epoch": 0.2504177951939146,
      "grad_norm": 0.9132838249206543,
      "learning_rate": 8.779331162491603e-06,
      "loss": 0.7687,
      "step": 8691
    },
    {
      "epoch": 0.2504466086555639,
      "grad_norm": 0.8682758212089539,
      "learning_rate": 8.779025644216978e-06,
      "loss": 0.7636,
      "step": 8692
    },
    {
      "epoch": 0.2504754221172132,
      "grad_norm": 0.8928659558296204,
      "learning_rate": 8.778720093030822e-06,
      "loss": 0.7811,
      "step": 8693
    },
    {
      "epoch": 0.2505042355788624,
      "grad_norm": 0.8924381732940674,
      "learning_rate": 8.778414508935788e-06,
      "loss": 0.7678,
      "step": 8694
    },
    {
      "epoch": 0.2505330490405117,
      "grad_norm": 0.8482975959777832,
      "learning_rate": 8.778108891934543e-06,
      "loss": 0.7748,
      "step": 8695
    },
    {
      "epoch": 0.250561862502161,
      "grad_norm": 0.8760583400726318,
      "learning_rate": 8.777803242029745e-06,
      "loss": 0.7969,
      "step": 8696
    },
    {
      "epoch": 0.2505906759638103,
      "grad_norm": 0.9119689464569092,
      "learning_rate": 8.777497559224058e-06,
      "loss": 0.8074,
      "step": 8697
    },
    {
      "epoch": 0.2506194894254596,
      "grad_norm": 0.8847499489784241,
      "learning_rate": 8.777191843520142e-06,
      "loss": 0.7758,
      "step": 8698
    },
    {
      "epoch": 0.2506483028871089,
      "grad_norm": 0.8801088929176331,
      "learning_rate": 8.776886094920662e-06,
      "loss": 0.7561,
      "step": 8699
    },
    {
      "epoch": 0.25067711634875817,
      "grad_norm": 0.9152742624282837,
      "learning_rate": 8.77658031342828e-06,
      "loss": 0.7818,
      "step": 8700
    },
    {
      "epoch": 0.2507059298104074,
      "grad_norm": 0.8708339929580688,
      "learning_rate": 8.776274499045658e-06,
      "loss": 0.7629,
      "step": 8701
    },
    {
      "epoch": 0.2507347432720567,
      "grad_norm": 0.8613559603691101,
      "learning_rate": 8.77596865177546e-06,
      "loss": 0.7605,
      "step": 8702
    },
    {
      "epoch": 0.250763556733706,
      "grad_norm": 0.8858427405357361,
      "learning_rate": 8.77566277162035e-06,
      "loss": 0.7794,
      "step": 8703
    },
    {
      "epoch": 0.2507923701953553,
      "grad_norm": 0.9113913178443909,
      "learning_rate": 8.77535685858299e-06,
      "loss": 0.7607,
      "step": 8704
    },
    {
      "epoch": 0.25082118365700456,
      "grad_norm": 0.9000673890113831,
      "learning_rate": 8.775050912666049e-06,
      "loss": 0.7992,
      "step": 8705
    },
    {
      "epoch": 0.25084999711865386,
      "grad_norm": 0.8644458055496216,
      "learning_rate": 8.774744933872186e-06,
      "loss": 0.7799,
      "step": 8706
    },
    {
      "epoch": 0.2508788105803031,
      "grad_norm": 0.8769914507865906,
      "learning_rate": 8.774438922204068e-06,
      "loss": 0.7718,
      "step": 8707
    },
    {
      "epoch": 0.2509076240419524,
      "grad_norm": 0.9350939393043518,
      "learning_rate": 8.774132877664361e-06,
      "loss": 0.7776,
      "step": 8708
    },
    {
      "epoch": 0.2509364375036017,
      "grad_norm": 0.9052518606185913,
      "learning_rate": 8.773826800255727e-06,
      "loss": 0.7599,
      "step": 8709
    },
    {
      "epoch": 0.25096525096525096,
      "grad_norm": 0.880072832107544,
      "learning_rate": 8.773520689980836e-06,
      "loss": 0.771,
      "step": 8710
    },
    {
      "epoch": 0.25099406442690025,
      "grad_norm": 0.8911394476890564,
      "learning_rate": 8.773214546842352e-06,
      "loss": 0.7786,
      "step": 8711
    },
    {
      "epoch": 0.25102287788854954,
      "grad_norm": 0.8844578862190247,
      "learning_rate": 8.772908370842941e-06,
      "loss": 0.7602,
      "step": 8712
    },
    {
      "epoch": 0.25105169135019884,
      "grad_norm": 0.9075611233711243,
      "learning_rate": 8.772602161985268e-06,
      "loss": 0.7863,
      "step": 8713
    },
    {
      "epoch": 0.25108050481184807,
      "grad_norm": 0.855862557888031,
      "learning_rate": 8.772295920272004e-06,
      "loss": 0.8055,
      "step": 8714
    },
    {
      "epoch": 0.25110931827349736,
      "grad_norm": 0.9309369921684265,
      "learning_rate": 8.771989645705811e-06,
      "loss": 0.7629,
      "step": 8715
    },
    {
      "epoch": 0.25113813173514665,
      "grad_norm": 0.8999558687210083,
      "learning_rate": 8.771683338289361e-06,
      "loss": 0.7939,
      "step": 8716
    },
    {
      "epoch": 0.25116694519679594,
      "grad_norm": 0.9032249450683594,
      "learning_rate": 8.771376998025318e-06,
      "loss": 0.7763,
      "step": 8717
    },
    {
      "epoch": 0.25119575865844523,
      "grad_norm": 0.8664863705635071,
      "learning_rate": 8.77107062491635e-06,
      "loss": 0.765,
      "step": 8718
    },
    {
      "epoch": 0.2512245721200945,
      "grad_norm": 0.8897820711135864,
      "learning_rate": 8.770764218965131e-06,
      "loss": 0.7664,
      "step": 8719
    },
    {
      "epoch": 0.2512533855817438,
      "grad_norm": 0.9098570942878723,
      "learning_rate": 8.770457780174322e-06,
      "loss": 0.7582,
      "step": 8720
    },
    {
      "epoch": 0.25128219904339305,
      "grad_norm": 0.8981025815010071,
      "learning_rate": 8.770151308546595e-06,
      "loss": 0.77,
      "step": 8721
    },
    {
      "epoch": 0.25131101250504234,
      "grad_norm": 0.860034704208374,
      "learning_rate": 8.76984480408462e-06,
      "loss": 0.7545,
      "step": 8722
    },
    {
      "epoch": 0.25133982596669163,
      "grad_norm": 0.8921576738357544,
      "learning_rate": 8.769538266791064e-06,
      "loss": 0.7963,
      "step": 8723
    },
    {
      "epoch": 0.2513686394283409,
      "grad_norm": 0.8744685649871826,
      "learning_rate": 8.769231696668597e-06,
      "loss": 0.7944,
      "step": 8724
    },
    {
      "epoch": 0.2513974528899902,
      "grad_norm": 0.8796738982200623,
      "learning_rate": 8.768925093719891e-06,
      "loss": 0.7633,
      "step": 8725
    },
    {
      "epoch": 0.2514262663516395,
      "grad_norm": 0.8686650991439819,
      "learning_rate": 8.768618457947614e-06,
      "loss": 0.7788,
      "step": 8726
    },
    {
      "epoch": 0.2514550798132888,
      "grad_norm": 0.8540209531784058,
      "learning_rate": 8.768311789354439e-06,
      "loss": 0.7438,
      "step": 8727
    },
    {
      "epoch": 0.25148389327493803,
      "grad_norm": 0.8648778796195984,
      "learning_rate": 8.768005087943035e-06,
      "loss": 0.7664,
      "step": 8728
    },
    {
      "epoch": 0.2515127067365873,
      "grad_norm": 0.8651167154312134,
      "learning_rate": 8.767698353716072e-06,
      "loss": 0.7525,
      "step": 8729
    },
    {
      "epoch": 0.2515415201982366,
      "grad_norm": 0.8678566217422485,
      "learning_rate": 8.767391586676224e-06,
      "loss": 0.7726,
      "step": 8730
    },
    {
      "epoch": 0.2515703336598859,
      "grad_norm": 0.8492189049720764,
      "learning_rate": 8.767084786826161e-06,
      "loss": 0.7516,
      "step": 8731
    },
    {
      "epoch": 0.2515991471215352,
      "grad_norm": 0.8824723958969116,
      "learning_rate": 8.766777954168555e-06,
      "loss": 0.7407,
      "step": 8732
    },
    {
      "epoch": 0.2516279605831845,
      "grad_norm": 0.8885572552680969,
      "learning_rate": 8.766471088706078e-06,
      "loss": 0.7848,
      "step": 8733
    },
    {
      "epoch": 0.2516567740448337,
      "grad_norm": 0.8294559717178345,
      "learning_rate": 8.766164190441405e-06,
      "loss": 0.7504,
      "step": 8734
    },
    {
      "epoch": 0.251685587506483,
      "grad_norm": 0.8853511810302734,
      "learning_rate": 8.765857259377204e-06,
      "loss": 0.7764,
      "step": 8735
    },
    {
      "epoch": 0.2517144009681323,
      "grad_norm": 0.8797395825386047,
      "learning_rate": 8.765550295516152e-06,
      "loss": 0.755,
      "step": 8736
    },
    {
      "epoch": 0.2517432144297816,
      "grad_norm": 0.8915903568267822,
      "learning_rate": 8.76524329886092e-06,
      "loss": 0.7981,
      "step": 8737
    },
    {
      "epoch": 0.2517720278914309,
      "grad_norm": 0.8992477655410767,
      "learning_rate": 8.764936269414184e-06,
      "loss": 0.7633,
      "step": 8738
    },
    {
      "epoch": 0.2518008413530802,
      "grad_norm": 0.8898319005966187,
      "learning_rate": 8.764629207178616e-06,
      "loss": 0.7757,
      "step": 8739
    },
    {
      "epoch": 0.25182965481472946,
      "grad_norm": 0.866642415523529,
      "learning_rate": 8.764322112156892e-06,
      "loss": 0.7712,
      "step": 8740
    },
    {
      "epoch": 0.2518584682763787,
      "grad_norm": 0.90838623046875,
      "learning_rate": 8.764014984351683e-06,
      "loss": 0.7741,
      "step": 8741
    },
    {
      "epoch": 0.251887281738028,
      "grad_norm": 0.8381341099739075,
      "learning_rate": 8.763707823765668e-06,
      "loss": 0.7438,
      "step": 8742
    },
    {
      "epoch": 0.2519160951996773,
      "grad_norm": 0.8864477872848511,
      "learning_rate": 8.763400630401521e-06,
      "loss": 0.7833,
      "step": 8743
    },
    {
      "epoch": 0.25194490866132657,
      "grad_norm": 1.069328784942627,
      "learning_rate": 8.763093404261916e-06,
      "loss": 0.7684,
      "step": 8744
    },
    {
      "epoch": 0.25197372212297586,
      "grad_norm": 0.9174985289573669,
      "learning_rate": 8.762786145349526e-06,
      "loss": 0.775,
      "step": 8745
    },
    {
      "epoch": 0.25200253558462515,
      "grad_norm": 0.8386560082435608,
      "learning_rate": 8.762478853667034e-06,
      "loss": 0.7671,
      "step": 8746
    },
    {
      "epoch": 0.25203134904627444,
      "grad_norm": 0.8786177635192871,
      "learning_rate": 8.762171529217112e-06,
      "loss": 0.7581,
      "step": 8747
    },
    {
      "epoch": 0.2520601625079237,
      "grad_norm": 0.9072234630584717,
      "learning_rate": 8.761864172002435e-06,
      "loss": 0.7457,
      "step": 8748
    },
    {
      "epoch": 0.25208897596957297,
      "grad_norm": 0.8894769549369812,
      "learning_rate": 8.761556782025681e-06,
      "loss": 0.7602,
      "step": 8749
    },
    {
      "epoch": 0.25211778943122226,
      "grad_norm": 0.8571528196334839,
      "learning_rate": 8.76124935928953e-06,
      "loss": 0.7564,
      "step": 8750
    },
    {
      "epoch": 0.25214660289287155,
      "grad_norm": 0.9281788468360901,
      "learning_rate": 8.760941903796655e-06,
      "loss": 0.777,
      "step": 8751
    },
    {
      "epoch": 0.25217541635452084,
      "grad_norm": 0.9062421321868896,
      "learning_rate": 8.760634415549738e-06,
      "loss": 0.7651,
      "step": 8752
    },
    {
      "epoch": 0.25220422981617013,
      "grad_norm": 0.8684877753257751,
      "learning_rate": 8.760326894551452e-06,
      "loss": 0.7709,
      "step": 8753
    },
    {
      "epoch": 0.2522330432778194,
      "grad_norm": 0.8779075145721436,
      "learning_rate": 8.760019340804478e-06,
      "loss": 0.7645,
      "step": 8754
    },
    {
      "epoch": 0.25226185673946866,
      "grad_norm": 0.9124706983566284,
      "learning_rate": 8.759711754311495e-06,
      "loss": 0.772,
      "step": 8755
    },
    {
      "epoch": 0.25229067020111795,
      "grad_norm": 0.8548992276191711,
      "learning_rate": 8.75940413507518e-06,
      "loss": 0.752,
      "step": 8756
    },
    {
      "epoch": 0.25231948366276724,
      "grad_norm": 0.914189875125885,
      "learning_rate": 8.759096483098216e-06,
      "loss": 0.7513,
      "step": 8757
    },
    {
      "epoch": 0.25234829712441653,
      "grad_norm": 0.8938772082328796,
      "learning_rate": 8.758788798383275e-06,
      "loss": 0.7624,
      "step": 8758
    },
    {
      "epoch": 0.2523771105860658,
      "grad_norm": 0.8674827814102173,
      "learning_rate": 8.758481080933043e-06,
      "loss": 0.7625,
      "step": 8759
    },
    {
      "epoch": 0.2524059240477151,
      "grad_norm": 0.8386251330375671,
      "learning_rate": 8.758173330750197e-06,
      "loss": 0.7646,
      "step": 8760
    },
    {
      "epoch": 0.25243473750936435,
      "grad_norm": 0.8771675229072571,
      "learning_rate": 8.757865547837421e-06,
      "loss": 0.7675,
      "step": 8761
    },
    {
      "epoch": 0.25246355097101364,
      "grad_norm": 0.8800970911979675,
      "learning_rate": 8.75755773219739e-06,
      "loss": 0.7371,
      "step": 8762
    },
    {
      "epoch": 0.25249236443266293,
      "grad_norm": 0.8739857077598572,
      "learning_rate": 8.757249883832787e-06,
      "loss": 0.7686,
      "step": 8763
    },
    {
      "epoch": 0.2525211778943122,
      "grad_norm": 0.8837625980377197,
      "learning_rate": 8.756942002746295e-06,
      "loss": 0.7693,
      "step": 8764
    },
    {
      "epoch": 0.2525499913559615,
      "grad_norm": 0.8890054225921631,
      "learning_rate": 8.756634088940592e-06,
      "loss": 0.7599,
      "step": 8765
    },
    {
      "epoch": 0.2525788048176108,
      "grad_norm": 0.917208194732666,
      "learning_rate": 8.756326142418364e-06,
      "loss": 0.7573,
      "step": 8766
    },
    {
      "epoch": 0.2526076182792601,
      "grad_norm": 0.8938680291175842,
      "learning_rate": 8.756018163182288e-06,
      "loss": 0.7719,
      "step": 8767
    },
    {
      "epoch": 0.25263643174090933,
      "grad_norm": 0.894180953502655,
      "learning_rate": 8.755710151235048e-06,
      "loss": 0.7824,
      "step": 8768
    },
    {
      "epoch": 0.2526652452025586,
      "grad_norm": 0.9035047888755798,
      "learning_rate": 8.75540210657933e-06,
      "loss": 0.7922,
      "step": 8769
    },
    {
      "epoch": 0.2526940586642079,
      "grad_norm": 0.9306754469871521,
      "learning_rate": 8.755094029217809e-06,
      "loss": 0.7762,
      "step": 8770
    },
    {
      "epoch": 0.2527228721258572,
      "grad_norm": 0.885085940361023,
      "learning_rate": 8.754785919153176e-06,
      "loss": 0.7733,
      "step": 8771
    },
    {
      "epoch": 0.2527516855875065,
      "grad_norm": 0.8708937764167786,
      "learning_rate": 8.754477776388112e-06,
      "loss": 0.7613,
      "step": 8772
    },
    {
      "epoch": 0.2527804990491558,
      "grad_norm": 0.8567492961883545,
      "learning_rate": 8.754169600925297e-06,
      "loss": 0.7403,
      "step": 8773
    },
    {
      "epoch": 0.2528093125108051,
      "grad_norm": 0.871647834777832,
      "learning_rate": 8.75386139276742e-06,
      "loss": 0.7623,
      "step": 8774
    },
    {
      "epoch": 0.2528381259724543,
      "grad_norm": 0.9185482263565063,
      "learning_rate": 8.75355315191716e-06,
      "loss": 0.7755,
      "step": 8775
    },
    {
      "epoch": 0.2528669394341036,
      "grad_norm": 0.859205961227417,
      "learning_rate": 8.753244878377206e-06,
      "loss": 0.7584,
      "step": 8776
    },
    {
      "epoch": 0.2528957528957529,
      "grad_norm": 0.8705272674560547,
      "learning_rate": 8.752936572150241e-06,
      "loss": 0.7556,
      "step": 8777
    },
    {
      "epoch": 0.2529245663574022,
      "grad_norm": 0.8495835065841675,
      "learning_rate": 8.75262823323895e-06,
      "loss": 0.7765,
      "step": 8778
    },
    {
      "epoch": 0.25295337981905147,
      "grad_norm": 0.8974508047103882,
      "learning_rate": 8.752319861646017e-06,
      "loss": 0.7818,
      "step": 8779
    },
    {
      "epoch": 0.25298219328070076,
      "grad_norm": 0.8780465722084045,
      "learning_rate": 8.75201145737413e-06,
      "loss": 0.7814,
      "step": 8780
    },
    {
      "epoch": 0.25301100674235005,
      "grad_norm": 0.8457291126251221,
      "learning_rate": 8.751703020425975e-06,
      "loss": 0.7613,
      "step": 8781
    },
    {
      "epoch": 0.2530398202039993,
      "grad_norm": 0.8765530586242676,
      "learning_rate": 8.751394550804235e-06,
      "loss": 0.7682,
      "step": 8782
    },
    {
      "epoch": 0.2530686336656486,
      "grad_norm": 0.8650594353675842,
      "learning_rate": 8.7510860485116e-06,
      "loss": 0.7623,
      "step": 8783
    },
    {
      "epoch": 0.25309744712729787,
      "grad_norm": 0.8682571649551392,
      "learning_rate": 8.750777513550755e-06,
      "loss": 0.7578,
      "step": 8784
    },
    {
      "epoch": 0.25312626058894716,
      "grad_norm": 0.9044787287712097,
      "learning_rate": 8.750468945924385e-06,
      "loss": 0.7617,
      "step": 8785
    },
    {
      "epoch": 0.25315507405059645,
      "grad_norm": 0.8930280208587646,
      "learning_rate": 8.750160345635183e-06,
      "loss": 0.7812,
      "step": 8786
    },
    {
      "epoch": 0.25318388751224574,
      "grad_norm": 0.8477242588996887,
      "learning_rate": 8.74985171268583e-06,
      "loss": 0.7626,
      "step": 8787
    },
    {
      "epoch": 0.253212700973895,
      "grad_norm": 0.8619908094406128,
      "learning_rate": 8.749543047079019e-06,
      "loss": 0.7478,
      "step": 8788
    },
    {
      "epoch": 0.25324151443554427,
      "grad_norm": 0.8913817405700684,
      "learning_rate": 8.749234348817436e-06,
      "loss": 0.7488,
      "step": 8789
    },
    {
      "epoch": 0.25327032789719356,
      "grad_norm": 0.8550992608070374,
      "learning_rate": 8.74892561790377e-06,
      "loss": 0.7492,
      "step": 8790
    },
    {
      "epoch": 0.25329914135884285,
      "grad_norm": 0.886997640132904,
      "learning_rate": 8.748616854340709e-06,
      "loss": 0.7561,
      "step": 8791
    },
    {
      "epoch": 0.25332795482049214,
      "grad_norm": 0.8519837856292725,
      "learning_rate": 8.748308058130944e-06,
      "loss": 0.749,
      "step": 8792
    },
    {
      "epoch": 0.25335676828214143,
      "grad_norm": 0.9149165749549866,
      "learning_rate": 8.74799922927716e-06,
      "loss": 0.7637,
      "step": 8793
    },
    {
      "epoch": 0.2533855817437907,
      "grad_norm": 0.8324651122093201,
      "learning_rate": 8.74769036778205e-06,
      "loss": 0.7484,
      "step": 8794
    },
    {
      "epoch": 0.25341439520543996,
      "grad_norm": 0.8522900938987732,
      "learning_rate": 8.747381473648305e-06,
      "loss": 0.7674,
      "step": 8795
    },
    {
      "epoch": 0.25344320866708925,
      "grad_norm": 0.8848430514335632,
      "learning_rate": 8.747072546878611e-06,
      "loss": 0.755,
      "step": 8796
    },
    {
      "epoch": 0.25347202212873854,
      "grad_norm": 0.8687821626663208,
      "learning_rate": 8.746763587475662e-06,
      "loss": 0.7517,
      "step": 8797
    },
    {
      "epoch": 0.25350083559038783,
      "grad_norm": 0.853875994682312,
      "learning_rate": 8.746454595442147e-06,
      "loss": 0.7853,
      "step": 8798
    },
    {
      "epoch": 0.2535296490520371,
      "grad_norm": 0.8634081482887268,
      "learning_rate": 8.746145570780758e-06,
      "loss": 0.731,
      "step": 8799
    },
    {
      "epoch": 0.2535584625136864,
      "grad_norm": 0.8736326098442078,
      "learning_rate": 8.745836513494185e-06,
      "loss": 0.7627,
      "step": 8800
    },
    {
      "epoch": 0.2535872759753357,
      "grad_norm": 0.8714659214019775,
      "learning_rate": 8.745527423585122e-06,
      "loss": 0.7713,
      "step": 8801
    },
    {
      "epoch": 0.25361608943698494,
      "grad_norm": 0.9033617377281189,
      "learning_rate": 8.74521830105626e-06,
      "loss": 0.7746,
      "step": 8802
    },
    {
      "epoch": 0.25364490289863423,
      "grad_norm": 0.8283765316009521,
      "learning_rate": 8.744909145910288e-06,
      "loss": 0.7604,
      "step": 8803
    },
    {
      "epoch": 0.2536737163602835,
      "grad_norm": 0.8840207457542419,
      "learning_rate": 8.744599958149902e-06,
      "loss": 0.7682,
      "step": 8804
    },
    {
      "epoch": 0.2537025298219328,
      "grad_norm": 0.8546888828277588,
      "learning_rate": 8.744290737777794e-06,
      "loss": 0.7636,
      "step": 8805
    },
    {
      "epoch": 0.2537313432835821,
      "grad_norm": 0.8483685851097107,
      "learning_rate": 8.743981484796655e-06,
      "loss": 0.7464,
      "step": 8806
    },
    {
      "epoch": 0.2537601567452314,
      "grad_norm": 0.8751692175865173,
      "learning_rate": 8.743672199209182e-06,
      "loss": 0.7889,
      "step": 8807
    },
    {
      "epoch": 0.2537889702068807,
      "grad_norm": 0.8527483344078064,
      "learning_rate": 8.743362881018065e-06,
      "loss": 0.7561,
      "step": 8808
    },
    {
      "epoch": 0.2538177836685299,
      "grad_norm": 0.8775988221168518,
      "learning_rate": 8.743053530226e-06,
      "loss": 0.7817,
      "step": 8809
    },
    {
      "epoch": 0.2538465971301792,
      "grad_norm": 0.8646843433380127,
      "learning_rate": 8.74274414683568e-06,
      "loss": 0.763,
      "step": 8810
    },
    {
      "epoch": 0.2538754105918285,
      "grad_norm": 0.8849129676818848,
      "learning_rate": 8.742434730849801e-06,
      "loss": 0.751,
      "step": 8811
    },
    {
      "epoch": 0.2539042240534778,
      "grad_norm": 0.888267993927002,
      "learning_rate": 8.742125282271055e-06,
      "loss": 0.7808,
      "step": 8812
    },
    {
      "epoch": 0.2539330375151271,
      "grad_norm": 0.8800472617149353,
      "learning_rate": 8.741815801102138e-06,
      "loss": 0.7749,
      "step": 8813
    },
    {
      "epoch": 0.25396185097677637,
      "grad_norm": 0.8884788155555725,
      "learning_rate": 8.741506287345747e-06,
      "loss": 0.7696,
      "step": 8814
    },
    {
      "epoch": 0.2539906644384256,
      "grad_norm": 0.8965123295783997,
      "learning_rate": 8.741196741004576e-06,
      "loss": 0.7846,
      "step": 8815
    },
    {
      "epoch": 0.2540194779000749,
      "grad_norm": 0.8711471557617188,
      "learning_rate": 8.740887162081322e-06,
      "loss": 0.7851,
      "step": 8816
    },
    {
      "epoch": 0.2540482913617242,
      "grad_norm": 0.8428049087524414,
      "learning_rate": 8.74057755057868e-06,
      "loss": 0.7929,
      "step": 8817
    },
    {
      "epoch": 0.2540771048233735,
      "grad_norm": 0.9181279540061951,
      "learning_rate": 8.740267906499347e-06,
      "loss": 0.7662,
      "step": 8818
    },
    {
      "epoch": 0.25410591828502277,
      "grad_norm": 5.220722675323486,
      "learning_rate": 8.73995822984602e-06,
      "loss": 0.7415,
      "step": 8819
    },
    {
      "epoch": 0.25413473174667206,
      "grad_norm": 0.9017191529273987,
      "learning_rate": 8.739648520621393e-06,
      "loss": 0.7762,
      "step": 8820
    },
    {
      "epoch": 0.25416354520832135,
      "grad_norm": 0.9133745431900024,
      "learning_rate": 8.739338778828169e-06,
      "loss": 0.7483,
      "step": 8821
    },
    {
      "epoch": 0.2541923586699706,
      "grad_norm": 0.8810449838638306,
      "learning_rate": 8.739029004469039e-06,
      "loss": 0.7581,
      "step": 8822
    },
    {
      "epoch": 0.2542211721316199,
      "grad_norm": 0.9549296498298645,
      "learning_rate": 8.738719197546706e-06,
      "loss": 0.7574,
      "step": 8823
    },
    {
      "epoch": 0.25424998559326917,
      "grad_norm": 0.9329493045806885,
      "learning_rate": 8.738409358063865e-06,
      "loss": 0.765,
      "step": 8824
    },
    {
      "epoch": 0.25427879905491846,
      "grad_norm": 0.901195228099823,
      "learning_rate": 8.738099486023217e-06,
      "loss": 0.7759,
      "step": 8825
    },
    {
      "epoch": 0.25430761251656775,
      "grad_norm": 6.889765739440918,
      "learning_rate": 8.737789581427457e-06,
      "loss": 0.8135,
      "step": 8826
    },
    {
      "epoch": 0.25433642597821704,
      "grad_norm": 0.9806538820266724,
      "learning_rate": 8.737479644279287e-06,
      "loss": 0.7691,
      "step": 8827
    },
    {
      "epoch": 0.25436523943986633,
      "grad_norm": 0.9925232529640198,
      "learning_rate": 8.737169674581407e-06,
      "loss": 0.772,
      "step": 8828
    },
    {
      "epoch": 0.25439405290151557,
      "grad_norm": 1.0196884870529175,
      "learning_rate": 8.736859672336512e-06,
      "loss": 0.7984,
      "step": 8829
    },
    {
      "epoch": 0.25442286636316486,
      "grad_norm": 1.036624550819397,
      "learning_rate": 8.736549637547307e-06,
      "loss": 0.7971,
      "step": 8830
    },
    {
      "epoch": 0.25445167982481415,
      "grad_norm": 2.445265769958496,
      "learning_rate": 8.73623957021649e-06,
      "loss": 0.7564,
      "step": 8831
    },
    {
      "epoch": 0.25448049328646344,
      "grad_norm": 1.1775667667388916,
      "learning_rate": 8.73592947034676e-06,
      "loss": 0.7777,
      "step": 8832
    },
    {
      "epoch": 0.25450930674811273,
      "grad_norm": 2.514387607574463,
      "learning_rate": 8.73561933794082e-06,
      "loss": 0.7756,
      "step": 8833
    },
    {
      "epoch": 0.254538120209762,
      "grad_norm": 0.9691184759140015,
      "learning_rate": 8.735309173001369e-06,
      "loss": 0.7542,
      "step": 8834
    },
    {
      "epoch": 0.2545669336714113,
      "grad_norm": 0.9210229516029358,
      "learning_rate": 8.73499897553111e-06,
      "loss": 0.777,
      "step": 8835
    },
    {
      "epoch": 0.25459574713306055,
      "grad_norm": 0.976550817489624,
      "learning_rate": 8.734688745532742e-06,
      "loss": 0.7653,
      "step": 8836
    },
    {
      "epoch": 0.25462456059470984,
      "grad_norm": 0.9659836292266846,
      "learning_rate": 8.73437848300897e-06,
      "loss": 0.7788,
      "step": 8837
    },
    {
      "epoch": 0.2546533740563591,
      "grad_norm": 0.9537115097045898,
      "learning_rate": 8.734068187962492e-06,
      "loss": 0.7915,
      "step": 8838
    },
    {
      "epoch": 0.2546821875180084,
      "grad_norm": 0.9086401462554932,
      "learning_rate": 8.733757860396015e-06,
      "loss": 0.737,
      "step": 8839
    },
    {
      "epoch": 0.2547110009796577,
      "grad_norm": 0.9761729836463928,
      "learning_rate": 8.73344750031224e-06,
      "loss": 0.7667,
      "step": 8840
    },
    {
      "epoch": 0.254739814441307,
      "grad_norm": 0.9353480935096741,
      "learning_rate": 8.733137107713867e-06,
      "loss": 0.767,
      "step": 8841
    },
    {
      "epoch": 0.25476862790295624,
      "grad_norm": 0.9147904515266418,
      "learning_rate": 8.732826682603603e-06,
      "loss": 0.7769,
      "step": 8842
    },
    {
      "epoch": 0.2547974413646055,
      "grad_norm": 0.9872839450836182,
      "learning_rate": 8.732516224984152e-06,
      "loss": 0.8179,
      "step": 8843
    },
    {
      "epoch": 0.2548262548262548,
      "grad_norm": 0.9075226187705994,
      "learning_rate": 8.732205734858212e-06,
      "loss": 0.7672,
      "step": 8844
    },
    {
      "epoch": 0.2548550682879041,
      "grad_norm": 0.9519957900047302,
      "learning_rate": 8.731895212228492e-06,
      "loss": 0.7927,
      "step": 8845
    },
    {
      "epoch": 0.2548838817495534,
      "grad_norm": 0.97532057762146,
      "learning_rate": 8.731584657097695e-06,
      "loss": 0.8156,
      "step": 8846
    },
    {
      "epoch": 0.2549126952112027,
      "grad_norm": 0.9093114137649536,
      "learning_rate": 8.731274069468527e-06,
      "loss": 0.7592,
      "step": 8847
    },
    {
      "epoch": 0.254941508672852,
      "grad_norm": 0.9149566292762756,
      "learning_rate": 8.730963449343693e-06,
      "loss": 0.7733,
      "step": 8848
    },
    {
      "epoch": 0.2549703221345012,
      "grad_norm": 0.8861602544784546,
      "learning_rate": 8.730652796725895e-06,
      "loss": 0.7619,
      "step": 8849
    },
    {
      "epoch": 0.2549991355961505,
      "grad_norm": 0.9008292555809021,
      "learning_rate": 8.73034211161784e-06,
      "loss": 0.7607,
      "step": 8850
    },
    {
      "epoch": 0.2550279490577998,
      "grad_norm": 0.9071536064147949,
      "learning_rate": 8.730031394022235e-06,
      "loss": 0.7978,
      "step": 8851
    },
    {
      "epoch": 0.2550567625194491,
      "grad_norm": 0.9076855182647705,
      "learning_rate": 8.729720643941786e-06,
      "loss": 0.7744,
      "step": 8852
    },
    {
      "epoch": 0.2550855759810984,
      "grad_norm": 0.8904415369033813,
      "learning_rate": 8.729409861379196e-06,
      "loss": 0.7537,
      "step": 8853
    },
    {
      "epoch": 0.25511438944274767,
      "grad_norm": 0.8587258458137512,
      "learning_rate": 8.729099046337178e-06,
      "loss": 0.7835,
      "step": 8854
    },
    {
      "epoch": 0.25514320290439696,
      "grad_norm": 0.8743646144866943,
      "learning_rate": 8.728788198818432e-06,
      "loss": 0.7356,
      "step": 8855
    },
    {
      "epoch": 0.2551720163660462,
      "grad_norm": 0.8903196454048157,
      "learning_rate": 8.72847731882567e-06,
      "loss": 0.7586,
      "step": 8856
    },
    {
      "epoch": 0.2552008298276955,
      "grad_norm": 0.863987147808075,
      "learning_rate": 8.728166406361596e-06,
      "loss": 0.7503,
      "step": 8857
    },
    {
      "epoch": 0.2552296432893448,
      "grad_norm": 0.9189414381980896,
      "learning_rate": 8.72785546142892e-06,
      "loss": 0.7588,
      "step": 8858
    },
    {
      "epoch": 0.25525845675099407,
      "grad_norm": 0.8980805277824402,
      "learning_rate": 8.72754448403035e-06,
      "loss": 0.7515,
      "step": 8859
    },
    {
      "epoch": 0.25528727021264336,
      "grad_norm": 0.8612525463104248,
      "learning_rate": 8.727233474168592e-06,
      "loss": 0.7251,
      "step": 8860
    },
    {
      "epoch": 0.25531608367429265,
      "grad_norm": 0.8693912625312805,
      "learning_rate": 8.726922431846358e-06,
      "loss": 0.7629,
      "step": 8861
    },
    {
      "epoch": 0.25534489713594194,
      "grad_norm": 0.8487643003463745,
      "learning_rate": 8.726611357066357e-06,
      "loss": 0.7439,
      "step": 8862
    },
    {
      "epoch": 0.2553737105975912,
      "grad_norm": 0.8743125796318054,
      "learning_rate": 8.726300249831292e-06,
      "loss": 0.7496,
      "step": 8863
    },
    {
      "epoch": 0.25540252405924047,
      "grad_norm": 0.9215828776359558,
      "learning_rate": 8.72598911014388e-06,
      "loss": 0.7945,
      "step": 8864
    },
    {
      "epoch": 0.25543133752088976,
      "grad_norm": 0.9047238826751709,
      "learning_rate": 8.725677938006826e-06,
      "loss": 0.7752,
      "step": 8865
    },
    {
      "epoch": 0.25546015098253905,
      "grad_norm": 0.891234815120697,
      "learning_rate": 8.725366733422842e-06,
      "loss": 0.7746,
      "step": 8866
    },
    {
      "epoch": 0.25548896444418834,
      "grad_norm": 0.8954598903656006,
      "learning_rate": 8.725055496394637e-06,
      "loss": 0.7815,
      "step": 8867
    },
    {
      "epoch": 0.25551777790583763,
      "grad_norm": 0.8650959730148315,
      "learning_rate": 8.724744226924924e-06,
      "loss": 0.756,
      "step": 8868
    },
    {
      "epoch": 0.25554659136748686,
      "grad_norm": 0.8872672319412231,
      "learning_rate": 8.724432925016411e-06,
      "loss": 0.7445,
      "step": 8869
    },
    {
      "epoch": 0.25557540482913615,
      "grad_norm": 0.9101532697677612,
      "learning_rate": 8.72412159067181e-06,
      "loss": 0.7623,
      "step": 8870
    },
    {
      "epoch": 0.25560421829078545,
      "grad_norm": 0.8695492148399353,
      "learning_rate": 8.723810223893834e-06,
      "loss": 0.7784,
      "step": 8871
    },
    {
      "epoch": 0.25563303175243474,
      "grad_norm": 0.8643758296966553,
      "learning_rate": 8.723498824685192e-06,
      "loss": 0.7501,
      "step": 8872
    },
    {
      "epoch": 0.255661845214084,
      "grad_norm": 0.8780376315116882,
      "learning_rate": 8.723187393048598e-06,
      "loss": 0.7701,
      "step": 8873
    },
    {
      "epoch": 0.2556906586757333,
      "grad_norm": 0.8899958729743958,
      "learning_rate": 8.722875928986762e-06,
      "loss": 0.748,
      "step": 8874
    },
    {
      "epoch": 0.2557194721373826,
      "grad_norm": 0.8969407677650452,
      "learning_rate": 8.7225644325024e-06,
      "loss": 0.7799,
      "step": 8875
    },
    {
      "epoch": 0.25574828559903184,
      "grad_norm": 0.9015818238258362,
      "learning_rate": 8.722252903598222e-06,
      "loss": 0.7735,
      "step": 8876
    },
    {
      "epoch": 0.25577709906068113,
      "grad_norm": 0.910262942314148,
      "learning_rate": 8.721941342276943e-06,
      "loss": 0.774,
      "step": 8877
    },
    {
      "epoch": 0.2558059125223304,
      "grad_norm": 0.8573765158653259,
      "learning_rate": 8.721629748541276e-06,
      "loss": 0.7442,
      "step": 8878
    },
    {
      "epoch": 0.2558347259839797,
      "grad_norm": 0.8737073540687561,
      "learning_rate": 8.721318122393933e-06,
      "loss": 0.7633,
      "step": 8879
    },
    {
      "epoch": 0.255863539445629,
      "grad_norm": 0.87447190284729,
      "learning_rate": 8.72100646383763e-06,
      "loss": 0.7448,
      "step": 8880
    },
    {
      "epoch": 0.2558923529072783,
      "grad_norm": 0.8723428845405579,
      "learning_rate": 8.72069477287508e-06,
      "loss": 0.7262,
      "step": 8881
    },
    {
      "epoch": 0.2559211663689276,
      "grad_norm": 0.8369721174240112,
      "learning_rate": 8.720383049508998e-06,
      "loss": 0.7716,
      "step": 8882
    },
    {
      "epoch": 0.2559499798305768,
      "grad_norm": 0.8543573617935181,
      "learning_rate": 8.720071293742099e-06,
      "loss": 0.7653,
      "step": 8883
    },
    {
      "epoch": 0.2559787932922261,
      "grad_norm": 0.9123809933662415,
      "learning_rate": 8.719759505577097e-06,
      "loss": 0.7523,
      "step": 8884
    },
    {
      "epoch": 0.2560076067538754,
      "grad_norm": 0.8570719361305237,
      "learning_rate": 8.719447685016709e-06,
      "loss": 0.7702,
      "step": 8885
    },
    {
      "epoch": 0.2560364202155247,
      "grad_norm": 0.8882539868354797,
      "learning_rate": 8.719135832063647e-06,
      "loss": 0.7522,
      "step": 8886
    },
    {
      "epoch": 0.256065233677174,
      "grad_norm": 0.8510768413543701,
      "learning_rate": 8.718823946720632e-06,
      "loss": 0.7328,
      "step": 8887
    },
    {
      "epoch": 0.2560940471388233,
      "grad_norm": 0.8330646753311157,
      "learning_rate": 8.718512028990378e-06,
      "loss": 0.7648,
      "step": 8888
    },
    {
      "epoch": 0.25612286060047257,
      "grad_norm": 0.895424485206604,
      "learning_rate": 8.7182000788756e-06,
      "loss": 0.7832,
      "step": 8889
    },
    {
      "epoch": 0.2561516740621218,
      "grad_norm": 0.8554991483688354,
      "learning_rate": 8.717888096379018e-06,
      "loss": 0.7488,
      "step": 8890
    },
    {
      "epoch": 0.2561804875237711,
      "grad_norm": 0.8780282735824585,
      "learning_rate": 8.717576081503347e-06,
      "loss": 0.7708,
      "step": 8891
    },
    {
      "epoch": 0.2562093009854204,
      "grad_norm": 0.8653927445411682,
      "learning_rate": 8.717264034251302e-06,
      "loss": 0.7478,
      "step": 8892
    },
    {
      "epoch": 0.2562381144470697,
      "grad_norm": 0.8666006922721863,
      "learning_rate": 8.716951954625606e-06,
      "loss": 0.7469,
      "step": 8893
    },
    {
      "epoch": 0.25626692790871897,
      "grad_norm": 0.8645737767219543,
      "learning_rate": 8.716639842628973e-06,
      "loss": 0.7564,
      "step": 8894
    },
    {
      "epoch": 0.25629574137036826,
      "grad_norm": 0.852192223072052,
      "learning_rate": 8.716327698264123e-06,
      "loss": 0.7649,
      "step": 8895
    },
    {
      "epoch": 0.2563245548320175,
      "grad_norm": 0.8945765495300293,
      "learning_rate": 8.716015521533773e-06,
      "loss": 0.7773,
      "step": 8896
    },
    {
      "epoch": 0.2563533682936668,
      "grad_norm": 0.8732379674911499,
      "learning_rate": 8.715703312440642e-06,
      "loss": 0.7521,
      "step": 8897
    },
    {
      "epoch": 0.2563821817553161,
      "grad_norm": 0.8853276968002319,
      "learning_rate": 8.71539107098745e-06,
      "loss": 0.7724,
      "step": 8898
    },
    {
      "epoch": 0.25641099521696537,
      "grad_norm": 0.8356859683990479,
      "learning_rate": 8.715078797176917e-06,
      "loss": 0.7578,
      "step": 8899
    },
    {
      "epoch": 0.25643980867861466,
      "grad_norm": 0.8790056705474854,
      "learning_rate": 8.714766491011759e-06,
      "loss": 0.7925,
      "step": 8900
    },
    {
      "epoch": 0.25646862214026395,
      "grad_norm": 0.8159679174423218,
      "learning_rate": 8.714454152494701e-06,
      "loss": 0.7426,
      "step": 8901
    },
    {
      "epoch": 0.25649743560191324,
      "grad_norm": 0.8504329919815063,
      "learning_rate": 8.714141781628459e-06,
      "loss": 0.7604,
      "step": 8902
    },
    {
      "epoch": 0.2565262490635625,
      "grad_norm": 0.8629888296127319,
      "learning_rate": 8.713829378415754e-06,
      "loss": 0.762,
      "step": 8903
    },
    {
      "epoch": 0.25655506252521176,
      "grad_norm": 0.8518761992454529,
      "learning_rate": 8.713516942859308e-06,
      "loss": 0.7495,
      "step": 8904
    },
    {
      "epoch": 0.25658387598686105,
      "grad_norm": 0.8542816638946533,
      "learning_rate": 8.713204474961842e-06,
      "loss": 0.772,
      "step": 8905
    },
    {
      "epoch": 0.25661268944851034,
      "grad_norm": 0.8718855977058411,
      "learning_rate": 8.712891974726076e-06,
      "loss": 0.7899,
      "step": 8906
    },
    {
      "epoch": 0.25664150291015964,
      "grad_norm": 0.8881014585494995,
      "learning_rate": 8.712579442154733e-06,
      "loss": 0.7575,
      "step": 8907
    },
    {
      "epoch": 0.2566703163718089,
      "grad_norm": 0.8815657496452332,
      "learning_rate": 8.712266877250534e-06,
      "loss": 0.7531,
      "step": 8908
    },
    {
      "epoch": 0.2566991298334582,
      "grad_norm": 0.8634742498397827,
      "learning_rate": 8.711954280016204e-06,
      "loss": 0.7699,
      "step": 8909
    },
    {
      "epoch": 0.25672794329510745,
      "grad_norm": 0.8642695546150208,
      "learning_rate": 8.71164165045446e-06,
      "loss": 0.7665,
      "step": 8910
    },
    {
      "epoch": 0.25675675675675674,
      "grad_norm": 0.8858514428138733,
      "learning_rate": 8.711328988568028e-06,
      "loss": 0.7629,
      "step": 8911
    },
    {
      "epoch": 0.25678557021840603,
      "grad_norm": 0.8650756478309631,
      "learning_rate": 8.711016294359632e-06,
      "loss": 0.7414,
      "step": 8912
    },
    {
      "epoch": 0.2568143836800553,
      "grad_norm": 0.8625120520591736,
      "learning_rate": 8.710703567831992e-06,
      "loss": 0.7723,
      "step": 8913
    },
    {
      "epoch": 0.2568431971417046,
      "grad_norm": 1.1469552516937256,
      "learning_rate": 8.710390808987834e-06,
      "loss": 0.7604,
      "step": 8914
    },
    {
      "epoch": 0.2568720106033539,
      "grad_norm": 0.8488196134567261,
      "learning_rate": 8.71007801782988e-06,
      "loss": 0.771,
      "step": 8915
    },
    {
      "epoch": 0.2569008240650032,
      "grad_norm": 0.8927820324897766,
      "learning_rate": 8.709765194360854e-06,
      "loss": 0.7798,
      "step": 8916
    },
    {
      "epoch": 0.25692963752665243,
      "grad_norm": 0.8516178131103516,
      "learning_rate": 8.709452338583486e-06,
      "loss": 0.7688,
      "step": 8917
    },
    {
      "epoch": 0.2569584509883017,
      "grad_norm": 0.8309195041656494,
      "learning_rate": 8.709139450500495e-06,
      "loss": 0.7773,
      "step": 8918
    },
    {
      "epoch": 0.256987264449951,
      "grad_norm": 0.8715848922729492,
      "learning_rate": 8.708826530114604e-06,
      "loss": 0.7557,
      "step": 8919
    },
    {
      "epoch": 0.2570160779116003,
      "grad_norm": 0.8681570887565613,
      "learning_rate": 8.708513577428544e-06,
      "loss": 0.7657,
      "step": 8920
    },
    {
      "epoch": 0.2570448913732496,
      "grad_norm": 0.8598151206970215,
      "learning_rate": 8.708200592445037e-06,
      "loss": 0.7634,
      "step": 8921
    },
    {
      "epoch": 0.2570737048348989,
      "grad_norm": 0.8744776248931885,
      "learning_rate": 8.707887575166811e-06,
      "loss": 0.7701,
      "step": 8922
    },
    {
      "epoch": 0.2571025182965481,
      "grad_norm": 0.9109063148498535,
      "learning_rate": 8.70757452559659e-06,
      "loss": 0.771,
      "step": 8923
    },
    {
      "epoch": 0.2571313317581974,
      "grad_norm": 0.8296816349029541,
      "learning_rate": 8.707261443737099e-06,
      "loss": 0.7714,
      "step": 8924
    },
    {
      "epoch": 0.2571601452198467,
      "grad_norm": 0.8893393278121948,
      "learning_rate": 8.70694832959107e-06,
      "loss": 0.7598,
      "step": 8925
    },
    {
      "epoch": 0.257188958681496,
      "grad_norm": 0.878208339214325,
      "learning_rate": 8.706635183161225e-06,
      "loss": 0.7458,
      "step": 8926
    },
    {
      "epoch": 0.2572177721431453,
      "grad_norm": 0.8492916226387024,
      "learning_rate": 8.706322004450291e-06,
      "loss": 0.7452,
      "step": 8927
    },
    {
      "epoch": 0.2572465856047946,
      "grad_norm": 0.8440157771110535,
      "learning_rate": 8.706008793461001e-06,
      "loss": 0.7552,
      "step": 8928
    },
    {
      "epoch": 0.25727539906644387,
      "grad_norm": 0.8741102814674377,
      "learning_rate": 8.705695550196077e-06,
      "loss": 0.7751,
      "step": 8929
    },
    {
      "epoch": 0.2573042125280931,
      "grad_norm": 0.8716955184936523,
      "learning_rate": 8.70538227465825e-06,
      "loss": 0.7528,
      "step": 8930
    },
    {
      "epoch": 0.2573330259897424,
      "grad_norm": 0.8335434794425964,
      "learning_rate": 8.705068966850247e-06,
      "loss": 0.7574,
      "step": 8931
    },
    {
      "epoch": 0.2573618394513917,
      "grad_norm": 0.8555911779403687,
      "learning_rate": 8.704755626774796e-06,
      "loss": 0.7759,
      "step": 8932
    },
    {
      "epoch": 0.257390652913041,
      "grad_norm": 0.8663693070411682,
      "learning_rate": 8.704442254434629e-06,
      "loss": 0.7778,
      "step": 8933
    },
    {
      "epoch": 0.25741946637469026,
      "grad_norm": 0.8446105122566223,
      "learning_rate": 8.704128849832471e-06,
      "loss": 0.7249,
      "step": 8934
    },
    {
      "epoch": 0.25744827983633956,
      "grad_norm": 0.8503268361091614,
      "learning_rate": 8.703815412971055e-06,
      "loss": 0.7458,
      "step": 8935
    },
    {
      "epoch": 0.25747709329798885,
      "grad_norm": 0.8763978481292725,
      "learning_rate": 8.703501943853108e-06,
      "loss": 0.7385,
      "step": 8936
    },
    {
      "epoch": 0.2575059067596381,
      "grad_norm": 0.8731247186660767,
      "learning_rate": 8.70318844248136e-06,
      "loss": 0.7732,
      "step": 8937
    },
    {
      "epoch": 0.25753472022128737,
      "grad_norm": 0.8545655012130737,
      "learning_rate": 8.702874908858544e-06,
      "loss": 0.7462,
      "step": 8938
    },
    {
      "epoch": 0.25756353368293666,
      "grad_norm": 0.9003003835678101,
      "learning_rate": 8.702561342987391e-06,
      "loss": 0.7646,
      "step": 8939
    },
    {
      "epoch": 0.25759234714458595,
      "grad_norm": 0.9316080212593079,
      "learning_rate": 8.70224774487063e-06,
      "loss": 0.767,
      "step": 8940
    },
    {
      "epoch": 0.25762116060623524,
      "grad_norm": 0.847370445728302,
      "learning_rate": 8.701934114510987e-06,
      "loss": 0.7424,
      "step": 8941
    },
    {
      "epoch": 0.25764997406788454,
      "grad_norm": 0.8781221508979797,
      "learning_rate": 8.701620451911203e-06,
      "loss": 0.7934,
      "step": 8942
    },
    {
      "epoch": 0.2576787875295338,
      "grad_norm": 0.9265986680984497,
      "learning_rate": 8.701306757074003e-06,
      "loss": 0.7624,
      "step": 8943
    },
    {
      "epoch": 0.25770760099118306,
      "grad_norm": 0.8627541661262512,
      "learning_rate": 8.700993030002123e-06,
      "loss": 0.7805,
      "step": 8944
    },
    {
      "epoch": 0.25773641445283235,
      "grad_norm": 0.8863325119018555,
      "learning_rate": 8.700679270698293e-06,
      "loss": 0.7484,
      "step": 8945
    },
    {
      "epoch": 0.25776522791448164,
      "grad_norm": 0.9031478762626648,
      "learning_rate": 8.700365479165246e-06,
      "loss": 0.7607,
      "step": 8946
    },
    {
      "epoch": 0.25779404137613093,
      "grad_norm": 0.847183883190155,
      "learning_rate": 8.700051655405713e-06,
      "loss": 0.7508,
      "step": 8947
    },
    {
      "epoch": 0.2578228548377802,
      "grad_norm": 0.8578971028327942,
      "learning_rate": 8.69973779942243e-06,
      "loss": 0.7553,
      "step": 8948
    },
    {
      "epoch": 0.2578516682994295,
      "grad_norm": 0.914364218711853,
      "learning_rate": 8.699423911218127e-06,
      "loss": 0.7466,
      "step": 8949
    },
    {
      "epoch": 0.25788048176107875,
      "grad_norm": 0.8855462670326233,
      "learning_rate": 8.699109990795542e-06,
      "loss": 0.754,
      "step": 8950
    },
    {
      "epoch": 0.25790929522272804,
      "grad_norm": 0.8449440598487854,
      "learning_rate": 8.698796038157407e-06,
      "loss": 0.7585,
      "step": 8951
    },
    {
      "epoch": 0.25793810868437733,
      "grad_norm": 0.8654137849807739,
      "learning_rate": 8.698482053306455e-06,
      "loss": 0.7627,
      "step": 8952
    },
    {
      "epoch": 0.2579669221460266,
      "grad_norm": 0.8488703370094299,
      "learning_rate": 8.69816803624542e-06,
      "loss": 0.7429,
      "step": 8953
    },
    {
      "epoch": 0.2579957356076759,
      "grad_norm": 0.8790054321289062,
      "learning_rate": 8.697853986977042e-06,
      "loss": 0.7594,
      "step": 8954
    },
    {
      "epoch": 0.2580245490693252,
      "grad_norm": 0.8640992045402527,
      "learning_rate": 8.69753990550405e-06,
      "loss": 0.772,
      "step": 8955
    },
    {
      "epoch": 0.2580533625309745,
      "grad_norm": 0.858239471912384,
      "learning_rate": 8.697225791829181e-06,
      "loss": 0.7676,
      "step": 8956
    },
    {
      "epoch": 0.25808217599262373,
      "grad_norm": 0.8651562929153442,
      "learning_rate": 8.696911645955172e-06,
      "loss": 0.7858,
      "step": 8957
    },
    {
      "epoch": 0.258110989454273,
      "grad_norm": 0.8366879224777222,
      "learning_rate": 8.696597467884758e-06,
      "loss": 0.7867,
      "step": 8958
    },
    {
      "epoch": 0.2581398029159223,
      "grad_norm": 0.8583800196647644,
      "learning_rate": 8.696283257620677e-06,
      "loss": 0.7613,
      "step": 8959
    },
    {
      "epoch": 0.2581686163775716,
      "grad_norm": 0.8695074915885925,
      "learning_rate": 8.695969015165662e-06,
      "loss": 0.7842,
      "step": 8960
    },
    {
      "epoch": 0.2581974298392209,
      "grad_norm": 0.8479824066162109,
      "learning_rate": 8.695654740522452e-06,
      "loss": 0.7667,
      "step": 8961
    },
    {
      "epoch": 0.2582262433008702,
      "grad_norm": 0.8676059246063232,
      "learning_rate": 8.695340433693782e-06,
      "loss": 0.7533,
      "step": 8962
    },
    {
      "epoch": 0.2582550567625195,
      "grad_norm": 0.8241903185844421,
      "learning_rate": 8.695026094682393e-06,
      "loss": 0.7623,
      "step": 8963
    },
    {
      "epoch": 0.2582838702241687,
      "grad_norm": 0.8691244125366211,
      "learning_rate": 8.694711723491018e-06,
      "loss": 0.749,
      "step": 8964
    },
    {
      "epoch": 0.258312683685818,
      "grad_norm": 0.9017667770385742,
      "learning_rate": 8.6943973201224e-06,
      "loss": 0.7948,
      "step": 8965
    },
    {
      "epoch": 0.2583414971474673,
      "grad_norm": 0.8663371801376343,
      "learning_rate": 8.694082884579273e-06,
      "loss": 0.762,
      "step": 8966
    },
    {
      "epoch": 0.2583703106091166,
      "grad_norm": 0.8832405209541321,
      "learning_rate": 8.693768416864377e-06,
      "loss": 0.774,
      "step": 8967
    },
    {
      "epoch": 0.2583991240707659,
      "grad_norm": 0.863275945186615,
      "learning_rate": 8.69345391698045e-06,
      "loss": 0.7726,
      "step": 8968
    },
    {
      "epoch": 0.25842793753241516,
      "grad_norm": 0.858721911907196,
      "learning_rate": 8.693139384930231e-06,
      "loss": 0.7753,
      "step": 8969
    },
    {
      "epoch": 0.2584567509940644,
      "grad_norm": 0.8513925075531006,
      "learning_rate": 8.692824820716462e-06,
      "loss": 0.759,
      "step": 8970
    },
    {
      "epoch": 0.2584855644557137,
      "grad_norm": 0.8457010388374329,
      "learning_rate": 8.69251022434188e-06,
      "loss": 0.7535,
      "step": 8971
    },
    {
      "epoch": 0.258514377917363,
      "grad_norm": 0.8608881831169128,
      "learning_rate": 8.692195595809224e-06,
      "loss": 0.7494,
      "step": 8972
    },
    {
      "epoch": 0.25854319137901227,
      "grad_norm": 0.8594235777854919,
      "learning_rate": 8.691880935121232e-06,
      "loss": 0.7643,
      "step": 8973
    },
    {
      "epoch": 0.25857200484066156,
      "grad_norm": 0.8092212080955505,
      "learning_rate": 8.691566242280653e-06,
      "loss": 0.7626,
      "step": 8974
    },
    {
      "epoch": 0.25860081830231085,
      "grad_norm": 0.8388246893882751,
      "learning_rate": 8.691251517290221e-06,
      "loss": 0.7404,
      "step": 8975
    },
    {
      "epoch": 0.25862963176396014,
      "grad_norm": 0.8677012920379639,
      "learning_rate": 8.690936760152675e-06,
      "loss": 0.7644,
      "step": 8976
    },
    {
      "epoch": 0.2586584452256094,
      "grad_norm": 0.91659015417099,
      "learning_rate": 8.690621970870764e-06,
      "loss": 0.763,
      "step": 8977
    },
    {
      "epoch": 0.25868725868725867,
      "grad_norm": 0.8613842725753784,
      "learning_rate": 8.690307149447222e-06,
      "loss": 0.7548,
      "step": 8978
    },
    {
      "epoch": 0.25871607214890796,
      "grad_norm": 0.8430839776992798,
      "learning_rate": 8.689992295884793e-06,
      "loss": 0.7755,
      "step": 8979
    },
    {
      "epoch": 0.25874488561055725,
      "grad_norm": 0.9019685983657837,
      "learning_rate": 8.689677410186221e-06,
      "loss": 0.7708,
      "step": 8980
    },
    {
      "epoch": 0.25877369907220654,
      "grad_norm": 0.8576926589012146,
      "learning_rate": 8.689362492354248e-06,
      "loss": 0.7586,
      "step": 8981
    },
    {
      "epoch": 0.25880251253385583,
      "grad_norm": 0.8444124460220337,
      "learning_rate": 8.689047542391613e-06,
      "loss": 0.7598,
      "step": 8982
    },
    {
      "epoch": 0.2588313259955051,
      "grad_norm": 0.835210382938385,
      "learning_rate": 8.688732560301064e-06,
      "loss": 0.7699,
      "step": 8983
    },
    {
      "epoch": 0.25886013945715436,
      "grad_norm": 0.8446855545043945,
      "learning_rate": 8.68841754608534e-06,
      "loss": 0.7498,
      "step": 8984
    },
    {
      "epoch": 0.25888895291880365,
      "grad_norm": 0.8531416654586792,
      "learning_rate": 8.688102499747189e-06,
      "loss": 0.7642,
      "step": 8985
    },
    {
      "epoch": 0.25891776638045294,
      "grad_norm": 0.8430548310279846,
      "learning_rate": 8.687787421289349e-06,
      "loss": 0.7843,
      "step": 8986
    },
    {
      "epoch": 0.25894657984210223,
      "grad_norm": 0.8612497448921204,
      "learning_rate": 8.687472310714568e-06,
      "loss": 0.7551,
      "step": 8987
    },
    {
      "epoch": 0.2589753933037515,
      "grad_norm": 0.8670886158943176,
      "learning_rate": 8.68715716802559e-06,
      "loss": 0.76,
      "step": 8988
    },
    {
      "epoch": 0.2590042067654008,
      "grad_norm": 0.8554362654685974,
      "learning_rate": 8.686841993225159e-06,
      "loss": 0.7467,
      "step": 8989
    },
    {
      "epoch": 0.2590330202270501,
      "grad_norm": 0.8728551268577576,
      "learning_rate": 8.686526786316021e-06,
      "loss": 0.7694,
      "step": 8990
    },
    {
      "epoch": 0.25906183368869934,
      "grad_norm": 0.8266108632087708,
      "learning_rate": 8.686211547300917e-06,
      "loss": 0.7565,
      "step": 8991
    },
    {
      "epoch": 0.25909064715034863,
      "grad_norm": 0.8310301303863525,
      "learning_rate": 8.685896276182595e-06,
      "loss": 0.7477,
      "step": 8992
    },
    {
      "epoch": 0.2591194606119979,
      "grad_norm": 0.8528779149055481,
      "learning_rate": 8.685580972963803e-06,
      "loss": 0.7581,
      "step": 8993
    },
    {
      "epoch": 0.2591482740736472,
      "grad_norm": 0.8845685124397278,
      "learning_rate": 8.685265637647284e-06,
      "loss": 0.7852,
      "step": 8994
    },
    {
      "epoch": 0.2591770875352965,
      "grad_norm": 0.876662015914917,
      "learning_rate": 8.684950270235786e-06,
      "loss": 0.7629,
      "step": 8995
    },
    {
      "epoch": 0.2592059009969458,
      "grad_norm": 0.8884968161582947,
      "learning_rate": 8.684634870732054e-06,
      "loss": 0.7235,
      "step": 8996
    },
    {
      "epoch": 0.25923471445859503,
      "grad_norm": 0.8655797243118286,
      "learning_rate": 8.684319439138837e-06,
      "loss": 0.7397,
      "step": 8997
    },
    {
      "epoch": 0.2592635279202443,
      "grad_norm": 0.8588024973869324,
      "learning_rate": 8.684003975458878e-06,
      "loss": 0.7442,
      "step": 8998
    },
    {
      "epoch": 0.2592923413818936,
      "grad_norm": 0.8755157589912415,
      "learning_rate": 8.68368847969493e-06,
      "loss": 0.7712,
      "step": 8999
    },
    {
      "epoch": 0.2593211548435429,
      "grad_norm": 0.8653261065483093,
      "learning_rate": 8.683372951849736e-06,
      "loss": 0.7499,
      "step": 9000
    },
    {
      "epoch": 0.2593499683051922,
      "grad_norm": 0.8614424467086792,
      "learning_rate": 8.683057391926047e-06,
      "loss": 0.7635,
      "step": 9001
    },
    {
      "epoch": 0.2593787817668415,
      "grad_norm": 0.8623330593109131,
      "learning_rate": 8.682741799926609e-06,
      "loss": 0.7632,
      "step": 9002
    },
    {
      "epoch": 0.2594075952284908,
      "grad_norm": 0.9040571451187134,
      "learning_rate": 8.68242617585417e-06,
      "loss": 0.7896,
      "step": 9003
    },
    {
      "epoch": 0.25943640869014,
      "grad_norm": 0.8794023990631104,
      "learning_rate": 8.682110519711481e-06,
      "loss": 0.7701,
      "step": 9004
    },
    {
      "epoch": 0.2594652221517893,
      "grad_norm": 0.8787522912025452,
      "learning_rate": 8.68179483150129e-06,
      "loss": 0.7607,
      "step": 9005
    },
    {
      "epoch": 0.2594940356134386,
      "grad_norm": 0.8597403764724731,
      "learning_rate": 8.681479111226347e-06,
      "loss": 0.7546,
      "step": 9006
    },
    {
      "epoch": 0.2595228490750879,
      "grad_norm": 0.9492928981781006,
      "learning_rate": 8.681163358889401e-06,
      "loss": 0.7805,
      "step": 9007
    },
    {
      "epoch": 0.25955166253673717,
      "grad_norm": 0.8931429386138916,
      "learning_rate": 8.6808475744932e-06,
      "loss": 0.7703,
      "step": 9008
    },
    {
      "epoch": 0.25958047599838646,
      "grad_norm": 0.880236804485321,
      "learning_rate": 8.680531758040499e-06,
      "loss": 0.7632,
      "step": 9009
    },
    {
      "epoch": 0.25960928946003575,
      "grad_norm": 0.8765155673027039,
      "learning_rate": 8.680215909534044e-06,
      "loss": 0.7733,
      "step": 9010
    },
    {
      "epoch": 0.259638102921685,
      "grad_norm": 0.8932042717933655,
      "learning_rate": 8.679900028976589e-06,
      "loss": 0.7744,
      "step": 9011
    },
    {
      "epoch": 0.2596669163833343,
      "grad_norm": 0.9066758751869202,
      "learning_rate": 8.67958411637088e-06,
      "loss": 0.7932,
      "step": 9012
    },
    {
      "epoch": 0.25969572984498357,
      "grad_norm": 0.854742705821991,
      "learning_rate": 8.679268171719675e-06,
      "loss": 0.776,
      "step": 9013
    },
    {
      "epoch": 0.25972454330663286,
      "grad_norm": 0.8877142667770386,
      "learning_rate": 8.67895219502572e-06,
      "loss": 0.765,
      "step": 9014
    },
    {
      "epoch": 0.25975335676828215,
      "grad_norm": 0.9040001034736633,
      "learning_rate": 8.67863618629177e-06,
      "loss": 0.7771,
      "step": 9015
    },
    {
      "epoch": 0.25978217022993144,
      "grad_norm": 0.8717069625854492,
      "learning_rate": 8.678320145520576e-06,
      "loss": 0.7627,
      "step": 9016
    },
    {
      "epoch": 0.25981098369158073,
      "grad_norm": 0.8643332123756409,
      "learning_rate": 8.67800407271489e-06,
      "loss": 0.7624,
      "step": 9017
    },
    {
      "epoch": 0.25983979715322997,
      "grad_norm": 1.3755505084991455,
      "learning_rate": 8.677687967877467e-06,
      "loss": 0.7651,
      "step": 9018
    },
    {
      "epoch": 0.25986861061487926,
      "grad_norm": 0.9437612295150757,
      "learning_rate": 8.677371831011056e-06,
      "loss": 0.7559,
      "step": 9019
    },
    {
      "epoch": 0.25989742407652855,
      "grad_norm": 0.8298661112785339,
      "learning_rate": 8.677055662118412e-06,
      "loss": 0.7671,
      "step": 9020
    },
    {
      "epoch": 0.25992623753817784,
      "grad_norm": 0.844119131565094,
      "learning_rate": 8.67673946120229e-06,
      "loss": 0.768,
      "step": 9021
    },
    {
      "epoch": 0.25995505099982713,
      "grad_norm": 0.8708070516586304,
      "learning_rate": 8.676423228265442e-06,
      "loss": 0.763,
      "step": 9022
    },
    {
      "epoch": 0.2599838644614764,
      "grad_norm": 0.8600725531578064,
      "learning_rate": 8.676106963310625e-06,
      "loss": 0.7657,
      "step": 9023
    },
    {
      "epoch": 0.26001267792312566,
      "grad_norm": 0.8719276189804077,
      "learning_rate": 8.67579066634059e-06,
      "loss": 0.7886,
      "step": 9024
    },
    {
      "epoch": 0.26004149138477495,
      "grad_norm": 0.8423277735710144,
      "learning_rate": 8.675474337358092e-06,
      "loss": 0.7519,
      "step": 9025
    },
    {
      "epoch": 0.26007030484642424,
      "grad_norm": 0.8995413780212402,
      "learning_rate": 8.675157976365886e-06,
      "loss": 0.7448,
      "step": 9026
    },
    {
      "epoch": 0.26009911830807353,
      "grad_norm": 0.8619572520256042,
      "learning_rate": 8.674841583366728e-06,
      "loss": 0.7685,
      "step": 9027
    },
    {
      "epoch": 0.2601279317697228,
      "grad_norm": 0.880414605140686,
      "learning_rate": 8.674525158363377e-06,
      "loss": 0.7726,
      "step": 9028
    },
    {
      "epoch": 0.2601567452313721,
      "grad_norm": 0.8830369710922241,
      "learning_rate": 8.67420870135858e-06,
      "loss": 0.7395,
      "step": 9029
    },
    {
      "epoch": 0.2601855586930214,
      "grad_norm": 0.8811079859733582,
      "learning_rate": 8.6738922123551e-06,
      "loss": 0.7756,
      "step": 9030
    },
    {
      "epoch": 0.26021437215467064,
      "grad_norm": 0.9285899996757507,
      "learning_rate": 8.67357569135569e-06,
      "loss": 0.7595,
      "step": 9031
    },
    {
      "epoch": 0.2602431856163199,
      "grad_norm": 0.9189123511314392,
      "learning_rate": 8.673259138363111e-06,
      "loss": 0.7503,
      "step": 9032
    },
    {
      "epoch": 0.2602719990779692,
      "grad_norm": 0.8836223483085632,
      "learning_rate": 8.672942553380115e-06,
      "loss": 0.7646,
      "step": 9033
    },
    {
      "epoch": 0.2603008125396185,
      "grad_norm": 0.8925027251243591,
      "learning_rate": 8.672625936409462e-06,
      "loss": 0.7669,
      "step": 9034
    },
    {
      "epoch": 0.2603296260012678,
      "grad_norm": 0.8864418864250183,
      "learning_rate": 8.672309287453908e-06,
      "loss": 0.7753,
      "step": 9035
    },
    {
      "epoch": 0.2603584394629171,
      "grad_norm": 0.8435968160629272,
      "learning_rate": 8.671992606516211e-06,
      "loss": 0.7743,
      "step": 9036
    },
    {
      "epoch": 0.2603872529245664,
      "grad_norm": 0.9023154973983765,
      "learning_rate": 8.67167589359913e-06,
      "loss": 0.7544,
      "step": 9037
    },
    {
      "epoch": 0.2604160663862156,
      "grad_norm": 0.892447829246521,
      "learning_rate": 8.67135914870542e-06,
      "loss": 0.757,
      "step": 9038
    },
    {
      "epoch": 0.2604448798478649,
      "grad_norm": 0.8441299200057983,
      "learning_rate": 8.671042371837842e-06,
      "loss": 0.7658,
      "step": 9039
    },
    {
      "epoch": 0.2604736933095142,
      "grad_norm": 0.891260027885437,
      "learning_rate": 8.670725562999156e-06,
      "loss": 0.7323,
      "step": 9040
    },
    {
      "epoch": 0.2605025067711635,
      "grad_norm": 0.8447741866111755,
      "learning_rate": 8.670408722192118e-06,
      "loss": 0.7171,
      "step": 9041
    },
    {
      "epoch": 0.2605313202328128,
      "grad_norm": 0.8524680137634277,
      "learning_rate": 8.67009184941949e-06,
      "loss": 0.7607,
      "step": 9042
    },
    {
      "epoch": 0.26056013369446207,
      "grad_norm": 0.8606159090995789,
      "learning_rate": 8.669774944684033e-06,
      "loss": 0.7788,
      "step": 9043
    },
    {
      "epoch": 0.26058894715611136,
      "grad_norm": 0.8305630087852478,
      "learning_rate": 8.669458007988502e-06,
      "loss": 0.7459,
      "step": 9044
    },
    {
      "epoch": 0.2606177606177606,
      "grad_norm": 0.9094341397285461,
      "learning_rate": 8.669141039335661e-06,
      "loss": 0.7582,
      "step": 9045
    },
    {
      "epoch": 0.2606465740794099,
      "grad_norm": 0.8495832681655884,
      "learning_rate": 8.668824038728269e-06,
      "loss": 0.7572,
      "step": 9046
    },
    {
      "epoch": 0.2606753875410592,
      "grad_norm": 0.8531652092933655,
      "learning_rate": 8.668507006169088e-06,
      "loss": 0.7623,
      "step": 9047
    },
    {
      "epoch": 0.26070420100270847,
      "grad_norm": 0.8401124477386475,
      "learning_rate": 8.668189941660876e-06,
      "loss": 0.7628,
      "step": 9048
    },
    {
      "epoch": 0.26073301446435776,
      "grad_norm": 0.8203069567680359,
      "learning_rate": 8.6678728452064e-06,
      "loss": 0.756,
      "step": 9049
    },
    {
      "epoch": 0.26076182792600705,
      "grad_norm": 0.8680259585380554,
      "learning_rate": 8.667555716808413e-06,
      "loss": 0.7475,
      "step": 9050
    },
    {
      "epoch": 0.2607906413876563,
      "grad_norm": 0.853187084197998,
      "learning_rate": 8.667238556469685e-06,
      "loss": 0.7685,
      "step": 9051
    },
    {
      "epoch": 0.2608194548493056,
      "grad_norm": 1.1754977703094482,
      "learning_rate": 8.666921364192975e-06,
      "loss": 0.752,
      "step": 9052
    },
    {
      "epoch": 0.26084826831095487,
      "grad_norm": 0.846054196357727,
      "learning_rate": 8.666604139981045e-06,
      "loss": 0.7443,
      "step": 9053
    },
    {
      "epoch": 0.26087708177260416,
      "grad_norm": 0.8729150891304016,
      "learning_rate": 8.666286883836658e-06,
      "loss": 0.7607,
      "step": 9054
    },
    {
      "epoch": 0.26090589523425345,
      "grad_norm": 0.8788877725601196,
      "learning_rate": 8.665969595762578e-06,
      "loss": 0.7718,
      "step": 9055
    },
    {
      "epoch": 0.26093470869590274,
      "grad_norm": 0.8737162351608276,
      "learning_rate": 8.665652275761569e-06,
      "loss": 0.774,
      "step": 9056
    },
    {
      "epoch": 0.26096352215755203,
      "grad_norm": 0.9487441778182983,
      "learning_rate": 8.66533492383639e-06,
      "loss": 0.7777,
      "step": 9057
    },
    {
      "epoch": 0.26099233561920127,
      "grad_norm": 0.8469837307929993,
      "learning_rate": 8.665017539989808e-06,
      "loss": 0.7666,
      "step": 9058
    },
    {
      "epoch": 0.26102114908085056,
      "grad_norm": 4.196361064910889,
      "learning_rate": 8.664700124224589e-06,
      "loss": 0.786,
      "step": 9059
    },
    {
      "epoch": 0.26104996254249985,
      "grad_norm": 0.9051464200019836,
      "learning_rate": 8.664382676543494e-06,
      "loss": 0.7803,
      "step": 9060
    },
    {
      "epoch": 0.26107877600414914,
      "grad_norm": 0.9066441059112549,
      "learning_rate": 8.664065196949289e-06,
      "loss": 0.7548,
      "step": 9061
    },
    {
      "epoch": 0.26110758946579843,
      "grad_norm": 0.9436838626861572,
      "learning_rate": 8.66374768544474e-06,
      "loss": 0.7651,
      "step": 9062
    },
    {
      "epoch": 0.2611364029274477,
      "grad_norm": 0.9150324463844299,
      "learning_rate": 8.663430142032608e-06,
      "loss": 0.7671,
      "step": 9063
    },
    {
      "epoch": 0.261165216389097,
      "grad_norm": 0.897318422794342,
      "learning_rate": 8.663112566715663e-06,
      "loss": 0.7656,
      "step": 9064
    },
    {
      "epoch": 0.26119402985074625,
      "grad_norm": 1.0556461811065674,
      "learning_rate": 8.662794959496669e-06,
      "loss": 0.7752,
      "step": 9065
    },
    {
      "epoch": 0.26122284331239554,
      "grad_norm": 0.8980481624603271,
      "learning_rate": 8.662477320378395e-06,
      "loss": 0.788,
      "step": 9066
    },
    {
      "epoch": 0.2612516567740448,
      "grad_norm": 0.8682122826576233,
      "learning_rate": 8.662159649363604e-06,
      "loss": 0.7664,
      "step": 9067
    },
    {
      "epoch": 0.2612804702356941,
      "grad_norm": 0.8976044654846191,
      "learning_rate": 8.66184194645506e-06,
      "loss": 0.7715,
      "step": 9068
    },
    {
      "epoch": 0.2613092836973434,
      "grad_norm": 0.8939946889877319,
      "learning_rate": 8.661524211655536e-06,
      "loss": 0.7727,
      "step": 9069
    },
    {
      "epoch": 0.2613380971589927,
      "grad_norm": 0.885776162147522,
      "learning_rate": 8.661206444967796e-06,
      "loss": 0.7477,
      "step": 9070
    },
    {
      "epoch": 0.261366910620642,
      "grad_norm": 0.9188781976699829,
      "learning_rate": 8.660888646394606e-06,
      "loss": 0.7931,
      "step": 9071
    },
    {
      "epoch": 0.2613957240822912,
      "grad_norm": 0.8732932209968567,
      "learning_rate": 8.660570815938736e-06,
      "loss": 0.7726,
      "step": 9072
    },
    {
      "epoch": 0.2614245375439405,
      "grad_norm": 1.8939950466156006,
      "learning_rate": 8.660252953602956e-06,
      "loss": 0.7655,
      "step": 9073
    },
    {
      "epoch": 0.2614533510055898,
      "grad_norm": 0.8968273401260376,
      "learning_rate": 8.659935059390029e-06,
      "loss": 0.734,
      "step": 9074
    },
    {
      "epoch": 0.2614821644672391,
      "grad_norm": 0.8983557820320129,
      "learning_rate": 8.659617133302727e-06,
      "loss": 0.8088,
      "step": 9075
    },
    {
      "epoch": 0.2615109779288884,
      "grad_norm": 0.9070143103599548,
      "learning_rate": 8.659299175343819e-06,
      "loss": 0.7689,
      "step": 9076
    },
    {
      "epoch": 0.2615397913905377,
      "grad_norm": 0.8650517463684082,
      "learning_rate": 8.658981185516071e-06,
      "loss": 0.7485,
      "step": 9077
    },
    {
      "epoch": 0.2615686048521869,
      "grad_norm": 0.8579095602035522,
      "learning_rate": 8.658663163822256e-06,
      "loss": 0.7609,
      "step": 9078
    },
    {
      "epoch": 0.2615974183138362,
      "grad_norm": 0.9110891222953796,
      "learning_rate": 8.658345110265143e-06,
      "loss": 0.766,
      "step": 9079
    },
    {
      "epoch": 0.2616262317754855,
      "grad_norm": 0.9016582369804382,
      "learning_rate": 8.6580270248475e-06,
      "loss": 0.7763,
      "step": 9080
    },
    {
      "epoch": 0.2616550452371348,
      "grad_norm": 0.8539788722991943,
      "learning_rate": 8.657708907572098e-06,
      "loss": 0.764,
      "step": 9081
    },
    {
      "epoch": 0.2616838586987841,
      "grad_norm": 0.9031374454498291,
      "learning_rate": 8.657390758441709e-06,
      "loss": 0.7636,
      "step": 9082
    },
    {
      "epoch": 0.26171267216043337,
      "grad_norm": 0.8629840612411499,
      "learning_rate": 8.6570725774591e-06,
      "loss": 0.7633,
      "step": 9083
    },
    {
      "epoch": 0.26174148562208266,
      "grad_norm": 0.8706642985343933,
      "learning_rate": 8.656754364627047e-06,
      "loss": 0.7387,
      "step": 9084
    },
    {
      "epoch": 0.2617702990837319,
      "grad_norm": 0.9092886447906494,
      "learning_rate": 8.656436119948318e-06,
      "loss": 0.7683,
      "step": 9085
    },
    {
      "epoch": 0.2617991125453812,
      "grad_norm": 0.8563299179077148,
      "learning_rate": 8.656117843425686e-06,
      "loss": 0.7715,
      "step": 9086
    },
    {
      "epoch": 0.2618279260070305,
      "grad_norm": 0.9264022707939148,
      "learning_rate": 8.655799535061922e-06,
      "loss": 0.7776,
      "step": 9087
    },
    {
      "epoch": 0.26185673946867977,
      "grad_norm": 0.8787239789962769,
      "learning_rate": 8.655481194859799e-06,
      "loss": 0.7629,
      "step": 9088
    },
    {
      "epoch": 0.26188555293032906,
      "grad_norm": 0.8431001305580139,
      "learning_rate": 8.655162822822087e-06,
      "loss": 0.776,
      "step": 9089
    },
    {
      "epoch": 0.26191436639197835,
      "grad_norm": 0.8798233866691589,
      "learning_rate": 8.654844418951562e-06,
      "loss": 0.7682,
      "step": 9090
    },
    {
      "epoch": 0.26194317985362764,
      "grad_norm": 0.8651798963546753,
      "learning_rate": 8.654525983250997e-06,
      "loss": 0.7639,
      "step": 9091
    },
    {
      "epoch": 0.2619719933152769,
      "grad_norm": 1.2423722743988037,
      "learning_rate": 8.65420751572316e-06,
      "loss": 0.7823,
      "step": 9092
    },
    {
      "epoch": 0.26200080677692617,
      "grad_norm": 0.9218684434890747,
      "learning_rate": 8.653889016370833e-06,
      "loss": 0.7465,
      "step": 9093
    },
    {
      "epoch": 0.26202962023857546,
      "grad_norm": 0.8786112070083618,
      "learning_rate": 8.653570485196784e-06,
      "loss": 0.758,
      "step": 9094
    },
    {
      "epoch": 0.26205843370022475,
      "grad_norm": 0.8605167865753174,
      "learning_rate": 8.653251922203787e-06,
      "loss": 0.7648,
      "step": 9095
    },
    {
      "epoch": 0.26208724716187404,
      "grad_norm": 0.8651785850524902,
      "learning_rate": 8.652933327394617e-06,
      "loss": 0.7553,
      "step": 9096
    },
    {
      "epoch": 0.26211606062352333,
      "grad_norm": 0.8936665058135986,
      "learning_rate": 8.65261470077205e-06,
      "loss": 0.7499,
      "step": 9097
    },
    {
      "epoch": 0.2621448740851726,
      "grad_norm": 0.872284471988678,
      "learning_rate": 8.652296042338861e-06,
      "loss": 0.7573,
      "step": 9098
    },
    {
      "epoch": 0.26217368754682185,
      "grad_norm": 0.9138461947441101,
      "learning_rate": 8.651977352097825e-06,
      "loss": 0.7785,
      "step": 9099
    },
    {
      "epoch": 0.26220250100847114,
      "grad_norm": 0.9127200245857239,
      "learning_rate": 8.651658630051716e-06,
      "loss": 0.7681,
      "step": 9100
    },
    {
      "epoch": 0.26223131447012044,
      "grad_norm": 0.8963853716850281,
      "learning_rate": 8.651339876203312e-06,
      "loss": 0.7703,
      "step": 9101
    },
    {
      "epoch": 0.2622601279317697,
      "grad_norm": 0.8614944815635681,
      "learning_rate": 8.651021090555384e-06,
      "loss": 0.7704,
      "step": 9102
    },
    {
      "epoch": 0.262288941393419,
      "grad_norm": 0.8518067598342896,
      "learning_rate": 8.650702273110716e-06,
      "loss": 0.7448,
      "step": 9103
    },
    {
      "epoch": 0.2623177548550683,
      "grad_norm": 0.8787591457366943,
      "learning_rate": 8.650383423872078e-06,
      "loss": 0.785,
      "step": 9104
    },
    {
      "epoch": 0.26234656831671754,
      "grad_norm": 0.8798320293426514,
      "learning_rate": 8.650064542842252e-06,
      "loss": 0.785,
      "step": 9105
    },
    {
      "epoch": 0.26237538177836683,
      "grad_norm": 0.8985924124717712,
      "learning_rate": 8.64974563002401e-06,
      "loss": 0.7564,
      "step": 9106
    },
    {
      "epoch": 0.2624041952400161,
      "grad_norm": 0.8706004023551941,
      "learning_rate": 8.649426685420133e-06,
      "loss": 0.743,
      "step": 9107
    },
    {
      "epoch": 0.2624330087016654,
      "grad_norm": 0.908129096031189,
      "learning_rate": 8.649107709033397e-06,
      "loss": 0.7691,
      "step": 9108
    },
    {
      "epoch": 0.2624618221633147,
      "grad_norm": 0.8188726305961609,
      "learning_rate": 8.648788700866582e-06,
      "loss": 0.7517,
      "step": 9109
    },
    {
      "epoch": 0.262490635624964,
      "grad_norm": 0.8553943634033203,
      "learning_rate": 8.648469660922462e-06,
      "loss": 0.7592,
      "step": 9110
    },
    {
      "epoch": 0.2625194490866133,
      "grad_norm": 0.9087488055229187,
      "learning_rate": 8.648150589203822e-06,
      "loss": 0.7521,
      "step": 9111
    },
    {
      "epoch": 0.2625482625482625,
      "grad_norm": 0.8746472001075745,
      "learning_rate": 8.647831485713436e-06,
      "loss": 0.7686,
      "step": 9112
    },
    {
      "epoch": 0.2625770760099118,
      "grad_norm": 0.8550740480422974,
      "learning_rate": 8.647512350454084e-06,
      "loss": 0.7552,
      "step": 9113
    },
    {
      "epoch": 0.2626058894715611,
      "grad_norm": 1.41226065158844,
      "learning_rate": 8.647193183428546e-06,
      "loss": 0.7606,
      "step": 9114
    },
    {
      "epoch": 0.2626347029332104,
      "grad_norm": 0.8785160183906555,
      "learning_rate": 8.6468739846396e-06,
      "loss": 0.759,
      "step": 9115
    },
    {
      "epoch": 0.2626635163948597,
      "grad_norm": 0.8926711082458496,
      "learning_rate": 8.646554754090028e-06,
      "loss": 0.7742,
      "step": 9116
    },
    {
      "epoch": 0.262692329856509,
      "grad_norm": 0.820232629776001,
      "learning_rate": 8.646235491782609e-06,
      "loss": 0.766,
      "step": 9117
    },
    {
      "epoch": 0.26272114331815827,
      "grad_norm": 2.340024471282959,
      "learning_rate": 8.645916197720123e-06,
      "loss": 0.7678,
      "step": 9118
    },
    {
      "epoch": 0.2627499567798075,
      "grad_norm": 0.8660359978675842,
      "learning_rate": 8.645596871905353e-06,
      "loss": 0.7639,
      "step": 9119
    },
    {
      "epoch": 0.2627787702414568,
      "grad_norm": 0.864619791507721,
      "learning_rate": 8.645277514341078e-06,
      "loss": 0.7734,
      "step": 9120
    },
    {
      "epoch": 0.2628075837031061,
      "grad_norm": 0.8425736427307129,
      "learning_rate": 8.64495812503008e-06,
      "loss": 0.7542,
      "step": 9121
    },
    {
      "epoch": 0.2628363971647554,
      "grad_norm": 0.8689376711845398,
      "learning_rate": 8.64463870397514e-06,
      "loss": 0.7612,
      "step": 9122
    },
    {
      "epoch": 0.26286521062640467,
      "grad_norm": 0.9054428935050964,
      "learning_rate": 8.644319251179042e-06,
      "loss": 0.7472,
      "step": 9123
    },
    {
      "epoch": 0.26289402408805396,
      "grad_norm": 0.945077121257782,
      "learning_rate": 8.643999766644563e-06,
      "loss": 0.7841,
      "step": 9124
    },
    {
      "epoch": 0.26292283754970325,
      "grad_norm": 0.8525294065475464,
      "learning_rate": 8.643680250374491e-06,
      "loss": 0.7618,
      "step": 9125
    },
    {
      "epoch": 0.2629516510113525,
      "grad_norm": 0.8896158337593079,
      "learning_rate": 8.643360702371605e-06,
      "loss": 0.7725,
      "step": 9126
    },
    {
      "epoch": 0.2629804644730018,
      "grad_norm": 0.8810176849365234,
      "learning_rate": 8.64304112263869e-06,
      "loss": 0.7693,
      "step": 9127
    },
    {
      "epoch": 0.26300927793465106,
      "grad_norm": 0.9210554957389832,
      "learning_rate": 8.64272151117853e-06,
      "loss": 0.7833,
      "step": 9128
    },
    {
      "epoch": 0.26303809139630036,
      "grad_norm": 0.887563169002533,
      "learning_rate": 8.642401867993904e-06,
      "loss": 0.7836,
      "step": 9129
    },
    {
      "epoch": 0.26306690485794965,
      "grad_norm": 0.8345038890838623,
      "learning_rate": 8.6420821930876e-06,
      "loss": 0.7551,
      "step": 9130
    },
    {
      "epoch": 0.26309571831959894,
      "grad_norm": 0.8976230621337891,
      "learning_rate": 8.641762486462401e-06,
      "loss": 0.7691,
      "step": 9131
    },
    {
      "epoch": 0.26312453178124817,
      "grad_norm": 0.9293687343597412,
      "learning_rate": 8.641442748121091e-06,
      "loss": 0.7498,
      "step": 9132
    },
    {
      "epoch": 0.26315334524289746,
      "grad_norm": 0.9126543998718262,
      "learning_rate": 8.641122978066457e-06,
      "loss": 0.7759,
      "step": 9133
    },
    {
      "epoch": 0.26318215870454675,
      "grad_norm": 0.8943958282470703,
      "learning_rate": 8.640803176301279e-06,
      "loss": 0.7656,
      "step": 9134
    },
    {
      "epoch": 0.26321097216619604,
      "grad_norm": 0.9030399322509766,
      "learning_rate": 8.640483342828345e-06,
      "loss": 0.7533,
      "step": 9135
    },
    {
      "epoch": 0.26323978562784534,
      "grad_norm": 0.9399629235267639,
      "learning_rate": 8.640163477650442e-06,
      "loss": 0.7878,
      "step": 9136
    },
    {
      "epoch": 0.2632685990894946,
      "grad_norm": 0.8816527724266052,
      "learning_rate": 8.639843580770352e-06,
      "loss": 0.7455,
      "step": 9137
    },
    {
      "epoch": 0.2632974125511439,
      "grad_norm": 0.8767692446708679,
      "learning_rate": 8.639523652190863e-06,
      "loss": 0.7258,
      "step": 9138
    },
    {
      "epoch": 0.26332622601279315,
      "grad_norm": 0.9576400518417358,
      "learning_rate": 8.639203691914762e-06,
      "loss": 0.7734,
      "step": 9139
    },
    {
      "epoch": 0.26335503947444244,
      "grad_norm": 0.939460277557373,
      "learning_rate": 8.638883699944835e-06,
      "loss": 0.7485,
      "step": 9140
    },
    {
      "epoch": 0.26338385293609173,
      "grad_norm": 0.8814387917518616,
      "learning_rate": 8.63856367628387e-06,
      "loss": 0.7438,
      "step": 9141
    },
    {
      "epoch": 0.263412666397741,
      "grad_norm": 0.8950255513191223,
      "learning_rate": 8.63824362093465e-06,
      "loss": 0.7724,
      "step": 9142
    },
    {
      "epoch": 0.2634414798593903,
      "grad_norm": 0.9132986664772034,
      "learning_rate": 8.637923533899965e-06,
      "loss": 0.7703,
      "step": 9143
    },
    {
      "epoch": 0.2634702933210396,
      "grad_norm": 0.9104281067848206,
      "learning_rate": 8.637603415182603e-06,
      "loss": 0.7754,
      "step": 9144
    },
    {
      "epoch": 0.2634991067826889,
      "grad_norm": 0.860718309879303,
      "learning_rate": 8.637283264785352e-06,
      "loss": 0.7833,
      "step": 9145
    },
    {
      "epoch": 0.26352792024433813,
      "grad_norm": 0.9221789836883545,
      "learning_rate": 8.636963082711e-06,
      "loss": 0.7697,
      "step": 9146
    },
    {
      "epoch": 0.2635567337059874,
      "grad_norm": 0.8881267309188843,
      "learning_rate": 8.636642868962335e-06,
      "loss": 0.7675,
      "step": 9147
    },
    {
      "epoch": 0.2635855471676367,
      "grad_norm": 0.8602201342582703,
      "learning_rate": 8.636322623542145e-06,
      "loss": 0.7526,
      "step": 9148
    },
    {
      "epoch": 0.263614360629286,
      "grad_norm": 0.9225722551345825,
      "learning_rate": 8.63600234645322e-06,
      "loss": 0.7406,
      "step": 9149
    },
    {
      "epoch": 0.2636431740909353,
      "grad_norm": 0.8816801905632019,
      "learning_rate": 8.63568203769835e-06,
      "loss": 0.7414,
      "step": 9150
    },
    {
      "epoch": 0.2636719875525846,
      "grad_norm": 0.9322827458381653,
      "learning_rate": 8.635361697280325e-06,
      "loss": 0.7374,
      "step": 9151
    },
    {
      "epoch": 0.2637008010142339,
      "grad_norm": 0.9095190763473511,
      "learning_rate": 8.635041325201932e-06,
      "loss": 0.7749,
      "step": 9152
    },
    {
      "epoch": 0.2637296144758831,
      "grad_norm": 0.9356779456138611,
      "learning_rate": 8.634720921465963e-06,
      "loss": 0.8035,
      "step": 9153
    },
    {
      "epoch": 0.2637584279375324,
      "grad_norm": 0.9302730560302734,
      "learning_rate": 8.634400486075207e-06,
      "loss": 0.7851,
      "step": 9154
    },
    {
      "epoch": 0.2637872413991817,
      "grad_norm": 0.9157454371452332,
      "learning_rate": 8.634080019032459e-06,
      "loss": 0.7665,
      "step": 9155
    },
    {
      "epoch": 0.263816054860831,
      "grad_norm": 0.9027175307273865,
      "learning_rate": 8.633759520340503e-06,
      "loss": 0.7636,
      "step": 9156
    },
    {
      "epoch": 0.2638448683224803,
      "grad_norm": 0.878035306930542,
      "learning_rate": 8.633438990002136e-06,
      "loss": 0.7482,
      "step": 9157
    },
    {
      "epoch": 0.26387368178412957,
      "grad_norm": 0.8669511079788208,
      "learning_rate": 8.633118428020147e-06,
      "loss": 0.7642,
      "step": 9158
    },
    {
      "epoch": 0.2639024952457788,
      "grad_norm": 0.8682480454444885,
      "learning_rate": 8.632797834397328e-06,
      "loss": 0.738,
      "step": 9159
    },
    {
      "epoch": 0.2639313087074281,
      "grad_norm": 0.9332042336463928,
      "learning_rate": 8.632477209136475e-06,
      "loss": 0.7809,
      "step": 9160
    },
    {
      "epoch": 0.2639601221690774,
      "grad_norm": 0.825905442237854,
      "learning_rate": 8.632156552240372e-06,
      "loss": 0.7677,
      "step": 9161
    },
    {
      "epoch": 0.2639889356307267,
      "grad_norm": 0.8436309099197388,
      "learning_rate": 8.631835863711817e-06,
      "loss": 0.8011,
      "step": 9162
    },
    {
      "epoch": 0.26401774909237596,
      "grad_norm": 0.894512414932251,
      "learning_rate": 8.631515143553604e-06,
      "loss": 0.7704,
      "step": 9163
    },
    {
      "epoch": 0.26404656255402525,
      "grad_norm": 0.8559101819992065,
      "learning_rate": 8.631194391768523e-06,
      "loss": 0.773,
      "step": 9164
    },
    {
      "epoch": 0.26407537601567455,
      "grad_norm": 0.860054075717926,
      "learning_rate": 8.63087360835937e-06,
      "loss": 0.7432,
      "step": 9165
    },
    {
      "epoch": 0.2641041894773238,
      "grad_norm": 0.8603527545928955,
      "learning_rate": 8.630552793328936e-06,
      "loss": 0.7708,
      "step": 9166
    },
    {
      "epoch": 0.26413300293897307,
      "grad_norm": 0.8425829410552979,
      "learning_rate": 8.630231946680016e-06,
      "loss": 0.7543,
      "step": 9167
    },
    {
      "epoch": 0.26416181640062236,
      "grad_norm": 0.8372218608856201,
      "learning_rate": 8.629911068415407e-06,
      "loss": 0.7628,
      "step": 9168
    },
    {
      "epoch": 0.26419062986227165,
      "grad_norm": 0.881159245967865,
      "learning_rate": 8.629590158537899e-06,
      "loss": 0.773,
      "step": 9169
    },
    {
      "epoch": 0.26421944332392094,
      "grad_norm": 0.8659399747848511,
      "learning_rate": 8.62926921705029e-06,
      "loss": 0.7524,
      "step": 9170
    },
    {
      "epoch": 0.26424825678557023,
      "grad_norm": 0.8836742639541626,
      "learning_rate": 8.628948243955373e-06,
      "loss": 0.7567,
      "step": 9171
    },
    {
      "epoch": 0.2642770702472195,
      "grad_norm": 0.8806372880935669,
      "learning_rate": 8.628627239255945e-06,
      "loss": 0.7768,
      "step": 9172
    },
    {
      "epoch": 0.26430588370886876,
      "grad_norm": 0.9237977266311646,
      "learning_rate": 8.628306202954801e-06,
      "loss": 0.7766,
      "step": 9173
    },
    {
      "epoch": 0.26433469717051805,
      "grad_norm": 0.8438715934753418,
      "learning_rate": 8.627985135054738e-06,
      "loss": 0.7653,
      "step": 9174
    },
    {
      "epoch": 0.26436351063216734,
      "grad_norm": 0.8647121787071228,
      "learning_rate": 8.62766403555855e-06,
      "loss": 0.7551,
      "step": 9175
    },
    {
      "epoch": 0.26439232409381663,
      "grad_norm": 0.88582843542099,
      "learning_rate": 8.627342904469035e-06,
      "loss": 0.7431,
      "step": 9176
    },
    {
      "epoch": 0.2644211375554659,
      "grad_norm": 0.8545365333557129,
      "learning_rate": 8.627021741788988e-06,
      "loss": 0.7918,
      "step": 9177
    },
    {
      "epoch": 0.2644499510171152,
      "grad_norm": 0.8488179445266724,
      "learning_rate": 8.62670054752121e-06,
      "loss": 0.7736,
      "step": 9178
    },
    {
      "epoch": 0.2644787644787645,
      "grad_norm": 0.8600279688835144,
      "learning_rate": 8.626379321668492e-06,
      "loss": 0.7601,
      "step": 9179
    },
    {
      "epoch": 0.26450757794041374,
      "grad_norm": 0.8527200222015381,
      "learning_rate": 8.626058064233638e-06,
      "loss": 0.7648,
      "step": 9180
    },
    {
      "epoch": 0.26453639140206303,
      "grad_norm": 0.8754947185516357,
      "learning_rate": 8.625736775219442e-06,
      "loss": 0.7736,
      "step": 9181
    },
    {
      "epoch": 0.2645652048637123,
      "grad_norm": 0.8248688578605652,
      "learning_rate": 8.625415454628703e-06,
      "loss": 0.7744,
      "step": 9182
    },
    {
      "epoch": 0.2645940183253616,
      "grad_norm": 0.8991623520851135,
      "learning_rate": 8.62509410246422e-06,
      "loss": 0.7873,
      "step": 9183
    },
    {
      "epoch": 0.2646228317870109,
      "grad_norm": 0.9081084132194519,
      "learning_rate": 8.624772718728792e-06,
      "loss": 0.7756,
      "step": 9184
    },
    {
      "epoch": 0.2646516452486602,
      "grad_norm": 0.8579061031341553,
      "learning_rate": 8.624451303425216e-06,
      "loss": 0.7482,
      "step": 9185
    },
    {
      "epoch": 0.26468045871030943,
      "grad_norm": 0.8720661997795105,
      "learning_rate": 8.624129856556291e-06,
      "loss": 0.7759,
      "step": 9186
    },
    {
      "epoch": 0.2647092721719587,
      "grad_norm": 0.9007221460342407,
      "learning_rate": 8.623808378124819e-06,
      "loss": 0.7583,
      "step": 9187
    },
    {
      "epoch": 0.264738085633608,
      "grad_norm": 0.8381876945495605,
      "learning_rate": 8.6234868681336e-06,
      "loss": 0.747,
      "step": 9188
    },
    {
      "epoch": 0.2647668990952573,
      "grad_norm": 0.8815913796424866,
      "learning_rate": 8.62316532658543e-06,
      "loss": 0.7602,
      "step": 9189
    },
    {
      "epoch": 0.2647957125569066,
      "grad_norm": 0.9095628261566162,
      "learning_rate": 8.622843753483112e-06,
      "loss": 0.7624,
      "step": 9190
    },
    {
      "epoch": 0.2648245260185559,
      "grad_norm": 0.8842479586601257,
      "learning_rate": 8.622522148829448e-06,
      "loss": 0.7737,
      "step": 9191
    },
    {
      "epoch": 0.2648533394802052,
      "grad_norm": 0.8613831996917725,
      "learning_rate": 8.622200512627237e-06,
      "loss": 0.7403,
      "step": 9192
    },
    {
      "epoch": 0.2648821529418544,
      "grad_norm": 0.8984231948852539,
      "learning_rate": 8.62187884487928e-06,
      "loss": 0.7857,
      "step": 9193
    },
    {
      "epoch": 0.2649109664035037,
      "grad_norm": 0.8690957427024841,
      "learning_rate": 8.62155714558838e-06,
      "loss": 0.7515,
      "step": 9194
    },
    {
      "epoch": 0.264939779865153,
      "grad_norm": 0.8315949440002441,
      "learning_rate": 8.621235414757337e-06,
      "loss": 0.7708,
      "step": 9195
    },
    {
      "epoch": 0.2649685933268023,
      "grad_norm": 0.8260254859924316,
      "learning_rate": 8.620913652388951e-06,
      "loss": 0.7621,
      "step": 9196
    },
    {
      "epoch": 0.2649974067884516,
      "grad_norm": 0.8638678789138794,
      "learning_rate": 8.62059185848603e-06,
      "loss": 0.7567,
      "step": 9197
    },
    {
      "epoch": 0.26502622025010086,
      "grad_norm": 0.8397841453552246,
      "learning_rate": 8.62027003305137e-06,
      "loss": 0.7586,
      "step": 9198
    },
    {
      "epoch": 0.26505503371175015,
      "grad_norm": 0.8564886450767517,
      "learning_rate": 8.619948176087779e-06,
      "loss": 0.7699,
      "step": 9199
    },
    {
      "epoch": 0.2650838471733994,
      "grad_norm": 0.8183078169822693,
      "learning_rate": 8.619626287598058e-06,
      "loss": 0.7745,
      "step": 9200
    },
    {
      "epoch": 0.2651126606350487,
      "grad_norm": 0.8743904829025269,
      "learning_rate": 8.61930436758501e-06,
      "loss": 0.7564,
      "step": 9201
    },
    {
      "epoch": 0.26514147409669797,
      "grad_norm": 0.8908920884132385,
      "learning_rate": 8.618982416051437e-06,
      "loss": 0.7847,
      "step": 9202
    },
    {
      "epoch": 0.26517028755834726,
      "grad_norm": 0.8605183959007263,
      "learning_rate": 8.618660433000147e-06,
      "loss": 0.7526,
      "step": 9203
    },
    {
      "epoch": 0.26519910101999655,
      "grad_norm": 0.8666764497756958,
      "learning_rate": 8.618338418433943e-06,
      "loss": 0.7568,
      "step": 9204
    },
    {
      "epoch": 0.26522791448164584,
      "grad_norm": 1.0624620914459229,
      "learning_rate": 8.618016372355626e-06,
      "loss": 0.7685,
      "step": 9205
    },
    {
      "epoch": 0.26525672794329513,
      "grad_norm": 0.9017935395240784,
      "learning_rate": 8.617694294768003e-06,
      "loss": 0.8037,
      "step": 9206
    },
    {
      "epoch": 0.26528554140494437,
      "grad_norm": 0.844092071056366,
      "learning_rate": 8.617372185673883e-06,
      "loss": 0.7633,
      "step": 9207
    },
    {
      "epoch": 0.26531435486659366,
      "grad_norm": 0.8666446208953857,
      "learning_rate": 8.617050045076064e-06,
      "loss": 0.7428,
      "step": 9208
    },
    {
      "epoch": 0.26534316832824295,
      "grad_norm": 0.9980646967887878,
      "learning_rate": 8.616727872977355e-06,
      "loss": 0.7574,
      "step": 9209
    },
    {
      "epoch": 0.26537198178989224,
      "grad_norm": 0.860288679599762,
      "learning_rate": 8.616405669380561e-06,
      "loss": 0.7679,
      "step": 9210
    },
    {
      "epoch": 0.26540079525154153,
      "grad_norm": 0.8954000473022461,
      "learning_rate": 8.616083434288492e-06,
      "loss": 0.7757,
      "step": 9211
    },
    {
      "epoch": 0.2654296087131908,
      "grad_norm": 0.8750998377799988,
      "learning_rate": 8.615761167703946e-06,
      "loss": 0.778,
      "step": 9212
    },
    {
      "epoch": 0.26545842217484006,
      "grad_norm": 0.8551232814788818,
      "learning_rate": 8.61543886962974e-06,
      "loss": 0.7576,
      "step": 9213
    },
    {
      "epoch": 0.26548723563648935,
      "grad_norm": 0.8628852367401123,
      "learning_rate": 8.615116540068672e-06,
      "loss": 0.7781,
      "step": 9214
    },
    {
      "epoch": 0.26551604909813864,
      "grad_norm": 0.8489047288894653,
      "learning_rate": 8.614794179023554e-06,
      "loss": 0.7659,
      "step": 9215
    },
    {
      "epoch": 0.26554486255978793,
      "grad_norm": 0.8584075570106506,
      "learning_rate": 8.614471786497192e-06,
      "loss": 0.7579,
      "step": 9216
    },
    {
      "epoch": 0.2655736760214372,
      "grad_norm": 0.8405713438987732,
      "learning_rate": 8.614149362492393e-06,
      "loss": 0.7873,
      "step": 9217
    },
    {
      "epoch": 0.2656024894830865,
      "grad_norm": 0.8908379673957825,
      "learning_rate": 8.613826907011967e-06,
      "loss": 0.784,
      "step": 9218
    },
    {
      "epoch": 0.2656313029447358,
      "grad_norm": 0.8901435732841492,
      "learning_rate": 8.61350442005872e-06,
      "loss": 0.7531,
      "step": 9219
    },
    {
      "epoch": 0.26566011640638504,
      "grad_norm": 0.8303987383842468,
      "learning_rate": 8.613181901635463e-06,
      "loss": 0.7325,
      "step": 9220
    },
    {
      "epoch": 0.26568892986803433,
      "grad_norm": 0.8867891430854797,
      "learning_rate": 8.612859351745002e-06,
      "loss": 0.7811,
      "step": 9221
    },
    {
      "epoch": 0.2657177433296836,
      "grad_norm": 0.9066758751869202,
      "learning_rate": 8.61253677039015e-06,
      "loss": 0.7698,
      "step": 9222
    },
    {
      "epoch": 0.2657465567913329,
      "grad_norm": 0.891703188419342,
      "learning_rate": 8.612214157573712e-06,
      "loss": 0.7515,
      "step": 9223
    },
    {
      "epoch": 0.2657753702529822,
      "grad_norm": 0.8526853322982788,
      "learning_rate": 8.6118915132985e-06,
      "loss": 0.7632,
      "step": 9224
    },
    {
      "epoch": 0.2658041837146315,
      "grad_norm": 0.8910826444625854,
      "learning_rate": 8.611568837567323e-06,
      "loss": 0.7635,
      "step": 9225
    },
    {
      "epoch": 0.2658329971762808,
      "grad_norm": 0.8663515448570251,
      "learning_rate": 8.611246130382993e-06,
      "loss": 0.7537,
      "step": 9226
    },
    {
      "epoch": 0.26586181063793,
      "grad_norm": 0.8835845589637756,
      "learning_rate": 8.610923391748319e-06,
      "loss": 0.7577,
      "step": 9227
    },
    {
      "epoch": 0.2658906240995793,
      "grad_norm": 0.8775052428245544,
      "learning_rate": 8.61060062166611e-06,
      "loss": 0.7697,
      "step": 9228
    },
    {
      "epoch": 0.2659194375612286,
      "grad_norm": 0.8789345026016235,
      "learning_rate": 8.610277820139182e-06,
      "loss": 0.7598,
      "step": 9229
    },
    {
      "epoch": 0.2659482510228779,
      "grad_norm": 0.8754284977912903,
      "learning_rate": 8.609954987170338e-06,
      "loss": 0.7347,
      "step": 9230
    },
    {
      "epoch": 0.2659770644845272,
      "grad_norm": 0.8947954177856445,
      "learning_rate": 8.6096321227624e-06,
      "loss": 0.7553,
      "step": 9231
    },
    {
      "epoch": 0.26600587794617647,
      "grad_norm": 0.8773005604743958,
      "learning_rate": 8.609309226918172e-06,
      "loss": 0.7869,
      "step": 9232
    },
    {
      "epoch": 0.26603469140782576,
      "grad_norm": 0.8519028425216675,
      "learning_rate": 8.60898629964047e-06,
      "loss": 0.7526,
      "step": 9233
    },
    {
      "epoch": 0.266063504869475,
      "grad_norm": 0.8434095978736877,
      "learning_rate": 8.608663340932105e-06,
      "loss": 0.7753,
      "step": 9234
    },
    {
      "epoch": 0.2660923183311243,
      "grad_norm": 0.8352457284927368,
      "learning_rate": 8.608340350795888e-06,
      "loss": 0.7721,
      "step": 9235
    },
    {
      "epoch": 0.2661211317927736,
      "grad_norm": 0.8516271114349365,
      "learning_rate": 8.608017329234633e-06,
      "loss": 0.7639,
      "step": 9236
    },
    {
      "epoch": 0.26614994525442287,
      "grad_norm": 0.8878002166748047,
      "learning_rate": 8.607694276251155e-06,
      "loss": 0.7752,
      "step": 9237
    },
    {
      "epoch": 0.26617875871607216,
      "grad_norm": 0.8298854827880859,
      "learning_rate": 8.607371191848266e-06,
      "loss": 0.7401,
      "step": 9238
    },
    {
      "epoch": 0.26620757217772145,
      "grad_norm": 0.8388106822967529,
      "learning_rate": 8.607048076028782e-06,
      "loss": 0.7487,
      "step": 9239
    },
    {
      "epoch": 0.2662363856393707,
      "grad_norm": 0.870577871799469,
      "learning_rate": 8.606724928795512e-06,
      "loss": 0.7295,
      "step": 9240
    },
    {
      "epoch": 0.26626519910102,
      "grad_norm": 0.9019443392753601,
      "learning_rate": 8.606401750151274e-06,
      "loss": 0.771,
      "step": 9241
    },
    {
      "epoch": 0.26629401256266927,
      "grad_norm": 0.8517506718635559,
      "learning_rate": 8.606078540098884e-06,
      "loss": 0.7423,
      "step": 9242
    },
    {
      "epoch": 0.26632282602431856,
      "grad_norm": 0.9184760451316833,
      "learning_rate": 8.605755298641152e-06,
      "loss": 0.8066,
      "step": 9243
    },
    {
      "epoch": 0.26635163948596785,
      "grad_norm": 0.9263590574264526,
      "learning_rate": 8.605432025780898e-06,
      "loss": 0.7761,
      "step": 9244
    },
    {
      "epoch": 0.26638045294761714,
      "grad_norm": 0.9223425984382629,
      "learning_rate": 8.605108721520933e-06,
      "loss": 0.7455,
      "step": 9245
    },
    {
      "epoch": 0.26640926640926643,
      "grad_norm": 0.8286913633346558,
      "learning_rate": 8.604785385864077e-06,
      "loss": 0.7691,
      "step": 9246
    },
    {
      "epoch": 0.26643807987091567,
      "grad_norm": 0.8805283904075623,
      "learning_rate": 8.604462018813142e-06,
      "loss": 0.7523,
      "step": 9247
    },
    {
      "epoch": 0.26646689333256496,
      "grad_norm": 0.8790541887283325,
      "learning_rate": 8.604138620370946e-06,
      "loss": 0.7436,
      "step": 9248
    },
    {
      "epoch": 0.26649570679421425,
      "grad_norm": 0.8394067287445068,
      "learning_rate": 8.603815190540305e-06,
      "loss": 0.7815,
      "step": 9249
    },
    {
      "epoch": 0.26652452025586354,
      "grad_norm": 0.8480826616287231,
      "learning_rate": 8.603491729324038e-06,
      "loss": 0.7651,
      "step": 9250
    },
    {
      "epoch": 0.26655333371751283,
      "grad_norm": 0.9616087675094604,
      "learning_rate": 8.60316823672496e-06,
      "loss": 0.7972,
      "step": 9251
    },
    {
      "epoch": 0.2665821471791621,
      "grad_norm": 0.9309529662132263,
      "learning_rate": 8.602844712745887e-06,
      "loss": 0.7575,
      "step": 9252
    },
    {
      "epoch": 0.2666109606408114,
      "grad_norm": 0.9862611293792725,
      "learning_rate": 8.602521157389638e-06,
      "loss": 0.7494,
      "step": 9253
    },
    {
      "epoch": 0.26663977410246065,
      "grad_norm": 0.8831889033317566,
      "learning_rate": 8.602197570659033e-06,
      "loss": 0.7866,
      "step": 9254
    },
    {
      "epoch": 0.26666858756410994,
      "grad_norm": 0.8823934197425842,
      "learning_rate": 8.601873952556887e-06,
      "loss": 0.7547,
      "step": 9255
    },
    {
      "epoch": 0.26669740102575923,
      "grad_norm": 0.8770024180412292,
      "learning_rate": 8.60155030308602e-06,
      "loss": 0.7743,
      "step": 9256
    },
    {
      "epoch": 0.2667262144874085,
      "grad_norm": 0.948030948638916,
      "learning_rate": 8.601226622249247e-06,
      "loss": 0.7748,
      "step": 9257
    },
    {
      "epoch": 0.2667550279490578,
      "grad_norm": 0.8501965999603271,
      "learning_rate": 8.600902910049393e-06,
      "loss": 0.7708,
      "step": 9258
    },
    {
      "epoch": 0.2667838414107071,
      "grad_norm": 0.8583133220672607,
      "learning_rate": 8.600579166489274e-06,
      "loss": 0.7547,
      "step": 9259
    },
    {
      "epoch": 0.2668126548723564,
      "grad_norm": 0.8574173450469971,
      "learning_rate": 8.60025539157171e-06,
      "loss": 0.7481,
      "step": 9260
    },
    {
      "epoch": 0.2668414683340056,
      "grad_norm": 0.8701392412185669,
      "learning_rate": 8.59993158529952e-06,
      "loss": 0.753,
      "step": 9261
    },
    {
      "epoch": 0.2668702817956549,
      "grad_norm": 0.86662358045578,
      "learning_rate": 8.599607747675524e-06,
      "loss": 0.7574,
      "step": 9262
    },
    {
      "epoch": 0.2668990952573042,
      "grad_norm": 0.8487371206283569,
      "learning_rate": 8.599283878702543e-06,
      "loss": 0.7573,
      "step": 9263
    },
    {
      "epoch": 0.2669279087189535,
      "grad_norm": 0.8752646446228027,
      "learning_rate": 8.598959978383398e-06,
      "loss": 0.7619,
      "step": 9264
    },
    {
      "epoch": 0.2669567221806028,
      "grad_norm": 0.8688684105873108,
      "learning_rate": 8.598636046720909e-06,
      "loss": 0.7734,
      "step": 9265
    },
    {
      "epoch": 0.2669855356422521,
      "grad_norm": 0.8444758057594299,
      "learning_rate": 8.598312083717897e-06,
      "loss": 0.7869,
      "step": 9266
    },
    {
      "epoch": 0.2670143491039013,
      "grad_norm": 0.810836911201477,
      "learning_rate": 8.597988089377182e-06,
      "loss": 0.7696,
      "step": 9267
    },
    {
      "epoch": 0.2670431625655506,
      "grad_norm": 0.8502755165100098,
      "learning_rate": 8.597664063701591e-06,
      "loss": 0.7621,
      "step": 9268
    },
    {
      "epoch": 0.2670719760271999,
      "grad_norm": 0.8599169254302979,
      "learning_rate": 8.59734000669394e-06,
      "loss": 0.7796,
      "step": 9269
    },
    {
      "epoch": 0.2671007894888492,
      "grad_norm": 0.8267770409584045,
      "learning_rate": 8.597015918357055e-06,
      "loss": 0.7646,
      "step": 9270
    },
    {
      "epoch": 0.2671296029504985,
      "grad_norm": 0.8568524718284607,
      "learning_rate": 8.596691798693757e-06,
      "loss": 0.7611,
      "step": 9271
    },
    {
      "epoch": 0.26715841641214777,
      "grad_norm": 0.8649187684059143,
      "learning_rate": 8.596367647706866e-06,
      "loss": 0.7833,
      "step": 9272
    },
    {
      "epoch": 0.26718722987379706,
      "grad_norm": 0.8503270149230957,
      "learning_rate": 8.596043465399212e-06,
      "loss": 0.7319,
      "step": 9273
    },
    {
      "epoch": 0.2672160433354463,
      "grad_norm": 0.8414262533187866,
      "learning_rate": 8.595719251773611e-06,
      "loss": 0.7686,
      "step": 9274
    },
    {
      "epoch": 0.2672448567970956,
      "grad_norm": 0.8359240889549255,
      "learning_rate": 8.595395006832892e-06,
      "loss": 0.7513,
      "step": 9275
    },
    {
      "epoch": 0.2672736702587449,
      "grad_norm": 0.8607792258262634,
      "learning_rate": 8.595070730579874e-06,
      "loss": 0.7681,
      "step": 9276
    },
    {
      "epoch": 0.26730248372039417,
      "grad_norm": 0.8370307683944702,
      "learning_rate": 8.594746423017387e-06,
      "loss": 0.7571,
      "step": 9277
    },
    {
      "epoch": 0.26733129718204346,
      "grad_norm": 0.8746548295021057,
      "learning_rate": 8.594422084148248e-06,
      "loss": 0.7539,
      "step": 9278
    },
    {
      "epoch": 0.26736011064369275,
      "grad_norm": 0.8587143421173096,
      "learning_rate": 8.594097713975289e-06,
      "loss": 0.7647,
      "step": 9279
    },
    {
      "epoch": 0.26738892410534204,
      "grad_norm": 0.8770630955696106,
      "learning_rate": 8.59377331250133e-06,
      "loss": 0.7833,
      "step": 9280
    },
    {
      "epoch": 0.2674177375669913,
      "grad_norm": 0.8778852224349976,
      "learning_rate": 8.593448879729198e-06,
      "loss": 0.7894,
      "step": 9281
    },
    {
      "epoch": 0.26744655102864057,
      "grad_norm": 0.8507195115089417,
      "learning_rate": 8.59312441566172e-06,
      "loss": 0.7696,
      "step": 9282
    },
    {
      "epoch": 0.26747536449028986,
      "grad_norm": 0.8392956852912903,
      "learning_rate": 8.592799920301718e-06,
      "loss": 0.765,
      "step": 9283
    },
    {
      "epoch": 0.26750417795193915,
      "grad_norm": 0.8563588857650757,
      "learning_rate": 8.592475393652023e-06,
      "loss": 0.7605,
      "step": 9284
    },
    {
      "epoch": 0.26753299141358844,
      "grad_norm": 0.8621544241905212,
      "learning_rate": 8.592150835715456e-06,
      "loss": 0.7678,
      "step": 9285
    },
    {
      "epoch": 0.26756180487523773,
      "grad_norm": 0.8435004949569702,
      "learning_rate": 8.591826246494845e-06,
      "loss": 0.7864,
      "step": 9286
    },
    {
      "epoch": 0.267590618336887,
      "grad_norm": 0.8360397219657898,
      "learning_rate": 8.59150162599302e-06,
      "loss": 0.7703,
      "step": 9287
    },
    {
      "epoch": 0.26761943179853626,
      "grad_norm": 0.8397249579429626,
      "learning_rate": 8.591176974212807e-06,
      "loss": 0.7535,
      "step": 9288
    },
    {
      "epoch": 0.26764824526018555,
      "grad_norm": 0.8618574738502502,
      "learning_rate": 8.59085229115703e-06,
      "loss": 0.7724,
      "step": 9289
    },
    {
      "epoch": 0.26767705872183484,
      "grad_norm": 0.8363451957702637,
      "learning_rate": 8.59052757682852e-06,
      "loss": 0.7521,
      "step": 9290
    },
    {
      "epoch": 0.26770587218348413,
      "grad_norm": 0.8390358090400696,
      "learning_rate": 8.590202831230106e-06,
      "loss": 0.7491,
      "step": 9291
    },
    {
      "epoch": 0.2677346856451334,
      "grad_norm": 0.8227835893630981,
      "learning_rate": 8.589878054364611e-06,
      "loss": 0.7595,
      "step": 9292
    },
    {
      "epoch": 0.2677634991067827,
      "grad_norm": 0.8700658679008484,
      "learning_rate": 8.589553246234868e-06,
      "loss": 0.7613,
      "step": 9293
    },
    {
      "epoch": 0.26779231256843194,
      "grad_norm": 0.8363258242607117,
      "learning_rate": 8.589228406843706e-06,
      "loss": 0.7716,
      "step": 9294
    },
    {
      "epoch": 0.26782112603008124,
      "grad_norm": 0.8909713625907898,
      "learning_rate": 8.58890353619395e-06,
      "loss": 0.759,
      "step": 9295
    },
    {
      "epoch": 0.2678499394917305,
      "grad_norm": 0.8571624755859375,
      "learning_rate": 8.588578634288433e-06,
      "loss": 0.7518,
      "step": 9296
    },
    {
      "epoch": 0.2678787529533798,
      "grad_norm": 0.8486679196357727,
      "learning_rate": 8.588253701129982e-06,
      "loss": 0.7546,
      "step": 9297
    },
    {
      "epoch": 0.2679075664150291,
      "grad_norm": 0.8278305530548096,
      "learning_rate": 8.587928736721432e-06,
      "loss": 0.7647,
      "step": 9298
    },
    {
      "epoch": 0.2679363798766784,
      "grad_norm": 0.9042659401893616,
      "learning_rate": 8.587603741065606e-06,
      "loss": 0.7518,
      "step": 9299
    },
    {
      "epoch": 0.2679651933383277,
      "grad_norm": 0.8523558974266052,
      "learning_rate": 8.587278714165338e-06,
      "loss": 0.7469,
      "step": 9300
    },
    {
      "epoch": 0.2679940067999769,
      "grad_norm": 0.8691900372505188,
      "learning_rate": 8.586953656023459e-06,
      "loss": 0.7625,
      "step": 9301
    },
    {
      "epoch": 0.2680228202616262,
      "grad_norm": 0.858277440071106,
      "learning_rate": 8.586628566642799e-06,
      "loss": 0.7489,
      "step": 9302
    },
    {
      "epoch": 0.2680516337232755,
      "grad_norm": 0.8560664653778076,
      "learning_rate": 8.586303446026189e-06,
      "loss": 0.7442,
      "step": 9303
    },
    {
      "epoch": 0.2680804471849248,
      "grad_norm": 0.7999840974807739,
      "learning_rate": 8.585978294176462e-06,
      "loss": 0.7466,
      "step": 9304
    },
    {
      "epoch": 0.2681092606465741,
      "grad_norm": 0.8524704575538635,
      "learning_rate": 8.585653111096448e-06,
      "loss": 0.7603,
      "step": 9305
    },
    {
      "epoch": 0.2681380741082234,
      "grad_norm": 0.8279326558113098,
      "learning_rate": 8.58532789678898e-06,
      "loss": 0.7488,
      "step": 9306
    },
    {
      "epoch": 0.26816688756987267,
      "grad_norm": 0.8465743064880371,
      "learning_rate": 8.585002651256891e-06,
      "loss": 0.7443,
      "step": 9307
    },
    {
      "epoch": 0.2681957010315219,
      "grad_norm": 0.8854657411575317,
      "learning_rate": 8.584677374503012e-06,
      "loss": 0.7553,
      "step": 9308
    },
    {
      "epoch": 0.2682245144931712,
      "grad_norm": 0.8823369145393372,
      "learning_rate": 8.584352066530175e-06,
      "loss": 0.7731,
      "step": 9309
    },
    {
      "epoch": 0.2682533279548205,
      "grad_norm": 0.8683966994285583,
      "learning_rate": 8.584026727341217e-06,
      "loss": 0.728,
      "step": 9310
    },
    {
      "epoch": 0.2682821414164698,
      "grad_norm": 0.8583192825317383,
      "learning_rate": 8.583701356938967e-06,
      "loss": 0.7736,
      "step": 9311
    },
    {
      "epoch": 0.26831095487811907,
      "grad_norm": 0.8633987307548523,
      "learning_rate": 8.58337595532626e-06,
      "loss": 0.7701,
      "step": 9312
    },
    {
      "epoch": 0.26833976833976836,
      "grad_norm": 0.8098134398460388,
      "learning_rate": 8.583050522505934e-06,
      "loss": 0.7518,
      "step": 9313
    },
    {
      "epoch": 0.26836858180141765,
      "grad_norm": 0.8388540744781494,
      "learning_rate": 8.582725058480817e-06,
      "loss": 0.7566,
      "step": 9314
    },
    {
      "epoch": 0.2683973952630669,
      "grad_norm": 0.832461953163147,
      "learning_rate": 8.582399563253745e-06,
      "loss": 0.7775,
      "step": 9315
    },
    {
      "epoch": 0.2684262087247162,
      "grad_norm": 0.8230301141738892,
      "learning_rate": 8.582074036827557e-06,
      "loss": 0.7654,
      "step": 9316
    },
    {
      "epoch": 0.26845502218636547,
      "grad_norm": 0.8670015335083008,
      "learning_rate": 8.581748479205082e-06,
      "loss": 0.7428,
      "step": 9317
    },
    {
      "epoch": 0.26848383564801476,
      "grad_norm": 0.8591680526733398,
      "learning_rate": 8.58142289038916e-06,
      "loss": 0.7793,
      "step": 9318
    },
    {
      "epoch": 0.26851264910966405,
      "grad_norm": 0.8635407090187073,
      "learning_rate": 8.581097270382624e-06,
      "loss": 0.7789,
      "step": 9319
    },
    {
      "epoch": 0.26854146257131334,
      "grad_norm": 0.885004997253418,
      "learning_rate": 8.580771619188311e-06,
      "loss": 0.7727,
      "step": 9320
    },
    {
      "epoch": 0.2685702760329626,
      "grad_norm": 0.8646416068077087,
      "learning_rate": 8.580445936809056e-06,
      "loss": 0.7701,
      "step": 9321
    },
    {
      "epoch": 0.26859908949461186,
      "grad_norm": 0.8801450729370117,
      "learning_rate": 8.580120223247697e-06,
      "loss": 0.7376,
      "step": 9322
    },
    {
      "epoch": 0.26862790295626116,
      "grad_norm": 0.850837767124176,
      "learning_rate": 8.579794478507068e-06,
      "loss": 0.7693,
      "step": 9323
    },
    {
      "epoch": 0.26865671641791045,
      "grad_norm": 0.8705349564552307,
      "learning_rate": 8.57946870259001e-06,
      "loss": 0.7729,
      "step": 9324
    },
    {
      "epoch": 0.26868552987955974,
      "grad_norm": 0.8920247554779053,
      "learning_rate": 8.579142895499356e-06,
      "loss": 0.7856,
      "step": 9325
    },
    {
      "epoch": 0.268714343341209,
      "grad_norm": 0.8909059762954712,
      "learning_rate": 8.578817057237946e-06,
      "loss": 0.7489,
      "step": 9326
    },
    {
      "epoch": 0.2687431568028583,
      "grad_norm": 0.8403888940811157,
      "learning_rate": 8.578491187808617e-06,
      "loss": 0.7789,
      "step": 9327
    },
    {
      "epoch": 0.26877197026450755,
      "grad_norm": 0.8741852045059204,
      "learning_rate": 8.578165287214206e-06,
      "loss": 0.7567,
      "step": 9328
    },
    {
      "epoch": 0.26880078372615684,
      "grad_norm": 0.8807492852210999,
      "learning_rate": 8.577839355457554e-06,
      "loss": 0.7486,
      "step": 9329
    },
    {
      "epoch": 0.26882959718780614,
      "grad_norm": 0.8358745574951172,
      "learning_rate": 8.577513392541496e-06,
      "loss": 0.7648,
      "step": 9330
    },
    {
      "epoch": 0.2688584106494554,
      "grad_norm": 0.843850314617157,
      "learning_rate": 8.577187398468872e-06,
      "loss": 0.7528,
      "step": 9331
    },
    {
      "epoch": 0.2688872241111047,
      "grad_norm": 0.8648015856742859,
      "learning_rate": 8.576861373242525e-06,
      "loss": 0.7453,
      "step": 9332
    },
    {
      "epoch": 0.268916037572754,
      "grad_norm": 0.8396766781806946,
      "learning_rate": 8.576535316865287e-06,
      "loss": 0.7615,
      "step": 9333
    },
    {
      "epoch": 0.2689448510344033,
      "grad_norm": 0.8695725202560425,
      "learning_rate": 8.576209229340004e-06,
      "loss": 0.7686,
      "step": 9334
    },
    {
      "epoch": 0.26897366449605253,
      "grad_norm": 0.8670309782028198,
      "learning_rate": 8.575883110669512e-06,
      "loss": 0.764,
      "step": 9335
    },
    {
      "epoch": 0.2690024779577018,
      "grad_norm": 0.8759284615516663,
      "learning_rate": 8.575556960856655e-06,
      "loss": 0.7589,
      "step": 9336
    },
    {
      "epoch": 0.2690312914193511,
      "grad_norm": 0.8842720985412598,
      "learning_rate": 8.575230779904269e-06,
      "loss": 0.7761,
      "step": 9337
    },
    {
      "epoch": 0.2690601048810004,
      "grad_norm": 0.8733423352241516,
      "learning_rate": 8.574904567815196e-06,
      "loss": 0.7691,
      "step": 9338
    },
    {
      "epoch": 0.2690889183426497,
      "grad_norm": 0.8136606812477112,
      "learning_rate": 8.574578324592279e-06,
      "loss": 0.7593,
      "step": 9339
    },
    {
      "epoch": 0.269117731804299,
      "grad_norm": 0.8222967386245728,
      "learning_rate": 8.57425205023836e-06,
      "loss": 0.738,
      "step": 9340
    },
    {
      "epoch": 0.2691465452659483,
      "grad_norm": 0.8501191735267639,
      "learning_rate": 8.573925744756278e-06,
      "loss": 0.7583,
      "step": 9341
    },
    {
      "epoch": 0.2691753587275975,
      "grad_norm": 0.896643877029419,
      "learning_rate": 8.573599408148874e-06,
      "loss": 0.77,
      "step": 9342
    },
    {
      "epoch": 0.2692041721892468,
      "grad_norm": 0.8246228098869324,
      "learning_rate": 8.57327304041899e-06,
      "loss": 0.781,
      "step": 9343
    },
    {
      "epoch": 0.2692329856508961,
      "grad_norm": 0.8680832982063293,
      "learning_rate": 8.572946641569473e-06,
      "loss": 0.7514,
      "step": 9344
    },
    {
      "epoch": 0.2692617991125454,
      "grad_norm": 0.828753650188446,
      "learning_rate": 8.57262021160316e-06,
      "loss": 0.7431,
      "step": 9345
    },
    {
      "epoch": 0.2692906125741947,
      "grad_norm": 0.8840882182121277,
      "learning_rate": 8.572293750522898e-06,
      "loss": 0.7665,
      "step": 9346
    },
    {
      "epoch": 0.26931942603584397,
      "grad_norm": 0.8467057943344116,
      "learning_rate": 8.571967258331527e-06,
      "loss": 0.7624,
      "step": 9347
    },
    {
      "epoch": 0.2693482394974932,
      "grad_norm": 0.8358088731765747,
      "learning_rate": 8.571640735031892e-06,
      "loss": 0.7807,
      "step": 9348
    },
    {
      "epoch": 0.2693770529591425,
      "grad_norm": 0.8539056777954102,
      "learning_rate": 8.571314180626837e-06,
      "loss": 0.7709,
      "step": 9349
    },
    {
      "epoch": 0.2694058664207918,
      "grad_norm": 0.8617292046546936,
      "learning_rate": 8.570987595119207e-06,
      "loss": 0.7532,
      "step": 9350
    },
    {
      "epoch": 0.2694346798824411,
      "grad_norm": 0.8550117611885071,
      "learning_rate": 8.570660978511842e-06,
      "loss": 0.7717,
      "step": 9351
    },
    {
      "epoch": 0.26946349334409037,
      "grad_norm": 0.8103318810462952,
      "learning_rate": 8.57033433080759e-06,
      "loss": 0.7331,
      "step": 9352
    },
    {
      "epoch": 0.26949230680573966,
      "grad_norm": 0.817530632019043,
      "learning_rate": 8.570007652009295e-06,
      "loss": 0.7653,
      "step": 9353
    },
    {
      "epoch": 0.26952112026738895,
      "grad_norm": 0.8406291604042053,
      "learning_rate": 8.569680942119804e-06,
      "loss": 0.7537,
      "step": 9354
    },
    {
      "epoch": 0.2695499337290382,
      "grad_norm": 0.8468755483627319,
      "learning_rate": 8.569354201141958e-06,
      "loss": 0.7794,
      "step": 9355
    },
    {
      "epoch": 0.2695787471906875,
      "grad_norm": 0.8531625270843506,
      "learning_rate": 8.569027429078606e-06,
      "loss": 0.754,
      "step": 9356
    },
    {
      "epoch": 0.26960756065233676,
      "grad_norm": 0.8766564130783081,
      "learning_rate": 8.568700625932594e-06,
      "loss": 0.7693,
      "step": 9357
    },
    {
      "epoch": 0.26963637411398605,
      "grad_norm": 0.8734040260314941,
      "learning_rate": 8.568373791706765e-06,
      "loss": 0.7469,
      "step": 9358
    },
    {
      "epoch": 0.26966518757563535,
      "grad_norm": 0.871962308883667,
      "learning_rate": 8.568046926403967e-06,
      "loss": 0.7554,
      "step": 9359
    },
    {
      "epoch": 0.26969400103728464,
      "grad_norm": 0.8813119530677795,
      "learning_rate": 8.56772003002705e-06,
      "loss": 0.7666,
      "step": 9360
    },
    {
      "epoch": 0.2697228144989339,
      "grad_norm": 0.8893218040466309,
      "learning_rate": 8.567393102578854e-06,
      "loss": 0.7666,
      "step": 9361
    },
    {
      "epoch": 0.26975162796058316,
      "grad_norm": 0.8686156868934631,
      "learning_rate": 8.567066144062232e-06,
      "loss": 0.7478,
      "step": 9362
    },
    {
      "epoch": 0.26978044142223245,
      "grad_norm": 0.8421167731285095,
      "learning_rate": 8.56673915448003e-06,
      "loss": 0.7575,
      "step": 9363
    },
    {
      "epoch": 0.26980925488388174,
      "grad_norm": 0.8451358079910278,
      "learning_rate": 8.566412133835096e-06,
      "loss": 0.7474,
      "step": 9364
    },
    {
      "epoch": 0.26983806834553103,
      "grad_norm": 0.8950861692428589,
      "learning_rate": 8.566085082130276e-06,
      "loss": 0.7532,
      "step": 9365
    },
    {
      "epoch": 0.2698668818071803,
      "grad_norm": 0.9049935340881348,
      "learning_rate": 8.56575799936842e-06,
      "loss": 0.7779,
      "step": 9366
    },
    {
      "epoch": 0.2698956952688296,
      "grad_norm": 0.8830804824829102,
      "learning_rate": 8.565430885552377e-06,
      "loss": 0.7298,
      "step": 9367
    },
    {
      "epoch": 0.2699245087304789,
      "grad_norm": 0.8536524772644043,
      "learning_rate": 8.565103740684995e-06,
      "loss": 0.7589,
      "step": 9368
    },
    {
      "epoch": 0.26995332219212814,
      "grad_norm": 0.8782687187194824,
      "learning_rate": 8.564776564769122e-06,
      "loss": 0.7472,
      "step": 9369
    },
    {
      "epoch": 0.26998213565377743,
      "grad_norm": 0.8753001093864441,
      "learning_rate": 8.56444935780761e-06,
      "loss": 0.7644,
      "step": 9370
    },
    {
      "epoch": 0.2700109491154267,
      "grad_norm": 0.8681100010871887,
      "learning_rate": 8.564122119803307e-06,
      "loss": 0.7596,
      "step": 9371
    },
    {
      "epoch": 0.270039762577076,
      "grad_norm": 0.8284718990325928,
      "learning_rate": 8.563794850759062e-06,
      "loss": 0.7428,
      "step": 9372
    },
    {
      "epoch": 0.2700685760387253,
      "grad_norm": 0.8301769495010376,
      "learning_rate": 8.563467550677727e-06,
      "loss": 0.7678,
      "step": 9373
    },
    {
      "epoch": 0.2700973895003746,
      "grad_norm": 0.922504723072052,
      "learning_rate": 8.563140219562153e-06,
      "loss": 0.76,
      "step": 9374
    },
    {
      "epoch": 0.27012620296202383,
      "grad_norm": 0.866425096988678,
      "learning_rate": 8.562812857415188e-06,
      "loss": 0.7837,
      "step": 9375
    },
    {
      "epoch": 0.2701550164236731,
      "grad_norm": 0.9047057032585144,
      "learning_rate": 8.562485464239685e-06,
      "loss": 0.7836,
      "step": 9376
    },
    {
      "epoch": 0.2701838298853224,
      "grad_norm": 0.834027111530304,
      "learning_rate": 8.562158040038494e-06,
      "loss": 0.7544,
      "step": 9377
    },
    {
      "epoch": 0.2702126433469717,
      "grad_norm": 0.904848575592041,
      "learning_rate": 8.56183058481447e-06,
      "loss": 0.7852,
      "step": 9378
    },
    {
      "epoch": 0.270241456808621,
      "grad_norm": 0.8555902242660522,
      "learning_rate": 8.561503098570458e-06,
      "loss": 0.7864,
      "step": 9379
    },
    {
      "epoch": 0.2702702702702703,
      "grad_norm": 0.9097204804420471,
      "learning_rate": 8.561175581309318e-06,
      "loss": 0.7698,
      "step": 9380
    },
    {
      "epoch": 0.2702990837319196,
      "grad_norm": 0.8741019368171692,
      "learning_rate": 8.560848033033898e-06,
      "loss": 0.7718,
      "step": 9381
    },
    {
      "epoch": 0.2703278971935688,
      "grad_norm": 0.8373787999153137,
      "learning_rate": 8.56052045374705e-06,
      "loss": 0.7397,
      "step": 9382
    },
    {
      "epoch": 0.2703567106552181,
      "grad_norm": 0.8843507766723633,
      "learning_rate": 8.560192843451627e-06,
      "loss": 0.7498,
      "step": 9383
    },
    {
      "epoch": 0.2703855241168674,
      "grad_norm": 0.9149443507194519,
      "learning_rate": 8.559865202150484e-06,
      "loss": 0.7783,
      "step": 9384
    },
    {
      "epoch": 0.2704143375785167,
      "grad_norm": 0.8939208984375,
      "learning_rate": 8.559537529846476e-06,
      "loss": 0.78,
      "step": 9385
    },
    {
      "epoch": 0.270443151040166,
      "grad_norm": 0.8482691049575806,
      "learning_rate": 8.559209826542452e-06,
      "loss": 0.7591,
      "step": 9386
    },
    {
      "epoch": 0.27047196450181527,
      "grad_norm": 0.9415745735168457,
      "learning_rate": 8.558882092241267e-06,
      "loss": 0.7586,
      "step": 9387
    },
    {
      "epoch": 0.27050077796346456,
      "grad_norm": 0.8435539603233337,
      "learning_rate": 8.558554326945779e-06,
      "loss": 0.7366,
      "step": 9388
    },
    {
      "epoch": 0.2705295914251138,
      "grad_norm": 0.9286542534828186,
      "learning_rate": 8.55822653065884e-06,
      "loss": 0.7554,
      "step": 9389
    },
    {
      "epoch": 0.2705584048867631,
      "grad_norm": 0.8848845362663269,
      "learning_rate": 8.557898703383304e-06,
      "loss": 0.7509,
      "step": 9390
    },
    {
      "epoch": 0.2705872183484124,
      "grad_norm": 0.9168110489845276,
      "learning_rate": 8.557570845122028e-06,
      "loss": 0.7297,
      "step": 9391
    },
    {
      "epoch": 0.27061603181006166,
      "grad_norm": 0.8849375247955322,
      "learning_rate": 8.557242955877865e-06,
      "loss": 0.7678,
      "step": 9392
    },
    {
      "epoch": 0.27064484527171095,
      "grad_norm": 0.8975483775138855,
      "learning_rate": 8.556915035653672e-06,
      "loss": 0.767,
      "step": 9393
    },
    {
      "epoch": 0.27067365873336025,
      "grad_norm": 0.9357759356498718,
      "learning_rate": 8.556587084452305e-06,
      "loss": 0.7591,
      "step": 9394
    },
    {
      "epoch": 0.27070247219500954,
      "grad_norm": 0.8238155245780945,
      "learning_rate": 8.556259102276619e-06,
      "loss": 0.7499,
      "step": 9395
    },
    {
      "epoch": 0.27073128565665877,
      "grad_norm": 0.8094214797019958,
      "learning_rate": 8.555931089129472e-06,
      "loss": 0.7562,
      "step": 9396
    },
    {
      "epoch": 0.27076009911830806,
      "grad_norm": 0.9656396508216858,
      "learning_rate": 8.555603045013721e-06,
      "loss": 0.7964,
      "step": 9397
    },
    {
      "epoch": 0.27078891257995735,
      "grad_norm": 0.869867205619812,
      "learning_rate": 8.555274969932218e-06,
      "loss": 0.7378,
      "step": 9398
    },
    {
      "epoch": 0.27081772604160664,
      "grad_norm": 0.8724068403244019,
      "learning_rate": 8.554946863887828e-06,
      "loss": 0.7781,
      "step": 9399
    },
    {
      "epoch": 0.27084653950325593,
      "grad_norm": 0.8803803324699402,
      "learning_rate": 8.554618726883403e-06,
      "loss": 0.7516,
      "step": 9400
    },
    {
      "epoch": 0.2708753529649052,
      "grad_norm": 0.8585349321365356,
      "learning_rate": 8.554290558921802e-06,
      "loss": 0.7521,
      "step": 9401
    },
    {
      "epoch": 0.27090416642655446,
      "grad_norm": 0.8514919281005859,
      "learning_rate": 8.553962360005883e-06,
      "loss": 0.7484,
      "step": 9402
    },
    {
      "epoch": 0.27093297988820375,
      "grad_norm": 0.8390305638313293,
      "learning_rate": 8.553634130138505e-06,
      "loss": 0.7408,
      "step": 9403
    },
    {
      "epoch": 0.27096179334985304,
      "grad_norm": 0.8979265093803406,
      "learning_rate": 8.553305869322526e-06,
      "loss": 0.7754,
      "step": 9404
    },
    {
      "epoch": 0.27099060681150233,
      "grad_norm": 0.8927518129348755,
      "learning_rate": 8.552977577560804e-06,
      "loss": 0.7664,
      "step": 9405
    },
    {
      "epoch": 0.2710194202731516,
      "grad_norm": 0.8246616125106812,
      "learning_rate": 8.5526492548562e-06,
      "loss": 0.7683,
      "step": 9406
    },
    {
      "epoch": 0.2710482337348009,
      "grad_norm": 0.8500185608863831,
      "learning_rate": 8.552320901211572e-06,
      "loss": 0.7528,
      "step": 9407
    },
    {
      "epoch": 0.2710770471964502,
      "grad_norm": 0.8916294574737549,
      "learning_rate": 8.55199251662978e-06,
      "loss": 0.7576,
      "step": 9408
    },
    {
      "epoch": 0.27110586065809944,
      "grad_norm": 0.8735697865486145,
      "learning_rate": 8.551664101113684e-06,
      "loss": 0.7675,
      "step": 9409
    },
    {
      "epoch": 0.27113467411974873,
      "grad_norm": 0.8708329200744629,
      "learning_rate": 8.551335654666144e-06,
      "loss": 0.7639,
      "step": 9410
    },
    {
      "epoch": 0.271163487581398,
      "grad_norm": 0.877440333366394,
      "learning_rate": 8.55100717729002e-06,
      "loss": 0.7756,
      "step": 9411
    },
    {
      "epoch": 0.2711923010430473,
      "grad_norm": 0.8766985535621643,
      "learning_rate": 8.550678668988174e-06,
      "loss": 0.7721,
      "step": 9412
    },
    {
      "epoch": 0.2712211145046966,
      "grad_norm": 0.8551262617111206,
      "learning_rate": 8.550350129763466e-06,
      "loss": 0.7568,
      "step": 9413
    },
    {
      "epoch": 0.2712499279663459,
      "grad_norm": 0.8347599506378174,
      "learning_rate": 8.550021559618758e-06,
      "loss": 0.7491,
      "step": 9414
    },
    {
      "epoch": 0.2712787414279952,
      "grad_norm": 0.8327603340148926,
      "learning_rate": 8.54969295855691e-06,
      "loss": 0.7391,
      "step": 9415
    },
    {
      "epoch": 0.2713075548896444,
      "grad_norm": 0.8505902290344238,
      "learning_rate": 8.549364326580784e-06,
      "loss": 0.7591,
      "step": 9416
    },
    {
      "epoch": 0.2713363683512937,
      "grad_norm": 0.8717712759971619,
      "learning_rate": 8.549035663693243e-06,
      "loss": 0.7451,
      "step": 9417
    },
    {
      "epoch": 0.271365181812943,
      "grad_norm": 0.8787648677825928,
      "learning_rate": 8.548706969897151e-06,
      "loss": 0.7459,
      "step": 9418
    },
    {
      "epoch": 0.2713939952745923,
      "grad_norm": 0.8341912627220154,
      "learning_rate": 8.548378245195368e-06,
      "loss": 0.7545,
      "step": 9419
    },
    {
      "epoch": 0.2714228087362416,
      "grad_norm": 0.8927015662193298,
      "learning_rate": 8.548049489590757e-06,
      "loss": 0.7699,
      "step": 9420
    },
    {
      "epoch": 0.2714516221978909,
      "grad_norm": 0.8424375057220459,
      "learning_rate": 8.54772070308618e-06,
      "loss": 0.7327,
      "step": 9421
    },
    {
      "epoch": 0.27148043565954016,
      "grad_norm": 0.8316469192504883,
      "learning_rate": 8.547391885684505e-06,
      "loss": 0.7498,
      "step": 9422
    },
    {
      "epoch": 0.2715092491211894,
      "grad_norm": 0.8575525283813477,
      "learning_rate": 8.547063037388592e-06,
      "loss": 0.7534,
      "step": 9423
    },
    {
      "epoch": 0.2715380625828387,
      "grad_norm": 0.8330456018447876,
      "learning_rate": 8.546734158201304e-06,
      "loss": 0.7786,
      "step": 9424
    },
    {
      "epoch": 0.271566876044488,
      "grad_norm": 0.8532165884971619,
      "learning_rate": 8.546405248125509e-06,
      "loss": 0.7987,
      "step": 9425
    },
    {
      "epoch": 0.27159568950613727,
      "grad_norm": 0.8920151591300964,
      "learning_rate": 8.546076307164068e-06,
      "loss": 0.7671,
      "step": 9426
    },
    {
      "epoch": 0.27162450296778656,
      "grad_norm": 0.8335199356079102,
      "learning_rate": 8.545747335319848e-06,
      "loss": 0.7508,
      "step": 9427
    },
    {
      "epoch": 0.27165331642943585,
      "grad_norm": 0.830872654914856,
      "learning_rate": 8.545418332595714e-06,
      "loss": 0.7349,
      "step": 9428
    },
    {
      "epoch": 0.2716821298910851,
      "grad_norm": 0.8698587417602539,
      "learning_rate": 8.545089298994528e-06,
      "loss": 0.7515,
      "step": 9429
    },
    {
      "epoch": 0.2717109433527344,
      "grad_norm": 0.8822008371353149,
      "learning_rate": 8.544760234519161e-06,
      "loss": 0.7596,
      "step": 9430
    },
    {
      "epoch": 0.27173975681438367,
      "grad_norm": 0.8558059334754944,
      "learning_rate": 8.544431139172474e-06,
      "loss": 0.7223,
      "step": 9431
    },
    {
      "epoch": 0.27176857027603296,
      "grad_norm": 0.8500528931617737,
      "learning_rate": 8.544102012957335e-06,
      "loss": 0.7389,
      "step": 9432
    },
    {
      "epoch": 0.27179738373768225,
      "grad_norm": 0.8625898361206055,
      "learning_rate": 8.54377285587661e-06,
      "loss": 0.7574,
      "step": 9433
    },
    {
      "epoch": 0.27182619719933154,
      "grad_norm": 0.8312978148460388,
      "learning_rate": 8.543443667933166e-06,
      "loss": 0.7626,
      "step": 9434
    },
    {
      "epoch": 0.27185501066098083,
      "grad_norm": 0.8581592440605164,
      "learning_rate": 8.54311444912987e-06,
      "loss": 0.7758,
      "step": 9435
    },
    {
      "epoch": 0.27188382412263007,
      "grad_norm": 0.8440650105476379,
      "learning_rate": 8.542785199469589e-06,
      "loss": 0.7597,
      "step": 9436
    },
    {
      "epoch": 0.27191263758427936,
      "grad_norm": 0.8835232853889465,
      "learning_rate": 8.542455918955191e-06,
      "loss": 0.7826,
      "step": 9437
    },
    {
      "epoch": 0.27194145104592865,
      "grad_norm": 0.832935631275177,
      "learning_rate": 8.542126607589542e-06,
      "loss": 0.746,
      "step": 9438
    },
    {
      "epoch": 0.27197026450757794,
      "grad_norm": 0.8394131064414978,
      "learning_rate": 8.541797265375512e-06,
      "loss": 0.7506,
      "step": 9439
    },
    {
      "epoch": 0.27199907796922723,
      "grad_norm": 0.8485683798789978,
      "learning_rate": 8.541467892315968e-06,
      "loss": 0.778,
      "step": 9440
    },
    {
      "epoch": 0.2720278914308765,
      "grad_norm": 0.8081462383270264,
      "learning_rate": 8.541138488413777e-06,
      "loss": 0.7619,
      "step": 9441
    },
    {
      "epoch": 0.2720567048925258,
      "grad_norm": 0.8614296913146973,
      "learning_rate": 8.540809053671812e-06,
      "loss": 0.7731,
      "step": 9442
    },
    {
      "epoch": 0.27208551835417505,
      "grad_norm": 0.8625158667564392,
      "learning_rate": 8.54047958809294e-06,
      "loss": 0.7835,
      "step": 9443
    },
    {
      "epoch": 0.27211433181582434,
      "grad_norm": 0.8322508335113525,
      "learning_rate": 8.540150091680029e-06,
      "loss": 0.7569,
      "step": 9444
    },
    {
      "epoch": 0.27214314527747363,
      "grad_norm": 0.836929202079773,
      "learning_rate": 8.53982056443595e-06,
      "loss": 0.7888,
      "step": 9445
    },
    {
      "epoch": 0.2721719587391229,
      "grad_norm": 0.848296046257019,
      "learning_rate": 8.539491006363571e-06,
      "loss": 0.7898,
      "step": 9446
    },
    {
      "epoch": 0.2722007722007722,
      "grad_norm": 0.8725848197937012,
      "learning_rate": 8.539161417465766e-06,
      "loss": 0.7897,
      "step": 9447
    },
    {
      "epoch": 0.2722295856624215,
      "grad_norm": 0.8203243017196655,
      "learning_rate": 8.538831797745401e-06,
      "loss": 0.7375,
      "step": 9448
    },
    {
      "epoch": 0.2722583991240708,
      "grad_norm": 0.8786287903785706,
      "learning_rate": 8.538502147205348e-06,
      "loss": 0.7701,
      "step": 9449
    },
    {
      "epoch": 0.27228721258572003,
      "grad_norm": 0.8412380814552307,
      "learning_rate": 8.53817246584848e-06,
      "loss": 0.7623,
      "step": 9450
    },
    {
      "epoch": 0.2723160260473693,
      "grad_norm": 0.843164324760437,
      "learning_rate": 8.537842753677666e-06,
      "loss": 0.7591,
      "step": 9451
    },
    {
      "epoch": 0.2723448395090186,
      "grad_norm": 0.8107947111129761,
      "learning_rate": 8.53751301069578e-06,
      "loss": 0.7768,
      "step": 9452
    },
    {
      "epoch": 0.2723736529706679,
      "grad_norm": 0.8243147134780884,
      "learning_rate": 8.537183236905688e-06,
      "loss": 0.7306,
      "step": 9453
    },
    {
      "epoch": 0.2724024664323172,
      "grad_norm": 0.8509328961372375,
      "learning_rate": 8.536853432310269e-06,
      "loss": 0.7795,
      "step": 9454
    },
    {
      "epoch": 0.2724312798939665,
      "grad_norm": 0.8757891058921814,
      "learning_rate": 8.536523596912392e-06,
      "loss": 0.7769,
      "step": 9455
    },
    {
      "epoch": 0.2724600933556157,
      "grad_norm": 0.8582304120063782,
      "learning_rate": 8.53619373071493e-06,
      "loss": 0.7726,
      "step": 9456
    },
    {
      "epoch": 0.272488906817265,
      "grad_norm": 0.8472487330436707,
      "learning_rate": 8.535863833720753e-06,
      "loss": 0.795,
      "step": 9457
    },
    {
      "epoch": 0.2725177202789143,
      "grad_norm": 0.8458384871482849,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.7825,
      "step": 9458
    },
    {
      "epoch": 0.2725465337405636,
      "grad_norm": 0.8559544682502747,
      "learning_rate": 8.535203947353758e-06,
      "loss": 0.7482,
      "step": 9459
    },
    {
      "epoch": 0.2725753472022129,
      "grad_norm": 0.9363059401512146,
      "learning_rate": 8.534873957986684e-06,
      "loss": 0.743,
      "step": 9460
    },
    {
      "epoch": 0.27260416066386217,
      "grad_norm": 0.8706473112106323,
      "learning_rate": 8.534543937834392e-06,
      "loss": 0.7706,
      "step": 9461
    },
    {
      "epoch": 0.27263297412551146,
      "grad_norm": 0.9028312563896179,
      "learning_rate": 8.534213886899757e-06,
      "loss": 0.7589,
      "step": 9462
    },
    {
      "epoch": 0.2726617875871607,
      "grad_norm": 0.870698869228363,
      "learning_rate": 8.53388380518565e-06,
      "loss": 0.7371,
      "step": 9463
    },
    {
      "epoch": 0.27269060104881,
      "grad_norm": 0.8343819975852966,
      "learning_rate": 8.533553692694947e-06,
      "loss": 0.7709,
      "step": 9464
    },
    {
      "epoch": 0.2727194145104593,
      "grad_norm": 0.878890335559845,
      "learning_rate": 8.533223549430524e-06,
      "loss": 0.7578,
      "step": 9465
    },
    {
      "epoch": 0.27274822797210857,
      "grad_norm": 0.8390188813209534,
      "learning_rate": 8.532893375395256e-06,
      "loss": 0.7046,
      "step": 9466
    },
    {
      "epoch": 0.27277704143375786,
      "grad_norm": 0.8438717722892761,
      "learning_rate": 8.53256317059202e-06,
      "loss": 0.7443,
      "step": 9467
    },
    {
      "epoch": 0.27280585489540715,
      "grad_norm": 0.8432992696762085,
      "learning_rate": 8.532232935023689e-06,
      "loss": 0.7723,
      "step": 9468
    },
    {
      "epoch": 0.27283466835705644,
      "grad_norm": 0.885456919670105,
      "learning_rate": 8.53190266869314e-06,
      "loss": 0.7952,
      "step": 9469
    },
    {
      "epoch": 0.2728634818187057,
      "grad_norm": 0.837083101272583,
      "learning_rate": 8.531572371603249e-06,
      "loss": 0.7822,
      "step": 9470
    },
    {
      "epoch": 0.27289229528035497,
      "grad_norm": 0.8681614398956299,
      "learning_rate": 8.531242043756894e-06,
      "loss": 0.7506,
      "step": 9471
    },
    {
      "epoch": 0.27292110874200426,
      "grad_norm": 0.8647327423095703,
      "learning_rate": 8.53091168515695e-06,
      "loss": 0.7619,
      "step": 9472
    },
    {
      "epoch": 0.27294992220365355,
      "grad_norm": 0.8183277249336243,
      "learning_rate": 8.530581295806295e-06,
      "loss": 0.7599,
      "step": 9473
    },
    {
      "epoch": 0.27297873566530284,
      "grad_norm": 0.8360071182250977,
      "learning_rate": 8.530250875707806e-06,
      "loss": 0.7592,
      "step": 9474
    },
    {
      "epoch": 0.27300754912695213,
      "grad_norm": 0.8633584976196289,
      "learning_rate": 8.529920424864361e-06,
      "loss": 0.7564,
      "step": 9475
    },
    {
      "epoch": 0.2730363625886014,
      "grad_norm": 0.8317269682884216,
      "learning_rate": 8.529589943278838e-06,
      "loss": 0.7499,
      "step": 9476
    },
    {
      "epoch": 0.27306517605025066,
      "grad_norm": 0.8513309955596924,
      "learning_rate": 8.529259430954113e-06,
      "loss": 0.798,
      "step": 9477
    },
    {
      "epoch": 0.27309398951189995,
      "grad_norm": 0.8954856395721436,
      "learning_rate": 8.528928887893068e-06,
      "loss": 0.7513,
      "step": 9478
    },
    {
      "epoch": 0.27312280297354924,
      "grad_norm": 0.825687050819397,
      "learning_rate": 8.52859831409858e-06,
      "loss": 0.7418,
      "step": 9479
    },
    {
      "epoch": 0.27315161643519853,
      "grad_norm": 0.8350855112075806,
      "learning_rate": 8.528267709573528e-06,
      "loss": 0.7688,
      "step": 9480
    },
    {
      "epoch": 0.2731804298968478,
      "grad_norm": 0.8073598742485046,
      "learning_rate": 8.527937074320791e-06,
      "loss": 0.7466,
      "step": 9481
    },
    {
      "epoch": 0.2732092433584971,
      "grad_norm": 0.8311944007873535,
      "learning_rate": 8.52760640834325e-06,
      "loss": 0.7372,
      "step": 9482
    },
    {
      "epoch": 0.27323805682014635,
      "grad_norm": 0.8939474821090698,
      "learning_rate": 8.52727571164378e-06,
      "loss": 0.7627,
      "step": 9483
    },
    {
      "epoch": 0.27326687028179564,
      "grad_norm": 0.8103615045547485,
      "learning_rate": 8.52694498422527e-06,
      "loss": 0.7641,
      "step": 9484
    },
    {
      "epoch": 0.27329568374344493,
      "grad_norm": 0.8657459020614624,
      "learning_rate": 8.52661422609059e-06,
      "loss": 0.7574,
      "step": 9485
    },
    {
      "epoch": 0.2733244972050942,
      "grad_norm": 0.8789820075035095,
      "learning_rate": 8.526283437242627e-06,
      "loss": 0.7403,
      "step": 9486
    },
    {
      "epoch": 0.2733533106667435,
      "grad_norm": 0.8638284206390381,
      "learning_rate": 8.525952617684262e-06,
      "loss": 0.7386,
      "step": 9487
    },
    {
      "epoch": 0.2733821241283928,
      "grad_norm": 0.866565465927124,
      "learning_rate": 8.525621767418372e-06,
      "loss": 0.7339,
      "step": 9488
    },
    {
      "epoch": 0.2734109375900421,
      "grad_norm": 0.8795670866966248,
      "learning_rate": 8.52529088644784e-06,
      "loss": 0.7559,
      "step": 9489
    },
    {
      "epoch": 0.2734397510516913,
      "grad_norm": 0.889515221118927,
      "learning_rate": 8.524959974775551e-06,
      "loss": 0.798,
      "step": 9490
    },
    {
      "epoch": 0.2734685645133406,
      "grad_norm": 0.8274075388908386,
      "learning_rate": 8.524629032404383e-06,
      "loss": 0.7462,
      "step": 9491
    },
    {
      "epoch": 0.2734973779749899,
      "grad_norm": 0.8952105045318604,
      "learning_rate": 8.52429805933722e-06,
      "loss": 0.7614,
      "step": 9492
    },
    {
      "epoch": 0.2735261914366392,
      "grad_norm": 0.8482698202133179,
      "learning_rate": 8.523967055576943e-06,
      "loss": 0.7739,
      "step": 9493
    },
    {
      "epoch": 0.2735550048982885,
      "grad_norm": 0.8550708293914795,
      "learning_rate": 8.523636021126436e-06,
      "loss": 0.7767,
      "step": 9494
    },
    {
      "epoch": 0.2735838183599378,
      "grad_norm": 0.8808062672615051,
      "learning_rate": 8.523304955988582e-06,
      "loss": 0.7486,
      "step": 9495
    },
    {
      "epoch": 0.27361263182158707,
      "grad_norm": 0.8215646147727966,
      "learning_rate": 8.522973860166263e-06,
      "loss": 0.7574,
      "step": 9496
    },
    {
      "epoch": 0.2736414452832363,
      "grad_norm": 0.842922568321228,
      "learning_rate": 8.522642733662365e-06,
      "loss": 0.759,
      "step": 9497
    },
    {
      "epoch": 0.2736702587448856,
      "grad_norm": 0.9146668314933777,
      "learning_rate": 8.52231157647977e-06,
      "loss": 0.787,
      "step": 9498
    },
    {
      "epoch": 0.2736990722065349,
      "grad_norm": 0.8368055820465088,
      "learning_rate": 8.52198038862136e-06,
      "loss": 0.7437,
      "step": 9499
    },
    {
      "epoch": 0.2737278856681842,
      "grad_norm": 0.8353203535079956,
      "learning_rate": 8.521649170090022e-06,
      "loss": 0.7651,
      "step": 9500
    },
    {
      "epoch": 0.27375669912983347,
      "grad_norm": 0.842663586139679,
      "learning_rate": 8.521317920888642e-06,
      "loss": 0.7674,
      "step": 9501
    },
    {
      "epoch": 0.27378551259148276,
      "grad_norm": 0.8236181735992432,
      "learning_rate": 8.520986641020101e-06,
      "loss": 0.7472,
      "step": 9502
    },
    {
      "epoch": 0.273814326053132,
      "grad_norm": 0.8066144585609436,
      "learning_rate": 8.520655330487287e-06,
      "loss": 0.7408,
      "step": 9503
    },
    {
      "epoch": 0.2738431395147813,
      "grad_norm": 0.8759569525718689,
      "learning_rate": 8.520323989293084e-06,
      "loss": 0.7618,
      "step": 9504
    },
    {
      "epoch": 0.2738719529764306,
      "grad_norm": 0.8574915528297424,
      "learning_rate": 8.519992617440379e-06,
      "loss": 0.7613,
      "step": 9505
    },
    {
      "epoch": 0.27390076643807987,
      "grad_norm": 0.830014705657959,
      "learning_rate": 8.519661214932056e-06,
      "loss": 0.7658,
      "step": 9506
    },
    {
      "epoch": 0.27392957989972916,
      "grad_norm": 0.863457441329956,
      "learning_rate": 8.519329781771003e-06,
      "loss": 0.7728,
      "step": 9507
    },
    {
      "epoch": 0.27395839336137845,
      "grad_norm": 0.8418831825256348,
      "learning_rate": 8.518998317960106e-06,
      "loss": 0.7644,
      "step": 9508
    },
    {
      "epoch": 0.27398720682302774,
      "grad_norm": 0.8369187116622925,
      "learning_rate": 8.518666823502249e-06,
      "loss": 0.768,
      "step": 9509
    },
    {
      "epoch": 0.274016020284677,
      "grad_norm": 3.0533196926116943,
      "learning_rate": 8.518335298400324e-06,
      "loss": 0.7666,
      "step": 9510
    },
    {
      "epoch": 0.27404483374632627,
      "grad_norm": 0.8507221937179565,
      "learning_rate": 8.518003742657214e-06,
      "loss": 0.7618,
      "step": 9511
    },
    {
      "epoch": 0.27407364720797556,
      "grad_norm": 0.8450868129730225,
      "learning_rate": 8.517672156275808e-06,
      "loss": 0.7701,
      "step": 9512
    },
    {
      "epoch": 0.27410246066962485,
      "grad_norm": 0.8607785701751709,
      "learning_rate": 8.517340539258996e-06,
      "loss": 0.772,
      "step": 9513
    },
    {
      "epoch": 0.27413127413127414,
      "grad_norm": 0.8829210996627808,
      "learning_rate": 8.517008891609662e-06,
      "loss": 0.7568,
      "step": 9514
    },
    {
      "epoch": 0.27416008759292343,
      "grad_norm": 0.8293053507804871,
      "learning_rate": 8.516677213330697e-06,
      "loss": 0.7635,
      "step": 9515
    },
    {
      "epoch": 0.2741889010545727,
      "grad_norm": 0.858014702796936,
      "learning_rate": 8.516345504424988e-06,
      "loss": 0.7463,
      "step": 9516
    },
    {
      "epoch": 0.27421771451622196,
      "grad_norm": 0.8719833493232727,
      "learning_rate": 8.516013764895425e-06,
      "loss": 0.781,
      "step": 9517
    },
    {
      "epoch": 0.27424652797787125,
      "grad_norm": 0.8425535559654236,
      "learning_rate": 8.515681994744897e-06,
      "loss": 0.7512,
      "step": 9518
    },
    {
      "epoch": 0.27427534143952054,
      "grad_norm": 0.8684535026550293,
      "learning_rate": 8.515350193976294e-06,
      "loss": 0.7745,
      "step": 9519
    },
    {
      "epoch": 0.2743041549011698,
      "grad_norm": 0.8616682291030884,
      "learning_rate": 8.515018362592506e-06,
      "loss": 0.7897,
      "step": 9520
    },
    {
      "epoch": 0.2743329683628191,
      "grad_norm": 0.8417472839355469,
      "learning_rate": 8.51468650059642e-06,
      "loss": 0.7535,
      "step": 9521
    },
    {
      "epoch": 0.2743617818244684,
      "grad_norm": 0.8374044299125671,
      "learning_rate": 8.514354607990927e-06,
      "loss": 0.7762,
      "step": 9522
    },
    {
      "epoch": 0.2743905952861177,
      "grad_norm": 0.8915882706642151,
      "learning_rate": 8.51402268477892e-06,
      "loss": 0.771,
      "step": 9523
    },
    {
      "epoch": 0.27441940874776694,
      "grad_norm": 0.8752023577690125,
      "learning_rate": 8.513690730963288e-06,
      "loss": 0.7599,
      "step": 9524
    },
    {
      "epoch": 0.2744482222094162,
      "grad_norm": 0.8612401485443115,
      "learning_rate": 8.513358746546922e-06,
      "loss": 0.7271,
      "step": 9525
    },
    {
      "epoch": 0.2744770356710655,
      "grad_norm": 0.8844572305679321,
      "learning_rate": 8.513026731532713e-06,
      "loss": 0.7734,
      "step": 9526
    },
    {
      "epoch": 0.2745058491327148,
      "grad_norm": 0.8704320788383484,
      "learning_rate": 8.512694685923553e-06,
      "loss": 0.7722,
      "step": 9527
    },
    {
      "epoch": 0.2745346625943641,
      "grad_norm": 0.8319047093391418,
      "learning_rate": 8.512362609722333e-06,
      "loss": 0.7656,
      "step": 9528
    },
    {
      "epoch": 0.2745634760560134,
      "grad_norm": 0.8752738237380981,
      "learning_rate": 8.512030502931948e-06,
      "loss": 0.8066,
      "step": 9529
    },
    {
      "epoch": 0.2745922895176626,
      "grad_norm": 0.871161937713623,
      "learning_rate": 8.511698365555286e-06,
      "loss": 0.7816,
      "step": 9530
    },
    {
      "epoch": 0.2746211029793119,
      "grad_norm": 0.8505608439445496,
      "learning_rate": 8.511366197595241e-06,
      "loss": 0.7553,
      "step": 9531
    },
    {
      "epoch": 0.2746499164409612,
      "grad_norm": 0.8473109006881714,
      "learning_rate": 8.511033999054708e-06,
      "loss": 0.7663,
      "step": 9532
    },
    {
      "epoch": 0.2746787299026105,
      "grad_norm": 0.864298403263092,
      "learning_rate": 8.510701769936578e-06,
      "loss": 0.7642,
      "step": 9533
    },
    {
      "epoch": 0.2747075433642598,
      "grad_norm": 0.8360415697097778,
      "learning_rate": 8.510369510243746e-06,
      "loss": 0.7615,
      "step": 9534
    },
    {
      "epoch": 0.2747363568259091,
      "grad_norm": 0.8519731163978577,
      "learning_rate": 8.510037219979104e-06,
      "loss": 0.7634,
      "step": 9535
    },
    {
      "epoch": 0.27476517028755837,
      "grad_norm": 0.8464840650558472,
      "learning_rate": 8.509704899145548e-06,
      "loss": 0.7725,
      "step": 9536
    },
    {
      "epoch": 0.2747939837492076,
      "grad_norm": 0.8873175978660583,
      "learning_rate": 8.509372547745968e-06,
      "loss": 0.7655,
      "step": 9537
    },
    {
      "epoch": 0.2748227972108569,
      "grad_norm": 0.8504034280776978,
      "learning_rate": 8.509040165783262e-06,
      "loss": 0.7757,
      "step": 9538
    },
    {
      "epoch": 0.2748516106725062,
      "grad_norm": 0.8335299491882324,
      "learning_rate": 8.508707753260326e-06,
      "loss": 0.7485,
      "step": 9539
    },
    {
      "epoch": 0.2748804241341555,
      "grad_norm": 0.8368737697601318,
      "learning_rate": 8.50837531018005e-06,
      "loss": 0.7711,
      "step": 9540
    },
    {
      "epoch": 0.27490923759580477,
      "grad_norm": 0.8442294597625732,
      "learning_rate": 8.508042836545333e-06,
      "loss": 0.7564,
      "step": 9541
    },
    {
      "epoch": 0.27493805105745406,
      "grad_norm": 0.8514871597290039,
      "learning_rate": 8.507710332359071e-06,
      "loss": 0.7584,
      "step": 9542
    },
    {
      "epoch": 0.27496686451910335,
      "grad_norm": 0.8303901553153992,
      "learning_rate": 8.507377797624158e-06,
      "loss": 0.7532,
      "step": 9543
    },
    {
      "epoch": 0.2749956779807526,
      "grad_norm": 0.8592755198478699,
      "learning_rate": 8.50704523234349e-06,
      "loss": 0.7524,
      "step": 9544
    },
    {
      "epoch": 0.2750244914424019,
      "grad_norm": 0.9409433603286743,
      "learning_rate": 8.506712636519963e-06,
      "loss": 0.7704,
      "step": 9545
    },
    {
      "epoch": 0.27505330490405117,
      "grad_norm": 0.8720154762268066,
      "learning_rate": 8.506380010156476e-06,
      "loss": 0.781,
      "step": 9546
    },
    {
      "epoch": 0.27508211836570046,
      "grad_norm": 0.8203563690185547,
      "learning_rate": 8.506047353255924e-06,
      "loss": 0.7571,
      "step": 9547
    },
    {
      "epoch": 0.27511093182734975,
      "grad_norm": 0.9038714170455933,
      "learning_rate": 8.505714665821203e-06,
      "loss": 0.7775,
      "step": 9548
    },
    {
      "epoch": 0.27513974528899904,
      "grad_norm": 0.8520273566246033,
      "learning_rate": 8.505381947855214e-06,
      "loss": 0.7462,
      "step": 9549
    },
    {
      "epoch": 0.27516855875064833,
      "grad_norm": 0.8269723653793335,
      "learning_rate": 8.505049199360849e-06,
      "loss": 0.7372,
      "step": 9550
    },
    {
      "epoch": 0.27519737221229756,
      "grad_norm": 0.8387449383735657,
      "learning_rate": 8.504716420341012e-06,
      "loss": 0.7482,
      "step": 9551
    },
    {
      "epoch": 0.27522618567394685,
      "grad_norm": 0.910239040851593,
      "learning_rate": 8.504383610798597e-06,
      "loss": 0.7736,
      "step": 9552
    },
    {
      "epoch": 0.27525499913559615,
      "grad_norm": 0.8559702038764954,
      "learning_rate": 8.504050770736504e-06,
      "loss": 0.7365,
      "step": 9553
    },
    {
      "epoch": 0.27528381259724544,
      "grad_norm": 0.824100136756897,
      "learning_rate": 8.503717900157632e-06,
      "loss": 0.7575,
      "step": 9554
    },
    {
      "epoch": 0.2753126260588947,
      "grad_norm": 0.8949701189994812,
      "learning_rate": 8.50338499906488e-06,
      "loss": 0.7903,
      "step": 9555
    },
    {
      "epoch": 0.275341439520544,
      "grad_norm": 0.9067307114601135,
      "learning_rate": 8.503052067461147e-06,
      "loss": 0.7901,
      "step": 9556
    },
    {
      "epoch": 0.27537025298219325,
      "grad_norm": 0.8784382939338684,
      "learning_rate": 8.50271910534933e-06,
      "loss": 0.75,
      "step": 9557
    },
    {
      "epoch": 0.27539906644384254,
      "grad_norm": 0.9219068884849548,
      "learning_rate": 8.502386112732333e-06,
      "loss": 0.794,
      "step": 9558
    },
    {
      "epoch": 0.27542787990549183,
      "grad_norm": 0.8485240340232849,
      "learning_rate": 8.502053089613053e-06,
      "loss": 0.7707,
      "step": 9559
    },
    {
      "epoch": 0.2754566933671411,
      "grad_norm": 0.8140101432800293,
      "learning_rate": 8.501720035994392e-06,
      "loss": 0.7447,
      "step": 9560
    },
    {
      "epoch": 0.2754855068287904,
      "grad_norm": 0.8628678321838379,
      "learning_rate": 8.50138695187925e-06,
      "loss": 0.7895,
      "step": 9561
    },
    {
      "epoch": 0.2755143202904397,
      "grad_norm": 0.9618430137634277,
      "learning_rate": 8.501053837270527e-06,
      "loss": 0.7527,
      "step": 9562
    },
    {
      "epoch": 0.275543133752089,
      "grad_norm": 0.8918135762214661,
      "learning_rate": 8.500720692171124e-06,
      "loss": 0.7781,
      "step": 9563
    },
    {
      "epoch": 0.27557194721373823,
      "grad_norm": 0.8251803517341614,
      "learning_rate": 8.500387516583946e-06,
      "loss": 0.7601,
      "step": 9564
    },
    {
      "epoch": 0.2756007606753875,
      "grad_norm": 0.8902843594551086,
      "learning_rate": 8.50005431051189e-06,
      "loss": 0.779,
      "step": 9565
    },
    {
      "epoch": 0.2756295741370368,
      "grad_norm": 0.9330669045448303,
      "learning_rate": 8.49972107395786e-06,
      "loss": 0.7576,
      "step": 9566
    },
    {
      "epoch": 0.2756583875986861,
      "grad_norm": 0.8396316766738892,
      "learning_rate": 8.49938780692476e-06,
      "loss": 0.7611,
      "step": 9567
    },
    {
      "epoch": 0.2756872010603354,
      "grad_norm": 0.8720664978027344,
      "learning_rate": 8.499054509415488e-06,
      "loss": 0.7582,
      "step": 9568
    },
    {
      "epoch": 0.2757160145219847,
      "grad_norm": 0.8612166047096252,
      "learning_rate": 8.498721181432947e-06,
      "loss": 0.7437,
      "step": 9569
    },
    {
      "epoch": 0.275744827983634,
      "grad_norm": 0.8905137181282043,
      "learning_rate": 8.498387822980045e-06,
      "loss": 0.7649,
      "step": 9570
    },
    {
      "epoch": 0.2757736414452832,
      "grad_norm": 0.8669300675392151,
      "learning_rate": 8.49805443405968e-06,
      "loss": 0.7413,
      "step": 9571
    },
    {
      "epoch": 0.2758024549069325,
      "grad_norm": 0.8484676480293274,
      "learning_rate": 8.49772101467476e-06,
      "loss": 0.7399,
      "step": 9572
    },
    {
      "epoch": 0.2758312683685818,
      "grad_norm": 0.8338367938995361,
      "learning_rate": 8.497387564828187e-06,
      "loss": 0.7432,
      "step": 9573
    },
    {
      "epoch": 0.2758600818302311,
      "grad_norm": 0.9050333499908447,
      "learning_rate": 8.497054084522863e-06,
      "loss": 0.7662,
      "step": 9574
    },
    {
      "epoch": 0.2758888952918804,
      "grad_norm": 0.849844753742218,
      "learning_rate": 8.496720573761691e-06,
      "loss": 0.757,
      "step": 9575
    },
    {
      "epoch": 0.27591770875352967,
      "grad_norm": 0.8395954370498657,
      "learning_rate": 8.496387032547583e-06,
      "loss": 0.7557,
      "step": 9576
    },
    {
      "epoch": 0.27594652221517896,
      "grad_norm": 0.8354830741882324,
      "learning_rate": 8.496053460883437e-06,
      "loss": 0.7424,
      "step": 9577
    },
    {
      "epoch": 0.2759753356768282,
      "grad_norm": 0.8721692562103271,
      "learning_rate": 8.49571985877216e-06,
      "loss": 0.7379,
      "step": 9578
    },
    {
      "epoch": 0.2760041491384775,
      "grad_norm": 0.8676145672798157,
      "learning_rate": 8.495386226216657e-06,
      "loss": 0.784,
      "step": 9579
    },
    {
      "epoch": 0.2760329626001268,
      "grad_norm": 0.8486842513084412,
      "learning_rate": 8.495052563219835e-06,
      "loss": 0.7538,
      "step": 9580
    },
    {
      "epoch": 0.27606177606177607,
      "grad_norm": 0.8449128866195679,
      "learning_rate": 8.494718869784599e-06,
      "loss": 0.7476,
      "step": 9581
    },
    {
      "epoch": 0.27609058952342536,
      "grad_norm": 0.8136500716209412,
      "learning_rate": 8.494385145913853e-06,
      "loss": 0.7444,
      "step": 9582
    },
    {
      "epoch": 0.27611940298507465,
      "grad_norm": 0.8576998114585876,
      "learning_rate": 8.494051391610508e-06,
      "loss": 0.7384,
      "step": 9583
    },
    {
      "epoch": 0.2761482164467239,
      "grad_norm": 0.8404962420463562,
      "learning_rate": 8.493717606877468e-06,
      "loss": 0.7448,
      "step": 9584
    },
    {
      "epoch": 0.2761770299083732,
      "grad_norm": 0.8137091994285583,
      "learning_rate": 8.49338379171764e-06,
      "loss": 0.7485,
      "step": 9585
    },
    {
      "epoch": 0.27620584337002246,
      "grad_norm": 0.8226326107978821,
      "learning_rate": 8.49304994613393e-06,
      "loss": 0.7596,
      "step": 9586
    },
    {
      "epoch": 0.27623465683167175,
      "grad_norm": 0.826029360294342,
      "learning_rate": 8.492716070129247e-06,
      "loss": 0.7582,
      "step": 9587
    },
    {
      "epoch": 0.27626347029332105,
      "grad_norm": 0.8419234156608582,
      "learning_rate": 8.4923821637065e-06,
      "loss": 0.7842,
      "step": 9588
    },
    {
      "epoch": 0.27629228375497034,
      "grad_norm": 0.8686518669128418,
      "learning_rate": 8.492048226868594e-06,
      "loss": 0.7775,
      "step": 9589
    },
    {
      "epoch": 0.2763210972166196,
      "grad_norm": 0.8195846080780029,
      "learning_rate": 8.491714259618438e-06,
      "loss": 0.7622,
      "step": 9590
    },
    {
      "epoch": 0.27634991067826886,
      "grad_norm": 0.8294890522956848,
      "learning_rate": 8.491380261958942e-06,
      "loss": 0.7523,
      "step": 9591
    },
    {
      "epoch": 0.27637872413991815,
      "grad_norm": 0.870557963848114,
      "learning_rate": 8.491046233893015e-06,
      "loss": 0.77,
      "step": 9592
    },
    {
      "epoch": 0.27640753760156744,
      "grad_norm": 0.8358743190765381,
      "learning_rate": 8.490712175423564e-06,
      "loss": 0.7279,
      "step": 9593
    },
    {
      "epoch": 0.27643635106321673,
      "grad_norm": 0.8902546763420105,
      "learning_rate": 8.490378086553498e-06,
      "loss": 0.7773,
      "step": 9594
    },
    {
      "epoch": 0.276465164524866,
      "grad_norm": 0.8035788536071777,
      "learning_rate": 8.490043967285731e-06,
      "loss": 0.7492,
      "step": 9595
    },
    {
      "epoch": 0.2764939779865153,
      "grad_norm": 0.844886302947998,
      "learning_rate": 8.489709817623168e-06,
      "loss": 0.7664,
      "step": 9596
    },
    {
      "epoch": 0.2765227914481646,
      "grad_norm": 0.8107315897941589,
      "learning_rate": 8.489375637568722e-06,
      "loss": 0.7509,
      "step": 9597
    },
    {
      "epoch": 0.27655160490981384,
      "grad_norm": 0.80388343334198,
      "learning_rate": 8.489041427125302e-06,
      "loss": 0.767,
      "step": 9598
    },
    {
      "epoch": 0.27658041837146313,
      "grad_norm": 0.8310754299163818,
      "learning_rate": 8.48870718629582e-06,
      "loss": 0.7413,
      "step": 9599
    },
    {
      "epoch": 0.2766092318331124,
      "grad_norm": 0.849821150302887,
      "learning_rate": 8.488372915083183e-06,
      "loss": 0.7729,
      "step": 9600
    },
    {
      "epoch": 0.2766380452947617,
      "grad_norm": 0.8283755779266357,
      "learning_rate": 8.488038613490305e-06,
      "loss": 0.7554,
      "step": 9601
    },
    {
      "epoch": 0.276666858756411,
      "grad_norm": 0.8439601063728333,
      "learning_rate": 8.4877042815201e-06,
      "loss": 0.7488,
      "step": 9602
    },
    {
      "epoch": 0.2766956722180603,
      "grad_norm": 0.8474754691123962,
      "learning_rate": 8.487369919175477e-06,
      "loss": 0.7348,
      "step": 9603
    },
    {
      "epoch": 0.2767244856797096,
      "grad_norm": 0.843235969543457,
      "learning_rate": 8.487035526459347e-06,
      "loss": 0.7636,
      "step": 9604
    },
    {
      "epoch": 0.2767532991413588,
      "grad_norm": 0.8678752183914185,
      "learning_rate": 8.486701103374624e-06,
      "loss": 0.7549,
      "step": 9605
    },
    {
      "epoch": 0.2767821126030081,
      "grad_norm": 0.8528637290000916,
      "learning_rate": 8.486366649924217e-06,
      "loss": 0.7697,
      "step": 9606
    },
    {
      "epoch": 0.2768109260646574,
      "grad_norm": 0.8468543887138367,
      "learning_rate": 8.486032166111046e-06,
      "loss": 0.7525,
      "step": 9607
    },
    {
      "epoch": 0.2768397395263067,
      "grad_norm": 0.8363198637962341,
      "learning_rate": 8.485697651938018e-06,
      "loss": 0.7462,
      "step": 9608
    },
    {
      "epoch": 0.276868552987956,
      "grad_norm": 0.8850324749946594,
      "learning_rate": 8.485363107408047e-06,
      "loss": 0.7698,
      "step": 9609
    },
    {
      "epoch": 0.2768973664496053,
      "grad_norm": 0.8668087720870972,
      "learning_rate": 8.485028532524046e-06,
      "loss": 0.7615,
      "step": 9610
    },
    {
      "epoch": 0.2769261799112545,
      "grad_norm": 0.8791693449020386,
      "learning_rate": 8.484693927288934e-06,
      "loss": 0.7711,
      "step": 9611
    },
    {
      "epoch": 0.2769549933729038,
      "grad_norm": 0.8921028971672058,
      "learning_rate": 8.484359291705619e-06,
      "loss": 0.7846,
      "step": 9612
    },
    {
      "epoch": 0.2769838068345531,
      "grad_norm": 0.8446207642555237,
      "learning_rate": 8.484024625777017e-06,
      "loss": 0.7611,
      "step": 9613
    },
    {
      "epoch": 0.2770126202962024,
      "grad_norm": 0.8617554903030396,
      "learning_rate": 8.483689929506045e-06,
      "loss": 0.7814,
      "step": 9614
    },
    {
      "epoch": 0.2770414337578517,
      "grad_norm": 0.8482651710510254,
      "learning_rate": 8.483355202895615e-06,
      "loss": 0.7463,
      "step": 9615
    },
    {
      "epoch": 0.27707024721950096,
      "grad_norm": 0.8207958340644836,
      "learning_rate": 8.483020445948645e-06,
      "loss": 0.7447,
      "step": 9616
    },
    {
      "epoch": 0.27709906068115026,
      "grad_norm": 0.8132447600364685,
      "learning_rate": 8.482685658668046e-06,
      "loss": 0.7609,
      "step": 9617
    },
    {
      "epoch": 0.2771278741427995,
      "grad_norm": 0.906756579875946,
      "learning_rate": 8.482350841056739e-06,
      "loss": 0.759,
      "step": 9618
    },
    {
      "epoch": 0.2771566876044488,
      "grad_norm": 0.8282272815704346,
      "learning_rate": 8.482015993117635e-06,
      "loss": 0.7483,
      "step": 9619
    },
    {
      "epoch": 0.2771855010660981,
      "grad_norm": 0.8490203022956848,
      "learning_rate": 8.481681114853653e-06,
      "loss": 0.7641,
      "step": 9620
    },
    {
      "epoch": 0.27721431452774736,
      "grad_norm": 0.8376784324645996,
      "learning_rate": 8.481346206267709e-06,
      "loss": 0.7284,
      "step": 9621
    },
    {
      "epoch": 0.27724312798939665,
      "grad_norm": 0.8907074332237244,
      "learning_rate": 8.48101126736272e-06,
      "loss": 0.7439,
      "step": 9622
    },
    {
      "epoch": 0.27727194145104594,
      "grad_norm": 0.9309623837471008,
      "learning_rate": 8.480676298141602e-06,
      "loss": 0.7688,
      "step": 9623
    },
    {
      "epoch": 0.27730075491269524,
      "grad_norm": 0.8958252668380737,
      "learning_rate": 8.480341298607272e-06,
      "loss": 0.7574,
      "step": 9624
    },
    {
      "epoch": 0.27732956837434447,
      "grad_norm": 0.8186765313148499,
      "learning_rate": 8.480006268762651e-06,
      "loss": 0.764,
      "step": 9625
    },
    {
      "epoch": 0.27735838183599376,
      "grad_norm": 0.8432775735855103,
      "learning_rate": 8.479671208610654e-06,
      "loss": 0.736,
      "step": 9626
    },
    {
      "epoch": 0.27738719529764305,
      "grad_norm": 0.878073513507843,
      "learning_rate": 8.479336118154198e-06,
      "loss": 0.7535,
      "step": 9627
    },
    {
      "epoch": 0.27741600875929234,
      "grad_norm": 0.903453528881073,
      "learning_rate": 8.479000997396201e-06,
      "loss": 0.7689,
      "step": 9628
    },
    {
      "epoch": 0.27744482222094163,
      "grad_norm": 0.88802570104599,
      "learning_rate": 8.478665846339585e-06,
      "loss": 0.7473,
      "step": 9629
    },
    {
      "epoch": 0.2774736356825909,
      "grad_norm": 0.7987120151519775,
      "learning_rate": 8.478330664987266e-06,
      "loss": 0.7516,
      "step": 9630
    },
    {
      "epoch": 0.2775024491442402,
      "grad_norm": 0.8823845982551575,
      "learning_rate": 8.477995453342164e-06,
      "loss": 0.7799,
      "step": 9631
    },
    {
      "epoch": 0.27753126260588945,
      "grad_norm": 0.8541367650032043,
      "learning_rate": 8.4776602114072e-06,
      "loss": 0.7337,
      "step": 9632
    },
    {
      "epoch": 0.27756007606753874,
      "grad_norm": 0.8313595056533813,
      "learning_rate": 8.47732493918529e-06,
      "loss": 0.7643,
      "step": 9633
    },
    {
      "epoch": 0.27758888952918803,
      "grad_norm": 0.8173457384109497,
      "learning_rate": 8.476989636679356e-06,
      "loss": 0.7864,
      "step": 9634
    },
    {
      "epoch": 0.2776177029908373,
      "grad_norm": 0.8309022784233093,
      "learning_rate": 8.476654303892318e-06,
      "loss": 0.7763,
      "step": 9635
    },
    {
      "epoch": 0.2776465164524866,
      "grad_norm": 0.9289020895957947,
      "learning_rate": 8.476318940827097e-06,
      "loss": 0.729,
      "step": 9636
    },
    {
      "epoch": 0.2776753299141359,
      "grad_norm": 0.9019743204116821,
      "learning_rate": 8.475983547486613e-06,
      "loss": 0.7906,
      "step": 9637
    },
    {
      "epoch": 0.27770414337578514,
      "grad_norm": 0.8505951762199402,
      "learning_rate": 8.475648123873788e-06,
      "loss": 0.7555,
      "step": 9638
    },
    {
      "epoch": 0.27773295683743443,
      "grad_norm": 0.8392530083656311,
      "learning_rate": 8.47531266999154e-06,
      "loss": 0.7463,
      "step": 9639
    },
    {
      "epoch": 0.2777617702990837,
      "grad_norm": 0.8449445962905884,
      "learning_rate": 8.474977185842794e-06,
      "loss": 0.7776,
      "step": 9640
    },
    {
      "epoch": 0.277790583760733,
      "grad_norm": 0.8874892592430115,
      "learning_rate": 8.47464167143047e-06,
      "loss": 0.7642,
      "step": 9641
    },
    {
      "epoch": 0.2778193972223823,
      "grad_norm": 0.8625447154045105,
      "learning_rate": 8.47430612675749e-06,
      "loss": 0.7992,
      "step": 9642
    },
    {
      "epoch": 0.2778482106840316,
      "grad_norm": 0.8432364463806152,
      "learning_rate": 8.473970551826778e-06,
      "loss": 0.7659,
      "step": 9643
    },
    {
      "epoch": 0.2778770241456809,
      "grad_norm": 0.8112524151802063,
      "learning_rate": 8.473634946641253e-06,
      "loss": 0.7906,
      "step": 9644
    },
    {
      "epoch": 0.2779058376073301,
      "grad_norm": 0.8542381525039673,
      "learning_rate": 8.473299311203841e-06,
      "loss": 0.7631,
      "step": 9645
    },
    {
      "epoch": 0.2779346510689794,
      "grad_norm": 0.8507365584373474,
      "learning_rate": 8.472963645517467e-06,
      "loss": 0.7437,
      "step": 9646
    },
    {
      "epoch": 0.2779634645306287,
      "grad_norm": 0.8523592352867126,
      "learning_rate": 8.472627949585047e-06,
      "loss": 0.7526,
      "step": 9647
    },
    {
      "epoch": 0.277992277992278,
      "grad_norm": 0.8594306707382202,
      "learning_rate": 8.47229222340951e-06,
      "loss": 0.7591,
      "step": 9648
    },
    {
      "epoch": 0.2780210914539273,
      "grad_norm": 0.8862290382385254,
      "learning_rate": 8.471956466993781e-06,
      "loss": 0.7451,
      "step": 9649
    },
    {
      "epoch": 0.2780499049155766,
      "grad_norm": 0.8344051241874695,
      "learning_rate": 8.47162068034078e-06,
      "loss": 0.7326,
      "step": 9650
    },
    {
      "epoch": 0.27807871837722586,
      "grad_norm": 0.8337432146072388,
      "learning_rate": 8.471284863453434e-06,
      "loss": 0.76,
      "step": 9651
    },
    {
      "epoch": 0.2781075318388751,
      "grad_norm": 0.8537147045135498,
      "learning_rate": 8.470949016334666e-06,
      "loss": 0.7244,
      "step": 9652
    },
    {
      "epoch": 0.2781363453005244,
      "grad_norm": 0.8228144645690918,
      "learning_rate": 8.470613138987402e-06,
      "loss": 0.7789,
      "step": 9653
    },
    {
      "epoch": 0.2781651587621737,
      "grad_norm": 0.8493305444717407,
      "learning_rate": 8.470277231414568e-06,
      "loss": 0.7661,
      "step": 9654
    },
    {
      "epoch": 0.27819397222382297,
      "grad_norm": 0.8679812550544739,
      "learning_rate": 8.469941293619086e-06,
      "loss": 0.7774,
      "step": 9655
    },
    {
      "epoch": 0.27822278568547226,
      "grad_norm": 0.8437119126319885,
      "learning_rate": 8.469605325603886e-06,
      "loss": 0.7523,
      "step": 9656
    },
    {
      "epoch": 0.27825159914712155,
      "grad_norm": 0.8199747204780579,
      "learning_rate": 8.469269327371892e-06,
      "loss": 0.7435,
      "step": 9657
    },
    {
      "epoch": 0.27828041260877084,
      "grad_norm": 0.8598024249076843,
      "learning_rate": 8.468933298926028e-06,
      "loss": 0.7596,
      "step": 9658
    },
    {
      "epoch": 0.2783092260704201,
      "grad_norm": 0.8289695382118225,
      "learning_rate": 8.468597240269226e-06,
      "loss": 0.7413,
      "step": 9659
    },
    {
      "epoch": 0.27833803953206937,
      "grad_norm": 0.8441044092178345,
      "learning_rate": 8.468261151404407e-06,
      "loss": 0.7817,
      "step": 9660
    },
    {
      "epoch": 0.27836685299371866,
      "grad_norm": 0.7999101877212524,
      "learning_rate": 8.4679250323345e-06,
      "loss": 0.7539,
      "step": 9661
    },
    {
      "epoch": 0.27839566645536795,
      "grad_norm": 0.849098801612854,
      "learning_rate": 8.467588883062434e-06,
      "loss": 0.7529,
      "step": 9662
    },
    {
      "epoch": 0.27842447991701724,
      "grad_norm": 0.9048650860786438,
      "learning_rate": 8.467252703591134e-06,
      "loss": 0.7731,
      "step": 9663
    },
    {
      "epoch": 0.27845329337866653,
      "grad_norm": 0.8451233506202698,
      "learning_rate": 8.46691649392353e-06,
      "loss": 0.7454,
      "step": 9664
    },
    {
      "epoch": 0.27848210684031577,
      "grad_norm": 0.8405987620353699,
      "learning_rate": 8.466580254062548e-06,
      "loss": 0.7559,
      "step": 9665
    },
    {
      "epoch": 0.27851092030196506,
      "grad_norm": 0.8884863257408142,
      "learning_rate": 8.466243984011117e-06,
      "loss": 0.766,
      "step": 9666
    },
    {
      "epoch": 0.27853973376361435,
      "grad_norm": 0.879807710647583,
      "learning_rate": 8.465907683772167e-06,
      "loss": 0.7593,
      "step": 9667
    },
    {
      "epoch": 0.27856854722526364,
      "grad_norm": 0.8945529460906982,
      "learning_rate": 8.465571353348627e-06,
      "loss": 0.7438,
      "step": 9668
    },
    {
      "epoch": 0.27859736068691293,
      "grad_norm": 0.8397984504699707,
      "learning_rate": 8.465234992743424e-06,
      "loss": 0.7449,
      "step": 9669
    },
    {
      "epoch": 0.2786261741485622,
      "grad_norm": 0.8334026336669922,
      "learning_rate": 8.464898601959485e-06,
      "loss": 0.7609,
      "step": 9670
    },
    {
      "epoch": 0.2786549876102115,
      "grad_norm": 0.8850739598274231,
      "learning_rate": 8.464562180999746e-06,
      "loss": 0.7656,
      "step": 9671
    },
    {
      "epoch": 0.27868380107186075,
      "grad_norm": 0.8908992409706116,
      "learning_rate": 8.464225729867133e-06,
      "loss": 0.748,
      "step": 9672
    },
    {
      "epoch": 0.27871261453351004,
      "grad_norm": 0.8626006245613098,
      "learning_rate": 8.463889248564577e-06,
      "loss": 0.7782,
      "step": 9673
    },
    {
      "epoch": 0.27874142799515933,
      "grad_norm": 0.860435962677002,
      "learning_rate": 8.463552737095008e-06,
      "loss": 0.7592,
      "step": 9674
    },
    {
      "epoch": 0.2787702414568086,
      "grad_norm": 0.8853819370269775,
      "learning_rate": 8.463216195461356e-06,
      "loss": 0.7643,
      "step": 9675
    },
    {
      "epoch": 0.2787990549184579,
      "grad_norm": 0.8428152799606323,
      "learning_rate": 8.462879623666555e-06,
      "loss": 0.7581,
      "step": 9676
    },
    {
      "epoch": 0.2788278683801072,
      "grad_norm": 0.8442355394363403,
      "learning_rate": 8.462543021713532e-06,
      "loss": 0.7536,
      "step": 9677
    },
    {
      "epoch": 0.2788566818417565,
      "grad_norm": 0.8902413845062256,
      "learning_rate": 8.462206389605221e-06,
      "loss": 0.7691,
      "step": 9678
    },
    {
      "epoch": 0.27888549530340573,
      "grad_norm": 0.8524680733680725,
      "learning_rate": 8.461869727344555e-06,
      "loss": 0.7511,
      "step": 9679
    },
    {
      "epoch": 0.278914308765055,
      "grad_norm": 0.8396785259246826,
      "learning_rate": 8.461533034934463e-06,
      "loss": 0.7373,
      "step": 9680
    },
    {
      "epoch": 0.2789431222267043,
      "grad_norm": 1.875181794166565,
      "learning_rate": 8.461196312377879e-06,
      "loss": 0.7673,
      "step": 9681
    },
    {
      "epoch": 0.2789719356883536,
      "grad_norm": 0.8391575813293457,
      "learning_rate": 8.460859559677735e-06,
      "loss": 0.7603,
      "step": 9682
    },
    {
      "epoch": 0.2790007491500029,
      "grad_norm": 0.8488449454307556,
      "learning_rate": 8.460522776836962e-06,
      "loss": 0.7458,
      "step": 9683
    },
    {
      "epoch": 0.2790295626116522,
      "grad_norm": 0.8193902969360352,
      "learning_rate": 8.460185963858497e-06,
      "loss": 0.7631,
      "step": 9684
    },
    {
      "epoch": 0.2790583760733015,
      "grad_norm": 0.8622328639030457,
      "learning_rate": 8.459849120745272e-06,
      "loss": 0.7634,
      "step": 9685
    },
    {
      "epoch": 0.2790871895349507,
      "grad_norm": 0.8443180918693542,
      "learning_rate": 8.459512247500217e-06,
      "loss": 0.7753,
      "step": 9686
    },
    {
      "epoch": 0.2791160029966,
      "grad_norm": 0.8148727416992188,
      "learning_rate": 8.45917534412627e-06,
      "loss": 0.7473,
      "step": 9687
    },
    {
      "epoch": 0.2791448164582493,
      "grad_norm": 0.8157315850257874,
      "learning_rate": 8.458838410626364e-06,
      "loss": 0.7453,
      "step": 9688
    },
    {
      "epoch": 0.2791736299198986,
      "grad_norm": 0.8693846464157104,
      "learning_rate": 8.458501447003432e-06,
      "loss": 0.7561,
      "step": 9689
    },
    {
      "epoch": 0.27920244338154787,
      "grad_norm": 0.8770554661750793,
      "learning_rate": 8.45816445326041e-06,
      "loss": 0.7757,
      "step": 9690
    },
    {
      "epoch": 0.27923125684319716,
      "grad_norm": 0.8266143798828125,
      "learning_rate": 8.457827429400233e-06,
      "loss": 0.7491,
      "step": 9691
    },
    {
      "epoch": 0.2792600703048464,
      "grad_norm": 0.8482769131660461,
      "learning_rate": 8.457490375425835e-06,
      "loss": 0.755,
      "step": 9692
    },
    {
      "epoch": 0.2792888837664957,
      "grad_norm": 0.8759729862213135,
      "learning_rate": 8.457153291340153e-06,
      "loss": 0.7295,
      "step": 9693
    },
    {
      "epoch": 0.279317697228145,
      "grad_norm": 0.8761910200119019,
      "learning_rate": 8.456816177146121e-06,
      "loss": 0.7734,
      "step": 9694
    },
    {
      "epoch": 0.27934651068979427,
      "grad_norm": 0.827558696269989,
      "learning_rate": 8.456479032846674e-06,
      "loss": 0.743,
      "step": 9695
    },
    {
      "epoch": 0.27937532415144356,
      "grad_norm": 0.8462837338447571,
      "learning_rate": 8.456141858444754e-06,
      "loss": 0.7581,
      "step": 9696
    },
    {
      "epoch": 0.27940413761309285,
      "grad_norm": 0.9102317690849304,
      "learning_rate": 8.45580465394329e-06,
      "loss": 0.7554,
      "step": 9697
    },
    {
      "epoch": 0.27943295107474214,
      "grad_norm": 0.8484868407249451,
      "learning_rate": 8.455467419345223e-06,
      "loss": 0.7598,
      "step": 9698
    },
    {
      "epoch": 0.2794617645363914,
      "grad_norm": 0.8681208491325378,
      "learning_rate": 8.455130154653488e-06,
      "loss": 0.7654,
      "step": 9699
    },
    {
      "epoch": 0.27949057799804067,
      "grad_norm": 0.8469645977020264,
      "learning_rate": 8.454792859871025e-06,
      "loss": 0.7397,
      "step": 9700
    },
    {
      "epoch": 0.27951939145968996,
      "grad_norm": 0.8553259372711182,
      "learning_rate": 8.454455535000767e-06,
      "loss": 0.7611,
      "step": 9701
    },
    {
      "epoch": 0.27954820492133925,
      "grad_norm": 0.8053003549575806,
      "learning_rate": 8.454118180045657e-06,
      "loss": 0.7357,
      "step": 9702
    },
    {
      "epoch": 0.27957701838298854,
      "grad_norm": 0.8873389959335327,
      "learning_rate": 8.45378079500863e-06,
      "loss": 0.7589,
      "step": 9703
    },
    {
      "epoch": 0.27960583184463783,
      "grad_norm": 0.8653327822685242,
      "learning_rate": 8.453443379892626e-06,
      "loss": 0.7679,
      "step": 9704
    },
    {
      "epoch": 0.2796346453062871,
      "grad_norm": 0.8257697820663452,
      "learning_rate": 8.45310593470058e-06,
      "loss": 0.7494,
      "step": 9705
    },
    {
      "epoch": 0.27966345876793636,
      "grad_norm": 0.7849225401878357,
      "learning_rate": 8.452768459435435e-06,
      "loss": 0.7486,
      "step": 9706
    },
    {
      "epoch": 0.27969227222958565,
      "grad_norm": 0.8586663603782654,
      "learning_rate": 8.452430954100125e-06,
      "loss": 0.7249,
      "step": 9707
    },
    {
      "epoch": 0.27972108569123494,
      "grad_norm": 0.8336762189865112,
      "learning_rate": 8.452093418697596e-06,
      "loss": 0.7543,
      "step": 9708
    },
    {
      "epoch": 0.27974989915288423,
      "grad_norm": 0.824109673500061,
      "learning_rate": 8.451755853230783e-06,
      "loss": 0.7431,
      "step": 9709
    },
    {
      "epoch": 0.2797787126145335,
      "grad_norm": 0.8383349180221558,
      "learning_rate": 8.451418257702628e-06,
      "loss": 0.7749,
      "step": 9710
    },
    {
      "epoch": 0.2798075260761828,
      "grad_norm": 0.8542144298553467,
      "learning_rate": 8.451080632116068e-06,
      "loss": 0.7533,
      "step": 9711
    },
    {
      "epoch": 0.2798363395378321,
      "grad_norm": 0.81328284740448,
      "learning_rate": 8.450742976474047e-06,
      "loss": 0.7593,
      "step": 9712
    },
    {
      "epoch": 0.27986515299948134,
      "grad_norm": 0.8716728687286377,
      "learning_rate": 8.450405290779504e-06,
      "loss": 0.7579,
      "step": 9713
    },
    {
      "epoch": 0.2798939664611306,
      "grad_norm": 0.8369583487510681,
      "learning_rate": 8.45006757503538e-06,
      "loss": 0.7667,
      "step": 9714
    },
    {
      "epoch": 0.2799227799227799,
      "grad_norm": 0.8447775840759277,
      "learning_rate": 8.449729829244616e-06,
      "loss": 0.7484,
      "step": 9715
    },
    {
      "epoch": 0.2799515933844292,
      "grad_norm": 0.7967288494110107,
      "learning_rate": 8.449392053410154e-06,
      "loss": 0.7607,
      "step": 9716
    },
    {
      "epoch": 0.2799804068460785,
      "grad_norm": 0.8941354751586914,
      "learning_rate": 8.449054247534936e-06,
      "loss": 0.7709,
      "step": 9717
    },
    {
      "epoch": 0.2800092203077278,
      "grad_norm": 0.8648818135261536,
      "learning_rate": 8.448716411621901e-06,
      "loss": 0.7695,
      "step": 9718
    },
    {
      "epoch": 0.280038033769377,
      "grad_norm": 0.8455047607421875,
      "learning_rate": 8.448378545673995e-06,
      "loss": 0.7496,
      "step": 9719
    },
    {
      "epoch": 0.2800668472310263,
      "grad_norm": 0.841845691204071,
      "learning_rate": 8.448040649694158e-06,
      "loss": 0.7636,
      "step": 9720
    },
    {
      "epoch": 0.2800956606926756,
      "grad_norm": 0.8669489622116089,
      "learning_rate": 8.447702723685335e-06,
      "loss": 0.7882,
      "step": 9721
    },
    {
      "epoch": 0.2801244741543249,
      "grad_norm": 0.8654326796531677,
      "learning_rate": 8.447364767650468e-06,
      "loss": 0.7747,
      "step": 9722
    },
    {
      "epoch": 0.2801532876159742,
      "grad_norm": 0.8308976292610168,
      "learning_rate": 8.4470267815925e-06,
      "loss": 0.7361,
      "step": 9723
    },
    {
      "epoch": 0.2801821010776235,
      "grad_norm": 0.8555483222007751,
      "learning_rate": 8.446688765514373e-06,
      "loss": 0.7598,
      "step": 9724
    },
    {
      "epoch": 0.28021091453927277,
      "grad_norm": 0.8348924517631531,
      "learning_rate": 8.446350719419032e-06,
      "loss": 0.771,
      "step": 9725
    },
    {
      "epoch": 0.280239728000922,
      "grad_norm": 0.8393352031707764,
      "learning_rate": 8.446012643309424e-06,
      "loss": 0.7501,
      "step": 9726
    },
    {
      "epoch": 0.2802685414625713,
      "grad_norm": 0.8595114946365356,
      "learning_rate": 8.445674537188488e-06,
      "loss": 0.7523,
      "step": 9727
    },
    {
      "epoch": 0.2802973549242206,
      "grad_norm": 0.840069591999054,
      "learning_rate": 8.445336401059173e-06,
      "loss": 0.7323,
      "step": 9728
    },
    {
      "epoch": 0.2803261683858699,
      "grad_norm": 0.850426971912384,
      "learning_rate": 8.444998234924421e-06,
      "loss": 0.7345,
      "step": 9729
    },
    {
      "epoch": 0.28035498184751917,
      "grad_norm": 0.860846757888794,
      "learning_rate": 8.444660038787178e-06,
      "loss": 0.7463,
      "step": 9730
    },
    {
      "epoch": 0.28038379530916846,
      "grad_norm": 0.8818159103393555,
      "learning_rate": 8.44432181265039e-06,
      "loss": 0.7234,
      "step": 9731
    },
    {
      "epoch": 0.28041260877081775,
      "grad_norm": 0.8162270784378052,
      "learning_rate": 8.443983556517002e-06,
      "loss": 0.7462,
      "step": 9732
    },
    {
      "epoch": 0.280441422232467,
      "grad_norm": 0.8598573803901672,
      "learning_rate": 8.44364527038996e-06,
      "loss": 0.7561,
      "step": 9733
    },
    {
      "epoch": 0.2804702356941163,
      "grad_norm": 0.840991735458374,
      "learning_rate": 8.44330695427221e-06,
      "loss": 0.724,
      "step": 9734
    },
    {
      "epoch": 0.28049904915576557,
      "grad_norm": 0.8560243248939514,
      "learning_rate": 8.442968608166698e-06,
      "loss": 0.7507,
      "step": 9735
    },
    {
      "epoch": 0.28052786261741486,
      "grad_norm": 0.9121756553649902,
      "learning_rate": 8.442630232076372e-06,
      "loss": 0.7594,
      "step": 9736
    },
    {
      "epoch": 0.28055667607906415,
      "grad_norm": 0.8704620003700256,
      "learning_rate": 8.442291826004176e-06,
      "loss": 0.736,
      "step": 9737
    },
    {
      "epoch": 0.28058548954071344,
      "grad_norm": 0.8907340168952942,
      "learning_rate": 8.441953389953062e-06,
      "loss": 0.7982,
      "step": 9738
    },
    {
      "epoch": 0.28061430300236273,
      "grad_norm": 0.8749295473098755,
      "learning_rate": 8.441614923925974e-06,
      "loss": 0.7464,
      "step": 9739
    },
    {
      "epoch": 0.28064311646401197,
      "grad_norm": 0.8496185541152954,
      "learning_rate": 8.44127642792586e-06,
      "loss": 0.7534,
      "step": 9740
    },
    {
      "epoch": 0.28067192992566126,
      "grad_norm": 0.8835356831550598,
      "learning_rate": 8.440937901955668e-06,
      "loss": 0.7252,
      "step": 9741
    },
    {
      "epoch": 0.28070074338731055,
      "grad_norm": 0.8461697101593018,
      "learning_rate": 8.440599346018345e-06,
      "loss": 0.7532,
      "step": 9742
    },
    {
      "epoch": 0.28072955684895984,
      "grad_norm": 0.8852660655975342,
      "learning_rate": 8.440260760116843e-06,
      "loss": 0.7544,
      "step": 9743
    },
    {
      "epoch": 0.28075837031060913,
      "grad_norm": 0.8776226043701172,
      "learning_rate": 8.43992214425411e-06,
      "loss": 0.7617,
      "step": 9744
    },
    {
      "epoch": 0.2807871837722584,
      "grad_norm": 0.8769751191139221,
      "learning_rate": 8.439583498433092e-06,
      "loss": 0.7567,
      "step": 9745
    },
    {
      "epoch": 0.28081599723390765,
      "grad_norm": 0.842337429523468,
      "learning_rate": 8.43924482265674e-06,
      "loss": 0.7853,
      "step": 9746
    },
    {
      "epoch": 0.28084481069555695,
      "grad_norm": 0.8275361657142639,
      "learning_rate": 8.438906116928005e-06,
      "loss": 0.7422,
      "step": 9747
    },
    {
      "epoch": 0.28087362415720624,
      "grad_norm": 0.875183641910553,
      "learning_rate": 8.438567381249833e-06,
      "loss": 0.7453,
      "step": 9748
    },
    {
      "epoch": 0.2809024376188555,
      "grad_norm": 0.8724499940872192,
      "learning_rate": 8.438228615625179e-06,
      "loss": 0.7729,
      "step": 9749
    },
    {
      "epoch": 0.2809312510805048,
      "grad_norm": 0.845343291759491,
      "learning_rate": 8.437889820056988e-06,
      "loss": 0.776,
      "step": 9750
    },
    {
      "epoch": 0.2809600645421541,
      "grad_norm": 0.8594716191291809,
      "learning_rate": 8.437550994548216e-06,
      "loss": 0.78,
      "step": 9751
    },
    {
      "epoch": 0.2809888780038034,
      "grad_norm": 0.8255227208137512,
      "learning_rate": 8.43721213910181e-06,
      "loss": 0.7468,
      "step": 9752
    },
    {
      "epoch": 0.28101769146545263,
      "grad_norm": 0.8376170992851257,
      "learning_rate": 8.436873253720722e-06,
      "loss": 0.7739,
      "step": 9753
    },
    {
      "epoch": 0.2810465049271019,
      "grad_norm": 0.825579047203064,
      "learning_rate": 8.436534338407902e-06,
      "loss": 0.7485,
      "step": 9754
    },
    {
      "epoch": 0.2810753183887512,
      "grad_norm": 0.8650928139686584,
      "learning_rate": 8.436195393166306e-06,
      "loss": 0.7461,
      "step": 9755
    },
    {
      "epoch": 0.2811041318504005,
      "grad_norm": 0.8546355366706848,
      "learning_rate": 8.435856417998882e-06,
      "loss": 0.7722,
      "step": 9756
    },
    {
      "epoch": 0.2811329453120498,
      "grad_norm": 0.8035034537315369,
      "learning_rate": 8.435517412908584e-06,
      "loss": 0.7553,
      "step": 9757
    },
    {
      "epoch": 0.2811617587736991,
      "grad_norm": 0.8499542474746704,
      "learning_rate": 8.435178377898362e-06,
      "loss": 0.7718,
      "step": 9758
    },
    {
      "epoch": 0.2811905722353484,
      "grad_norm": 0.8304734826087952,
      "learning_rate": 8.434839312971172e-06,
      "loss": 0.7564,
      "step": 9759
    },
    {
      "epoch": 0.2812193856969976,
      "grad_norm": 0.8471037149429321,
      "learning_rate": 8.434500218129963e-06,
      "loss": 0.773,
      "step": 9760
    },
    {
      "epoch": 0.2812481991586469,
      "grad_norm": 0.8414168953895569,
      "learning_rate": 8.434161093377691e-06,
      "loss": 0.7638,
      "step": 9761
    },
    {
      "epoch": 0.2812770126202962,
      "grad_norm": 0.8678140640258789,
      "learning_rate": 8.433821938717311e-06,
      "loss": 0.7484,
      "step": 9762
    },
    {
      "epoch": 0.2813058260819455,
      "grad_norm": 0.8287981748580933,
      "learning_rate": 8.433482754151772e-06,
      "loss": 0.7597,
      "step": 9763
    },
    {
      "epoch": 0.2813346395435948,
      "grad_norm": 0.8447609543800354,
      "learning_rate": 8.433143539684032e-06,
      "loss": 0.7598,
      "step": 9764
    },
    {
      "epoch": 0.28136345300524407,
      "grad_norm": 0.870475709438324,
      "learning_rate": 8.432804295317044e-06,
      "loss": 0.7675,
      "step": 9765
    },
    {
      "epoch": 0.28139226646689336,
      "grad_norm": 0.8585666418075562,
      "learning_rate": 8.43246502105376e-06,
      "loss": 0.7477,
      "step": 9766
    },
    {
      "epoch": 0.2814210799285426,
      "grad_norm": 0.8424393534660339,
      "learning_rate": 8.432125716897136e-06,
      "loss": 0.7455,
      "step": 9767
    },
    {
      "epoch": 0.2814498933901919,
      "grad_norm": 0.8199095129966736,
      "learning_rate": 8.431786382850132e-06,
      "loss": 0.7669,
      "step": 9768
    },
    {
      "epoch": 0.2814787068518412,
      "grad_norm": 0.8465152978897095,
      "learning_rate": 8.431447018915695e-06,
      "loss": 0.7602,
      "step": 9769
    },
    {
      "epoch": 0.28150752031349047,
      "grad_norm": 0.8640962839126587,
      "learning_rate": 8.431107625096787e-06,
      "loss": 0.7866,
      "step": 9770
    },
    {
      "epoch": 0.28153633377513976,
      "grad_norm": 0.8472276926040649,
      "learning_rate": 8.430768201396362e-06,
      "loss": 0.7594,
      "step": 9771
    },
    {
      "epoch": 0.28156514723678905,
      "grad_norm": 0.8159722089767456,
      "learning_rate": 8.430428747817374e-06,
      "loss": 0.7489,
      "step": 9772
    },
    {
      "epoch": 0.2815939606984383,
      "grad_norm": 0.9294787645339966,
      "learning_rate": 8.43008926436278e-06,
      "loss": 0.757,
      "step": 9773
    },
    {
      "epoch": 0.2816227741600876,
      "grad_norm": 0.8458787202835083,
      "learning_rate": 8.42974975103554e-06,
      "loss": 0.7765,
      "step": 9774
    },
    {
      "epoch": 0.28165158762173687,
      "grad_norm": 0.8733122944831848,
      "learning_rate": 8.429410207838607e-06,
      "loss": 0.7655,
      "step": 9775
    },
    {
      "epoch": 0.28168040108338616,
      "grad_norm": 0.8472979068756104,
      "learning_rate": 8.429070634774939e-06,
      "loss": 0.7418,
      "step": 9776
    },
    {
      "epoch": 0.28170921454503545,
      "grad_norm": 0.8768532872200012,
      "learning_rate": 8.428731031847493e-06,
      "loss": 0.7736,
      "step": 9777
    },
    {
      "epoch": 0.28173802800668474,
      "grad_norm": 0.848952054977417,
      "learning_rate": 8.428391399059228e-06,
      "loss": 0.741,
      "step": 9778
    },
    {
      "epoch": 0.28176684146833403,
      "grad_norm": 0.8564133048057556,
      "learning_rate": 8.4280517364131e-06,
      "loss": 0.7637,
      "step": 9779
    },
    {
      "epoch": 0.28179565492998326,
      "grad_norm": 0.8329480886459351,
      "learning_rate": 8.427712043912069e-06,
      "loss": 0.7681,
      "step": 9780
    },
    {
      "epoch": 0.28182446839163255,
      "grad_norm": 0.8117567896842957,
      "learning_rate": 8.427372321559092e-06,
      "loss": 0.7389,
      "step": 9781
    },
    {
      "epoch": 0.28185328185328185,
      "grad_norm": 0.8398391604423523,
      "learning_rate": 8.42703256935713e-06,
      "loss": 0.7612,
      "step": 9782
    },
    {
      "epoch": 0.28188209531493114,
      "grad_norm": 0.8092480301856995,
      "learning_rate": 8.426692787309139e-06,
      "loss": 0.7452,
      "step": 9783
    },
    {
      "epoch": 0.2819109087765804,
      "grad_norm": 0.8660645484924316,
      "learning_rate": 8.426352975418078e-06,
      "loss": 0.7616,
      "step": 9784
    },
    {
      "epoch": 0.2819397222382297,
      "grad_norm": 0.8387265801429749,
      "learning_rate": 8.42601313368691e-06,
      "loss": 0.7818,
      "step": 9785
    },
    {
      "epoch": 0.281968535699879,
      "grad_norm": 0.805063784122467,
      "learning_rate": 8.42567326211859e-06,
      "loss": 0.7392,
      "step": 9786
    },
    {
      "epoch": 0.28199734916152824,
      "grad_norm": 0.850865364074707,
      "learning_rate": 8.425333360716083e-06,
      "loss": 0.7662,
      "step": 9787
    },
    {
      "epoch": 0.28202616262317753,
      "grad_norm": 0.8638531565666199,
      "learning_rate": 8.424993429482344e-06,
      "loss": 0.7493,
      "step": 9788
    },
    {
      "epoch": 0.2820549760848268,
      "grad_norm": 0.8532534837722778,
      "learning_rate": 8.424653468420337e-06,
      "loss": 0.7692,
      "step": 9789
    },
    {
      "epoch": 0.2820837895464761,
      "grad_norm": 0.8613641858100891,
      "learning_rate": 8.424313477533022e-06,
      "loss": 0.7669,
      "step": 9790
    },
    {
      "epoch": 0.2821126030081254,
      "grad_norm": 0.8274796605110168,
      "learning_rate": 8.42397345682336e-06,
      "loss": 0.7403,
      "step": 9791
    },
    {
      "epoch": 0.2821414164697747,
      "grad_norm": 0.8122320175170898,
      "learning_rate": 8.42363340629431e-06,
      "loss": 0.7553,
      "step": 9792
    },
    {
      "epoch": 0.282170229931424,
      "grad_norm": 0.8128442764282227,
      "learning_rate": 8.423293325948838e-06,
      "loss": 0.7474,
      "step": 9793
    },
    {
      "epoch": 0.2821990433930732,
      "grad_norm": 0.803383469581604,
      "learning_rate": 8.4229532157899e-06,
      "loss": 0.7666,
      "step": 9794
    },
    {
      "epoch": 0.2822278568547225,
      "grad_norm": 0.8310620188713074,
      "learning_rate": 8.422613075820465e-06,
      "loss": 0.7623,
      "step": 9795
    },
    {
      "epoch": 0.2822566703163718,
      "grad_norm": 0.8291082978248596,
      "learning_rate": 8.422272906043487e-06,
      "loss": 0.7512,
      "step": 9796
    },
    {
      "epoch": 0.2822854837780211,
      "grad_norm": 0.81293123960495,
      "learning_rate": 8.421932706461937e-06,
      "loss": 0.7509,
      "step": 9797
    },
    {
      "epoch": 0.2823142972396704,
      "grad_norm": 0.8421802520751953,
      "learning_rate": 8.421592477078772e-06,
      "loss": 0.7551,
      "step": 9798
    },
    {
      "epoch": 0.2823431107013197,
      "grad_norm": 0.8028570413589478,
      "learning_rate": 8.421252217896956e-06,
      "loss": 0.7575,
      "step": 9799
    },
    {
      "epoch": 0.2823719241629689,
      "grad_norm": 0.7992745637893677,
      "learning_rate": 8.420911928919455e-06,
      "loss": 0.743,
      "step": 9800
    },
    {
      "epoch": 0.2824007376246182,
      "grad_norm": 0.8035697340965271,
      "learning_rate": 8.42057161014923e-06,
      "loss": 0.749,
      "step": 9801
    },
    {
      "epoch": 0.2824295510862675,
      "grad_norm": 0.8042325973510742,
      "learning_rate": 8.420231261589245e-06,
      "loss": 0.7661,
      "step": 9802
    },
    {
      "epoch": 0.2824583645479168,
      "grad_norm": 0.8080511689186096,
      "learning_rate": 8.419890883242467e-06,
      "loss": 0.7783,
      "step": 9803
    },
    {
      "epoch": 0.2824871780095661,
      "grad_norm": 0.9178035855293274,
      "learning_rate": 8.419550475111856e-06,
      "loss": 0.7657,
      "step": 9804
    },
    {
      "epoch": 0.28251599147121537,
      "grad_norm": 0.8529528379440308,
      "learning_rate": 8.419210037200379e-06,
      "loss": 0.7669,
      "step": 9805
    },
    {
      "epoch": 0.28254480493286466,
      "grad_norm": 0.8274950385093689,
      "learning_rate": 8.418869569511001e-06,
      "loss": 0.7393,
      "step": 9806
    },
    {
      "epoch": 0.2825736183945139,
      "grad_norm": 0.9043546319007874,
      "learning_rate": 8.418529072046686e-06,
      "loss": 0.7592,
      "step": 9807
    },
    {
      "epoch": 0.2826024318561632,
      "grad_norm": 0.8394898176193237,
      "learning_rate": 8.4181885448104e-06,
      "loss": 0.7833,
      "step": 9808
    },
    {
      "epoch": 0.2826312453178125,
      "grad_norm": 0.7762308120727539,
      "learning_rate": 8.41784798780511e-06,
      "loss": 0.7378,
      "step": 9809
    },
    {
      "epoch": 0.28266005877946176,
      "grad_norm": 0.8479437828063965,
      "learning_rate": 8.41750740103378e-06,
      "loss": 0.7942,
      "step": 9810
    },
    {
      "epoch": 0.28268887224111106,
      "grad_norm": 0.8739098310470581,
      "learning_rate": 8.417166784499376e-06,
      "loss": 0.7577,
      "step": 9811
    },
    {
      "epoch": 0.28271768570276035,
      "grad_norm": 0.8318173885345459,
      "learning_rate": 8.416826138204865e-06,
      "loss": 0.7463,
      "step": 9812
    },
    {
      "epoch": 0.28274649916440964,
      "grad_norm": 0.8350359797477722,
      "learning_rate": 8.416485462153216e-06,
      "loss": 0.7506,
      "step": 9813
    },
    {
      "epoch": 0.2827753126260589,
      "grad_norm": 0.846991240978241,
      "learning_rate": 8.41614475634739e-06,
      "loss": 0.7569,
      "step": 9814
    },
    {
      "epoch": 0.28280412608770816,
      "grad_norm": 0.8272070288658142,
      "learning_rate": 8.415804020790362e-06,
      "loss": 0.748,
      "step": 9815
    },
    {
      "epoch": 0.28283293954935745,
      "grad_norm": 0.8775806427001953,
      "learning_rate": 8.415463255485094e-06,
      "loss": 0.7582,
      "step": 9816
    },
    {
      "epoch": 0.28286175301100674,
      "grad_norm": 0.850021243095398,
      "learning_rate": 8.415122460434556e-06,
      "loss": 0.741,
      "step": 9817
    },
    {
      "epoch": 0.28289056647265604,
      "grad_norm": 0.8055320978164673,
      "learning_rate": 8.414781635641714e-06,
      "loss": 0.761,
      "step": 9818
    },
    {
      "epoch": 0.2829193799343053,
      "grad_norm": 0.8506917953491211,
      "learning_rate": 8.414440781109538e-06,
      "loss": 0.7301,
      "step": 9819
    },
    {
      "epoch": 0.2829481933959546,
      "grad_norm": 0.833897054195404,
      "learning_rate": 8.414099896840995e-06,
      "loss": 0.7596,
      "step": 9820
    },
    {
      "epoch": 0.28297700685760385,
      "grad_norm": 0.8430997729301453,
      "learning_rate": 8.413758982839057e-06,
      "loss": 0.7401,
      "step": 9821
    },
    {
      "epoch": 0.28300582031925314,
      "grad_norm": 0.8277217745780945,
      "learning_rate": 8.413418039106689e-06,
      "loss": 0.7715,
      "step": 9822
    },
    {
      "epoch": 0.28303463378090243,
      "grad_norm": 0.8708632588386536,
      "learning_rate": 8.413077065646862e-06,
      "loss": 0.7594,
      "step": 9823
    },
    {
      "epoch": 0.2830634472425517,
      "grad_norm": 0.8255499005317688,
      "learning_rate": 8.412736062462547e-06,
      "loss": 0.7484,
      "step": 9824
    },
    {
      "epoch": 0.283092260704201,
      "grad_norm": 0.8561577200889587,
      "learning_rate": 8.412395029556712e-06,
      "loss": 0.751,
      "step": 9825
    },
    {
      "epoch": 0.2831210741658503,
      "grad_norm": 0.8518970608711243,
      "learning_rate": 8.412053966932326e-06,
      "loss": 0.752,
      "step": 9826
    },
    {
      "epoch": 0.28314988762749954,
      "grad_norm": 0.8791094422340393,
      "learning_rate": 8.411712874592363e-06,
      "loss": 0.766,
      "step": 9827
    },
    {
      "epoch": 0.28317870108914883,
      "grad_norm": 0.8390752673149109,
      "learning_rate": 8.41137175253979e-06,
      "loss": 0.7812,
      "step": 9828
    },
    {
      "epoch": 0.2832075145507981,
      "grad_norm": 0.827892005443573,
      "learning_rate": 8.41103060077758e-06,
      "loss": 0.7186,
      "step": 9829
    },
    {
      "epoch": 0.2832363280124474,
      "grad_norm": 0.8365288376808167,
      "learning_rate": 8.410689419308702e-06,
      "loss": 0.7513,
      "step": 9830
    },
    {
      "epoch": 0.2832651414740967,
      "grad_norm": 0.8826614022254944,
      "learning_rate": 8.410348208136129e-06,
      "loss": 0.7705,
      "step": 9831
    },
    {
      "epoch": 0.283293954935746,
      "grad_norm": 0.8217954039573669,
      "learning_rate": 8.410006967262832e-06,
      "loss": 0.7426,
      "step": 9832
    },
    {
      "epoch": 0.2833227683973953,
      "grad_norm": 0.8444023728370667,
      "learning_rate": 8.409665696691783e-06,
      "loss": 0.7666,
      "step": 9833
    },
    {
      "epoch": 0.2833515818590445,
      "grad_norm": 0.8492793440818787,
      "learning_rate": 8.409324396425956e-06,
      "loss": 0.758,
      "step": 9834
    },
    {
      "epoch": 0.2833803953206938,
      "grad_norm": 0.8523628115653992,
      "learning_rate": 8.408983066468319e-06,
      "loss": 0.788,
      "step": 9835
    },
    {
      "epoch": 0.2834092087823431,
      "grad_norm": 0.8408512473106384,
      "learning_rate": 8.408641706821849e-06,
      "loss": 0.753,
      "step": 9836
    },
    {
      "epoch": 0.2834380222439924,
      "grad_norm": 0.8902846574783325,
      "learning_rate": 8.408300317489516e-06,
      "loss": 0.7673,
      "step": 9837
    },
    {
      "epoch": 0.2834668357056417,
      "grad_norm": 0.8236010670661926,
      "learning_rate": 8.407958898474295e-06,
      "loss": 0.7475,
      "step": 9838
    },
    {
      "epoch": 0.283495649167291,
      "grad_norm": 0.8449414372444153,
      "learning_rate": 8.407617449779158e-06,
      "loss": 0.7511,
      "step": 9839
    },
    {
      "epoch": 0.28352446262894027,
      "grad_norm": 0.8254203796386719,
      "learning_rate": 8.40727597140708e-06,
      "loss": 0.7648,
      "step": 9840
    },
    {
      "epoch": 0.2835532760905895,
      "grad_norm": 0.8423912525177002,
      "learning_rate": 8.406934463361034e-06,
      "loss": 0.7834,
      "step": 9841
    },
    {
      "epoch": 0.2835820895522388,
      "grad_norm": 0.8246486186981201,
      "learning_rate": 8.406592925643995e-06,
      "loss": 0.7618,
      "step": 9842
    },
    {
      "epoch": 0.2836109030138881,
      "grad_norm": 0.8136628866195679,
      "learning_rate": 8.406251358258937e-06,
      "loss": 0.728,
      "step": 9843
    },
    {
      "epoch": 0.2836397164755374,
      "grad_norm": 0.8460549116134644,
      "learning_rate": 8.405909761208835e-06,
      "loss": 0.7568,
      "step": 9844
    },
    {
      "epoch": 0.28366852993718666,
      "grad_norm": 0.8597212433815002,
      "learning_rate": 8.405568134496662e-06,
      "loss": 0.7374,
      "step": 9845
    },
    {
      "epoch": 0.28369734339883595,
      "grad_norm": 0.8553836345672607,
      "learning_rate": 8.405226478125397e-06,
      "loss": 0.7558,
      "step": 9846
    },
    {
      "epoch": 0.28372615686048525,
      "grad_norm": 0.8217753767967224,
      "learning_rate": 8.404884792098012e-06,
      "loss": 0.7756,
      "step": 9847
    },
    {
      "epoch": 0.2837549703221345,
      "grad_norm": 0.8910686373710632,
      "learning_rate": 8.404543076417484e-06,
      "loss": 0.7718,
      "step": 9848
    },
    {
      "epoch": 0.28378378378378377,
      "grad_norm": 0.8676488399505615,
      "learning_rate": 8.40420133108679e-06,
      "loss": 0.766,
      "step": 9849
    },
    {
      "epoch": 0.28381259724543306,
      "grad_norm": 0.8160657286643982,
      "learning_rate": 8.403859556108904e-06,
      "loss": 0.7406,
      "step": 9850
    },
    {
      "epoch": 0.28384141070708235,
      "grad_norm": 0.899724543094635,
      "learning_rate": 8.403517751486806e-06,
      "loss": 0.7683,
      "step": 9851
    },
    {
      "epoch": 0.28387022416873164,
      "grad_norm": 0.8624120950698853,
      "learning_rate": 8.403175917223467e-06,
      "loss": 0.731,
      "step": 9852
    },
    {
      "epoch": 0.28389903763038093,
      "grad_norm": 0.8010725975036621,
      "learning_rate": 8.40283405332187e-06,
      "loss": 0.7476,
      "step": 9853
    },
    {
      "epoch": 0.28392785109203017,
      "grad_norm": 0.8240339756011963,
      "learning_rate": 8.402492159784993e-06,
      "loss": 0.7718,
      "step": 9854
    },
    {
      "epoch": 0.28395666455367946,
      "grad_norm": 0.9021530151367188,
      "learning_rate": 8.402150236615807e-06,
      "loss": 0.7547,
      "step": 9855
    },
    {
      "epoch": 0.28398547801532875,
      "grad_norm": 0.8258063793182373,
      "learning_rate": 8.401808283817291e-06,
      "loss": 0.7706,
      "step": 9856
    },
    {
      "epoch": 0.28401429147697804,
      "grad_norm": 0.8004977703094482,
      "learning_rate": 8.40146630139243e-06,
      "loss": 0.7552,
      "step": 9857
    },
    {
      "epoch": 0.28404310493862733,
      "grad_norm": 0.9037795066833496,
      "learning_rate": 8.401124289344195e-06,
      "loss": 0.7795,
      "step": 9858
    },
    {
      "epoch": 0.2840719184002766,
      "grad_norm": 0.8453738689422607,
      "learning_rate": 8.40078224767557e-06,
      "loss": 0.7475,
      "step": 9859
    },
    {
      "epoch": 0.2841007318619259,
      "grad_norm": 0.8356261253356934,
      "learning_rate": 8.400440176389529e-06,
      "loss": 0.7358,
      "step": 9860
    },
    {
      "epoch": 0.28412954532357515,
      "grad_norm": 0.7963055372238159,
      "learning_rate": 8.400098075489054e-06,
      "loss": 0.7396,
      "step": 9861
    },
    {
      "epoch": 0.28415835878522444,
      "grad_norm": 0.8354527950286865,
      "learning_rate": 8.399755944977123e-06,
      "loss": 0.7248,
      "step": 9862
    },
    {
      "epoch": 0.28418717224687373,
      "grad_norm": 0.8398975133895874,
      "learning_rate": 8.399413784856716e-06,
      "loss": 0.7636,
      "step": 9863
    },
    {
      "epoch": 0.284215985708523,
      "grad_norm": 0.8060164451599121,
      "learning_rate": 8.399071595130815e-06,
      "loss": 0.7399,
      "step": 9864
    },
    {
      "epoch": 0.2842447991701723,
      "grad_norm": 0.8048872947692871,
      "learning_rate": 8.398729375802397e-06,
      "loss": 0.7319,
      "step": 9865
    },
    {
      "epoch": 0.2842736126318216,
      "grad_norm": 0.8569262623786926,
      "learning_rate": 8.398387126874443e-06,
      "loss": 0.7788,
      "step": 9866
    },
    {
      "epoch": 0.2843024260934709,
      "grad_norm": 0.8251283764839172,
      "learning_rate": 8.398044848349935e-06,
      "loss": 0.7695,
      "step": 9867
    },
    {
      "epoch": 0.28433123955512013,
      "grad_norm": 0.8459335565567017,
      "learning_rate": 8.397702540231852e-06,
      "loss": 0.7574,
      "step": 9868
    },
    {
      "epoch": 0.2843600530167694,
      "grad_norm": 0.8257839679718018,
      "learning_rate": 8.397360202523178e-06,
      "loss": 0.723,
      "step": 9869
    },
    {
      "epoch": 0.2843888664784187,
      "grad_norm": 0.8269270658493042,
      "learning_rate": 8.397017835226893e-06,
      "loss": 0.7421,
      "step": 9870
    },
    {
      "epoch": 0.284417679940068,
      "grad_norm": 0.8504173755645752,
      "learning_rate": 8.396675438345976e-06,
      "loss": 0.7784,
      "step": 9871
    },
    {
      "epoch": 0.2844464934017173,
      "grad_norm": 0.8201809525489807,
      "learning_rate": 8.396333011883415e-06,
      "loss": 0.7482,
      "step": 9872
    },
    {
      "epoch": 0.2844753068633666,
      "grad_norm": 0.8442239761352539,
      "learning_rate": 8.395990555842187e-06,
      "loss": 0.749,
      "step": 9873
    },
    {
      "epoch": 0.2845041203250159,
      "grad_norm": 0.8918122053146362,
      "learning_rate": 8.395648070225273e-06,
      "loss": 0.7378,
      "step": 9874
    },
    {
      "epoch": 0.2845329337866651,
      "grad_norm": 0.8903546333312988,
      "learning_rate": 8.395305555035662e-06,
      "loss": 0.7605,
      "step": 9875
    },
    {
      "epoch": 0.2845617472483144,
      "grad_norm": 0.8677001595497131,
      "learning_rate": 8.394963010276333e-06,
      "loss": 0.7582,
      "step": 9876
    },
    {
      "epoch": 0.2845905607099637,
      "grad_norm": 0.8417643904685974,
      "learning_rate": 8.394620435950269e-06,
      "loss": 0.7586,
      "step": 9877
    },
    {
      "epoch": 0.284619374171613,
      "grad_norm": 0.9365625381469727,
      "learning_rate": 8.394277832060456e-06,
      "loss": 0.7323,
      "step": 9878
    },
    {
      "epoch": 0.2846481876332623,
      "grad_norm": 0.824874758720398,
      "learning_rate": 8.393935198609875e-06,
      "loss": 0.7494,
      "step": 9879
    },
    {
      "epoch": 0.28467700109491156,
      "grad_norm": 0.8118094205856323,
      "learning_rate": 8.393592535601511e-06,
      "loss": 0.7555,
      "step": 9880
    },
    {
      "epoch": 0.2847058145565608,
      "grad_norm": 0.8549738526344299,
      "learning_rate": 8.393249843038349e-06,
      "loss": 0.7574,
      "step": 9881
    },
    {
      "epoch": 0.2847346280182101,
      "grad_norm": 0.8972455263137817,
      "learning_rate": 8.392907120923372e-06,
      "loss": 0.7752,
      "step": 9882
    },
    {
      "epoch": 0.2847634414798594,
      "grad_norm": 0.7916013598442078,
      "learning_rate": 8.392564369259568e-06,
      "loss": 0.7423,
      "step": 9883
    },
    {
      "epoch": 0.28479225494150867,
      "grad_norm": 0.8420073390007019,
      "learning_rate": 8.392221588049918e-06,
      "loss": 0.7745,
      "step": 9884
    },
    {
      "epoch": 0.28482106840315796,
      "grad_norm": 0.85512375831604,
      "learning_rate": 8.391878777297409e-06,
      "loss": 0.7586,
      "step": 9885
    },
    {
      "epoch": 0.28484988186480725,
      "grad_norm": 0.836094081401825,
      "learning_rate": 8.391535937005026e-06,
      "loss": 0.7427,
      "step": 9886
    },
    {
      "epoch": 0.28487869532645654,
      "grad_norm": 0.7849149703979492,
      "learning_rate": 8.391193067175755e-06,
      "loss": 0.7506,
      "step": 9887
    },
    {
      "epoch": 0.2849075087881058,
      "grad_norm": 0.8524054884910583,
      "learning_rate": 8.390850167812583e-06,
      "loss": 0.7522,
      "step": 9888
    },
    {
      "epoch": 0.28493632224975507,
      "grad_norm": 0.833973228931427,
      "learning_rate": 8.390507238918497e-06,
      "loss": 0.7406,
      "step": 9889
    },
    {
      "epoch": 0.28496513571140436,
      "grad_norm": 0.8113918304443359,
      "learning_rate": 8.39016428049648e-06,
      "loss": 0.7422,
      "step": 9890
    },
    {
      "epoch": 0.28499394917305365,
      "grad_norm": 0.8411144614219666,
      "learning_rate": 8.389821292549523e-06,
      "loss": 0.7325,
      "step": 9891
    },
    {
      "epoch": 0.28502276263470294,
      "grad_norm": 0.8208893537521362,
      "learning_rate": 8.389478275080609e-06,
      "loss": 0.7689,
      "step": 9892
    },
    {
      "epoch": 0.28505157609635223,
      "grad_norm": 0.8018105030059814,
      "learning_rate": 8.389135228092728e-06,
      "loss": 0.746,
      "step": 9893
    },
    {
      "epoch": 0.2850803895580015,
      "grad_norm": 0.8443918824195862,
      "learning_rate": 8.388792151588867e-06,
      "loss": 0.7633,
      "step": 9894
    },
    {
      "epoch": 0.28510920301965076,
      "grad_norm": 0.8170769810676575,
      "learning_rate": 8.388449045572016e-06,
      "loss": 0.7701,
      "step": 9895
    },
    {
      "epoch": 0.28513801648130005,
      "grad_norm": 0.8121708035469055,
      "learning_rate": 8.388105910045159e-06,
      "loss": 0.755,
      "step": 9896
    },
    {
      "epoch": 0.28516682994294934,
      "grad_norm": 0.8416796922683716,
      "learning_rate": 8.387762745011287e-06,
      "loss": 0.767,
      "step": 9897
    },
    {
      "epoch": 0.28519564340459863,
      "grad_norm": 0.8621059656143188,
      "learning_rate": 8.387419550473388e-06,
      "loss": 0.7595,
      "step": 9898
    },
    {
      "epoch": 0.2852244568662479,
      "grad_norm": 0.8426170349121094,
      "learning_rate": 8.38707632643445e-06,
      "loss": 0.7596,
      "step": 9899
    },
    {
      "epoch": 0.2852532703278972,
      "grad_norm": 0.8366062641143799,
      "learning_rate": 8.386733072897464e-06,
      "loss": 0.7688,
      "step": 9900
    },
    {
      "epoch": 0.2852820837895465,
      "grad_norm": 0.8506807684898376,
      "learning_rate": 8.38638978986542e-06,
      "loss": 0.7591,
      "step": 9901
    },
    {
      "epoch": 0.28531089725119574,
      "grad_norm": 0.8258047699928284,
      "learning_rate": 8.386046477341302e-06,
      "loss": 0.74,
      "step": 9902
    },
    {
      "epoch": 0.28533971071284503,
      "grad_norm": 0.8418492078781128,
      "learning_rate": 8.385703135328108e-06,
      "loss": 0.7811,
      "step": 9903
    },
    {
      "epoch": 0.2853685241744943,
      "grad_norm": 0.8431321978569031,
      "learning_rate": 8.385359763828822e-06,
      "loss": 0.7599,
      "step": 9904
    },
    {
      "epoch": 0.2853973376361436,
      "grad_norm": 0.8204811811447144,
      "learning_rate": 8.385016362846437e-06,
      "loss": 0.7512,
      "step": 9905
    },
    {
      "epoch": 0.2854261510977929,
      "grad_norm": 0.8715309500694275,
      "learning_rate": 8.384672932383942e-06,
      "loss": 0.7536,
      "step": 9906
    },
    {
      "epoch": 0.2854549645594422,
      "grad_norm": 0.8072412014007568,
      "learning_rate": 8.384329472444332e-06,
      "loss": 0.7541,
      "step": 9907
    },
    {
      "epoch": 0.28548377802109143,
      "grad_norm": 0.8691192865371704,
      "learning_rate": 8.383985983030594e-06,
      "loss": 0.7289,
      "step": 9908
    },
    {
      "epoch": 0.2855125914827407,
      "grad_norm": 0.837632417678833,
      "learning_rate": 8.38364246414572e-06,
      "loss": 0.7618,
      "step": 9909
    },
    {
      "epoch": 0.28554140494439,
      "grad_norm": 0.822999894618988,
      "learning_rate": 8.383298915792703e-06,
      "loss": 0.7713,
      "step": 9910
    },
    {
      "epoch": 0.2855702184060393,
      "grad_norm": 0.835826575756073,
      "learning_rate": 8.382955337974534e-06,
      "loss": 0.7682,
      "step": 9911
    },
    {
      "epoch": 0.2855990318676886,
      "grad_norm": 0.8279374241828918,
      "learning_rate": 8.382611730694207e-06,
      "loss": 0.7719,
      "step": 9912
    },
    {
      "epoch": 0.2856278453293379,
      "grad_norm": 0.7849985957145691,
      "learning_rate": 8.382268093954712e-06,
      "loss": 0.7176,
      "step": 9913
    },
    {
      "epoch": 0.2856566587909872,
      "grad_norm": 0.8120601773262024,
      "learning_rate": 8.381924427759046e-06,
      "loss": 0.7408,
      "step": 9914
    },
    {
      "epoch": 0.2856854722526364,
      "grad_norm": 0.8616014122962952,
      "learning_rate": 8.381580732110195e-06,
      "loss": 0.7588,
      "step": 9915
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.8434789776802063,
      "learning_rate": 8.381237007011158e-06,
      "loss": 0.7433,
      "step": 9916
    },
    {
      "epoch": 0.285743099175935,
      "grad_norm": 0.8531871438026428,
      "learning_rate": 8.380893252464926e-06,
      "loss": 0.7895,
      "step": 9917
    },
    {
      "epoch": 0.2857719126375843,
      "grad_norm": 0.8636019229888916,
      "learning_rate": 8.380549468474495e-06,
      "loss": 0.7545,
      "step": 9918
    },
    {
      "epoch": 0.28580072609923357,
      "grad_norm": 0.8344849944114685,
      "learning_rate": 8.380205655042855e-06,
      "loss": 0.7403,
      "step": 9919
    },
    {
      "epoch": 0.28582953956088286,
      "grad_norm": 0.8616579174995422,
      "learning_rate": 8.379861812173004e-06,
      "loss": 0.7832,
      "step": 9920
    },
    {
      "epoch": 0.28585835302253215,
      "grad_norm": 0.9012340903282166,
      "learning_rate": 8.379517939867936e-06,
      "loss": 0.7685,
      "step": 9921
    },
    {
      "epoch": 0.2858871664841814,
      "grad_norm": 0.8039771914482117,
      "learning_rate": 8.379174038130645e-06,
      "loss": 0.7361,
      "step": 9922
    },
    {
      "epoch": 0.2859159799458307,
      "grad_norm": 0.8318992257118225,
      "learning_rate": 8.378830106964124e-06,
      "loss": 0.7651,
      "step": 9923
    },
    {
      "epoch": 0.28594479340747997,
      "grad_norm": 0.8711714148521423,
      "learning_rate": 8.378486146371372e-06,
      "loss": 0.7597,
      "step": 9924
    },
    {
      "epoch": 0.28597360686912926,
      "grad_norm": 0.8549885153770447,
      "learning_rate": 8.378142156355382e-06,
      "loss": 0.7548,
      "step": 9925
    },
    {
      "epoch": 0.28600242033077855,
      "grad_norm": 0.8604803085327148,
      "learning_rate": 8.377798136919152e-06,
      "loss": 0.7678,
      "step": 9926
    },
    {
      "epoch": 0.28603123379242784,
      "grad_norm": 0.7992481589317322,
      "learning_rate": 8.377454088065675e-06,
      "loss": 0.7381,
      "step": 9927
    },
    {
      "epoch": 0.28606004725407713,
      "grad_norm": 0.8443212509155273,
      "learning_rate": 8.37711000979795e-06,
      "loss": 0.769,
      "step": 9928
    },
    {
      "epoch": 0.28608886071572637,
      "grad_norm": 0.8357501029968262,
      "learning_rate": 8.376765902118973e-06,
      "loss": 0.7528,
      "step": 9929
    },
    {
      "epoch": 0.28611767417737566,
      "grad_norm": 0.7841864228248596,
      "learning_rate": 8.376421765031741e-06,
      "loss": 0.7549,
      "step": 9930
    },
    {
      "epoch": 0.28614648763902495,
      "grad_norm": 0.8523451089859009,
      "learning_rate": 8.37607759853925e-06,
      "loss": 0.7495,
      "step": 9931
    },
    {
      "epoch": 0.28617530110067424,
      "grad_norm": 0.8105366826057434,
      "learning_rate": 8.3757334026445e-06,
      "loss": 0.7596,
      "step": 9932
    },
    {
      "epoch": 0.28620411456232353,
      "grad_norm": 0.8715230822563171,
      "learning_rate": 8.375389177350484e-06,
      "loss": 0.7551,
      "step": 9933
    },
    {
      "epoch": 0.2862329280239728,
      "grad_norm": 0.8363797068595886,
      "learning_rate": 8.375044922660202e-06,
      "loss": 0.7579,
      "step": 9934
    },
    {
      "epoch": 0.28626174148562206,
      "grad_norm": 0.8398805260658264,
      "learning_rate": 8.374700638576655e-06,
      "loss": 0.7448,
      "step": 9935
    },
    {
      "epoch": 0.28629055494727135,
      "grad_norm": 0.835160493850708,
      "learning_rate": 8.374356325102835e-06,
      "loss": 0.7572,
      "step": 9936
    },
    {
      "epoch": 0.28631936840892064,
      "grad_norm": 0.8678650856018066,
      "learning_rate": 8.374011982241747e-06,
      "loss": 0.7322,
      "step": 9937
    },
    {
      "epoch": 0.28634818187056993,
      "grad_norm": 0.9045292735099792,
      "learning_rate": 8.373667609996388e-06,
      "loss": 0.7561,
      "step": 9938
    },
    {
      "epoch": 0.2863769953322192,
      "grad_norm": 0.819122314453125,
      "learning_rate": 8.373323208369757e-06,
      "loss": 0.742,
      "step": 9939
    },
    {
      "epoch": 0.2864058087938685,
      "grad_norm": 0.8548039197921753,
      "learning_rate": 8.37297877736485e-06,
      "loss": 0.7576,
      "step": 9940
    },
    {
      "epoch": 0.2864346222555178,
      "grad_norm": 0.8877045512199402,
      "learning_rate": 8.372634316984672e-06,
      "loss": 0.7794,
      "step": 9941
    },
    {
      "epoch": 0.28646343571716704,
      "grad_norm": 0.8698693513870239,
      "learning_rate": 8.37228982723222e-06,
      "loss": 0.7575,
      "step": 9942
    },
    {
      "epoch": 0.2864922491788163,
      "grad_norm": 0.8151934146881104,
      "learning_rate": 8.371945308110494e-06,
      "loss": 0.7493,
      "step": 9943
    },
    {
      "epoch": 0.2865210626404656,
      "grad_norm": 0.8209232687950134,
      "learning_rate": 8.371600759622496e-06,
      "loss": 0.773,
      "step": 9944
    },
    {
      "epoch": 0.2865498761021149,
      "grad_norm": 0.8599560856819153,
      "learning_rate": 8.371256181771226e-06,
      "loss": 0.7524,
      "step": 9945
    },
    {
      "epoch": 0.2865786895637642,
      "grad_norm": 0.8669360280036926,
      "learning_rate": 8.370911574559683e-06,
      "loss": 0.7313,
      "step": 9946
    },
    {
      "epoch": 0.2866075030254135,
      "grad_norm": 0.8263177275657654,
      "learning_rate": 8.370566937990872e-06,
      "loss": 0.7369,
      "step": 9947
    },
    {
      "epoch": 0.2866363164870628,
      "grad_norm": 0.888221025466919,
      "learning_rate": 8.370222272067793e-06,
      "loss": 0.7746,
      "step": 9948
    },
    {
      "epoch": 0.286665129948712,
      "grad_norm": 0.8637178540229797,
      "learning_rate": 8.369877576793443e-06,
      "loss": 0.7733,
      "step": 9949
    },
    {
      "epoch": 0.2866939434103613,
      "grad_norm": 0.9085138440132141,
      "learning_rate": 8.369532852170832e-06,
      "loss": 0.7727,
      "step": 9950
    },
    {
      "epoch": 0.2867227568720106,
      "grad_norm": 0.8873505592346191,
      "learning_rate": 8.369188098202957e-06,
      "loss": 0.7485,
      "step": 9951
    },
    {
      "epoch": 0.2867515703336599,
      "grad_norm": 0.8837985992431641,
      "learning_rate": 8.368843314892819e-06,
      "loss": 0.7765,
      "step": 9952
    },
    {
      "epoch": 0.2867803837953092,
      "grad_norm": 0.8835050463676453,
      "learning_rate": 8.368498502243425e-06,
      "loss": 0.7621,
      "step": 9953
    },
    {
      "epoch": 0.28680919725695847,
      "grad_norm": 0.9136900305747986,
      "learning_rate": 8.368153660257776e-06,
      "loss": 0.7765,
      "step": 9954
    },
    {
      "epoch": 0.28683801071860776,
      "grad_norm": 0.8542464971542358,
      "learning_rate": 8.367808788938877e-06,
      "loss": 0.7573,
      "step": 9955
    },
    {
      "epoch": 0.286866824180257,
      "grad_norm": 0.8518571853637695,
      "learning_rate": 8.367463888289728e-06,
      "loss": 0.7463,
      "step": 9956
    },
    {
      "epoch": 0.2868956376419063,
      "grad_norm": 0.8875941634178162,
      "learning_rate": 8.367118958313336e-06,
      "loss": 0.7515,
      "step": 9957
    },
    {
      "epoch": 0.2869244511035556,
      "grad_norm": 0.91974276304245,
      "learning_rate": 8.366773999012705e-06,
      "loss": 0.7615,
      "step": 9958
    },
    {
      "epoch": 0.28695326456520487,
      "grad_norm": 0.8661124110221863,
      "learning_rate": 8.366429010390837e-06,
      "loss": 0.7574,
      "step": 9959
    },
    {
      "epoch": 0.28698207802685416,
      "grad_norm": 0.8975598812103271,
      "learning_rate": 8.366083992450738e-06,
      "loss": 0.7749,
      "step": 9960
    },
    {
      "epoch": 0.28701089148850345,
      "grad_norm": 0.858052670955658,
      "learning_rate": 8.36573894519541e-06,
      "loss": 0.7395,
      "step": 9961
    },
    {
      "epoch": 0.2870397049501527,
      "grad_norm": 0.8296306133270264,
      "learning_rate": 8.365393868627863e-06,
      "loss": 0.7569,
      "step": 9962
    },
    {
      "epoch": 0.287068518411802,
      "grad_norm": 0.8478489518165588,
      "learning_rate": 8.365048762751097e-06,
      "loss": 0.7295,
      "step": 9963
    },
    {
      "epoch": 0.28709733187345127,
      "grad_norm": 0.9196361899375916,
      "learning_rate": 8.364703627568123e-06,
      "loss": 0.7488,
      "step": 9964
    },
    {
      "epoch": 0.28712614533510056,
      "grad_norm": 0.8152012825012207,
      "learning_rate": 8.364358463081943e-06,
      "loss": 0.7498,
      "step": 9965
    },
    {
      "epoch": 0.28715495879674985,
      "grad_norm": 0.8564147353172302,
      "learning_rate": 8.36401326929556e-06,
      "loss": 0.7555,
      "step": 9966
    },
    {
      "epoch": 0.28718377225839914,
      "grad_norm": 0.8348056077957153,
      "learning_rate": 8.36366804621199e-06,
      "loss": 0.7561,
      "step": 9967
    },
    {
      "epoch": 0.28721258572004843,
      "grad_norm": 0.8452224135398865,
      "learning_rate": 8.363322793834232e-06,
      "loss": 0.7611,
      "step": 9968
    },
    {
      "epoch": 0.28724139918169767,
      "grad_norm": 0.811263918876648,
      "learning_rate": 8.362977512165293e-06,
      "loss": 0.7535,
      "step": 9969
    },
    {
      "epoch": 0.28727021264334696,
      "grad_norm": 0.8456927537918091,
      "learning_rate": 8.362632201208182e-06,
      "loss": 0.7723,
      "step": 9970
    },
    {
      "epoch": 0.28729902610499625,
      "grad_norm": 0.8270636796951294,
      "learning_rate": 8.362286860965905e-06,
      "loss": 0.761,
      "step": 9971
    },
    {
      "epoch": 0.28732783956664554,
      "grad_norm": 0.834593653678894,
      "learning_rate": 8.361941491441473e-06,
      "loss": 0.7423,
      "step": 9972
    },
    {
      "epoch": 0.28735665302829483,
      "grad_norm": 0.8351224660873413,
      "learning_rate": 8.36159609263789e-06,
      "loss": 0.7648,
      "step": 9973
    },
    {
      "epoch": 0.2873854664899441,
      "grad_norm": 0.8498583436012268,
      "learning_rate": 8.361250664558166e-06,
      "loss": 0.7477,
      "step": 9974
    },
    {
      "epoch": 0.2874142799515934,
      "grad_norm": 0.8164024353027344,
      "learning_rate": 8.360905207205308e-06,
      "loss": 0.7432,
      "step": 9975
    },
    {
      "epoch": 0.28744309341324265,
      "grad_norm": 0.8333367705345154,
      "learning_rate": 8.360559720582325e-06,
      "loss": 0.78,
      "step": 9976
    },
    {
      "epoch": 0.28747190687489194,
      "grad_norm": 0.8075093030929565,
      "learning_rate": 8.360214204692227e-06,
      "loss": 0.7711,
      "step": 9977
    },
    {
      "epoch": 0.2875007203365412,
      "grad_norm": 0.8085695505142212,
      "learning_rate": 8.359868659538022e-06,
      "loss": 0.7519,
      "step": 9978
    },
    {
      "epoch": 0.2875295337981905,
      "grad_norm": 0.8200368881225586,
      "learning_rate": 8.35952308512272e-06,
      "loss": 0.7473,
      "step": 9979
    },
    {
      "epoch": 0.2875583472598398,
      "grad_norm": 0.8900620341300964,
      "learning_rate": 8.35917748144933e-06,
      "loss": 0.7703,
      "step": 9980
    },
    {
      "epoch": 0.2875871607214891,
      "grad_norm": 0.8222236633300781,
      "learning_rate": 8.358831848520862e-06,
      "loss": 0.7498,
      "step": 9981
    },
    {
      "epoch": 0.2876159741831384,
      "grad_norm": 0.8393956422805786,
      "learning_rate": 8.358486186340326e-06,
      "loss": 0.7677,
      "step": 9982
    },
    {
      "epoch": 0.2876447876447876,
      "grad_norm": 0.8087325096130371,
      "learning_rate": 8.358140494910734e-06,
      "loss": 0.757,
      "step": 9983
    },
    {
      "epoch": 0.2876736011064369,
      "grad_norm": 0.8231093287467957,
      "learning_rate": 8.357794774235094e-06,
      "loss": 0.769,
      "step": 9984
    },
    {
      "epoch": 0.2877024145680862,
      "grad_norm": 0.8377342820167542,
      "learning_rate": 8.357449024316418e-06,
      "loss": 0.7567,
      "step": 9985
    },
    {
      "epoch": 0.2877312280297355,
      "grad_norm": 0.8482453227043152,
      "learning_rate": 8.357103245157716e-06,
      "loss": 0.7692,
      "step": 9986
    },
    {
      "epoch": 0.2877600414913848,
      "grad_norm": 0.834065854549408,
      "learning_rate": 8.356757436762002e-06,
      "loss": 0.7325,
      "step": 9987
    },
    {
      "epoch": 0.2877888549530341,
      "grad_norm": 0.8094228506088257,
      "learning_rate": 8.356411599132286e-06,
      "loss": 0.7628,
      "step": 9988
    },
    {
      "epoch": 0.2878176684146833,
      "grad_norm": 0.8067699074745178,
      "learning_rate": 8.35606573227158e-06,
      "loss": 0.7529,
      "step": 9989
    },
    {
      "epoch": 0.2878464818763326,
      "grad_norm": 0.807611346244812,
      "learning_rate": 8.355719836182896e-06,
      "loss": 0.7694,
      "step": 9990
    },
    {
      "epoch": 0.2878752953379819,
      "grad_norm": 0.8026532530784607,
      "learning_rate": 8.355373910869249e-06,
      "loss": 0.76,
      "step": 9991
    },
    {
      "epoch": 0.2879041087996312,
      "grad_norm": 0.8212977647781372,
      "learning_rate": 8.355027956333648e-06,
      "loss": 0.7545,
      "step": 9992
    },
    {
      "epoch": 0.2879329222612805,
      "grad_norm": 0.8001276850700378,
      "learning_rate": 8.354681972579106e-06,
      "loss": 0.7491,
      "step": 9993
    },
    {
      "epoch": 0.28796173572292977,
      "grad_norm": 0.7776724100112915,
      "learning_rate": 8.354335959608638e-06,
      "loss": 0.7205,
      "step": 9994
    },
    {
      "epoch": 0.28799054918457906,
      "grad_norm": 0.8326725363731384,
      "learning_rate": 8.353989917425258e-06,
      "loss": 0.7652,
      "step": 9995
    },
    {
      "epoch": 0.2880193626462283,
      "grad_norm": 0.8058792352676392,
      "learning_rate": 8.353643846031978e-06,
      "loss": 0.7614,
      "step": 9996
    },
    {
      "epoch": 0.2880481761078776,
      "grad_norm": 0.8200294375419617,
      "learning_rate": 8.353297745431811e-06,
      "loss": 0.7612,
      "step": 9997
    },
    {
      "epoch": 0.2880769895695269,
      "grad_norm": 0.8083036541938782,
      "learning_rate": 8.352951615627775e-06,
      "loss": 0.74,
      "step": 9998
    },
    {
      "epoch": 0.28810580303117617,
      "grad_norm": 0.8354085683822632,
      "learning_rate": 8.35260545662288e-06,
      "loss": 0.7366,
      "step": 9999
    },
    {
      "epoch": 0.28813461649282546,
      "grad_norm": 0.8465505838394165,
      "learning_rate": 8.352259268420146e-06,
      "loss": 0.7436,
      "step": 10000
    },
    {
      "epoch": 0.28816342995447475,
      "grad_norm": 0.8440613150596619,
      "learning_rate": 8.351913051022582e-06,
      "loss": 0.7798,
      "step": 10001
    },
    {
      "epoch": 0.28819224341612404,
      "grad_norm": 0.811385452747345,
      "learning_rate": 8.351566804433207e-06,
      "loss": 0.764,
      "step": 10002
    },
    {
      "epoch": 0.2882210568777733,
      "grad_norm": 0.831434428691864,
      "learning_rate": 8.351220528655035e-06,
      "loss": 0.7624,
      "step": 10003
    },
    {
      "epoch": 0.28824987033942256,
      "grad_norm": 0.8385694026947021,
      "learning_rate": 8.350874223691083e-06,
      "loss": 0.7727,
      "step": 10004
    },
    {
      "epoch": 0.28827868380107186,
      "grad_norm": 0.8952437043190002,
      "learning_rate": 8.350527889544367e-06,
      "loss": 0.7692,
      "step": 10005
    },
    {
      "epoch": 0.28830749726272115,
      "grad_norm": 0.8151806592941284,
      "learning_rate": 8.3501815262179e-06,
      "loss": 0.742,
      "step": 10006
    },
    {
      "epoch": 0.28833631072437044,
      "grad_norm": 0.8043649792671204,
      "learning_rate": 8.349835133714703e-06,
      "loss": 0.7651,
      "step": 10007
    },
    {
      "epoch": 0.28836512418601973,
      "grad_norm": 0.8732089400291443,
      "learning_rate": 8.34948871203779e-06,
      "loss": 0.7557,
      "step": 10008
    },
    {
      "epoch": 0.288393937647669,
      "grad_norm": 0.8472023606300354,
      "learning_rate": 8.349142261190178e-06,
      "loss": 0.7476,
      "step": 10009
    },
    {
      "epoch": 0.28842275110931825,
      "grad_norm": 0.8134183287620544,
      "learning_rate": 8.348795781174885e-06,
      "loss": 0.7809,
      "step": 10010
    },
    {
      "epoch": 0.28845156457096754,
      "grad_norm": 0.8139396905899048,
      "learning_rate": 8.348449271994928e-06,
      "loss": 0.742,
      "step": 10011
    },
    {
      "epoch": 0.28848037803261684,
      "grad_norm": 0.8272242546081543,
      "learning_rate": 8.348102733653324e-06,
      "loss": 0.7571,
      "step": 10012
    },
    {
      "epoch": 0.2885091914942661,
      "grad_norm": 0.7903175354003906,
      "learning_rate": 8.347756166153095e-06,
      "loss": 0.7577,
      "step": 10013
    },
    {
      "epoch": 0.2885380049559154,
      "grad_norm": 0.768747091293335,
      "learning_rate": 8.347409569497254e-06,
      "loss": 0.7503,
      "step": 10014
    },
    {
      "epoch": 0.2885668184175647,
      "grad_norm": 0.8170306086540222,
      "learning_rate": 8.347062943688823e-06,
      "loss": 0.7442,
      "step": 10015
    },
    {
      "epoch": 0.28859563187921394,
      "grad_norm": 0.8262405395507812,
      "learning_rate": 8.34671628873082e-06,
      "loss": 0.7688,
      "step": 10016
    },
    {
      "epoch": 0.28862444534086323,
      "grad_norm": 0.8375255465507507,
      "learning_rate": 8.346369604626262e-06,
      "loss": 0.7756,
      "step": 10017
    },
    {
      "epoch": 0.2886532588025125,
      "grad_norm": 0.8238489627838135,
      "learning_rate": 8.346022891378171e-06,
      "loss": 0.7382,
      "step": 10018
    },
    {
      "epoch": 0.2886820722641618,
      "grad_norm": 0.7902451157569885,
      "learning_rate": 8.345676148989565e-06,
      "loss": 0.7476,
      "step": 10019
    },
    {
      "epoch": 0.2887108857258111,
      "grad_norm": 0.8296172618865967,
      "learning_rate": 8.345329377463464e-06,
      "loss": 0.755,
      "step": 10020
    },
    {
      "epoch": 0.2887396991874604,
      "grad_norm": 0.8227055668830872,
      "learning_rate": 8.344982576802889e-06,
      "loss": 0.7427,
      "step": 10021
    },
    {
      "epoch": 0.2887685126491097,
      "grad_norm": 0.8656448125839233,
      "learning_rate": 8.344635747010859e-06,
      "loss": 0.7885,
      "step": 10022
    },
    {
      "epoch": 0.2887973261107589,
      "grad_norm": 0.8166444301605225,
      "learning_rate": 8.344288888090396e-06,
      "loss": 0.7629,
      "step": 10023
    },
    {
      "epoch": 0.2888261395724082,
      "grad_norm": 0.8513675928115845,
      "learning_rate": 8.343942000044518e-06,
      "loss": 0.7526,
      "step": 10024
    },
    {
      "epoch": 0.2888549530340575,
      "grad_norm": 0.8379186987876892,
      "learning_rate": 8.34359508287625e-06,
      "loss": 0.7561,
      "step": 10025
    },
    {
      "epoch": 0.2888837664957068,
      "grad_norm": 0.7993063926696777,
      "learning_rate": 8.34324813658861e-06,
      "loss": 0.76,
      "step": 10026
    },
    {
      "epoch": 0.2889125799573561,
      "grad_norm": 0.8517819046974182,
      "learning_rate": 8.34290116118462e-06,
      "loss": 0.7587,
      "step": 10027
    },
    {
      "epoch": 0.2889413934190054,
      "grad_norm": 0.7969347238540649,
      "learning_rate": 8.342554156667303e-06,
      "loss": 0.7433,
      "step": 10028
    },
    {
      "epoch": 0.28897020688065467,
      "grad_norm": 0.8209611177444458,
      "learning_rate": 8.342207123039681e-06,
      "loss": 0.7821,
      "step": 10029
    },
    {
      "epoch": 0.2889990203423039,
      "grad_norm": 0.8410317301750183,
      "learning_rate": 8.341860060304776e-06,
      "loss": 0.7845,
      "step": 10030
    },
    {
      "epoch": 0.2890278338039532,
      "grad_norm": 0.8496650457382202,
      "learning_rate": 8.34151296846561e-06,
      "loss": 0.7436,
      "step": 10031
    },
    {
      "epoch": 0.2890566472656025,
      "grad_norm": 0.8285030126571655,
      "learning_rate": 8.341165847525209e-06,
      "loss": 0.7445,
      "step": 10032
    },
    {
      "epoch": 0.2890854607272518,
      "grad_norm": 0.8465824723243713,
      "learning_rate": 8.34081869748659e-06,
      "loss": 0.7584,
      "step": 10033
    },
    {
      "epoch": 0.28911427418890107,
      "grad_norm": 0.8221558332443237,
      "learning_rate": 8.340471518352781e-06,
      "loss": 0.7543,
      "step": 10034
    },
    {
      "epoch": 0.28914308765055036,
      "grad_norm": 0.8016360402107239,
      "learning_rate": 8.340124310126805e-06,
      "loss": 0.7626,
      "step": 10035
    },
    {
      "epoch": 0.2891719011121996,
      "grad_norm": 0.8448188900947571,
      "learning_rate": 8.339777072811684e-06,
      "loss": 0.7419,
      "step": 10036
    },
    {
      "epoch": 0.2892007145738489,
      "grad_norm": 0.8880829215049744,
      "learning_rate": 8.339429806410444e-06,
      "loss": 0.7661,
      "step": 10037
    },
    {
      "epoch": 0.2892295280354982,
      "grad_norm": 0.830018937587738,
      "learning_rate": 8.339082510926108e-06,
      "loss": 0.7775,
      "step": 10038
    },
    {
      "epoch": 0.28925834149714746,
      "grad_norm": 0.8301506638526917,
      "learning_rate": 8.338735186361702e-06,
      "loss": 0.738,
      "step": 10039
    },
    {
      "epoch": 0.28928715495879676,
      "grad_norm": 0.9049792289733887,
      "learning_rate": 8.33838783272025e-06,
      "loss": 0.7524,
      "step": 10040
    },
    {
      "epoch": 0.28931596842044605,
      "grad_norm": 0.8621746301651001,
      "learning_rate": 8.338040450004776e-06,
      "loss": 0.7563,
      "step": 10041
    },
    {
      "epoch": 0.28934478188209534,
      "grad_norm": 0.8159582018852234,
      "learning_rate": 8.337693038218307e-06,
      "loss": 0.7494,
      "step": 10042
    },
    {
      "epoch": 0.28937359534374457,
      "grad_norm": 0.8551239967346191,
      "learning_rate": 8.33734559736387e-06,
      "loss": 0.7478,
      "step": 10043
    },
    {
      "epoch": 0.28940240880539386,
      "grad_norm": 0.7952032685279846,
      "learning_rate": 8.336998127444487e-06,
      "loss": 0.7537,
      "step": 10044
    },
    {
      "epoch": 0.28943122226704315,
      "grad_norm": 0.8430699110031128,
      "learning_rate": 8.336650628463186e-06,
      "loss": 0.7582,
      "step": 10045
    },
    {
      "epoch": 0.28946003572869244,
      "grad_norm": 0.8870272040367126,
      "learning_rate": 8.336303100422993e-06,
      "loss": 0.7686,
      "step": 10046
    },
    {
      "epoch": 0.28948884919034173,
      "grad_norm": 0.8533257842063904,
      "learning_rate": 8.335955543326936e-06,
      "loss": 0.7654,
      "step": 10047
    },
    {
      "epoch": 0.289517662651991,
      "grad_norm": 0.8383223414421082,
      "learning_rate": 8.33560795717804e-06,
      "loss": 0.7661,
      "step": 10048
    },
    {
      "epoch": 0.2895464761136403,
      "grad_norm": 0.8559989929199219,
      "learning_rate": 8.335260341979334e-06,
      "loss": 0.7667,
      "step": 10049
    },
    {
      "epoch": 0.28957528957528955,
      "grad_norm": 0.8011196851730347,
      "learning_rate": 8.334912697733846e-06,
      "loss": 0.7493,
      "step": 10050
    },
    {
      "epoch": 0.28960410303693884,
      "grad_norm": 0.8197978734970093,
      "learning_rate": 8.3345650244446e-06,
      "loss": 0.7677,
      "step": 10051
    },
    {
      "epoch": 0.28963291649858813,
      "grad_norm": 0.8583512306213379,
      "learning_rate": 8.334217322114626e-06,
      "loss": 0.7606,
      "step": 10052
    },
    {
      "epoch": 0.2896617299602374,
      "grad_norm": 0.8222407698631287,
      "learning_rate": 8.333869590746953e-06,
      "loss": 0.7446,
      "step": 10053
    },
    {
      "epoch": 0.2896905434218867,
      "grad_norm": 0.7858763933181763,
      "learning_rate": 8.333521830344609e-06,
      "loss": 0.766,
      "step": 10054
    },
    {
      "epoch": 0.289719356883536,
      "grad_norm": 0.8527705669403076,
      "learning_rate": 8.333174040910621e-06,
      "loss": 0.8023,
      "step": 10055
    },
    {
      "epoch": 0.2897481703451853,
      "grad_norm": 0.8355134725570679,
      "learning_rate": 8.332826222448018e-06,
      "loss": 0.7505,
      "step": 10056
    },
    {
      "epoch": 0.28977698380683453,
      "grad_norm": 0.8215440511703491,
      "learning_rate": 8.332478374959832e-06,
      "loss": 0.7483,
      "step": 10057
    },
    {
      "epoch": 0.2898057972684838,
      "grad_norm": 0.8353074193000793,
      "learning_rate": 8.33213049844909e-06,
      "loss": 0.7408,
      "step": 10058
    },
    {
      "epoch": 0.2898346107301331,
      "grad_norm": 0.789740264415741,
      "learning_rate": 8.331782592918822e-06,
      "loss": 0.7532,
      "step": 10059
    },
    {
      "epoch": 0.2898634241917824,
      "grad_norm": 0.8336406350135803,
      "learning_rate": 8.33143465837206e-06,
      "loss": 0.7523,
      "step": 10060
    },
    {
      "epoch": 0.2898922376534317,
      "grad_norm": 0.827938437461853,
      "learning_rate": 8.33108669481183e-06,
      "loss": 0.7563,
      "step": 10061
    },
    {
      "epoch": 0.289921051115081,
      "grad_norm": 0.8278096914291382,
      "learning_rate": 8.330738702241166e-06,
      "loss": 0.7518,
      "step": 10062
    },
    {
      "epoch": 0.2899498645767302,
      "grad_norm": 0.8163110613822937,
      "learning_rate": 8.330390680663098e-06,
      "loss": 0.7355,
      "step": 10063
    },
    {
      "epoch": 0.2899786780383795,
      "grad_norm": 0.8460318446159363,
      "learning_rate": 8.330042630080655e-06,
      "loss": 0.7749,
      "step": 10064
    },
    {
      "epoch": 0.2900074915000288,
      "grad_norm": 0.838749885559082,
      "learning_rate": 8.32969455049687e-06,
      "loss": 0.7504,
      "step": 10065
    },
    {
      "epoch": 0.2900363049616781,
      "grad_norm": 0.829856812953949,
      "learning_rate": 8.329346441914774e-06,
      "loss": 0.7623,
      "step": 10066
    },
    {
      "epoch": 0.2900651184233274,
      "grad_norm": 0.8119767308235168,
      "learning_rate": 8.328998304337399e-06,
      "loss": 0.7473,
      "step": 10067
    },
    {
      "epoch": 0.2900939318849767,
      "grad_norm": 0.88938969373703,
      "learning_rate": 8.328650137767775e-06,
      "loss": 0.7787,
      "step": 10068
    },
    {
      "epoch": 0.29012274534662597,
      "grad_norm": 0.8300673961639404,
      "learning_rate": 8.328301942208938e-06,
      "loss": 0.757,
      "step": 10069
    },
    {
      "epoch": 0.2901515588082752,
      "grad_norm": 0.8265442252159119,
      "learning_rate": 8.327953717663917e-06,
      "loss": 0.7432,
      "step": 10070
    },
    {
      "epoch": 0.2901803722699245,
      "grad_norm": 0.8104313015937805,
      "learning_rate": 8.327605464135744e-06,
      "loss": 0.7537,
      "step": 10071
    },
    {
      "epoch": 0.2902091857315738,
      "grad_norm": 0.8798257112503052,
      "learning_rate": 8.327257181627457e-06,
      "loss": 0.7662,
      "step": 10072
    },
    {
      "epoch": 0.2902379991932231,
      "grad_norm": 0.7938525676727295,
      "learning_rate": 8.326908870142085e-06,
      "loss": 0.7506,
      "step": 10073
    },
    {
      "epoch": 0.29026681265487236,
      "grad_norm": 0.873704195022583,
      "learning_rate": 8.326560529682661e-06,
      "loss": 0.7732,
      "step": 10074
    },
    {
      "epoch": 0.29029562611652165,
      "grad_norm": 0.8052147030830383,
      "learning_rate": 8.326212160252221e-06,
      "loss": 0.7424,
      "step": 10075
    },
    {
      "epoch": 0.29032443957817095,
      "grad_norm": 0.8437504172325134,
      "learning_rate": 8.325863761853798e-06,
      "loss": 0.747,
      "step": 10076
    },
    {
      "epoch": 0.2903532530398202,
      "grad_norm": 0.817603588104248,
      "learning_rate": 8.325515334490427e-06,
      "loss": 0.7592,
      "step": 10077
    },
    {
      "epoch": 0.29038206650146947,
      "grad_norm": 0.8121086955070496,
      "learning_rate": 8.32516687816514e-06,
      "loss": 0.7691,
      "step": 10078
    },
    {
      "epoch": 0.29041087996311876,
      "grad_norm": 0.8008537292480469,
      "learning_rate": 8.324818392880975e-06,
      "loss": 0.7508,
      "step": 10079
    },
    {
      "epoch": 0.29043969342476805,
      "grad_norm": 0.8249675631523132,
      "learning_rate": 8.324469878640966e-06,
      "loss": 0.7457,
      "step": 10080
    },
    {
      "epoch": 0.29046850688641734,
      "grad_norm": 0.8608337044715881,
      "learning_rate": 8.324121335448147e-06,
      "loss": 0.7686,
      "step": 10081
    },
    {
      "epoch": 0.29049732034806663,
      "grad_norm": 0.8060774803161621,
      "learning_rate": 8.323772763305555e-06,
      "loss": 0.7501,
      "step": 10082
    },
    {
      "epoch": 0.2905261338097159,
      "grad_norm": 0.841805636882782,
      "learning_rate": 8.323424162216223e-06,
      "loss": 0.7694,
      "step": 10083
    },
    {
      "epoch": 0.29055494727136516,
      "grad_norm": 0.8870923519134521,
      "learning_rate": 8.32307553218319e-06,
      "loss": 0.751,
      "step": 10084
    },
    {
      "epoch": 0.29058376073301445,
      "grad_norm": 0.82708340883255,
      "learning_rate": 8.32272687320949e-06,
      "loss": 0.7715,
      "step": 10085
    },
    {
      "epoch": 0.29061257419466374,
      "grad_norm": 0.798630952835083,
      "learning_rate": 8.32237818529816e-06,
      "loss": 0.7681,
      "step": 10086
    },
    {
      "epoch": 0.29064138765631303,
      "grad_norm": 0.8211667537689209,
      "learning_rate": 8.32202946845224e-06,
      "loss": 0.752,
      "step": 10087
    },
    {
      "epoch": 0.2906702011179623,
      "grad_norm": 0.8327888250350952,
      "learning_rate": 8.32168072267476e-06,
      "loss": 0.7508,
      "step": 10088
    },
    {
      "epoch": 0.2906990145796116,
      "grad_norm": 0.83570396900177,
      "learning_rate": 8.321331947968762e-06,
      "loss": 0.7599,
      "step": 10089
    },
    {
      "epoch": 0.29072782804126085,
      "grad_norm": 0.8534596562385559,
      "learning_rate": 8.320983144337286e-06,
      "loss": 0.7315,
      "step": 10090
    },
    {
      "epoch": 0.29075664150291014,
      "grad_norm": 0.8399370908737183,
      "learning_rate": 8.320634311783364e-06,
      "loss": 0.7423,
      "step": 10091
    },
    {
      "epoch": 0.29078545496455943,
      "grad_norm": 0.8192590475082397,
      "learning_rate": 8.320285450310038e-06,
      "loss": 0.7491,
      "step": 10092
    },
    {
      "epoch": 0.2908142684262087,
      "grad_norm": 0.8199911117553711,
      "learning_rate": 8.319936559920345e-06,
      "loss": 0.7572,
      "step": 10093
    },
    {
      "epoch": 0.290843081887858,
      "grad_norm": 0.8541300296783447,
      "learning_rate": 8.319587640617322e-06,
      "loss": 0.7487,
      "step": 10094
    },
    {
      "epoch": 0.2908718953495073,
      "grad_norm": 0.8188740015029907,
      "learning_rate": 8.319238692404009e-06,
      "loss": 0.7394,
      "step": 10095
    },
    {
      "epoch": 0.2909007088111566,
      "grad_norm": 0.8368216753005981,
      "learning_rate": 8.318889715283446e-06,
      "loss": 0.7752,
      "step": 10096
    },
    {
      "epoch": 0.29092952227280583,
      "grad_norm": 0.8344018459320068,
      "learning_rate": 8.31854070925867e-06,
      "loss": 0.7627,
      "step": 10097
    },
    {
      "epoch": 0.2909583357344551,
      "grad_norm": 0.8321034908294678,
      "learning_rate": 8.318191674332724e-06,
      "loss": 0.7613,
      "step": 10098
    },
    {
      "epoch": 0.2909871491961044,
      "grad_norm": 0.8540560007095337,
      "learning_rate": 8.317842610508645e-06,
      "loss": 0.7718,
      "step": 10099
    },
    {
      "epoch": 0.2910159626577537,
      "grad_norm": 0.8611857295036316,
      "learning_rate": 8.317493517789472e-06,
      "loss": 0.7651,
      "step": 10100
    },
    {
      "epoch": 0.291044776119403,
      "grad_norm": 0.8072512149810791,
      "learning_rate": 8.317144396178247e-06,
      "loss": 0.7387,
      "step": 10101
    },
    {
      "epoch": 0.2910735895810523,
      "grad_norm": 0.8593738079071045,
      "learning_rate": 8.31679524567801e-06,
      "loss": 0.7584,
      "step": 10102
    },
    {
      "epoch": 0.2911024030427016,
      "grad_norm": 0.8433664441108704,
      "learning_rate": 8.316446066291802e-06,
      "loss": 0.7664,
      "step": 10103
    },
    {
      "epoch": 0.2911312165043508,
      "grad_norm": 0.8272392153739929,
      "learning_rate": 8.316096858022665e-06,
      "loss": 0.752,
      "step": 10104
    },
    {
      "epoch": 0.2911600299660001,
      "grad_norm": 0.8061601519584656,
      "learning_rate": 8.31574762087364e-06,
      "loss": 0.7457,
      "step": 10105
    },
    {
      "epoch": 0.2911888434276494,
      "grad_norm": 0.9068531394004822,
      "learning_rate": 8.315398354847766e-06,
      "loss": 0.7701,
      "step": 10106
    },
    {
      "epoch": 0.2912176568892987,
      "grad_norm": 0.8459575176239014,
      "learning_rate": 8.315049059948086e-06,
      "loss": 0.7321,
      "step": 10107
    },
    {
      "epoch": 0.291246470350948,
      "grad_norm": 0.868429958820343,
      "learning_rate": 8.314699736177643e-06,
      "loss": 0.7605,
      "step": 10108
    },
    {
      "epoch": 0.29127528381259726,
      "grad_norm": 0.831237256526947,
      "learning_rate": 8.314350383539481e-06,
      "loss": 0.7559,
      "step": 10109
    },
    {
      "epoch": 0.29130409727424655,
      "grad_norm": 0.8435024619102478,
      "learning_rate": 8.314001002036638e-06,
      "loss": 0.7178,
      "step": 10110
    },
    {
      "epoch": 0.2913329107358958,
      "grad_norm": 0.853885293006897,
      "learning_rate": 8.31365159167216e-06,
      "loss": 0.7717,
      "step": 10111
    },
    {
      "epoch": 0.2913617241975451,
      "grad_norm": 0.8277021050453186,
      "learning_rate": 8.313302152449089e-06,
      "loss": 0.7687,
      "step": 10112
    },
    {
      "epoch": 0.29139053765919437,
      "grad_norm": 0.8183115720748901,
      "learning_rate": 8.312952684370468e-06,
      "loss": 0.7309,
      "step": 10113
    },
    {
      "epoch": 0.29141935112084366,
      "grad_norm": 0.7854560017585754,
      "learning_rate": 8.31260318743934e-06,
      "loss": 0.7582,
      "step": 10114
    },
    {
      "epoch": 0.29144816458249295,
      "grad_norm": 0.8030362129211426,
      "learning_rate": 8.312253661658751e-06,
      "loss": 0.7307,
      "step": 10115
    },
    {
      "epoch": 0.29147697804414224,
      "grad_norm": 0.7597546577453613,
      "learning_rate": 8.311904107031742e-06,
      "loss": 0.7416,
      "step": 10116
    },
    {
      "epoch": 0.2915057915057915,
      "grad_norm": 0.8193445205688477,
      "learning_rate": 8.311554523561362e-06,
      "loss": 0.7319,
      "step": 10117
    },
    {
      "epoch": 0.29153460496744077,
      "grad_norm": 0.8421133756637573,
      "learning_rate": 8.31120491125065e-06,
      "loss": 0.7631,
      "step": 10118
    },
    {
      "epoch": 0.29156341842909006,
      "grad_norm": 0.8212715983390808,
      "learning_rate": 8.310855270102653e-06,
      "loss": 0.7468,
      "step": 10119
    },
    {
      "epoch": 0.29159223189073935,
      "grad_norm": 0.8318383097648621,
      "learning_rate": 8.310505600120417e-06,
      "loss": 0.7476,
      "step": 10120
    },
    {
      "epoch": 0.29162104535238864,
      "grad_norm": 0.8366042375564575,
      "learning_rate": 8.310155901306989e-06,
      "loss": 0.7429,
      "step": 10121
    },
    {
      "epoch": 0.29164985881403793,
      "grad_norm": 0.8391713500022888,
      "learning_rate": 8.30980617366541e-06,
      "loss": 0.7393,
      "step": 10122
    },
    {
      "epoch": 0.2916786722756872,
      "grad_norm": 0.8238121867179871,
      "learning_rate": 8.309456417198726e-06,
      "loss": 0.7502,
      "step": 10123
    },
    {
      "epoch": 0.29170748573733646,
      "grad_norm": 0.836247980594635,
      "learning_rate": 8.309106631909987e-06,
      "loss": 0.7452,
      "step": 10124
    },
    {
      "epoch": 0.29173629919898575,
      "grad_norm": 0.8448177576065063,
      "learning_rate": 8.308756817802235e-06,
      "loss": 0.7325,
      "step": 10125
    },
    {
      "epoch": 0.29176511266063504,
      "grad_norm": 0.8467022776603699,
      "learning_rate": 8.30840697487852e-06,
      "loss": 0.7661,
      "step": 10126
    },
    {
      "epoch": 0.29179392612228433,
      "grad_norm": 0.8312366008758545,
      "learning_rate": 8.308057103141889e-06,
      "loss": 0.7487,
      "step": 10127
    },
    {
      "epoch": 0.2918227395839336,
      "grad_norm": 0.8840606808662415,
      "learning_rate": 8.307707202595386e-06,
      "loss": 0.7584,
      "step": 10128
    },
    {
      "epoch": 0.2918515530455829,
      "grad_norm": 0.8308945298194885,
      "learning_rate": 8.30735727324206e-06,
      "loss": 0.7593,
      "step": 10129
    },
    {
      "epoch": 0.2918803665072322,
      "grad_norm": 0.8222327828407288,
      "learning_rate": 8.307007315084958e-06,
      "loss": 0.7518,
      "step": 10130
    },
    {
      "epoch": 0.29190917996888144,
      "grad_norm": 0.8222982287406921,
      "learning_rate": 8.306657328127128e-06,
      "loss": 0.7857,
      "step": 10131
    },
    {
      "epoch": 0.29193799343053073,
      "grad_norm": 0.8663648366928101,
      "learning_rate": 8.306307312371617e-06,
      "loss": 0.7671,
      "step": 10132
    },
    {
      "epoch": 0.29196680689218,
      "grad_norm": 1.0819799900054932,
      "learning_rate": 8.305957267821476e-06,
      "loss": 0.7808,
      "step": 10133
    },
    {
      "epoch": 0.2919956203538293,
      "grad_norm": 0.8264220952987671,
      "learning_rate": 8.305607194479752e-06,
      "loss": 0.7639,
      "step": 10134
    },
    {
      "epoch": 0.2920244338154786,
      "grad_norm": 0.8408275842666626,
      "learning_rate": 8.305257092349494e-06,
      "loss": 0.7586,
      "step": 10135
    },
    {
      "epoch": 0.2920532472771279,
      "grad_norm": 0.8541795611381531,
      "learning_rate": 8.30490696143375e-06,
      "loss": 0.7357,
      "step": 10136
    },
    {
      "epoch": 0.2920820607387772,
      "grad_norm": 0.8135554790496826,
      "learning_rate": 8.30455680173557e-06,
      "loss": 0.7619,
      "step": 10137
    },
    {
      "epoch": 0.2921108742004264,
      "grad_norm": 0.8253082633018494,
      "learning_rate": 8.304206613258004e-06,
      "loss": 0.7496,
      "step": 10138
    },
    {
      "epoch": 0.2921396876620757,
      "grad_norm": 0.859679102897644,
      "learning_rate": 8.303856396004101e-06,
      "loss": 0.7522,
      "step": 10139
    },
    {
      "epoch": 0.292168501123725,
      "grad_norm": 0.8289759755134583,
      "learning_rate": 8.303506149976911e-06,
      "loss": 0.7439,
      "step": 10140
    },
    {
      "epoch": 0.2921973145853743,
      "grad_norm": 0.8081464767456055,
      "learning_rate": 8.303155875179487e-06,
      "loss": 0.7345,
      "step": 10141
    },
    {
      "epoch": 0.2922261280470236,
      "grad_norm": 0.8861159086227417,
      "learning_rate": 8.302805571614875e-06,
      "loss": 0.7469,
      "step": 10142
    },
    {
      "epoch": 0.29225494150867287,
      "grad_norm": 0.8161733746528625,
      "learning_rate": 8.302455239286129e-06,
      "loss": 0.7619,
      "step": 10143
    },
    {
      "epoch": 0.2922837549703221,
      "grad_norm": 0.8473708629608154,
      "learning_rate": 8.302104878196298e-06,
      "loss": 0.7843,
      "step": 10144
    },
    {
      "epoch": 0.2923125684319714,
      "grad_norm": 0.8419815897941589,
      "learning_rate": 8.301754488348436e-06,
      "loss": 0.7453,
      "step": 10145
    },
    {
      "epoch": 0.2923413818936207,
      "grad_norm": 0.8364497423171997,
      "learning_rate": 8.301404069745592e-06,
      "loss": 0.7448,
      "step": 10146
    },
    {
      "epoch": 0.29237019535527,
      "grad_norm": 0.8214044570922852,
      "learning_rate": 8.30105362239082e-06,
      "loss": 0.7347,
      "step": 10147
    },
    {
      "epoch": 0.29239900881691927,
      "grad_norm": 0.823218047618866,
      "learning_rate": 8.30070314628717e-06,
      "loss": 0.7597,
      "step": 10148
    },
    {
      "epoch": 0.29242782227856856,
      "grad_norm": 0.8651007413864136,
      "learning_rate": 8.300352641437693e-06,
      "loss": 0.7554,
      "step": 10149
    },
    {
      "epoch": 0.29245663574021785,
      "grad_norm": 0.8643461465835571,
      "learning_rate": 8.300002107845447e-06,
      "loss": 0.7376,
      "step": 10150
    },
    {
      "epoch": 0.2924854492018671,
      "grad_norm": 1.2559268474578857,
      "learning_rate": 8.29965154551348e-06,
      "loss": 0.7365,
      "step": 10151
    },
    {
      "epoch": 0.2925142626635164,
      "grad_norm": 0.8060210347175598,
      "learning_rate": 8.299300954444845e-06,
      "loss": 0.765,
      "step": 10152
    },
    {
      "epoch": 0.29254307612516567,
      "grad_norm": 0.8635280132293701,
      "learning_rate": 8.2989503346426e-06,
      "loss": 0.7573,
      "step": 10153
    },
    {
      "epoch": 0.29257188958681496,
      "grad_norm": 0.86842942237854,
      "learning_rate": 8.298599686109794e-06,
      "loss": 0.7364,
      "step": 10154
    },
    {
      "epoch": 0.29260070304846425,
      "grad_norm": 0.859062671661377,
      "learning_rate": 8.298249008849479e-06,
      "loss": 0.7635,
      "step": 10155
    },
    {
      "epoch": 0.29262951651011354,
      "grad_norm": 0.8338059186935425,
      "learning_rate": 8.297898302864715e-06,
      "loss": 0.7383,
      "step": 10156
    },
    {
      "epoch": 0.29265832997176283,
      "grad_norm": 0.8406912684440613,
      "learning_rate": 8.297547568158556e-06,
      "loss": 0.7737,
      "step": 10157
    },
    {
      "epoch": 0.29268714343341207,
      "grad_norm": 0.8306726813316345,
      "learning_rate": 8.29719680473405e-06,
      "loss": 0.7598,
      "step": 10158
    },
    {
      "epoch": 0.29271595689506136,
      "grad_norm": 0.8186433911323547,
      "learning_rate": 8.296846012594258e-06,
      "loss": 0.7508,
      "step": 10159
    },
    {
      "epoch": 0.29274477035671065,
      "grad_norm": 0.8517801761627197,
      "learning_rate": 8.29649519174223e-06,
      "loss": 0.7394,
      "step": 10160
    },
    {
      "epoch": 0.29277358381835994,
      "grad_norm": 0.8460848331451416,
      "learning_rate": 8.296144342181028e-06,
      "loss": 0.7535,
      "step": 10161
    },
    {
      "epoch": 0.29280239728000923,
      "grad_norm": 0.8505507707595825,
      "learning_rate": 8.295793463913701e-06,
      "loss": 0.7554,
      "step": 10162
    },
    {
      "epoch": 0.2928312107416585,
      "grad_norm": 0.7878891229629517,
      "learning_rate": 8.295442556943308e-06,
      "loss": 0.7572,
      "step": 10163
    },
    {
      "epoch": 0.2928600242033078,
      "grad_norm": 0.8201469779014587,
      "learning_rate": 8.295091621272905e-06,
      "loss": 0.7335,
      "step": 10164
    },
    {
      "epoch": 0.29288883766495705,
      "grad_norm": 0.8453593254089355,
      "learning_rate": 8.294740656905548e-06,
      "loss": 0.764,
      "step": 10165
    },
    {
      "epoch": 0.29291765112660634,
      "grad_norm": 0.8387647271156311,
      "learning_rate": 8.294389663844293e-06,
      "loss": 0.7722,
      "step": 10166
    },
    {
      "epoch": 0.29294646458825563,
      "grad_norm": 0.8396385908126831,
      "learning_rate": 8.294038642092197e-06,
      "loss": 0.7779,
      "step": 10167
    },
    {
      "epoch": 0.2929752780499049,
      "grad_norm": 0.8195225596427917,
      "learning_rate": 8.293687591652318e-06,
      "loss": 0.7704,
      "step": 10168
    },
    {
      "epoch": 0.2930040915115542,
      "grad_norm": 0.8923097848892212,
      "learning_rate": 8.293336512527713e-06,
      "loss": 0.753,
      "step": 10169
    },
    {
      "epoch": 0.2930329049732035,
      "grad_norm": 0.8254028558731079,
      "learning_rate": 8.292985404721438e-06,
      "loss": 0.7549,
      "step": 10170
    },
    {
      "epoch": 0.29306171843485274,
      "grad_norm": 0.8631695508956909,
      "learning_rate": 8.292634268236553e-06,
      "loss": 0.7589,
      "step": 10171
    },
    {
      "epoch": 0.293090531896502,
      "grad_norm": 0.8311550617218018,
      "learning_rate": 8.292283103076112e-06,
      "loss": 0.7389,
      "step": 10172
    },
    {
      "epoch": 0.2931193453581513,
      "grad_norm": 0.8771064281463623,
      "learning_rate": 8.29193190924318e-06,
      "loss": 0.7472,
      "step": 10173
    },
    {
      "epoch": 0.2931481588198006,
      "grad_norm": 0.8316687345504761,
      "learning_rate": 8.29158068674081e-06,
      "loss": 0.7564,
      "step": 10174
    },
    {
      "epoch": 0.2931769722814499,
      "grad_norm": 0.9049981236457825,
      "learning_rate": 8.291229435572064e-06,
      "loss": 0.7509,
      "step": 10175
    },
    {
      "epoch": 0.2932057857430992,
      "grad_norm": 0.8856701850891113,
      "learning_rate": 8.29087815574e-06,
      "loss": 0.755,
      "step": 10176
    },
    {
      "epoch": 0.2932345992047485,
      "grad_norm": 0.8752689361572266,
      "learning_rate": 8.290526847247673e-06,
      "loss": 0.7653,
      "step": 10177
    },
    {
      "epoch": 0.2932634126663977,
      "grad_norm": 0.8532607555389404,
      "learning_rate": 8.29017551009815e-06,
      "loss": 0.761,
      "step": 10178
    },
    {
      "epoch": 0.293292226128047,
      "grad_norm": 0.8265278935432434,
      "learning_rate": 8.289824144294486e-06,
      "loss": 0.7636,
      "step": 10179
    },
    {
      "epoch": 0.2933210395896963,
      "grad_norm": 0.8524651527404785,
      "learning_rate": 8.289472749839743e-06,
      "loss": 0.7505,
      "step": 10180
    },
    {
      "epoch": 0.2933498530513456,
      "grad_norm": 0.8305153250694275,
      "learning_rate": 8.28912132673698e-06,
      "loss": 0.7458,
      "step": 10181
    },
    {
      "epoch": 0.2933786665129949,
      "grad_norm": 0.8582872152328491,
      "learning_rate": 8.288769874989257e-06,
      "loss": 0.7582,
      "step": 10182
    },
    {
      "epoch": 0.29340747997464417,
      "grad_norm": 0.821124255657196,
      "learning_rate": 8.288418394599637e-06,
      "loss": 0.7548,
      "step": 10183
    },
    {
      "epoch": 0.29343629343629346,
      "grad_norm": 0.833588182926178,
      "learning_rate": 8.288066885571183e-06,
      "loss": 0.7493,
      "step": 10184
    },
    {
      "epoch": 0.2934651068979427,
      "grad_norm": 0.8266794681549072,
      "learning_rate": 8.287715347906948e-06,
      "loss": 0.7642,
      "step": 10185
    },
    {
      "epoch": 0.293493920359592,
      "grad_norm": 0.8163244128227234,
      "learning_rate": 8.287363781610003e-06,
      "loss": 0.7596,
      "step": 10186
    },
    {
      "epoch": 0.2935227338212413,
      "grad_norm": 0.8108016848564148,
      "learning_rate": 8.287012186683404e-06,
      "loss": 0.7115,
      "step": 10187
    },
    {
      "epoch": 0.29355154728289057,
      "grad_norm": 0.7995719313621521,
      "learning_rate": 8.286660563130213e-06,
      "loss": 0.7556,
      "step": 10188
    },
    {
      "epoch": 0.29358036074453986,
      "grad_norm": 0.8237337470054626,
      "learning_rate": 8.286308910953497e-06,
      "loss": 0.7681,
      "step": 10189
    },
    {
      "epoch": 0.29360917420618915,
      "grad_norm": 0.8327677249908447,
      "learning_rate": 8.285957230156315e-06,
      "loss": 0.7757,
      "step": 10190
    },
    {
      "epoch": 0.29363798766783844,
      "grad_norm": 0.840946614742279,
      "learning_rate": 8.285605520741729e-06,
      "loss": 0.7671,
      "step": 10191
    },
    {
      "epoch": 0.2936668011294877,
      "grad_norm": 0.80652916431427,
      "learning_rate": 8.285253782712804e-06,
      "loss": 0.7637,
      "step": 10192
    },
    {
      "epoch": 0.29369561459113697,
      "grad_norm": 0.8249050974845886,
      "learning_rate": 8.2849020160726e-06,
      "loss": 0.765,
      "step": 10193
    },
    {
      "epoch": 0.29372442805278626,
      "grad_norm": 0.8064612746238708,
      "learning_rate": 8.284550220824186e-06,
      "loss": 0.7389,
      "step": 10194
    },
    {
      "epoch": 0.29375324151443555,
      "grad_norm": 0.8229057192802429,
      "learning_rate": 8.284198396970624e-06,
      "loss": 0.7666,
      "step": 10195
    },
    {
      "epoch": 0.29378205497608484,
      "grad_norm": 0.8468176126480103,
      "learning_rate": 8.283846544514975e-06,
      "loss": 0.7454,
      "step": 10196
    },
    {
      "epoch": 0.29381086843773413,
      "grad_norm": 0.8145420551300049,
      "learning_rate": 8.283494663460306e-06,
      "loss": 0.7761,
      "step": 10197
    },
    {
      "epoch": 0.29383968189938336,
      "grad_norm": 0.778091311454773,
      "learning_rate": 8.283142753809679e-06,
      "loss": 0.7453,
      "step": 10198
    },
    {
      "epoch": 0.29386849536103266,
      "grad_norm": 0.8363803029060364,
      "learning_rate": 8.282790815566162e-06,
      "loss": 0.7709,
      "step": 10199
    },
    {
      "epoch": 0.29389730882268195,
      "grad_norm": 0.8043680787086487,
      "learning_rate": 8.282438848732818e-06,
      "loss": 0.7643,
      "step": 10200
    },
    {
      "epoch": 0.29392612228433124,
      "grad_norm": 0.8610881567001343,
      "learning_rate": 8.282086853312714e-06,
      "loss": 0.7476,
      "step": 10201
    },
    {
      "epoch": 0.29395493574598053,
      "grad_norm": 0.8801898956298828,
      "learning_rate": 8.281734829308914e-06,
      "loss": 0.7952,
      "step": 10202
    },
    {
      "epoch": 0.2939837492076298,
      "grad_norm": 0.820689857006073,
      "learning_rate": 8.281382776724484e-06,
      "loss": 0.7448,
      "step": 10203
    },
    {
      "epoch": 0.2940125626692791,
      "grad_norm": 0.8195933699607849,
      "learning_rate": 8.28103069556249e-06,
      "loss": 0.7486,
      "step": 10204
    },
    {
      "epoch": 0.29404137613092834,
      "grad_norm": 0.8972507119178772,
      "learning_rate": 8.280678585826e-06,
      "loss": 0.7636,
      "step": 10205
    },
    {
      "epoch": 0.29407018959257764,
      "grad_norm": 0.817145586013794,
      "learning_rate": 8.280326447518076e-06,
      "loss": 0.7451,
      "step": 10206
    },
    {
      "epoch": 0.2940990030542269,
      "grad_norm": 0.8131103515625,
      "learning_rate": 8.279974280641791e-06,
      "loss": 0.7568,
      "step": 10207
    },
    {
      "epoch": 0.2941278165158762,
      "grad_norm": 0.832417368888855,
      "learning_rate": 8.279622085200208e-06,
      "loss": 0.7355,
      "step": 10208
    },
    {
      "epoch": 0.2941566299775255,
      "grad_norm": 0.8118257522583008,
      "learning_rate": 8.279269861196393e-06,
      "loss": 0.7488,
      "step": 10209
    },
    {
      "epoch": 0.2941854434391748,
      "grad_norm": 0.8211923241615295,
      "learning_rate": 8.278917608633416e-06,
      "loss": 0.7469,
      "step": 10210
    },
    {
      "epoch": 0.2942142569008241,
      "grad_norm": 0.8057779669761658,
      "learning_rate": 8.278565327514345e-06,
      "loss": 0.7552,
      "step": 10211
    },
    {
      "epoch": 0.2942430703624733,
      "grad_norm": 0.8391096591949463,
      "learning_rate": 8.278213017842246e-06,
      "loss": 0.7668,
      "step": 10212
    },
    {
      "epoch": 0.2942718838241226,
      "grad_norm": 0.905119776725769,
      "learning_rate": 8.277860679620192e-06,
      "loss": 0.7929,
      "step": 10213
    },
    {
      "epoch": 0.2943006972857719,
      "grad_norm": 0.8201538920402527,
      "learning_rate": 8.277508312851247e-06,
      "loss": 0.7889,
      "step": 10214
    },
    {
      "epoch": 0.2943295107474212,
      "grad_norm": 0.8158507347106934,
      "learning_rate": 8.27715591753848e-06,
      "loss": 0.7792,
      "step": 10215
    },
    {
      "epoch": 0.2943583242090705,
      "grad_norm": 0.8340404629707336,
      "learning_rate": 8.27680349368496e-06,
      "loss": 0.7485,
      "step": 10216
    },
    {
      "epoch": 0.2943871376707198,
      "grad_norm": 0.8355147242546082,
      "learning_rate": 8.276451041293758e-06,
      "loss": 0.7511,
      "step": 10217
    },
    {
      "epoch": 0.29441595113236907,
      "grad_norm": 0.9235586524009705,
      "learning_rate": 8.276098560367944e-06,
      "loss": 0.7647,
      "step": 10218
    },
    {
      "epoch": 0.2944447645940183,
      "grad_norm": 0.8016037344932556,
      "learning_rate": 8.275746050910585e-06,
      "loss": 0.7526,
      "step": 10219
    },
    {
      "epoch": 0.2944735780556676,
      "grad_norm": 0.8799397349357605,
      "learning_rate": 8.275393512924753e-06,
      "loss": 0.7243,
      "step": 10220
    },
    {
      "epoch": 0.2945023915173169,
      "grad_norm": 0.8748942613601685,
      "learning_rate": 8.275040946413518e-06,
      "loss": 0.7345,
      "step": 10221
    },
    {
      "epoch": 0.2945312049789662,
      "grad_norm": 0.8254528045654297,
      "learning_rate": 8.274688351379949e-06,
      "loss": 0.7629,
      "step": 10222
    },
    {
      "epoch": 0.29456001844061547,
      "grad_norm": 0.8370658159255981,
      "learning_rate": 8.274335727827118e-06,
      "loss": 0.7846,
      "step": 10223
    },
    {
      "epoch": 0.29458883190226476,
      "grad_norm": 0.8991815447807312,
      "learning_rate": 8.2739830757581e-06,
      "loss": 0.7713,
      "step": 10224
    },
    {
      "epoch": 0.294617645363914,
      "grad_norm": 0.8333629369735718,
      "learning_rate": 8.273630395175958e-06,
      "loss": 0.7329,
      "step": 10225
    },
    {
      "epoch": 0.2946464588255633,
      "grad_norm": 0.8256475925445557,
      "learning_rate": 8.273277686083768e-06,
      "loss": 0.7628,
      "step": 10226
    },
    {
      "epoch": 0.2946752722872126,
      "grad_norm": 0.8876357674598694,
      "learning_rate": 8.272924948484601e-06,
      "loss": 0.766,
      "step": 10227
    },
    {
      "epoch": 0.29470408574886187,
      "grad_norm": 0.8915969133377075,
      "learning_rate": 8.272572182381531e-06,
      "loss": 0.7636,
      "step": 10228
    },
    {
      "epoch": 0.29473289921051116,
      "grad_norm": 0.8379279971122742,
      "learning_rate": 8.272219387777628e-06,
      "loss": 0.7702,
      "step": 10229
    },
    {
      "epoch": 0.29476171267216045,
      "grad_norm": 0.7979731559753418,
      "learning_rate": 8.271866564675966e-06,
      "loss": 0.749,
      "step": 10230
    },
    {
      "epoch": 0.29479052613380974,
      "grad_norm": 0.8218042254447937,
      "learning_rate": 8.271513713079616e-06,
      "loss": 0.7596,
      "step": 10231
    },
    {
      "epoch": 0.294819339595459,
      "grad_norm": 0.8693106770515442,
      "learning_rate": 8.271160832991652e-06,
      "loss": 0.7541,
      "step": 10232
    },
    {
      "epoch": 0.29484815305710826,
      "grad_norm": 0.8207653164863586,
      "learning_rate": 8.270807924415144e-06,
      "loss": 0.753,
      "step": 10233
    },
    {
      "epoch": 0.29487696651875756,
      "grad_norm": 0.8446674346923828,
      "learning_rate": 8.270454987353173e-06,
      "loss": 0.7875,
      "step": 10234
    },
    {
      "epoch": 0.29490577998040685,
      "grad_norm": 0.81545090675354,
      "learning_rate": 8.270102021808807e-06,
      "loss": 0.7737,
      "step": 10235
    },
    {
      "epoch": 0.29493459344205614,
      "grad_norm": 0.8257784247398376,
      "learning_rate": 8.269749027785122e-06,
      "loss": 0.7542,
      "step": 10236
    },
    {
      "epoch": 0.2949634069037054,
      "grad_norm": 0.8149206638336182,
      "learning_rate": 8.26939600528519e-06,
      "loss": 0.7717,
      "step": 10237
    },
    {
      "epoch": 0.2949922203653547,
      "grad_norm": 0.8528736233711243,
      "learning_rate": 8.269042954312089e-06,
      "loss": 0.7802,
      "step": 10238
    },
    {
      "epoch": 0.29502103382700395,
      "grad_norm": 0.8180153369903564,
      "learning_rate": 8.26868987486889e-06,
      "loss": 0.7695,
      "step": 10239
    },
    {
      "epoch": 0.29504984728865324,
      "grad_norm": 0.8209069967269897,
      "learning_rate": 8.268336766958671e-06,
      "loss": 0.7614,
      "step": 10240
    },
    {
      "epoch": 0.29507866075030253,
      "grad_norm": 0.8177564144134521,
      "learning_rate": 8.267983630584504e-06,
      "loss": 0.7541,
      "step": 10241
    },
    {
      "epoch": 0.2951074742119518,
      "grad_norm": 0.8262794017791748,
      "learning_rate": 8.267630465749468e-06,
      "loss": 0.7714,
      "step": 10242
    },
    {
      "epoch": 0.2951362876736011,
      "grad_norm": 0.8257580995559692,
      "learning_rate": 8.267277272456638e-06,
      "loss": 0.734,
      "step": 10243
    },
    {
      "epoch": 0.2951651011352504,
      "grad_norm": 0.8311225175857544,
      "learning_rate": 8.266924050709087e-06,
      "loss": 0.765,
      "step": 10244
    },
    {
      "epoch": 0.2951939145968997,
      "grad_norm": 0.8092928528785706,
      "learning_rate": 8.266570800509892e-06,
      "loss": 0.7346,
      "step": 10245
    },
    {
      "epoch": 0.29522272805854893,
      "grad_norm": 0.8285135626792908,
      "learning_rate": 8.266217521862134e-06,
      "loss": 0.7731,
      "step": 10246
    },
    {
      "epoch": 0.2952515415201982,
      "grad_norm": 0.8061242699623108,
      "learning_rate": 8.265864214768883e-06,
      "loss": 0.7583,
      "step": 10247
    },
    {
      "epoch": 0.2952803549818475,
      "grad_norm": 0.8364967703819275,
      "learning_rate": 8.265510879233223e-06,
      "loss": 0.7665,
      "step": 10248
    },
    {
      "epoch": 0.2953091684434968,
      "grad_norm": 0.822763979434967,
      "learning_rate": 8.265157515258227e-06,
      "loss": 0.7579,
      "step": 10249
    },
    {
      "epoch": 0.2953379819051461,
      "grad_norm": 0.8431181907653809,
      "learning_rate": 8.26480412284697e-06,
      "loss": 0.7576,
      "step": 10250
    },
    {
      "epoch": 0.2953667953667954,
      "grad_norm": 0.8335471153259277,
      "learning_rate": 8.264450702002536e-06,
      "loss": 0.7351,
      "step": 10251
    },
    {
      "epoch": 0.2953956088284446,
      "grad_norm": 0.8148729801177979,
      "learning_rate": 8.264097252727997e-06,
      "loss": 0.7414,
      "step": 10252
    },
    {
      "epoch": 0.2954244222900939,
      "grad_norm": 0.8420764803886414,
      "learning_rate": 8.263743775026435e-06,
      "loss": 0.7354,
      "step": 10253
    },
    {
      "epoch": 0.2954532357517432,
      "grad_norm": 0.8809353113174438,
      "learning_rate": 8.263390268900928e-06,
      "loss": 0.7632,
      "step": 10254
    },
    {
      "epoch": 0.2954820492133925,
      "grad_norm": 0.8202419281005859,
      "learning_rate": 8.263036734354554e-06,
      "loss": 0.7556,
      "step": 10255
    },
    {
      "epoch": 0.2955108626750418,
      "grad_norm": 0.8913553953170776,
      "learning_rate": 8.26268317139039e-06,
      "loss": 0.7704,
      "step": 10256
    },
    {
      "epoch": 0.2955396761366911,
      "grad_norm": 0.8550801277160645,
      "learning_rate": 8.26232958001152e-06,
      "loss": 0.7826,
      "step": 10257
    },
    {
      "epoch": 0.29556848959834037,
      "grad_norm": 0.8336955308914185,
      "learning_rate": 8.261975960221016e-06,
      "loss": 0.754,
      "step": 10258
    },
    {
      "epoch": 0.2955973030599896,
      "grad_norm": 0.8373169302940369,
      "learning_rate": 8.261622312021966e-06,
      "loss": 0.749,
      "step": 10259
    },
    {
      "epoch": 0.2956261165216389,
      "grad_norm": 0.8023114800453186,
      "learning_rate": 8.261268635417445e-06,
      "loss": 0.7301,
      "step": 10260
    },
    {
      "epoch": 0.2956549299832882,
      "grad_norm": 0.8701940774917603,
      "learning_rate": 8.260914930410535e-06,
      "loss": 0.7499,
      "step": 10261
    },
    {
      "epoch": 0.2956837434449375,
      "grad_norm": 0.8286354541778564,
      "learning_rate": 8.260561197004314e-06,
      "loss": 0.7515,
      "step": 10262
    },
    {
      "epoch": 0.29571255690658677,
      "grad_norm": 0.8778207302093506,
      "learning_rate": 8.260207435201866e-06,
      "loss": 0.7609,
      "step": 10263
    },
    {
      "epoch": 0.29574137036823606,
      "grad_norm": 0.8334822654724121,
      "learning_rate": 8.259853645006269e-06,
      "loss": 0.7439,
      "step": 10264
    },
    {
      "epoch": 0.29577018382988535,
      "grad_norm": 4.574763298034668,
      "learning_rate": 8.259499826420606e-06,
      "loss": 0.759,
      "step": 10265
    },
    {
      "epoch": 0.2957989972915346,
      "grad_norm": 0.8669148683547974,
      "learning_rate": 8.259145979447958e-06,
      "loss": 0.7842,
      "step": 10266
    },
    {
      "epoch": 0.2958278107531839,
      "grad_norm": 0.8286969661712646,
      "learning_rate": 8.258792104091407e-06,
      "loss": 0.7518,
      "step": 10267
    },
    {
      "epoch": 0.29585662421483316,
      "grad_norm": 0.8453723192214966,
      "learning_rate": 8.258438200354034e-06,
      "loss": 0.7514,
      "step": 10268
    },
    {
      "epoch": 0.29588543767648245,
      "grad_norm": 0.8314445614814758,
      "learning_rate": 8.25808426823892e-06,
      "loss": 0.7582,
      "step": 10269
    },
    {
      "epoch": 0.29591425113813175,
      "grad_norm": 0.8440044522285461,
      "learning_rate": 8.25773030774915e-06,
      "loss": 0.7629,
      "step": 10270
    },
    {
      "epoch": 0.29594306459978104,
      "grad_norm": 0.877219021320343,
      "learning_rate": 8.257376318887805e-06,
      "loss": 0.769,
      "step": 10271
    },
    {
      "epoch": 0.2959718780614303,
      "grad_norm": 0.8622033596038818,
      "learning_rate": 8.25702230165797e-06,
      "loss": 0.7432,
      "step": 10272
    },
    {
      "epoch": 0.29600069152307956,
      "grad_norm": 0.789928674697876,
      "learning_rate": 8.256668256062725e-06,
      "loss": 0.731,
      "step": 10273
    },
    {
      "epoch": 0.29602950498472885,
      "grad_norm": 0.8595908284187317,
      "learning_rate": 8.256314182105156e-06,
      "loss": 0.7905,
      "step": 10274
    },
    {
      "epoch": 0.29605831844637814,
      "grad_norm": 0.855340838432312,
      "learning_rate": 8.255960079788345e-06,
      "loss": 0.7654,
      "step": 10275
    },
    {
      "epoch": 0.29608713190802743,
      "grad_norm": 0.8615542650222778,
      "learning_rate": 8.255605949115376e-06,
      "loss": 0.7669,
      "step": 10276
    },
    {
      "epoch": 0.2961159453696767,
      "grad_norm": 0.8920412659645081,
      "learning_rate": 8.255251790089334e-06,
      "loss": 0.7542,
      "step": 10277
    },
    {
      "epoch": 0.296144758831326,
      "grad_norm": 0.8544942736625671,
      "learning_rate": 8.254897602713305e-06,
      "loss": 0.7748,
      "step": 10278
    },
    {
      "epoch": 0.29617357229297525,
      "grad_norm": 0.8634146451950073,
      "learning_rate": 8.254543386990368e-06,
      "loss": 0.7639,
      "step": 10279
    },
    {
      "epoch": 0.29620238575462454,
      "grad_norm": 0.854729413986206,
      "learning_rate": 8.254189142923614e-06,
      "loss": 0.7924,
      "step": 10280
    },
    {
      "epoch": 0.29623119921627383,
      "grad_norm": 0.8378247022628784,
      "learning_rate": 8.253834870516124e-06,
      "loss": 0.76,
      "step": 10281
    },
    {
      "epoch": 0.2962600126779231,
      "grad_norm": 0.8367696404457092,
      "learning_rate": 8.253480569770986e-06,
      "loss": 0.765,
      "step": 10282
    },
    {
      "epoch": 0.2962888261395724,
      "grad_norm": 0.8566484451293945,
      "learning_rate": 8.253126240691285e-06,
      "loss": 0.7648,
      "step": 10283
    },
    {
      "epoch": 0.2963176396012217,
      "grad_norm": 0.8554096221923828,
      "learning_rate": 8.252771883280107e-06,
      "loss": 0.7762,
      "step": 10284
    },
    {
      "epoch": 0.296346453062871,
      "grad_norm": 0.821898341178894,
      "learning_rate": 8.252417497540535e-06,
      "loss": 0.759,
      "step": 10285
    },
    {
      "epoch": 0.29637526652452023,
      "grad_norm": 0.8834109902381897,
      "learning_rate": 8.252063083475658e-06,
      "loss": 0.7856,
      "step": 10286
    },
    {
      "epoch": 0.2964040799861695,
      "grad_norm": 0.7929681539535522,
      "learning_rate": 8.251708641088565e-06,
      "loss": 0.7512,
      "step": 10287
    },
    {
      "epoch": 0.2964328934478188,
      "grad_norm": 0.8428348302841187,
      "learning_rate": 8.251354170382337e-06,
      "loss": 0.7666,
      "step": 10288
    },
    {
      "epoch": 0.2964617069094681,
      "grad_norm": 0.8215131163597107,
      "learning_rate": 8.250999671360066e-06,
      "loss": 0.7325,
      "step": 10289
    },
    {
      "epoch": 0.2964905203711174,
      "grad_norm": 0.8394251465797424,
      "learning_rate": 8.250645144024837e-06,
      "loss": 0.7657,
      "step": 10290
    },
    {
      "epoch": 0.2965193338327667,
      "grad_norm": 0.8286503553390503,
      "learning_rate": 8.25029058837974e-06,
      "loss": 0.7522,
      "step": 10291
    },
    {
      "epoch": 0.296548147294416,
      "grad_norm": 0.8123524188995361,
      "learning_rate": 8.249936004427859e-06,
      "loss": 0.7651,
      "step": 10292
    },
    {
      "epoch": 0.2965769607560652,
      "grad_norm": 0.8570563793182373,
      "learning_rate": 8.249581392172285e-06,
      "loss": 0.789,
      "step": 10293
    },
    {
      "epoch": 0.2966057742177145,
      "grad_norm": 0.8506864309310913,
      "learning_rate": 8.249226751616105e-06,
      "loss": 0.7478,
      "step": 10294
    },
    {
      "epoch": 0.2966345876793638,
      "grad_norm": 0.8235844373703003,
      "learning_rate": 8.24887208276241e-06,
      "loss": 0.7887,
      "step": 10295
    },
    {
      "epoch": 0.2966634011410131,
      "grad_norm": 0.8730305433273315,
      "learning_rate": 8.248517385614286e-06,
      "loss": 0.7811,
      "step": 10296
    },
    {
      "epoch": 0.2966922146026624,
      "grad_norm": 0.8330619931221008,
      "learning_rate": 8.248162660174824e-06,
      "loss": 0.7582,
      "step": 10297
    },
    {
      "epoch": 0.29672102806431166,
      "grad_norm": 0.8123430609703064,
      "learning_rate": 8.247807906447108e-06,
      "loss": 0.7672,
      "step": 10298
    },
    {
      "epoch": 0.29674984152596096,
      "grad_norm": 0.7835545539855957,
      "learning_rate": 8.247453124434234e-06,
      "loss": 0.7266,
      "step": 10299
    },
    {
      "epoch": 0.2967786549876102,
      "grad_norm": 0.8509670495986938,
      "learning_rate": 8.247098314139291e-06,
      "loss": 0.7557,
      "step": 10300
    },
    {
      "epoch": 0.2968074684492595,
      "grad_norm": 0.8276417851448059,
      "learning_rate": 8.246743475565367e-06,
      "loss": 0.7423,
      "step": 10301
    },
    {
      "epoch": 0.2968362819109088,
      "grad_norm": 0.7894048094749451,
      "learning_rate": 8.246388608715552e-06,
      "loss": 0.747,
      "step": 10302
    },
    {
      "epoch": 0.29686509537255806,
      "grad_norm": 0.8371739983558655,
      "learning_rate": 8.246033713592938e-06,
      "loss": 0.758,
      "step": 10303
    },
    {
      "epoch": 0.29689390883420735,
      "grad_norm": 0.8238900899887085,
      "learning_rate": 8.245678790200614e-06,
      "loss": 0.7529,
      "step": 10304
    },
    {
      "epoch": 0.29692272229585664,
      "grad_norm": 0.8489658236503601,
      "learning_rate": 8.245323838541674e-06,
      "loss": 0.7535,
      "step": 10305
    },
    {
      "epoch": 0.2969515357575059,
      "grad_norm": 0.8299725651741028,
      "learning_rate": 8.244968858619207e-06,
      "loss": 0.7757,
      "step": 10306
    },
    {
      "epoch": 0.29698034921915517,
      "grad_norm": 0.7960236668586731,
      "learning_rate": 8.244613850436303e-06,
      "loss": 0.7702,
      "step": 10307
    },
    {
      "epoch": 0.29700916268080446,
      "grad_norm": 0.8086405992507935,
      "learning_rate": 8.244258813996057e-06,
      "loss": 0.7518,
      "step": 10308
    },
    {
      "epoch": 0.29703797614245375,
      "grad_norm": 0.833763599395752,
      "learning_rate": 8.243903749301561e-06,
      "loss": 0.7602,
      "step": 10309
    },
    {
      "epoch": 0.29706678960410304,
      "grad_norm": 0.8249545693397522,
      "learning_rate": 8.243548656355907e-06,
      "loss": 0.7777,
      "step": 10310
    },
    {
      "epoch": 0.29709560306575233,
      "grad_norm": 0.8012622594833374,
      "learning_rate": 8.243193535162184e-06,
      "loss": 0.7519,
      "step": 10311
    },
    {
      "epoch": 0.2971244165274016,
      "grad_norm": 0.7987781167030334,
      "learning_rate": 8.242838385723489e-06,
      "loss": 0.7591,
      "step": 10312
    },
    {
      "epoch": 0.29715322998905086,
      "grad_norm": 0.8172581195831299,
      "learning_rate": 8.242483208042911e-06,
      "loss": 0.7408,
      "step": 10313
    },
    {
      "epoch": 0.29718204345070015,
      "grad_norm": 0.908581554889679,
      "learning_rate": 8.242128002123549e-06,
      "loss": 0.7722,
      "step": 10314
    },
    {
      "epoch": 0.29721085691234944,
      "grad_norm": 0.8138546943664551,
      "learning_rate": 8.24177276796849e-06,
      "loss": 0.7478,
      "step": 10315
    },
    {
      "epoch": 0.29723967037399873,
      "grad_norm": 0.8129280805587769,
      "learning_rate": 8.24141750558083e-06,
      "loss": 0.7559,
      "step": 10316
    },
    {
      "epoch": 0.297268483835648,
      "grad_norm": 0.7936309576034546,
      "learning_rate": 8.241062214963668e-06,
      "loss": 0.7411,
      "step": 10317
    },
    {
      "epoch": 0.2972972972972973,
      "grad_norm": 0.8187017440795898,
      "learning_rate": 8.240706896120093e-06,
      "loss": 0.7684,
      "step": 10318
    },
    {
      "epoch": 0.2973261107589466,
      "grad_norm": 0.842604398727417,
      "learning_rate": 8.240351549053196e-06,
      "loss": 0.752,
      "step": 10319
    },
    {
      "epoch": 0.29735492422059584,
      "grad_norm": 0.836192786693573,
      "learning_rate": 8.23999617376608e-06,
      "loss": 0.7424,
      "step": 10320
    },
    {
      "epoch": 0.29738373768224513,
      "grad_norm": 0.8075115084648132,
      "learning_rate": 8.239640770261836e-06,
      "loss": 0.7654,
      "step": 10321
    },
    {
      "epoch": 0.2974125511438944,
      "grad_norm": 0.8127316832542419,
      "learning_rate": 8.23928533854356e-06,
      "loss": 0.7554,
      "step": 10322
    },
    {
      "epoch": 0.2974413646055437,
      "grad_norm": 0.8660789728164673,
      "learning_rate": 8.238929878614345e-06,
      "loss": 0.7645,
      "step": 10323
    },
    {
      "epoch": 0.297470178067193,
      "grad_norm": 0.8084139227867126,
      "learning_rate": 8.23857439047729e-06,
      "loss": 0.7475,
      "step": 10324
    },
    {
      "epoch": 0.2974989915288423,
      "grad_norm": 0.8180336952209473,
      "learning_rate": 8.238218874135487e-06,
      "loss": 0.772,
      "step": 10325
    },
    {
      "epoch": 0.2975278049904916,
      "grad_norm": 0.807819664478302,
      "learning_rate": 8.237863329592038e-06,
      "loss": 0.7509,
      "step": 10326
    },
    {
      "epoch": 0.2975566184521408,
      "grad_norm": 0.8403026461601257,
      "learning_rate": 8.237507756850033e-06,
      "loss": 0.7579,
      "step": 10327
    },
    {
      "epoch": 0.2975854319137901,
      "grad_norm": 0.8158670663833618,
      "learning_rate": 8.237152155912574e-06,
      "loss": 0.7641,
      "step": 10328
    },
    {
      "epoch": 0.2976142453754394,
      "grad_norm": 0.7925803661346436,
      "learning_rate": 8.236796526782752e-06,
      "loss": 0.7634,
      "step": 10329
    },
    {
      "epoch": 0.2976430588370887,
      "grad_norm": 0.8844529986381531,
      "learning_rate": 8.236440869463671e-06,
      "loss": 0.7614,
      "step": 10330
    },
    {
      "epoch": 0.297671872298738,
      "grad_norm": 1.8992531299591064,
      "learning_rate": 8.236085183958427e-06,
      "loss": 0.7544,
      "step": 10331
    },
    {
      "epoch": 0.2977006857603873,
      "grad_norm": 0.8360555768013,
      "learning_rate": 8.235729470270113e-06,
      "loss": 0.7437,
      "step": 10332
    },
    {
      "epoch": 0.2977294992220365,
      "grad_norm": 0.8439463376998901,
      "learning_rate": 8.235373728401831e-06,
      "loss": 0.7537,
      "step": 10333
    },
    {
      "epoch": 0.2977583126836858,
      "grad_norm": 0.820537805557251,
      "learning_rate": 8.235017958356676e-06,
      "loss": 0.7593,
      "step": 10334
    },
    {
      "epoch": 0.2977871261453351,
      "grad_norm": 0.8176907896995544,
      "learning_rate": 8.23466216013775e-06,
      "loss": 0.7534,
      "step": 10335
    },
    {
      "epoch": 0.2978159396069844,
      "grad_norm": 0.8275842666625977,
      "learning_rate": 8.23430633374815e-06,
      "loss": 0.729,
      "step": 10336
    },
    {
      "epoch": 0.29784475306863367,
      "grad_norm": 0.8433172106742859,
      "learning_rate": 8.233950479190976e-06,
      "loss": 0.8035,
      "step": 10337
    },
    {
      "epoch": 0.29787356653028296,
      "grad_norm": 0.8124526143074036,
      "learning_rate": 8.233594596469325e-06,
      "loss": 0.7628,
      "step": 10338
    },
    {
      "epoch": 0.29790237999193225,
      "grad_norm": 0.8600448369979858,
      "learning_rate": 8.233238685586297e-06,
      "loss": 0.7664,
      "step": 10339
    },
    {
      "epoch": 0.2979311934535815,
      "grad_norm": 0.876675546169281,
      "learning_rate": 8.232882746544993e-06,
      "loss": 0.7616,
      "step": 10340
    },
    {
      "epoch": 0.2979600069152308,
      "grad_norm": 0.8418536186218262,
      "learning_rate": 8.232526779348514e-06,
      "loss": 0.7488,
      "step": 10341
    },
    {
      "epoch": 0.29798882037688007,
      "grad_norm": 0.794249951839447,
      "learning_rate": 8.232170783999956e-06,
      "loss": 0.7184,
      "step": 10342
    },
    {
      "epoch": 0.29801763383852936,
      "grad_norm": 0.8739024996757507,
      "learning_rate": 8.231814760502421e-06,
      "loss": 0.7741,
      "step": 10343
    },
    {
      "epoch": 0.29804644730017865,
      "grad_norm": 0.8182096481323242,
      "learning_rate": 8.23145870885901e-06,
      "loss": 0.7638,
      "step": 10344
    },
    {
      "epoch": 0.29807526076182794,
      "grad_norm": 0.8050069212913513,
      "learning_rate": 8.231102629072825e-06,
      "loss": 0.7583,
      "step": 10345
    },
    {
      "epoch": 0.29810407422347723,
      "grad_norm": 0.7975626587867737,
      "learning_rate": 8.230746521146963e-06,
      "loss": 0.765,
      "step": 10346
    },
    {
      "epoch": 0.29813288768512647,
      "grad_norm": 0.8405431509017944,
      "learning_rate": 8.230390385084532e-06,
      "loss": 0.7566,
      "step": 10347
    },
    {
      "epoch": 0.29816170114677576,
      "grad_norm": 0.803274929523468,
      "learning_rate": 8.230034220888629e-06,
      "loss": 0.7507,
      "step": 10348
    },
    {
      "epoch": 0.29819051460842505,
      "grad_norm": 0.8254438042640686,
      "learning_rate": 8.229678028562356e-06,
      "loss": 0.7354,
      "step": 10349
    },
    {
      "epoch": 0.29821932807007434,
      "grad_norm": 0.8078831434249878,
      "learning_rate": 8.229321808108817e-06,
      "loss": 0.7431,
      "step": 10350
    },
    {
      "epoch": 0.29824814153172363,
      "grad_norm": 0.826337993144989,
      "learning_rate": 8.228965559531113e-06,
      "loss": 0.7837,
      "step": 10351
    },
    {
      "epoch": 0.2982769549933729,
      "grad_norm": 0.832973837852478,
      "learning_rate": 8.228609282832345e-06,
      "loss": 0.7588,
      "step": 10352
    },
    {
      "epoch": 0.2983057684550222,
      "grad_norm": 0.7823649048805237,
      "learning_rate": 8.22825297801562e-06,
      "loss": 0.7485,
      "step": 10353
    },
    {
      "epoch": 0.29833458191667145,
      "grad_norm": 0.8054549098014832,
      "learning_rate": 8.227896645084037e-06,
      "loss": 0.7475,
      "step": 10354
    },
    {
      "epoch": 0.29836339537832074,
      "grad_norm": 0.8196922540664673,
      "learning_rate": 8.227540284040701e-06,
      "loss": 0.7336,
      "step": 10355
    },
    {
      "epoch": 0.29839220883997003,
      "grad_norm": 0.7814051508903503,
      "learning_rate": 8.227183894888716e-06,
      "loss": 0.7472,
      "step": 10356
    },
    {
      "epoch": 0.2984210223016193,
      "grad_norm": 0.7696799039840698,
      "learning_rate": 8.226827477631185e-06,
      "loss": 0.7278,
      "step": 10357
    },
    {
      "epoch": 0.2984498357632686,
      "grad_norm": 0.8538980484008789,
      "learning_rate": 8.226471032271212e-06,
      "loss": 0.7682,
      "step": 10358
    },
    {
      "epoch": 0.2984786492249179,
      "grad_norm": 0.7924761176109314,
      "learning_rate": 8.226114558811902e-06,
      "loss": 0.7394,
      "step": 10359
    },
    {
      "epoch": 0.29850746268656714,
      "grad_norm": 0.8399515151977539,
      "learning_rate": 8.22575805725636e-06,
      "loss": 0.7413,
      "step": 10360
    },
    {
      "epoch": 0.29853627614821643,
      "grad_norm": 0.8364174962043762,
      "learning_rate": 8.225401527607689e-06,
      "loss": 0.7599,
      "step": 10361
    },
    {
      "epoch": 0.2985650896098657,
      "grad_norm": 0.8840613961219788,
      "learning_rate": 8.225044969868995e-06,
      "loss": 0.7565,
      "step": 10362
    },
    {
      "epoch": 0.298593903071515,
      "grad_norm": 0.8254703879356384,
      "learning_rate": 8.224688384043383e-06,
      "loss": 0.7564,
      "step": 10363
    },
    {
      "epoch": 0.2986227165331643,
      "grad_norm": 0.8198409676551819,
      "learning_rate": 8.224331770133958e-06,
      "loss": 0.7362,
      "step": 10364
    },
    {
      "epoch": 0.2986515299948136,
      "grad_norm": 0.8334674835205078,
      "learning_rate": 8.223975128143828e-06,
      "loss": 0.7426,
      "step": 10365
    },
    {
      "epoch": 0.2986803434564629,
      "grad_norm": 0.8007071018218994,
      "learning_rate": 8.223618458076097e-06,
      "loss": 0.7668,
      "step": 10366
    },
    {
      "epoch": 0.2987091569181121,
      "grad_norm": 0.8012126684188843,
      "learning_rate": 8.223261759933872e-06,
      "loss": 0.7262,
      "step": 10367
    },
    {
      "epoch": 0.2987379703797614,
      "grad_norm": 0.8546137809753418,
      "learning_rate": 8.222905033720257e-06,
      "loss": 0.7511,
      "step": 10368
    },
    {
      "epoch": 0.2987667838414107,
      "grad_norm": 0.8468483090400696,
      "learning_rate": 8.222548279438364e-06,
      "loss": 0.7452,
      "step": 10369
    },
    {
      "epoch": 0.29879559730306,
      "grad_norm": 0.8274365663528442,
      "learning_rate": 8.222191497091295e-06,
      "loss": 0.7538,
      "step": 10370
    },
    {
      "epoch": 0.2988244107647093,
      "grad_norm": 0.852275550365448,
      "learning_rate": 8.22183468668216e-06,
      "loss": 0.7422,
      "step": 10371
    },
    {
      "epoch": 0.29885322422635857,
      "grad_norm": 0.8012613654136658,
      "learning_rate": 8.221477848214063e-06,
      "loss": 0.7337,
      "step": 10372
    },
    {
      "epoch": 0.29888203768800786,
      "grad_norm": 0.7960861325263977,
      "learning_rate": 8.221120981690116e-06,
      "loss": 0.7308,
      "step": 10373
    },
    {
      "epoch": 0.2989108511496571,
      "grad_norm": 0.7839807271957397,
      "learning_rate": 8.220764087113426e-06,
      "loss": 0.7616,
      "step": 10374
    },
    {
      "epoch": 0.2989396646113064,
      "grad_norm": 0.8122572302818298,
      "learning_rate": 8.220407164487101e-06,
      "loss": 0.7771,
      "step": 10375
    },
    {
      "epoch": 0.2989684780729557,
      "grad_norm": 0.856698751449585,
      "learning_rate": 8.220050213814246e-06,
      "loss": 0.8046,
      "step": 10376
    },
    {
      "epoch": 0.29899729153460497,
      "grad_norm": 0.8167356252670288,
      "learning_rate": 8.219693235097973e-06,
      "loss": 0.742,
      "step": 10377
    },
    {
      "epoch": 0.29902610499625426,
      "grad_norm": 0.8292368054389954,
      "learning_rate": 8.219336228341393e-06,
      "loss": 0.7762,
      "step": 10378
    },
    {
      "epoch": 0.29905491845790355,
      "grad_norm": 0.815496027469635,
      "learning_rate": 8.21897919354761e-06,
      "loss": 0.7725,
      "step": 10379
    },
    {
      "epoch": 0.29908373191955284,
      "grad_norm": 0.7748602032661438,
      "learning_rate": 8.218622130719738e-06,
      "loss": 0.7512,
      "step": 10380
    },
    {
      "epoch": 0.2991125453812021,
      "grad_norm": 0.8591178059577942,
      "learning_rate": 8.218265039860884e-06,
      "loss": 0.7626,
      "step": 10381
    },
    {
      "epoch": 0.29914135884285137,
      "grad_norm": 0.8683648705482483,
      "learning_rate": 8.217907920974159e-06,
      "loss": 0.7715,
      "step": 10382
    },
    {
      "epoch": 0.29917017230450066,
      "grad_norm": 0.81267911195755,
      "learning_rate": 8.217550774062672e-06,
      "loss": 0.7577,
      "step": 10383
    },
    {
      "epoch": 0.29919898576614995,
      "grad_norm": 0.8256098031997681,
      "learning_rate": 8.217193599129533e-06,
      "loss": 0.7526,
      "step": 10384
    },
    {
      "epoch": 0.29922779922779924,
      "grad_norm": 0.7852615714073181,
      "learning_rate": 8.216836396177856e-06,
      "loss": 0.7555,
      "step": 10385
    },
    {
      "epoch": 0.29925661268944853,
      "grad_norm": 0.8207653164863586,
      "learning_rate": 8.216479165210748e-06,
      "loss": 0.7573,
      "step": 10386
    },
    {
      "epoch": 0.29928542615109777,
      "grad_norm": 0.8213577270507812,
      "learning_rate": 8.216121906231324e-06,
      "loss": 0.7503,
      "step": 10387
    },
    {
      "epoch": 0.29931423961274706,
      "grad_norm": 0.8958659172058105,
      "learning_rate": 8.215764619242691e-06,
      "loss": 0.755,
      "step": 10388
    },
    {
      "epoch": 0.29934305307439635,
      "grad_norm": 0.8311572074890137,
      "learning_rate": 8.215407304247963e-06,
      "loss": 0.7538,
      "step": 10389
    },
    {
      "epoch": 0.29937186653604564,
      "grad_norm": 0.8835358023643494,
      "learning_rate": 8.215049961250252e-06,
      "loss": 0.7533,
      "step": 10390
    },
    {
      "epoch": 0.29940067999769493,
      "grad_norm": 0.7964689135551453,
      "learning_rate": 8.21469259025267e-06,
      "loss": 0.7388,
      "step": 10391
    },
    {
      "epoch": 0.2994294934593442,
      "grad_norm": 0.7949269413948059,
      "learning_rate": 8.21433519125833e-06,
      "loss": 0.7253,
      "step": 10392
    },
    {
      "epoch": 0.2994583069209935,
      "grad_norm": 0.8688085079193115,
      "learning_rate": 8.213977764270342e-06,
      "loss": 0.7685,
      "step": 10393
    },
    {
      "epoch": 0.29948712038264275,
      "grad_norm": 0.8063207268714905,
      "learning_rate": 8.21362030929182e-06,
      "loss": 0.7694,
      "step": 10394
    },
    {
      "epoch": 0.29951593384429204,
      "grad_norm": 0.7804579138755798,
      "learning_rate": 8.21326282632588e-06,
      "loss": 0.731,
      "step": 10395
    },
    {
      "epoch": 0.29954474730594133,
      "grad_norm": 0.7823679447174072,
      "learning_rate": 8.212905315375629e-06,
      "loss": 0.7603,
      "step": 10396
    },
    {
      "epoch": 0.2995735607675906,
      "grad_norm": 0.8223229050636292,
      "learning_rate": 8.212547776444188e-06,
      "loss": 0.7775,
      "step": 10397
    },
    {
      "epoch": 0.2996023742292399,
      "grad_norm": 0.8140460252761841,
      "learning_rate": 8.212190209534665e-06,
      "loss": 0.7543,
      "step": 10398
    },
    {
      "epoch": 0.2996311876908892,
      "grad_norm": 0.8137475252151489,
      "learning_rate": 8.211832614650179e-06,
      "loss": 0.7455,
      "step": 10399
    },
    {
      "epoch": 0.2996600011525385,
      "grad_norm": 0.8330724835395813,
      "learning_rate": 8.211474991793839e-06,
      "loss": 0.7683,
      "step": 10400
    },
    {
      "epoch": 0.2996888146141877,
      "grad_norm": 0.7828154563903809,
      "learning_rate": 8.211117340968762e-06,
      "loss": 0.731,
      "step": 10401
    },
    {
      "epoch": 0.299717628075837,
      "grad_norm": 0.8520700931549072,
      "learning_rate": 8.210759662178065e-06,
      "loss": 0.7363,
      "step": 10402
    },
    {
      "epoch": 0.2997464415374863,
      "grad_norm": 0.8673796057701111,
      "learning_rate": 8.21040195542486e-06,
      "loss": 0.7577,
      "step": 10403
    },
    {
      "epoch": 0.2997752549991356,
      "grad_norm": 0.8206591606140137,
      "learning_rate": 8.210044220712262e-06,
      "loss": 0.7655,
      "step": 10404
    },
    {
      "epoch": 0.2998040684607849,
      "grad_norm": 0.8347724080085754,
      "learning_rate": 8.209686458043389e-06,
      "loss": 0.7586,
      "step": 10405
    },
    {
      "epoch": 0.2998328819224342,
      "grad_norm": 0.7791216969490051,
      "learning_rate": 8.209328667421355e-06,
      "loss": 0.7398,
      "step": 10406
    },
    {
      "epoch": 0.29986169538408347,
      "grad_norm": 0.8403484225273132,
      "learning_rate": 8.208970848849277e-06,
      "loss": 0.7482,
      "step": 10407
    },
    {
      "epoch": 0.2998905088457327,
      "grad_norm": 0.8182687759399414,
      "learning_rate": 8.208613002330271e-06,
      "loss": 0.7483,
      "step": 10408
    },
    {
      "epoch": 0.299919322307382,
      "grad_norm": 0.8061969876289368,
      "learning_rate": 8.208255127867451e-06,
      "loss": 0.7677,
      "step": 10409
    },
    {
      "epoch": 0.2999481357690313,
      "grad_norm": 0.8058501482009888,
      "learning_rate": 8.20789722546394e-06,
      "loss": 0.7391,
      "step": 10410
    },
    {
      "epoch": 0.2999769492306806,
      "grad_norm": 0.8040851950645447,
      "learning_rate": 8.207539295122847e-06,
      "loss": 0.7557,
      "step": 10411
    },
    {
      "epoch": 0.30000576269232987,
      "grad_norm": 0.8474425673484802,
      "learning_rate": 8.207181336847295e-06,
      "loss": 0.7783,
      "step": 10412
    },
    {
      "epoch": 0.30003457615397916,
      "grad_norm": 0.8161541819572449,
      "learning_rate": 8.206823350640399e-06,
      "loss": 0.7433,
      "step": 10413
    },
    {
      "epoch": 0.3000633896156284,
      "grad_norm": 0.7740404605865479,
      "learning_rate": 8.206465336505276e-06,
      "loss": 0.7273,
      "step": 10414
    },
    {
      "epoch": 0.3000922030772777,
      "grad_norm": 0.819034218788147,
      "learning_rate": 8.206107294445048e-06,
      "loss": 0.7487,
      "step": 10415
    },
    {
      "epoch": 0.300121016538927,
      "grad_norm": 0.8271703124046326,
      "learning_rate": 8.205749224462827e-06,
      "loss": 0.7552,
      "step": 10416
    },
    {
      "epoch": 0.30014983000057627,
      "grad_norm": 0.786234974861145,
      "learning_rate": 8.205391126561737e-06,
      "loss": 0.7494,
      "step": 10417
    },
    {
      "epoch": 0.30017864346222556,
      "grad_norm": 0.8339555263519287,
      "learning_rate": 8.205033000744895e-06,
      "loss": 0.7661,
      "step": 10418
    },
    {
      "epoch": 0.30020745692387485,
      "grad_norm": 0.8369381427764893,
      "learning_rate": 8.204674847015419e-06,
      "loss": 0.7425,
      "step": 10419
    },
    {
      "epoch": 0.30023627038552414,
      "grad_norm": 0.8507848978042603,
      "learning_rate": 8.204316665376428e-06,
      "loss": 0.7404,
      "step": 10420
    },
    {
      "epoch": 0.3002650838471734,
      "grad_norm": 0.800215482711792,
      "learning_rate": 8.203958455831041e-06,
      "loss": 0.7635,
      "step": 10421
    },
    {
      "epoch": 0.30029389730882267,
      "grad_norm": 0.8641466498374939,
      "learning_rate": 8.20360021838238e-06,
      "loss": 0.754,
      "step": 10422
    },
    {
      "epoch": 0.30032271077047196,
      "grad_norm": 0.8531571626663208,
      "learning_rate": 8.203241953033564e-06,
      "loss": 0.745,
      "step": 10423
    },
    {
      "epoch": 0.30035152423212125,
      "grad_norm": 0.7772640585899353,
      "learning_rate": 8.202883659787711e-06,
      "loss": 0.7455,
      "step": 10424
    },
    {
      "epoch": 0.30038033769377054,
      "grad_norm": 0.7962172031402588,
      "learning_rate": 8.202525338647945e-06,
      "loss": 0.7518,
      "step": 10425
    },
    {
      "epoch": 0.30040915115541983,
      "grad_norm": 0.8308166265487671,
      "learning_rate": 8.202166989617384e-06,
      "loss": 0.7652,
      "step": 10426
    },
    {
      "epoch": 0.3004379646170691,
      "grad_norm": 0.8854640126228333,
      "learning_rate": 8.20180861269915e-06,
      "loss": 0.7463,
      "step": 10427
    },
    {
      "epoch": 0.30046677807871836,
      "grad_norm": 0.8204176425933838,
      "learning_rate": 8.201450207896362e-06,
      "loss": 0.7484,
      "step": 10428
    },
    {
      "epoch": 0.30049559154036765,
      "grad_norm": 0.8282819390296936,
      "learning_rate": 8.201091775212145e-06,
      "loss": 0.7514,
      "step": 10429
    },
    {
      "epoch": 0.30052440500201694,
      "grad_norm": 0.8669887781143188,
      "learning_rate": 8.200733314649615e-06,
      "loss": 0.7505,
      "step": 10430
    },
    {
      "epoch": 0.3005532184636662,
      "grad_norm": 0.8056787848472595,
      "learning_rate": 8.2003748262119e-06,
      "loss": 0.7321,
      "step": 10431
    },
    {
      "epoch": 0.3005820319253155,
      "grad_norm": 0.8462845087051392,
      "learning_rate": 8.20001630990212e-06,
      "loss": 0.7573,
      "step": 10432
    },
    {
      "epoch": 0.3006108453869648,
      "grad_norm": 0.8169171214103699,
      "learning_rate": 8.199657765723396e-06,
      "loss": 0.7402,
      "step": 10433
    },
    {
      "epoch": 0.3006396588486141,
      "grad_norm": 0.8516300916671753,
      "learning_rate": 8.19929919367885e-06,
      "loss": 0.7656,
      "step": 10434
    },
    {
      "epoch": 0.30066847231026333,
      "grad_norm": 0.835036039352417,
      "learning_rate": 8.198940593771609e-06,
      "loss": 0.7634,
      "step": 10435
    },
    {
      "epoch": 0.3006972857719126,
      "grad_norm": 0.8004975914955139,
      "learning_rate": 8.19858196600479e-06,
      "loss": 0.7699,
      "step": 10436
    },
    {
      "epoch": 0.3007260992335619,
      "grad_norm": 0.821940004825592,
      "learning_rate": 8.198223310381524e-06,
      "loss": 0.737,
      "step": 10437
    },
    {
      "epoch": 0.3007549126952112,
      "grad_norm": 0.8597584366798401,
      "learning_rate": 8.197864626904925e-06,
      "loss": 0.7803,
      "step": 10438
    },
    {
      "epoch": 0.3007837261568605,
      "grad_norm": 0.8030967712402344,
      "learning_rate": 8.197505915578126e-06,
      "loss": 0.7506,
      "step": 10439
    },
    {
      "epoch": 0.3008125396185098,
      "grad_norm": 0.8324010968208313,
      "learning_rate": 8.197147176404244e-06,
      "loss": 0.767,
      "step": 10440
    },
    {
      "epoch": 0.300841353080159,
      "grad_norm": 0.8188304901123047,
      "learning_rate": 8.196788409386407e-06,
      "loss": 0.7537,
      "step": 10441
    },
    {
      "epoch": 0.3008701665418083,
      "grad_norm": 0.7820459604263306,
      "learning_rate": 8.196429614527737e-06,
      "loss": 0.769,
      "step": 10442
    },
    {
      "epoch": 0.3008989800034576,
      "grad_norm": 0.7669125199317932,
      "learning_rate": 8.196070791831363e-06,
      "loss": 0.7391,
      "step": 10443
    },
    {
      "epoch": 0.3009277934651069,
      "grad_norm": 0.7857956886291504,
      "learning_rate": 8.195711941300406e-06,
      "loss": 0.7487,
      "step": 10444
    },
    {
      "epoch": 0.3009566069267562,
      "grad_norm": 0.8311423063278198,
      "learning_rate": 8.195353062937993e-06,
      "loss": 0.74,
      "step": 10445
    },
    {
      "epoch": 0.3009854203884055,
      "grad_norm": 0.8071320652961731,
      "learning_rate": 8.194994156747247e-06,
      "loss": 0.7652,
      "step": 10446
    },
    {
      "epoch": 0.30101423385005477,
      "grad_norm": 0.803146481513977,
      "learning_rate": 8.194635222731298e-06,
      "loss": 0.7543,
      "step": 10447
    },
    {
      "epoch": 0.301043047311704,
      "grad_norm": 0.8089199662208557,
      "learning_rate": 8.19427626089327e-06,
      "loss": 0.7488,
      "step": 10448
    },
    {
      "epoch": 0.3010718607733533,
      "grad_norm": 0.8531984090805054,
      "learning_rate": 8.193917271236287e-06,
      "loss": 0.744,
      "step": 10449
    },
    {
      "epoch": 0.3011006742350026,
      "grad_norm": 0.8132701516151428,
      "learning_rate": 8.193558253763479e-06,
      "loss": 0.7718,
      "step": 10450
    },
    {
      "epoch": 0.3011294876966519,
      "grad_norm": 0.7800969481468201,
      "learning_rate": 8.19319920847797e-06,
      "loss": 0.75,
      "step": 10451
    },
    {
      "epoch": 0.30115830115830117,
      "grad_norm": 0.8203515410423279,
      "learning_rate": 8.192840135382889e-06,
      "loss": 0.7385,
      "step": 10452
    },
    {
      "epoch": 0.30118711461995046,
      "grad_norm": 0.839316189289093,
      "learning_rate": 8.19248103448136e-06,
      "loss": 0.7566,
      "step": 10453
    },
    {
      "epoch": 0.30121592808159975,
      "grad_norm": 0.8451997637748718,
      "learning_rate": 8.192121905776513e-06,
      "loss": 0.7656,
      "step": 10454
    },
    {
      "epoch": 0.301244741543249,
      "grad_norm": 0.8358224630355835,
      "learning_rate": 8.191762749271477e-06,
      "loss": 0.7568,
      "step": 10455
    },
    {
      "epoch": 0.3012735550048983,
      "grad_norm": 0.7966776490211487,
      "learning_rate": 8.191403564969377e-06,
      "loss": 0.7515,
      "step": 10456
    },
    {
      "epoch": 0.30130236846654757,
      "grad_norm": 0.8461130857467651,
      "learning_rate": 8.191044352873342e-06,
      "loss": 0.7645,
      "step": 10457
    },
    {
      "epoch": 0.30133118192819686,
      "grad_norm": 0.8149253726005554,
      "learning_rate": 8.1906851129865e-06,
      "loss": 0.7599,
      "step": 10458
    },
    {
      "epoch": 0.30135999538984615,
      "grad_norm": 0.828729510307312,
      "learning_rate": 8.190325845311982e-06,
      "loss": 0.7755,
      "step": 10459
    },
    {
      "epoch": 0.30138880885149544,
      "grad_norm": 0.7833572030067444,
      "learning_rate": 8.189966549852913e-06,
      "loss": 0.7568,
      "step": 10460
    },
    {
      "epoch": 0.30141762231314473,
      "grad_norm": 0.7788742780685425,
      "learning_rate": 8.189607226612426e-06,
      "loss": 0.7389,
      "step": 10461
    },
    {
      "epoch": 0.30144643577479396,
      "grad_norm": 0.8429156541824341,
      "learning_rate": 8.189247875593646e-06,
      "loss": 0.7535,
      "step": 10462
    },
    {
      "epoch": 0.30147524923644325,
      "grad_norm": 0.8293814659118652,
      "learning_rate": 8.188888496799707e-06,
      "loss": 0.7426,
      "step": 10463
    },
    {
      "epoch": 0.30150406269809255,
      "grad_norm": 0.7936646342277527,
      "learning_rate": 8.188529090233738e-06,
      "loss": 0.757,
      "step": 10464
    },
    {
      "epoch": 0.30153287615974184,
      "grad_norm": 0.7810634970664978,
      "learning_rate": 8.188169655898866e-06,
      "loss": 0.7537,
      "step": 10465
    },
    {
      "epoch": 0.3015616896213911,
      "grad_norm": 0.8071728944778442,
      "learning_rate": 8.187810193798223e-06,
      "loss": 0.7731,
      "step": 10466
    },
    {
      "epoch": 0.3015905030830404,
      "grad_norm": 0.8131280541419983,
      "learning_rate": 8.187450703934941e-06,
      "loss": 0.734,
      "step": 10467
    },
    {
      "epoch": 0.30161931654468965,
      "grad_norm": 0.8234683275222778,
      "learning_rate": 8.187091186312151e-06,
      "loss": 0.7407,
      "step": 10468
    },
    {
      "epoch": 0.30164813000633894,
      "grad_norm": 0.8223254084587097,
      "learning_rate": 8.18673164093298e-06,
      "loss": 0.7465,
      "step": 10469
    },
    {
      "epoch": 0.30167694346798823,
      "grad_norm": 0.8047912120819092,
      "learning_rate": 8.186372067800563e-06,
      "loss": 0.7538,
      "step": 10470
    },
    {
      "epoch": 0.3017057569296375,
      "grad_norm": 0.8345626592636108,
      "learning_rate": 8.18601246691803e-06,
      "loss": 0.7512,
      "step": 10471
    },
    {
      "epoch": 0.3017345703912868,
      "grad_norm": 0.8604221343994141,
      "learning_rate": 8.185652838288513e-06,
      "loss": 0.7511,
      "step": 10472
    },
    {
      "epoch": 0.3017633838529361,
      "grad_norm": 0.8648780584335327,
      "learning_rate": 8.185293181915145e-06,
      "loss": 0.7786,
      "step": 10473
    },
    {
      "epoch": 0.3017921973145854,
      "grad_norm": 0.7960134744644165,
      "learning_rate": 8.18493349780106e-06,
      "loss": 0.7529,
      "step": 10474
    },
    {
      "epoch": 0.30182101077623463,
      "grad_norm": 0.8169814944267273,
      "learning_rate": 8.184573785949384e-06,
      "loss": 0.7629,
      "step": 10475
    },
    {
      "epoch": 0.3018498242378839,
      "grad_norm": 0.8221723437309265,
      "learning_rate": 8.184214046363255e-06,
      "loss": 0.7585,
      "step": 10476
    },
    {
      "epoch": 0.3018786376995332,
      "grad_norm": 0.8236778974533081,
      "learning_rate": 8.183854279045806e-06,
      "loss": 0.7392,
      "step": 10477
    },
    {
      "epoch": 0.3019074511611825,
      "grad_norm": 0.8428910970687866,
      "learning_rate": 8.183494484000168e-06,
      "loss": 0.759,
      "step": 10478
    },
    {
      "epoch": 0.3019362646228318,
      "grad_norm": 0.8537213206291199,
      "learning_rate": 8.183134661229475e-06,
      "loss": 0.7481,
      "step": 10479
    },
    {
      "epoch": 0.3019650780844811,
      "grad_norm": 0.7881990075111389,
      "learning_rate": 8.182774810736862e-06,
      "loss": 0.7062,
      "step": 10480
    },
    {
      "epoch": 0.3019938915461304,
      "grad_norm": 0.8046098351478577,
      "learning_rate": 8.182414932525461e-06,
      "loss": 0.7444,
      "step": 10481
    },
    {
      "epoch": 0.3020227050077796,
      "grad_norm": 0.7819213271141052,
      "learning_rate": 8.182055026598408e-06,
      "loss": 0.7584,
      "step": 10482
    },
    {
      "epoch": 0.3020515184694289,
      "grad_norm": 0.8617668747901917,
      "learning_rate": 8.181695092958835e-06,
      "loss": 0.7566,
      "step": 10483
    },
    {
      "epoch": 0.3020803319310782,
      "grad_norm": 0.8361926078796387,
      "learning_rate": 8.18133513160988e-06,
      "loss": 0.7322,
      "step": 10484
    },
    {
      "epoch": 0.3021091453927275,
      "grad_norm": 0.7925942540168762,
      "learning_rate": 8.180975142554676e-06,
      "loss": 0.7689,
      "step": 10485
    },
    {
      "epoch": 0.3021379588543768,
      "grad_norm": 0.8246526718139648,
      "learning_rate": 8.180615125796358e-06,
      "loss": 0.7723,
      "step": 10486
    },
    {
      "epoch": 0.30216677231602607,
      "grad_norm": 0.8235265612602234,
      "learning_rate": 8.18025508133806e-06,
      "loss": 0.7485,
      "step": 10487
    },
    {
      "epoch": 0.30219558577767536,
      "grad_norm": 0.8034875988960266,
      "learning_rate": 8.179895009182922e-06,
      "loss": 0.7782,
      "step": 10488
    },
    {
      "epoch": 0.3022243992393246,
      "grad_norm": 0.8446217775344849,
      "learning_rate": 8.179534909334077e-06,
      "loss": 0.7523,
      "step": 10489
    },
    {
      "epoch": 0.3022532127009739,
      "grad_norm": 0.8061464428901672,
      "learning_rate": 8.17917478179466e-06,
      "loss": 0.7415,
      "step": 10490
    },
    {
      "epoch": 0.3022820261626232,
      "grad_norm": 0.7877923250198364,
      "learning_rate": 8.178814626567808e-06,
      "loss": 0.7308,
      "step": 10491
    },
    {
      "epoch": 0.30231083962427246,
      "grad_norm": 0.8096185922622681,
      "learning_rate": 8.178454443656661e-06,
      "loss": 0.7532,
      "step": 10492
    },
    {
      "epoch": 0.30233965308592176,
      "grad_norm": 0.8377954959869385,
      "learning_rate": 8.17809423306435e-06,
      "loss": 0.7681,
      "step": 10493
    },
    {
      "epoch": 0.30236846654757105,
      "grad_norm": 0.8029561042785645,
      "learning_rate": 8.177733994794016e-06,
      "loss": 0.7425,
      "step": 10494
    },
    {
      "epoch": 0.3023972800092203,
      "grad_norm": 0.8537072539329529,
      "learning_rate": 8.177373728848797e-06,
      "loss": 0.7487,
      "step": 10495
    },
    {
      "epoch": 0.3024260934708696,
      "grad_norm": 0.830953061580658,
      "learning_rate": 8.177013435231829e-06,
      "loss": 0.7445,
      "step": 10496
    },
    {
      "epoch": 0.30245490693251886,
      "grad_norm": 0.816742479801178,
      "learning_rate": 8.176653113946249e-06,
      "loss": 0.7333,
      "step": 10497
    },
    {
      "epoch": 0.30248372039416815,
      "grad_norm": 0.8037758469581604,
      "learning_rate": 8.176292764995196e-06,
      "loss": 0.7343,
      "step": 10498
    },
    {
      "epoch": 0.30251253385581744,
      "grad_norm": 0.7847071886062622,
      "learning_rate": 8.175932388381808e-06,
      "loss": 0.7313,
      "step": 10499
    },
    {
      "epoch": 0.30254134731746674,
      "grad_norm": 0.8289444446563721,
      "learning_rate": 8.175571984109223e-06,
      "loss": 0.7728,
      "step": 10500
    },
    {
      "epoch": 0.302570160779116,
      "grad_norm": 0.7993621230125427,
      "learning_rate": 8.175211552180581e-06,
      "loss": 0.7423,
      "step": 10501
    },
    {
      "epoch": 0.30259897424076526,
      "grad_norm": 0.7953037619590759,
      "learning_rate": 8.17485109259902e-06,
      "loss": 0.7771,
      "step": 10502
    },
    {
      "epoch": 0.30262778770241455,
      "grad_norm": 0.7998183369636536,
      "learning_rate": 8.174490605367682e-06,
      "loss": 0.7275,
      "step": 10503
    },
    {
      "epoch": 0.30265660116406384,
      "grad_norm": 0.7893660068511963,
      "learning_rate": 8.174130090489701e-06,
      "loss": 0.7632,
      "step": 10504
    },
    {
      "epoch": 0.30268541462571313,
      "grad_norm": 0.7866781949996948,
      "learning_rate": 8.173769547968222e-06,
      "loss": 0.7516,
      "step": 10505
    },
    {
      "epoch": 0.3027142280873624,
      "grad_norm": 0.7979158163070679,
      "learning_rate": 8.173408977806382e-06,
      "loss": 0.7501,
      "step": 10506
    },
    {
      "epoch": 0.3027430415490117,
      "grad_norm": 0.7867904901504517,
      "learning_rate": 8.173048380007323e-06,
      "loss": 0.7414,
      "step": 10507
    },
    {
      "epoch": 0.302771855010661,
      "grad_norm": 0.7944717407226562,
      "learning_rate": 8.172687754574182e-06,
      "loss": 0.757,
      "step": 10508
    },
    {
      "epoch": 0.30280066847231024,
      "grad_norm": 0.827896773815155,
      "learning_rate": 8.172327101510104e-06,
      "loss": 0.7622,
      "step": 10509
    },
    {
      "epoch": 0.30282948193395953,
      "grad_norm": 0.8291407227516174,
      "learning_rate": 8.171966420818227e-06,
      "loss": 0.7512,
      "step": 10510
    },
    {
      "epoch": 0.3028582953956088,
      "grad_norm": 0.7703992128372192,
      "learning_rate": 8.171605712501696e-06,
      "loss": 0.7525,
      "step": 10511
    },
    {
      "epoch": 0.3028871088572581,
      "grad_norm": 0.8662728071212769,
      "learning_rate": 8.171244976563647e-06,
      "loss": 0.7522,
      "step": 10512
    },
    {
      "epoch": 0.3029159223189074,
      "grad_norm": 0.830669105052948,
      "learning_rate": 8.170884213007225e-06,
      "loss": 0.7535,
      "step": 10513
    },
    {
      "epoch": 0.3029447357805567,
      "grad_norm": 0.8109483122825623,
      "learning_rate": 8.170523421835573e-06,
      "loss": 0.7461,
      "step": 10514
    },
    {
      "epoch": 0.302973549242206,
      "grad_norm": 0.8299656510353088,
      "learning_rate": 8.170162603051828e-06,
      "loss": 0.7493,
      "step": 10515
    },
    {
      "epoch": 0.3030023627038552,
      "grad_norm": 0.8282266855239868,
      "learning_rate": 8.169801756659138e-06,
      "loss": 0.7644,
      "step": 10516
    },
    {
      "epoch": 0.3030311761655045,
      "grad_norm": 0.8298437595367432,
      "learning_rate": 8.169440882660645e-06,
      "loss": 0.7588,
      "step": 10517
    },
    {
      "epoch": 0.3030599896271538,
      "grad_norm": 0.8007273077964783,
      "learning_rate": 8.169079981059488e-06,
      "loss": 0.7404,
      "step": 10518
    },
    {
      "epoch": 0.3030888030888031,
      "grad_norm": 0.81175696849823,
      "learning_rate": 8.168719051858812e-06,
      "loss": 0.7557,
      "step": 10519
    },
    {
      "epoch": 0.3031176165504524,
      "grad_norm": 0.8175522089004517,
      "learning_rate": 8.168358095061759e-06,
      "loss": 0.7673,
      "step": 10520
    },
    {
      "epoch": 0.3031464300121017,
      "grad_norm": 0.8282006978988647,
      "learning_rate": 8.167997110671477e-06,
      "loss": 0.7553,
      "step": 10521
    },
    {
      "epoch": 0.3031752434737509,
      "grad_norm": 0.8275969624519348,
      "learning_rate": 8.167636098691107e-06,
      "loss": 0.744,
      "step": 10522
    },
    {
      "epoch": 0.3032040569354002,
      "grad_norm": 0.7723819613456726,
      "learning_rate": 8.167275059123791e-06,
      "loss": 0.749,
      "step": 10523
    },
    {
      "epoch": 0.3032328703970495,
      "grad_norm": 0.8091643452644348,
      "learning_rate": 8.166913991972677e-06,
      "loss": 0.7445,
      "step": 10524
    },
    {
      "epoch": 0.3032616838586988,
      "grad_norm": 0.8129962086677551,
      "learning_rate": 8.166552897240908e-06,
      "loss": 0.7602,
      "step": 10525
    },
    {
      "epoch": 0.3032904973203481,
      "grad_norm": 0.8485960960388184,
      "learning_rate": 8.166191774931628e-06,
      "loss": 0.7499,
      "step": 10526
    },
    {
      "epoch": 0.30331931078199736,
      "grad_norm": 0.7735818028450012,
      "learning_rate": 8.165830625047983e-06,
      "loss": 0.7383,
      "step": 10527
    },
    {
      "epoch": 0.30334812424364666,
      "grad_norm": 0.8471791744232178,
      "learning_rate": 8.165469447593118e-06,
      "loss": 0.7523,
      "step": 10528
    },
    {
      "epoch": 0.3033769377052959,
      "grad_norm": 0.8461520671844482,
      "learning_rate": 8.165108242570178e-06,
      "loss": 0.751,
      "step": 10529
    },
    {
      "epoch": 0.3034057511669452,
      "grad_norm": 0.791534423828125,
      "learning_rate": 8.164747009982309e-06,
      "loss": 0.7643,
      "step": 10530
    },
    {
      "epoch": 0.30343456462859447,
      "grad_norm": 0.8087489604949951,
      "learning_rate": 8.164385749832657e-06,
      "loss": 0.7494,
      "step": 10531
    },
    {
      "epoch": 0.30346337809024376,
      "grad_norm": 0.8439077734947205,
      "learning_rate": 8.16402446212437e-06,
      "loss": 0.7524,
      "step": 10532
    },
    {
      "epoch": 0.30349219155189305,
      "grad_norm": 0.8308500647544861,
      "learning_rate": 8.163663146860592e-06,
      "loss": 0.7299,
      "step": 10533
    },
    {
      "epoch": 0.30352100501354234,
      "grad_norm": 0.8176578879356384,
      "learning_rate": 8.16330180404447e-06,
      "loss": 0.7368,
      "step": 10534
    },
    {
      "epoch": 0.30354981847519164,
      "grad_norm": 0.7931567430496216,
      "learning_rate": 8.16294043367915e-06,
      "loss": 0.733,
      "step": 10535
    },
    {
      "epoch": 0.30357863193684087,
      "grad_norm": 0.8448998928070068,
      "learning_rate": 8.162579035767782e-06,
      "loss": 0.7792,
      "step": 10536
    },
    {
      "epoch": 0.30360744539849016,
      "grad_norm": 0.8109278082847595,
      "learning_rate": 8.162217610313513e-06,
      "loss": 0.7457,
      "step": 10537
    },
    {
      "epoch": 0.30363625886013945,
      "grad_norm": 0.8470053672790527,
      "learning_rate": 8.161856157319487e-06,
      "loss": 0.7597,
      "step": 10538
    },
    {
      "epoch": 0.30366507232178874,
      "grad_norm": 0.8142629265785217,
      "learning_rate": 8.161494676788857e-06,
      "loss": 0.7505,
      "step": 10539
    },
    {
      "epoch": 0.30369388578343803,
      "grad_norm": 0.8125433325767517,
      "learning_rate": 8.16113316872477e-06,
      "loss": 0.7277,
      "step": 10540
    },
    {
      "epoch": 0.3037226992450873,
      "grad_norm": 0.7988589406013489,
      "learning_rate": 8.16077163313037e-06,
      "loss": 0.7526,
      "step": 10541
    },
    {
      "epoch": 0.3037515127067366,
      "grad_norm": 0.8190787434577942,
      "learning_rate": 8.160410070008811e-06,
      "loss": 0.7506,
      "step": 10542
    },
    {
      "epoch": 0.30378032616838585,
      "grad_norm": 0.8036193251609802,
      "learning_rate": 8.160048479363237e-06,
      "loss": 0.7355,
      "step": 10543
    },
    {
      "epoch": 0.30380913963003514,
      "grad_norm": 1.0251927375793457,
      "learning_rate": 8.159686861196801e-06,
      "loss": 0.7656,
      "step": 10544
    },
    {
      "epoch": 0.30383795309168443,
      "grad_norm": 0.8065954446792603,
      "learning_rate": 8.159325215512653e-06,
      "loss": 0.7471,
      "step": 10545
    },
    {
      "epoch": 0.3038667665533337,
      "grad_norm": 0.8471518158912659,
      "learning_rate": 8.158963542313937e-06,
      "loss": 0.776,
      "step": 10546
    },
    {
      "epoch": 0.303895580014983,
      "grad_norm": 0.810218870639801,
      "learning_rate": 8.158601841603808e-06,
      "loss": 0.7558,
      "step": 10547
    },
    {
      "epoch": 0.3039243934766323,
      "grad_norm": 0.8390134572982788,
      "learning_rate": 8.158240113385415e-06,
      "loss": 0.75,
      "step": 10548
    },
    {
      "epoch": 0.30395320693828154,
      "grad_norm": 0.8521206974983215,
      "learning_rate": 8.157878357661905e-06,
      "loss": 0.7469,
      "step": 10549
    },
    {
      "epoch": 0.30398202039993083,
      "grad_norm": 0.8219888806343079,
      "learning_rate": 8.157516574436435e-06,
      "loss": 0.7611,
      "step": 10550
    },
    {
      "epoch": 0.3040108338615801,
      "grad_norm": 0.8143085837364197,
      "learning_rate": 8.157154763712149e-06,
      "loss": 0.7524,
      "step": 10551
    },
    {
      "epoch": 0.3040396473232294,
      "grad_norm": 0.8148396015167236,
      "learning_rate": 8.156792925492202e-06,
      "loss": 0.7748,
      "step": 10552
    },
    {
      "epoch": 0.3040684607848787,
      "grad_norm": 0.8161784410476685,
      "learning_rate": 8.156431059779745e-06,
      "loss": 0.777,
      "step": 10553
    },
    {
      "epoch": 0.304097274246528,
      "grad_norm": 0.8733042478561401,
      "learning_rate": 8.15606916657793e-06,
      "loss": 0.7611,
      "step": 10554
    },
    {
      "epoch": 0.3041260877081773,
      "grad_norm": 0.8118149638175964,
      "learning_rate": 8.155707245889905e-06,
      "loss": 0.7463,
      "step": 10555
    },
    {
      "epoch": 0.3041549011698265,
      "grad_norm": 0.7962005734443665,
      "learning_rate": 8.155345297718824e-06,
      "loss": 0.7664,
      "step": 10556
    },
    {
      "epoch": 0.3041837146314758,
      "grad_norm": 0.8171854615211487,
      "learning_rate": 8.154983322067842e-06,
      "loss": 0.7663,
      "step": 10557
    },
    {
      "epoch": 0.3042125280931251,
      "grad_norm": 0.8190751671791077,
      "learning_rate": 8.154621318940108e-06,
      "loss": 0.7982,
      "step": 10558
    },
    {
      "epoch": 0.3042413415547744,
      "grad_norm": 0.8266130089759827,
      "learning_rate": 8.154259288338776e-06,
      "loss": 0.7629,
      "step": 10559
    },
    {
      "epoch": 0.3042701550164237,
      "grad_norm": 0.805456280708313,
      "learning_rate": 8.153897230266998e-06,
      "loss": 0.7617,
      "step": 10560
    },
    {
      "epoch": 0.304298968478073,
      "grad_norm": 0.8151456713676453,
      "learning_rate": 8.153535144727927e-06,
      "loss": 0.7465,
      "step": 10561
    },
    {
      "epoch": 0.30432778193972226,
      "grad_norm": 0.8119978308677673,
      "learning_rate": 8.153173031724719e-06,
      "loss": 0.7598,
      "step": 10562
    },
    {
      "epoch": 0.3043565954013715,
      "grad_norm": 0.8319162726402283,
      "learning_rate": 8.152810891260525e-06,
      "loss": 0.7366,
      "step": 10563
    },
    {
      "epoch": 0.3043854088630208,
      "grad_norm": 0.8148282766342163,
      "learning_rate": 8.152448723338501e-06,
      "loss": 0.7418,
      "step": 10564
    },
    {
      "epoch": 0.3044142223246701,
      "grad_norm": 0.8177221417427063,
      "learning_rate": 8.1520865279618e-06,
      "loss": 0.7497,
      "step": 10565
    },
    {
      "epoch": 0.30444303578631937,
      "grad_norm": 0.8126566410064697,
      "learning_rate": 8.151724305133574e-06,
      "loss": 0.7516,
      "step": 10566
    },
    {
      "epoch": 0.30447184924796866,
      "grad_norm": 0.78323894739151,
      "learning_rate": 8.151362054856982e-06,
      "loss": 0.7349,
      "step": 10567
    },
    {
      "epoch": 0.30450066270961795,
      "grad_norm": 0.8072283267974854,
      "learning_rate": 8.150999777135176e-06,
      "loss": 0.7446,
      "step": 10568
    },
    {
      "epoch": 0.30452947617126724,
      "grad_norm": 0.8309691548347473,
      "learning_rate": 8.150637471971313e-06,
      "loss": 0.755,
      "step": 10569
    },
    {
      "epoch": 0.3045582896329165,
      "grad_norm": 0.8058789968490601,
      "learning_rate": 8.150275139368543e-06,
      "loss": 0.7444,
      "step": 10570
    },
    {
      "epoch": 0.30458710309456577,
      "grad_norm": 0.792915940284729,
      "learning_rate": 8.149912779330032e-06,
      "loss": 0.7357,
      "step": 10571
    },
    {
      "epoch": 0.30461591655621506,
      "grad_norm": 0.792998194694519,
      "learning_rate": 8.149550391858924e-06,
      "loss": 0.7603,
      "step": 10572
    },
    {
      "epoch": 0.30464473001786435,
      "grad_norm": 0.8160053491592407,
      "learning_rate": 8.149187976958383e-06,
      "loss": 0.7481,
      "step": 10573
    },
    {
      "epoch": 0.30467354347951364,
      "grad_norm": 0.8132138252258301,
      "learning_rate": 8.148825534631562e-06,
      "loss": 0.7371,
      "step": 10574
    },
    {
      "epoch": 0.30470235694116293,
      "grad_norm": 0.8263844847679138,
      "learning_rate": 8.148463064881618e-06,
      "loss": 0.7538,
      "step": 10575
    },
    {
      "epoch": 0.30473117040281217,
      "grad_norm": 0.7960144877433777,
      "learning_rate": 8.14810056771171e-06,
      "loss": 0.7641,
      "step": 10576
    },
    {
      "epoch": 0.30475998386446146,
      "grad_norm": 0.8410415053367615,
      "learning_rate": 8.14773804312499e-06,
      "loss": 0.7742,
      "step": 10577
    },
    {
      "epoch": 0.30478879732611075,
      "grad_norm": 0.7925176024436951,
      "learning_rate": 8.14737549112462e-06,
      "loss": 0.7332,
      "step": 10578
    },
    {
      "epoch": 0.30481761078776004,
      "grad_norm": 0.7891781330108643,
      "learning_rate": 8.147012911713757e-06,
      "loss": 0.7436,
      "step": 10579
    },
    {
      "epoch": 0.30484642424940933,
      "grad_norm": 0.7960652112960815,
      "learning_rate": 8.146650304895555e-06,
      "loss": 0.7756,
      "step": 10580
    },
    {
      "epoch": 0.3048752377110586,
      "grad_norm": 0.8194069862365723,
      "learning_rate": 8.146287670673176e-06,
      "loss": 0.7471,
      "step": 10581
    },
    {
      "epoch": 0.3049040511727079,
      "grad_norm": 0.8372876644134521,
      "learning_rate": 8.145925009049776e-06,
      "loss": 0.7571,
      "step": 10582
    },
    {
      "epoch": 0.30493286463435715,
      "grad_norm": 0.8054112195968628,
      "learning_rate": 8.145562320028512e-06,
      "loss": 0.76,
      "step": 10583
    },
    {
      "epoch": 0.30496167809600644,
      "grad_norm": 0.7997879981994629,
      "learning_rate": 8.145199603612546e-06,
      "loss": 0.7347,
      "step": 10584
    },
    {
      "epoch": 0.30499049155765573,
      "grad_norm": 0.8031381964683533,
      "learning_rate": 8.144836859805036e-06,
      "loss": 0.7659,
      "step": 10585
    },
    {
      "epoch": 0.305019305019305,
      "grad_norm": 0.818662166595459,
      "learning_rate": 8.144474088609142e-06,
      "loss": 0.7351,
      "step": 10586
    },
    {
      "epoch": 0.3050481184809543,
      "grad_norm": 0.8055047988891602,
      "learning_rate": 8.144111290028019e-06,
      "loss": 0.7479,
      "step": 10587
    },
    {
      "epoch": 0.3050769319426036,
      "grad_norm": 0.8078935146331787,
      "learning_rate": 8.143748464064831e-06,
      "loss": 0.7474,
      "step": 10588
    },
    {
      "epoch": 0.3051057454042529,
      "grad_norm": 0.8063830137252808,
      "learning_rate": 8.143385610722738e-06,
      "loss": 0.7511,
      "step": 10589
    },
    {
      "epoch": 0.30513455886590213,
      "grad_norm": 0.8059578537940979,
      "learning_rate": 8.143022730004897e-06,
      "loss": 0.7484,
      "step": 10590
    },
    {
      "epoch": 0.3051633723275514,
      "grad_norm": 0.7912415266036987,
      "learning_rate": 8.142659821914468e-06,
      "loss": 0.7565,
      "step": 10591
    },
    {
      "epoch": 0.3051921857892007,
      "grad_norm": 0.821083664894104,
      "learning_rate": 8.142296886454617e-06,
      "loss": 0.7319,
      "step": 10592
    },
    {
      "epoch": 0.30522099925085,
      "grad_norm": 0.8044906258583069,
      "learning_rate": 8.1419339236285e-06,
      "loss": 0.7321,
      "step": 10593
    },
    {
      "epoch": 0.3052498127124993,
      "grad_norm": 0.7900819182395935,
      "learning_rate": 8.141570933439278e-06,
      "loss": 0.7407,
      "step": 10594
    },
    {
      "epoch": 0.3052786261741486,
      "grad_norm": 0.8491023182868958,
      "learning_rate": 8.141207915890114e-06,
      "loss": 0.7412,
      "step": 10595
    },
    {
      "epoch": 0.3053074396357978,
      "grad_norm": 0.8227972388267517,
      "learning_rate": 8.140844870984171e-06,
      "loss": 0.7555,
      "step": 10596
    },
    {
      "epoch": 0.3053362530974471,
      "grad_norm": 0.8495236039161682,
      "learning_rate": 8.140481798724606e-06,
      "loss": 0.7824,
      "step": 10597
    },
    {
      "epoch": 0.3053650665590964,
      "grad_norm": 0.7946985363960266,
      "learning_rate": 8.140118699114587e-06,
      "loss": 0.7406,
      "step": 10598
    },
    {
      "epoch": 0.3053938800207457,
      "grad_norm": 0.7950883507728577,
      "learning_rate": 8.139755572157273e-06,
      "loss": 0.7259,
      "step": 10599
    },
    {
      "epoch": 0.305422693482395,
      "grad_norm": 0.800528347492218,
      "learning_rate": 8.139392417855825e-06,
      "loss": 0.7406,
      "step": 10600
    },
    {
      "epoch": 0.30545150694404427,
      "grad_norm": 0.7890335917472839,
      "learning_rate": 8.139029236213407e-06,
      "loss": 0.7488,
      "step": 10601
    },
    {
      "epoch": 0.30548032040569356,
      "grad_norm": 0.8768587708473206,
      "learning_rate": 8.138666027233182e-06,
      "loss": 0.7666,
      "step": 10602
    },
    {
      "epoch": 0.3055091338673428,
      "grad_norm": 0.8098974227905273,
      "learning_rate": 8.138302790918314e-06,
      "loss": 0.7823,
      "step": 10603
    },
    {
      "epoch": 0.3055379473289921,
      "grad_norm": 0.7930355072021484,
      "learning_rate": 8.137939527271966e-06,
      "loss": 0.7569,
      "step": 10604
    },
    {
      "epoch": 0.3055667607906414,
      "grad_norm": 0.8080809712409973,
      "learning_rate": 8.137576236297302e-06,
      "loss": 0.7281,
      "step": 10605
    },
    {
      "epoch": 0.30559557425229067,
      "grad_norm": 0.795950710773468,
      "learning_rate": 8.137212917997485e-06,
      "loss": 0.7328,
      "step": 10606
    },
    {
      "epoch": 0.30562438771393996,
      "grad_norm": 0.8187181353569031,
      "learning_rate": 8.136849572375679e-06,
      "loss": 0.7452,
      "step": 10607
    },
    {
      "epoch": 0.30565320117558925,
      "grad_norm": 0.812970757484436,
      "learning_rate": 8.13648619943505e-06,
      "loss": 0.7739,
      "step": 10608
    },
    {
      "epoch": 0.30568201463723854,
      "grad_norm": 0.8229576349258423,
      "learning_rate": 8.136122799178759e-06,
      "loss": 0.738,
      "step": 10609
    },
    {
      "epoch": 0.3057108280988878,
      "grad_norm": 0.846027135848999,
      "learning_rate": 8.135759371609977e-06,
      "loss": 0.7592,
      "step": 10610
    },
    {
      "epoch": 0.30573964156053707,
      "grad_norm": 2.6977505683898926,
      "learning_rate": 8.135395916731863e-06,
      "loss": 0.772,
      "step": 10611
    },
    {
      "epoch": 0.30576845502218636,
      "grad_norm": 0.8416024446487427,
      "learning_rate": 8.135032434547585e-06,
      "loss": 0.7611,
      "step": 10612
    },
    {
      "epoch": 0.30579726848383565,
      "grad_norm": 0.8164721131324768,
      "learning_rate": 8.134668925060309e-06,
      "loss": 0.7637,
      "step": 10613
    },
    {
      "epoch": 0.30582608194548494,
      "grad_norm": 0.8877723813056946,
      "learning_rate": 8.134305388273201e-06,
      "loss": 0.7717,
      "step": 10614
    },
    {
      "epoch": 0.30585489540713423,
      "grad_norm": 0.796614944934845,
      "learning_rate": 8.133941824189425e-06,
      "loss": 0.7434,
      "step": 10615
    },
    {
      "epoch": 0.3058837088687835,
      "grad_norm": 0.8083622455596924,
      "learning_rate": 8.13357823281215e-06,
      "loss": 0.7711,
      "step": 10616
    },
    {
      "epoch": 0.30591252233043276,
      "grad_norm": 0.8220793008804321,
      "learning_rate": 8.133214614144537e-06,
      "loss": 0.7639,
      "step": 10617
    },
    {
      "epoch": 0.30594133579208205,
      "grad_norm": 0.7884430885314941,
      "learning_rate": 8.132850968189761e-06,
      "loss": 0.7541,
      "step": 10618
    },
    {
      "epoch": 0.30597014925373134,
      "grad_norm": 0.8618866801261902,
      "learning_rate": 8.132487294950984e-06,
      "loss": 0.768,
      "step": 10619
    },
    {
      "epoch": 0.30599896271538063,
      "grad_norm": 0.8089079856872559,
      "learning_rate": 8.132123594431374e-06,
      "loss": 0.737,
      "step": 10620
    },
    {
      "epoch": 0.3060277761770299,
      "grad_norm": 0.834607720375061,
      "learning_rate": 8.131759866634096e-06,
      "loss": 0.7777,
      "step": 10621
    },
    {
      "epoch": 0.3060565896386792,
      "grad_norm": 0.8216825723648071,
      "learning_rate": 8.131396111562324e-06,
      "loss": 0.7324,
      "step": 10622
    },
    {
      "epoch": 0.30608540310032845,
      "grad_norm": 0.869624137878418,
      "learning_rate": 8.13103232921922e-06,
      "loss": 0.7704,
      "step": 10623
    },
    {
      "epoch": 0.30611421656197774,
      "grad_norm": 0.8679978251457214,
      "learning_rate": 8.130668519607954e-06,
      "loss": 0.7459,
      "step": 10624
    },
    {
      "epoch": 0.306143030023627,
      "grad_norm": 0.8371934294700623,
      "learning_rate": 8.130304682731696e-06,
      "loss": 0.7599,
      "step": 10625
    },
    {
      "epoch": 0.3061718434852763,
      "grad_norm": 0.8087931871414185,
      "learning_rate": 8.129940818593612e-06,
      "loss": 0.7497,
      "step": 10626
    },
    {
      "epoch": 0.3062006569469256,
      "grad_norm": 0.8327988982200623,
      "learning_rate": 8.129576927196872e-06,
      "loss": 0.7461,
      "step": 10627
    },
    {
      "epoch": 0.3062294704085749,
      "grad_norm": 0.8216053247451782,
      "learning_rate": 8.129213008544647e-06,
      "loss": 0.7515,
      "step": 10628
    },
    {
      "epoch": 0.3062582838702242,
      "grad_norm": 0.8046631813049316,
      "learning_rate": 8.128849062640104e-06,
      "loss": 0.7363,
      "step": 10629
    },
    {
      "epoch": 0.3062870973318734,
      "grad_norm": 0.8541102409362793,
      "learning_rate": 8.128485089486413e-06,
      "loss": 0.7652,
      "step": 10630
    },
    {
      "epoch": 0.3063159107935227,
      "grad_norm": 0.7741805911064148,
      "learning_rate": 8.128121089086745e-06,
      "loss": 0.7325,
      "step": 10631
    },
    {
      "epoch": 0.306344724255172,
      "grad_norm": 0.8267723917961121,
      "learning_rate": 8.127757061444269e-06,
      "loss": 0.7679,
      "step": 10632
    },
    {
      "epoch": 0.3063735377168213,
      "grad_norm": 0.8192306756973267,
      "learning_rate": 8.127393006562156e-06,
      "loss": 0.7953,
      "step": 10633
    },
    {
      "epoch": 0.3064023511784706,
      "grad_norm": 0.8179599642753601,
      "learning_rate": 8.127028924443575e-06,
      "loss": 0.7474,
      "step": 10634
    },
    {
      "epoch": 0.3064311646401199,
      "grad_norm": 0.8044435977935791,
      "learning_rate": 8.126664815091699e-06,
      "loss": 0.7573,
      "step": 10635
    },
    {
      "epoch": 0.30645997810176917,
      "grad_norm": 0.7759252786636353,
      "learning_rate": 8.126300678509698e-06,
      "loss": 0.7287,
      "step": 10636
    },
    {
      "epoch": 0.3064887915634184,
      "grad_norm": 0.7776702046394348,
      "learning_rate": 8.125936514700742e-06,
      "loss": 0.7584,
      "step": 10637
    },
    {
      "epoch": 0.3065176050250677,
      "grad_norm": 0.7979080677032471,
      "learning_rate": 8.125572323668005e-06,
      "loss": 0.7574,
      "step": 10638
    },
    {
      "epoch": 0.306546418486717,
      "grad_norm": 0.8102908730506897,
      "learning_rate": 8.125208105414656e-06,
      "loss": 0.7481,
      "step": 10639
    },
    {
      "epoch": 0.3065752319483663,
      "grad_norm": 0.8050668835639954,
      "learning_rate": 8.124843859943869e-06,
      "loss": 0.7592,
      "step": 10640
    },
    {
      "epoch": 0.30660404541001557,
      "grad_norm": 0.8353232145309448,
      "learning_rate": 8.124479587258817e-06,
      "loss": 0.7446,
      "step": 10641
    },
    {
      "epoch": 0.30663285887166486,
      "grad_norm": 0.7945769429206848,
      "learning_rate": 8.12411528736267e-06,
      "loss": 0.7376,
      "step": 10642
    },
    {
      "epoch": 0.30666167233331415,
      "grad_norm": 0.7808191180229187,
      "learning_rate": 8.123750960258603e-06,
      "loss": 0.7564,
      "step": 10643
    },
    {
      "epoch": 0.3066904857949634,
      "grad_norm": 0.8168830275535583,
      "learning_rate": 8.123386605949789e-06,
      "loss": 0.7608,
      "step": 10644
    },
    {
      "epoch": 0.3067192992566127,
      "grad_norm": 0.7908010482788086,
      "learning_rate": 8.123022224439396e-06,
      "loss": 0.7334,
      "step": 10645
    },
    {
      "epoch": 0.30674811271826197,
      "grad_norm": 0.7887358665466309,
      "learning_rate": 8.122657815730604e-06,
      "loss": 0.7702,
      "step": 10646
    },
    {
      "epoch": 0.30677692617991126,
      "grad_norm": 0.8039499521255493,
      "learning_rate": 8.122293379826583e-06,
      "loss": 0.7478,
      "step": 10647
    },
    {
      "epoch": 0.30680573964156055,
      "grad_norm": 0.7766205668449402,
      "learning_rate": 8.121928916730507e-06,
      "loss": 0.7493,
      "step": 10648
    },
    {
      "epoch": 0.30683455310320984,
      "grad_norm": 1.0230121612548828,
      "learning_rate": 8.121564426445551e-06,
      "loss": 0.751,
      "step": 10649
    },
    {
      "epoch": 0.3068633665648591,
      "grad_norm": 0.7797535061836243,
      "learning_rate": 8.121199908974891e-06,
      "loss": 0.7214,
      "step": 10650
    },
    {
      "epoch": 0.30689218002650837,
      "grad_norm": 0.8165822625160217,
      "learning_rate": 8.120835364321699e-06,
      "loss": 0.7317,
      "step": 10651
    },
    {
      "epoch": 0.30692099348815766,
      "grad_norm": 0.808199405670166,
      "learning_rate": 8.120470792489151e-06,
      "loss": 0.742,
      "step": 10652
    },
    {
      "epoch": 0.30694980694980695,
      "grad_norm": 0.7979610562324524,
      "learning_rate": 8.12010619348042e-06,
      "loss": 0.7493,
      "step": 10653
    },
    {
      "epoch": 0.30697862041145624,
      "grad_norm": 0.8199052214622498,
      "learning_rate": 8.119741567298685e-06,
      "loss": 0.7325,
      "step": 10654
    },
    {
      "epoch": 0.30700743387310553,
      "grad_norm": 0.8126615285873413,
      "learning_rate": 8.119376913947119e-06,
      "loss": 0.7484,
      "step": 10655
    },
    {
      "epoch": 0.3070362473347548,
      "grad_norm": 0.8407812714576721,
      "learning_rate": 8.119012233428898e-06,
      "loss": 0.742,
      "step": 10656
    },
    {
      "epoch": 0.30706506079640405,
      "grad_norm": 0.847724199295044,
      "learning_rate": 8.118647525747198e-06,
      "loss": 0.7547,
      "step": 10657
    },
    {
      "epoch": 0.30709387425805335,
      "grad_norm": 0.8151255249977112,
      "learning_rate": 8.118282790905196e-06,
      "loss": 0.7497,
      "step": 10658
    },
    {
      "epoch": 0.30712268771970264,
      "grad_norm": 0.8306108713150024,
      "learning_rate": 8.117918028906068e-06,
      "loss": 0.7552,
      "step": 10659
    },
    {
      "epoch": 0.3071515011813519,
      "grad_norm": 0.7764747738838196,
      "learning_rate": 8.117553239752992e-06,
      "loss": 0.7395,
      "step": 10660
    },
    {
      "epoch": 0.3071803146430012,
      "grad_norm": 0.8349770903587341,
      "learning_rate": 8.117188423449143e-06,
      "loss": 0.7683,
      "step": 10661
    },
    {
      "epoch": 0.3072091281046505,
      "grad_norm": 0.859160304069519,
      "learning_rate": 8.116823579997697e-06,
      "loss": 0.75,
      "step": 10662
    },
    {
      "epoch": 0.3072379415662998,
      "grad_norm": 0.781129777431488,
      "learning_rate": 8.116458709401836e-06,
      "loss": 0.714,
      "step": 10663
    },
    {
      "epoch": 0.30726675502794903,
      "grad_norm": 0.9008992314338684,
      "learning_rate": 8.116093811664732e-06,
      "loss": 0.7678,
      "step": 10664
    },
    {
      "epoch": 0.3072955684895983,
      "grad_norm": 0.8480977416038513,
      "learning_rate": 8.11572888678957e-06,
      "loss": 0.7485,
      "step": 10665
    },
    {
      "epoch": 0.3073243819512476,
      "grad_norm": 0.7728533744812012,
      "learning_rate": 8.11536393477952e-06,
      "loss": 0.7126,
      "step": 10666
    },
    {
      "epoch": 0.3073531954128969,
      "grad_norm": 0.8209190964698792,
      "learning_rate": 8.114998955637766e-06,
      "loss": 0.7509,
      "step": 10667
    },
    {
      "epoch": 0.3073820088745462,
      "grad_norm": 0.819621741771698,
      "learning_rate": 8.114633949367486e-06,
      "loss": 0.748,
      "step": 10668
    },
    {
      "epoch": 0.3074108223361955,
      "grad_norm": 0.8376238346099854,
      "learning_rate": 8.114268915971855e-06,
      "loss": 0.7542,
      "step": 10669
    },
    {
      "epoch": 0.3074396357978448,
      "grad_norm": 0.8329547047615051,
      "learning_rate": 8.113903855454058e-06,
      "loss": 0.7634,
      "step": 10670
    },
    {
      "epoch": 0.307468449259494,
      "grad_norm": 0.9016777276992798,
      "learning_rate": 8.113538767817269e-06,
      "loss": 0.75,
      "step": 10671
    },
    {
      "epoch": 0.3074972627211433,
      "grad_norm": 0.8208999633789062,
      "learning_rate": 8.11317365306467e-06,
      "loss": 0.7659,
      "step": 10672
    },
    {
      "epoch": 0.3075260761827926,
      "grad_norm": 0.8060023188591003,
      "learning_rate": 8.112808511199442e-06,
      "loss": 0.7609,
      "step": 10673
    },
    {
      "epoch": 0.3075548896444419,
      "grad_norm": 0.8493107557296753,
      "learning_rate": 8.112443342224761e-06,
      "loss": 0.7485,
      "step": 10674
    },
    {
      "epoch": 0.3075837031060912,
      "grad_norm": 0.8447122573852539,
      "learning_rate": 8.112078146143812e-06,
      "loss": 0.7792,
      "step": 10675
    },
    {
      "epoch": 0.30761251656774047,
      "grad_norm": 0.8123989701271057,
      "learning_rate": 8.111712922959773e-06,
      "loss": 0.7577,
      "step": 10676
    },
    {
      "epoch": 0.3076413300293897,
      "grad_norm": 0.8023298978805542,
      "learning_rate": 8.111347672675825e-06,
      "loss": 0.7397,
      "step": 10677
    },
    {
      "epoch": 0.307670143491039,
      "grad_norm": 0.8041682243347168,
      "learning_rate": 8.110982395295148e-06,
      "loss": 0.7379,
      "step": 10678
    },
    {
      "epoch": 0.3076989569526883,
      "grad_norm": 0.8128987550735474,
      "learning_rate": 8.110617090820926e-06,
      "loss": 0.7396,
      "step": 10679
    },
    {
      "epoch": 0.3077277704143376,
      "grad_norm": 0.8259178996086121,
      "learning_rate": 8.110251759256336e-06,
      "loss": 0.7414,
      "step": 10680
    },
    {
      "epoch": 0.30775658387598687,
      "grad_norm": 0.8514415621757507,
      "learning_rate": 8.109886400604563e-06,
      "loss": 0.7768,
      "step": 10681
    },
    {
      "epoch": 0.30778539733763616,
      "grad_norm": 0.8273023366928101,
      "learning_rate": 8.10952101486879e-06,
      "loss": 0.7607,
      "step": 10682
    },
    {
      "epoch": 0.30781421079928545,
      "grad_norm": 0.8236276507377625,
      "learning_rate": 8.109155602052195e-06,
      "loss": 0.7431,
      "step": 10683
    },
    {
      "epoch": 0.3078430242609347,
      "grad_norm": 0.8002662062644958,
      "learning_rate": 8.108790162157965e-06,
      "loss": 0.7517,
      "step": 10684
    },
    {
      "epoch": 0.307871837722584,
      "grad_norm": 0.7842856049537659,
      "learning_rate": 8.108424695189278e-06,
      "loss": 0.75,
      "step": 10685
    },
    {
      "epoch": 0.30790065118423326,
      "grad_norm": 0.780396580696106,
      "learning_rate": 8.108059201149321e-06,
      "loss": 0.7476,
      "step": 10686
    },
    {
      "epoch": 0.30792946464588256,
      "grad_norm": 0.8362133502960205,
      "learning_rate": 8.107693680041274e-06,
      "loss": 0.7211,
      "step": 10687
    },
    {
      "epoch": 0.30795827810753185,
      "grad_norm": 0.7927781343460083,
      "learning_rate": 8.107328131868323e-06,
      "loss": 0.7283,
      "step": 10688
    },
    {
      "epoch": 0.30798709156918114,
      "grad_norm": 0.7972254157066345,
      "learning_rate": 8.106962556633649e-06,
      "loss": 0.7557,
      "step": 10689
    },
    {
      "epoch": 0.30801590503083043,
      "grad_norm": 0.8234474062919617,
      "learning_rate": 8.106596954340438e-06,
      "loss": 0.7711,
      "step": 10690
    },
    {
      "epoch": 0.30804471849247966,
      "grad_norm": 0.8225436210632324,
      "learning_rate": 8.106231324991871e-06,
      "loss": 0.7584,
      "step": 10691
    },
    {
      "epoch": 0.30807353195412895,
      "grad_norm": 0.8032797574996948,
      "learning_rate": 8.105865668591138e-06,
      "loss": 0.7672,
      "step": 10692
    },
    {
      "epoch": 0.30810234541577824,
      "grad_norm": 0.8154138326644897,
      "learning_rate": 8.105499985141416e-06,
      "loss": 0.7564,
      "step": 10693
    },
    {
      "epoch": 0.30813115887742754,
      "grad_norm": 0.7894315123558044,
      "learning_rate": 8.105134274645896e-06,
      "loss": 0.7545,
      "step": 10694
    },
    {
      "epoch": 0.3081599723390768,
      "grad_norm": 0.7959744930267334,
      "learning_rate": 8.104768537107759e-06,
      "loss": 0.7552,
      "step": 10695
    },
    {
      "epoch": 0.3081887858007261,
      "grad_norm": 0.7805479168891907,
      "learning_rate": 8.104402772530193e-06,
      "loss": 0.7232,
      "step": 10696
    },
    {
      "epoch": 0.3082175992623754,
      "grad_norm": 0.8417384624481201,
      "learning_rate": 8.104036980916382e-06,
      "loss": 0.7491,
      "step": 10697
    },
    {
      "epoch": 0.30824641272402464,
      "grad_norm": 0.8432399034500122,
      "learning_rate": 8.10367116226951e-06,
      "loss": 0.7611,
      "step": 10698
    },
    {
      "epoch": 0.30827522618567393,
      "grad_norm": 0.7916463613510132,
      "learning_rate": 8.103305316592765e-06,
      "loss": 0.7486,
      "step": 10699
    },
    {
      "epoch": 0.3083040396473232,
      "grad_norm": 0.7994959354400635,
      "learning_rate": 8.102939443889332e-06,
      "loss": 0.7613,
      "step": 10700
    },
    {
      "epoch": 0.3083328531089725,
      "grad_norm": 0.7782138586044312,
      "learning_rate": 8.102573544162402e-06,
      "loss": 0.7808,
      "step": 10701
    },
    {
      "epoch": 0.3083616665706218,
      "grad_norm": 0.8043104410171509,
      "learning_rate": 8.102207617415154e-06,
      "loss": 0.7565,
      "step": 10702
    },
    {
      "epoch": 0.3083904800322711,
      "grad_norm": 0.8252952098846436,
      "learning_rate": 8.101841663650781e-06,
      "loss": 0.7852,
      "step": 10703
    },
    {
      "epoch": 0.30841929349392033,
      "grad_norm": 0.7776844501495361,
      "learning_rate": 8.101475682872468e-06,
      "loss": 0.7506,
      "step": 10704
    },
    {
      "epoch": 0.3084481069555696,
      "grad_norm": 0.7960137128829956,
      "learning_rate": 8.101109675083401e-06,
      "loss": 0.7536,
      "step": 10705
    },
    {
      "epoch": 0.3084769204172189,
      "grad_norm": 0.8186833262443542,
      "learning_rate": 8.100743640286768e-06,
      "loss": 0.7686,
      "step": 10706
    },
    {
      "epoch": 0.3085057338788682,
      "grad_norm": 0.8013653755187988,
      "learning_rate": 8.10037757848576e-06,
      "loss": 0.7539,
      "step": 10707
    },
    {
      "epoch": 0.3085345473405175,
      "grad_norm": 0.7817067503929138,
      "learning_rate": 8.10001148968356e-06,
      "loss": 0.7556,
      "step": 10708
    },
    {
      "epoch": 0.3085633608021668,
      "grad_norm": 0.8171495199203491,
      "learning_rate": 8.099645373883361e-06,
      "loss": 0.7625,
      "step": 10709
    },
    {
      "epoch": 0.3085921742638161,
      "grad_norm": 0.8194803595542908,
      "learning_rate": 8.099279231088348e-06,
      "loss": 0.7311,
      "step": 10710
    },
    {
      "epoch": 0.3086209877254653,
      "grad_norm": 0.7652603983879089,
      "learning_rate": 8.098913061301711e-06,
      "loss": 0.7419,
      "step": 10711
    },
    {
      "epoch": 0.3086498011871146,
      "grad_norm": 0.8353564739227295,
      "learning_rate": 8.09854686452664e-06,
      "loss": 0.7564,
      "step": 10712
    },
    {
      "epoch": 0.3086786146487639,
      "grad_norm": 0.8109682202339172,
      "learning_rate": 8.098180640766322e-06,
      "loss": 0.7389,
      "step": 10713
    },
    {
      "epoch": 0.3087074281104132,
      "grad_norm": 0.7756155133247375,
      "learning_rate": 8.097814390023948e-06,
      "loss": 0.725,
      "step": 10714
    },
    {
      "epoch": 0.3087362415720625,
      "grad_norm": 0.8490676879882812,
      "learning_rate": 8.097448112302709e-06,
      "loss": 0.7501,
      "step": 10715
    },
    {
      "epoch": 0.30876505503371177,
      "grad_norm": 0.8040912747383118,
      "learning_rate": 8.097081807605792e-06,
      "loss": 0.7632,
      "step": 10716
    },
    {
      "epoch": 0.30879386849536106,
      "grad_norm": 0.7861469388008118,
      "learning_rate": 8.096715475936386e-06,
      "loss": 0.7855,
      "step": 10717
    },
    {
      "epoch": 0.3088226819570103,
      "grad_norm": 0.8209878206253052,
      "learning_rate": 8.096349117297686e-06,
      "loss": 0.7578,
      "step": 10718
    },
    {
      "epoch": 0.3088514954186596,
      "grad_norm": 0.8317371010780334,
      "learning_rate": 8.09598273169288e-06,
      "loss": 0.7578,
      "step": 10719
    },
    {
      "epoch": 0.3088803088803089,
      "grad_norm": 0.7863743901252747,
      "learning_rate": 8.09561631912516e-06,
      "loss": 0.7412,
      "step": 10720
    },
    {
      "epoch": 0.30890912234195816,
      "grad_norm": 0.8219156265258789,
      "learning_rate": 8.095249879597716e-06,
      "loss": 0.7396,
      "step": 10721
    },
    {
      "epoch": 0.30893793580360746,
      "grad_norm": 0.8307347893714905,
      "learning_rate": 8.09488341311374e-06,
      "loss": 0.7557,
      "step": 10722
    },
    {
      "epoch": 0.30896674926525675,
      "grad_norm": 0.8448805809020996,
      "learning_rate": 8.094516919676423e-06,
      "loss": 0.7874,
      "step": 10723
    },
    {
      "epoch": 0.30899556272690604,
      "grad_norm": 0.8310681581497192,
      "learning_rate": 8.094150399288956e-06,
      "loss": 0.7415,
      "step": 10724
    },
    {
      "epoch": 0.30902437618855527,
      "grad_norm": 0.8268681764602661,
      "learning_rate": 8.093783851954533e-06,
      "loss": 0.739,
      "step": 10725
    },
    {
      "epoch": 0.30905318965020456,
      "grad_norm": 0.8123823404312134,
      "learning_rate": 8.093417277676346e-06,
      "loss": 0.7379,
      "step": 10726
    },
    {
      "epoch": 0.30908200311185385,
      "grad_norm": 0.8244998455047607,
      "learning_rate": 8.093050676457585e-06,
      "loss": 0.7573,
      "step": 10727
    },
    {
      "epoch": 0.30911081657350314,
      "grad_norm": 0.8230948448181152,
      "learning_rate": 8.092684048301447e-06,
      "loss": 0.7452,
      "step": 10728
    },
    {
      "epoch": 0.30913963003515244,
      "grad_norm": 0.8005564212799072,
      "learning_rate": 8.092317393211119e-06,
      "loss": 0.7284,
      "step": 10729
    },
    {
      "epoch": 0.3091684434968017,
      "grad_norm": 0.8176239132881165,
      "learning_rate": 8.0919507111898e-06,
      "loss": 0.7641,
      "step": 10730
    },
    {
      "epoch": 0.30919725695845096,
      "grad_norm": 0.7931923270225525,
      "learning_rate": 8.091584002240682e-06,
      "loss": 0.7315,
      "step": 10731
    },
    {
      "epoch": 0.30922607042010025,
      "grad_norm": 0.8046744465827942,
      "learning_rate": 8.091217266366957e-06,
      "loss": 0.7338,
      "step": 10732
    },
    {
      "epoch": 0.30925488388174954,
      "grad_norm": 0.8400354385375977,
      "learning_rate": 8.090850503571818e-06,
      "loss": 0.76,
      "step": 10733
    },
    {
      "epoch": 0.30928369734339883,
      "grad_norm": 0.7997647523880005,
      "learning_rate": 8.090483713858463e-06,
      "loss": 0.7561,
      "step": 10734
    },
    {
      "epoch": 0.3093125108050481,
      "grad_norm": 0.7968940138816833,
      "learning_rate": 8.090116897230083e-06,
      "loss": 0.7353,
      "step": 10735
    },
    {
      "epoch": 0.3093413242666974,
      "grad_norm": 0.8244019150733948,
      "learning_rate": 8.089750053689874e-06,
      "loss": 0.7402,
      "step": 10736
    },
    {
      "epoch": 0.3093701377283467,
      "grad_norm": 0.8088309168815613,
      "learning_rate": 8.08938318324103e-06,
      "loss": 0.7684,
      "step": 10737
    },
    {
      "epoch": 0.30939895118999594,
      "grad_norm": 1.1921509504318237,
      "learning_rate": 8.089016285886747e-06,
      "loss": 0.725,
      "step": 10738
    },
    {
      "epoch": 0.30942776465164523,
      "grad_norm": 0.7873346209526062,
      "learning_rate": 8.088649361630224e-06,
      "loss": 0.7783,
      "step": 10739
    },
    {
      "epoch": 0.3094565781132945,
      "grad_norm": 0.783416211605072,
      "learning_rate": 8.088282410474649e-06,
      "loss": 0.7272,
      "step": 10740
    },
    {
      "epoch": 0.3094853915749438,
      "grad_norm": 0.7951030731201172,
      "learning_rate": 8.08791543242322e-06,
      "loss": 0.7505,
      "step": 10741
    },
    {
      "epoch": 0.3095142050365931,
      "grad_norm": 0.830958366394043,
      "learning_rate": 8.087548427479136e-06,
      "loss": 0.7465,
      "step": 10742
    },
    {
      "epoch": 0.3095430184982424,
      "grad_norm": 0.8229153156280518,
      "learning_rate": 8.08718139564559e-06,
      "loss": 0.755,
      "step": 10743
    },
    {
      "epoch": 0.3095718319598917,
      "grad_norm": 0.8436763882637024,
      "learning_rate": 8.086814336925783e-06,
      "loss": 0.7559,
      "step": 10744
    },
    {
      "epoch": 0.3096006454215409,
      "grad_norm": 0.809137225151062,
      "learning_rate": 8.086447251322908e-06,
      "loss": 0.7501,
      "step": 10745
    },
    {
      "epoch": 0.3096294588831902,
      "grad_norm": 0.8362744450569153,
      "learning_rate": 8.086080138840163e-06,
      "loss": 0.7583,
      "step": 10746
    },
    {
      "epoch": 0.3096582723448395,
      "grad_norm": 0.8385539054870605,
      "learning_rate": 8.085712999480742e-06,
      "loss": 0.7373,
      "step": 10747
    },
    {
      "epoch": 0.3096870858064888,
      "grad_norm": 0.8070892095565796,
      "learning_rate": 8.085345833247849e-06,
      "loss": 0.7368,
      "step": 10748
    },
    {
      "epoch": 0.3097158992681381,
      "grad_norm": 0.8306432366371155,
      "learning_rate": 8.084978640144677e-06,
      "loss": 0.7603,
      "step": 10749
    },
    {
      "epoch": 0.3097447127297874,
      "grad_norm": 0.8795518279075623,
      "learning_rate": 8.084611420174424e-06,
      "loss": 0.762,
      "step": 10750
    },
    {
      "epoch": 0.30977352619143667,
      "grad_norm": 1.0363178253173828,
      "learning_rate": 8.084244173340288e-06,
      "loss": 0.7441,
      "step": 10751
    },
    {
      "epoch": 0.3098023396530859,
      "grad_norm": 0.8209866881370544,
      "learning_rate": 8.08387689964547e-06,
      "loss": 0.7638,
      "step": 10752
    },
    {
      "epoch": 0.3098311531147352,
      "grad_norm": 0.835652232170105,
      "learning_rate": 8.083509599093167e-06,
      "loss": 0.7553,
      "step": 10753
    },
    {
      "epoch": 0.3098599665763845,
      "grad_norm": 1.4161992073059082,
      "learning_rate": 8.083142271686577e-06,
      "loss": 0.766,
      "step": 10754
    },
    {
      "epoch": 0.3098887800380338,
      "grad_norm": 0.8151569962501526,
      "learning_rate": 8.0827749174289e-06,
      "loss": 0.7579,
      "step": 10755
    },
    {
      "epoch": 0.30991759349968306,
      "grad_norm": 0.8041853904724121,
      "learning_rate": 8.082407536323335e-06,
      "loss": 0.7826,
      "step": 10756
    },
    {
      "epoch": 0.30994640696133235,
      "grad_norm": 0.7865723371505737,
      "learning_rate": 8.082040128373084e-06,
      "loss": 0.7429,
      "step": 10757
    },
    {
      "epoch": 0.3099752204229816,
      "grad_norm": 0.8001589179039001,
      "learning_rate": 8.081672693581342e-06,
      "loss": 0.7487,
      "step": 10758
    },
    {
      "epoch": 0.3100040338846309,
      "grad_norm": 0.8244328498840332,
      "learning_rate": 8.081305231951311e-06,
      "loss": 0.763,
      "step": 10759
    },
    {
      "epoch": 0.31003284734628017,
      "grad_norm": 0.8448318243026733,
      "learning_rate": 8.080937743486192e-06,
      "loss": 0.7514,
      "step": 10760
    },
    {
      "epoch": 0.31006166080792946,
      "grad_norm": 0.8093527555465698,
      "learning_rate": 8.080570228189187e-06,
      "loss": 0.7534,
      "step": 10761
    },
    {
      "epoch": 0.31009047426957875,
      "grad_norm": 0.8192700743675232,
      "learning_rate": 8.080202686063494e-06,
      "loss": 0.761,
      "step": 10762
    },
    {
      "epoch": 0.31011928773122804,
      "grad_norm": 0.8091453313827515,
      "learning_rate": 8.079835117112313e-06,
      "loss": 0.7501,
      "step": 10763
    },
    {
      "epoch": 0.31014810119287733,
      "grad_norm": 0.8520186543464661,
      "learning_rate": 8.07946752133885e-06,
      "loss": 0.7548,
      "step": 10764
    },
    {
      "epoch": 0.31017691465452657,
      "grad_norm": 0.7930030822753906,
      "learning_rate": 8.079099898746298e-06,
      "loss": 0.7575,
      "step": 10765
    },
    {
      "epoch": 0.31020572811617586,
      "grad_norm": 0.8081322312355042,
      "learning_rate": 8.078732249337867e-06,
      "loss": 0.7612,
      "step": 10766
    },
    {
      "epoch": 0.31023454157782515,
      "grad_norm": 0.7882506251335144,
      "learning_rate": 8.078364573116756e-06,
      "loss": 0.7613,
      "step": 10767
    },
    {
      "epoch": 0.31026335503947444,
      "grad_norm": 0.8075588345527649,
      "learning_rate": 8.077996870086167e-06,
      "loss": 0.7477,
      "step": 10768
    },
    {
      "epoch": 0.31029216850112373,
      "grad_norm": 0.8144681453704834,
      "learning_rate": 8.077629140249301e-06,
      "loss": 0.753,
      "step": 10769
    },
    {
      "epoch": 0.310320981962773,
      "grad_norm": 0.8711029291152954,
      "learning_rate": 8.077261383609363e-06,
      "loss": 0.7544,
      "step": 10770
    },
    {
      "epoch": 0.3103497954244223,
      "grad_norm": 0.822454571723938,
      "learning_rate": 8.076893600169553e-06,
      "loss": 0.7518,
      "step": 10771
    },
    {
      "epoch": 0.31037860888607155,
      "grad_norm": 0.771652340888977,
      "learning_rate": 8.076525789933077e-06,
      "loss": 0.743,
      "step": 10772
    },
    {
      "epoch": 0.31040742234772084,
      "grad_norm": 0.7838506698608398,
      "learning_rate": 8.076157952903134e-06,
      "loss": 0.7249,
      "step": 10773
    },
    {
      "epoch": 0.31043623580937013,
      "grad_norm": 0.847453236579895,
      "learning_rate": 8.075790089082934e-06,
      "loss": 0.7498,
      "step": 10774
    },
    {
      "epoch": 0.3104650492710194,
      "grad_norm": 0.7946906089782715,
      "learning_rate": 8.075422198475675e-06,
      "loss": 0.748,
      "step": 10775
    },
    {
      "epoch": 0.3104938627326687,
      "grad_norm": 0.7761605978012085,
      "learning_rate": 8.075054281084562e-06,
      "loss": 0.7313,
      "step": 10776
    },
    {
      "epoch": 0.310522676194318,
      "grad_norm": 0.8705773949623108,
      "learning_rate": 8.074686336912802e-06,
      "loss": 0.743,
      "step": 10777
    },
    {
      "epoch": 0.3105514896559673,
      "grad_norm": 0.8435922861099243,
      "learning_rate": 8.074318365963598e-06,
      "loss": 0.7615,
      "step": 10778
    },
    {
      "epoch": 0.31058030311761653,
      "grad_norm": 0.7917904853820801,
      "learning_rate": 8.073950368240151e-06,
      "loss": 0.7748,
      "step": 10779
    },
    {
      "epoch": 0.3106091165792658,
      "grad_norm": 0.774467945098877,
      "learning_rate": 8.073582343745671e-06,
      "loss": 0.7172,
      "step": 10780
    },
    {
      "epoch": 0.3106379300409151,
      "grad_norm": 0.8257917761802673,
      "learning_rate": 8.073214292483363e-06,
      "loss": 0.7554,
      "step": 10781
    },
    {
      "epoch": 0.3106667435025644,
      "grad_norm": 0.8562591671943665,
      "learning_rate": 8.072846214456429e-06,
      "loss": 0.7597,
      "step": 10782
    },
    {
      "epoch": 0.3106955569642137,
      "grad_norm": 0.784739077091217,
      "learning_rate": 8.072478109668076e-06,
      "loss": 0.7269,
      "step": 10783
    },
    {
      "epoch": 0.310724370425863,
      "grad_norm": 0.7939310073852539,
      "learning_rate": 8.07210997812151e-06,
      "loss": 0.7476,
      "step": 10784
    },
    {
      "epoch": 0.3107531838875122,
      "grad_norm": 0.843666136264801,
      "learning_rate": 8.071741819819938e-06,
      "loss": 0.764,
      "step": 10785
    },
    {
      "epoch": 0.3107819973491615,
      "grad_norm": 0.830560564994812,
      "learning_rate": 8.071373634766563e-06,
      "loss": 0.7294,
      "step": 10786
    },
    {
      "epoch": 0.3108108108108108,
      "grad_norm": 0.806062638759613,
      "learning_rate": 8.071005422964598e-06,
      "loss": 0.7332,
      "step": 10787
    },
    {
      "epoch": 0.3108396242724601,
      "grad_norm": 0.7940459251403809,
      "learning_rate": 8.070637184417243e-06,
      "loss": 0.7511,
      "step": 10788
    },
    {
      "epoch": 0.3108684377341094,
      "grad_norm": 0.8249983787536621,
      "learning_rate": 8.070268919127708e-06,
      "loss": 0.7361,
      "step": 10789
    },
    {
      "epoch": 0.3108972511957587,
      "grad_norm": 0.8657320737838745,
      "learning_rate": 8.069900627099198e-06,
      "loss": 0.7517,
      "step": 10790
    },
    {
      "epoch": 0.31092606465740796,
      "grad_norm": 0.8300461173057556,
      "learning_rate": 8.069532308334925e-06,
      "loss": 0.7756,
      "step": 10791
    },
    {
      "epoch": 0.3109548781190572,
      "grad_norm": 0.7970863580703735,
      "learning_rate": 8.069163962838093e-06,
      "loss": 0.75,
      "step": 10792
    },
    {
      "epoch": 0.3109836915807065,
      "grad_norm": 0.8772076368331909,
      "learning_rate": 8.068795590611911e-06,
      "loss": 0.7521,
      "step": 10793
    },
    {
      "epoch": 0.3110125050423558,
      "grad_norm": 0.8404179215431213,
      "learning_rate": 8.068427191659588e-06,
      "loss": 0.7838,
      "step": 10794
    },
    {
      "epoch": 0.31104131850400507,
      "grad_norm": 0.8401931524276733,
      "learning_rate": 8.06805876598433e-06,
      "loss": 0.7286,
      "step": 10795
    },
    {
      "epoch": 0.31107013196565436,
      "grad_norm": 0.8073196411132812,
      "learning_rate": 8.067690313589347e-06,
      "loss": 0.7614,
      "step": 10796
    },
    {
      "epoch": 0.31109894542730365,
      "grad_norm": 0.8133311867713928,
      "learning_rate": 8.06732183447785e-06,
      "loss": 0.7581,
      "step": 10797
    },
    {
      "epoch": 0.31112775888895294,
      "grad_norm": 0.8584988117218018,
      "learning_rate": 8.066953328653045e-06,
      "loss": 0.7682,
      "step": 10798
    },
    {
      "epoch": 0.3111565723506022,
      "grad_norm": 0.8046073913574219,
      "learning_rate": 8.06658479611814e-06,
      "loss": 0.7294,
      "step": 10799
    },
    {
      "epoch": 0.31118538581225147,
      "grad_norm": 0.8325737714767456,
      "learning_rate": 8.06621623687635e-06,
      "loss": 0.7862,
      "step": 10800
    },
    {
      "epoch": 0.31121419927390076,
      "grad_norm": 0.777445912361145,
      "learning_rate": 8.06584765093088e-06,
      "loss": 0.7714,
      "step": 10801
    },
    {
      "epoch": 0.31124301273555005,
      "grad_norm": 0.7628124952316284,
      "learning_rate": 8.065479038284942e-06,
      "loss": 0.7308,
      "step": 10802
    },
    {
      "epoch": 0.31127182619719934,
      "grad_norm": 0.7781811356544495,
      "learning_rate": 8.065110398941746e-06,
      "loss": 0.7408,
      "step": 10803
    },
    {
      "epoch": 0.31130063965884863,
      "grad_norm": 0.7945171594619751,
      "learning_rate": 8.064741732904504e-06,
      "loss": 0.741,
      "step": 10804
    },
    {
      "epoch": 0.3113294531204979,
      "grad_norm": 0.8138388395309448,
      "learning_rate": 8.064373040176422e-06,
      "loss": 0.7624,
      "step": 10805
    },
    {
      "epoch": 0.31135826658214716,
      "grad_norm": 0.8145878911018372,
      "learning_rate": 8.064004320760717e-06,
      "loss": 0.768,
      "step": 10806
    },
    {
      "epoch": 0.31138708004379645,
      "grad_norm": 0.7801358103752136,
      "learning_rate": 8.063635574660595e-06,
      "loss": 0.7632,
      "step": 10807
    },
    {
      "epoch": 0.31141589350544574,
      "grad_norm": 0.8011848330497742,
      "learning_rate": 8.06326680187927e-06,
      "loss": 0.7534,
      "step": 10808
    },
    {
      "epoch": 0.31144470696709503,
      "grad_norm": 0.7969697117805481,
      "learning_rate": 8.062898002419954e-06,
      "loss": 0.7516,
      "step": 10809
    },
    {
      "epoch": 0.3114735204287443,
      "grad_norm": 0.8270362019538879,
      "learning_rate": 8.062529176285858e-06,
      "loss": 0.7362,
      "step": 10810
    },
    {
      "epoch": 0.3115023338903936,
      "grad_norm": 0.7891209721565247,
      "learning_rate": 8.062160323480195e-06,
      "loss": 0.7607,
      "step": 10811
    },
    {
      "epoch": 0.31153114735204285,
      "grad_norm": 0.8154779076576233,
      "learning_rate": 8.061791444006175e-06,
      "loss": 0.737,
      "step": 10812
    },
    {
      "epoch": 0.31155996081369214,
      "grad_norm": 0.8007762432098389,
      "learning_rate": 8.061422537867015e-06,
      "loss": 0.7668,
      "step": 10813
    },
    {
      "epoch": 0.31158877427534143,
      "grad_norm": 0.8235228657722473,
      "learning_rate": 8.061053605065922e-06,
      "loss": 0.7617,
      "step": 10814
    },
    {
      "epoch": 0.3116175877369907,
      "grad_norm": 0.8408417105674744,
      "learning_rate": 8.060684645606114e-06,
      "loss": 0.7674,
      "step": 10815
    },
    {
      "epoch": 0.31164640119864,
      "grad_norm": 0.8031232357025146,
      "learning_rate": 8.060315659490802e-06,
      "loss": 0.7695,
      "step": 10816
    },
    {
      "epoch": 0.3116752146602893,
      "grad_norm": 2.550950765609741,
      "learning_rate": 8.0599466467232e-06,
      "loss": 0.7435,
      "step": 10817
    },
    {
      "epoch": 0.3117040281219386,
      "grad_norm": 0.8218332529067993,
      "learning_rate": 8.05957760730652e-06,
      "loss": 0.7642,
      "step": 10818
    },
    {
      "epoch": 0.3117328415835878,
      "grad_norm": 0.8377723693847656,
      "learning_rate": 8.059208541243978e-06,
      "loss": 0.7502,
      "step": 10819
    },
    {
      "epoch": 0.3117616550452371,
      "grad_norm": 0.8229957222938538,
      "learning_rate": 8.058839448538789e-06,
      "loss": 0.7724,
      "step": 10820
    },
    {
      "epoch": 0.3117904685068864,
      "grad_norm": 0.8273907899856567,
      "learning_rate": 8.058470329194165e-06,
      "loss": 0.7524,
      "step": 10821
    },
    {
      "epoch": 0.3118192819685357,
      "grad_norm": 0.8042860627174377,
      "learning_rate": 8.05810118321332e-06,
      "loss": 0.7547,
      "step": 10822
    },
    {
      "epoch": 0.311848095430185,
      "grad_norm": 0.8145886063575745,
      "learning_rate": 8.057732010599476e-06,
      "loss": 0.7503,
      "step": 10823
    },
    {
      "epoch": 0.3118769088918343,
      "grad_norm": 0.8158321380615234,
      "learning_rate": 8.057362811355838e-06,
      "loss": 0.7476,
      "step": 10824
    },
    {
      "epoch": 0.31190572235348357,
      "grad_norm": 0.8159922361373901,
      "learning_rate": 8.056993585485626e-06,
      "loss": 0.7541,
      "step": 10825
    },
    {
      "epoch": 0.3119345358151328,
      "grad_norm": 0.792829155921936,
      "learning_rate": 8.056624332992058e-06,
      "loss": 0.7485,
      "step": 10826
    },
    {
      "epoch": 0.3119633492767821,
      "grad_norm": 0.8379937410354614,
      "learning_rate": 8.056255053878348e-06,
      "loss": 0.7382,
      "step": 10827
    },
    {
      "epoch": 0.3119921627384314,
      "grad_norm": 0.8341211676597595,
      "learning_rate": 8.05588574814771e-06,
      "loss": 0.7541,
      "step": 10828
    },
    {
      "epoch": 0.3120209762000807,
      "grad_norm": 0.8242160081863403,
      "learning_rate": 8.055516415803363e-06,
      "loss": 0.7679,
      "step": 10829
    },
    {
      "epoch": 0.31204978966172997,
      "grad_norm": 0.8217971324920654,
      "learning_rate": 8.055147056848522e-06,
      "loss": 0.7644,
      "step": 10830
    },
    {
      "epoch": 0.31207860312337926,
      "grad_norm": 0.8475967049598694,
      "learning_rate": 8.054777671286403e-06,
      "loss": 0.7406,
      "step": 10831
    },
    {
      "epoch": 0.31210741658502855,
      "grad_norm": 0.8404272198677063,
      "learning_rate": 8.054408259120225e-06,
      "loss": 0.7544,
      "step": 10832
    },
    {
      "epoch": 0.3121362300466778,
      "grad_norm": 0.7819145917892456,
      "learning_rate": 8.054038820353205e-06,
      "loss": 0.7264,
      "step": 10833
    },
    {
      "epoch": 0.3121650435083271,
      "grad_norm": 0.7983687520027161,
      "learning_rate": 8.05366935498856e-06,
      "loss": 0.7552,
      "step": 10834
    },
    {
      "epoch": 0.31219385696997637,
      "grad_norm": 0.8207574486732483,
      "learning_rate": 8.053299863029507e-06,
      "loss": 0.7476,
      "step": 10835
    },
    {
      "epoch": 0.31222267043162566,
      "grad_norm": 0.8506303429603577,
      "learning_rate": 8.052930344479264e-06,
      "loss": 0.7634,
      "step": 10836
    },
    {
      "epoch": 0.31225148389327495,
      "grad_norm": 0.7777572274208069,
      "learning_rate": 8.052560799341049e-06,
      "loss": 0.764,
      "step": 10837
    },
    {
      "epoch": 0.31228029735492424,
      "grad_norm": 0.8096075057983398,
      "learning_rate": 8.052191227618083e-06,
      "loss": 0.739,
      "step": 10838
    },
    {
      "epoch": 0.3123091108165735,
      "grad_norm": 0.8452168107032776,
      "learning_rate": 8.05182162931358e-06,
      "loss": 0.7642,
      "step": 10839
    },
    {
      "epoch": 0.31233792427822277,
      "grad_norm": 0.7852778434753418,
      "learning_rate": 8.051452004430763e-06,
      "loss": 0.7276,
      "step": 10840
    },
    {
      "epoch": 0.31236673773987206,
      "grad_norm": 0.8079907894134521,
      "learning_rate": 8.051082352972848e-06,
      "loss": 0.7532,
      "step": 10841
    },
    {
      "epoch": 0.31239555120152135,
      "grad_norm": 0.8374592661857605,
      "learning_rate": 8.050712674943056e-06,
      "loss": 0.747,
      "step": 10842
    },
    {
      "epoch": 0.31242436466317064,
      "grad_norm": 0.7835307121276855,
      "learning_rate": 8.050342970344607e-06,
      "loss": 0.7389,
      "step": 10843
    },
    {
      "epoch": 0.31245317812481993,
      "grad_norm": 0.7897706031799316,
      "learning_rate": 8.049973239180719e-06,
      "loss": 0.741,
      "step": 10844
    },
    {
      "epoch": 0.3124819915864692,
      "grad_norm": 0.7633383274078369,
      "learning_rate": 8.049603481454615e-06,
      "loss": 0.7361,
      "step": 10845
    },
    {
      "epoch": 0.31251080504811846,
      "grad_norm": 0.8395198583602905,
      "learning_rate": 8.049233697169511e-06,
      "loss": 0.7283,
      "step": 10846
    },
    {
      "epoch": 0.31253961850976775,
      "grad_norm": 0.8809974789619446,
      "learning_rate": 8.04886388632863e-06,
      "loss": 0.7649,
      "step": 10847
    },
    {
      "epoch": 0.31256843197141704,
      "grad_norm": 0.8140408992767334,
      "learning_rate": 8.048494048935192e-06,
      "loss": 0.7194,
      "step": 10848
    },
    {
      "epoch": 0.31259724543306633,
      "grad_norm": 0.8044030666351318,
      "learning_rate": 8.048124184992418e-06,
      "loss": 0.7478,
      "step": 10849
    },
    {
      "epoch": 0.3126260588947156,
      "grad_norm": 0.798810601234436,
      "learning_rate": 8.047754294503532e-06,
      "loss": 0.7518,
      "step": 10850
    },
    {
      "epoch": 0.3126548723563649,
      "grad_norm": 0.8291030526161194,
      "learning_rate": 8.04738437747175e-06,
      "loss": 0.7512,
      "step": 10851
    },
    {
      "epoch": 0.3126836858180142,
      "grad_norm": 0.7958788275718689,
      "learning_rate": 8.047014433900297e-06,
      "loss": 0.7047,
      "step": 10852
    },
    {
      "epoch": 0.31271249927966344,
      "grad_norm": 0.8051261901855469,
      "learning_rate": 8.046644463792394e-06,
      "loss": 0.7304,
      "step": 10853
    },
    {
      "epoch": 0.3127413127413127,
      "grad_norm": 0.8172518610954285,
      "learning_rate": 8.046274467151264e-06,
      "loss": 0.7455,
      "step": 10854
    },
    {
      "epoch": 0.312770126202962,
      "grad_norm": 0.8123369216918945,
      "learning_rate": 8.045904443980129e-06,
      "loss": 0.7772,
      "step": 10855
    },
    {
      "epoch": 0.3127989396646113,
      "grad_norm": 0.7949446439743042,
      "learning_rate": 8.045534394282211e-06,
      "loss": 0.7441,
      "step": 10856
    },
    {
      "epoch": 0.3128277531262606,
      "grad_norm": 0.7738912105560303,
      "learning_rate": 8.045164318060731e-06,
      "loss": 0.7437,
      "step": 10857
    },
    {
      "epoch": 0.3128565665879099,
      "grad_norm": 0.7849189043045044,
      "learning_rate": 8.044794215318916e-06,
      "loss": 0.7247,
      "step": 10858
    },
    {
      "epoch": 0.3128853800495592,
      "grad_norm": 0.7861728668212891,
      "learning_rate": 8.044424086059985e-06,
      "loss": 0.7545,
      "step": 10859
    },
    {
      "epoch": 0.3129141935112084,
      "grad_norm": 0.8182209730148315,
      "learning_rate": 8.044053930287165e-06,
      "loss": 0.7513,
      "step": 10860
    },
    {
      "epoch": 0.3129430069728577,
      "grad_norm": 0.7985143661499023,
      "learning_rate": 8.043683748003677e-06,
      "loss": 0.7398,
      "step": 10861
    },
    {
      "epoch": 0.312971820434507,
      "grad_norm": 0.7909280061721802,
      "learning_rate": 8.043313539212747e-06,
      "loss": 0.7442,
      "step": 10862
    },
    {
      "epoch": 0.3130006338961563,
      "grad_norm": 0.8350248336791992,
      "learning_rate": 8.042943303917598e-06,
      "loss": 0.7715,
      "step": 10863
    },
    {
      "epoch": 0.3130294473578056,
      "grad_norm": 0.8755092620849609,
      "learning_rate": 8.042573042121455e-06,
      "loss": 0.7771,
      "step": 10864
    },
    {
      "epoch": 0.31305826081945487,
      "grad_norm": 0.8139822483062744,
      "learning_rate": 8.042202753827544e-06,
      "loss": 0.742,
      "step": 10865
    },
    {
      "epoch": 0.3130870742811041,
      "grad_norm": 0.823562741279602,
      "learning_rate": 8.041832439039085e-06,
      "loss": 0.7751,
      "step": 10866
    },
    {
      "epoch": 0.3131158877427534,
      "grad_norm": 0.8331836462020874,
      "learning_rate": 8.041462097759307e-06,
      "loss": 0.7458,
      "step": 10867
    },
    {
      "epoch": 0.3131447012044027,
      "grad_norm": 0.8461818099021912,
      "learning_rate": 8.041091729991434e-06,
      "loss": 0.7577,
      "step": 10868
    },
    {
      "epoch": 0.313173514666052,
      "grad_norm": 0.8127824068069458,
      "learning_rate": 8.040721335738693e-06,
      "loss": 0.7239,
      "step": 10869
    },
    {
      "epoch": 0.31320232812770127,
      "grad_norm": 0.8168846964836121,
      "learning_rate": 8.040350915004309e-06,
      "loss": 0.7535,
      "step": 10870
    },
    {
      "epoch": 0.31323114158935056,
      "grad_norm": 0.8185516595840454,
      "learning_rate": 8.039980467791509e-06,
      "loss": 0.793,
      "step": 10871
    },
    {
      "epoch": 0.31325995505099985,
      "grad_norm": 0.8321619629859924,
      "learning_rate": 8.039609994103515e-06,
      "loss": 0.7549,
      "step": 10872
    },
    {
      "epoch": 0.3132887685126491,
      "grad_norm": 0.7962097525596619,
      "learning_rate": 8.039239493943557e-06,
      "loss": 0.7512,
      "step": 10873
    },
    {
      "epoch": 0.3133175819742984,
      "grad_norm": 0.7862381339073181,
      "learning_rate": 8.038868967314864e-06,
      "loss": 0.7414,
      "step": 10874
    },
    {
      "epoch": 0.31334639543594767,
      "grad_norm": 0.8009669184684753,
      "learning_rate": 8.038498414220657e-06,
      "loss": 0.7514,
      "step": 10875
    },
    {
      "epoch": 0.31337520889759696,
      "grad_norm": 0.8563210964202881,
      "learning_rate": 8.038127834664168e-06,
      "loss": 0.7635,
      "step": 10876
    },
    {
      "epoch": 0.31340402235924625,
      "grad_norm": 0.7692787647247314,
      "learning_rate": 8.037757228648621e-06,
      "loss": 0.7539,
      "step": 10877
    },
    {
      "epoch": 0.31343283582089554,
      "grad_norm": 0.7952061891555786,
      "learning_rate": 8.037386596177245e-06,
      "loss": 0.7572,
      "step": 10878
    },
    {
      "epoch": 0.31346164928254483,
      "grad_norm": 0.8112700581550598,
      "learning_rate": 8.03701593725327e-06,
      "loss": 0.7662,
      "step": 10879
    },
    {
      "epoch": 0.31349046274419406,
      "grad_norm": 0.8069347143173218,
      "learning_rate": 8.036645251879921e-06,
      "loss": 0.7635,
      "step": 10880
    },
    {
      "epoch": 0.31351927620584336,
      "grad_norm": 0.7636970281600952,
      "learning_rate": 8.036274540060426e-06,
      "loss": 0.722,
      "step": 10881
    },
    {
      "epoch": 0.31354808966749265,
      "grad_norm": 0.7869192957878113,
      "learning_rate": 8.035903801798019e-06,
      "loss": 0.7638,
      "step": 10882
    },
    {
      "epoch": 0.31357690312914194,
      "grad_norm": 0.7799513936042786,
      "learning_rate": 8.035533037095921e-06,
      "loss": 0.7407,
      "step": 10883
    },
    {
      "epoch": 0.31360571659079123,
      "grad_norm": 0.8353095650672913,
      "learning_rate": 8.035162245957367e-06,
      "loss": 0.7835,
      "step": 10884
    },
    {
      "epoch": 0.3136345300524405,
      "grad_norm": 0.8275323510169983,
      "learning_rate": 8.034791428385582e-06,
      "loss": 0.7466,
      "step": 10885
    },
    {
      "epoch": 0.3136633435140898,
      "grad_norm": 0.7915062308311462,
      "learning_rate": 8.034420584383797e-06,
      "loss": 0.758,
      "step": 10886
    },
    {
      "epoch": 0.31369215697573904,
      "grad_norm": 0.7981797456741333,
      "learning_rate": 8.034049713955244e-06,
      "loss": 0.7416,
      "step": 10887
    },
    {
      "epoch": 0.31372097043738834,
      "grad_norm": 0.791276752948761,
      "learning_rate": 8.03367881710315e-06,
      "loss": 0.7276,
      "step": 10888
    },
    {
      "epoch": 0.3137497838990376,
      "grad_norm": 0.8068981170654297,
      "learning_rate": 8.033307893830746e-06,
      "loss": 0.7663,
      "step": 10889
    },
    {
      "epoch": 0.3137785973606869,
      "grad_norm": 0.8215310573577881,
      "learning_rate": 8.032936944141262e-06,
      "loss": 0.7563,
      "step": 10890
    },
    {
      "epoch": 0.3138074108223362,
      "grad_norm": 0.8157197833061218,
      "learning_rate": 8.032565968037929e-06,
      "loss": 0.7521,
      "step": 10891
    },
    {
      "epoch": 0.3138362242839855,
      "grad_norm": 0.8051162362098694,
      "learning_rate": 8.032194965523979e-06,
      "loss": 0.7482,
      "step": 10892
    },
    {
      "epoch": 0.31386503774563473,
      "grad_norm": 0.7957286238670349,
      "learning_rate": 8.03182393660264e-06,
      "loss": 0.7648,
      "step": 10893
    },
    {
      "epoch": 0.313893851207284,
      "grad_norm": 0.7897598147392273,
      "learning_rate": 8.031452881277149e-06,
      "loss": 0.7382,
      "step": 10894
    },
    {
      "epoch": 0.3139226646689333,
      "grad_norm": 0.8472633361816406,
      "learning_rate": 8.031081799550731e-06,
      "loss": 0.7517,
      "step": 10895
    },
    {
      "epoch": 0.3139514781305826,
      "grad_norm": 0.8429309725761414,
      "learning_rate": 8.03071069142662e-06,
      "loss": 0.7406,
      "step": 10896
    },
    {
      "epoch": 0.3139802915922319,
      "grad_norm": 0.794934093952179,
      "learning_rate": 8.030339556908049e-06,
      "loss": 0.7475,
      "step": 10897
    },
    {
      "epoch": 0.3140091050538812,
      "grad_norm": 0.803617000579834,
      "learning_rate": 8.02996839599825e-06,
      "loss": 0.7429,
      "step": 10898
    },
    {
      "epoch": 0.3140379185155305,
      "grad_norm": 0.8420127630233765,
      "learning_rate": 8.029597208700455e-06,
      "loss": 0.746,
      "step": 10899
    },
    {
      "epoch": 0.3140667319771797,
      "grad_norm": 0.7764767408370972,
      "learning_rate": 8.029225995017897e-06,
      "loss": 0.7673,
      "step": 10900
    },
    {
      "epoch": 0.314095545438829,
      "grad_norm": 0.7931874394416809,
      "learning_rate": 8.02885475495381e-06,
      "loss": 0.7454,
      "step": 10901
    },
    {
      "epoch": 0.3141243589004783,
      "grad_norm": 0.7895917892456055,
      "learning_rate": 8.028483488511422e-06,
      "loss": 0.7633,
      "step": 10902
    },
    {
      "epoch": 0.3141531723621276,
      "grad_norm": 0.8055709004402161,
      "learning_rate": 8.028112195693974e-06,
      "loss": 0.7543,
      "step": 10903
    },
    {
      "epoch": 0.3141819858237769,
      "grad_norm": 0.8186283707618713,
      "learning_rate": 8.027740876504693e-06,
      "loss": 0.742,
      "step": 10904
    },
    {
      "epoch": 0.31421079928542617,
      "grad_norm": 0.7798434495925903,
      "learning_rate": 8.027369530946818e-06,
      "loss": 0.739,
      "step": 10905
    },
    {
      "epoch": 0.31423961274707546,
      "grad_norm": 0.7665755152702332,
      "learning_rate": 8.02699815902358e-06,
      "loss": 0.7275,
      "step": 10906
    },
    {
      "epoch": 0.3142684262087247,
      "grad_norm": 0.7764081954956055,
      "learning_rate": 8.026626760738213e-06,
      "loss": 0.7371,
      "step": 10907
    },
    {
      "epoch": 0.314297239670374,
      "grad_norm": 0.8004605174064636,
      "learning_rate": 8.026255336093954e-06,
      "loss": 0.7515,
      "step": 10908
    },
    {
      "epoch": 0.3143260531320233,
      "grad_norm": 0.8188742399215698,
      "learning_rate": 8.025883885094035e-06,
      "loss": 0.7527,
      "step": 10909
    },
    {
      "epoch": 0.31435486659367257,
      "grad_norm": 0.788254976272583,
      "learning_rate": 8.025512407741692e-06,
      "loss": 0.749,
      "step": 10910
    },
    {
      "epoch": 0.31438368005532186,
      "grad_norm": 0.838051438331604,
      "learning_rate": 8.02514090404016e-06,
      "loss": 0.7804,
      "step": 10911
    },
    {
      "epoch": 0.31441249351697115,
      "grad_norm": 0.8156901597976685,
      "learning_rate": 8.024769373992676e-06,
      "loss": 0.7589,
      "step": 10912
    },
    {
      "epoch": 0.31444130697862044,
      "grad_norm": 0.7888239026069641,
      "learning_rate": 8.024397817602473e-06,
      "loss": 0.7644,
      "step": 10913
    },
    {
      "epoch": 0.3144701204402697,
      "grad_norm": 0.7648945450782776,
      "learning_rate": 8.02402623487279e-06,
      "loss": 0.7446,
      "step": 10914
    },
    {
      "epoch": 0.31449893390191896,
      "grad_norm": 0.7885923981666565,
      "learning_rate": 8.02365462580686e-06,
      "loss": 0.7521,
      "step": 10915
    },
    {
      "epoch": 0.31452774736356826,
      "grad_norm": 0.7960216403007507,
      "learning_rate": 8.023282990407921e-06,
      "loss": 0.7334,
      "step": 10916
    },
    {
      "epoch": 0.31455656082521755,
      "grad_norm": 0.8438016176223755,
      "learning_rate": 8.02291132867921e-06,
      "loss": 0.7554,
      "step": 10917
    },
    {
      "epoch": 0.31458537428686684,
      "grad_norm": 0.8090221285820007,
      "learning_rate": 8.022539640623962e-06,
      "loss": 0.7452,
      "step": 10918
    },
    {
      "epoch": 0.3146141877485161,
      "grad_norm": 0.7961806058883667,
      "learning_rate": 8.022167926245416e-06,
      "loss": 0.7498,
      "step": 10919
    },
    {
      "epoch": 0.31464300121016536,
      "grad_norm": 0.7798007726669312,
      "learning_rate": 8.021796185546808e-06,
      "loss": 0.7207,
      "step": 10920
    },
    {
      "epoch": 0.31467181467181465,
      "grad_norm": 0.8142949938774109,
      "learning_rate": 8.021424418531376e-06,
      "loss": 0.7613,
      "step": 10921
    },
    {
      "epoch": 0.31470062813346394,
      "grad_norm": 0.7894954681396484,
      "learning_rate": 8.02105262520236e-06,
      "loss": 0.7551,
      "step": 10922
    },
    {
      "epoch": 0.31472944159511324,
      "grad_norm": 0.8327106833457947,
      "learning_rate": 8.020680805562993e-06,
      "loss": 0.7626,
      "step": 10923
    },
    {
      "epoch": 0.3147582550567625,
      "grad_norm": 0.8054222464561462,
      "learning_rate": 8.020308959616516e-06,
      "loss": 0.7678,
      "step": 10924
    },
    {
      "epoch": 0.3147870685184118,
      "grad_norm": 0.8339789509773254,
      "learning_rate": 8.019937087366167e-06,
      "loss": 0.7848,
      "step": 10925
    },
    {
      "epoch": 0.3148158819800611,
      "grad_norm": 0.8644517064094543,
      "learning_rate": 8.019565188815185e-06,
      "loss": 0.7887,
      "step": 10926
    },
    {
      "epoch": 0.31484469544171034,
      "grad_norm": 0.7937217354774475,
      "learning_rate": 8.019193263966809e-06,
      "loss": 0.7556,
      "step": 10927
    },
    {
      "epoch": 0.31487350890335963,
      "grad_norm": 0.8265551924705505,
      "learning_rate": 8.018821312824278e-06,
      "loss": 0.7539,
      "step": 10928
    },
    {
      "epoch": 0.3149023223650089,
      "grad_norm": 0.8127135038375854,
      "learning_rate": 8.01844933539083e-06,
      "loss": 0.7256,
      "step": 10929
    },
    {
      "epoch": 0.3149311358266582,
      "grad_norm": 0.8318156599998474,
      "learning_rate": 8.018077331669707e-06,
      "loss": 0.7509,
      "step": 10930
    },
    {
      "epoch": 0.3149599492883075,
      "grad_norm": 0.847121000289917,
      "learning_rate": 8.017705301664147e-06,
      "loss": 0.7325,
      "step": 10931
    },
    {
      "epoch": 0.3149887627499568,
      "grad_norm": 0.8470549583435059,
      "learning_rate": 8.01733324537739e-06,
      "loss": 0.7315,
      "step": 10932
    },
    {
      "epoch": 0.3150175762116061,
      "grad_norm": 0.7741166353225708,
      "learning_rate": 8.016961162812676e-06,
      "loss": 0.719,
      "step": 10933
    },
    {
      "epoch": 0.3150463896732553,
      "grad_norm": 0.8523608446121216,
      "learning_rate": 8.016589053973246e-06,
      "loss": 0.7516,
      "step": 10934
    },
    {
      "epoch": 0.3150752031349046,
      "grad_norm": 0.8782697916030884,
      "learning_rate": 8.016216918862342e-06,
      "loss": 0.7573,
      "step": 10935
    },
    {
      "epoch": 0.3151040165965539,
      "grad_norm": 0.811414897441864,
      "learning_rate": 8.015844757483204e-06,
      "loss": 0.7515,
      "step": 10936
    },
    {
      "epoch": 0.3151328300582032,
      "grad_norm": 0.83203125,
      "learning_rate": 8.015472569839074e-06,
      "loss": 0.7596,
      "step": 10937
    },
    {
      "epoch": 0.3151616435198525,
      "grad_norm": 0.8304694890975952,
      "learning_rate": 8.01510035593319e-06,
      "loss": 0.7382,
      "step": 10938
    },
    {
      "epoch": 0.3151904569815018,
      "grad_norm": 0.771511971950531,
      "learning_rate": 8.014728115768797e-06,
      "loss": 0.7357,
      "step": 10939
    },
    {
      "epoch": 0.31521927044315107,
      "grad_norm": 4.037568092346191,
      "learning_rate": 8.014355849349134e-06,
      "loss": 0.7503,
      "step": 10940
    },
    {
      "epoch": 0.3152480839048003,
      "grad_norm": 0.7778560519218445,
      "learning_rate": 8.013983556677448e-06,
      "loss": 0.741,
      "step": 10941
    },
    {
      "epoch": 0.3152768973664496,
      "grad_norm": 0.7688624262809753,
      "learning_rate": 8.013611237756975e-06,
      "loss": 0.7516,
      "step": 10942
    },
    {
      "epoch": 0.3153057108280989,
      "grad_norm": 0.8324155211448669,
      "learning_rate": 8.013238892590962e-06,
      "loss": 0.751,
      "step": 10943
    },
    {
      "epoch": 0.3153345242897482,
      "grad_norm": 0.8190929889678955,
      "learning_rate": 8.01286652118265e-06,
      "loss": 0.7976,
      "step": 10944
    },
    {
      "epoch": 0.31536333775139747,
      "grad_norm": 0.8144834041595459,
      "learning_rate": 8.012494123535283e-06,
      "loss": 0.7431,
      "step": 10945
    },
    {
      "epoch": 0.31539215121304676,
      "grad_norm": 0.8081961274147034,
      "learning_rate": 8.012121699652103e-06,
      "loss": 0.7505,
      "step": 10946
    },
    {
      "epoch": 0.315420964674696,
      "grad_norm": 0.7979898452758789,
      "learning_rate": 8.011749249536354e-06,
      "loss": 0.7528,
      "step": 10947
    },
    {
      "epoch": 0.3154497781363453,
      "grad_norm": 0.8976014852523804,
      "learning_rate": 8.01137677319128e-06,
      "loss": 0.7692,
      "step": 10948
    },
    {
      "epoch": 0.3154785915979946,
      "grad_norm": 0.8271259069442749,
      "learning_rate": 8.011004270620126e-06,
      "loss": 0.744,
      "step": 10949
    },
    {
      "epoch": 0.31550740505964386,
      "grad_norm": 0.8568337559700012,
      "learning_rate": 8.010631741826131e-06,
      "loss": 0.7841,
      "step": 10950
    },
    {
      "epoch": 0.31553621852129315,
      "grad_norm": 0.8024575114250183,
      "learning_rate": 8.010259186812545e-06,
      "loss": 0.7506,
      "step": 10951
    },
    {
      "epoch": 0.31556503198294245,
      "grad_norm": 0.8382140398025513,
      "learning_rate": 8.009886605582612e-06,
      "loss": 0.7708,
      "step": 10952
    },
    {
      "epoch": 0.31559384544459174,
      "grad_norm": 0.7904840111732483,
      "learning_rate": 8.009513998139575e-06,
      "loss": 0.7315,
      "step": 10953
    },
    {
      "epoch": 0.31562265890624097,
      "grad_norm": 1.081939458847046,
      "learning_rate": 8.00914136448668e-06,
      "loss": 0.7685,
      "step": 10954
    },
    {
      "epoch": 0.31565147236789026,
      "grad_norm": 0.7889933586120605,
      "learning_rate": 8.00876870462717e-06,
      "loss": 0.7256,
      "step": 10955
    },
    {
      "epoch": 0.31568028582953955,
      "grad_norm": 0.898186981678009,
      "learning_rate": 8.008396018564293e-06,
      "loss": 0.7501,
      "step": 10956
    },
    {
      "epoch": 0.31570909929118884,
      "grad_norm": 0.831335723400116,
      "learning_rate": 8.008023306301293e-06,
      "loss": 0.7376,
      "step": 10957
    },
    {
      "epoch": 0.31573791275283813,
      "grad_norm": 0.8289527893066406,
      "learning_rate": 8.007650567841416e-06,
      "loss": 0.7452,
      "step": 10958
    },
    {
      "epoch": 0.3157667262144874,
      "grad_norm": 0.8045763969421387,
      "learning_rate": 8.007277803187912e-06,
      "loss": 0.7443,
      "step": 10959
    },
    {
      "epoch": 0.3157955396761367,
      "grad_norm": 0.7863439917564392,
      "learning_rate": 8.006905012344024e-06,
      "loss": 0.7632,
      "step": 10960
    },
    {
      "epoch": 0.31582435313778595,
      "grad_norm": 0.8367076516151428,
      "learning_rate": 8.006532195312998e-06,
      "loss": 0.7252,
      "step": 10961
    },
    {
      "epoch": 0.31585316659943524,
      "grad_norm": 0.7749225497245789,
      "learning_rate": 8.006159352098082e-06,
      "loss": 0.7422,
      "step": 10962
    },
    {
      "epoch": 0.31588198006108453,
      "grad_norm": 0.7892942428588867,
      "learning_rate": 8.005786482702525e-06,
      "loss": 0.7342,
      "step": 10963
    },
    {
      "epoch": 0.3159107935227338,
      "grad_norm": 0.7871493697166443,
      "learning_rate": 8.00541358712957e-06,
      "loss": 0.7337,
      "step": 10964
    },
    {
      "epoch": 0.3159396069843831,
      "grad_norm": 0.8241455554962158,
      "learning_rate": 8.00504066538247e-06,
      "loss": 0.7693,
      "step": 10965
    },
    {
      "epoch": 0.3159684204460324,
      "grad_norm": 0.7741226553916931,
      "learning_rate": 8.004667717464467e-06,
      "loss": 0.7463,
      "step": 10966
    },
    {
      "epoch": 0.3159972339076817,
      "grad_norm": 0.8280811905860901,
      "learning_rate": 8.00429474337881e-06,
      "loss": 0.7623,
      "step": 10967
    },
    {
      "epoch": 0.31602604736933093,
      "grad_norm": 0.78017657995224,
      "learning_rate": 8.003921743128753e-06,
      "loss": 0.7481,
      "step": 10968
    },
    {
      "epoch": 0.3160548608309802,
      "grad_norm": 0.7927071452140808,
      "learning_rate": 8.003548716717538e-06,
      "loss": 0.735,
      "step": 10969
    },
    {
      "epoch": 0.3160836742926295,
      "grad_norm": 0.8105362057685852,
      "learning_rate": 8.003175664148417e-06,
      "loss": 0.7745,
      "step": 10970
    },
    {
      "epoch": 0.3161124877542788,
      "grad_norm": 0.825835108757019,
      "learning_rate": 8.002802585424639e-06,
      "loss": 0.7645,
      "step": 10971
    },
    {
      "epoch": 0.3161413012159281,
      "grad_norm": 0.8059589862823486,
      "learning_rate": 8.002429480549453e-06,
      "loss": 0.7504,
      "step": 10972
    },
    {
      "epoch": 0.3161701146775774,
      "grad_norm": 0.7958929538726807,
      "learning_rate": 8.002056349526103e-06,
      "loss": 0.7412,
      "step": 10973
    },
    {
      "epoch": 0.3161989281392266,
      "grad_norm": 0.7987133860588074,
      "learning_rate": 8.001683192357847e-06,
      "loss": 0.7267,
      "step": 10974
    },
    {
      "epoch": 0.3162277416008759,
      "grad_norm": 0.7989544868469238,
      "learning_rate": 8.00131000904793e-06,
      "loss": 0.7489,
      "step": 10975
    },
    {
      "epoch": 0.3162565550625252,
      "grad_norm": 0.7960219383239746,
      "learning_rate": 8.000936799599604e-06,
      "loss": 0.7446,
      "step": 10976
    },
    {
      "epoch": 0.3162853685241745,
      "grad_norm": 0.8150654435157776,
      "learning_rate": 8.000563564016115e-06,
      "loss": 0.7432,
      "step": 10977
    },
    {
      "epoch": 0.3163141819858238,
      "grad_norm": 0.7948225736618042,
      "learning_rate": 8.00019030230072e-06,
      "loss": 0.7447,
      "step": 10978
    },
    {
      "epoch": 0.3163429954474731,
      "grad_norm": 0.7623810768127441,
      "learning_rate": 7.999817014456666e-06,
      "loss": 0.7512,
      "step": 10979
    },
    {
      "epoch": 0.31637180890912237,
      "grad_norm": 2.084416389465332,
      "learning_rate": 7.999443700487204e-06,
      "loss": 0.7829,
      "step": 10980
    },
    {
      "epoch": 0.3164006223707716,
      "grad_norm": 0.8015614151954651,
      "learning_rate": 7.999070360395586e-06,
      "loss": 0.7608,
      "step": 10981
    },
    {
      "epoch": 0.3164294358324209,
      "grad_norm": 0.8237153887748718,
      "learning_rate": 7.998696994185063e-06,
      "loss": 0.7789,
      "step": 10982
    },
    {
      "epoch": 0.3164582492940702,
      "grad_norm": 0.7964189648628235,
      "learning_rate": 7.998323601858887e-06,
      "loss": 0.7372,
      "step": 10983
    },
    {
      "epoch": 0.3164870627557195,
      "grad_norm": 0.8134584426879883,
      "learning_rate": 7.99795018342031e-06,
      "loss": 0.7495,
      "step": 10984
    },
    {
      "epoch": 0.31651587621736876,
      "grad_norm": 0.7928292155265808,
      "learning_rate": 7.997576738872583e-06,
      "loss": 0.7552,
      "step": 10985
    },
    {
      "epoch": 0.31654468967901805,
      "grad_norm": 0.771957278251648,
      "learning_rate": 7.99720326821896e-06,
      "loss": 0.7508,
      "step": 10986
    },
    {
      "epoch": 0.31657350314066734,
      "grad_norm": 0.7820997834205627,
      "learning_rate": 7.99682977146269e-06,
      "loss": 0.7628,
      "step": 10987
    },
    {
      "epoch": 0.3166023166023166,
      "grad_norm": 0.821540892124176,
      "learning_rate": 7.996456248607031e-06,
      "loss": 0.757,
      "step": 10988
    },
    {
      "epoch": 0.31663113006396587,
      "grad_norm": 0.7500148415565491,
      "learning_rate": 7.996082699655234e-06,
      "loss": 0.7549,
      "step": 10989
    },
    {
      "epoch": 0.31665994352561516,
      "grad_norm": 0.8291401863098145,
      "learning_rate": 7.99570912461055e-06,
      "loss": 0.7544,
      "step": 10990
    },
    {
      "epoch": 0.31668875698726445,
      "grad_norm": 0.7957682609558105,
      "learning_rate": 7.995335523476235e-06,
      "loss": 0.7482,
      "step": 10991
    },
    {
      "epoch": 0.31671757044891374,
      "grad_norm": 0.7778523564338684,
      "learning_rate": 7.99496189625554e-06,
      "loss": 0.7477,
      "step": 10992
    },
    {
      "epoch": 0.31674638391056303,
      "grad_norm": 0.8240798115730286,
      "learning_rate": 7.994588242951725e-06,
      "loss": 0.755,
      "step": 10993
    },
    {
      "epoch": 0.3167751973722123,
      "grad_norm": 0.7723858952522278,
      "learning_rate": 7.994214563568036e-06,
      "loss": 0.7339,
      "step": 10994
    },
    {
      "epoch": 0.31680401083386156,
      "grad_norm": 0.8150185942649841,
      "learning_rate": 7.993840858107733e-06,
      "loss": 0.7563,
      "step": 10995
    },
    {
      "epoch": 0.31683282429551085,
      "grad_norm": 0.773155152797699,
      "learning_rate": 7.993467126574067e-06,
      "loss": 0.7363,
      "step": 10996
    },
    {
      "epoch": 0.31686163775716014,
      "grad_norm": 0.7961469888687134,
      "learning_rate": 7.993093368970297e-06,
      "loss": 0.7547,
      "step": 10997
    },
    {
      "epoch": 0.31689045121880943,
      "grad_norm": 0.821017861366272,
      "learning_rate": 7.992719585299673e-06,
      "loss": 0.7298,
      "step": 10998
    },
    {
      "epoch": 0.3169192646804587,
      "grad_norm": 0.8343141674995422,
      "learning_rate": 7.992345775565455e-06,
      "loss": 0.7606,
      "step": 10999
    },
    {
      "epoch": 0.316948078142108,
      "grad_norm": 0.7919997572898865,
      "learning_rate": 7.991971939770898e-06,
      "loss": 0.7395,
      "step": 11000
    },
    {
      "epoch": 0.31697689160375725,
      "grad_norm": 0.7938151955604553,
      "learning_rate": 7.991598077919253e-06,
      "loss": 0.751,
      "step": 11001
    },
    {
      "epoch": 0.31700570506540654,
      "grad_norm": 0.7985039949417114,
      "learning_rate": 7.991224190013783e-06,
      "loss": 0.804,
      "step": 11002
    },
    {
      "epoch": 0.31703451852705583,
      "grad_norm": 0.806166410446167,
      "learning_rate": 7.990850276057738e-06,
      "loss": 0.7286,
      "step": 11003
    },
    {
      "epoch": 0.3170633319887051,
      "grad_norm": 0.8047322630882263,
      "learning_rate": 7.990476336054376e-06,
      "loss": 0.7615,
      "step": 11004
    },
    {
      "epoch": 0.3170921454503544,
      "grad_norm": 0.7925033569335938,
      "learning_rate": 7.990102370006957e-06,
      "loss": 0.7584,
      "step": 11005
    },
    {
      "epoch": 0.3171209589120037,
      "grad_norm": 0.8101633787155151,
      "learning_rate": 7.989728377918732e-06,
      "loss": 0.7552,
      "step": 11006
    },
    {
      "epoch": 0.317149772373653,
      "grad_norm": 0.804405689239502,
      "learning_rate": 7.989354359792963e-06,
      "loss": 0.7369,
      "step": 11007
    },
    {
      "epoch": 0.31717858583530223,
      "grad_norm": 0.802935004234314,
      "learning_rate": 7.988980315632907e-06,
      "loss": 0.7492,
      "step": 11008
    },
    {
      "epoch": 0.3172073992969515,
      "grad_norm": 0.8521443009376526,
      "learning_rate": 7.98860624544182e-06,
      "loss": 0.7672,
      "step": 11009
    },
    {
      "epoch": 0.3172362127586008,
      "grad_norm": 0.8456365466117859,
      "learning_rate": 7.988232149222958e-06,
      "loss": 0.7595,
      "step": 11010
    },
    {
      "epoch": 0.3172650262202501,
      "grad_norm": 0.7828723192214966,
      "learning_rate": 7.987858026979585e-06,
      "loss": 0.7288,
      "step": 11011
    },
    {
      "epoch": 0.3172938396818994,
      "grad_norm": 0.7951561808586121,
      "learning_rate": 7.987483878714952e-06,
      "loss": 0.7322,
      "step": 11012
    },
    {
      "epoch": 0.3173226531435487,
      "grad_norm": 0.8041170835494995,
      "learning_rate": 7.987109704432323e-06,
      "loss": 0.743,
      "step": 11013
    },
    {
      "epoch": 0.317351466605198,
      "grad_norm": 0.8080066442489624,
      "learning_rate": 7.986735504134952e-06,
      "loss": 0.7609,
      "step": 11014
    },
    {
      "epoch": 0.3173802800668472,
      "grad_norm": 0.8091317415237427,
      "learning_rate": 7.986361277826102e-06,
      "loss": 0.7621,
      "step": 11015
    },
    {
      "epoch": 0.3174090935284965,
      "grad_norm": 0.7801126837730408,
      "learning_rate": 7.98598702550903e-06,
      "loss": 0.7624,
      "step": 11016
    },
    {
      "epoch": 0.3174379069901458,
      "grad_norm": 0.7707124948501587,
      "learning_rate": 7.985612747186997e-06,
      "loss": 0.7467,
      "step": 11017
    },
    {
      "epoch": 0.3174667204517951,
      "grad_norm": 0.7938570976257324,
      "learning_rate": 7.985238442863262e-06,
      "loss": 0.7509,
      "step": 11018
    },
    {
      "epoch": 0.31749553391344437,
      "grad_norm": 0.8066354990005493,
      "learning_rate": 7.984864112541082e-06,
      "loss": 0.747,
      "step": 11019
    },
    {
      "epoch": 0.31752434737509366,
      "grad_norm": 0.7689207196235657,
      "learning_rate": 7.984489756223722e-06,
      "loss": 0.77,
      "step": 11020
    },
    {
      "epoch": 0.31755316083674295,
      "grad_norm": 0.7915639877319336,
      "learning_rate": 7.984115373914436e-06,
      "loss": 0.741,
      "step": 11021
    },
    {
      "epoch": 0.3175819742983922,
      "grad_norm": 0.8182823657989502,
      "learning_rate": 7.983740965616491e-06,
      "loss": 0.7425,
      "step": 11022
    },
    {
      "epoch": 0.3176107877600415,
      "grad_norm": 0.8063064217567444,
      "learning_rate": 7.983366531333143e-06,
      "loss": 0.7696,
      "step": 11023
    },
    {
      "epoch": 0.31763960122169077,
      "grad_norm": 0.7910730242729187,
      "learning_rate": 7.982992071067656e-06,
      "loss": 0.7524,
      "step": 11024
    },
    {
      "epoch": 0.31766841468334006,
      "grad_norm": 0.7763526439666748,
      "learning_rate": 7.98261758482329e-06,
      "loss": 0.7539,
      "step": 11025
    },
    {
      "epoch": 0.31769722814498935,
      "grad_norm": 0.7883604168891907,
      "learning_rate": 7.982243072603306e-06,
      "loss": 0.7494,
      "step": 11026
    },
    {
      "epoch": 0.31772604160663864,
      "grad_norm": 0.8069026470184326,
      "learning_rate": 7.981868534410967e-06,
      "loss": 0.7763,
      "step": 11027
    },
    {
      "epoch": 0.3177548550682879,
      "grad_norm": 0.8170228004455566,
      "learning_rate": 7.981493970249531e-06,
      "loss": 0.7299,
      "step": 11028
    },
    {
      "epoch": 0.31778366852993717,
      "grad_norm": 0.8216277956962585,
      "learning_rate": 7.981119380122267e-06,
      "loss": 0.7509,
      "step": 11029
    },
    {
      "epoch": 0.31781248199158646,
      "grad_norm": 0.8479229807853699,
      "learning_rate": 7.980744764032431e-06,
      "loss": 0.744,
      "step": 11030
    },
    {
      "epoch": 0.31784129545323575,
      "grad_norm": 0.9036546945571899,
      "learning_rate": 7.980370121983287e-06,
      "loss": 0.7709,
      "step": 11031
    },
    {
      "epoch": 0.31787010891488504,
      "grad_norm": 0.8516362905502319,
      "learning_rate": 7.9799954539781e-06,
      "loss": 0.7819,
      "step": 11032
    },
    {
      "epoch": 0.31789892237653433,
      "grad_norm": 0.8188378214836121,
      "learning_rate": 7.979620760020131e-06,
      "loss": 0.7697,
      "step": 11033
    },
    {
      "epoch": 0.3179277358381836,
      "grad_norm": 0.8042387962341309,
      "learning_rate": 7.979246040112642e-06,
      "loss": 0.7291,
      "step": 11034
    },
    {
      "epoch": 0.31795654929983286,
      "grad_norm": 0.8384959697723389,
      "learning_rate": 7.9788712942589e-06,
      "loss": 0.762,
      "step": 11035
    },
    {
      "epoch": 0.31798536276148215,
      "grad_norm": 0.8833949565887451,
      "learning_rate": 7.978496522462167e-06,
      "loss": 0.7469,
      "step": 11036
    },
    {
      "epoch": 0.31801417622313144,
      "grad_norm": 0.8189323544502258,
      "learning_rate": 7.978121724725708e-06,
      "loss": 0.7204,
      "step": 11037
    },
    {
      "epoch": 0.31804298968478073,
      "grad_norm": 0.8401877284049988,
      "learning_rate": 7.977746901052783e-06,
      "loss": 0.7379,
      "step": 11038
    },
    {
      "epoch": 0.31807180314643,
      "grad_norm": 0.8628679513931274,
      "learning_rate": 7.97737205144666e-06,
      "loss": 0.7695,
      "step": 11039
    },
    {
      "epoch": 0.3181006166080793,
      "grad_norm": 0.816411554813385,
      "learning_rate": 7.976997175910603e-06,
      "loss": 0.7409,
      "step": 11040
    },
    {
      "epoch": 0.3181294300697286,
      "grad_norm": 0.7903165221214294,
      "learning_rate": 7.976622274447876e-06,
      "loss": 0.7562,
      "step": 11041
    },
    {
      "epoch": 0.31815824353137784,
      "grad_norm": 0.7740365862846375,
      "learning_rate": 7.976247347061745e-06,
      "loss": 0.7391,
      "step": 11042
    },
    {
      "epoch": 0.31818705699302713,
      "grad_norm": 0.7991631627082825,
      "learning_rate": 7.975872393755476e-06,
      "loss": 0.7443,
      "step": 11043
    },
    {
      "epoch": 0.3182158704546764,
      "grad_norm": 0.803077220916748,
      "learning_rate": 7.97549741453233e-06,
      "loss": 0.735,
      "step": 11044
    },
    {
      "epoch": 0.3182446839163257,
      "grad_norm": 0.7990317940711975,
      "learning_rate": 7.97512240939558e-06,
      "loss": 0.7415,
      "step": 11045
    },
    {
      "epoch": 0.318273497377975,
      "grad_norm": 0.7869670987129211,
      "learning_rate": 7.974747378348484e-06,
      "loss": 0.7718,
      "step": 11046
    },
    {
      "epoch": 0.3183023108396243,
      "grad_norm": 0.7630481123924255,
      "learning_rate": 7.974372321394315e-06,
      "loss": 0.7445,
      "step": 11047
    },
    {
      "epoch": 0.3183311243012736,
      "grad_norm": 0.7875491380691528,
      "learning_rate": 7.973997238536335e-06,
      "loss": 0.7405,
      "step": 11048
    },
    {
      "epoch": 0.3183599377629228,
      "grad_norm": 0.804770290851593,
      "learning_rate": 7.973622129777812e-06,
      "loss": 0.7438,
      "step": 11049
    },
    {
      "epoch": 0.3183887512245721,
      "grad_norm": 0.8046684861183167,
      "learning_rate": 7.973246995122012e-06,
      "loss": 0.7404,
      "step": 11050
    },
    {
      "epoch": 0.3184175646862214,
      "grad_norm": 0.803697943687439,
      "learning_rate": 7.972871834572203e-06,
      "loss": 0.7618,
      "step": 11051
    },
    {
      "epoch": 0.3184463781478707,
      "grad_norm": 0.7683225870132446,
      "learning_rate": 7.972496648131654e-06,
      "loss": 0.7271,
      "step": 11052
    },
    {
      "epoch": 0.31847519160952,
      "grad_norm": 0.8219700455665588,
      "learning_rate": 7.972121435803629e-06,
      "loss": 0.7577,
      "step": 11053
    },
    {
      "epoch": 0.31850400507116927,
      "grad_norm": 0.8283135890960693,
      "learning_rate": 7.971746197591398e-06,
      "loss": 0.7221,
      "step": 11054
    },
    {
      "epoch": 0.3185328185328185,
      "grad_norm": 0.781808614730835,
      "learning_rate": 7.971370933498227e-06,
      "loss": 0.7675,
      "step": 11055
    },
    {
      "epoch": 0.3185616319944678,
      "grad_norm": 0.7859870791435242,
      "learning_rate": 7.970995643527386e-06,
      "loss": 0.7443,
      "step": 11056
    },
    {
      "epoch": 0.3185904454561171,
      "grad_norm": 0.8323488235473633,
      "learning_rate": 7.970620327682144e-06,
      "loss": 0.748,
      "step": 11057
    },
    {
      "epoch": 0.3186192589177664,
      "grad_norm": 0.8042402267456055,
      "learning_rate": 7.970244985965767e-06,
      "loss": 0.7471,
      "step": 11058
    },
    {
      "epoch": 0.31864807237941567,
      "grad_norm": 0.7824038863182068,
      "learning_rate": 7.969869618381528e-06,
      "loss": 0.7419,
      "step": 11059
    },
    {
      "epoch": 0.31867688584106496,
      "grad_norm": 0.8287235498428345,
      "learning_rate": 7.96949422493269e-06,
      "loss": 0.7693,
      "step": 11060
    },
    {
      "epoch": 0.31870569930271425,
      "grad_norm": 0.9902170896530151,
      "learning_rate": 7.969118805622529e-06,
      "loss": 0.752,
      "step": 11061
    },
    {
      "epoch": 0.3187345127643635,
      "grad_norm": 0.7788481116294861,
      "learning_rate": 7.96874336045431e-06,
      "loss": 0.7637,
      "step": 11062
    },
    {
      "epoch": 0.3187633262260128,
      "grad_norm": 0.8242411017417908,
      "learning_rate": 7.968367889431301e-06,
      "loss": 0.7468,
      "step": 11063
    },
    {
      "epoch": 0.31879213968766207,
      "grad_norm": 0.7757481336593628,
      "learning_rate": 7.967992392556778e-06,
      "loss": 0.7492,
      "step": 11064
    },
    {
      "epoch": 0.31882095314931136,
      "grad_norm": 0.7796785235404968,
      "learning_rate": 7.967616869834007e-06,
      "loss": 0.7406,
      "step": 11065
    },
    {
      "epoch": 0.31884976661096065,
      "grad_norm": 0.7917008996009827,
      "learning_rate": 7.96724132126626e-06,
      "loss": 0.7491,
      "step": 11066
    },
    {
      "epoch": 0.31887858007260994,
      "grad_norm": 0.8186780214309692,
      "learning_rate": 7.96686574685681e-06,
      "loss": 0.7422,
      "step": 11067
    },
    {
      "epoch": 0.31890739353425923,
      "grad_norm": 0.7973890900611877,
      "learning_rate": 7.966490146608921e-06,
      "loss": 0.7697,
      "step": 11068
    },
    {
      "epoch": 0.31893620699590847,
      "grad_norm": 0.8008295893669128,
      "learning_rate": 7.96611452052587e-06,
      "loss": 0.764,
      "step": 11069
    },
    {
      "epoch": 0.31896502045755776,
      "grad_norm": 0.8137802481651306,
      "learning_rate": 7.965738868610927e-06,
      "loss": 0.7448,
      "step": 11070
    },
    {
      "epoch": 0.31899383391920705,
      "grad_norm": 0.7753378748893738,
      "learning_rate": 7.96536319086736e-06,
      "loss": 0.7263,
      "step": 11071
    },
    {
      "epoch": 0.31902264738085634,
      "grad_norm": 0.7729461789131165,
      "learning_rate": 7.964987487298448e-06,
      "loss": 0.7545,
      "step": 11072
    },
    {
      "epoch": 0.31905146084250563,
      "grad_norm": 0.7912499308586121,
      "learning_rate": 7.964611757907456e-06,
      "loss": 0.7477,
      "step": 11073
    },
    {
      "epoch": 0.3190802743041549,
      "grad_norm": 0.8167281150817871,
      "learning_rate": 7.96423600269766e-06,
      "loss": 0.7397,
      "step": 11074
    },
    {
      "epoch": 0.3191090877658042,
      "grad_norm": 0.7800610065460205,
      "learning_rate": 7.963860221672333e-06,
      "loss": 0.7369,
      "step": 11075
    },
    {
      "epoch": 0.31913790122745345,
      "grad_norm": 0.8039007186889648,
      "learning_rate": 7.963484414834746e-06,
      "loss": 0.7262,
      "step": 11076
    },
    {
      "epoch": 0.31916671468910274,
      "grad_norm": 0.8105089068412781,
      "learning_rate": 7.963108582188171e-06,
      "loss": 0.7422,
      "step": 11077
    },
    {
      "epoch": 0.31919552815075203,
      "grad_norm": 0.8472611308097839,
      "learning_rate": 7.962732723735882e-06,
      "loss": 0.7545,
      "step": 11078
    },
    {
      "epoch": 0.3192243416124013,
      "grad_norm": 0.8467971086502075,
      "learning_rate": 7.962356839481154e-06,
      "loss": 0.7607,
      "step": 11079
    },
    {
      "epoch": 0.3192531550740506,
      "grad_norm": 0.8075046539306641,
      "learning_rate": 7.961980929427257e-06,
      "loss": 0.7337,
      "step": 11080
    },
    {
      "epoch": 0.3192819685356999,
      "grad_norm": 0.7584012150764465,
      "learning_rate": 7.961604993577468e-06,
      "loss": 0.7484,
      "step": 11081
    },
    {
      "epoch": 0.31931078199734914,
      "grad_norm": 0.8172497749328613,
      "learning_rate": 7.96122903193506e-06,
      "loss": 0.7546,
      "step": 11082
    },
    {
      "epoch": 0.3193395954589984,
      "grad_norm": 0.8444250226020813,
      "learning_rate": 7.96085304450331e-06,
      "loss": 0.7273,
      "step": 11083
    },
    {
      "epoch": 0.3193684089206477,
      "grad_norm": 0.8431944847106934,
      "learning_rate": 7.960477031285487e-06,
      "loss": 0.7485,
      "step": 11084
    },
    {
      "epoch": 0.319397222382297,
      "grad_norm": 0.809755802154541,
      "learning_rate": 7.960100992284868e-06,
      "loss": 0.7512,
      "step": 11085
    },
    {
      "epoch": 0.3194260358439463,
      "grad_norm": 0.887972354888916,
      "learning_rate": 7.959724927504731e-06,
      "loss": 0.761,
      "step": 11086
    },
    {
      "epoch": 0.3194548493055956,
      "grad_norm": 0.8370357155799866,
      "learning_rate": 7.959348836948346e-06,
      "loss": 0.7433,
      "step": 11087
    },
    {
      "epoch": 0.3194836627672449,
      "grad_norm": 0.8221194744110107,
      "learning_rate": 7.958972720618992e-06,
      "loss": 0.7446,
      "step": 11088
    },
    {
      "epoch": 0.3195124762288941,
      "grad_norm": 0.8513370752334595,
      "learning_rate": 7.958596578519941e-06,
      "loss": 0.7551,
      "step": 11089
    },
    {
      "epoch": 0.3195412896905434,
      "grad_norm": 0.8568114638328552,
      "learning_rate": 7.958220410654474e-06,
      "loss": 0.7609,
      "step": 11090
    },
    {
      "epoch": 0.3195701031521927,
      "grad_norm": 0.8181893229484558,
      "learning_rate": 7.957844217025866e-06,
      "loss": 0.7633,
      "step": 11091
    },
    {
      "epoch": 0.319598916613842,
      "grad_norm": 0.7920733690261841,
      "learning_rate": 7.957467997637388e-06,
      "loss": 0.7367,
      "step": 11092
    },
    {
      "epoch": 0.3196277300754913,
      "grad_norm": 0.8126123547554016,
      "learning_rate": 7.957091752492321e-06,
      "loss": 0.7506,
      "step": 11093
    },
    {
      "epoch": 0.31965654353714057,
      "grad_norm": 0.8425920605659485,
      "learning_rate": 7.956715481593942e-06,
      "loss": 0.7554,
      "step": 11094
    },
    {
      "epoch": 0.31968535699878986,
      "grad_norm": 0.7803601622581482,
      "learning_rate": 7.956339184945526e-06,
      "loss": 0.7336,
      "step": 11095
    },
    {
      "epoch": 0.3197141704604391,
      "grad_norm": 0.7967623472213745,
      "learning_rate": 7.955962862550351e-06,
      "loss": 0.7407,
      "step": 11096
    },
    {
      "epoch": 0.3197429839220884,
      "grad_norm": 0.7926453351974487,
      "learning_rate": 7.955586514411695e-06,
      "loss": 0.7137,
      "step": 11097
    },
    {
      "epoch": 0.3197717973837377,
      "grad_norm": 0.8153645396232605,
      "learning_rate": 7.955210140532834e-06,
      "loss": 0.7564,
      "step": 11098
    },
    {
      "epoch": 0.31980061084538697,
      "grad_norm": 0.7959955334663391,
      "learning_rate": 7.954833740917047e-06,
      "loss": 0.7243,
      "step": 11099
    },
    {
      "epoch": 0.31982942430703626,
      "grad_norm": 0.8096619844436646,
      "learning_rate": 7.954457315567614e-06,
      "loss": 0.7397,
      "step": 11100
    },
    {
      "epoch": 0.31985823776868555,
      "grad_norm": 0.8308321833610535,
      "learning_rate": 7.954080864487807e-06,
      "loss": 0.7225,
      "step": 11101
    },
    {
      "epoch": 0.31988705123033484,
      "grad_norm": 0.7963953614234924,
      "learning_rate": 7.953704387680913e-06,
      "loss": 0.7458,
      "step": 11102
    },
    {
      "epoch": 0.3199158646919841,
      "grad_norm": 0.7763944268226624,
      "learning_rate": 7.953327885150203e-06,
      "loss": 0.7404,
      "step": 11103
    },
    {
      "epoch": 0.31994467815363337,
      "grad_norm": 0.8183842301368713,
      "learning_rate": 7.952951356898961e-06,
      "loss": 0.74,
      "step": 11104
    },
    {
      "epoch": 0.31997349161528266,
      "grad_norm": 0.8118444085121155,
      "learning_rate": 7.952574802930465e-06,
      "loss": 0.7614,
      "step": 11105
    },
    {
      "epoch": 0.32000230507693195,
      "grad_norm": 0.8056778311729431,
      "learning_rate": 7.952198223247992e-06,
      "loss": 0.7378,
      "step": 11106
    },
    {
      "epoch": 0.32003111853858124,
      "grad_norm": 0.8170649409294128,
      "learning_rate": 7.951821617854825e-06,
      "loss": 0.7657,
      "step": 11107
    },
    {
      "epoch": 0.32005993200023053,
      "grad_norm": 0.7867599129676819,
      "learning_rate": 7.951444986754244e-06,
      "loss": 0.7351,
      "step": 11108
    },
    {
      "epoch": 0.32008874546187976,
      "grad_norm": 0.7972534894943237,
      "learning_rate": 7.951068329949525e-06,
      "loss": 0.758,
      "step": 11109
    },
    {
      "epoch": 0.32011755892352906,
      "grad_norm": 0.84911048412323,
      "learning_rate": 7.950691647443952e-06,
      "loss": 0.7573,
      "step": 11110
    },
    {
      "epoch": 0.32014637238517835,
      "grad_norm": 0.8071699142456055,
      "learning_rate": 7.950314939240804e-06,
      "loss": 0.7474,
      "step": 11111
    },
    {
      "epoch": 0.32017518584682764,
      "grad_norm": 0.7979209423065186,
      "learning_rate": 7.949938205343362e-06,
      "loss": 0.7467,
      "step": 11112
    },
    {
      "epoch": 0.3202039993084769,
      "grad_norm": 0.844475507736206,
      "learning_rate": 7.949561445754908e-06,
      "loss": 0.7266,
      "step": 11113
    },
    {
      "epoch": 0.3202328127701262,
      "grad_norm": 0.8519406318664551,
      "learning_rate": 7.949184660478721e-06,
      "loss": 0.7474,
      "step": 11114
    },
    {
      "epoch": 0.3202616262317755,
      "grad_norm": 0.7954272031784058,
      "learning_rate": 7.948807849518084e-06,
      "loss": 0.7655,
      "step": 11115
    },
    {
      "epoch": 0.32029043969342474,
      "grad_norm": 0.8232630491256714,
      "learning_rate": 7.94843101287628e-06,
      "loss": 0.7484,
      "step": 11116
    },
    {
      "epoch": 0.32031925315507404,
      "grad_norm": 0.8775096535682678,
      "learning_rate": 7.948054150556586e-06,
      "loss": 0.7651,
      "step": 11117
    },
    {
      "epoch": 0.3203480666167233,
      "grad_norm": 0.8967607617378235,
      "learning_rate": 7.94767726256229e-06,
      "loss": 0.7422,
      "step": 11118
    },
    {
      "epoch": 0.3203768800783726,
      "grad_norm": 0.8020403981208801,
      "learning_rate": 7.947300348896672e-06,
      "loss": 0.7471,
      "step": 11119
    },
    {
      "epoch": 0.3204056935400219,
      "grad_norm": 0.8650214076042175,
      "learning_rate": 7.946923409563012e-06,
      "loss": 0.7822,
      "step": 11120
    },
    {
      "epoch": 0.3204345070016712,
      "grad_norm": 0.8617588877677917,
      "learning_rate": 7.946546444564595e-06,
      "loss": 0.759,
      "step": 11121
    },
    {
      "epoch": 0.3204633204633205,
      "grad_norm": 6.919222831726074,
      "learning_rate": 7.946169453904706e-06,
      "loss": 0.7811,
      "step": 11122
    },
    {
      "epoch": 0.3204921339249697,
      "grad_norm": 0.8234766721725464,
      "learning_rate": 7.945792437586625e-06,
      "loss": 0.7765,
      "step": 11123
    },
    {
      "epoch": 0.320520947386619,
      "grad_norm": 0.7935124635696411,
      "learning_rate": 7.945415395613636e-06,
      "loss": 0.7486,
      "step": 11124
    },
    {
      "epoch": 0.3205497608482683,
      "grad_norm": 0.8062594532966614,
      "learning_rate": 7.945038327989024e-06,
      "loss": 0.7505,
      "step": 11125
    },
    {
      "epoch": 0.3205785743099176,
      "grad_norm": 0.82124924659729,
      "learning_rate": 7.944661234716072e-06,
      "loss": 0.7284,
      "step": 11126
    },
    {
      "epoch": 0.3206073877715669,
      "grad_norm": 0.8306527733802795,
      "learning_rate": 7.944284115798065e-06,
      "loss": 0.7567,
      "step": 11127
    },
    {
      "epoch": 0.3206362012332162,
      "grad_norm": 0.8601223230361938,
      "learning_rate": 7.943906971238286e-06,
      "loss": 0.7819,
      "step": 11128
    },
    {
      "epoch": 0.3206650146948654,
      "grad_norm": 0.7969762086868286,
      "learning_rate": 7.94352980104002e-06,
      "loss": 0.7688,
      "step": 11129
    },
    {
      "epoch": 0.3206938281565147,
      "grad_norm": 0.8772279024124146,
      "learning_rate": 7.943152605206551e-06,
      "loss": 0.7436,
      "step": 11130
    },
    {
      "epoch": 0.320722641618164,
      "grad_norm": 0.8343045115470886,
      "learning_rate": 7.942775383741166e-06,
      "loss": 0.755,
      "step": 11131
    },
    {
      "epoch": 0.3207514550798133,
      "grad_norm": 0.8492089509963989,
      "learning_rate": 7.94239813664715e-06,
      "loss": 0.7505,
      "step": 11132
    },
    {
      "epoch": 0.3207802685414626,
      "grad_norm": 0.8787469267845154,
      "learning_rate": 7.942020863927784e-06,
      "loss": 0.7557,
      "step": 11133
    },
    {
      "epoch": 0.32080908200311187,
      "grad_norm": 0.8329540491104126,
      "learning_rate": 7.941643565586362e-06,
      "loss": 0.7617,
      "step": 11134
    },
    {
      "epoch": 0.32083789546476116,
      "grad_norm": 0.8352074027061462,
      "learning_rate": 7.941266241626162e-06,
      "loss": 0.7533,
      "step": 11135
    },
    {
      "epoch": 0.3208667089264104,
      "grad_norm": 0.852798581123352,
      "learning_rate": 7.940888892050473e-06,
      "loss": 0.7437,
      "step": 11136
    },
    {
      "epoch": 0.3208955223880597,
      "grad_norm": 0.8274326920509338,
      "learning_rate": 7.94051151686258e-06,
      "loss": 0.7466,
      "step": 11137
    },
    {
      "epoch": 0.320924335849709,
      "grad_norm": 0.8497942090034485,
      "learning_rate": 7.940134116065776e-06,
      "loss": 0.7459,
      "step": 11138
    },
    {
      "epoch": 0.32095314931135827,
      "grad_norm": 0.8021315336227417,
      "learning_rate": 7.93975668966334e-06,
      "loss": 0.7402,
      "step": 11139
    },
    {
      "epoch": 0.32098196277300756,
      "grad_norm": 0.8187866806983948,
      "learning_rate": 7.939379237658563e-06,
      "loss": 0.7569,
      "step": 11140
    },
    {
      "epoch": 0.32101077623465685,
      "grad_norm": 0.8217072486877441,
      "learning_rate": 7.93900176005473e-06,
      "loss": 0.7677,
      "step": 11141
    },
    {
      "epoch": 0.32103958969630614,
      "grad_norm": 0.8281592726707458,
      "learning_rate": 7.93862425685513e-06,
      "loss": 0.7397,
      "step": 11142
    },
    {
      "epoch": 0.3210684031579554,
      "grad_norm": 0.8451306223869324,
      "learning_rate": 7.93824672806305e-06,
      "loss": 0.7436,
      "step": 11143
    },
    {
      "epoch": 0.32109721661960466,
      "grad_norm": 0.823359489440918,
      "learning_rate": 7.937869173681779e-06,
      "loss": 0.7646,
      "step": 11144
    },
    {
      "epoch": 0.32112603008125395,
      "grad_norm": 0.8168091773986816,
      "learning_rate": 7.937491593714604e-06,
      "loss": 0.7593,
      "step": 11145
    },
    {
      "epoch": 0.32115484354290325,
      "grad_norm": 0.8485236167907715,
      "learning_rate": 7.937113988164814e-06,
      "loss": 0.746,
      "step": 11146
    },
    {
      "epoch": 0.32118365700455254,
      "grad_norm": 0.8063970804214478,
      "learning_rate": 7.936736357035695e-06,
      "loss": 0.7511,
      "step": 11147
    },
    {
      "epoch": 0.3212124704662018,
      "grad_norm": 0.8211365342140198,
      "learning_rate": 7.936358700330542e-06,
      "loss": 0.7408,
      "step": 11148
    },
    {
      "epoch": 0.3212412839278511,
      "grad_norm": 0.8528679609298706,
      "learning_rate": 7.935981018052637e-06,
      "loss": 0.7436,
      "step": 11149
    },
    {
      "epoch": 0.32127009738950035,
      "grad_norm": 0.8195213675498962,
      "learning_rate": 7.935603310205274e-06,
      "loss": 0.757,
      "step": 11150
    },
    {
      "epoch": 0.32129891085114964,
      "grad_norm": 0.816724419593811,
      "learning_rate": 7.935225576791739e-06,
      "loss": 0.741,
      "step": 11151
    },
    {
      "epoch": 0.32132772431279893,
      "grad_norm": 0.8030626177787781,
      "learning_rate": 7.934847817815323e-06,
      "loss": 0.7461,
      "step": 11152
    },
    {
      "epoch": 0.3213565377744482,
      "grad_norm": 0.8261647820472717,
      "learning_rate": 7.934470033279318e-06,
      "loss": 0.7649,
      "step": 11153
    },
    {
      "epoch": 0.3213853512360975,
      "grad_norm": 0.8418678641319275,
      "learning_rate": 7.93409222318701e-06,
      "loss": 0.7736,
      "step": 11154
    },
    {
      "epoch": 0.3214141646977468,
      "grad_norm": 7.272024154663086,
      "learning_rate": 7.933714387541695e-06,
      "loss": 0.772,
      "step": 11155
    },
    {
      "epoch": 0.32144297815939604,
      "grad_norm": 0.8322027325630188,
      "learning_rate": 7.933336526346657e-06,
      "loss": 0.7634,
      "step": 11156
    },
    {
      "epoch": 0.32147179162104533,
      "grad_norm": 0.8549715876579285,
      "learning_rate": 7.932958639605192e-06,
      "loss": 0.7674,
      "step": 11157
    },
    {
      "epoch": 0.3215006050826946,
      "grad_norm": 0.8849143385887146,
      "learning_rate": 7.932580727320591e-06,
      "loss": 0.765,
      "step": 11158
    },
    {
      "epoch": 0.3215294185443439,
      "grad_norm": 0.9023061394691467,
      "learning_rate": 7.932202789496141e-06,
      "loss": 0.7506,
      "step": 11159
    },
    {
      "epoch": 0.3215582320059932,
      "grad_norm": 0.879815399646759,
      "learning_rate": 7.931824826135137e-06,
      "loss": 0.7411,
      "step": 11160
    },
    {
      "epoch": 0.3215870454676425,
      "grad_norm": 0.8984388709068298,
      "learning_rate": 7.931446837240868e-06,
      "loss": 0.7655,
      "step": 11161
    },
    {
      "epoch": 0.3216158589292918,
      "grad_norm": 0.9786787629127502,
      "learning_rate": 7.931068822816628e-06,
      "loss": 0.7459,
      "step": 11162
    },
    {
      "epoch": 0.321644672390941,
      "grad_norm": 0.8712018132209778,
      "learning_rate": 7.930690782865708e-06,
      "loss": 0.7651,
      "step": 11163
    },
    {
      "epoch": 0.3216734858525903,
      "grad_norm": 1.1623696088790894,
      "learning_rate": 7.930312717391402e-06,
      "loss": 0.7593,
      "step": 11164
    },
    {
      "epoch": 0.3217022993142396,
      "grad_norm": 0.8276840448379517,
      "learning_rate": 7.929934626397002e-06,
      "loss": 0.7551,
      "step": 11165
    },
    {
      "epoch": 0.3217311127758889,
      "grad_norm": 0.8128806352615356,
      "learning_rate": 7.929556509885798e-06,
      "loss": 0.7583,
      "step": 11166
    },
    {
      "epoch": 0.3217599262375382,
      "grad_norm": 0.8034794926643372,
      "learning_rate": 7.929178367861088e-06,
      "loss": 0.742,
      "step": 11167
    },
    {
      "epoch": 0.3217887396991875,
      "grad_norm": 3.2017300128936768,
      "learning_rate": 7.928800200326161e-06,
      "loss": 0.7875,
      "step": 11168
    },
    {
      "epoch": 0.32181755316083677,
      "grad_norm": 0.8156041502952576,
      "learning_rate": 7.928422007284313e-06,
      "loss": 0.75,
      "step": 11169
    },
    {
      "epoch": 0.321846366622486,
      "grad_norm": 0.8329769968986511,
      "learning_rate": 7.928043788738836e-06,
      "loss": 0.7551,
      "step": 11170
    },
    {
      "epoch": 0.3218751800841353,
      "grad_norm": 0.8189477920532227,
      "learning_rate": 7.927665544693024e-06,
      "loss": 0.7681,
      "step": 11171
    },
    {
      "epoch": 0.3219039935457846,
      "grad_norm": 0.8088990449905396,
      "learning_rate": 7.927287275150171e-06,
      "loss": 0.7735,
      "step": 11172
    },
    {
      "epoch": 0.3219328070074339,
      "grad_norm": 0.8034080862998962,
      "learning_rate": 7.926908980113574e-06,
      "loss": 0.7691,
      "step": 11173
    },
    {
      "epoch": 0.32196162046908317,
      "grad_norm": 0.8660274147987366,
      "learning_rate": 7.926530659586524e-06,
      "loss": 0.78,
      "step": 11174
    },
    {
      "epoch": 0.32199043393073246,
      "grad_norm": 0.8084828853607178,
      "learning_rate": 7.926152313572318e-06,
      "loss": 0.7269,
      "step": 11175
    },
    {
      "epoch": 0.32201924739238175,
      "grad_norm": 0.8136535882949829,
      "learning_rate": 7.925773942074251e-06,
      "loss": 0.7577,
      "step": 11176
    },
    {
      "epoch": 0.322048060854031,
      "grad_norm": 0.7521231174468994,
      "learning_rate": 7.925395545095617e-06,
      "loss": 0.7452,
      "step": 11177
    },
    {
      "epoch": 0.3220768743156803,
      "grad_norm": 0.8053721189498901,
      "learning_rate": 7.925017122639713e-06,
      "loss": 0.7501,
      "step": 11178
    },
    {
      "epoch": 0.32210568777732956,
      "grad_norm": 0.8125750422477722,
      "learning_rate": 7.924638674709833e-06,
      "loss": 0.7291,
      "step": 11179
    },
    {
      "epoch": 0.32213450123897885,
      "grad_norm": 0.818817675113678,
      "learning_rate": 7.924260201309273e-06,
      "loss": 0.7373,
      "step": 11180
    },
    {
      "epoch": 0.32216331470062815,
      "grad_norm": 0.8173324465751648,
      "learning_rate": 7.923881702441332e-06,
      "loss": 0.7558,
      "step": 11181
    },
    {
      "epoch": 0.32219212816227744,
      "grad_norm": 0.8178030848503113,
      "learning_rate": 7.923503178109302e-06,
      "loss": 0.7724,
      "step": 11182
    },
    {
      "epoch": 0.32222094162392667,
      "grad_norm": 0.8124175667762756,
      "learning_rate": 7.923124628316483e-06,
      "loss": 0.7347,
      "step": 11183
    },
    {
      "epoch": 0.32224975508557596,
      "grad_norm": 0.7905527353286743,
      "learning_rate": 7.92274605306617e-06,
      "loss": 0.756,
      "step": 11184
    },
    {
      "epoch": 0.32227856854722525,
      "grad_norm": 0.8305066227912903,
      "learning_rate": 7.922367452361661e-06,
      "loss": 0.7493,
      "step": 11185
    },
    {
      "epoch": 0.32230738200887454,
      "grad_norm": 0.8531407713890076,
      "learning_rate": 7.921988826206252e-06,
      "loss": 0.7541,
      "step": 11186
    },
    {
      "epoch": 0.32233619547052383,
      "grad_norm": 0.7908920049667358,
      "learning_rate": 7.921610174603243e-06,
      "loss": 0.758,
      "step": 11187
    },
    {
      "epoch": 0.3223650089321731,
      "grad_norm": 0.8220016956329346,
      "learning_rate": 7.921231497555928e-06,
      "loss": 0.73,
      "step": 11188
    },
    {
      "epoch": 0.3223938223938224,
      "grad_norm": 0.8219693303108215,
      "learning_rate": 7.920852795067607e-06,
      "loss": 0.7591,
      "step": 11189
    },
    {
      "epoch": 0.32242263585547165,
      "grad_norm": 0.8089961409568787,
      "learning_rate": 7.92047406714158e-06,
      "loss": 0.7756,
      "step": 11190
    },
    {
      "epoch": 0.32245144931712094,
      "grad_norm": 0.8473843336105347,
      "learning_rate": 7.920095313781141e-06,
      "loss": 0.7529,
      "step": 11191
    },
    {
      "epoch": 0.32248026277877023,
      "grad_norm": 0.8416839241981506,
      "learning_rate": 7.919716534989593e-06,
      "loss": 0.7618,
      "step": 11192
    },
    {
      "epoch": 0.3225090762404195,
      "grad_norm": 0.8133653998374939,
      "learning_rate": 7.919337730770228e-06,
      "loss": 0.7598,
      "step": 11193
    },
    {
      "epoch": 0.3225378897020688,
      "grad_norm": 0.8007431030273438,
      "learning_rate": 7.918958901126355e-06,
      "loss": 0.7407,
      "step": 11194
    },
    {
      "epoch": 0.3225667031637181,
      "grad_norm": 0.8134908676147461,
      "learning_rate": 7.918580046061265e-06,
      "loss": 0.7522,
      "step": 11195
    },
    {
      "epoch": 0.3225955166253674,
      "grad_norm": 0.8035034537315369,
      "learning_rate": 7.91820116557826e-06,
      "loss": 0.741,
      "step": 11196
    },
    {
      "epoch": 0.32262433008701663,
      "grad_norm": 0.7704001665115356,
      "learning_rate": 7.917822259680643e-06,
      "loss": 0.7621,
      "step": 11197
    },
    {
      "epoch": 0.3226531435486659,
      "grad_norm": 0.78178471326828,
      "learning_rate": 7.917443328371708e-06,
      "loss": 0.741,
      "step": 11198
    },
    {
      "epoch": 0.3226819570103152,
      "grad_norm": 0.7805017828941345,
      "learning_rate": 7.91706437165476e-06,
      "loss": 0.7466,
      "step": 11199
    },
    {
      "epoch": 0.3227107704719645,
      "grad_norm": 0.7958698868751526,
      "learning_rate": 7.916685389533096e-06,
      "loss": 0.7567,
      "step": 11200
    },
    {
      "epoch": 0.3227395839336138,
      "grad_norm": 0.7880593538284302,
      "learning_rate": 7.916306382010017e-06,
      "loss": 0.7367,
      "step": 11201
    },
    {
      "epoch": 0.3227683973952631,
      "grad_norm": 0.831623911857605,
      "learning_rate": 7.915927349088826e-06,
      "loss": 0.7583,
      "step": 11202
    },
    {
      "epoch": 0.3227972108569124,
      "grad_norm": 0.7862048149108887,
      "learning_rate": 7.915548290772823e-06,
      "loss": 0.7482,
      "step": 11203
    },
    {
      "epoch": 0.3228260243185616,
      "grad_norm": 0.7946721315383911,
      "learning_rate": 7.915169207065307e-06,
      "loss": 0.7852,
      "step": 11204
    },
    {
      "epoch": 0.3228548377802109,
      "grad_norm": 0.8260758519172668,
      "learning_rate": 7.914790097969584e-06,
      "loss": 0.7633,
      "step": 11205
    },
    {
      "epoch": 0.3228836512418602,
      "grad_norm": 0.9055474400520325,
      "learning_rate": 7.914410963488952e-06,
      "loss": 0.7553,
      "step": 11206
    },
    {
      "epoch": 0.3229124647035095,
      "grad_norm": 0.8081784248352051,
      "learning_rate": 7.914031803626712e-06,
      "loss": 0.7688,
      "step": 11207
    },
    {
      "epoch": 0.3229412781651588,
      "grad_norm": 0.823360800743103,
      "learning_rate": 7.91365261838617e-06,
      "loss": 0.7541,
      "step": 11208
    },
    {
      "epoch": 0.32297009162680806,
      "grad_norm": 0.7800561189651489,
      "learning_rate": 7.913273407770625e-06,
      "loss": 0.7195,
      "step": 11209
    },
    {
      "epoch": 0.3229989050884573,
      "grad_norm": 0.7969197630882263,
      "learning_rate": 7.912894171783383e-06,
      "loss": 0.7429,
      "step": 11210
    },
    {
      "epoch": 0.3230277185501066,
      "grad_norm": 1.462323784828186,
      "learning_rate": 7.912514910427742e-06,
      "loss": 0.776,
      "step": 11211
    },
    {
      "epoch": 0.3230565320117559,
      "grad_norm": 0.7773100137710571,
      "learning_rate": 7.91213562370701e-06,
      "loss": 0.7242,
      "step": 11212
    },
    {
      "epoch": 0.32308534547340517,
      "grad_norm": 0.8124592304229736,
      "learning_rate": 7.911756311624486e-06,
      "loss": 0.764,
      "step": 11213
    },
    {
      "epoch": 0.32311415893505446,
      "grad_norm": 0.8008064031600952,
      "learning_rate": 7.911376974183475e-06,
      "loss": 0.748,
      "step": 11214
    },
    {
      "epoch": 0.32314297239670375,
      "grad_norm": 0.7896736860275269,
      "learning_rate": 7.910997611387281e-06,
      "loss": 0.7442,
      "step": 11215
    },
    {
      "epoch": 0.32317178585835304,
      "grad_norm": 0.8151198625564575,
      "learning_rate": 7.910618223239209e-06,
      "loss": 0.715,
      "step": 11216
    },
    {
      "epoch": 0.3232005993200023,
      "grad_norm": 0.7982407212257385,
      "learning_rate": 7.91023880974256e-06,
      "loss": 0.7681,
      "step": 11217
    },
    {
      "epoch": 0.32322941278165157,
      "grad_norm": 0.7903236150741577,
      "learning_rate": 7.909859370900642e-06,
      "loss": 0.7431,
      "step": 11218
    },
    {
      "epoch": 0.32325822624330086,
      "grad_norm": 0.8553128838539124,
      "learning_rate": 7.909479906716758e-06,
      "loss": 0.7519,
      "step": 11219
    },
    {
      "epoch": 0.32328703970495015,
      "grad_norm": 0.7974734306335449,
      "learning_rate": 7.90910041719421e-06,
      "loss": 0.7597,
      "step": 11220
    },
    {
      "epoch": 0.32331585316659944,
      "grad_norm": 0.8051288723945618,
      "learning_rate": 7.908720902336307e-06,
      "loss": 0.7611,
      "step": 11221
    },
    {
      "epoch": 0.32334466662824873,
      "grad_norm": 0.7892063856124878,
      "learning_rate": 7.908341362146352e-06,
      "loss": 0.7582,
      "step": 11222
    },
    {
      "epoch": 0.323373480089898,
      "grad_norm": 0.7885623574256897,
      "learning_rate": 7.90796179662765e-06,
      "loss": 0.7403,
      "step": 11223
    },
    {
      "epoch": 0.32340229355154726,
      "grad_norm": 0.7940977215766907,
      "learning_rate": 7.90758220578351e-06,
      "loss": 0.7453,
      "step": 11224
    },
    {
      "epoch": 0.32343110701319655,
      "grad_norm": 0.8491561412811279,
      "learning_rate": 7.907202589617234e-06,
      "loss": 0.7545,
      "step": 11225
    },
    {
      "epoch": 0.32345992047484584,
      "grad_norm": 0.8336657881736755,
      "learning_rate": 7.90682294813213e-06,
      "loss": 0.734,
      "step": 11226
    },
    {
      "epoch": 0.32348873393649513,
      "grad_norm": 0.8075718283653259,
      "learning_rate": 7.906443281331504e-06,
      "loss": 0.7615,
      "step": 11227
    },
    {
      "epoch": 0.3235175473981444,
      "grad_norm": 0.7802321910858154,
      "learning_rate": 7.906063589218664e-06,
      "loss": 0.7545,
      "step": 11228
    },
    {
      "epoch": 0.3235463608597937,
      "grad_norm": 0.8423953056335449,
      "learning_rate": 7.905683871796912e-06,
      "loss": 0.7479,
      "step": 11229
    },
    {
      "epoch": 0.323575174321443,
      "grad_norm": 0.7933407425880432,
      "learning_rate": 7.90530412906956e-06,
      "loss": 0.7196,
      "step": 11230
    },
    {
      "epoch": 0.32360398778309224,
      "grad_norm": 0.8168944716453552,
      "learning_rate": 7.904924361039913e-06,
      "loss": 0.7527,
      "step": 11231
    },
    {
      "epoch": 0.32363280124474153,
      "grad_norm": 0.8501297235488892,
      "learning_rate": 7.904544567711279e-06,
      "loss": 0.7706,
      "step": 11232
    },
    {
      "epoch": 0.3236616147063908,
      "grad_norm": 0.8292848467826843,
      "learning_rate": 7.904164749086964e-06,
      "loss": 0.7775,
      "step": 11233
    },
    {
      "epoch": 0.3236904281680401,
      "grad_norm": 0.8010437488555908,
      "learning_rate": 7.903784905170277e-06,
      "loss": 0.7326,
      "step": 11234
    },
    {
      "epoch": 0.3237192416296894,
      "grad_norm": 0.8144823908805847,
      "learning_rate": 7.90340503596453e-06,
      "loss": 0.7607,
      "step": 11235
    },
    {
      "epoch": 0.3237480550913387,
      "grad_norm": 0.8421857357025146,
      "learning_rate": 7.903025141473024e-06,
      "loss": 0.7587,
      "step": 11236
    },
    {
      "epoch": 0.32377686855298793,
      "grad_norm": 0.8159623146057129,
      "learning_rate": 7.90264522169907e-06,
      "loss": 0.7362,
      "step": 11237
    },
    {
      "epoch": 0.3238056820146372,
      "grad_norm": 0.7952765226364136,
      "learning_rate": 7.90226527664598e-06,
      "loss": 0.7472,
      "step": 11238
    },
    {
      "epoch": 0.3238344954762865,
      "grad_norm": 0.839614748954773,
      "learning_rate": 7.90188530631706e-06,
      "loss": 0.7603,
      "step": 11239
    },
    {
      "epoch": 0.3238633089379358,
      "grad_norm": 0.817672610282898,
      "learning_rate": 7.90150531071562e-06,
      "loss": 0.7726,
      "step": 11240
    },
    {
      "epoch": 0.3238921223995851,
      "grad_norm": 0.8434412479400635,
      "learning_rate": 7.901125289844969e-06,
      "loss": 0.7212,
      "step": 11241
    },
    {
      "epoch": 0.3239209358612344,
      "grad_norm": 0.7872346639633179,
      "learning_rate": 7.900745243708416e-06,
      "loss": 0.7435,
      "step": 11242
    },
    {
      "epoch": 0.3239497493228837,
      "grad_norm": 0.8435353636741638,
      "learning_rate": 7.900365172309272e-06,
      "loss": 0.7452,
      "step": 11243
    },
    {
      "epoch": 0.3239785627845329,
      "grad_norm": 0.8102597594261169,
      "learning_rate": 7.899985075650848e-06,
      "loss": 0.7434,
      "step": 11244
    },
    {
      "epoch": 0.3240073762461822,
      "grad_norm": 0.819255530834198,
      "learning_rate": 7.899604953736451e-06,
      "loss": 0.74,
      "step": 11245
    },
    {
      "epoch": 0.3240361897078315,
      "grad_norm": 0.8513647317886353,
      "learning_rate": 7.899224806569395e-06,
      "loss": 0.7714,
      "step": 11246
    },
    {
      "epoch": 0.3240650031694808,
      "grad_norm": 0.8356141448020935,
      "learning_rate": 7.898844634152988e-06,
      "loss": 0.737,
      "step": 11247
    },
    {
      "epoch": 0.32409381663113007,
      "grad_norm": 0.8117710947990417,
      "learning_rate": 7.898464436490542e-06,
      "loss": 0.7637,
      "step": 11248
    },
    {
      "epoch": 0.32412263009277936,
      "grad_norm": 0.8141085505485535,
      "learning_rate": 7.89808421358537e-06,
      "loss": 0.7398,
      "step": 11249
    },
    {
      "epoch": 0.32415144355442865,
      "grad_norm": 0.8284193873405457,
      "learning_rate": 7.89770396544078e-06,
      "loss": 0.7464,
      "step": 11250
    },
    {
      "epoch": 0.3241802570160779,
      "grad_norm": 0.8446379899978638,
      "learning_rate": 7.897323692060087e-06,
      "loss": 0.7432,
      "step": 11251
    },
    {
      "epoch": 0.3242090704777272,
      "grad_norm": 0.8081342577934265,
      "learning_rate": 7.896943393446597e-06,
      "loss": 0.7619,
      "step": 11252
    },
    {
      "epoch": 0.32423788393937647,
      "grad_norm": 0.7810166478157043,
      "learning_rate": 7.896563069603628e-06,
      "loss": 0.7647,
      "step": 11253
    },
    {
      "epoch": 0.32426669740102576,
      "grad_norm": 0.7847560048103333,
      "learning_rate": 7.89618272053449e-06,
      "loss": 0.7303,
      "step": 11254
    },
    {
      "epoch": 0.32429551086267505,
      "grad_norm": 0.8504945635795593,
      "learning_rate": 7.895802346242497e-06,
      "loss": 0.7449,
      "step": 11255
    },
    {
      "epoch": 0.32432432432432434,
      "grad_norm": 0.8040239810943604,
      "learning_rate": 7.89542194673096e-06,
      "loss": 0.7362,
      "step": 11256
    },
    {
      "epoch": 0.32435313778597363,
      "grad_norm": 0.8376503586769104,
      "learning_rate": 7.89504152200319e-06,
      "loss": 0.7615,
      "step": 11257
    },
    {
      "epoch": 0.32438195124762287,
      "grad_norm": 0.8412485718727112,
      "learning_rate": 7.894661072062504e-06,
      "loss": 0.7682,
      "step": 11258
    },
    {
      "epoch": 0.32441076470927216,
      "grad_norm": 0.8379024863243103,
      "learning_rate": 7.894280596912214e-06,
      "loss": 0.7414,
      "step": 11259
    },
    {
      "epoch": 0.32443957817092145,
      "grad_norm": 0.8011726140975952,
      "learning_rate": 7.893900096555633e-06,
      "loss": 0.7569,
      "step": 11260
    },
    {
      "epoch": 0.32446839163257074,
      "grad_norm": 0.7984116077423096,
      "learning_rate": 7.893519570996074e-06,
      "loss": 0.7734,
      "step": 11261
    },
    {
      "epoch": 0.32449720509422003,
      "grad_norm": 0.8377450108528137,
      "learning_rate": 7.893139020236851e-06,
      "loss": 0.7596,
      "step": 11262
    },
    {
      "epoch": 0.3245260185558693,
      "grad_norm": 0.7520282864570618,
      "learning_rate": 7.892758444281282e-06,
      "loss": 0.7601,
      "step": 11263
    },
    {
      "epoch": 0.32455483201751856,
      "grad_norm": 3.0469844341278076,
      "learning_rate": 7.892377843132677e-06,
      "loss": 0.7595,
      "step": 11264
    },
    {
      "epoch": 0.32458364547916785,
      "grad_norm": 0.768405020236969,
      "learning_rate": 7.891997216794354e-06,
      "loss": 0.7483,
      "step": 11265
    },
    {
      "epoch": 0.32461245894081714,
      "grad_norm": 0.7691212892532349,
      "learning_rate": 7.891616565269624e-06,
      "loss": 0.7392,
      "step": 11266
    },
    {
      "epoch": 0.32464127240246643,
      "grad_norm": 0.7913263440132141,
      "learning_rate": 7.891235888561804e-06,
      "loss": 0.7597,
      "step": 11267
    },
    {
      "epoch": 0.3246700858641157,
      "grad_norm": 0.7965723872184753,
      "learning_rate": 7.890855186674211e-06,
      "loss": 0.7444,
      "step": 11268
    },
    {
      "epoch": 0.324698899325765,
      "grad_norm": 1.0060590505599976,
      "learning_rate": 7.890474459610161e-06,
      "loss": 0.7308,
      "step": 11269
    },
    {
      "epoch": 0.3247277127874143,
      "grad_norm": 0.8021911382675171,
      "learning_rate": 7.890093707372966e-06,
      "loss": 0.7319,
      "step": 11270
    },
    {
      "epoch": 0.32475652624906354,
      "grad_norm": 0.8211544156074524,
      "learning_rate": 7.889712929965943e-06,
      "loss": 0.7413,
      "step": 11271
    },
    {
      "epoch": 0.32478533971071283,
      "grad_norm": 0.7940809726715088,
      "learning_rate": 7.889332127392409e-06,
      "loss": 0.7401,
      "step": 11272
    },
    {
      "epoch": 0.3248141531723621,
      "grad_norm": 0.7773241400718689,
      "learning_rate": 7.88895129965568e-06,
      "loss": 0.7381,
      "step": 11273
    },
    {
      "epoch": 0.3248429666340114,
      "grad_norm": 0.7926507592201233,
      "learning_rate": 7.888570446759074e-06,
      "loss": 0.7326,
      "step": 11274
    },
    {
      "epoch": 0.3248717800956607,
      "grad_norm": 0.8237621188163757,
      "learning_rate": 7.888189568705908e-06,
      "loss": 0.7553,
      "step": 11275
    },
    {
      "epoch": 0.32490059355731,
      "grad_norm": 0.8228040337562561,
      "learning_rate": 7.887808665499497e-06,
      "loss": 0.7218,
      "step": 11276
    },
    {
      "epoch": 0.3249294070189593,
      "grad_norm": 0.8043169379234314,
      "learning_rate": 7.88742773714316e-06,
      "loss": 0.7647,
      "step": 11277
    },
    {
      "epoch": 0.3249582204806085,
      "grad_norm": 0.7504684925079346,
      "learning_rate": 7.887046783640214e-06,
      "loss": 0.742,
      "step": 11278
    },
    {
      "epoch": 0.3249870339422578,
      "grad_norm": 0.7939964532852173,
      "learning_rate": 7.886665804993973e-06,
      "loss": 0.7667,
      "step": 11279
    },
    {
      "epoch": 0.3250158474039071,
      "grad_norm": 0.8434250354766846,
      "learning_rate": 7.886284801207762e-06,
      "loss": 0.754,
      "step": 11280
    },
    {
      "epoch": 0.3250446608655564,
      "grad_norm": 0.8018321990966797,
      "learning_rate": 7.885903772284895e-06,
      "loss": 0.7403,
      "step": 11281
    },
    {
      "epoch": 0.3250734743272057,
      "grad_norm": 0.8015230298042297,
      "learning_rate": 7.88552271822869e-06,
      "loss": 0.7496,
      "step": 11282
    },
    {
      "epoch": 0.32510228778885497,
      "grad_norm": 0.7959204316139221,
      "learning_rate": 7.885141639042468e-06,
      "loss": 0.7328,
      "step": 11283
    },
    {
      "epoch": 0.32513110125050426,
      "grad_norm": 0.8254463076591492,
      "learning_rate": 7.884760534729547e-06,
      "loss": 0.7386,
      "step": 11284
    },
    {
      "epoch": 0.3251599147121535,
      "grad_norm": 0.8477780818939209,
      "learning_rate": 7.884379405293244e-06,
      "loss": 0.7637,
      "step": 11285
    },
    {
      "epoch": 0.3251887281738028,
      "grad_norm": 0.8264339566230774,
      "learning_rate": 7.88399825073688e-06,
      "loss": 0.7511,
      "step": 11286
    },
    {
      "epoch": 0.3252175416354521,
      "grad_norm": 0.7782816886901855,
      "learning_rate": 7.883617071063774e-06,
      "loss": 0.7328,
      "step": 11287
    },
    {
      "epoch": 0.32524635509710137,
      "grad_norm": 0.8061347007751465,
      "learning_rate": 7.883235866277246e-06,
      "loss": 0.7741,
      "step": 11288
    },
    {
      "epoch": 0.32527516855875066,
      "grad_norm": 0.8100230693817139,
      "learning_rate": 7.882854636380615e-06,
      "loss": 0.7509,
      "step": 11289
    },
    {
      "epoch": 0.32530398202039995,
      "grad_norm": 0.7823702692985535,
      "learning_rate": 7.882473381377202e-06,
      "loss": 0.7471,
      "step": 11290
    },
    {
      "epoch": 0.3253327954820492,
      "grad_norm": 0.7903842329978943,
      "learning_rate": 7.88209210127033e-06,
      "loss": 0.7769,
      "step": 11291
    },
    {
      "epoch": 0.3253616089436985,
      "grad_norm": 0.8095833659172058,
      "learning_rate": 7.881710796063314e-06,
      "loss": 0.7403,
      "step": 11292
    },
    {
      "epoch": 0.32539042240534777,
      "grad_norm": 0.80476975440979,
      "learning_rate": 7.881329465759477e-06,
      "loss": 0.7309,
      "step": 11293
    },
    {
      "epoch": 0.32541923586699706,
      "grad_norm": 0.8046413660049438,
      "learning_rate": 7.880948110362143e-06,
      "loss": 0.7509,
      "step": 11294
    },
    {
      "epoch": 0.32544804932864635,
      "grad_norm": 0.7912353873252869,
      "learning_rate": 7.88056672987463e-06,
      "loss": 0.7423,
      "step": 11295
    },
    {
      "epoch": 0.32547686279029564,
      "grad_norm": 0.7802984714508057,
      "learning_rate": 7.88018532430026e-06,
      "loss": 0.7494,
      "step": 11296
    },
    {
      "epoch": 0.32550567625194493,
      "grad_norm": 0.7886490225791931,
      "learning_rate": 7.879803893642355e-06,
      "loss": 0.7518,
      "step": 11297
    },
    {
      "epoch": 0.32553448971359417,
      "grad_norm": 0.7597782611846924,
      "learning_rate": 7.879422437904236e-06,
      "loss": 0.7334,
      "step": 11298
    },
    {
      "epoch": 0.32556330317524346,
      "grad_norm": 0.8006235361099243,
      "learning_rate": 7.879040957089229e-06,
      "loss": 0.7282,
      "step": 11299
    },
    {
      "epoch": 0.32559211663689275,
      "grad_norm": 0.8023361563682556,
      "learning_rate": 7.87865945120065e-06,
      "loss": 0.7642,
      "step": 11300
    },
    {
      "epoch": 0.32562093009854204,
      "grad_norm": 0.8021963238716125,
      "learning_rate": 7.878277920241826e-06,
      "loss": 0.7656,
      "step": 11301
    },
    {
      "epoch": 0.32564974356019133,
      "grad_norm": 0.7624968886375427,
      "learning_rate": 7.877896364216078e-06,
      "loss": 0.7322,
      "step": 11302
    },
    {
      "epoch": 0.3256785570218406,
      "grad_norm": 0.8178533315658569,
      "learning_rate": 7.87751478312673e-06,
      "loss": 0.7679,
      "step": 11303
    },
    {
      "epoch": 0.3257073704834899,
      "grad_norm": 0.7679972052574158,
      "learning_rate": 7.877133176977104e-06,
      "loss": 0.7685,
      "step": 11304
    },
    {
      "epoch": 0.32573618394513915,
      "grad_norm": 0.7638117671012878,
      "learning_rate": 7.876751545770524e-06,
      "loss": 0.7311,
      "step": 11305
    },
    {
      "epoch": 0.32576499740678844,
      "grad_norm": 0.7717370390892029,
      "learning_rate": 7.876369889510316e-06,
      "loss": 0.7415,
      "step": 11306
    },
    {
      "epoch": 0.3257938108684377,
      "grad_norm": 0.7960564494132996,
      "learning_rate": 7.8759882081998e-06,
      "loss": 0.7389,
      "step": 11307
    },
    {
      "epoch": 0.325822624330087,
      "grad_norm": 0.8421401977539062,
      "learning_rate": 7.8756065018423e-06,
      "loss": 0.758,
      "step": 11308
    },
    {
      "epoch": 0.3258514377917363,
      "grad_norm": 0.7995235323905945,
      "learning_rate": 7.875224770441146e-06,
      "loss": 0.7749,
      "step": 11309
    },
    {
      "epoch": 0.3258802512533856,
      "grad_norm": 0.8050480484962463,
      "learning_rate": 7.874843013999656e-06,
      "loss": 0.7551,
      "step": 11310
    },
    {
      "epoch": 0.3259090647150349,
      "grad_norm": 0.8030425906181335,
      "learning_rate": 7.874461232521156e-06,
      "loss": 0.7629,
      "step": 11311
    },
    {
      "epoch": 0.3259378781766841,
      "grad_norm": 0.8254221081733704,
      "learning_rate": 7.874079426008974e-06,
      "loss": 0.7512,
      "step": 11312
    },
    {
      "epoch": 0.3259666916383334,
      "grad_norm": 0.8030336499214172,
      "learning_rate": 7.873697594466432e-06,
      "loss": 0.7531,
      "step": 11313
    },
    {
      "epoch": 0.3259955050999827,
      "grad_norm": 0.7791382074356079,
      "learning_rate": 7.873315737896856e-06,
      "loss": 0.7502,
      "step": 11314
    },
    {
      "epoch": 0.326024318561632,
      "grad_norm": 0.8012187480926514,
      "learning_rate": 7.872933856303575e-06,
      "loss": 0.7435,
      "step": 11315
    },
    {
      "epoch": 0.3260531320232813,
      "grad_norm": 0.8961107730865479,
      "learning_rate": 7.872551949689909e-06,
      "loss": 0.7148,
      "step": 11316
    },
    {
      "epoch": 0.3260819454849306,
      "grad_norm": 0.7910194993019104,
      "learning_rate": 7.872170018059188e-06,
      "loss": 0.7532,
      "step": 11317
    },
    {
      "epoch": 0.3261107589465798,
      "grad_norm": 0.7840735912322998,
      "learning_rate": 7.871788061414737e-06,
      "loss": 0.7352,
      "step": 11318
    },
    {
      "epoch": 0.3261395724082291,
      "grad_norm": 0.8028537631034851,
      "learning_rate": 7.871406079759885e-06,
      "loss": 0.7293,
      "step": 11319
    },
    {
      "epoch": 0.3261683858698784,
      "grad_norm": 0.8205002546310425,
      "learning_rate": 7.871024073097954e-06,
      "loss": 0.7616,
      "step": 11320
    },
    {
      "epoch": 0.3261971993315277,
      "grad_norm": 0.84340900182724,
      "learning_rate": 7.870642041432275e-06,
      "loss": 0.7378,
      "step": 11321
    },
    {
      "epoch": 0.326226012793177,
      "grad_norm": 0.837433934211731,
      "learning_rate": 7.87025998476617e-06,
      "loss": 0.7441,
      "step": 11322
    },
    {
      "epoch": 0.32625482625482627,
      "grad_norm": 0.7909320592880249,
      "learning_rate": 7.869877903102974e-06,
      "loss": 0.7416,
      "step": 11323
    },
    {
      "epoch": 0.32628363971647556,
      "grad_norm": 0.79595547914505,
      "learning_rate": 7.869495796446008e-06,
      "loss": 0.7362,
      "step": 11324
    },
    {
      "epoch": 0.3263124531781248,
      "grad_norm": 0.8144220113754272,
      "learning_rate": 7.869113664798604e-06,
      "loss": 0.7449,
      "step": 11325
    },
    {
      "epoch": 0.3263412666397741,
      "grad_norm": 0.8049570918083191,
      "learning_rate": 7.868731508164084e-06,
      "loss": 0.775,
      "step": 11326
    },
    {
      "epoch": 0.3263700801014234,
      "grad_norm": 5.0436015129089355,
      "learning_rate": 7.868349326545783e-06,
      "loss": 0.7758,
      "step": 11327
    },
    {
      "epoch": 0.32639889356307267,
      "grad_norm": 0.8046895265579224,
      "learning_rate": 7.867967119947027e-06,
      "loss": 0.7439,
      "step": 11328
    },
    {
      "epoch": 0.32642770702472196,
      "grad_norm": 0.7996934652328491,
      "learning_rate": 7.867584888371145e-06,
      "loss": 0.7362,
      "step": 11329
    },
    {
      "epoch": 0.32645652048637125,
      "grad_norm": 0.8703468441963196,
      "learning_rate": 7.867202631821464e-06,
      "loss": 0.7358,
      "step": 11330
    },
    {
      "epoch": 0.32648533394802054,
      "grad_norm": 0.8753935694694519,
      "learning_rate": 7.866820350301314e-06,
      "loss": 0.7724,
      "step": 11331
    },
    {
      "epoch": 0.3265141474096698,
      "grad_norm": 0.8301600813865662,
      "learning_rate": 7.866438043814025e-06,
      "loss": 0.7687,
      "step": 11332
    },
    {
      "epoch": 0.32654296087131907,
      "grad_norm": 0.8900853991508484,
      "learning_rate": 7.866055712362924e-06,
      "loss": 0.7708,
      "step": 11333
    },
    {
      "epoch": 0.32657177433296836,
      "grad_norm": 0.8818381428718567,
      "learning_rate": 7.865673355951345e-06,
      "loss": 0.765,
      "step": 11334
    },
    {
      "epoch": 0.32660058779461765,
      "grad_norm": 0.7940084934234619,
      "learning_rate": 7.865290974582616e-06,
      "loss": 0.7365,
      "step": 11335
    },
    {
      "epoch": 0.32662940125626694,
      "grad_norm": 0.8048868179321289,
      "learning_rate": 7.864908568260066e-06,
      "loss": 0.732,
      "step": 11336
    },
    {
      "epoch": 0.32665821471791623,
      "grad_norm": 0.7940956950187683,
      "learning_rate": 7.864526136987025e-06,
      "loss": 0.7614,
      "step": 11337
    },
    {
      "epoch": 0.3266870281795655,
      "grad_norm": 0.805647075176239,
      "learning_rate": 7.864143680766826e-06,
      "loss": 0.7483,
      "step": 11338
    },
    {
      "epoch": 0.32671584164121475,
      "grad_norm": 0.8118994235992432,
      "learning_rate": 7.8637611996028e-06,
      "loss": 0.7677,
      "step": 11339
    },
    {
      "epoch": 0.32674465510286405,
      "grad_norm": 1.0246713161468506,
      "learning_rate": 7.863378693498274e-06,
      "loss": 0.7489,
      "step": 11340
    },
    {
      "epoch": 0.32677346856451334,
      "grad_norm": 0.7969879508018494,
      "learning_rate": 7.862996162456584e-06,
      "loss": 0.7666,
      "step": 11341
    },
    {
      "epoch": 0.3268022820261626,
      "grad_norm": 0.7745166420936584,
      "learning_rate": 7.862613606481057e-06,
      "loss": 0.7312,
      "step": 11342
    },
    {
      "epoch": 0.3268310954878119,
      "grad_norm": 0.8102993965148926,
      "learning_rate": 7.86223102557503e-06,
      "loss": 0.7625,
      "step": 11343
    },
    {
      "epoch": 0.3268599089494612,
      "grad_norm": 0.8097595572471619,
      "learning_rate": 7.86184841974183e-06,
      "loss": 0.7521,
      "step": 11344
    },
    {
      "epoch": 0.32688872241111044,
      "grad_norm": 0.7743856906890869,
      "learning_rate": 7.861465788984792e-06,
      "loss": 0.7742,
      "step": 11345
    },
    {
      "epoch": 0.32691753587275973,
      "grad_norm": 0.8128389120101929,
      "learning_rate": 7.861083133307248e-06,
      "loss": 0.7396,
      "step": 11346
    },
    {
      "epoch": 0.326946349334409,
      "grad_norm": 0.7893686890602112,
      "learning_rate": 7.860700452712528e-06,
      "loss": 0.7415,
      "step": 11347
    },
    {
      "epoch": 0.3269751627960583,
      "grad_norm": 0.8045784831047058,
      "learning_rate": 7.860317747203968e-06,
      "loss": 0.7721,
      "step": 11348
    },
    {
      "epoch": 0.3270039762577076,
      "grad_norm": 0.7845116853713989,
      "learning_rate": 7.859935016784901e-06,
      "loss": 0.7683,
      "step": 11349
    },
    {
      "epoch": 0.3270327897193569,
      "grad_norm": 0.8111539483070374,
      "learning_rate": 7.859552261458656e-06,
      "loss": 0.7472,
      "step": 11350
    },
    {
      "epoch": 0.3270616031810062,
      "grad_norm": 0.7936896681785583,
      "learning_rate": 7.85916948122857e-06,
      "loss": 0.7366,
      "step": 11351
    },
    {
      "epoch": 0.3270904166426554,
      "grad_norm": 0.7731183171272278,
      "learning_rate": 7.858786676097977e-06,
      "loss": 0.7408,
      "step": 11352
    },
    {
      "epoch": 0.3271192301043047,
      "grad_norm": 0.781114399433136,
      "learning_rate": 7.858403846070209e-06,
      "loss": 0.7426,
      "step": 11353
    },
    {
      "epoch": 0.327148043565954,
      "grad_norm": 0.8020622730255127,
      "learning_rate": 7.858020991148601e-06,
      "loss": 0.7478,
      "step": 11354
    },
    {
      "epoch": 0.3271768570276033,
      "grad_norm": 0.7808359861373901,
      "learning_rate": 7.857638111336489e-06,
      "loss": 0.7617,
      "step": 11355
    },
    {
      "epoch": 0.3272056704892526,
      "grad_norm": 0.8033491969108582,
      "learning_rate": 7.857255206637201e-06,
      "loss": 0.7772,
      "step": 11356
    },
    {
      "epoch": 0.3272344839509019,
      "grad_norm": 0.778137743473053,
      "learning_rate": 7.856872277054082e-06,
      "loss": 0.7378,
      "step": 11357
    },
    {
      "epoch": 0.32726329741255117,
      "grad_norm": 0.8042933940887451,
      "learning_rate": 7.856489322590458e-06,
      "loss": 0.7846,
      "step": 11358
    },
    {
      "epoch": 0.3272921108742004,
      "grad_norm": 0.8006391525268555,
      "learning_rate": 7.856106343249669e-06,
      "loss": 0.7384,
      "step": 11359
    },
    {
      "epoch": 0.3273209243358497,
      "grad_norm": 0.8027382493019104,
      "learning_rate": 7.855723339035047e-06,
      "loss": 0.7787,
      "step": 11360
    },
    {
      "epoch": 0.327349737797499,
      "grad_norm": 0.823296070098877,
      "learning_rate": 7.85534030994993e-06,
      "loss": 0.7683,
      "step": 11361
    },
    {
      "epoch": 0.3273785512591483,
      "grad_norm": 0.825249969959259,
      "learning_rate": 7.854957255997654e-06,
      "loss": 0.7584,
      "step": 11362
    },
    {
      "epoch": 0.32740736472079757,
      "grad_norm": 0.7988976240158081,
      "learning_rate": 7.854574177181554e-06,
      "loss": 0.7675,
      "step": 11363
    },
    {
      "epoch": 0.32743617818244686,
      "grad_norm": 0.7864082455635071,
      "learning_rate": 7.854191073504966e-06,
      "loss": 0.7083,
      "step": 11364
    },
    {
      "epoch": 0.32746499164409615,
      "grad_norm": 0.7975149154663086,
      "learning_rate": 7.853807944971227e-06,
      "loss": 0.7447,
      "step": 11365
    },
    {
      "epoch": 0.3274938051057454,
      "grad_norm": 0.7681595087051392,
      "learning_rate": 7.853424791583674e-06,
      "loss": 0.7536,
      "step": 11366
    },
    {
      "epoch": 0.3275226185673947,
      "grad_norm": 0.8037063479423523,
      "learning_rate": 7.853041613345644e-06,
      "loss": 0.7618,
      "step": 11367
    },
    {
      "epoch": 0.32755143202904397,
      "grad_norm": 0.8296350240707397,
      "learning_rate": 7.852658410260471e-06,
      "loss": 0.7488,
      "step": 11368
    },
    {
      "epoch": 0.32758024549069326,
      "grad_norm": 0.8257864713668823,
      "learning_rate": 7.852275182331497e-06,
      "loss": 0.7563,
      "step": 11369
    },
    {
      "epoch": 0.32760905895234255,
      "grad_norm": 0.7879478335380554,
      "learning_rate": 7.851891929562056e-06,
      "loss": 0.7277,
      "step": 11370
    },
    {
      "epoch": 0.32763787241399184,
      "grad_norm": 0.7982610464096069,
      "learning_rate": 7.851508651955489e-06,
      "loss": 0.7777,
      "step": 11371
    },
    {
      "epoch": 0.3276666858756411,
      "grad_norm": 0.8397880792617798,
      "learning_rate": 7.85112534951513e-06,
      "loss": 0.7656,
      "step": 11372
    },
    {
      "epoch": 0.32769549933729036,
      "grad_norm": 0.8345381617546082,
      "learning_rate": 7.850742022244323e-06,
      "loss": 0.7858,
      "step": 11373
    },
    {
      "epoch": 0.32772431279893965,
      "grad_norm": 0.7941411733627319,
      "learning_rate": 7.850358670146398e-06,
      "loss": 0.7439,
      "step": 11374
    },
    {
      "epoch": 0.32775312626058895,
      "grad_norm": 0.8425265550613403,
      "learning_rate": 7.849975293224701e-06,
      "loss": 0.738,
      "step": 11375
    },
    {
      "epoch": 0.32778193972223824,
      "grad_norm": 0.847861111164093,
      "learning_rate": 7.849591891482567e-06,
      "loss": 0.7145,
      "step": 11376
    },
    {
      "epoch": 0.3278107531838875,
      "grad_norm": 0.7941967248916626,
      "learning_rate": 7.849208464923337e-06,
      "loss": 0.7375,
      "step": 11377
    },
    {
      "epoch": 0.3278395666455368,
      "grad_norm": 0.7710018754005432,
      "learning_rate": 7.84882501355035e-06,
      "loss": 0.7261,
      "step": 11378
    },
    {
      "epoch": 0.32786838010718605,
      "grad_norm": 0.8105650544166565,
      "learning_rate": 7.848441537366943e-06,
      "loss": 0.7686,
      "step": 11379
    },
    {
      "epoch": 0.32789719356883534,
      "grad_norm": 0.8416699171066284,
      "learning_rate": 7.848058036376459e-06,
      "loss": 0.7275,
      "step": 11380
    },
    {
      "epoch": 0.32792600703048463,
      "grad_norm": 0.7951586246490479,
      "learning_rate": 7.847674510582236e-06,
      "loss": 0.7491,
      "step": 11381
    },
    {
      "epoch": 0.3279548204921339,
      "grad_norm": 0.8148095607757568,
      "learning_rate": 7.847290959987614e-06,
      "loss": 0.7212,
      "step": 11382
    },
    {
      "epoch": 0.3279836339537832,
      "grad_norm": 0.8306559920310974,
      "learning_rate": 7.846907384595935e-06,
      "loss": 0.7662,
      "step": 11383
    },
    {
      "epoch": 0.3280124474154325,
      "grad_norm": 0.8157174587249756,
      "learning_rate": 7.846523784410539e-06,
      "loss": 0.746,
      "step": 11384
    },
    {
      "epoch": 0.3280412608770818,
      "grad_norm": 0.7957122325897217,
      "learning_rate": 7.846140159434766e-06,
      "loss": 0.7464,
      "step": 11385
    },
    {
      "epoch": 0.32807007433873103,
      "grad_norm": 15.215590476989746,
      "learning_rate": 7.845756509671955e-06,
      "loss": 0.8506,
      "step": 11386
    },
    {
      "epoch": 0.3280988878003803,
      "grad_norm": 0.8107870221138,
      "learning_rate": 7.845372835125452e-06,
      "loss": 0.7326,
      "step": 11387
    },
    {
      "epoch": 0.3281277012620296,
      "grad_norm": 0.9740598797798157,
      "learning_rate": 7.844989135798597e-06,
      "loss": 0.7726,
      "step": 11388
    },
    {
      "epoch": 0.3281565147236789,
      "grad_norm": 1.0342230796813965,
      "learning_rate": 7.84460541169473e-06,
      "loss": 0.7829,
      "step": 11389
    },
    {
      "epoch": 0.3281853281853282,
      "grad_norm": 1.1074707508087158,
      "learning_rate": 7.84422166281719e-06,
      "loss": 0.7627,
      "step": 11390
    },
    {
      "epoch": 0.3282141416469775,
      "grad_norm": 1.1430366039276123,
      "learning_rate": 7.843837889169325e-06,
      "loss": 0.7846,
      "step": 11391
    },
    {
      "epoch": 0.3282429551086268,
      "grad_norm": 0.9635797739028931,
      "learning_rate": 7.843454090754476e-06,
      "loss": 0.781,
      "step": 11392
    },
    {
      "epoch": 0.328271768570276,
      "grad_norm": 0.8943400382995605,
      "learning_rate": 7.843070267575983e-06,
      "loss": 0.7756,
      "step": 11393
    },
    {
      "epoch": 0.3283005820319253,
      "grad_norm": 0.8807152509689331,
      "learning_rate": 7.84268641963719e-06,
      "loss": 0.751,
      "step": 11394
    },
    {
      "epoch": 0.3283293954935746,
      "grad_norm": 0.8978941440582275,
      "learning_rate": 7.84230254694144e-06,
      "loss": 0.7554,
      "step": 11395
    },
    {
      "epoch": 0.3283582089552239,
      "grad_norm": 0.8689761161804199,
      "learning_rate": 7.841918649492077e-06,
      "loss": 0.7887,
      "step": 11396
    },
    {
      "epoch": 0.3283870224168732,
      "grad_norm": 0.8312409520149231,
      "learning_rate": 7.841534727292442e-06,
      "loss": 0.7627,
      "step": 11397
    },
    {
      "epoch": 0.32841583587852247,
      "grad_norm": 0.8756749629974365,
      "learning_rate": 7.84115078034588e-06,
      "loss": 0.7644,
      "step": 11398
    },
    {
      "epoch": 0.3284446493401717,
      "grad_norm": 0.829990565776825,
      "learning_rate": 7.840766808655736e-06,
      "loss": 0.7516,
      "step": 11399
    },
    {
      "epoch": 0.328473462801821,
      "grad_norm": 0.8220329284667969,
      "learning_rate": 7.84038281222535e-06,
      "loss": 0.7392,
      "step": 11400
    },
    {
      "epoch": 0.3285022762634703,
      "grad_norm": 0.8318465948104858,
      "learning_rate": 7.839998791058072e-06,
      "loss": 0.7773,
      "step": 11401
    },
    {
      "epoch": 0.3285310897251196,
      "grad_norm": 0.8391769528388977,
      "learning_rate": 7.839614745157242e-06,
      "loss": 0.7794,
      "step": 11402
    },
    {
      "epoch": 0.32855990318676886,
      "grad_norm": 0.8520700335502625,
      "learning_rate": 7.839230674526209e-06,
      "loss": 0.7735,
      "step": 11403
    },
    {
      "epoch": 0.32858871664841816,
      "grad_norm": 0.7939357161521912,
      "learning_rate": 7.838846579168311e-06,
      "loss": 0.7246,
      "step": 11404
    },
    {
      "epoch": 0.32861753011006745,
      "grad_norm": 0.8187628984451294,
      "learning_rate": 7.8384624590869e-06,
      "loss": 0.7227,
      "step": 11405
    },
    {
      "epoch": 0.3286463435717167,
      "grad_norm": 0.8182387351989746,
      "learning_rate": 7.838078314285317e-06,
      "loss": 0.7338,
      "step": 11406
    },
    {
      "epoch": 0.32867515703336597,
      "grad_norm": 0.8357338905334473,
      "learning_rate": 7.837694144766909e-06,
      "loss": 0.7477,
      "step": 11407
    },
    {
      "epoch": 0.32870397049501526,
      "grad_norm": 0.8244689702987671,
      "learning_rate": 7.837309950535022e-06,
      "loss": 0.7391,
      "step": 11408
    },
    {
      "epoch": 0.32873278395666455,
      "grad_norm": 0.8402907252311707,
      "learning_rate": 7.836925731593002e-06,
      "loss": 0.7636,
      "step": 11409
    },
    {
      "epoch": 0.32876159741831384,
      "grad_norm": 0.7733492851257324,
      "learning_rate": 7.836541487944194e-06,
      "loss": 0.7351,
      "step": 11410
    },
    {
      "epoch": 0.32879041087996314,
      "grad_norm": 0.7954080700874329,
      "learning_rate": 7.836157219591945e-06,
      "loss": 0.7487,
      "step": 11411
    },
    {
      "epoch": 0.3288192243416124,
      "grad_norm": 0.8217668533325195,
      "learning_rate": 7.835772926539602e-06,
      "loss": 0.7643,
      "step": 11412
    },
    {
      "epoch": 0.32884803780326166,
      "grad_norm": 0.8237785696983337,
      "learning_rate": 7.835388608790513e-06,
      "loss": 0.7357,
      "step": 11413
    },
    {
      "epoch": 0.32887685126491095,
      "grad_norm": 0.768964409828186,
      "learning_rate": 7.83500426634802e-06,
      "loss": 0.7284,
      "step": 11414
    },
    {
      "epoch": 0.32890566472656024,
      "grad_norm": 0.8426988124847412,
      "learning_rate": 7.834619899215477e-06,
      "loss": 0.7603,
      "step": 11415
    },
    {
      "epoch": 0.32893447818820953,
      "grad_norm": 0.8171431422233582,
      "learning_rate": 7.834235507396228e-06,
      "loss": 0.7855,
      "step": 11416
    },
    {
      "epoch": 0.3289632916498588,
      "grad_norm": 0.7851941585540771,
      "learning_rate": 7.83385109089362e-06,
      "loss": 0.7223,
      "step": 11417
    },
    {
      "epoch": 0.3289921051115081,
      "grad_norm": 0.782404899597168,
      "learning_rate": 7.833466649711002e-06,
      "loss": 0.7544,
      "step": 11418
    },
    {
      "epoch": 0.3290209185731574,
      "grad_norm": 0.8257805109024048,
      "learning_rate": 7.833082183851722e-06,
      "loss": 0.7367,
      "step": 11419
    },
    {
      "epoch": 0.32904973203480664,
      "grad_norm": 0.8088416457176208,
      "learning_rate": 7.83269769331913e-06,
      "loss": 0.7526,
      "step": 11420
    },
    {
      "epoch": 0.32907854549645593,
      "grad_norm": 0.8309888243675232,
      "learning_rate": 7.83231317811657e-06,
      "loss": 0.7577,
      "step": 11421
    },
    {
      "epoch": 0.3291073589581052,
      "grad_norm": 0.8284411430358887,
      "learning_rate": 7.831928638247393e-06,
      "loss": 0.7414,
      "step": 11422
    },
    {
      "epoch": 0.3291361724197545,
      "grad_norm": 0.8201579451560974,
      "learning_rate": 7.83154407371495e-06,
      "loss": 0.7466,
      "step": 11423
    },
    {
      "epoch": 0.3291649858814038,
      "grad_norm": 0.812842845916748,
      "learning_rate": 7.831159484522588e-06,
      "loss": 0.7352,
      "step": 11424
    },
    {
      "epoch": 0.3291937993430531,
      "grad_norm": 0.7923663258552551,
      "learning_rate": 7.830774870673659e-06,
      "loss": 0.7535,
      "step": 11425
    },
    {
      "epoch": 0.32922261280470233,
      "grad_norm": 0.8076550960540771,
      "learning_rate": 7.83039023217151e-06,
      "loss": 0.7768,
      "step": 11426
    },
    {
      "epoch": 0.3292514262663516,
      "grad_norm": 0.8048204183578491,
      "learning_rate": 7.83000556901949e-06,
      "loss": 0.7738,
      "step": 11427
    },
    {
      "epoch": 0.3292802397280009,
      "grad_norm": 0.8544691801071167,
      "learning_rate": 7.829620881220949e-06,
      "loss": 0.7593,
      "step": 11428
    },
    {
      "epoch": 0.3293090531896502,
      "grad_norm": 0.8105443120002747,
      "learning_rate": 7.829236168779241e-06,
      "loss": 0.7567,
      "step": 11429
    },
    {
      "epoch": 0.3293378666512995,
      "grad_norm": 0.8058494329452515,
      "learning_rate": 7.828851431697716e-06,
      "loss": 0.7581,
      "step": 11430
    },
    {
      "epoch": 0.3293666801129488,
      "grad_norm": 0.8282619118690491,
      "learning_rate": 7.82846666997972e-06,
      "loss": 0.7475,
      "step": 11431
    },
    {
      "epoch": 0.3293954935745981,
      "grad_norm": 0.8374467492103577,
      "learning_rate": 7.828081883628608e-06,
      "loss": 0.7623,
      "step": 11432
    },
    {
      "epoch": 0.3294243070362473,
      "grad_norm": 0.7960215210914612,
      "learning_rate": 7.827697072647729e-06,
      "loss": 0.729,
      "step": 11433
    },
    {
      "epoch": 0.3294531204978966,
      "grad_norm": 0.804798424243927,
      "learning_rate": 7.827312237040435e-06,
      "loss": 0.7734,
      "step": 11434
    },
    {
      "epoch": 0.3294819339595459,
      "grad_norm": 0.8079067468643188,
      "learning_rate": 7.82692737681008e-06,
      "loss": 0.7758,
      "step": 11435
    },
    {
      "epoch": 0.3295107474211952,
      "grad_norm": 0.8452391624450684,
      "learning_rate": 7.826542491960009e-06,
      "loss": 0.7483,
      "step": 11436
    },
    {
      "epoch": 0.3295395608828445,
      "grad_norm": 0.8264631032943726,
      "learning_rate": 7.826157582493582e-06,
      "loss": 0.7393,
      "step": 11437
    },
    {
      "epoch": 0.32956837434449376,
      "grad_norm": 0.7971525192260742,
      "learning_rate": 7.825772648414147e-06,
      "loss": 0.755,
      "step": 11438
    },
    {
      "epoch": 0.32959718780614305,
      "grad_norm": 0.8439275026321411,
      "learning_rate": 7.825387689725057e-06,
      "loss": 0.7885,
      "step": 11439
    },
    {
      "epoch": 0.3296260012677923,
      "grad_norm": 0.8194226026535034,
      "learning_rate": 7.825002706429663e-06,
      "loss": 0.7278,
      "step": 11440
    },
    {
      "epoch": 0.3296548147294416,
      "grad_norm": 0.8027651309967041,
      "learning_rate": 7.82461769853132e-06,
      "loss": 0.7786,
      "step": 11441
    },
    {
      "epoch": 0.32968362819109087,
      "grad_norm": 3.9583115577697754,
      "learning_rate": 7.824232666033382e-06,
      "loss": 0.7546,
      "step": 11442
    },
    {
      "epoch": 0.32971244165274016,
      "grad_norm": 0.8193223476409912,
      "learning_rate": 7.8238476089392e-06,
      "loss": 0.7498,
      "step": 11443
    },
    {
      "epoch": 0.32974125511438945,
      "grad_norm": 0.8446735143661499,
      "learning_rate": 7.823462527252128e-06,
      "loss": 0.7689,
      "step": 11444
    },
    {
      "epoch": 0.32977006857603874,
      "grad_norm": 0.790183961391449,
      "learning_rate": 7.82307742097552e-06,
      "loss": 0.7498,
      "step": 11445
    },
    {
      "epoch": 0.32979888203768803,
      "grad_norm": 0.8027670979499817,
      "learning_rate": 7.822692290112729e-06,
      "loss": 0.7399,
      "step": 11446
    },
    {
      "epoch": 0.32982769549933727,
      "grad_norm": 0.7797123193740845,
      "learning_rate": 7.82230713466711e-06,
      "loss": 0.7341,
      "step": 11447
    },
    {
      "epoch": 0.32985650896098656,
      "grad_norm": 0.8371152877807617,
      "learning_rate": 7.821921954642018e-06,
      "loss": 0.7537,
      "step": 11448
    },
    {
      "epoch": 0.32988532242263585,
      "grad_norm": 0.8482974171638489,
      "learning_rate": 7.821536750040805e-06,
      "loss": 0.7675,
      "step": 11449
    },
    {
      "epoch": 0.32991413588428514,
      "grad_norm": 0.8255109190940857,
      "learning_rate": 7.821151520866829e-06,
      "loss": 0.7484,
      "step": 11450
    },
    {
      "epoch": 0.32994294934593443,
      "grad_norm": 0.8467826843261719,
      "learning_rate": 7.820766267123443e-06,
      "loss": 0.7395,
      "step": 11451
    },
    {
      "epoch": 0.3299717628075837,
      "grad_norm": 0.8101400136947632,
      "learning_rate": 7.820380988814002e-06,
      "loss": 0.7315,
      "step": 11452
    },
    {
      "epoch": 0.33000057626923296,
      "grad_norm": 0.7883486151695251,
      "learning_rate": 7.81999568594186e-06,
      "loss": 0.7533,
      "step": 11453
    },
    {
      "epoch": 0.33002938973088225,
      "grad_norm": 0.7949755787849426,
      "learning_rate": 7.819610358510379e-06,
      "loss": 0.7418,
      "step": 11454
    },
    {
      "epoch": 0.33005820319253154,
      "grad_norm": 0.8106580972671509,
      "learning_rate": 7.819225006522907e-06,
      "loss": 0.7402,
      "step": 11455
    },
    {
      "epoch": 0.33008701665418083,
      "grad_norm": 0.8901661038398743,
      "learning_rate": 7.818839629982806e-06,
      "loss": 0.7505,
      "step": 11456
    },
    {
      "epoch": 0.3301158301158301,
      "grad_norm": 0.7885951399803162,
      "learning_rate": 7.818454228893428e-06,
      "loss": 0.7058,
      "step": 11457
    },
    {
      "epoch": 0.3301446435774794,
      "grad_norm": 0.8345755338668823,
      "learning_rate": 7.81806880325813e-06,
      "loss": 0.738,
      "step": 11458
    },
    {
      "epoch": 0.3301734570391287,
      "grad_norm": 0.8254851698875427,
      "learning_rate": 7.817683353080271e-06,
      "loss": 0.7574,
      "step": 11459
    },
    {
      "epoch": 0.33020227050077794,
      "grad_norm": 0.8195738196372986,
      "learning_rate": 7.817297878363208e-06,
      "loss": 0.7296,
      "step": 11460
    },
    {
      "epoch": 0.33023108396242723,
      "grad_norm": 0.8046286106109619,
      "learning_rate": 7.816912379110294e-06,
      "loss": 0.7531,
      "step": 11461
    },
    {
      "epoch": 0.3302598974240765,
      "grad_norm": 0.8355334997177124,
      "learning_rate": 7.816526855324891e-06,
      "loss": 0.7389,
      "step": 11462
    },
    {
      "epoch": 0.3302887108857258,
      "grad_norm": 0.8354543447494507,
      "learning_rate": 7.816141307010355e-06,
      "loss": 0.7702,
      "step": 11463
    },
    {
      "epoch": 0.3303175243473751,
      "grad_norm": 0.7838195562362671,
      "learning_rate": 7.815755734170041e-06,
      "loss": 0.7217,
      "step": 11464
    },
    {
      "epoch": 0.3303463378090244,
      "grad_norm": 0.8329812288284302,
      "learning_rate": 7.815370136807313e-06,
      "loss": 0.7409,
      "step": 11465
    },
    {
      "epoch": 0.3303751512706737,
      "grad_norm": 0.8350663781166077,
      "learning_rate": 7.814984514925522e-06,
      "loss": 0.7321,
      "step": 11466
    },
    {
      "epoch": 0.3304039647323229,
      "grad_norm": 0.8169152736663818,
      "learning_rate": 7.814598868528032e-06,
      "loss": 0.7497,
      "step": 11467
    },
    {
      "epoch": 0.3304327781939722,
      "grad_norm": 0.8342455625534058,
      "learning_rate": 7.814213197618197e-06,
      "loss": 0.7535,
      "step": 11468
    },
    {
      "epoch": 0.3304615916556215,
      "grad_norm": 0.8188900351524353,
      "learning_rate": 7.81382750219938e-06,
      "loss": 0.7365,
      "step": 11469
    },
    {
      "epoch": 0.3304904051172708,
      "grad_norm": 0.8333708047866821,
      "learning_rate": 7.813441782274938e-06,
      "loss": 0.7234,
      "step": 11470
    },
    {
      "epoch": 0.3305192185789201,
      "grad_norm": 0.8158481121063232,
      "learning_rate": 7.81305603784823e-06,
      "loss": 0.7534,
      "step": 11471
    },
    {
      "epoch": 0.3305480320405694,
      "grad_norm": 0.7948095798492432,
      "learning_rate": 7.812670268922615e-06,
      "loss": 0.7511,
      "step": 11472
    },
    {
      "epoch": 0.33057684550221866,
      "grad_norm": 0.8254397511482239,
      "learning_rate": 7.812284475501456e-06,
      "loss": 0.72,
      "step": 11473
    },
    {
      "epoch": 0.3306056589638679,
      "grad_norm": 0.8334171175956726,
      "learning_rate": 7.81189865758811e-06,
      "loss": 0.735,
      "step": 11474
    },
    {
      "epoch": 0.3306344724255172,
      "grad_norm": 0.7898662090301514,
      "learning_rate": 7.811512815185936e-06,
      "loss": 0.7589,
      "step": 11475
    },
    {
      "epoch": 0.3306632858871665,
      "grad_norm": 0.8040724396705627,
      "learning_rate": 7.811126948298296e-06,
      "loss": 0.7262,
      "step": 11476
    },
    {
      "epoch": 0.33069209934881577,
      "grad_norm": 0.8026666045188904,
      "learning_rate": 7.810741056928553e-06,
      "loss": 0.7361,
      "step": 11477
    },
    {
      "epoch": 0.33072091281046506,
      "grad_norm": 0.8267593383789062,
      "learning_rate": 7.810355141080062e-06,
      "loss": 0.7364,
      "step": 11478
    },
    {
      "epoch": 0.33074972627211435,
      "grad_norm": 0.8028157353401184,
      "learning_rate": 7.809969200756189e-06,
      "loss": 0.7262,
      "step": 11479
    },
    {
      "epoch": 0.3307785397337636,
      "grad_norm": 0.8144214153289795,
      "learning_rate": 7.809583235960291e-06,
      "loss": 0.7566,
      "step": 11480
    },
    {
      "epoch": 0.3308073531954129,
      "grad_norm": 0.793403148651123,
      "learning_rate": 7.809197246695735e-06,
      "loss": 0.7645,
      "step": 11481
    },
    {
      "epoch": 0.33083616665706217,
      "grad_norm": 10.532869338989258,
      "learning_rate": 7.808811232965876e-06,
      "loss": 0.775,
      "step": 11482
    },
    {
      "epoch": 0.33086498011871146,
      "grad_norm": 0.788582444190979,
      "learning_rate": 7.80842519477408e-06,
      "loss": 0.736,
      "step": 11483
    },
    {
      "epoch": 0.33089379358036075,
      "grad_norm": 0.8137825131416321,
      "learning_rate": 7.808039132123708e-06,
      "loss": 0.7341,
      "step": 11484
    },
    {
      "epoch": 0.33092260704201004,
      "grad_norm": 0.8400722742080688,
      "learning_rate": 7.807653045018124e-06,
      "loss": 0.756,
      "step": 11485
    },
    {
      "epoch": 0.33095142050365933,
      "grad_norm": 0.8664617538452148,
      "learning_rate": 7.807266933460687e-06,
      "loss": 0.7253,
      "step": 11486
    },
    {
      "epoch": 0.33098023396530857,
      "grad_norm": 0.8378958106040955,
      "learning_rate": 7.80688079745476e-06,
      "loss": 0.7202,
      "step": 11487
    },
    {
      "epoch": 0.33100904742695786,
      "grad_norm": 0.8710066080093384,
      "learning_rate": 7.806494637003707e-06,
      "loss": 0.7456,
      "step": 11488
    },
    {
      "epoch": 0.33103786088860715,
      "grad_norm": 0.8407316207885742,
      "learning_rate": 7.806108452110895e-06,
      "loss": 0.7604,
      "step": 11489
    },
    {
      "epoch": 0.33106667435025644,
      "grad_norm": 0.8032335638999939,
      "learning_rate": 7.805722242779679e-06,
      "loss": 0.7452,
      "step": 11490
    },
    {
      "epoch": 0.33109548781190573,
      "grad_norm": 0.8687886595726013,
      "learning_rate": 7.805336009013428e-06,
      "loss": 0.7426,
      "step": 11491
    },
    {
      "epoch": 0.331124301273555,
      "grad_norm": 0.7904365658760071,
      "learning_rate": 7.804949750815506e-06,
      "loss": 0.7262,
      "step": 11492
    },
    {
      "epoch": 0.3311531147352043,
      "grad_norm": 0.7953555583953857,
      "learning_rate": 7.804563468189276e-06,
      "loss": 0.7545,
      "step": 11493
    },
    {
      "epoch": 0.33118192819685355,
      "grad_norm": 0.7944278717041016,
      "learning_rate": 7.8041771611381e-06,
      "loss": 0.7637,
      "step": 11494
    },
    {
      "epoch": 0.33121074165850284,
      "grad_norm": 0.8296322822570801,
      "learning_rate": 7.803790829665345e-06,
      "loss": 0.7358,
      "step": 11495
    },
    {
      "epoch": 0.33123955512015213,
      "grad_norm": 0.8609147071838379,
      "learning_rate": 7.803404473774375e-06,
      "loss": 0.7889,
      "step": 11496
    },
    {
      "epoch": 0.3312683685818014,
      "grad_norm": 0.7983195185661316,
      "learning_rate": 7.803018093468555e-06,
      "loss": 0.7759,
      "step": 11497
    },
    {
      "epoch": 0.3312971820434507,
      "grad_norm": 0.8041146993637085,
      "learning_rate": 7.802631688751248e-06,
      "loss": 0.7463,
      "step": 11498
    },
    {
      "epoch": 0.3313259955051,
      "grad_norm": 0.8081710934638977,
      "learning_rate": 7.80224525962582e-06,
      "loss": 0.7519,
      "step": 11499
    },
    {
      "epoch": 0.3313548089667493,
      "grad_norm": 0.8145257234573364,
      "learning_rate": 7.801858806095637e-06,
      "loss": 0.7387,
      "step": 11500
    },
    {
      "epoch": 0.3313836224283985,
      "grad_norm": 0.7960187196731567,
      "learning_rate": 7.801472328164067e-06,
      "loss": 0.718,
      "step": 11501
    },
    {
      "epoch": 0.3314124358900478,
      "grad_norm": 0.808123767375946,
      "learning_rate": 7.801085825834472e-06,
      "loss": 0.7634,
      "step": 11502
    },
    {
      "epoch": 0.3314412493516971,
      "grad_norm": 0.7709792256355286,
      "learning_rate": 7.800699299110219e-06,
      "loss": 0.7794,
      "step": 11503
    },
    {
      "epoch": 0.3314700628133464,
      "grad_norm": 1.4642757177352905,
      "learning_rate": 7.800312747994675e-06,
      "loss": 0.7582,
      "step": 11504
    },
    {
      "epoch": 0.3314988762749957,
      "grad_norm": 0.7745758295059204,
      "learning_rate": 7.799926172491208e-06,
      "loss": 0.7697,
      "step": 11505
    },
    {
      "epoch": 0.331527689736645,
      "grad_norm": 0.766148567199707,
      "learning_rate": 7.79953957260318e-06,
      "loss": 0.7353,
      "step": 11506
    },
    {
      "epoch": 0.3315565031982942,
      "grad_norm": 0.8024436831474304,
      "learning_rate": 7.799152948333963e-06,
      "loss": 0.751,
      "step": 11507
    },
    {
      "epoch": 0.3315853166599435,
      "grad_norm": 0.8018203973770142,
      "learning_rate": 7.798766299686922e-06,
      "loss": 0.7698,
      "step": 11508
    },
    {
      "epoch": 0.3316141301215928,
      "grad_norm": 0.8134492039680481,
      "learning_rate": 7.798379626665424e-06,
      "loss": 0.7688,
      "step": 11509
    },
    {
      "epoch": 0.3316429435832421,
      "grad_norm": 0.7849413752555847,
      "learning_rate": 7.797992929272837e-06,
      "loss": 0.7483,
      "step": 11510
    },
    {
      "epoch": 0.3316717570448914,
      "grad_norm": 0.8146234750747681,
      "learning_rate": 7.797606207512528e-06,
      "loss": 0.7879,
      "step": 11511
    },
    {
      "epoch": 0.33170057050654067,
      "grad_norm": 0.8156513571739197,
      "learning_rate": 7.797219461387863e-06,
      "loss": 0.7515,
      "step": 11512
    },
    {
      "epoch": 0.33172938396818996,
      "grad_norm": 0.8247729539871216,
      "learning_rate": 7.796832690902216e-06,
      "loss": 0.7399,
      "step": 11513
    },
    {
      "epoch": 0.3317581974298392,
      "grad_norm": 0.8692286014556885,
      "learning_rate": 7.796445896058954e-06,
      "loss": 0.7745,
      "step": 11514
    },
    {
      "epoch": 0.3317870108914885,
      "grad_norm": 0.7922722697257996,
      "learning_rate": 7.79605907686144e-06,
      "loss": 0.7247,
      "step": 11515
    },
    {
      "epoch": 0.3318158243531378,
      "grad_norm": 0.7748432159423828,
      "learning_rate": 7.795672233313047e-06,
      "loss": 0.7661,
      "step": 11516
    },
    {
      "epoch": 0.33184463781478707,
      "grad_norm": 0.8233796954154968,
      "learning_rate": 7.795285365417144e-06,
      "loss": 0.732,
      "step": 11517
    },
    {
      "epoch": 0.33187345127643636,
      "grad_norm": 0.870772659778595,
      "learning_rate": 7.794898473177099e-06,
      "loss": 0.7537,
      "step": 11518
    },
    {
      "epoch": 0.33190226473808565,
      "grad_norm": 0.7835409641265869,
      "learning_rate": 7.794511556596283e-06,
      "loss": 0.778,
      "step": 11519
    },
    {
      "epoch": 0.33193107819973494,
      "grad_norm": 0.7973385453224182,
      "learning_rate": 7.794124615678064e-06,
      "loss": 0.7594,
      "step": 11520
    },
    {
      "epoch": 0.3319598916613842,
      "grad_norm": 1.7480313777923584,
      "learning_rate": 7.793737650425815e-06,
      "loss": 0.7432,
      "step": 11521
    },
    {
      "epoch": 0.33198870512303347,
      "grad_norm": 0.8098764419555664,
      "learning_rate": 7.793350660842902e-06,
      "loss": 0.75,
      "step": 11522
    },
    {
      "epoch": 0.33201751858468276,
      "grad_norm": 0.786561906337738,
      "learning_rate": 7.792963646932696e-06,
      "loss": 0.7414,
      "step": 11523
    },
    {
      "epoch": 0.33204633204633205,
      "grad_norm": 0.7766151428222656,
      "learning_rate": 7.79257660869857e-06,
      "loss": 0.7603,
      "step": 11524
    },
    {
      "epoch": 0.33207514550798134,
      "grad_norm": 0.7988197207450867,
      "learning_rate": 7.792189546143893e-06,
      "loss": 0.7565,
      "step": 11525
    },
    {
      "epoch": 0.33210395896963063,
      "grad_norm": 0.8006371259689331,
      "learning_rate": 7.791802459272036e-06,
      "loss": 0.7058,
      "step": 11526
    },
    {
      "epoch": 0.3321327724312799,
      "grad_norm": 0.8003238439559937,
      "learning_rate": 7.791415348086371e-06,
      "loss": 0.7439,
      "step": 11527
    },
    {
      "epoch": 0.33216158589292916,
      "grad_norm": 0.8189533948898315,
      "learning_rate": 7.791028212590267e-06,
      "loss": 0.7555,
      "step": 11528
    },
    {
      "epoch": 0.33219039935457845,
      "grad_norm": 0.817482590675354,
      "learning_rate": 7.7906410527871e-06,
      "loss": 0.7783,
      "step": 11529
    },
    {
      "epoch": 0.33221921281622774,
      "grad_norm": 0.7850977182388306,
      "learning_rate": 7.790253868680238e-06,
      "loss": 0.7621,
      "step": 11530
    },
    {
      "epoch": 0.33224802627787703,
      "grad_norm": 0.881117582321167,
      "learning_rate": 7.789866660273053e-06,
      "loss": 0.7459,
      "step": 11531
    },
    {
      "epoch": 0.3322768397395263,
      "grad_norm": 0.7860058546066284,
      "learning_rate": 7.78947942756892e-06,
      "loss": 0.7209,
      "step": 11532
    },
    {
      "epoch": 0.3323056532011756,
      "grad_norm": 0.8282033205032349,
      "learning_rate": 7.789092170571208e-06,
      "loss": 0.7437,
      "step": 11533
    },
    {
      "epoch": 0.33233446666282485,
      "grad_norm": 0.7916606068611145,
      "learning_rate": 7.788704889283291e-06,
      "loss": 0.7417,
      "step": 11534
    },
    {
      "epoch": 0.33236328012447414,
      "grad_norm": 0.8343075513839722,
      "learning_rate": 7.788317583708545e-06,
      "loss": 0.7605,
      "step": 11535
    },
    {
      "epoch": 0.3323920935861234,
      "grad_norm": 0.8071552515029907,
      "learning_rate": 7.787930253850337e-06,
      "loss": 0.7452,
      "step": 11536
    },
    {
      "epoch": 0.3324209070477727,
      "grad_norm": 0.7905356287956238,
      "learning_rate": 7.787542899712043e-06,
      "loss": 0.7337,
      "step": 11537
    },
    {
      "epoch": 0.332449720509422,
      "grad_norm": 0.7758591175079346,
      "learning_rate": 7.78715552129704e-06,
      "loss": 0.7283,
      "step": 11538
    },
    {
      "epoch": 0.3324785339710713,
      "grad_norm": 0.8709844350814819,
      "learning_rate": 7.786768118608696e-06,
      "loss": 0.7578,
      "step": 11539
    },
    {
      "epoch": 0.3325073474327206,
      "grad_norm": 0.7822712063789368,
      "learning_rate": 7.78638069165039e-06,
      "loss": 0.7392,
      "step": 11540
    },
    {
      "epoch": 0.3325361608943698,
      "grad_norm": 0.8182845115661621,
      "learning_rate": 7.785993240425491e-06,
      "loss": 0.7438,
      "step": 11541
    },
    {
      "epoch": 0.3325649743560191,
      "grad_norm": 0.7708452939987183,
      "learning_rate": 7.785605764937376e-06,
      "loss": 0.7497,
      "step": 11542
    },
    {
      "epoch": 0.3325937878176684,
      "grad_norm": 0.7931838631629944,
      "learning_rate": 7.78521826518942e-06,
      "loss": 0.744,
      "step": 11543
    },
    {
      "epoch": 0.3326226012793177,
      "grad_norm": 0.8136195540428162,
      "learning_rate": 7.784830741184999e-06,
      "loss": 0.7656,
      "step": 11544
    },
    {
      "epoch": 0.332651414740967,
      "grad_norm": 0.8087642788887024,
      "learning_rate": 7.784443192927484e-06,
      "loss": 0.7602,
      "step": 11545
    },
    {
      "epoch": 0.3326802282026163,
      "grad_norm": 0.7917773723602295,
      "learning_rate": 7.784055620420252e-06,
      "loss": 0.751,
      "step": 11546
    },
    {
      "epoch": 0.33270904166426557,
      "grad_norm": 0.7548835277557373,
      "learning_rate": 7.783668023666679e-06,
      "loss": 0.7481,
      "step": 11547
    },
    {
      "epoch": 0.3327378551259148,
      "grad_norm": 0.8177057504653931,
      "learning_rate": 7.78328040267014e-06,
      "loss": 0.7873,
      "step": 11548
    },
    {
      "epoch": 0.3327666685875641,
      "grad_norm": 1.1136901378631592,
      "learning_rate": 7.782892757434012e-06,
      "loss": 0.7328,
      "step": 11549
    },
    {
      "epoch": 0.3327954820492134,
      "grad_norm": 0.8177999258041382,
      "learning_rate": 7.782505087961669e-06,
      "loss": 0.7486,
      "step": 11550
    },
    {
      "epoch": 0.3328242955108627,
      "grad_norm": 0.7956475615501404,
      "learning_rate": 7.782117394256487e-06,
      "loss": 0.7542,
      "step": 11551
    },
    {
      "epoch": 0.33285310897251197,
      "grad_norm": 0.7965140342712402,
      "learning_rate": 7.781729676321845e-06,
      "loss": 0.7535,
      "step": 11552
    },
    {
      "epoch": 0.33288192243416126,
      "grad_norm": 0.7777013182640076,
      "learning_rate": 7.78134193416112e-06,
      "loss": 0.7413,
      "step": 11553
    },
    {
      "epoch": 0.33291073589581055,
      "grad_norm": 0.7512498497962952,
      "learning_rate": 7.780954167777684e-06,
      "loss": 0.7272,
      "step": 11554
    },
    {
      "epoch": 0.3329395493574598,
      "grad_norm": 0.8257309198379517,
      "learning_rate": 7.780566377174918e-06,
      "loss": 0.7871,
      "step": 11555
    },
    {
      "epoch": 0.3329683628191091,
      "grad_norm": 0.8040263056755066,
      "learning_rate": 7.7801785623562e-06,
      "loss": 0.7516,
      "step": 11556
    },
    {
      "epoch": 0.33299717628075837,
      "grad_norm": 0.8185697793960571,
      "learning_rate": 7.779790723324904e-06,
      "loss": 0.7391,
      "step": 11557
    },
    {
      "epoch": 0.33302598974240766,
      "grad_norm": 0.7685295939445496,
      "learning_rate": 7.779402860084411e-06,
      "loss": 0.746,
      "step": 11558
    },
    {
      "epoch": 0.33305480320405695,
      "grad_norm": 0.8311576843261719,
      "learning_rate": 7.779014972638095e-06,
      "loss": 0.743,
      "step": 11559
    },
    {
      "epoch": 0.33308361666570624,
      "grad_norm": 0.7989012598991394,
      "learning_rate": 7.77862706098934e-06,
      "loss": 0.748,
      "step": 11560
    },
    {
      "epoch": 0.3331124301273555,
      "grad_norm": 0.8316221237182617,
      "learning_rate": 7.778239125141518e-06,
      "loss": 0.7507,
      "step": 11561
    },
    {
      "epoch": 0.33314124358900477,
      "grad_norm": 0.7946360111236572,
      "learning_rate": 7.777851165098012e-06,
      "loss": 0.7584,
      "step": 11562
    },
    {
      "epoch": 0.33317005705065406,
      "grad_norm": 0.7871268391609192,
      "learning_rate": 7.777463180862199e-06,
      "loss": 0.7383,
      "step": 11563
    },
    {
      "epoch": 0.33319887051230335,
      "grad_norm": 0.7925636768341064,
      "learning_rate": 7.777075172437458e-06,
      "loss": 0.7591,
      "step": 11564
    },
    {
      "epoch": 0.33322768397395264,
      "grad_norm": 0.7643693089485168,
      "learning_rate": 7.776687139827168e-06,
      "loss": 0.7601,
      "step": 11565
    },
    {
      "epoch": 0.33325649743560193,
      "grad_norm": 0.808378279209137,
      "learning_rate": 7.776299083034709e-06,
      "loss": 0.7635,
      "step": 11566
    },
    {
      "epoch": 0.3332853108972512,
      "grad_norm": 0.8004292845726013,
      "learning_rate": 7.775911002063462e-06,
      "loss": 0.721,
      "step": 11567
    },
    {
      "epoch": 0.33331412435890045,
      "grad_norm": 11.525429725646973,
      "learning_rate": 7.7755228969168e-06,
      "loss": 0.7598,
      "step": 11568
    },
    {
      "epoch": 0.33334293782054975,
      "grad_norm": 0.8338243961334229,
      "learning_rate": 7.775134767598112e-06,
      "loss": 0.7621,
      "step": 11569
    },
    {
      "epoch": 0.33337175128219904,
      "grad_norm": 0.8431177139282227,
      "learning_rate": 7.774746614110775e-06,
      "loss": 0.7407,
      "step": 11570
    },
    {
      "epoch": 0.3334005647438483,
      "grad_norm": 0.8483486771583557,
      "learning_rate": 7.774358436458168e-06,
      "loss": 0.752,
      "step": 11571
    },
    {
      "epoch": 0.3334293782054976,
      "grad_norm": 0.8597800731658936,
      "learning_rate": 7.773970234643671e-06,
      "loss": 0.7614,
      "step": 11572
    },
    {
      "epoch": 0.3334581916671469,
      "grad_norm": 0.8277650475502014,
      "learning_rate": 7.773582008670667e-06,
      "loss": 0.7438,
      "step": 11573
    },
    {
      "epoch": 0.3334870051287962,
      "grad_norm": 0.9122857451438904,
      "learning_rate": 7.773193758542537e-06,
      "loss": 0.782,
      "step": 11574
    },
    {
      "epoch": 0.33351581859044543,
      "grad_norm": 0.8134661316871643,
      "learning_rate": 7.77280548426266e-06,
      "loss": 0.7275,
      "step": 11575
    },
    {
      "epoch": 0.3335446320520947,
      "grad_norm": 1.6395710706710815,
      "learning_rate": 7.77241718583442e-06,
      "loss": 0.7468,
      "step": 11576
    },
    {
      "epoch": 0.333573445513744,
      "grad_norm": 3.3975000381469727,
      "learning_rate": 7.772028863261197e-06,
      "loss": 0.7488,
      "step": 11577
    },
    {
      "epoch": 0.3336022589753933,
      "grad_norm": 0.9106335639953613,
      "learning_rate": 7.771640516546373e-06,
      "loss": 0.7446,
      "step": 11578
    },
    {
      "epoch": 0.3336310724370426,
      "grad_norm": 0.8745089173316956,
      "learning_rate": 7.771252145693333e-06,
      "loss": 0.7409,
      "step": 11579
    },
    {
      "epoch": 0.3336598858986919,
      "grad_norm": 0.8408945202827454,
      "learning_rate": 7.770863750705454e-06,
      "loss": 0.7571,
      "step": 11580
    },
    {
      "epoch": 0.3336886993603412,
      "grad_norm": 0.9282495975494385,
      "learning_rate": 7.770475331586126e-06,
      "loss": 0.7674,
      "step": 11581
    },
    {
      "epoch": 0.3337175128219904,
      "grad_norm": 0.8494476675987244,
      "learning_rate": 7.770086888338724e-06,
      "loss": 0.7607,
      "step": 11582
    },
    {
      "epoch": 0.3337463262836397,
      "grad_norm": 0.9036874771118164,
      "learning_rate": 7.769698420966634e-06,
      "loss": 0.765,
      "step": 11583
    },
    {
      "epoch": 0.333775139745289,
      "grad_norm": 0.8471857905387878,
      "learning_rate": 7.769309929473241e-06,
      "loss": 0.763,
      "step": 11584
    },
    {
      "epoch": 0.3338039532069383,
      "grad_norm": 0.7953132390975952,
      "learning_rate": 7.768921413861928e-06,
      "loss": 0.7478,
      "step": 11585
    },
    {
      "epoch": 0.3338327666685876,
      "grad_norm": 0.8169087767601013,
      "learning_rate": 7.768532874136075e-06,
      "loss": 0.757,
      "step": 11586
    },
    {
      "epoch": 0.33386158013023687,
      "grad_norm": 0.8700182437896729,
      "learning_rate": 7.768144310299068e-06,
      "loss": 0.7499,
      "step": 11587
    },
    {
      "epoch": 0.3338903935918861,
      "grad_norm": 0.83025062084198,
      "learning_rate": 7.767755722354292e-06,
      "loss": 0.7527,
      "step": 11588
    },
    {
      "epoch": 0.3339192070535354,
      "grad_norm": 0.8663896322250366,
      "learning_rate": 7.76736711030513e-06,
      "loss": 0.7711,
      "step": 11589
    },
    {
      "epoch": 0.3339480205151847,
      "grad_norm": 0.8297219276428223,
      "learning_rate": 7.766978474154968e-06,
      "loss": 0.7433,
      "step": 11590
    },
    {
      "epoch": 0.333976833976834,
      "grad_norm": 0.847133994102478,
      "learning_rate": 7.766589813907189e-06,
      "loss": 0.7597,
      "step": 11591
    },
    {
      "epoch": 0.33400564743848327,
      "grad_norm": 0.8107408881187439,
      "learning_rate": 7.766201129565178e-06,
      "loss": 0.7403,
      "step": 11592
    },
    {
      "epoch": 0.33403446090013256,
      "grad_norm": 0.812193751335144,
      "learning_rate": 7.765812421132322e-06,
      "loss": 0.761,
      "step": 11593
    },
    {
      "epoch": 0.33406327436178185,
      "grad_norm": 0.8206763863563538,
      "learning_rate": 7.765423688612002e-06,
      "loss": 0.7339,
      "step": 11594
    },
    {
      "epoch": 0.3340920878234311,
      "grad_norm": 0.8169931769371033,
      "learning_rate": 7.765034932007607e-06,
      "loss": 0.7161,
      "step": 11595
    },
    {
      "epoch": 0.3341209012850804,
      "grad_norm": 0.7993503212928772,
      "learning_rate": 7.764646151322522e-06,
      "loss": 0.7568,
      "step": 11596
    },
    {
      "epoch": 0.33414971474672966,
      "grad_norm": 0.820515513420105,
      "learning_rate": 7.764257346560132e-06,
      "loss": 0.7646,
      "step": 11597
    },
    {
      "epoch": 0.33417852820837896,
      "grad_norm": 0.8170409202575684,
      "learning_rate": 7.763868517723825e-06,
      "loss": 0.7242,
      "step": 11598
    },
    {
      "epoch": 0.33420734167002825,
      "grad_norm": 0.7969456911087036,
      "learning_rate": 7.763479664816986e-06,
      "loss": 0.7224,
      "step": 11599
    },
    {
      "epoch": 0.33423615513167754,
      "grad_norm": 0.7903942465782166,
      "learning_rate": 7.763090787843e-06,
      "loss": 0.7235,
      "step": 11600
    },
    {
      "epoch": 0.33426496859332683,
      "grad_norm": 0.8018198013305664,
      "learning_rate": 7.762701886805254e-06,
      "loss": 0.7369,
      "step": 11601
    },
    {
      "epoch": 0.33429378205497606,
      "grad_norm": 0.7724486589431763,
      "learning_rate": 7.762312961707142e-06,
      "loss": 0.765,
      "step": 11602
    },
    {
      "epoch": 0.33432259551662535,
      "grad_norm": 0.7987584471702576,
      "learning_rate": 7.76192401255204e-06,
      "loss": 0.7322,
      "step": 11603
    },
    {
      "epoch": 0.33435140897827464,
      "grad_norm": 0.8158887624740601,
      "learning_rate": 7.761535039343345e-06,
      "loss": 0.7657,
      "step": 11604
    },
    {
      "epoch": 0.33438022243992394,
      "grad_norm": 0.8235997557640076,
      "learning_rate": 7.761146042084437e-06,
      "loss": 0.7513,
      "step": 11605
    },
    {
      "epoch": 0.3344090359015732,
      "grad_norm": 0.8274944424629211,
      "learning_rate": 7.760757020778707e-06,
      "loss": 0.7916,
      "step": 11606
    },
    {
      "epoch": 0.3344378493632225,
      "grad_norm": 0.813456654548645,
      "learning_rate": 7.760367975429545e-06,
      "loss": 0.7403,
      "step": 11607
    },
    {
      "epoch": 0.3344666628248718,
      "grad_norm": 0.866884708404541,
      "learning_rate": 7.759978906040335e-06,
      "loss": 0.7536,
      "step": 11608
    },
    {
      "epoch": 0.33449547628652104,
      "grad_norm": 0.7950873374938965,
      "learning_rate": 7.759589812614469e-06,
      "loss": 0.7522,
      "step": 11609
    },
    {
      "epoch": 0.33452428974817033,
      "grad_norm": 0.7899826765060425,
      "learning_rate": 7.759200695155336e-06,
      "loss": 0.749,
      "step": 11610
    },
    {
      "epoch": 0.3345531032098196,
      "grad_norm": 0.7718024253845215,
      "learning_rate": 7.758811553666323e-06,
      "loss": 0.7411,
      "step": 11611
    },
    {
      "epoch": 0.3345819166714689,
      "grad_norm": 0.8274055123329163,
      "learning_rate": 7.758422388150817e-06,
      "loss": 0.759,
      "step": 11612
    },
    {
      "epoch": 0.3346107301331182,
      "grad_norm": 0.7914291024208069,
      "learning_rate": 7.758033198612209e-06,
      "loss": 0.7662,
      "step": 11613
    },
    {
      "epoch": 0.3346395435947675,
      "grad_norm": 0.7641514539718628,
      "learning_rate": 7.75764398505389e-06,
      "loss": 0.7351,
      "step": 11614
    },
    {
      "epoch": 0.33466835705641673,
      "grad_norm": 0.822497546672821,
      "learning_rate": 7.757254747479251e-06,
      "loss": 0.7625,
      "step": 11615
    },
    {
      "epoch": 0.334697170518066,
      "grad_norm": 0.7646114230155945,
      "learning_rate": 7.756865485891676e-06,
      "loss": 0.7644,
      "step": 11616
    },
    {
      "epoch": 0.3347259839797153,
      "grad_norm": 0.8185021877288818,
      "learning_rate": 7.75647620029456e-06,
      "loss": 0.7683,
      "step": 11617
    },
    {
      "epoch": 0.3347547974413646,
      "grad_norm": 0.7909248471260071,
      "learning_rate": 7.75608689069129e-06,
      "loss": 0.7383,
      "step": 11618
    },
    {
      "epoch": 0.3347836109030139,
      "grad_norm": 0.7843075394630432,
      "learning_rate": 7.75569755708526e-06,
      "loss": 0.7543,
      "step": 11619
    },
    {
      "epoch": 0.3348124243646632,
      "grad_norm": 0.8312982320785522,
      "learning_rate": 7.75530819947986e-06,
      "loss": 0.7437,
      "step": 11620
    },
    {
      "epoch": 0.3348412378263125,
      "grad_norm": 0.7592137455940247,
      "learning_rate": 7.754918817878477e-06,
      "loss": 0.7183,
      "step": 11621
    },
    {
      "epoch": 0.3348700512879617,
      "grad_norm": 0.820790708065033,
      "learning_rate": 7.754529412284506e-06,
      "loss": 0.7461,
      "step": 11622
    },
    {
      "epoch": 0.334898864749611,
      "grad_norm": 0.7826203107833862,
      "learning_rate": 7.754139982701338e-06,
      "loss": 0.7309,
      "step": 11623
    },
    {
      "epoch": 0.3349276782112603,
      "grad_norm": 0.7865408658981323,
      "learning_rate": 7.753750529132365e-06,
      "loss": 0.7326,
      "step": 11624
    },
    {
      "epoch": 0.3349564916729096,
      "grad_norm": 0.7562050223350525,
      "learning_rate": 7.753361051580974e-06,
      "loss": 0.7355,
      "step": 11625
    },
    {
      "epoch": 0.3349853051345589,
      "grad_norm": 0.8035718202590942,
      "learning_rate": 7.752971550050562e-06,
      "loss": 0.7771,
      "step": 11626
    },
    {
      "epoch": 0.33501411859620817,
      "grad_norm": 0.7962263226509094,
      "learning_rate": 7.752582024544521e-06,
      "loss": 0.7623,
      "step": 11627
    },
    {
      "epoch": 0.33504293205785746,
      "grad_norm": 0.7792998552322388,
      "learning_rate": 7.752192475066243e-06,
      "loss": 0.7404,
      "step": 11628
    },
    {
      "epoch": 0.3350717455195067,
      "grad_norm": 0.8144372701644897,
      "learning_rate": 7.75180290161912e-06,
      "loss": 0.7631,
      "step": 11629
    },
    {
      "epoch": 0.335100558981156,
      "grad_norm": 0.7728071212768555,
      "learning_rate": 7.751413304206542e-06,
      "loss": 0.7766,
      "step": 11630
    },
    {
      "epoch": 0.3351293724428053,
      "grad_norm": 0.783244252204895,
      "learning_rate": 7.751023682831904e-06,
      "loss": 0.719,
      "step": 11631
    },
    {
      "epoch": 0.33515818590445456,
      "grad_norm": 0.7991316318511963,
      "learning_rate": 7.750634037498602e-06,
      "loss": 0.7621,
      "step": 11632
    },
    {
      "epoch": 0.33518699936610385,
      "grad_norm": 0.7764188647270203,
      "learning_rate": 7.750244368210028e-06,
      "loss": 0.7487,
      "step": 11633
    },
    {
      "epoch": 0.33521581282775315,
      "grad_norm": 0.7710868120193481,
      "learning_rate": 7.749854674969572e-06,
      "loss": 0.7507,
      "step": 11634
    },
    {
      "epoch": 0.33524462628940244,
      "grad_norm": 0.7939848899841309,
      "learning_rate": 7.749464957780632e-06,
      "loss": 0.7563,
      "step": 11635
    },
    {
      "epoch": 0.33527343975105167,
      "grad_norm": 0.7877445816993713,
      "learning_rate": 7.7490752166466e-06,
      "loss": 0.7456,
      "step": 11636
    },
    {
      "epoch": 0.33530225321270096,
      "grad_norm": 0.7471206784248352,
      "learning_rate": 7.748685451570872e-06,
      "loss": 0.7353,
      "step": 11637
    },
    {
      "epoch": 0.33533106667435025,
      "grad_norm": 0.7983543872833252,
      "learning_rate": 7.74829566255684e-06,
      "loss": 0.7587,
      "step": 11638
    },
    {
      "epoch": 0.33535988013599954,
      "grad_norm": 0.7968281507492065,
      "learning_rate": 7.747905849607901e-06,
      "loss": 0.8006,
      "step": 11639
    },
    {
      "epoch": 0.33538869359764883,
      "grad_norm": 0.7855402827262878,
      "learning_rate": 7.74751601272745e-06,
      "loss": 0.7625,
      "step": 11640
    },
    {
      "epoch": 0.3354175070592981,
      "grad_norm": 0.7667627930641174,
      "learning_rate": 7.74712615191888e-06,
      "loss": 0.741,
      "step": 11641
    },
    {
      "epoch": 0.33544632052094736,
      "grad_norm": 0.7806314826011658,
      "learning_rate": 7.746736267185588e-06,
      "loss": 0.7396,
      "step": 11642
    },
    {
      "epoch": 0.33547513398259665,
      "grad_norm": 0.7779011726379395,
      "learning_rate": 7.746346358530967e-06,
      "loss": 0.7315,
      "step": 11643
    },
    {
      "epoch": 0.33550394744424594,
      "grad_norm": 0.7842736840248108,
      "learning_rate": 7.745956425958417e-06,
      "loss": 0.714,
      "step": 11644
    },
    {
      "epoch": 0.33553276090589523,
      "grad_norm": 0.8008041977882385,
      "learning_rate": 7.74556646947133e-06,
      "loss": 0.74,
      "step": 11645
    },
    {
      "epoch": 0.3355615743675445,
      "grad_norm": 0.7687137126922607,
      "learning_rate": 7.745176489073103e-06,
      "loss": 0.7506,
      "step": 11646
    },
    {
      "epoch": 0.3355903878291938,
      "grad_norm": 0.7619121074676514,
      "learning_rate": 7.744786484767133e-06,
      "loss": 0.7434,
      "step": 11647
    },
    {
      "epoch": 0.3356192012908431,
      "grad_norm": 0.7946091890335083,
      "learning_rate": 7.744396456556817e-06,
      "loss": 0.7656,
      "step": 11648
    },
    {
      "epoch": 0.33564801475249234,
      "grad_norm": 0.8090528249740601,
      "learning_rate": 7.744006404445553e-06,
      "loss": 0.7572,
      "step": 11649
    },
    {
      "epoch": 0.33567682821414163,
      "grad_norm": 0.7831555008888245,
      "learning_rate": 7.743616328436734e-06,
      "loss": 0.7368,
      "step": 11650
    },
    {
      "epoch": 0.3357056416757909,
      "grad_norm": 0.7818573713302612,
      "learning_rate": 7.74322622853376e-06,
      "loss": 0.7482,
      "step": 11651
    },
    {
      "epoch": 0.3357344551374402,
      "grad_norm": 0.7627626061439514,
      "learning_rate": 7.742836104740028e-06,
      "loss": 0.7215,
      "step": 11652
    },
    {
      "epoch": 0.3357632685990895,
      "grad_norm": 0.7727810144424438,
      "learning_rate": 7.742445957058934e-06,
      "loss": 0.7462,
      "step": 11653
    },
    {
      "epoch": 0.3357920820607388,
      "grad_norm": 0.7922280430793762,
      "learning_rate": 7.74205578549388e-06,
      "loss": 0.7482,
      "step": 11654
    },
    {
      "epoch": 0.3358208955223881,
      "grad_norm": 0.8037744164466858,
      "learning_rate": 7.741665590048259e-06,
      "loss": 0.774,
      "step": 11655
    },
    {
      "epoch": 0.3358497089840373,
      "grad_norm": 0.7859479188919067,
      "learning_rate": 7.741275370725472e-06,
      "loss": 0.7547,
      "step": 11656
    },
    {
      "epoch": 0.3358785224456866,
      "grad_norm": 0.7971737384796143,
      "learning_rate": 7.740885127528916e-06,
      "loss": 0.7618,
      "step": 11657
    },
    {
      "epoch": 0.3359073359073359,
      "grad_norm": 0.838263988494873,
      "learning_rate": 7.74049486046199e-06,
      "loss": 0.7457,
      "step": 11658
    },
    {
      "epoch": 0.3359361493689852,
      "grad_norm": 0.7772573232650757,
      "learning_rate": 7.740104569528096e-06,
      "loss": 0.7539,
      "step": 11659
    },
    {
      "epoch": 0.3359649628306345,
      "grad_norm": 0.7743715047836304,
      "learning_rate": 7.739714254730629e-06,
      "loss": 0.7419,
      "step": 11660
    },
    {
      "epoch": 0.3359937762922838,
      "grad_norm": 0.7500496506690979,
      "learning_rate": 7.739323916072989e-06,
      "loss": 0.7138,
      "step": 11661
    },
    {
      "epoch": 0.336022589753933,
      "grad_norm": 0.8277460932731628,
      "learning_rate": 7.738933553558576e-06,
      "loss": 0.7795,
      "step": 11662
    },
    {
      "epoch": 0.3360514032155823,
      "grad_norm": 0.8291163444519043,
      "learning_rate": 7.73854316719079e-06,
      "loss": 0.765,
      "step": 11663
    },
    {
      "epoch": 0.3360802166772316,
      "grad_norm": 0.7782652378082275,
      "learning_rate": 7.73815275697303e-06,
      "loss": 0.7332,
      "step": 11664
    },
    {
      "epoch": 0.3361090301388809,
      "grad_norm": 0.8073041439056396,
      "learning_rate": 7.737762322908697e-06,
      "loss": 0.7506,
      "step": 11665
    },
    {
      "epoch": 0.3361378436005302,
      "grad_norm": 0.8039801120758057,
      "learning_rate": 7.737371865001191e-06,
      "loss": 0.7484,
      "step": 11666
    },
    {
      "epoch": 0.33616665706217946,
      "grad_norm": 0.8155747056007385,
      "learning_rate": 7.736981383253911e-06,
      "loss": 0.752,
      "step": 11667
    },
    {
      "epoch": 0.33619547052382875,
      "grad_norm": 0.7704663872718811,
      "learning_rate": 7.736590877670262e-06,
      "loss": 0.7405,
      "step": 11668
    },
    {
      "epoch": 0.336224283985478,
      "grad_norm": 0.8425151109695435,
      "learning_rate": 7.736200348253641e-06,
      "loss": 0.752,
      "step": 11669
    },
    {
      "epoch": 0.3362530974471273,
      "grad_norm": 0.8166927099227905,
      "learning_rate": 7.73580979500745e-06,
      "loss": 0.7284,
      "step": 11670
    },
    {
      "epoch": 0.33628191090877657,
      "grad_norm": 0.7684778571128845,
      "learning_rate": 7.73541921793509e-06,
      "loss": 0.774,
      "step": 11671
    },
    {
      "epoch": 0.33631072437042586,
      "grad_norm": 0.7598613500595093,
      "learning_rate": 7.735028617039964e-06,
      "loss": 0.7254,
      "step": 11672
    },
    {
      "epoch": 0.33633953783207515,
      "grad_norm": 0.8317673206329346,
      "learning_rate": 7.734637992325471e-06,
      "loss": 0.7479,
      "step": 11673
    },
    {
      "epoch": 0.33636835129372444,
      "grad_norm": 0.774455189704895,
      "learning_rate": 7.734247343795017e-06,
      "loss": 0.7219,
      "step": 11674
    },
    {
      "epoch": 0.33639716475537373,
      "grad_norm": 0.7792340517044067,
      "learning_rate": 7.733856671452e-06,
      "loss": 0.7449,
      "step": 11675
    },
    {
      "epoch": 0.33642597821702297,
      "grad_norm": 0.8027287721633911,
      "learning_rate": 7.733465975299825e-06,
      "loss": 0.7635,
      "step": 11676
    },
    {
      "epoch": 0.33645479167867226,
      "grad_norm": 0.8015628457069397,
      "learning_rate": 7.733075255341892e-06,
      "loss": 0.7438,
      "step": 11677
    },
    {
      "epoch": 0.33648360514032155,
      "grad_norm": 0.8269340395927429,
      "learning_rate": 7.73268451158161e-06,
      "loss": 0.767,
      "step": 11678
    },
    {
      "epoch": 0.33651241860197084,
      "grad_norm": 0.7951765656471252,
      "learning_rate": 7.732293744022375e-06,
      "loss": 0.7663,
      "step": 11679
    },
    {
      "epoch": 0.33654123206362013,
      "grad_norm": 0.8322068452835083,
      "learning_rate": 7.731902952667592e-06,
      "loss": 0.738,
      "step": 11680
    },
    {
      "epoch": 0.3365700455252694,
      "grad_norm": 0.8159688711166382,
      "learning_rate": 7.731512137520666e-06,
      "loss": 0.7341,
      "step": 11681
    },
    {
      "epoch": 0.3365988589869187,
      "grad_norm": 0.8193602561950684,
      "learning_rate": 7.731121298585001e-06,
      "loss": 0.7773,
      "step": 11682
    },
    {
      "epoch": 0.33662767244856795,
      "grad_norm": 0.7899747490882874,
      "learning_rate": 7.730730435863997e-06,
      "loss": 0.7606,
      "step": 11683
    },
    {
      "epoch": 0.33665648591021724,
      "grad_norm": 0.8090016841888428,
      "learning_rate": 7.730339549361063e-06,
      "loss": 0.7107,
      "step": 11684
    },
    {
      "epoch": 0.33668529937186653,
      "grad_norm": 0.8148398399353027,
      "learning_rate": 7.7299486390796e-06,
      "loss": 0.7494,
      "step": 11685
    },
    {
      "epoch": 0.3367141128335158,
      "grad_norm": 0.8347864747047424,
      "learning_rate": 7.729557705023012e-06,
      "loss": 0.7238,
      "step": 11686
    },
    {
      "epoch": 0.3367429262951651,
      "grad_norm": 0.801960289478302,
      "learning_rate": 7.729166747194706e-06,
      "loss": 0.7466,
      "step": 11687
    },
    {
      "epoch": 0.3367717397568144,
      "grad_norm": 0.78859543800354,
      "learning_rate": 7.728775765598086e-06,
      "loss": 0.731,
      "step": 11688
    },
    {
      "epoch": 0.33680055321846364,
      "grad_norm": 0.7793959379196167,
      "learning_rate": 7.728384760236556e-06,
      "loss": 0.7327,
      "step": 11689
    },
    {
      "epoch": 0.33682936668011293,
      "grad_norm": 0.7702480554580688,
      "learning_rate": 7.727993731113523e-06,
      "loss": 0.7565,
      "step": 11690
    },
    {
      "epoch": 0.3368581801417622,
      "grad_norm": 0.7959219217300415,
      "learning_rate": 7.72760267823239e-06,
      "loss": 0.7513,
      "step": 11691
    },
    {
      "epoch": 0.3368869936034115,
      "grad_norm": 0.7854728698730469,
      "learning_rate": 7.727211601596567e-06,
      "loss": 0.7264,
      "step": 11692
    },
    {
      "epoch": 0.3369158070650608,
      "grad_norm": 0.7610155940055847,
      "learning_rate": 7.726820501209456e-06,
      "loss": 0.7512,
      "step": 11693
    },
    {
      "epoch": 0.3369446205267101,
      "grad_norm": 0.8013639450073242,
      "learning_rate": 7.726429377074464e-06,
      "loss": 0.7501,
      "step": 11694
    },
    {
      "epoch": 0.3369734339883594,
      "grad_norm": 0.7786518931388855,
      "learning_rate": 7.726038229194999e-06,
      "loss": 0.7405,
      "step": 11695
    },
    {
      "epoch": 0.3370022474500086,
      "grad_norm": 0.795053243637085,
      "learning_rate": 7.725647057574465e-06,
      "loss": 0.7502,
      "step": 11696
    },
    {
      "epoch": 0.3370310609116579,
      "grad_norm": 0.7555896639823914,
      "learning_rate": 7.725255862216268e-06,
      "loss": 0.7175,
      "step": 11697
    },
    {
      "epoch": 0.3370598743733072,
      "grad_norm": 0.8081674575805664,
      "learning_rate": 7.72486464312382e-06,
      "loss": 0.7456,
      "step": 11698
    },
    {
      "epoch": 0.3370886878349565,
      "grad_norm": 0.7826142311096191,
      "learning_rate": 7.724473400300523e-06,
      "loss": 0.7361,
      "step": 11699
    },
    {
      "epoch": 0.3371175012966058,
      "grad_norm": 0.8472177386283875,
      "learning_rate": 7.724082133749786e-06,
      "loss": 0.7304,
      "step": 11700
    },
    {
      "epoch": 0.3371463147582551,
      "grad_norm": 0.7695299983024597,
      "learning_rate": 7.723690843475017e-06,
      "loss": 0.7394,
      "step": 11701
    },
    {
      "epoch": 0.33717512821990436,
      "grad_norm": 0.7757649421691895,
      "learning_rate": 7.723299529479623e-06,
      "loss": 0.7369,
      "step": 11702
    },
    {
      "epoch": 0.3372039416815536,
      "grad_norm": 0.7848504185676575,
      "learning_rate": 7.722908191767013e-06,
      "loss": 0.7488,
      "step": 11703
    },
    {
      "epoch": 0.3372327551432029,
      "grad_norm": 0.7840172052383423,
      "learning_rate": 7.722516830340594e-06,
      "loss": 0.7658,
      "step": 11704
    },
    {
      "epoch": 0.3372615686048522,
      "grad_norm": 0.7951980829238892,
      "learning_rate": 7.722125445203775e-06,
      "loss": 0.7323,
      "step": 11705
    },
    {
      "epoch": 0.33729038206650147,
      "grad_norm": 0.7995883226394653,
      "learning_rate": 7.721734036359964e-06,
      "loss": 0.7427,
      "step": 11706
    },
    {
      "epoch": 0.33731919552815076,
      "grad_norm": 0.7656635046005249,
      "learning_rate": 7.721342603812571e-06,
      "loss": 0.7526,
      "step": 11707
    },
    {
      "epoch": 0.33734800898980005,
      "grad_norm": 0.7734960913658142,
      "learning_rate": 7.720951147565005e-06,
      "loss": 0.7343,
      "step": 11708
    },
    {
      "epoch": 0.33737682245144934,
      "grad_norm": 0.7884484529495239,
      "learning_rate": 7.720559667620674e-06,
      "loss": 0.7142,
      "step": 11709
    },
    {
      "epoch": 0.3374056359130986,
      "grad_norm": 0.7609842419624329,
      "learning_rate": 7.720168163982989e-06,
      "loss": 0.7651,
      "step": 11710
    },
    {
      "epoch": 0.33743444937474787,
      "grad_norm": 0.7893567085266113,
      "learning_rate": 7.719776636655357e-06,
      "loss": 0.735,
      "step": 11711
    },
    {
      "epoch": 0.33746326283639716,
      "grad_norm": 0.7851643562316895,
      "learning_rate": 7.719385085641189e-06,
      "loss": 0.7419,
      "step": 11712
    },
    {
      "epoch": 0.33749207629804645,
      "grad_norm": 0.7853624820709229,
      "learning_rate": 7.718993510943895e-06,
      "loss": 0.7586,
      "step": 11713
    },
    {
      "epoch": 0.33752088975969574,
      "grad_norm": 0.7833537459373474,
      "learning_rate": 7.718601912566887e-06,
      "loss": 0.7451,
      "step": 11714
    },
    {
      "epoch": 0.33754970322134503,
      "grad_norm": 0.761049211025238,
      "learning_rate": 7.718210290513572e-06,
      "loss": 0.7442,
      "step": 11715
    },
    {
      "epoch": 0.33757851668299427,
      "grad_norm": 0.8095692992210388,
      "learning_rate": 7.717818644787365e-06,
      "loss": 0.7364,
      "step": 11716
    },
    {
      "epoch": 0.33760733014464356,
      "grad_norm": 0.8008050918579102,
      "learning_rate": 7.717426975391672e-06,
      "loss": 0.7339,
      "step": 11717
    },
    {
      "epoch": 0.33763614360629285,
      "grad_norm": 0.7889611721038818,
      "learning_rate": 7.717035282329908e-06,
      "loss": 0.7331,
      "step": 11718
    },
    {
      "epoch": 0.33766495706794214,
      "grad_norm": 0.8010131120681763,
      "learning_rate": 7.716643565605482e-06,
      "loss": 0.7401,
      "step": 11719
    },
    {
      "epoch": 0.33769377052959143,
      "grad_norm": 0.7780081629753113,
      "learning_rate": 7.716251825221808e-06,
      "loss": 0.7546,
      "step": 11720
    },
    {
      "epoch": 0.3377225839912407,
      "grad_norm": 0.7742897868156433,
      "learning_rate": 7.715860061182295e-06,
      "loss": 0.7547,
      "step": 11721
    },
    {
      "epoch": 0.33775139745289,
      "grad_norm": 0.7979384660720825,
      "learning_rate": 7.715468273490354e-06,
      "loss": 0.7592,
      "step": 11722
    },
    {
      "epoch": 0.33778021091453925,
      "grad_norm": 0.7435106039047241,
      "learning_rate": 7.7150764621494e-06,
      "loss": 0.7379,
      "step": 11723
    },
    {
      "epoch": 0.33780902437618854,
      "grad_norm": 1.7808412313461304,
      "learning_rate": 7.714684627162844e-06,
      "loss": 0.7661,
      "step": 11724
    },
    {
      "epoch": 0.33783783783783783,
      "grad_norm": 0.7833999395370483,
      "learning_rate": 7.7142927685341e-06,
      "loss": 0.7492,
      "step": 11725
    },
    {
      "epoch": 0.3378666512994871,
      "grad_norm": 0.7864347696304321,
      "learning_rate": 7.713900886266577e-06,
      "loss": 0.7462,
      "step": 11726
    },
    {
      "epoch": 0.3378954647611364,
      "grad_norm": 0.7555538415908813,
      "learning_rate": 7.71350898036369e-06,
      "loss": 0.728,
      "step": 11727
    },
    {
      "epoch": 0.3379242782227857,
      "grad_norm": 0.7509418725967407,
      "learning_rate": 7.713117050828854e-06,
      "loss": 0.7364,
      "step": 11728
    },
    {
      "epoch": 0.337953091684435,
      "grad_norm": 0.7996344566345215,
      "learning_rate": 7.71272509766548e-06,
      "loss": 0.7344,
      "step": 11729
    },
    {
      "epoch": 0.3379819051460842,
      "grad_norm": 0.7773727774620056,
      "learning_rate": 7.712333120876983e-06,
      "loss": 0.7472,
      "step": 11730
    },
    {
      "epoch": 0.3380107186077335,
      "grad_norm": 2.811257839202881,
      "learning_rate": 7.711941120466774e-06,
      "loss": 0.745,
      "step": 11731
    },
    {
      "epoch": 0.3380395320693828,
      "grad_norm": 0.8101869225502014,
      "learning_rate": 7.711549096438272e-06,
      "loss": 0.7521,
      "step": 11732
    },
    {
      "epoch": 0.3380683455310321,
      "grad_norm": 0.7905840277671814,
      "learning_rate": 7.711157048794884e-06,
      "loss": 0.7538,
      "step": 11733
    },
    {
      "epoch": 0.3380971589926814,
      "grad_norm": 0.7924931645393372,
      "learning_rate": 7.71076497754003e-06,
      "loss": 0.756,
      "step": 11734
    },
    {
      "epoch": 0.3381259724543307,
      "grad_norm": 0.763251006603241,
      "learning_rate": 7.710372882677124e-06,
      "loss": 0.7372,
      "step": 11735
    },
    {
      "epoch": 0.33815478591597997,
      "grad_norm": 0.7469669580459595,
      "learning_rate": 7.709980764209577e-06,
      "loss": 0.7267,
      "step": 11736
    },
    {
      "epoch": 0.3381835993776292,
      "grad_norm": 0.7824147343635559,
      "learning_rate": 7.70958862214081e-06,
      "loss": 0.731,
      "step": 11737
    },
    {
      "epoch": 0.3382124128392785,
      "grad_norm": 0.8129638433456421,
      "learning_rate": 7.709196456474232e-06,
      "loss": 0.7236,
      "step": 11738
    },
    {
      "epoch": 0.3382412263009278,
      "grad_norm": 0.7744210958480835,
      "learning_rate": 7.708804267213261e-06,
      "loss": 0.7328,
      "step": 11739
    },
    {
      "epoch": 0.3382700397625771,
      "grad_norm": 0.7828415632247925,
      "learning_rate": 7.708412054361313e-06,
      "loss": 0.7314,
      "step": 11740
    },
    {
      "epoch": 0.33829885322422637,
      "grad_norm": 0.8012816905975342,
      "learning_rate": 7.708019817921804e-06,
      "loss": 0.7447,
      "step": 11741
    },
    {
      "epoch": 0.33832766668587566,
      "grad_norm": 0.8446606397628784,
      "learning_rate": 7.70762755789815e-06,
      "loss": 0.7368,
      "step": 11742
    },
    {
      "epoch": 0.3383564801475249,
      "grad_norm": 0.7759518623352051,
      "learning_rate": 7.707235274293765e-06,
      "loss": 0.7482,
      "step": 11743
    },
    {
      "epoch": 0.3383852936091742,
      "grad_norm": 0.7523435950279236,
      "learning_rate": 7.706842967112069e-06,
      "loss": 0.7114,
      "step": 11744
    },
    {
      "epoch": 0.3384141070708235,
      "grad_norm": 0.7613064646720886,
      "learning_rate": 7.706450636356475e-06,
      "loss": 0.7529,
      "step": 11745
    },
    {
      "epoch": 0.33844292053247277,
      "grad_norm": 0.8244365453720093,
      "learning_rate": 7.706058282030402e-06,
      "loss": 0.7181,
      "step": 11746
    },
    {
      "epoch": 0.33847173399412206,
      "grad_norm": 0.8015828132629395,
      "learning_rate": 7.705665904137265e-06,
      "loss": 0.7435,
      "step": 11747
    },
    {
      "epoch": 0.33850054745577135,
      "grad_norm": 0.7625389695167542,
      "learning_rate": 7.705273502680483e-06,
      "loss": 0.7037,
      "step": 11748
    },
    {
      "epoch": 0.33852936091742064,
      "grad_norm": 0.807523250579834,
      "learning_rate": 7.704881077663475e-06,
      "loss": 0.7514,
      "step": 11749
    },
    {
      "epoch": 0.3385581743790699,
      "grad_norm": 0.8147759437561035,
      "learning_rate": 7.704488629089657e-06,
      "loss": 0.7578,
      "step": 11750
    },
    {
      "epoch": 0.33858698784071917,
      "grad_norm": 0.7852419018745422,
      "learning_rate": 7.704096156962444e-06,
      "loss": 0.7541,
      "step": 11751
    },
    {
      "epoch": 0.33861580130236846,
      "grad_norm": 0.8160160183906555,
      "learning_rate": 7.703703661285258e-06,
      "loss": 0.7674,
      "step": 11752
    },
    {
      "epoch": 0.33864461476401775,
      "grad_norm": 0.8183078169822693,
      "learning_rate": 7.703311142061516e-06,
      "loss": 0.7419,
      "step": 11753
    },
    {
      "epoch": 0.33867342822566704,
      "grad_norm": 0.7674232721328735,
      "learning_rate": 7.702918599294636e-06,
      "loss": 0.741,
      "step": 11754
    },
    {
      "epoch": 0.33870224168731633,
      "grad_norm": 0.7995744943618774,
      "learning_rate": 7.702526032988038e-06,
      "loss": 0.755,
      "step": 11755
    },
    {
      "epoch": 0.3387310551489656,
      "grad_norm": 0.7845680117607117,
      "learning_rate": 7.702133443145138e-06,
      "loss": 0.7538,
      "step": 11756
    },
    {
      "epoch": 0.33875986861061486,
      "grad_norm": 0.7605034112930298,
      "learning_rate": 7.701740829769356e-06,
      "loss": 0.7323,
      "step": 11757
    },
    {
      "epoch": 0.33878868207226415,
      "grad_norm": 0.759441614151001,
      "learning_rate": 7.701348192864115e-06,
      "loss": 0.7358,
      "step": 11758
    },
    {
      "epoch": 0.33881749553391344,
      "grad_norm": 0.7587072253227234,
      "learning_rate": 7.70095553243283e-06,
      "loss": 0.7174,
      "step": 11759
    },
    {
      "epoch": 0.33884630899556273,
      "grad_norm": 0.8031119704246521,
      "learning_rate": 7.700562848478924e-06,
      "loss": 0.7533,
      "step": 11760
    },
    {
      "epoch": 0.338875122457212,
      "grad_norm": 0.7691320776939392,
      "learning_rate": 7.700170141005811e-06,
      "loss": 0.7458,
      "step": 11761
    },
    {
      "epoch": 0.3389039359188613,
      "grad_norm": 0.7883211970329285,
      "learning_rate": 7.699777410016919e-06,
      "loss": 0.7417,
      "step": 11762
    },
    {
      "epoch": 0.3389327493805106,
      "grad_norm": 0.7496775984764099,
      "learning_rate": 7.699384655515663e-06,
      "loss": 0.7443,
      "step": 11763
    },
    {
      "epoch": 0.33896156284215984,
      "grad_norm": 0.7846361398696899,
      "learning_rate": 7.698991877505466e-06,
      "loss": 0.7542,
      "step": 11764
    },
    {
      "epoch": 0.3389903763038091,
      "grad_norm": 0.7718308568000793,
      "learning_rate": 7.698599075989748e-06,
      "loss": 0.7355,
      "step": 11765
    },
    {
      "epoch": 0.3390191897654584,
      "grad_norm": 0.7828617095947266,
      "learning_rate": 7.698206250971929e-06,
      "loss": 0.7671,
      "step": 11766
    },
    {
      "epoch": 0.3390480032271077,
      "grad_norm": 0.7731099724769592,
      "learning_rate": 7.697813402455431e-06,
      "loss": 0.742,
      "step": 11767
    },
    {
      "epoch": 0.339076816688757,
      "grad_norm": 0.7663701176643372,
      "learning_rate": 7.697420530443673e-06,
      "loss": 0.7586,
      "step": 11768
    },
    {
      "epoch": 0.3391056301504063,
      "grad_norm": 0.7878989577293396,
      "learning_rate": 7.697027634940081e-06,
      "loss": 0.7531,
      "step": 11769
    },
    {
      "epoch": 0.3391344436120555,
      "grad_norm": 0.7768248319625854,
      "learning_rate": 7.696634715948073e-06,
      "loss": 0.7499,
      "step": 11770
    },
    {
      "epoch": 0.3391632570737048,
      "grad_norm": 0.7709141969680786,
      "learning_rate": 7.696241773471073e-06,
      "loss": 0.7615,
      "step": 11771
    },
    {
      "epoch": 0.3391920705353541,
      "grad_norm": 3.6765940189361572,
      "learning_rate": 7.695848807512502e-06,
      "loss": 0.7822,
      "step": 11772
    },
    {
      "epoch": 0.3392208839970034,
      "grad_norm": 0.7861607670783997,
      "learning_rate": 7.695455818075784e-06,
      "loss": 0.7349,
      "step": 11773
    },
    {
      "epoch": 0.3392496974586527,
      "grad_norm": 0.8216802477836609,
      "learning_rate": 7.695062805164339e-06,
      "loss": 0.7342,
      "step": 11774
    },
    {
      "epoch": 0.339278510920302,
      "grad_norm": 0.8011531233787537,
      "learning_rate": 7.694669768781592e-06,
      "loss": 0.7559,
      "step": 11775
    },
    {
      "epoch": 0.33930732438195127,
      "grad_norm": 0.8137917518615723,
      "learning_rate": 7.694276708930965e-06,
      "loss": 0.7682,
      "step": 11776
    },
    {
      "epoch": 0.3393361378436005,
      "grad_norm": 0.7805644273757935,
      "learning_rate": 7.693883625615881e-06,
      "loss": 0.7396,
      "step": 11777
    },
    {
      "epoch": 0.3393649513052498,
      "grad_norm": 0.7655026316642761,
      "learning_rate": 7.693490518839764e-06,
      "loss": 0.7338,
      "step": 11778
    },
    {
      "epoch": 0.3393937647668991,
      "grad_norm": 0.7553138732910156,
      "learning_rate": 7.693097388606035e-06,
      "loss": 0.7117,
      "step": 11779
    },
    {
      "epoch": 0.3394225782285484,
      "grad_norm": 0.7767239212989807,
      "learning_rate": 7.692704234918123e-06,
      "loss": 0.7621,
      "step": 11780
    },
    {
      "epoch": 0.33945139169019767,
      "grad_norm": 0.7945842742919922,
      "learning_rate": 7.692311057779445e-06,
      "loss": 0.7221,
      "step": 11781
    },
    {
      "epoch": 0.33948020515184696,
      "grad_norm": 0.8024033904075623,
      "learning_rate": 7.691917857193431e-06,
      "loss": 0.7508,
      "step": 11782
    },
    {
      "epoch": 0.33950901861349625,
      "grad_norm": 0.7791299223899841,
      "learning_rate": 7.691524633163504e-06,
      "loss": 0.73,
      "step": 11783
    },
    {
      "epoch": 0.3395378320751455,
      "grad_norm": 0.7831947207450867,
      "learning_rate": 7.691131385693086e-06,
      "loss": 0.7274,
      "step": 11784
    },
    {
      "epoch": 0.3395666455367948,
      "grad_norm": 0.8064879179000854,
      "learning_rate": 7.690738114785607e-06,
      "loss": 0.7473,
      "step": 11785
    },
    {
      "epoch": 0.33959545899844407,
      "grad_norm": 0.751179575920105,
      "learning_rate": 7.690344820444487e-06,
      "loss": 0.7349,
      "step": 11786
    },
    {
      "epoch": 0.33962427246009336,
      "grad_norm": 0.7933776378631592,
      "learning_rate": 7.689951502673151e-06,
      "loss": 0.7387,
      "step": 11787
    },
    {
      "epoch": 0.33965308592174265,
      "grad_norm": 0.8144455552101135,
      "learning_rate": 7.68955816147503e-06,
      "loss": 0.7552,
      "step": 11788
    },
    {
      "epoch": 0.33968189938339194,
      "grad_norm": 0.8078182935714722,
      "learning_rate": 7.689164796853544e-06,
      "loss": 0.7247,
      "step": 11789
    },
    {
      "epoch": 0.33971071284504123,
      "grad_norm": 3.7144806385040283,
      "learning_rate": 7.688771408812121e-06,
      "loss": 0.7586,
      "step": 11790
    },
    {
      "epoch": 0.33973952630669046,
      "grad_norm": 1.2526931762695312,
      "learning_rate": 7.688377997354186e-06,
      "loss": 0.7558,
      "step": 11791
    },
    {
      "epoch": 0.33976833976833976,
      "grad_norm": 0.8421772122383118,
      "learning_rate": 7.687984562483168e-06,
      "loss": 0.7639,
      "step": 11792
    },
    {
      "epoch": 0.33979715322998905,
      "grad_norm": 0.8405575752258301,
      "learning_rate": 7.687591104202489e-06,
      "loss": 0.7408,
      "step": 11793
    },
    {
      "epoch": 0.33982596669163834,
      "grad_norm": 0.7724866271018982,
      "learning_rate": 7.68719762251558e-06,
      "loss": 0.7445,
      "step": 11794
    },
    {
      "epoch": 0.33985478015328763,
      "grad_norm": 0.8101367354393005,
      "learning_rate": 7.686804117425866e-06,
      "loss": 0.7478,
      "step": 11795
    },
    {
      "epoch": 0.3398835936149369,
      "grad_norm": 0.8160409927368164,
      "learning_rate": 7.686410588936772e-06,
      "loss": 0.7496,
      "step": 11796
    },
    {
      "epoch": 0.33991240707658615,
      "grad_norm": 0.7691290974617004,
      "learning_rate": 7.686017037051729e-06,
      "loss": 0.7561,
      "step": 11797
    },
    {
      "epoch": 0.33994122053823544,
      "grad_norm": 0.8271958231925964,
      "learning_rate": 7.685623461774163e-06,
      "loss": 0.7552,
      "step": 11798
    },
    {
      "epoch": 0.33997003399988474,
      "grad_norm": 0.8020553588867188,
      "learning_rate": 7.6852298631075e-06,
      "loss": 0.7599,
      "step": 11799
    },
    {
      "epoch": 0.339998847461534,
      "grad_norm": 0.8031203746795654,
      "learning_rate": 7.68483624105517e-06,
      "loss": 0.7536,
      "step": 11800
    },
    {
      "epoch": 0.3400276609231833,
      "grad_norm": 0.7779264450073242,
      "learning_rate": 7.6844425956206e-06,
      "loss": 0.7358,
      "step": 11801
    },
    {
      "epoch": 0.3400564743848326,
      "grad_norm": 0.7895647287368774,
      "learning_rate": 7.684048926807218e-06,
      "loss": 0.7305,
      "step": 11802
    },
    {
      "epoch": 0.3400852878464819,
      "grad_norm": 0.799248218536377,
      "learning_rate": 7.683655234618454e-06,
      "loss": 0.7307,
      "step": 11803
    },
    {
      "epoch": 0.34011410130813113,
      "grad_norm": 0.8053476810455322,
      "learning_rate": 7.683261519057734e-06,
      "loss": 0.7305,
      "step": 11804
    },
    {
      "epoch": 0.3401429147697804,
      "grad_norm": 0.7807640433311462,
      "learning_rate": 7.682867780128488e-06,
      "loss": 0.7421,
      "step": 11805
    },
    {
      "epoch": 0.3401717282314297,
      "grad_norm": 0.7844020128250122,
      "learning_rate": 7.682474017834147e-06,
      "loss": 0.7519,
      "step": 11806
    },
    {
      "epoch": 0.340200541693079,
      "grad_norm": 0.7944567203521729,
      "learning_rate": 7.68208023217814e-06,
      "loss": 0.7445,
      "step": 11807
    },
    {
      "epoch": 0.3402293551547283,
      "grad_norm": 0.7911540269851685,
      "learning_rate": 7.681686423163892e-06,
      "loss": 0.7651,
      "step": 11808
    },
    {
      "epoch": 0.3402581686163776,
      "grad_norm": 0.7968201041221619,
      "learning_rate": 7.681292590794839e-06,
      "loss": 0.7226,
      "step": 11809
    },
    {
      "epoch": 0.3402869820780269,
      "grad_norm": 0.7932791709899902,
      "learning_rate": 7.680898735074407e-06,
      "loss": 0.7476,
      "step": 11810
    },
    {
      "epoch": 0.3403157955396761,
      "grad_norm": 0.7848911285400391,
      "learning_rate": 7.680504856006026e-06,
      "loss": 0.7383,
      "step": 11811
    },
    {
      "epoch": 0.3403446090013254,
      "grad_norm": 0.7698956727981567,
      "learning_rate": 7.680110953593128e-06,
      "loss": 0.745,
      "step": 11812
    },
    {
      "epoch": 0.3403734224629747,
      "grad_norm": 0.757439136505127,
      "learning_rate": 7.679717027839142e-06,
      "loss": 0.7305,
      "step": 11813
    },
    {
      "epoch": 0.340402235924624,
      "grad_norm": 0.7766952514648438,
      "learning_rate": 7.6793230787475e-06,
      "loss": 0.7613,
      "step": 11814
    },
    {
      "epoch": 0.3404310493862733,
      "grad_norm": 0.7971848249435425,
      "learning_rate": 7.678929106321633e-06,
      "loss": 0.766,
      "step": 11815
    },
    {
      "epoch": 0.34045986284792257,
      "grad_norm": 0.8055784702301025,
      "learning_rate": 7.678535110564971e-06,
      "loss": 0.7508,
      "step": 11816
    },
    {
      "epoch": 0.34048867630957186,
      "grad_norm": 0.7474145293235779,
      "learning_rate": 7.678141091480945e-06,
      "loss": 0.7172,
      "step": 11817
    },
    {
      "epoch": 0.3405174897712211,
      "grad_norm": 0.7803160548210144,
      "learning_rate": 7.677747049072987e-06,
      "loss": 0.7567,
      "step": 11818
    },
    {
      "epoch": 0.3405463032328704,
      "grad_norm": 0.7972705364227295,
      "learning_rate": 7.67735298334453e-06,
      "loss": 0.7619,
      "step": 11819
    },
    {
      "epoch": 0.3405751166945197,
      "grad_norm": 0.7822277545928955,
      "learning_rate": 7.676958894299005e-06,
      "loss": 0.7701,
      "step": 11820
    },
    {
      "epoch": 0.34060393015616897,
      "grad_norm": 0.7736098766326904,
      "learning_rate": 7.676564781939844e-06,
      "loss": 0.7412,
      "step": 11821
    },
    {
      "epoch": 0.34063274361781826,
      "grad_norm": 0.7737981677055359,
      "learning_rate": 7.67617064627048e-06,
      "loss": 0.7264,
      "step": 11822
    },
    {
      "epoch": 0.34066155707946755,
      "grad_norm": 0.8309412598609924,
      "learning_rate": 7.675776487294342e-06,
      "loss": 0.763,
      "step": 11823
    },
    {
      "epoch": 0.3406903705411168,
      "grad_norm": 0.7664970755577087,
      "learning_rate": 7.675382305014867e-06,
      "loss": 0.7402,
      "step": 11824
    },
    {
      "epoch": 0.3407191840027661,
      "grad_norm": 0.7570613622665405,
      "learning_rate": 7.674988099435487e-06,
      "loss": 0.7495,
      "step": 11825
    },
    {
      "epoch": 0.34074799746441536,
      "grad_norm": 0.8033779263496399,
      "learning_rate": 7.674593870559635e-06,
      "loss": 0.7218,
      "step": 11826
    },
    {
      "epoch": 0.34077681092606465,
      "grad_norm": 0.7883049249649048,
      "learning_rate": 7.674199618390742e-06,
      "loss": 0.7398,
      "step": 11827
    },
    {
      "epoch": 0.34080562438771395,
      "grad_norm": 0.7949645519256592,
      "learning_rate": 7.673805342932246e-06,
      "loss": 0.7512,
      "step": 11828
    },
    {
      "epoch": 0.34083443784936324,
      "grad_norm": 0.8124500513076782,
      "learning_rate": 7.673411044187576e-06,
      "loss": 0.7711,
      "step": 11829
    },
    {
      "epoch": 0.3408632513110125,
      "grad_norm": 0.8146833777427673,
      "learning_rate": 7.67301672216017e-06,
      "loss": 0.7614,
      "step": 11830
    },
    {
      "epoch": 0.34089206477266176,
      "grad_norm": 0.7696194052696228,
      "learning_rate": 7.672622376853461e-06,
      "loss": 0.7651,
      "step": 11831
    },
    {
      "epoch": 0.34092087823431105,
      "grad_norm": 0.785097062587738,
      "learning_rate": 7.672228008270881e-06,
      "loss": 0.7108,
      "step": 11832
    },
    {
      "epoch": 0.34094969169596034,
      "grad_norm": 0.7801012992858887,
      "learning_rate": 7.671833616415868e-06,
      "loss": 0.7317,
      "step": 11833
    },
    {
      "epoch": 0.34097850515760963,
      "grad_norm": 0.767520010471344,
      "learning_rate": 7.671439201291853e-06,
      "loss": 0.7407,
      "step": 11834
    },
    {
      "epoch": 0.3410073186192589,
      "grad_norm": 0.7956241965293884,
      "learning_rate": 7.671044762902274e-06,
      "loss": 0.7519,
      "step": 11835
    },
    {
      "epoch": 0.3410361320809082,
      "grad_norm": 0.7972955703735352,
      "learning_rate": 7.670650301250567e-06,
      "loss": 0.739,
      "step": 11836
    },
    {
      "epoch": 0.3410649455425575,
      "grad_norm": 0.7584205269813538,
      "learning_rate": 7.670255816340163e-06,
      "loss": 0.7494,
      "step": 11837
    },
    {
      "epoch": 0.34109375900420674,
      "grad_norm": 0.8540799617767334,
      "learning_rate": 7.6698613081745e-06,
      "loss": 0.7213,
      "step": 11838
    },
    {
      "epoch": 0.34112257246585603,
      "grad_norm": 0.7990274429321289,
      "learning_rate": 7.669466776757016e-06,
      "loss": 0.7579,
      "step": 11839
    },
    {
      "epoch": 0.3411513859275053,
      "grad_norm": 0.8135292530059814,
      "learning_rate": 7.669072222091144e-06,
      "loss": 0.758,
      "step": 11840
    },
    {
      "epoch": 0.3411801993891546,
      "grad_norm": 0.78024822473526,
      "learning_rate": 7.66867764418032e-06,
      "loss": 0.7488,
      "step": 11841
    },
    {
      "epoch": 0.3412090128508039,
      "grad_norm": 0.7641000747680664,
      "learning_rate": 7.668283043027982e-06,
      "loss": 0.7242,
      "step": 11842
    },
    {
      "epoch": 0.3412378263124532,
      "grad_norm": 0.7950347065925598,
      "learning_rate": 7.667888418637568e-06,
      "loss": 0.7534,
      "step": 11843
    },
    {
      "epoch": 0.3412666397741025,
      "grad_norm": 0.7481628656387329,
      "learning_rate": 7.667493771012511e-06,
      "loss": 0.7362,
      "step": 11844
    },
    {
      "epoch": 0.3412954532357517,
      "grad_norm": 0.751731812953949,
      "learning_rate": 7.66709910015625e-06,
      "loss": 0.7211,
      "step": 11845
    },
    {
      "epoch": 0.341324266697401,
      "grad_norm": 0.7906537055969238,
      "learning_rate": 7.666704406072221e-06,
      "loss": 0.7583,
      "step": 11846
    },
    {
      "epoch": 0.3413530801590503,
      "grad_norm": 0.7903926968574524,
      "learning_rate": 7.666309688763863e-06,
      "loss": 0.7504,
      "step": 11847
    },
    {
      "epoch": 0.3413818936206996,
      "grad_norm": 0.8138660788536072,
      "learning_rate": 7.665914948234611e-06,
      "loss": 0.7752,
      "step": 11848
    },
    {
      "epoch": 0.3414107070823489,
      "grad_norm": 0.7699692249298096,
      "learning_rate": 7.665520184487907e-06,
      "loss": 0.7445,
      "step": 11849
    },
    {
      "epoch": 0.3414395205439982,
      "grad_norm": 0.796590268611908,
      "learning_rate": 7.665125397527186e-06,
      "loss": 0.7452,
      "step": 11850
    },
    {
      "epoch": 0.3414683340056474,
      "grad_norm": 0.7849773168563843,
      "learning_rate": 7.664730587355888e-06,
      "loss": 0.7365,
      "step": 11851
    },
    {
      "epoch": 0.3414971474672967,
      "grad_norm": 0.748786211013794,
      "learning_rate": 7.66433575397745e-06,
      "loss": 0.7235,
      "step": 11852
    },
    {
      "epoch": 0.341525960928946,
      "grad_norm": 0.7846525311470032,
      "learning_rate": 7.66394089739531e-06,
      "loss": 0.7591,
      "step": 11853
    },
    {
      "epoch": 0.3415547743905953,
      "grad_norm": 0.7887009382247925,
      "learning_rate": 7.663546017612908e-06,
      "loss": 0.7438,
      "step": 11854
    },
    {
      "epoch": 0.3415835878522446,
      "grad_norm": 0.7910418510437012,
      "learning_rate": 7.663151114633684e-06,
      "loss": 0.7571,
      "step": 11855
    },
    {
      "epoch": 0.34161240131389387,
      "grad_norm": 0.7817254066467285,
      "learning_rate": 7.662756188461075e-06,
      "loss": 0.7453,
      "step": 11856
    },
    {
      "epoch": 0.34164121477554316,
      "grad_norm": 0.7620415091514587,
      "learning_rate": 7.662361239098522e-06,
      "loss": 0.749,
      "step": 11857
    },
    {
      "epoch": 0.3416700282371924,
      "grad_norm": 0.7769989967346191,
      "learning_rate": 7.661966266549463e-06,
      "loss": 0.748,
      "step": 11858
    },
    {
      "epoch": 0.3416988416988417,
      "grad_norm": 0.8109506368637085,
      "learning_rate": 7.66157127081734e-06,
      "loss": 0.7637,
      "step": 11859
    },
    {
      "epoch": 0.341727655160491,
      "grad_norm": 0.7491449117660522,
      "learning_rate": 7.66117625190559e-06,
      "loss": 0.7421,
      "step": 11860
    },
    {
      "epoch": 0.34175646862214026,
      "grad_norm": 0.7873537540435791,
      "learning_rate": 7.660781209817658e-06,
      "loss": 0.7396,
      "step": 11861
    },
    {
      "epoch": 0.34178528208378955,
      "grad_norm": 0.7701275944709778,
      "learning_rate": 7.66038614455698e-06,
      "loss": 0.7692,
      "step": 11862
    },
    {
      "epoch": 0.34181409554543885,
      "grad_norm": 0.7862445116043091,
      "learning_rate": 7.659991056126998e-06,
      "loss": 0.743,
      "step": 11863
    },
    {
      "epoch": 0.34184290900708814,
      "grad_norm": 0.7921869158744812,
      "learning_rate": 7.659595944531152e-06,
      "loss": 0.7845,
      "step": 11864
    },
    {
      "epoch": 0.34187172246873737,
      "grad_norm": 0.7745088934898376,
      "learning_rate": 7.659200809772884e-06,
      "loss": 0.7269,
      "step": 11865
    },
    {
      "epoch": 0.34190053593038666,
      "grad_norm": 0.7811728119850159,
      "learning_rate": 7.658805651855637e-06,
      "loss": 0.7514,
      "step": 11866
    },
    {
      "epoch": 0.34192934939203595,
      "grad_norm": 0.761021077632904,
      "learning_rate": 7.658410470782848e-06,
      "loss": 0.7279,
      "step": 11867
    },
    {
      "epoch": 0.34195816285368524,
      "grad_norm": 0.7522125840187073,
      "learning_rate": 7.658015266557963e-06,
      "loss": 0.7435,
      "step": 11868
    },
    {
      "epoch": 0.34198697631533453,
      "grad_norm": 0.7815041542053223,
      "learning_rate": 7.657620039184422e-06,
      "loss": 0.7425,
      "step": 11869
    },
    {
      "epoch": 0.3420157897769838,
      "grad_norm": 0.7554120421409607,
      "learning_rate": 7.657224788665668e-06,
      "loss": 0.7219,
      "step": 11870
    },
    {
      "epoch": 0.3420446032386331,
      "grad_norm": 0.8020978569984436,
      "learning_rate": 7.65682951500514e-06,
      "loss": 0.7597,
      "step": 11871
    },
    {
      "epoch": 0.34207341670028235,
      "grad_norm": 0.7875996828079224,
      "learning_rate": 7.656434218206283e-06,
      "loss": 0.7268,
      "step": 11872
    },
    {
      "epoch": 0.34210223016193164,
      "grad_norm": 0.810599148273468,
      "learning_rate": 7.656038898272542e-06,
      "loss": 0.7575,
      "step": 11873
    },
    {
      "epoch": 0.34213104362358093,
      "grad_norm": 0.8075114488601685,
      "learning_rate": 7.655643555207355e-06,
      "loss": 0.7566,
      "step": 11874
    },
    {
      "epoch": 0.3421598570852302,
      "grad_norm": 0.7730861902236938,
      "learning_rate": 7.655248189014168e-06,
      "loss": 0.7348,
      "step": 11875
    },
    {
      "epoch": 0.3421886705468795,
      "grad_norm": 0.748870313167572,
      "learning_rate": 7.654852799696423e-06,
      "loss": 0.7496,
      "step": 11876
    },
    {
      "epoch": 0.3422174840085288,
      "grad_norm": 0.8066474199295044,
      "learning_rate": 7.654457387257565e-06,
      "loss": 0.7139,
      "step": 11877
    },
    {
      "epoch": 0.34224629747017804,
      "grad_norm": 0.8475056886672974,
      "learning_rate": 7.654061951701036e-06,
      "loss": 0.7466,
      "step": 11878
    },
    {
      "epoch": 0.34227511093182733,
      "grad_norm": 0.8010962605476379,
      "learning_rate": 7.65366649303028e-06,
      "loss": 0.7343,
      "step": 11879
    },
    {
      "epoch": 0.3423039243934766,
      "grad_norm": 0.7734107375144958,
      "learning_rate": 7.653271011248744e-06,
      "loss": 0.7632,
      "step": 11880
    },
    {
      "epoch": 0.3423327378551259,
      "grad_norm": 0.8127967119216919,
      "learning_rate": 7.652875506359868e-06,
      "loss": 0.7486,
      "step": 11881
    },
    {
      "epoch": 0.3423615513167752,
      "grad_norm": 0.8017286062240601,
      "learning_rate": 7.652479978367097e-06,
      "loss": 0.7395,
      "step": 11882
    },
    {
      "epoch": 0.3423903647784245,
      "grad_norm": 0.7780112624168396,
      "learning_rate": 7.652084427273878e-06,
      "loss": 0.74,
      "step": 11883
    },
    {
      "epoch": 0.3424191782400738,
      "grad_norm": 0.7610076665878296,
      "learning_rate": 7.651688853083656e-06,
      "loss": 0.7295,
      "step": 11884
    },
    {
      "epoch": 0.342447991701723,
      "grad_norm": 0.787010908126831,
      "learning_rate": 7.651293255799874e-06,
      "loss": 0.7567,
      "step": 11885
    },
    {
      "epoch": 0.3424768051633723,
      "grad_norm": 0.7852414846420288,
      "learning_rate": 7.650897635425977e-06,
      "loss": 0.7449,
      "step": 11886
    },
    {
      "epoch": 0.3425056186250216,
      "grad_norm": 0.7949974536895752,
      "learning_rate": 7.650501991965413e-06,
      "loss": 0.7715,
      "step": 11887
    },
    {
      "epoch": 0.3425344320866709,
      "grad_norm": 1.6257838010787964,
      "learning_rate": 7.650106325421625e-06,
      "loss": 0.7559,
      "step": 11888
    },
    {
      "epoch": 0.3425632455483202,
      "grad_norm": 0.7450538873672485,
      "learning_rate": 7.64971063579806e-06,
      "loss": 0.7249,
      "step": 11889
    },
    {
      "epoch": 0.3425920590099695,
      "grad_norm": 0.9822070598602295,
      "learning_rate": 7.649314923098164e-06,
      "loss": 0.7453,
      "step": 11890
    },
    {
      "epoch": 0.34262087247161876,
      "grad_norm": 0.7845355868339539,
      "learning_rate": 7.648919187325384e-06,
      "loss": 0.7387,
      "step": 11891
    },
    {
      "epoch": 0.342649685933268,
      "grad_norm": 0.7672374844551086,
      "learning_rate": 7.648523428483167e-06,
      "loss": 0.7589,
      "step": 11892
    },
    {
      "epoch": 0.3426784993949173,
      "grad_norm": 0.78507000207901,
      "learning_rate": 7.648127646574956e-06,
      "loss": 0.7447,
      "step": 11893
    },
    {
      "epoch": 0.3427073128565666,
      "grad_norm": 0.776984691619873,
      "learning_rate": 7.647731841604202e-06,
      "loss": 0.7365,
      "step": 11894
    },
    {
      "epoch": 0.3427361263182159,
      "grad_norm": 0.7739754915237427,
      "learning_rate": 7.64733601357435e-06,
      "loss": 0.7187,
      "step": 11895
    },
    {
      "epoch": 0.34276493977986516,
      "grad_norm": 0.7752223014831543,
      "learning_rate": 7.646940162488847e-06,
      "loss": 0.7484,
      "step": 11896
    },
    {
      "epoch": 0.34279375324151445,
      "grad_norm": 0.7642474174499512,
      "learning_rate": 7.646544288351143e-06,
      "loss": 0.735,
      "step": 11897
    },
    {
      "epoch": 0.34282256670316374,
      "grad_norm": 0.7917859554290771,
      "learning_rate": 7.646148391164683e-06,
      "loss": 0.7299,
      "step": 11898
    },
    {
      "epoch": 0.342851380164813,
      "grad_norm": 0.7700368165969849,
      "learning_rate": 7.645752470932916e-06,
      "loss": 0.7333,
      "step": 11899
    },
    {
      "epoch": 0.34288019362646227,
      "grad_norm": 0.7597877383232117,
      "learning_rate": 7.645356527659289e-06,
      "loss": 0.7441,
      "step": 11900
    },
    {
      "epoch": 0.34290900708811156,
      "grad_norm": 0.7685312032699585,
      "learning_rate": 7.64496056134725e-06,
      "loss": 0.7426,
      "step": 11901
    },
    {
      "epoch": 0.34293782054976085,
      "grad_norm": 0.763954222202301,
      "learning_rate": 7.644564572000251e-06,
      "loss": 0.7402,
      "step": 11902
    },
    {
      "epoch": 0.34296663401141014,
      "grad_norm": 0.7429169416427612,
      "learning_rate": 7.644168559621737e-06,
      "loss": 0.7405,
      "step": 11903
    },
    {
      "epoch": 0.34299544747305943,
      "grad_norm": 0.7865998148918152,
      "learning_rate": 7.643772524215158e-06,
      "loss": 0.7365,
      "step": 11904
    },
    {
      "epoch": 0.34302426093470867,
      "grad_norm": 0.770543098449707,
      "learning_rate": 7.643376465783963e-06,
      "loss": 0.76,
      "step": 11905
    },
    {
      "epoch": 0.34305307439635796,
      "grad_norm": 0.7553691864013672,
      "learning_rate": 7.642980384331601e-06,
      "loss": 0.7548,
      "step": 11906
    },
    {
      "epoch": 0.34308188785800725,
      "grad_norm": 0.7555032968521118,
      "learning_rate": 7.642584279861522e-06,
      "loss": 0.7461,
      "step": 11907
    },
    {
      "epoch": 0.34311070131965654,
      "grad_norm": 0.7914770841598511,
      "learning_rate": 7.642188152377175e-06,
      "loss": 0.7472,
      "step": 11908
    },
    {
      "epoch": 0.34313951478130583,
      "grad_norm": 0.7865982055664062,
      "learning_rate": 7.64179200188201e-06,
      "loss": 0.7242,
      "step": 11909
    },
    {
      "epoch": 0.3431683282429551,
      "grad_norm": 0.7711324691772461,
      "learning_rate": 7.641395828379482e-06,
      "loss": 0.741,
      "step": 11910
    },
    {
      "epoch": 0.3431971417046044,
      "grad_norm": 0.7770606279373169,
      "learning_rate": 7.640999631873032e-06,
      "loss": 0.7452,
      "step": 11911
    },
    {
      "epoch": 0.34322595516625365,
      "grad_norm": 0.8202487230300903,
      "learning_rate": 7.640603412366116e-06,
      "loss": 0.7444,
      "step": 11912
    },
    {
      "epoch": 0.34325476862790294,
      "grad_norm": 0.7951777577400208,
      "learning_rate": 7.640207169862186e-06,
      "loss": 0.7444,
      "step": 11913
    },
    {
      "epoch": 0.34328358208955223,
      "grad_norm": 0.753242552280426,
      "learning_rate": 7.639810904364689e-06,
      "loss": 0.7396,
      "step": 11914
    },
    {
      "epoch": 0.3433123955512015,
      "grad_norm": 0.8028579354286194,
      "learning_rate": 7.639414615877078e-06,
      "loss": 0.7488,
      "step": 11915
    },
    {
      "epoch": 0.3433412090128508,
      "grad_norm": 0.8141797184944153,
      "learning_rate": 7.639018304402806e-06,
      "loss": 0.7179,
      "step": 11916
    },
    {
      "epoch": 0.3433700224745001,
      "grad_norm": 0.7718095183372498,
      "learning_rate": 7.638621969945321e-06,
      "loss": 0.7495,
      "step": 11917
    },
    {
      "epoch": 0.3433988359361494,
      "grad_norm": 0.74391770362854,
      "learning_rate": 7.638225612508075e-06,
      "loss": 0.721,
      "step": 11918
    },
    {
      "epoch": 0.34342764939779863,
      "grad_norm": 0.7773288488388062,
      "learning_rate": 7.637829232094523e-06,
      "loss": 0.7283,
      "step": 11919
    },
    {
      "epoch": 0.3434564628594479,
      "grad_norm": 0.811148464679718,
      "learning_rate": 7.637432828708116e-06,
      "loss": 0.7291,
      "step": 11920
    },
    {
      "epoch": 0.3434852763210972,
      "grad_norm": 0.7766571640968323,
      "learning_rate": 7.637036402352305e-06,
      "loss": 0.7395,
      "step": 11921
    },
    {
      "epoch": 0.3435140897827465,
      "grad_norm": 0.7739644646644592,
      "learning_rate": 7.636639953030541e-06,
      "loss": 0.7594,
      "step": 11922
    },
    {
      "epoch": 0.3435429032443958,
      "grad_norm": 0.8057547807693481,
      "learning_rate": 7.636243480746281e-06,
      "loss": 0.7515,
      "step": 11923
    },
    {
      "epoch": 0.3435717167060451,
      "grad_norm": 0.7556611895561218,
      "learning_rate": 7.635846985502974e-06,
      "loss": 0.7327,
      "step": 11924
    },
    {
      "epoch": 0.3436005301676944,
      "grad_norm": 0.7913532853126526,
      "learning_rate": 7.635450467304076e-06,
      "loss": 0.7301,
      "step": 11925
    },
    {
      "epoch": 0.3436293436293436,
      "grad_norm": 0.7563570141792297,
      "learning_rate": 7.635053926153038e-06,
      "loss": 0.7367,
      "step": 11926
    },
    {
      "epoch": 0.3436581570909929,
      "grad_norm": 0.7852398753166199,
      "learning_rate": 7.634657362053312e-06,
      "loss": 0.7254,
      "step": 11927
    },
    {
      "epoch": 0.3436869705526422,
      "grad_norm": 0.7860859632492065,
      "learning_rate": 7.634260775008358e-06,
      "loss": 0.745,
      "step": 11928
    },
    {
      "epoch": 0.3437157840142915,
      "grad_norm": 0.7704750299453735,
      "learning_rate": 7.633864165021625e-06,
      "loss": 0.7144,
      "step": 11929
    },
    {
      "epoch": 0.34374459747594077,
      "grad_norm": 0.7847412824630737,
      "learning_rate": 7.633467532096567e-06,
      "loss": 0.7423,
      "step": 11930
    },
    {
      "epoch": 0.34377341093759006,
      "grad_norm": 0.802193284034729,
      "learning_rate": 7.63307087623664e-06,
      "loss": 0.7106,
      "step": 11931
    },
    {
      "epoch": 0.3438022243992393,
      "grad_norm": 0.8164847493171692,
      "learning_rate": 7.632674197445297e-06,
      "loss": 0.7165,
      "step": 11932
    },
    {
      "epoch": 0.3438310378608886,
      "grad_norm": 0.8006474375724792,
      "learning_rate": 7.632277495725995e-06,
      "loss": 0.7302,
      "step": 11933
    },
    {
      "epoch": 0.3438598513225379,
      "grad_norm": 0.7724356055259705,
      "learning_rate": 7.631880771082187e-06,
      "loss": 0.7357,
      "step": 11934
    },
    {
      "epoch": 0.34388866478418717,
      "grad_norm": 0.7612765431404114,
      "learning_rate": 7.631484023517329e-06,
      "loss": 0.7355,
      "step": 11935
    },
    {
      "epoch": 0.34391747824583646,
      "grad_norm": 0.8327381610870361,
      "learning_rate": 7.631087253034872e-06,
      "loss": 0.7403,
      "step": 11936
    },
    {
      "epoch": 0.34394629170748575,
      "grad_norm": 0.7944421172142029,
      "learning_rate": 7.630690459638279e-06,
      "loss": 0.7634,
      "step": 11937
    },
    {
      "epoch": 0.34397510516913504,
      "grad_norm": 0.7669183611869812,
      "learning_rate": 7.630293643331e-06,
      "loss": 0.7372,
      "step": 11938
    },
    {
      "epoch": 0.3440039186307843,
      "grad_norm": 0.7981516122817993,
      "learning_rate": 7.629896804116495e-06,
      "loss": 0.7497,
      "step": 11939
    },
    {
      "epoch": 0.34403273209243357,
      "grad_norm": 0.7773517370223999,
      "learning_rate": 7.629499941998215e-06,
      "loss": 0.7445,
      "step": 11940
    },
    {
      "epoch": 0.34406154555408286,
      "grad_norm": 0.7868274450302124,
      "learning_rate": 7.629103056979621e-06,
      "loss": 0.7535,
      "step": 11941
    },
    {
      "epoch": 0.34409035901573215,
      "grad_norm": 0.7699682712554932,
      "learning_rate": 7.628706149064164e-06,
      "loss": 0.7616,
      "step": 11942
    },
    {
      "epoch": 0.34411917247738144,
      "grad_norm": 0.7902195453643799,
      "learning_rate": 7.628309218255307e-06,
      "loss": 0.7302,
      "step": 11943
    },
    {
      "epoch": 0.34414798593903073,
      "grad_norm": 0.757581889629364,
      "learning_rate": 7.627912264556505e-06,
      "loss": 0.7336,
      "step": 11944
    },
    {
      "epoch": 0.34417679940068,
      "grad_norm": 0.7851883769035339,
      "learning_rate": 7.6275152879712125e-06,
      "loss": 0.7522,
      "step": 11945
    },
    {
      "epoch": 0.34420561286232926,
      "grad_norm": 0.7667372226715088,
      "learning_rate": 7.627118288502888e-06,
      "loss": 0.7318,
      "step": 11946
    },
    {
      "epoch": 0.34423442632397855,
      "grad_norm": 0.7873470187187195,
      "learning_rate": 7.62672126615499e-06,
      "loss": 0.7613,
      "step": 11947
    },
    {
      "epoch": 0.34426323978562784,
      "grad_norm": 0.802844762802124,
      "learning_rate": 7.626324220930976e-06,
      "loss": 0.7686,
      "step": 11948
    },
    {
      "epoch": 0.34429205324727713,
      "grad_norm": 0.8119134306907654,
      "learning_rate": 7.625927152834302e-06,
      "loss": 0.7697,
      "step": 11949
    },
    {
      "epoch": 0.3443208667089264,
      "grad_norm": 0.7882622480392456,
      "learning_rate": 7.6255300618684305e-06,
      "loss": 0.7697,
      "step": 11950
    },
    {
      "epoch": 0.3443496801705757,
      "grad_norm": 0.7651969790458679,
      "learning_rate": 7.625132948036815e-06,
      "loss": 0.747,
      "step": 11951
    },
    {
      "epoch": 0.344378493632225,
      "grad_norm": 0.7930699586868286,
      "learning_rate": 7.624735811342916e-06,
      "loss": 0.7329,
      "step": 11952
    },
    {
      "epoch": 0.34440730709387424,
      "grad_norm": 0.7680681347846985,
      "learning_rate": 7.624338651790191e-06,
      "loss": 0.7589,
      "step": 11953
    },
    {
      "epoch": 0.34443612055552353,
      "grad_norm": 0.7860912084579468,
      "learning_rate": 7.623941469382099e-06,
      "loss": 0.7404,
      "step": 11954
    },
    {
      "epoch": 0.3444649340171728,
      "grad_norm": 0.8106557130813599,
      "learning_rate": 7.623544264122101e-06,
      "loss": 0.7599,
      "step": 11955
    },
    {
      "epoch": 0.3444937474788221,
      "grad_norm": 0.7640606164932251,
      "learning_rate": 7.623147036013655e-06,
      "loss": 0.739,
      "step": 11956
    },
    {
      "epoch": 0.3445225609404714,
      "grad_norm": 0.7967695593833923,
      "learning_rate": 7.622749785060222e-06,
      "loss": 0.7376,
      "step": 11957
    },
    {
      "epoch": 0.3445513744021207,
      "grad_norm": 0.7840479612350464,
      "learning_rate": 7.622352511265257e-06,
      "loss": 0.7622,
      "step": 11958
    },
    {
      "epoch": 0.3445801878637699,
      "grad_norm": 0.7676406502723694,
      "learning_rate": 7.621955214632226e-06,
      "loss": 0.7423,
      "step": 11959
    },
    {
      "epoch": 0.3446090013254192,
      "grad_norm": 0.7999454140663147,
      "learning_rate": 7.621557895164584e-06,
      "loss": 0.7468,
      "step": 11960
    },
    {
      "epoch": 0.3446378147870685,
      "grad_norm": 0.7895113229751587,
      "learning_rate": 7.621160552865794e-06,
      "loss": 0.7639,
      "step": 11961
    },
    {
      "epoch": 0.3446666282487178,
      "grad_norm": 0.7635118365287781,
      "learning_rate": 7.620763187739315e-06,
      "loss": 0.7512,
      "step": 11962
    },
    {
      "epoch": 0.3446954417103671,
      "grad_norm": 0.7551798820495605,
      "learning_rate": 7.62036579978861e-06,
      "loss": 0.7237,
      "step": 11963
    },
    {
      "epoch": 0.3447242551720164,
      "grad_norm": 0.7571203708648682,
      "learning_rate": 7.619968389017138e-06,
      "loss": 0.7405,
      "step": 11964
    },
    {
      "epoch": 0.34475306863366567,
      "grad_norm": 0.8064730763435364,
      "learning_rate": 7.6195709554283594e-06,
      "loss": 0.7381,
      "step": 11965
    },
    {
      "epoch": 0.3447818820953149,
      "grad_norm": 0.7846805453300476,
      "learning_rate": 7.619173499025736e-06,
      "loss": 0.7469,
      "step": 11966
    },
    {
      "epoch": 0.3448106955569642,
      "grad_norm": 0.7492022514343262,
      "learning_rate": 7.61877601981273e-06,
      "loss": 0.7106,
      "step": 11967
    },
    {
      "epoch": 0.3448395090186135,
      "grad_norm": 0.7596668004989624,
      "learning_rate": 7.618378517792804e-06,
      "loss": 0.7373,
      "step": 11968
    },
    {
      "epoch": 0.3448683224802628,
      "grad_norm": 0.8154919147491455,
      "learning_rate": 7.6179809929694184e-06,
      "loss": 0.746,
      "step": 11969
    },
    {
      "epoch": 0.34489713594191207,
      "grad_norm": 0.8298592567443848,
      "learning_rate": 7.617583445346034e-06,
      "loss": 0.7847,
      "step": 11970
    },
    {
      "epoch": 0.34492594940356136,
      "grad_norm": 0.7708490490913391,
      "learning_rate": 7.617185874926115e-06,
      "loss": 0.7495,
      "step": 11971
    },
    {
      "epoch": 0.34495476286521065,
      "grad_norm": 0.7315658926963806,
      "learning_rate": 7.616788281713124e-06,
      "loss": 0.7273,
      "step": 11972
    },
    {
      "epoch": 0.3449835763268599,
      "grad_norm": 0.7992579936981201,
      "learning_rate": 7.616390665710522e-06,
      "loss": 0.7111,
      "step": 11973
    },
    {
      "epoch": 0.3450123897885092,
      "grad_norm": 0.7864288091659546,
      "learning_rate": 7.615993026921773e-06,
      "loss": 0.7531,
      "step": 11974
    },
    {
      "epoch": 0.34504120325015847,
      "grad_norm": 0.7591516375541687,
      "learning_rate": 7.615595365350339e-06,
      "loss": 0.7125,
      "step": 11975
    },
    {
      "epoch": 0.34507001671180776,
      "grad_norm": 0.7782360911369324,
      "learning_rate": 7.615197680999686e-06,
      "loss": 0.7448,
      "step": 11976
    },
    {
      "epoch": 0.34509883017345705,
      "grad_norm": 0.799591600894928,
      "learning_rate": 7.614799973873273e-06,
      "loss": 0.7592,
      "step": 11977
    },
    {
      "epoch": 0.34512764363510634,
      "grad_norm": 0.7859863042831421,
      "learning_rate": 7.614402243974568e-06,
      "loss": 0.7583,
      "step": 11978
    },
    {
      "epoch": 0.34515645709675563,
      "grad_norm": 0.7699653506278992,
      "learning_rate": 7.614004491307032e-06,
      "loss": 0.7459,
      "step": 11979
    },
    {
      "epoch": 0.34518527055840487,
      "grad_norm": 0.7805811762809753,
      "learning_rate": 7.61360671587413e-06,
      "loss": 0.761,
      "step": 11980
    },
    {
      "epoch": 0.34521408402005416,
      "grad_norm": 0.790963888168335,
      "learning_rate": 7.613208917679327e-06,
      "loss": 0.7657,
      "step": 11981
    },
    {
      "epoch": 0.34524289748170345,
      "grad_norm": 0.7967780828475952,
      "learning_rate": 7.612811096726085e-06,
      "loss": 0.7502,
      "step": 11982
    },
    {
      "epoch": 0.34527171094335274,
      "grad_norm": 0.7967286109924316,
      "learning_rate": 7.6124132530178706e-06,
      "loss": 0.7534,
      "step": 11983
    },
    {
      "epoch": 0.34530052440500203,
      "grad_norm": 0.7895054221153259,
      "learning_rate": 7.612015386558149e-06,
      "loss": 0.7438,
      "step": 11984
    },
    {
      "epoch": 0.3453293378666513,
      "grad_norm": 0.7874817252159119,
      "learning_rate": 7.611617497350384e-06,
      "loss": 0.7458,
      "step": 11985
    },
    {
      "epoch": 0.34535815132830056,
      "grad_norm": 0.7938742637634277,
      "learning_rate": 7.6112195853980405e-06,
      "loss": 0.745,
      "step": 11986
    },
    {
      "epoch": 0.34538696478994985,
      "grad_norm": 0.7711338400840759,
      "learning_rate": 7.610821650704586e-06,
      "loss": 0.7422,
      "step": 11987
    },
    {
      "epoch": 0.34541577825159914,
      "grad_norm": 0.782914400100708,
      "learning_rate": 7.610423693273485e-06,
      "loss": 0.7235,
      "step": 11988
    },
    {
      "epoch": 0.34544459171324843,
      "grad_norm": 0.7949322462081909,
      "learning_rate": 7.610025713108202e-06,
      "loss": 0.7325,
      "step": 11989
    },
    {
      "epoch": 0.3454734051748977,
      "grad_norm": 0.7753481268882751,
      "learning_rate": 7.609627710212204e-06,
      "loss": 0.7094,
      "step": 11990
    },
    {
      "epoch": 0.345502218636547,
      "grad_norm": 0.795489490032196,
      "learning_rate": 7.609229684588958e-06,
      "loss": 0.7484,
      "step": 11991
    },
    {
      "epoch": 0.3455310320981963,
      "grad_norm": 0.8111645579338074,
      "learning_rate": 7.60883163624193e-06,
      "loss": 0.7388,
      "step": 11992
    },
    {
      "epoch": 0.34555984555984554,
      "grad_norm": 0.7707281112670898,
      "learning_rate": 7.608433565174587e-06,
      "loss": 0.7458,
      "step": 11993
    },
    {
      "epoch": 0.3455886590214948,
      "grad_norm": 0.8284056186676025,
      "learning_rate": 7.6080354713903935e-06,
      "loss": 0.7615,
      "step": 11994
    },
    {
      "epoch": 0.3456174724831441,
      "grad_norm": 0.7856478095054626,
      "learning_rate": 7.607637354892818e-06,
      "loss": 0.7732,
      "step": 11995
    },
    {
      "epoch": 0.3456462859447934,
      "grad_norm": 0.7785935997962952,
      "learning_rate": 7.607239215685328e-06,
      "loss": 0.7457,
      "step": 11996
    },
    {
      "epoch": 0.3456750994064427,
      "grad_norm": 0.7649481892585754,
      "learning_rate": 7.606841053771392e-06,
      "loss": 0.7506,
      "step": 11997
    },
    {
      "epoch": 0.345703912868092,
      "grad_norm": 0.7610709071159363,
      "learning_rate": 7.606442869154475e-06,
      "loss": 0.7192,
      "step": 11998
    },
    {
      "epoch": 0.3457327263297413,
      "grad_norm": 0.7892111539840698,
      "learning_rate": 7.606044661838048e-06,
      "loss": 0.7342,
      "step": 11999
    },
    {
      "epoch": 0.3457615397913905,
      "grad_norm": 0.8268393278121948,
      "learning_rate": 7.605646431825575e-06,
      "loss": 0.7391,
      "step": 12000
    },
    {
      "epoch": 0.3457903532530398,
      "grad_norm": 0.7486904859542847,
      "learning_rate": 7.605248179120528e-06,
      "loss": 0.7323,
      "step": 12001
    },
    {
      "epoch": 0.3458191667146891,
      "grad_norm": 0.7326814532279968,
      "learning_rate": 7.604849903726371e-06,
      "loss": 0.7305,
      "step": 12002
    },
    {
      "epoch": 0.3458479801763384,
      "grad_norm": 0.7887474894523621,
      "learning_rate": 7.604451605646577e-06,
      "loss": 0.7375,
      "step": 12003
    },
    {
      "epoch": 0.3458767936379877,
      "grad_norm": 0.8022899031639099,
      "learning_rate": 7.604053284884614e-06,
      "loss": 0.7381,
      "step": 12004
    },
    {
      "epoch": 0.34590560709963697,
      "grad_norm": 0.7594028115272522,
      "learning_rate": 7.603654941443947e-06,
      "loss": 0.744,
      "step": 12005
    },
    {
      "epoch": 0.34593442056128626,
      "grad_norm": 0.7447358965873718,
      "learning_rate": 7.603256575328051e-06,
      "loss": 0.711,
      "step": 12006
    },
    {
      "epoch": 0.3459632340229355,
      "grad_norm": 0.7576807141304016,
      "learning_rate": 7.602858186540392e-06,
      "loss": 0.7356,
      "step": 12007
    },
    {
      "epoch": 0.3459920474845848,
      "grad_norm": 0.7948666214942932,
      "learning_rate": 7.602459775084439e-06,
      "loss": 0.7536,
      "step": 12008
    },
    {
      "epoch": 0.3460208609462341,
      "grad_norm": 0.776321291923523,
      "learning_rate": 7.602061340963662e-06,
      "loss": 0.7546,
      "step": 12009
    },
    {
      "epoch": 0.34604967440788337,
      "grad_norm": 0.7712318301200867,
      "learning_rate": 7.6016628841815344e-06,
      "loss": 0.7577,
      "step": 12010
    },
    {
      "epoch": 0.34607848786953266,
      "grad_norm": 0.8179230093955994,
      "learning_rate": 7.601264404741523e-06,
      "loss": 0.7409,
      "step": 12011
    },
    {
      "epoch": 0.34610730133118195,
      "grad_norm": 0.786106526851654,
      "learning_rate": 7.600865902647099e-06,
      "loss": 0.7391,
      "step": 12012
    },
    {
      "epoch": 0.3461361147928312,
      "grad_norm": 0.7833578586578369,
      "learning_rate": 7.600467377901733e-06,
      "loss": 0.7557,
      "step": 12013
    },
    {
      "epoch": 0.3461649282544805,
      "grad_norm": 0.7781774997711182,
      "learning_rate": 7.600068830508893e-06,
      "loss": 0.7682,
      "step": 12014
    },
    {
      "epoch": 0.34619374171612977,
      "grad_norm": 0.7956727743148804,
      "learning_rate": 7.599670260472056e-06,
      "loss": 0.7552,
      "step": 12015
    },
    {
      "epoch": 0.34622255517777906,
      "grad_norm": 0.7670694589614868,
      "learning_rate": 7.599271667794689e-06,
      "loss": 0.7145,
      "step": 12016
    },
    {
      "epoch": 0.34625136863942835,
      "grad_norm": 0.7741959691047668,
      "learning_rate": 7.598873052480265e-06,
      "loss": 0.7366,
      "step": 12017
    },
    {
      "epoch": 0.34628018210107764,
      "grad_norm": 0.7528734803199768,
      "learning_rate": 7.598474414532252e-06,
      "loss": 0.7473,
      "step": 12018
    },
    {
      "epoch": 0.34630899556272693,
      "grad_norm": 0.7956487536430359,
      "learning_rate": 7.5980757539541265e-06,
      "loss": 0.7721,
      "step": 12019
    },
    {
      "epoch": 0.34633780902437616,
      "grad_norm": 0.7683137059211731,
      "learning_rate": 7.597677070749357e-06,
      "loss": 0.7386,
      "step": 12020
    },
    {
      "epoch": 0.34636662248602545,
      "grad_norm": 0.777726411819458,
      "learning_rate": 7.597278364921417e-06,
      "loss": 0.749,
      "step": 12021
    },
    {
      "epoch": 0.34639543594767475,
      "grad_norm": 0.7684032320976257,
      "learning_rate": 7.59687963647378e-06,
      "loss": 0.7253,
      "step": 12022
    },
    {
      "epoch": 0.34642424940932404,
      "grad_norm": 0.7663682103157043,
      "learning_rate": 7.596480885409916e-06,
      "loss": 0.7367,
      "step": 12023
    },
    {
      "epoch": 0.3464530628709733,
      "grad_norm": 0.804231584072113,
      "learning_rate": 7.5960821117332985e-06,
      "loss": 0.7294,
      "step": 12024
    },
    {
      "epoch": 0.3464818763326226,
      "grad_norm": 0.7611682415008545,
      "learning_rate": 7.595683315447402e-06,
      "loss": 0.7765,
      "step": 12025
    },
    {
      "epoch": 0.3465106897942719,
      "grad_norm": 0.7589316964149475,
      "learning_rate": 7.595284496555698e-06,
      "loss": 0.7386,
      "step": 12026
    },
    {
      "epoch": 0.34653950325592114,
      "grad_norm": 0.788267195224762,
      "learning_rate": 7.5948856550616615e-06,
      "loss": 0.7402,
      "step": 12027
    },
    {
      "epoch": 0.34656831671757043,
      "grad_norm": 0.8055986166000366,
      "learning_rate": 7.594486790968765e-06,
      "loss": 0.7786,
      "step": 12028
    },
    {
      "epoch": 0.3465971301792197,
      "grad_norm": 0.7687458395957947,
      "learning_rate": 7.594087904280481e-06,
      "loss": 0.7399,
      "step": 12029
    },
    {
      "epoch": 0.346625943640869,
      "grad_norm": 0.762498676776886,
      "learning_rate": 7.593688995000285e-06,
      "loss": 0.7344,
      "step": 12030
    },
    {
      "epoch": 0.3466547571025183,
      "grad_norm": 0.7401641011238098,
      "learning_rate": 7.593290063131651e-06,
      "loss": 0.7401,
      "step": 12031
    },
    {
      "epoch": 0.3466835705641676,
      "grad_norm": 0.8091983199119568,
      "learning_rate": 7.592891108678051e-06,
      "loss": 0.7493,
      "step": 12032
    },
    {
      "epoch": 0.3467123840258169,
      "grad_norm": 0.7782248258590698,
      "learning_rate": 7.592492131642963e-06,
      "loss": 0.7416,
      "step": 12033
    },
    {
      "epoch": 0.3467411974874661,
      "grad_norm": 0.7810917496681213,
      "learning_rate": 7.592093132029861e-06,
      "loss": 0.7253,
      "step": 12034
    },
    {
      "epoch": 0.3467700109491154,
      "grad_norm": 0.7775231003761292,
      "learning_rate": 7.591694109842217e-06,
      "loss": 0.7436,
      "step": 12035
    },
    {
      "epoch": 0.3467988244107647,
      "grad_norm": 0.8095473647117615,
      "learning_rate": 7.59129506508351e-06,
      "loss": 0.7199,
      "step": 12036
    },
    {
      "epoch": 0.346827637872414,
      "grad_norm": 0.7525136470794678,
      "learning_rate": 7.590895997757211e-06,
      "loss": 0.7699,
      "step": 12037
    },
    {
      "epoch": 0.3468564513340633,
      "grad_norm": 0.7874327898025513,
      "learning_rate": 7.590496907866799e-06,
      "loss": 0.7485,
      "step": 12038
    },
    {
      "epoch": 0.3468852647957126,
      "grad_norm": 0.7810758948326111,
      "learning_rate": 7.5900977954157485e-06,
      "loss": 0.7438,
      "step": 12039
    },
    {
      "epoch": 0.3469140782573618,
      "grad_norm": 0.7631186246871948,
      "learning_rate": 7.589698660407535e-06,
      "loss": 0.7232,
      "step": 12040
    },
    {
      "epoch": 0.3469428917190111,
      "grad_norm": 0.7683423757553101,
      "learning_rate": 7.589299502845635e-06,
      "loss": 0.7703,
      "step": 12041
    },
    {
      "epoch": 0.3469717051806604,
      "grad_norm": 0.8095253109931946,
      "learning_rate": 7.588900322733526e-06,
      "loss": 0.7686,
      "step": 12042
    },
    {
      "epoch": 0.3470005186423097,
      "grad_norm": 0.800632655620575,
      "learning_rate": 7.5885011200746805e-06,
      "loss": 0.7319,
      "step": 12043
    },
    {
      "epoch": 0.347029332103959,
      "grad_norm": 0.7509306073188782,
      "learning_rate": 7.588101894872579e-06,
      "loss": 0.7299,
      "step": 12044
    },
    {
      "epoch": 0.34705814556560827,
      "grad_norm": 0.7828119397163391,
      "learning_rate": 7.587702647130698e-06,
      "loss": 0.7196,
      "step": 12045
    },
    {
      "epoch": 0.34708695902725756,
      "grad_norm": 0.7344641089439392,
      "learning_rate": 7.587303376852513e-06,
      "loss": 0.7217,
      "step": 12046
    },
    {
      "epoch": 0.3471157724889068,
      "grad_norm": 0.7554152607917786,
      "learning_rate": 7.586904084041502e-06,
      "loss": 0.7089,
      "step": 12047
    },
    {
      "epoch": 0.3471445859505561,
      "grad_norm": 0.7560082674026489,
      "learning_rate": 7.586504768701144e-06,
      "loss": 0.7403,
      "step": 12048
    },
    {
      "epoch": 0.3471733994122054,
      "grad_norm": 0.7825270891189575,
      "learning_rate": 7.586105430834913e-06,
      "loss": 0.7546,
      "step": 12049
    },
    {
      "epoch": 0.34720221287385467,
      "grad_norm": 0.7583812475204468,
      "learning_rate": 7.585706070446288e-06,
      "loss": 0.7654,
      "step": 12050
    },
    {
      "epoch": 0.34723102633550396,
      "grad_norm": 1.2386016845703125,
      "learning_rate": 7.58530668753875e-06,
      "loss": 0.7049,
      "step": 12051
    },
    {
      "epoch": 0.34725983979715325,
      "grad_norm": 0.7558614015579224,
      "learning_rate": 7.584907282115774e-06,
      "loss": 0.7338,
      "step": 12052
    },
    {
      "epoch": 0.34728865325880254,
      "grad_norm": 0.7726944088935852,
      "learning_rate": 7.5845078541808406e-06,
      "loss": 0.7473,
      "step": 12053
    },
    {
      "epoch": 0.3473174667204518,
      "grad_norm": 0.7715103626251221,
      "learning_rate": 7.584108403737427e-06,
      "loss": 0.7524,
      "step": 12054
    },
    {
      "epoch": 0.34734628018210106,
      "grad_norm": 0.7622461318969727,
      "learning_rate": 7.583708930789012e-06,
      "loss": 0.7203,
      "step": 12055
    },
    {
      "epoch": 0.34737509364375035,
      "grad_norm": 0.8413820862770081,
      "learning_rate": 7.583309435339074e-06,
      "loss": 0.715,
      "step": 12056
    },
    {
      "epoch": 0.34740390710539965,
      "grad_norm": 0.755655825138092,
      "learning_rate": 7.582909917391095e-06,
      "loss": 0.728,
      "step": 12057
    },
    {
      "epoch": 0.34743272056704894,
      "grad_norm": 0.742603600025177,
      "learning_rate": 7.582510376948552e-06,
      "loss": 0.7196,
      "step": 12058
    },
    {
      "epoch": 0.3474615340286982,
      "grad_norm": 0.7607391476631165,
      "learning_rate": 7.582110814014925e-06,
      "loss": 0.7451,
      "step": 12059
    },
    {
      "epoch": 0.3474903474903475,
      "grad_norm": 0.767845094203949,
      "learning_rate": 7.581711228593693e-06,
      "loss": 0.719,
      "step": 12060
    },
    {
      "epoch": 0.34751916095199675,
      "grad_norm": 0.760209321975708,
      "learning_rate": 7.581311620688339e-06,
      "loss": 0.724,
      "step": 12061
    },
    {
      "epoch": 0.34754797441364604,
      "grad_norm": 0.8115943670272827,
      "learning_rate": 7.58091199030234e-06,
      "loss": 0.7438,
      "step": 12062
    },
    {
      "epoch": 0.34757678787529533,
      "grad_norm": 0.7798794507980347,
      "learning_rate": 7.580512337439178e-06,
      "loss": 0.7316,
      "step": 12063
    },
    {
      "epoch": 0.3476056013369446,
      "grad_norm": 0.7956894636154175,
      "learning_rate": 7.580112662102331e-06,
      "loss": 0.745,
      "step": 12064
    },
    {
      "epoch": 0.3476344147985939,
      "grad_norm": 0.7772747874259949,
      "learning_rate": 7.579712964295284e-06,
      "loss": 0.762,
      "step": 12065
    },
    {
      "epoch": 0.3476632282602432,
      "grad_norm": 0.802843451499939,
      "learning_rate": 7.579313244021515e-06,
      "loss": 0.7653,
      "step": 12066
    },
    {
      "epoch": 0.34769204172189244,
      "grad_norm": 0.7837246060371399,
      "learning_rate": 7.5789135012845075e-06,
      "loss": 0.745,
      "step": 12067
    },
    {
      "epoch": 0.34772085518354173,
      "grad_norm": 0.7489647269248962,
      "learning_rate": 7.578513736087739e-06,
      "loss": 0.7417,
      "step": 12068
    },
    {
      "epoch": 0.347749668645191,
      "grad_norm": 0.809542179107666,
      "learning_rate": 7.578113948434695e-06,
      "loss": 0.749,
      "step": 12069
    },
    {
      "epoch": 0.3477784821068403,
      "grad_norm": 0.7752792239189148,
      "learning_rate": 7.577714138328855e-06,
      "loss": 0.7416,
      "step": 12070
    },
    {
      "epoch": 0.3478072955684896,
      "grad_norm": 0.7667393088340759,
      "learning_rate": 7.577314305773703e-06,
      "loss": 0.741,
      "step": 12071
    },
    {
      "epoch": 0.3478361090301389,
      "grad_norm": 0.7749785780906677,
      "learning_rate": 7.5769144507727185e-06,
      "loss": 0.7313,
      "step": 12072
    },
    {
      "epoch": 0.3478649224917882,
      "grad_norm": 0.7889901995658875,
      "learning_rate": 7.576514573329386e-06,
      "loss": 0.7601,
      "step": 12073
    },
    {
      "epoch": 0.3478937359534374,
      "grad_norm": 0.7678651809692383,
      "learning_rate": 7.5761146734471855e-06,
      "loss": 0.7204,
      "step": 12074
    },
    {
      "epoch": 0.3479225494150867,
      "grad_norm": 0.7545537352561951,
      "learning_rate": 7.575714751129604e-06,
      "loss": 0.7367,
      "step": 12075
    },
    {
      "epoch": 0.347951362876736,
      "grad_norm": 0.7901446223258972,
      "learning_rate": 7.5753148063801194e-06,
      "loss": 0.7792,
      "step": 12076
    },
    {
      "epoch": 0.3479801763383853,
      "grad_norm": 0.7704073786735535,
      "learning_rate": 7.574914839202218e-06,
      "loss": 0.7402,
      "step": 12077
    },
    {
      "epoch": 0.3480089898000346,
      "grad_norm": 0.7928324341773987,
      "learning_rate": 7.5745148495993834e-06,
      "loss": 0.7532,
      "step": 12078
    },
    {
      "epoch": 0.3480378032616839,
      "grad_norm": 0.8114069700241089,
      "learning_rate": 7.574114837575098e-06,
      "loss": 0.7565,
      "step": 12079
    },
    {
      "epoch": 0.34806661672333317,
      "grad_norm": 0.7376604080200195,
      "learning_rate": 7.573714803132843e-06,
      "loss": 0.7469,
      "step": 12080
    },
    {
      "epoch": 0.3480954301849824,
      "grad_norm": 0.7937977910041809,
      "learning_rate": 7.5733147462761055e-06,
      "loss": 0.741,
      "step": 12081
    },
    {
      "epoch": 0.3481242436466317,
      "grad_norm": 0.7500701546669006,
      "learning_rate": 7.572914667008371e-06,
      "loss": 0.7746,
      "step": 12082
    },
    {
      "epoch": 0.348153057108281,
      "grad_norm": 0.7836493849754333,
      "learning_rate": 7.57251456533312e-06,
      "loss": 0.7305,
      "step": 12083
    },
    {
      "epoch": 0.3481818705699303,
      "grad_norm": 0.7594601511955261,
      "learning_rate": 7.57211444125384e-06,
      "loss": 0.7328,
      "step": 12084
    },
    {
      "epoch": 0.34821068403157956,
      "grad_norm": 0.7733618021011353,
      "learning_rate": 7.571714294774013e-06,
      "loss": 0.7391,
      "step": 12085
    },
    {
      "epoch": 0.34823949749322886,
      "grad_norm": 0.7741925120353699,
      "learning_rate": 7.571314125897125e-06,
      "loss": 0.7395,
      "step": 12086
    },
    {
      "epoch": 0.34826831095487815,
      "grad_norm": 0.7716498374938965,
      "learning_rate": 7.570913934626663e-06,
      "loss": 0.7286,
      "step": 12087
    },
    {
      "epoch": 0.3482971244165274,
      "grad_norm": 3.9850661754608154,
      "learning_rate": 7.570513720966108e-06,
      "loss": 0.7418,
      "step": 12088
    },
    {
      "epoch": 0.3483259378781767,
      "grad_norm": 0.7986596822738647,
      "learning_rate": 7.57011348491895e-06,
      "loss": 0.7224,
      "step": 12089
    },
    {
      "epoch": 0.34835475133982596,
      "grad_norm": 0.7703083753585815,
      "learning_rate": 7.5697132264886745e-06,
      "loss": 0.7266,
      "step": 12090
    },
    {
      "epoch": 0.34838356480147525,
      "grad_norm": 0.8051521182060242,
      "learning_rate": 7.569312945678762e-06,
      "loss": 0.7577,
      "step": 12091
    },
    {
      "epoch": 0.34841237826312454,
      "grad_norm": 0.7647287845611572,
      "learning_rate": 7.568912642492705e-06,
      "loss": 0.7467,
      "step": 12092
    },
    {
      "epoch": 0.34844119172477384,
      "grad_norm": 0.8131792545318604,
      "learning_rate": 7.568512316933985e-06,
      "loss": 0.7619,
      "step": 12093
    },
    {
      "epoch": 0.34847000518642307,
      "grad_norm": 0.770778238773346,
      "learning_rate": 7.568111969006091e-06,
      "loss": 0.7348,
      "step": 12094
    },
    {
      "epoch": 0.34849881864807236,
      "grad_norm": 0.7689905166625977,
      "learning_rate": 7.567711598712509e-06,
      "loss": 0.7168,
      "step": 12095
    },
    {
      "epoch": 0.34852763210972165,
      "grad_norm": 0.7895147800445557,
      "learning_rate": 7.567311206056726e-06,
      "loss": 0.7714,
      "step": 12096
    },
    {
      "epoch": 0.34855644557137094,
      "grad_norm": 0.8150911331176758,
      "learning_rate": 7.566910791042228e-06,
      "loss": 0.7224,
      "step": 12097
    },
    {
      "epoch": 0.34858525903302023,
      "grad_norm": 0.8290809392929077,
      "learning_rate": 7.5665103536725036e-06,
      "loss": 0.7586,
      "step": 12098
    },
    {
      "epoch": 0.3486140724946695,
      "grad_norm": 0.7911881804466248,
      "learning_rate": 7.566109893951039e-06,
      "loss": 0.7459,
      "step": 12099
    },
    {
      "epoch": 0.3486428859563188,
      "grad_norm": 0.7896729707717896,
      "learning_rate": 7.565709411881323e-06,
      "loss": 0.7544,
      "step": 12100
    },
    {
      "epoch": 0.34867169941796805,
      "grad_norm": 0.8058376312255859,
      "learning_rate": 7.565308907466843e-06,
      "loss": 0.775,
      "step": 12101
    },
    {
      "epoch": 0.34870051287961734,
      "grad_norm": 0.8261611461639404,
      "learning_rate": 7.5649083807110855e-06,
      "loss": 0.7661,
      "step": 12102
    },
    {
      "epoch": 0.34872932634126663,
      "grad_norm": 0.796049177646637,
      "learning_rate": 7.5645078316175415e-06,
      "loss": 0.7404,
      "step": 12103
    },
    {
      "epoch": 0.3487581398029159,
      "grad_norm": 0.7655412554740906,
      "learning_rate": 7.564107260189697e-06,
      "loss": 0.7342,
      "step": 12104
    },
    {
      "epoch": 0.3487869532645652,
      "grad_norm": 0.7777022123336792,
      "learning_rate": 7.563706666431041e-06,
      "loss": 0.7207,
      "step": 12105
    },
    {
      "epoch": 0.3488157667262145,
      "grad_norm": 0.8097335696220398,
      "learning_rate": 7.563306050345063e-06,
      "loss": 0.7803,
      "step": 12106
    },
    {
      "epoch": 0.3488445801878638,
      "grad_norm": 0.7898794412612915,
      "learning_rate": 7.5629054119352526e-06,
      "loss": 0.7398,
      "step": 12107
    },
    {
      "epoch": 0.34887339364951303,
      "grad_norm": 0.8003599047660828,
      "learning_rate": 7.562504751205097e-06,
      "loss": 0.7446,
      "step": 12108
    },
    {
      "epoch": 0.3489022071111623,
      "grad_norm": 0.8604703545570374,
      "learning_rate": 7.562104068158087e-06,
      "loss": 0.7657,
      "step": 12109
    },
    {
      "epoch": 0.3489310205728116,
      "grad_norm": 0.7874539494514465,
      "learning_rate": 7.561703362797712e-06,
      "loss": 0.7355,
      "step": 12110
    },
    {
      "epoch": 0.3489598340344609,
      "grad_norm": 0.7811632752418518,
      "learning_rate": 7.561302635127461e-06,
      "loss": 0.7744,
      "step": 12111
    },
    {
      "epoch": 0.3489886474961102,
      "grad_norm": 0.8266266584396362,
      "learning_rate": 7.560901885150824e-06,
      "loss": 0.7565,
      "step": 12112
    },
    {
      "epoch": 0.3490174609577595,
      "grad_norm": 0.869818389415741,
      "learning_rate": 7.5605011128712925e-06,
      "loss": 0.7509,
      "step": 12113
    },
    {
      "epoch": 0.3490462744194088,
      "grad_norm": 0.7744073867797852,
      "learning_rate": 7.560100318292356e-06,
      "loss": 0.7541,
      "step": 12114
    },
    {
      "epoch": 0.349075087881058,
      "grad_norm": 0.8127523064613342,
      "learning_rate": 7.559699501417504e-06,
      "loss": 0.7629,
      "step": 12115
    },
    {
      "epoch": 0.3491039013427073,
      "grad_norm": 0.7798715829849243,
      "learning_rate": 7.559298662250228e-06,
      "loss": 0.7255,
      "step": 12116
    },
    {
      "epoch": 0.3491327148043566,
      "grad_norm": 0.8109699487686157,
      "learning_rate": 7.558897800794021e-06,
      "loss": 0.7488,
      "step": 12117
    },
    {
      "epoch": 0.3491615282660059,
      "grad_norm": 0.7630916833877563,
      "learning_rate": 7.558496917052371e-06,
      "loss": 0.724,
      "step": 12118
    },
    {
      "epoch": 0.3491903417276552,
      "grad_norm": 0.7807985544204712,
      "learning_rate": 7.558096011028769e-06,
      "loss": 0.7333,
      "step": 12119
    },
    {
      "epoch": 0.34921915518930446,
      "grad_norm": 0.823401927947998,
      "learning_rate": 7.5576950827267105e-06,
      "loss": 0.7423,
      "step": 12120
    },
    {
      "epoch": 0.3492479686509537,
      "grad_norm": 0.7804315090179443,
      "learning_rate": 7.557294132149682e-06,
      "loss": 0.7033,
      "step": 12121
    },
    {
      "epoch": 0.349276782112603,
      "grad_norm": 0.7927826642990112,
      "learning_rate": 7.5568931593011785e-06,
      "loss": 0.7258,
      "step": 12122
    },
    {
      "epoch": 0.3493055955742523,
      "grad_norm": 0.7861157655715942,
      "learning_rate": 7.5564921641846925e-06,
      "loss": 0.7527,
      "step": 12123
    },
    {
      "epoch": 0.34933440903590157,
      "grad_norm": 0.8043622374534607,
      "learning_rate": 7.556091146803714e-06,
      "loss": 0.7527,
      "step": 12124
    },
    {
      "epoch": 0.34936322249755086,
      "grad_norm": 0.8379008173942566,
      "learning_rate": 7.555690107161737e-06,
      "loss": 0.7735,
      "step": 12125
    },
    {
      "epoch": 0.34939203595920015,
      "grad_norm": 0.8333812952041626,
      "learning_rate": 7.555289045262255e-06,
      "loss": 0.7548,
      "step": 12126
    },
    {
      "epoch": 0.34942084942084944,
      "grad_norm": 0.7835637927055359,
      "learning_rate": 7.554887961108759e-06,
      "loss": 0.7557,
      "step": 12127
    },
    {
      "epoch": 0.3494496628824987,
      "grad_norm": 0.7978457808494568,
      "learning_rate": 7.5544868547047435e-06,
      "loss": 0.7262,
      "step": 12128
    },
    {
      "epoch": 0.34947847634414797,
      "grad_norm": 0.8265353441238403,
      "learning_rate": 7.554085726053701e-06,
      "loss": 0.7239,
      "step": 12129
    },
    {
      "epoch": 0.34950728980579726,
      "grad_norm": 0.8105274438858032,
      "learning_rate": 7.553684575159125e-06,
      "loss": 0.7363,
      "step": 12130
    },
    {
      "epoch": 0.34953610326744655,
      "grad_norm": 0.8198778033256531,
      "learning_rate": 7.55328340202451e-06,
      "loss": 0.7298,
      "step": 12131
    },
    {
      "epoch": 0.34956491672909584,
      "grad_norm": 0.7874972820281982,
      "learning_rate": 7.552882206653347e-06,
      "loss": 0.7464,
      "step": 12132
    },
    {
      "epoch": 0.34959373019074513,
      "grad_norm": 0.8072221875190735,
      "learning_rate": 7.552480989049131e-06,
      "loss": 0.7473,
      "step": 12133
    },
    {
      "epoch": 0.3496225436523944,
      "grad_norm": 0.8231728672981262,
      "learning_rate": 7.55207974921536e-06,
      "loss": 0.726,
      "step": 12134
    },
    {
      "epoch": 0.34965135711404366,
      "grad_norm": 0.8213169574737549,
      "learning_rate": 7.551678487155523e-06,
      "loss": 0.7535,
      "step": 12135
    },
    {
      "epoch": 0.34968017057569295,
      "grad_norm": 0.7784144878387451,
      "learning_rate": 7.55127720287312e-06,
      "loss": 0.7725,
      "step": 12136
    },
    {
      "epoch": 0.34970898403734224,
      "grad_norm": 0.8538054823875427,
      "learning_rate": 7.5508758963716414e-06,
      "loss": 0.7242,
      "step": 12137
    },
    {
      "epoch": 0.34973779749899153,
      "grad_norm": 0.8427159190177917,
      "learning_rate": 7.550474567654583e-06,
      "loss": 0.7411,
      "step": 12138
    },
    {
      "epoch": 0.3497666109606408,
      "grad_norm": 1.472836971282959,
      "learning_rate": 7.55007321672544e-06,
      "loss": 0.7363,
      "step": 12139
    },
    {
      "epoch": 0.3497954244222901,
      "grad_norm": 0.7863512635231018,
      "learning_rate": 7.54967184358771e-06,
      "loss": 0.7633,
      "step": 12140
    },
    {
      "epoch": 0.3498242378839394,
      "grad_norm": 0.8060302734375,
      "learning_rate": 7.549270448244888e-06,
      "loss": 0.7703,
      "step": 12141
    },
    {
      "epoch": 0.34985305134558864,
      "grad_norm": 0.7506824135780334,
      "learning_rate": 7.548869030700465e-06,
      "loss": 0.7323,
      "step": 12142
    },
    {
      "epoch": 0.34988186480723793,
      "grad_norm": 0.779499351978302,
      "learning_rate": 7.5484675909579445e-06,
      "loss": 0.7588,
      "step": 12143
    },
    {
      "epoch": 0.3499106782688872,
      "grad_norm": 0.7799529433250427,
      "learning_rate": 7.548066129020817e-06,
      "loss": 0.7433,
      "step": 12144
    },
    {
      "epoch": 0.3499394917305365,
      "grad_norm": 0.8081849217414856,
      "learning_rate": 7.54766464489258e-06,
      "loss": 0.742,
      "step": 12145
    },
    {
      "epoch": 0.3499683051921858,
      "grad_norm": 0.7796536684036255,
      "learning_rate": 7.5472631385767325e-06,
      "loss": 0.7183,
      "step": 12146
    },
    {
      "epoch": 0.3499971186538351,
      "grad_norm": 0.7621690034866333,
      "learning_rate": 7.546861610076769e-06,
      "loss": 0.7149,
      "step": 12147
    },
    {
      "epoch": 0.35002593211548433,
      "grad_norm": 0.7499916553497314,
      "learning_rate": 7.546460059396186e-06,
      "loss": 0.7253,
      "step": 12148
    },
    {
      "epoch": 0.3500547455771336,
      "grad_norm": 0.7472728490829468,
      "learning_rate": 7.546058486538483e-06,
      "loss": 0.7425,
      "step": 12149
    },
    {
      "epoch": 0.3500835590387829,
      "grad_norm": 0.7962438464164734,
      "learning_rate": 7.545656891507153e-06,
      "loss": 0.7373,
      "step": 12150
    },
    {
      "epoch": 0.3501123725004322,
      "grad_norm": 0.7938312888145447,
      "learning_rate": 7.545255274305698e-06,
      "loss": 0.7343,
      "step": 12151
    },
    {
      "epoch": 0.3501411859620815,
      "grad_norm": 0.7667459845542908,
      "learning_rate": 7.544853634937614e-06,
      "loss": 0.7237,
      "step": 12152
    },
    {
      "epoch": 0.3501699994237308,
      "grad_norm": 0.769934356212616,
      "learning_rate": 7.5444519734063985e-06,
      "loss": 0.7632,
      "step": 12153
    },
    {
      "epoch": 0.3501988128853801,
      "grad_norm": 0.7425171136856079,
      "learning_rate": 7.54405028971555e-06,
      "loss": 0.7134,
      "step": 12154
    },
    {
      "epoch": 0.3502276263470293,
      "grad_norm": 0.7797067165374756,
      "learning_rate": 7.543648583868566e-06,
      "loss": 0.7055,
      "step": 12155
    },
    {
      "epoch": 0.3502564398086786,
      "grad_norm": 0.7775472402572632,
      "learning_rate": 7.543246855868947e-06,
      "loss": 0.7598,
      "step": 12156
    },
    {
      "epoch": 0.3502852532703279,
      "grad_norm": 0.7821950316429138,
      "learning_rate": 7.542845105720187e-06,
      "loss": 0.7174,
      "step": 12157
    },
    {
      "epoch": 0.3503140667319772,
      "grad_norm": 0.8058508038520813,
      "learning_rate": 7.542443333425791e-06,
      "loss": 0.7328,
      "step": 12158
    },
    {
      "epoch": 0.35034288019362647,
      "grad_norm": 0.7755135297775269,
      "learning_rate": 7.5420415389892545e-06,
      "loss": 0.7371,
      "step": 12159
    },
    {
      "epoch": 0.35037169365527576,
      "grad_norm": 0.7731264233589172,
      "learning_rate": 7.541639722414077e-06,
      "loss": 0.7443,
      "step": 12160
    },
    {
      "epoch": 0.35040050711692505,
      "grad_norm": 0.763014018535614,
      "learning_rate": 7.541237883703757e-06,
      "loss": 0.7237,
      "step": 12161
    },
    {
      "epoch": 0.3504293205785743,
      "grad_norm": 0.7738236784934998,
      "learning_rate": 7.540836022861797e-06,
      "loss": 0.7405,
      "step": 12162
    },
    {
      "epoch": 0.3504581340402236,
      "grad_norm": 0.7468668818473816,
      "learning_rate": 7.5404341398916946e-06,
      "loss": 0.7262,
      "step": 12163
    },
    {
      "epoch": 0.35048694750187287,
      "grad_norm": 0.7707664370536804,
      "learning_rate": 7.54003223479695e-06,
      "loss": 0.7153,
      "step": 12164
    },
    {
      "epoch": 0.35051576096352216,
      "grad_norm": 0.7722881436347961,
      "learning_rate": 7.539630307581065e-06,
      "loss": 0.7432,
      "step": 12165
    },
    {
      "epoch": 0.35054457442517145,
      "grad_norm": 0.7618131637573242,
      "learning_rate": 7.539228358247536e-06,
      "loss": 0.7244,
      "step": 12166
    },
    {
      "epoch": 0.35057338788682074,
      "grad_norm": 0.7539277672767639,
      "learning_rate": 7.538826386799869e-06,
      "loss": 0.7291,
      "step": 12167
    },
    {
      "epoch": 0.35060220134847003,
      "grad_norm": 0.805169403553009,
      "learning_rate": 7.538424393241562e-06,
      "loss": 0.769,
      "step": 12168
    },
    {
      "epoch": 0.35063101481011927,
      "grad_norm": 0.7660933136940002,
      "learning_rate": 7.538022377576115e-06,
      "loss": 0.7199,
      "step": 12169
    },
    {
      "epoch": 0.35065982827176856,
      "grad_norm": 0.7724733352661133,
      "learning_rate": 7.53762033980703e-06,
      "loss": 0.7289,
      "step": 12170
    },
    {
      "epoch": 0.35068864173341785,
      "grad_norm": 0.7473323345184326,
      "learning_rate": 7.53721827993781e-06,
      "loss": 0.7326,
      "step": 12171
    },
    {
      "epoch": 0.35071745519506714,
      "grad_norm": 0.7759649157524109,
      "learning_rate": 7.536816197971955e-06,
      "loss": 0.7706,
      "step": 12172
    },
    {
      "epoch": 0.35074626865671643,
      "grad_norm": 0.7568910717964172,
      "learning_rate": 7.536414093912965e-06,
      "loss": 0.7193,
      "step": 12173
    },
    {
      "epoch": 0.3507750821183657,
      "grad_norm": 0.7631442546844482,
      "learning_rate": 7.536011967764345e-06,
      "loss": 0.7211,
      "step": 12174
    },
    {
      "epoch": 0.35080389558001496,
      "grad_norm": 0.8149234652519226,
      "learning_rate": 7.535609819529594e-06,
      "loss": 0.7486,
      "step": 12175
    },
    {
      "epoch": 0.35083270904166425,
      "grad_norm": 0.775956928730011,
      "learning_rate": 7.535207649212219e-06,
      "loss": 0.7428,
      "step": 12176
    },
    {
      "epoch": 0.35086152250331354,
      "grad_norm": 0.7430815100669861,
      "learning_rate": 7.534805456815717e-06,
      "loss": 0.7317,
      "step": 12177
    },
    {
      "epoch": 0.35089033596496283,
      "grad_norm": 0.7725895047187805,
      "learning_rate": 7.5344032423435955e-06,
      "loss": 0.7391,
      "step": 12178
    },
    {
      "epoch": 0.3509191494266121,
      "grad_norm": 0.8374216556549072,
      "learning_rate": 7.534001005799354e-06,
      "loss": 0.7351,
      "step": 12179
    },
    {
      "epoch": 0.3509479628882614,
      "grad_norm": 0.7691704630851746,
      "learning_rate": 7.533598747186497e-06,
      "loss": 0.7581,
      "step": 12180
    },
    {
      "epoch": 0.3509767763499107,
      "grad_norm": 0.7965707778930664,
      "learning_rate": 7.533196466508528e-06,
      "loss": 0.7243,
      "step": 12181
    },
    {
      "epoch": 0.35100558981155994,
      "grad_norm": 0.7488424181938171,
      "learning_rate": 7.53279416376895e-06,
      "loss": 0.7476,
      "step": 12182
    },
    {
      "epoch": 0.35103440327320923,
      "grad_norm": 0.759942889213562,
      "learning_rate": 7.532391838971266e-06,
      "loss": 0.7443,
      "step": 12183
    },
    {
      "epoch": 0.3510632167348585,
      "grad_norm": 0.7528688907623291,
      "learning_rate": 7.531989492118981e-06,
      "loss": 0.729,
      "step": 12184
    },
    {
      "epoch": 0.3510920301965078,
      "grad_norm": 0.7655367851257324,
      "learning_rate": 7.531587123215598e-06,
      "loss": 0.7429,
      "step": 12185
    },
    {
      "epoch": 0.3511208436581571,
      "grad_norm": 0.7583006024360657,
      "learning_rate": 7.531184732264623e-06,
      "loss": 0.7659,
      "step": 12186
    },
    {
      "epoch": 0.3511496571198064,
      "grad_norm": 0.7946438789367676,
      "learning_rate": 7.530782319269559e-06,
      "loss": 0.7555,
      "step": 12187
    },
    {
      "epoch": 0.3511784705814557,
      "grad_norm": 0.7774735689163208,
      "learning_rate": 7.530379884233909e-06,
      "loss": 0.7286,
      "step": 12188
    },
    {
      "epoch": 0.3512072840431049,
      "grad_norm": 0.756314218044281,
      "learning_rate": 7.529977427161182e-06,
      "loss": 0.7476,
      "step": 12189
    },
    {
      "epoch": 0.3512360975047542,
      "grad_norm": 0.7747510075569153,
      "learning_rate": 7.529574948054881e-06,
      "loss": 0.7457,
      "step": 12190
    },
    {
      "epoch": 0.3512649109664035,
      "grad_norm": 0.8020899295806885,
      "learning_rate": 7.529172446918509e-06,
      "loss": 0.7235,
      "step": 12191
    },
    {
      "epoch": 0.3512937244280528,
      "grad_norm": 0.7693765163421631,
      "learning_rate": 7.528769923755575e-06,
      "loss": 0.7482,
      "step": 12192
    },
    {
      "epoch": 0.3513225378897021,
      "grad_norm": 0.7843798995018005,
      "learning_rate": 7.528367378569581e-06,
      "loss": 0.7602,
      "step": 12193
    },
    {
      "epoch": 0.35135135135135137,
      "grad_norm": 0.7579762935638428,
      "learning_rate": 7.5279648113640345e-06,
      "loss": 0.7407,
      "step": 12194
    },
    {
      "epoch": 0.35138016481300066,
      "grad_norm": 0.8114287853240967,
      "learning_rate": 7.527562222142444e-06,
      "loss": 0.734,
      "step": 12195
    },
    {
      "epoch": 0.3514089782746499,
      "grad_norm": 0.7868173718452454,
      "learning_rate": 7.527159610908311e-06,
      "loss": 0.7203,
      "step": 12196
    },
    {
      "epoch": 0.3514377917362992,
      "grad_norm": 0.7466931939125061,
      "learning_rate": 7.526756977665144e-06,
      "loss": 0.7346,
      "step": 12197
    },
    {
      "epoch": 0.3514666051979485,
      "grad_norm": 0.8037872910499573,
      "learning_rate": 7.526354322416451e-06,
      "loss": 0.7362,
      "step": 12198
    },
    {
      "epoch": 0.35149541865959777,
      "grad_norm": 0.7719217538833618,
      "learning_rate": 7.525951645165736e-06,
      "loss": 0.7375,
      "step": 12199
    },
    {
      "epoch": 0.35152423212124706,
      "grad_norm": 0.7634820342063904,
      "learning_rate": 7.525548945916508e-06,
      "loss": 0.7322,
      "step": 12200
    },
    {
      "epoch": 0.35155304558289635,
      "grad_norm": 0.7753027677536011,
      "learning_rate": 7.525146224672273e-06,
      "loss": 0.7473,
      "step": 12201
    },
    {
      "epoch": 0.3515818590445456,
      "grad_norm": 0.8107560873031616,
      "learning_rate": 7.524743481436537e-06,
      "loss": 0.7397,
      "step": 12202
    },
    {
      "epoch": 0.3516106725061949,
      "grad_norm": 0.7819803953170776,
      "learning_rate": 7.524340716212811e-06,
      "loss": 0.728,
      "step": 12203
    },
    {
      "epoch": 0.35163948596784417,
      "grad_norm": 0.7755972743034363,
      "learning_rate": 7.523937929004601e-06,
      "loss": 0.7522,
      "step": 12204
    },
    {
      "epoch": 0.35166829942949346,
      "grad_norm": 0.7684467434883118,
      "learning_rate": 7.5235351198154124e-06,
      "loss": 0.7504,
      "step": 12205
    },
    {
      "epoch": 0.35169711289114275,
      "grad_norm": 0.7696886658668518,
      "learning_rate": 7.523132288648758e-06,
      "loss": 0.7367,
      "step": 12206
    },
    {
      "epoch": 0.35172592635279204,
      "grad_norm": 0.7723578214645386,
      "learning_rate": 7.522729435508143e-06,
      "loss": 0.7538,
      "step": 12207
    },
    {
      "epoch": 0.35175473981444133,
      "grad_norm": 0.7483147978782654,
      "learning_rate": 7.522326560397076e-06,
      "loss": 0.7371,
      "step": 12208
    },
    {
      "epoch": 0.35178355327609057,
      "grad_norm": 0.76796555519104,
      "learning_rate": 7.521923663319066e-06,
      "loss": 0.7263,
      "step": 12209
    },
    {
      "epoch": 0.35181236673773986,
      "grad_norm": 0.8098405003547668,
      "learning_rate": 7.521520744277624e-06,
      "loss": 0.7367,
      "step": 12210
    },
    {
      "epoch": 0.35184118019938915,
      "grad_norm": 0.7590709328651428,
      "learning_rate": 7.521117803276253e-06,
      "loss": 0.7523,
      "step": 12211
    },
    {
      "epoch": 0.35186999366103844,
      "grad_norm": 0.756274938583374,
      "learning_rate": 7.520714840318469e-06,
      "loss": 0.7339,
      "step": 12212
    },
    {
      "epoch": 0.35189880712268773,
      "grad_norm": 0.774425208568573,
      "learning_rate": 7.520311855407778e-06,
      "loss": 0.7359,
      "step": 12213
    },
    {
      "epoch": 0.351927620584337,
      "grad_norm": 0.8094663023948669,
      "learning_rate": 7.51990884854769e-06,
      "loss": 0.7603,
      "step": 12214
    },
    {
      "epoch": 0.3519564340459863,
      "grad_norm": 0.7504511475563049,
      "learning_rate": 7.519505819741715e-06,
      "loss": 0.7301,
      "step": 12215
    },
    {
      "epoch": 0.35198524750763555,
      "grad_norm": 0.7784312963485718,
      "learning_rate": 7.5191027689933626e-06,
      "loss": 0.7487,
      "step": 12216
    },
    {
      "epoch": 0.35201406096928484,
      "grad_norm": 0.8104366064071655,
      "learning_rate": 7.518699696306142e-06,
      "loss": 0.7548,
      "step": 12217
    },
    {
      "epoch": 0.3520428744309341,
      "grad_norm": 0.759586751461029,
      "learning_rate": 7.518296601683567e-06,
      "loss": 0.7445,
      "step": 12218
    },
    {
      "epoch": 0.3520716878925834,
      "grad_norm": 0.8220040202140808,
      "learning_rate": 7.517893485129145e-06,
      "loss": 0.7408,
      "step": 12219
    },
    {
      "epoch": 0.3521005013542327,
      "grad_norm": 0.7547824382781982,
      "learning_rate": 7.517490346646388e-06,
      "loss": 0.7495,
      "step": 12220
    },
    {
      "epoch": 0.352129314815882,
      "grad_norm": 0.7360476851463318,
      "learning_rate": 7.517087186238805e-06,
      "loss": 0.7272,
      "step": 12221
    },
    {
      "epoch": 0.35215812827753123,
      "grad_norm": 0.8175367116928101,
      "learning_rate": 7.516684003909911e-06,
      "loss": 0.7545,
      "step": 12222
    },
    {
      "epoch": 0.3521869417391805,
      "grad_norm": 0.7809702754020691,
      "learning_rate": 7.516280799663214e-06,
      "loss": 0.7406,
      "step": 12223
    },
    {
      "epoch": 0.3522157552008298,
      "grad_norm": 0.7643014788627625,
      "learning_rate": 7.515877573502227e-06,
      "loss": 0.7365,
      "step": 12224
    },
    {
      "epoch": 0.3522445686624791,
      "grad_norm": 4.990627288818359,
      "learning_rate": 7.515474325430461e-06,
      "loss": 0.7474,
      "step": 12225
    },
    {
      "epoch": 0.3522733821241284,
      "grad_norm": 0.7756986021995544,
      "learning_rate": 7.515071055451429e-06,
      "loss": 0.7231,
      "step": 12226
    },
    {
      "epoch": 0.3523021955857777,
      "grad_norm": 0.7832804322242737,
      "learning_rate": 7.5146677635686415e-06,
      "loss": 0.7362,
      "step": 12227
    },
    {
      "epoch": 0.352331009047427,
      "grad_norm": 0.7961354851722717,
      "learning_rate": 7.514264449785612e-06,
      "loss": 0.7445,
      "step": 12228
    },
    {
      "epoch": 0.3523598225090762,
      "grad_norm": 0.802431583404541,
      "learning_rate": 7.5138611141058515e-06,
      "loss": 0.7342,
      "step": 12229
    },
    {
      "epoch": 0.3523886359707255,
      "grad_norm": 0.8046864867210388,
      "learning_rate": 7.513457756532875e-06,
      "loss": 0.7334,
      "step": 12230
    },
    {
      "epoch": 0.3524174494323748,
      "grad_norm": 0.8183853626251221,
      "learning_rate": 7.513054377070193e-06,
      "loss": 0.7001,
      "step": 12231
    },
    {
      "epoch": 0.3524462628940241,
      "grad_norm": 0.7704580426216125,
      "learning_rate": 7.51265097572132e-06,
      "loss": 0.7463,
      "step": 12232
    },
    {
      "epoch": 0.3524750763556734,
      "grad_norm": 0.7901611924171448,
      "learning_rate": 7.512247552489768e-06,
      "loss": 0.72,
      "step": 12233
    },
    {
      "epoch": 0.35250388981732267,
      "grad_norm": 0.7872961163520813,
      "learning_rate": 7.511844107379053e-06,
      "loss": 0.7497,
      "step": 12234
    },
    {
      "epoch": 0.35253270327897196,
      "grad_norm": 0.7675230503082275,
      "learning_rate": 7.511440640392684e-06,
      "loss": 0.7572,
      "step": 12235
    },
    {
      "epoch": 0.3525615167406212,
      "grad_norm": 0.8211008310317993,
      "learning_rate": 7.511037151534179e-06,
      "loss": 0.749,
      "step": 12236
    },
    {
      "epoch": 0.3525903302022705,
      "grad_norm": 0.7861496210098267,
      "learning_rate": 7.510633640807051e-06,
      "loss": 0.7472,
      "step": 12237
    },
    {
      "epoch": 0.3526191436639198,
      "grad_norm": 0.768867015838623,
      "learning_rate": 7.5102301082148135e-06,
      "loss": 0.7455,
      "step": 12238
    },
    {
      "epoch": 0.35264795712556907,
      "grad_norm": 0.7803993821144104,
      "learning_rate": 7.509826553760982e-06,
      "loss": 0.7381,
      "step": 12239
    },
    {
      "epoch": 0.35267677058721836,
      "grad_norm": 0.9481029510498047,
      "learning_rate": 7.5094229774490675e-06,
      "loss": 0.7565,
      "step": 12240
    },
    {
      "epoch": 0.35270558404886765,
      "grad_norm": 0.7902301549911499,
      "learning_rate": 7.509019379282589e-06,
      "loss": 0.7383,
      "step": 12241
    },
    {
      "epoch": 0.35273439751051694,
      "grad_norm": 0.7654304504394531,
      "learning_rate": 7.508615759265059e-06,
      "loss": 0.7374,
      "step": 12242
    },
    {
      "epoch": 0.3527632109721662,
      "grad_norm": 0.7994462847709656,
      "learning_rate": 7.508212117399994e-06,
      "loss": 0.7487,
      "step": 12243
    },
    {
      "epoch": 0.35279202443381547,
      "grad_norm": 0.7876371741294861,
      "learning_rate": 7.507808453690908e-06,
      "loss": 0.7277,
      "step": 12244
    },
    {
      "epoch": 0.35282083789546476,
      "grad_norm": 0.7818271517753601,
      "learning_rate": 7.507404768141319e-06,
      "loss": 0.7393,
      "step": 12245
    },
    {
      "epoch": 0.35284965135711405,
      "grad_norm": 0.7783252596855164,
      "learning_rate": 7.50700106075474e-06,
      "loss": 0.7448,
      "step": 12246
    },
    {
      "epoch": 0.35287846481876334,
      "grad_norm": 0.7731917500495911,
      "learning_rate": 7.506597331534686e-06,
      "loss": 0.7194,
      "step": 12247
    },
    {
      "epoch": 0.35290727828041263,
      "grad_norm": 0.7736202478408813,
      "learning_rate": 7.5061935804846776e-06,
      "loss": 0.7359,
      "step": 12248
    },
    {
      "epoch": 0.35293609174206186,
      "grad_norm": 0.7597277164459229,
      "learning_rate": 7.505789807608228e-06,
      "loss": 0.7381,
      "step": 12249
    },
    {
      "epoch": 0.35296490520371115,
      "grad_norm": 0.824619710445404,
      "learning_rate": 7.505386012908854e-06,
      "loss": 0.763,
      "step": 12250
    },
    {
      "epoch": 0.35299371866536045,
      "grad_norm": 0.7648443579673767,
      "learning_rate": 7.504982196390072e-06,
      "loss": 0.7261,
      "step": 12251
    },
    {
      "epoch": 0.35302253212700974,
      "grad_norm": 0.7649303078651428,
      "learning_rate": 7.504578358055399e-06,
      "loss": 0.7516,
      "step": 12252
    },
    {
      "epoch": 0.353051345588659,
      "grad_norm": 0.7676900625228882,
      "learning_rate": 7.5041744979083516e-06,
      "loss": 0.7377,
      "step": 12253
    },
    {
      "epoch": 0.3530801590503083,
      "grad_norm": 0.7896332740783691,
      "learning_rate": 7.503770615952448e-06,
      "loss": 0.719,
      "step": 12254
    },
    {
      "epoch": 0.3531089725119576,
      "grad_norm": 0.7652387619018555,
      "learning_rate": 7.503366712191205e-06,
      "loss": 0.7442,
      "step": 12255
    },
    {
      "epoch": 0.35313778597360684,
      "grad_norm": 0.7982746362686157,
      "learning_rate": 7.502962786628142e-06,
      "loss": 0.7689,
      "step": 12256
    },
    {
      "epoch": 0.35316659943525613,
      "grad_norm": 0.7830382585525513,
      "learning_rate": 7.502558839266773e-06,
      "loss": 0.7498,
      "step": 12257
    },
    {
      "epoch": 0.3531954128969054,
      "grad_norm": 0.7539769411087036,
      "learning_rate": 7.502154870110621e-06,
      "loss": 0.7375,
      "step": 12258
    },
    {
      "epoch": 0.3532242263585547,
      "grad_norm": 0.7599236369132996,
      "learning_rate": 7.501750879163198e-06,
      "loss": 0.7197,
      "step": 12259
    },
    {
      "epoch": 0.353253039820204,
      "grad_norm": 0.7683375477790833,
      "learning_rate": 7.501346866428028e-06,
      "loss": 0.717,
      "step": 12260
    },
    {
      "epoch": 0.3532818532818533,
      "grad_norm": 0.7929191589355469,
      "learning_rate": 7.500942831908624e-06,
      "loss": 0.7391,
      "step": 12261
    },
    {
      "epoch": 0.3533106667435026,
      "grad_norm": 0.7971802949905396,
      "learning_rate": 7.500538775608511e-06,
      "loss": 0.7168,
      "step": 12262
    },
    {
      "epoch": 0.3533394802051518,
      "grad_norm": 0.7698802351951599,
      "learning_rate": 7.500134697531205e-06,
      "loss": 0.7543,
      "step": 12263
    },
    {
      "epoch": 0.3533682936668011,
      "grad_norm": 0.7635290622711182,
      "learning_rate": 7.499730597680224e-06,
      "loss": 0.7594,
      "step": 12264
    },
    {
      "epoch": 0.3533971071284504,
      "grad_norm": 0.7453317046165466,
      "learning_rate": 7.499326476059088e-06,
      "loss": 0.7471,
      "step": 12265
    },
    {
      "epoch": 0.3534259205900997,
      "grad_norm": 0.7550840973854065,
      "learning_rate": 7.498922332671316e-06,
      "loss": 0.7459,
      "step": 12266
    },
    {
      "epoch": 0.353454734051749,
      "grad_norm": 0.7864168882369995,
      "learning_rate": 7.498518167520431e-06,
      "loss": 0.7551,
      "step": 12267
    },
    {
      "epoch": 0.3534835475133983,
      "grad_norm": 0.7603254914283752,
      "learning_rate": 7.498113980609949e-06,
      "loss": 0.7342,
      "step": 12268
    },
    {
      "epoch": 0.35351236097504757,
      "grad_norm": 0.8066340684890747,
      "learning_rate": 7.4977097719433925e-06,
      "loss": 0.7685,
      "step": 12269
    },
    {
      "epoch": 0.3535411744366968,
      "grad_norm": 0.7588194608688354,
      "learning_rate": 7.497305541524279e-06,
      "loss": 0.7306,
      "step": 12270
    },
    {
      "epoch": 0.3535699878983461,
      "grad_norm": 1.6187312602996826,
      "learning_rate": 7.496901289356131e-06,
      "loss": 0.7641,
      "step": 12271
    },
    {
      "epoch": 0.3535988013599954,
      "grad_norm": 0.8107759952545166,
      "learning_rate": 7.496497015442469e-06,
      "loss": 0.7654,
      "step": 12272
    },
    {
      "epoch": 0.3536276148216447,
      "grad_norm": 0.8353114128112793,
      "learning_rate": 7.496092719786814e-06,
      "loss": 0.7485,
      "step": 12273
    },
    {
      "epoch": 0.35365642828329397,
      "grad_norm": 0.7461850047111511,
      "learning_rate": 7.495688402392687e-06,
      "loss": 0.7285,
      "step": 12274
    },
    {
      "epoch": 0.35368524174494326,
      "grad_norm": 0.7697840929031372,
      "learning_rate": 7.495284063263608e-06,
      "loss": 0.7498,
      "step": 12275
    },
    {
      "epoch": 0.3537140552065925,
      "grad_norm": 0.7971102595329285,
      "learning_rate": 7.4948797024031e-06,
      "loss": 0.7564,
      "step": 12276
    },
    {
      "epoch": 0.3537428686682418,
      "grad_norm": 0.7562421560287476,
      "learning_rate": 7.494475319814681e-06,
      "loss": 0.7321,
      "step": 12277
    },
    {
      "epoch": 0.3537716821298911,
      "grad_norm": 0.7575778961181641,
      "learning_rate": 7.4940709155018785e-06,
      "loss": 0.7329,
      "step": 12278
    },
    {
      "epoch": 0.35380049559154036,
      "grad_norm": 0.766775369644165,
      "learning_rate": 7.49366648946821e-06,
      "loss": 0.7544,
      "step": 12279
    },
    {
      "epoch": 0.35382930905318966,
      "grad_norm": 0.8164690732955933,
      "learning_rate": 7.4932620417171995e-06,
      "loss": 0.7358,
      "step": 12280
    },
    {
      "epoch": 0.35385812251483895,
      "grad_norm": 0.7570450305938721,
      "learning_rate": 7.492857572252369e-06,
      "loss": 0.735,
      "step": 12281
    },
    {
      "epoch": 0.35388693597648824,
      "grad_norm": 0.749635636806488,
      "learning_rate": 7.492453081077241e-06,
      "loss": 0.7222,
      "step": 12282
    },
    {
      "epoch": 0.3539157494381375,
      "grad_norm": 0.7770226001739502,
      "learning_rate": 7.492048568195338e-06,
      "loss": 0.7249,
      "step": 12283
    },
    {
      "epoch": 0.35394456289978676,
      "grad_norm": 0.7901058197021484,
      "learning_rate": 7.491644033610183e-06,
      "loss": 0.737,
      "step": 12284
    },
    {
      "epoch": 0.35397337636143605,
      "grad_norm": 0.7822508215904236,
      "learning_rate": 7.491239477325301e-06,
      "loss": 0.7482,
      "step": 12285
    },
    {
      "epoch": 0.35400218982308534,
      "grad_norm": 0.7872046828269958,
      "learning_rate": 7.490834899344211e-06,
      "loss": 0.732,
      "step": 12286
    },
    {
      "epoch": 0.35403100328473464,
      "grad_norm": 0.7804462313652039,
      "learning_rate": 7.490430299670441e-06,
      "loss": 0.7476,
      "step": 12287
    },
    {
      "epoch": 0.3540598167463839,
      "grad_norm": 0.7517085671424866,
      "learning_rate": 7.490025678307512e-06,
      "loss": 0.7237,
      "step": 12288
    },
    {
      "epoch": 0.3540886302080332,
      "grad_norm": 0.797857940196991,
      "learning_rate": 7.489621035258947e-06,
      "loss": 0.7362,
      "step": 12289
    },
    {
      "epoch": 0.35411744366968245,
      "grad_norm": 0.7728413343429565,
      "learning_rate": 7.4892163705282725e-06,
      "loss": 0.7341,
      "step": 12290
    },
    {
      "epoch": 0.35414625713133174,
      "grad_norm": 0.7873786687850952,
      "learning_rate": 7.488811684119013e-06,
      "loss": 0.7424,
      "step": 12291
    },
    {
      "epoch": 0.35417507059298103,
      "grad_norm": 0.7666369676589966,
      "learning_rate": 7.48840697603469e-06,
      "loss": 0.7459,
      "step": 12292
    },
    {
      "epoch": 0.3542038840546303,
      "grad_norm": 0.8200348615646362,
      "learning_rate": 7.4880022462788306e-06,
      "loss": 0.7432,
      "step": 12293
    },
    {
      "epoch": 0.3542326975162796,
      "grad_norm": 0.7624673247337341,
      "learning_rate": 7.487597494854959e-06,
      "loss": 0.7244,
      "step": 12294
    },
    {
      "epoch": 0.3542615109779289,
      "grad_norm": 0.7846153974533081,
      "learning_rate": 7.487192721766598e-06,
      "loss": 0.7266,
      "step": 12295
    },
    {
      "epoch": 0.3542903244395782,
      "grad_norm": 0.7591754794120789,
      "learning_rate": 7.486787927017276e-06,
      "loss": 0.7317,
      "step": 12296
    },
    {
      "epoch": 0.35431913790122743,
      "grad_norm": 0.7959897518157959,
      "learning_rate": 7.486383110610517e-06,
      "loss": 0.7344,
      "step": 12297
    },
    {
      "epoch": 0.3543479513628767,
      "grad_norm": 0.7818376421928406,
      "learning_rate": 7.485978272549846e-06,
      "loss": 0.7619,
      "step": 12298
    },
    {
      "epoch": 0.354376764824526,
      "grad_norm": 0.7639544606208801,
      "learning_rate": 7.48557341283879e-06,
      "loss": 0.7284,
      "step": 12299
    },
    {
      "epoch": 0.3544055782861753,
      "grad_norm": 0.7863299250602722,
      "learning_rate": 7.485168531480874e-06,
      "loss": 0.7641,
      "step": 12300
    },
    {
      "epoch": 0.3544343917478246,
      "grad_norm": 0.7645857334136963,
      "learning_rate": 7.484763628479624e-06,
      "loss": 0.76,
      "step": 12301
    },
    {
      "epoch": 0.3544632052094739,
      "grad_norm": 0.7842814922332764,
      "learning_rate": 7.4843587038385665e-06,
      "loss": 0.7122,
      "step": 12302
    },
    {
      "epoch": 0.3544920186711231,
      "grad_norm": 0.8114845752716064,
      "learning_rate": 7.48395375756123e-06,
      "loss": 0.7613,
      "step": 12303
    },
    {
      "epoch": 0.3545208321327724,
      "grad_norm": 0.7704209685325623,
      "learning_rate": 7.483548789651138e-06,
      "loss": 0.7129,
      "step": 12304
    },
    {
      "epoch": 0.3545496455944217,
      "grad_norm": 0.7943167090415955,
      "learning_rate": 7.4831438001118186e-06,
      "loss": 0.7463,
      "step": 12305
    },
    {
      "epoch": 0.354578459056071,
      "grad_norm": 0.77790766954422,
      "learning_rate": 7.4827387889468e-06,
      "loss": 0.7373,
      "step": 12306
    },
    {
      "epoch": 0.3546072725177203,
      "grad_norm": 0.758829653263092,
      "learning_rate": 7.482333756159605e-06,
      "loss": 0.7238,
      "step": 12307
    },
    {
      "epoch": 0.3546360859793696,
      "grad_norm": 0.843636155128479,
      "learning_rate": 7.481928701753768e-06,
      "loss": 0.7473,
      "step": 12308
    },
    {
      "epoch": 0.35466489944101887,
      "grad_norm": 0.8087693452835083,
      "learning_rate": 7.481523625732811e-06,
      "loss": 0.7509,
      "step": 12309
    },
    {
      "epoch": 0.3546937129026681,
      "grad_norm": 0.7694956064224243,
      "learning_rate": 7.481118528100266e-06,
      "loss": 0.7329,
      "step": 12310
    },
    {
      "epoch": 0.3547225263643174,
      "grad_norm": 0.8043522238731384,
      "learning_rate": 7.480713408859658e-06,
      "loss": 0.7312,
      "step": 12311
    },
    {
      "epoch": 0.3547513398259667,
      "grad_norm": 0.7522567510604858,
      "learning_rate": 7.480308268014516e-06,
      "loss": 0.7353,
      "step": 12312
    },
    {
      "epoch": 0.354780153287616,
      "grad_norm": 0.7686366438865662,
      "learning_rate": 7.479903105568367e-06,
      "loss": 0.7118,
      "step": 12313
    },
    {
      "epoch": 0.35480896674926526,
      "grad_norm": 2.607724666595459,
      "learning_rate": 7.479497921524741e-06,
      "loss": 0.7901,
      "step": 12314
    },
    {
      "epoch": 0.35483778021091456,
      "grad_norm": 0.7771077156066895,
      "learning_rate": 7.479092715887169e-06,
      "loss": 0.7495,
      "step": 12315
    },
    {
      "epoch": 0.35486659367256385,
      "grad_norm": 0.7690101265907288,
      "learning_rate": 7.478687488659175e-06,
      "loss": 0.7351,
      "step": 12316
    },
    {
      "epoch": 0.3548954071342131,
      "grad_norm": 0.7468268275260925,
      "learning_rate": 7.4782822398442925e-06,
      "loss": 0.7495,
      "step": 12317
    },
    {
      "epoch": 0.35492422059586237,
      "grad_norm": 0.7896434664726257,
      "learning_rate": 7.477876969446048e-06,
      "loss": 0.7568,
      "step": 12318
    },
    {
      "epoch": 0.35495303405751166,
      "grad_norm": 0.7872797250747681,
      "learning_rate": 7.4774716774679735e-06,
      "loss": 0.739,
      "step": 12319
    },
    {
      "epoch": 0.35498184751916095,
      "grad_norm": 0.7708626389503479,
      "learning_rate": 7.477066363913595e-06,
      "loss": 0.7379,
      "step": 12320
    },
    {
      "epoch": 0.35501066098081024,
      "grad_norm": 0.7607110142707825,
      "learning_rate": 7.476661028786446e-06,
      "loss": 0.763,
      "step": 12321
    },
    {
      "epoch": 0.35503947444245954,
      "grad_norm": 0.7594691514968872,
      "learning_rate": 7.476255672090056e-06,
      "loss": 0.7421,
      "step": 12322
    },
    {
      "epoch": 0.3550682879041088,
      "grad_norm": 0.787630558013916,
      "learning_rate": 7.475850293827953e-06,
      "loss": 0.7287,
      "step": 12323
    },
    {
      "epoch": 0.35509710136575806,
      "grad_norm": 0.7754614353179932,
      "learning_rate": 7.475444894003669e-06,
      "loss": 0.7526,
      "step": 12324
    },
    {
      "epoch": 0.35512591482740735,
      "grad_norm": 0.7495574355125427,
      "learning_rate": 7.475039472620735e-06,
      "loss": 0.7676,
      "step": 12325
    },
    {
      "epoch": 0.35515472828905664,
      "grad_norm": 0.7544833421707153,
      "learning_rate": 7.4746340296826815e-06,
      "loss": 0.7357,
      "step": 12326
    },
    {
      "epoch": 0.35518354175070593,
      "grad_norm": 0.7757566571235657,
      "learning_rate": 7.47422856519304e-06,
      "loss": 0.7301,
      "step": 12327
    },
    {
      "epoch": 0.3552123552123552,
      "grad_norm": 0.7937452793121338,
      "learning_rate": 7.47382307915534e-06,
      "loss": 0.7821,
      "step": 12328
    },
    {
      "epoch": 0.3552411686740045,
      "grad_norm": 0.7387498617172241,
      "learning_rate": 7.473417571573114e-06,
      "loss": 0.7375,
      "step": 12329
    },
    {
      "epoch": 0.35526998213565375,
      "grad_norm": 0.7833258509635925,
      "learning_rate": 7.4730120424498945e-06,
      "loss": 0.732,
      "step": 12330
    },
    {
      "epoch": 0.35529879559730304,
      "grad_norm": 0.7562470436096191,
      "learning_rate": 7.4726064917892104e-06,
      "loss": 0.7356,
      "step": 12331
    },
    {
      "epoch": 0.35532760905895233,
      "grad_norm": 0.765060544013977,
      "learning_rate": 7.472200919594597e-06,
      "loss": 0.7412,
      "step": 12332
    },
    {
      "epoch": 0.3553564225206016,
      "grad_norm": 0.7583305835723877,
      "learning_rate": 7.471795325869584e-06,
      "loss": 0.7579,
      "step": 12333
    },
    {
      "epoch": 0.3553852359822509,
      "grad_norm": 0.7470635175704956,
      "learning_rate": 7.471389710617705e-06,
      "loss": 0.7317,
      "step": 12334
    },
    {
      "epoch": 0.3554140494439002,
      "grad_norm": 0.7860587239265442,
      "learning_rate": 7.470984073842492e-06,
      "loss": 0.7775,
      "step": 12335
    },
    {
      "epoch": 0.3554428629055495,
      "grad_norm": 0.7790300846099854,
      "learning_rate": 7.470578415547478e-06,
      "loss": 0.7187,
      "step": 12336
    },
    {
      "epoch": 0.35547167636719873,
      "grad_norm": 0.7572951912879944,
      "learning_rate": 7.470172735736194e-06,
      "loss": 0.7279,
      "step": 12337
    },
    {
      "epoch": 0.355500489828848,
      "grad_norm": 0.7436115741729736,
      "learning_rate": 7.469767034412176e-06,
      "loss": 0.7425,
      "step": 12338
    },
    {
      "epoch": 0.3555293032904973,
      "grad_norm": 0.7932392954826355,
      "learning_rate": 7.4693613115789554e-06,
      "loss": 0.7375,
      "step": 12339
    },
    {
      "epoch": 0.3555581167521466,
      "grad_norm": 0.796216607093811,
      "learning_rate": 7.468955567240068e-06,
      "loss": 0.7357,
      "step": 12340
    },
    {
      "epoch": 0.3555869302137959,
      "grad_norm": 0.7711541056632996,
      "learning_rate": 7.468549801399044e-06,
      "loss": 0.7488,
      "step": 12341
    },
    {
      "epoch": 0.3556157436754452,
      "grad_norm": 0.8109049201011658,
      "learning_rate": 7.46814401405942e-06,
      "loss": 0.7554,
      "step": 12342
    },
    {
      "epoch": 0.3556445571370945,
      "grad_norm": 0.7711625099182129,
      "learning_rate": 7.467738205224728e-06,
      "loss": 0.7393,
      "step": 12343
    },
    {
      "epoch": 0.3556733705987437,
      "grad_norm": 0.7419314980506897,
      "learning_rate": 7.467332374898504e-06,
      "loss": 0.738,
      "step": 12344
    },
    {
      "epoch": 0.355702184060393,
      "grad_norm": 0.7933905124664307,
      "learning_rate": 7.46692652308428e-06,
      "loss": 0.7348,
      "step": 12345
    },
    {
      "epoch": 0.3557309975220423,
      "grad_norm": 0.7651790380477905,
      "learning_rate": 7.466520649785594e-06,
      "loss": 0.7279,
      "step": 12346
    },
    {
      "epoch": 0.3557598109836916,
      "grad_norm": 0.7637474536895752,
      "learning_rate": 7.466114755005977e-06,
      "loss": 0.7314,
      "step": 12347
    },
    {
      "epoch": 0.3557886244453409,
      "grad_norm": 0.7699828147888184,
      "learning_rate": 7.465708838748967e-06,
      "loss": 0.7522,
      "step": 12348
    },
    {
      "epoch": 0.35581743790699016,
      "grad_norm": 0.7951697111129761,
      "learning_rate": 7.465302901018097e-06,
      "loss": 0.7379,
      "step": 12349
    },
    {
      "epoch": 0.35584625136863945,
      "grad_norm": 0.7244309782981873,
      "learning_rate": 7.464896941816902e-06,
      "loss": 0.7222,
      "step": 12350
    },
    {
      "epoch": 0.3558750648302887,
      "grad_norm": 13.023500442504883,
      "learning_rate": 7.464490961148921e-06,
      "loss": 0.8161,
      "step": 12351
    },
    {
      "epoch": 0.355903878291938,
      "grad_norm": 0.8440571427345276,
      "learning_rate": 7.4640849590176865e-06,
      "loss": 0.7154,
      "step": 12352
    },
    {
      "epoch": 0.35593269175358727,
      "grad_norm": 0.9899745583534241,
      "learning_rate": 7.463678935426735e-06,
      "loss": 0.7771,
      "step": 12353
    },
    {
      "epoch": 0.35596150521523656,
      "grad_norm": 1.1785601377487183,
      "learning_rate": 7.463272890379603e-06,
      "loss": 0.7716,
      "step": 12354
    },
    {
      "epoch": 0.35599031867688585,
      "grad_norm": 0.8627496361732483,
      "learning_rate": 7.462866823879826e-06,
      "loss": 0.7325,
      "step": 12355
    },
    {
      "epoch": 0.35601913213853514,
      "grad_norm": 0.9177410006523132,
      "learning_rate": 7.4624607359309406e-06,
      "loss": 0.7261,
      "step": 12356
    },
    {
      "epoch": 0.3560479456001844,
      "grad_norm": 0.8661530017852783,
      "learning_rate": 7.462054626536485e-06,
      "loss": 0.7319,
      "step": 12357
    },
    {
      "epoch": 0.35607675906183367,
      "grad_norm": 0.857063889503479,
      "learning_rate": 7.461648495699993e-06,
      "loss": 0.7635,
      "step": 12358
    },
    {
      "epoch": 0.35610557252348296,
      "grad_norm": 0.9084339737892151,
      "learning_rate": 7.4612423434250045e-06,
      "loss": 0.754,
      "step": 12359
    },
    {
      "epoch": 0.35613438598513225,
      "grad_norm": 0.9966819286346436,
      "learning_rate": 7.460836169715056e-06,
      "loss": 0.7762,
      "step": 12360
    },
    {
      "epoch": 0.35616319944678154,
      "grad_norm": 1.0396770238876343,
      "learning_rate": 7.460429974573683e-06,
      "loss": 0.7486,
      "step": 12361
    },
    {
      "epoch": 0.35619201290843083,
      "grad_norm": 0.83632892370224,
      "learning_rate": 7.4600237580044255e-06,
      "loss": 0.7486,
      "step": 12362
    },
    {
      "epoch": 0.3562208263700801,
      "grad_norm": 0.8766055703163147,
      "learning_rate": 7.45961752001082e-06,
      "loss": 0.7166,
      "step": 12363
    },
    {
      "epoch": 0.35624963983172936,
      "grad_norm": 0.8844005465507507,
      "learning_rate": 7.459211260596404e-06,
      "loss": 0.7312,
      "step": 12364
    },
    {
      "epoch": 0.35627845329337865,
      "grad_norm": 0.9166683554649353,
      "learning_rate": 7.4588049797647175e-06,
      "loss": 0.7593,
      "step": 12365
    },
    {
      "epoch": 0.35630726675502794,
      "grad_norm": 0.7781643271446228,
      "learning_rate": 7.458398677519296e-06,
      "loss": 0.7138,
      "step": 12366
    },
    {
      "epoch": 0.35633608021667723,
      "grad_norm": 0.8955307602882385,
      "learning_rate": 7.4579923538636815e-06,
      "loss": 0.7429,
      "step": 12367
    },
    {
      "epoch": 0.3563648936783265,
      "grad_norm": 0.854034423828125,
      "learning_rate": 7.457586008801409e-06,
      "loss": 0.7566,
      "step": 12368
    },
    {
      "epoch": 0.3563937071399758,
      "grad_norm": 0.9788275361061096,
      "learning_rate": 7.45717964233602e-06,
      "loss": 0.7556,
      "step": 12369
    },
    {
      "epoch": 0.3564225206016251,
      "grad_norm": 0.7900778651237488,
      "learning_rate": 7.456773254471053e-06,
      "loss": 0.736,
      "step": 12370
    },
    {
      "epoch": 0.35645133406327434,
      "grad_norm": 0.8094109892845154,
      "learning_rate": 7.456366845210046e-06,
      "loss": 0.7336,
      "step": 12371
    },
    {
      "epoch": 0.35648014752492363,
      "grad_norm": 0.8588293790817261,
      "learning_rate": 7.455960414556538e-06,
      "loss": 0.7651,
      "step": 12372
    },
    {
      "epoch": 0.3565089609865729,
      "grad_norm": 0.8951162099838257,
      "learning_rate": 7.45555396251407e-06,
      "loss": 0.7495,
      "step": 12373
    },
    {
      "epoch": 0.3565377744482222,
      "grad_norm": 0.816375195980072,
      "learning_rate": 7.4551474890861845e-06,
      "loss": 0.7397,
      "step": 12374
    },
    {
      "epoch": 0.3565665879098715,
      "grad_norm": 0.8186528086662292,
      "learning_rate": 7.454740994276417e-06,
      "loss": 0.735,
      "step": 12375
    },
    {
      "epoch": 0.3565954013715208,
      "grad_norm": 0.8341619968414307,
      "learning_rate": 7.454334478088309e-06,
      "loss": 0.7465,
      "step": 12376
    },
    {
      "epoch": 0.3566242148331701,
      "grad_norm": 0.8392030000686646,
      "learning_rate": 7.4539279405254e-06,
      "loss": 0.7286,
      "step": 12377
    },
    {
      "epoch": 0.3566530282948193,
      "grad_norm": 0.7855136394500732,
      "learning_rate": 7.4535213815912335e-06,
      "loss": 0.7215,
      "step": 12378
    },
    {
      "epoch": 0.3566818417564686,
      "grad_norm": 0.8233960866928101,
      "learning_rate": 7.453114801289349e-06,
      "loss": 0.7441,
      "step": 12379
    },
    {
      "epoch": 0.3567106552181179,
      "grad_norm": 0.8088493943214417,
      "learning_rate": 7.4527081996232845e-06,
      "loss": 0.7414,
      "step": 12380
    },
    {
      "epoch": 0.3567394686797672,
      "grad_norm": 0.8084051012992859,
      "learning_rate": 7.452301576596584e-06,
      "loss": 0.7497,
      "step": 12381
    },
    {
      "epoch": 0.3567682821414165,
      "grad_norm": 0.8295155763626099,
      "learning_rate": 7.451894932212789e-06,
      "loss": 0.7603,
      "step": 12382
    },
    {
      "epoch": 0.3567970956030658,
      "grad_norm": 0.8523538708686829,
      "learning_rate": 7.45148826647544e-06,
      "loss": 0.762,
      "step": 12383
    },
    {
      "epoch": 0.356825909064715,
      "grad_norm": 0.7782810926437378,
      "learning_rate": 7.451081579388077e-06,
      "loss": 0.7506,
      "step": 12384
    },
    {
      "epoch": 0.3568547225263643,
      "grad_norm": 0.792449414730072,
      "learning_rate": 7.450674870954246e-06,
      "loss": 0.7278,
      "step": 12385
    },
    {
      "epoch": 0.3568835359880136,
      "grad_norm": 0.7969391942024231,
      "learning_rate": 7.450268141177487e-06,
      "loss": 0.7406,
      "step": 12386
    },
    {
      "epoch": 0.3569123494496629,
      "grad_norm": 0.7806527614593506,
      "learning_rate": 7.44986139006134e-06,
      "loss": 0.7155,
      "step": 12387
    },
    {
      "epoch": 0.35694116291131217,
      "grad_norm": 0.7379750609397888,
      "learning_rate": 7.449454617609351e-06,
      "loss": 0.7178,
      "step": 12388
    },
    {
      "epoch": 0.35696997637296146,
      "grad_norm": 0.7831646800041199,
      "learning_rate": 7.44904782382506e-06,
      "loss": 0.7135,
      "step": 12389
    },
    {
      "epoch": 0.35699878983461075,
      "grad_norm": 0.7849818468093872,
      "learning_rate": 7.44864100871201e-06,
      "loss": 0.7341,
      "step": 12390
    },
    {
      "epoch": 0.35702760329626,
      "grad_norm": 0.8404879570007324,
      "learning_rate": 7.448234172273745e-06,
      "loss": 0.7622,
      "step": 12391
    },
    {
      "epoch": 0.3570564167579093,
      "grad_norm": 0.7886494398117065,
      "learning_rate": 7.447827314513808e-06,
      "loss": 0.7217,
      "step": 12392
    },
    {
      "epoch": 0.35708523021955857,
      "grad_norm": 0.775594174861908,
      "learning_rate": 7.4474204354357414e-06,
      "loss": 0.759,
      "step": 12393
    },
    {
      "epoch": 0.35711404368120786,
      "grad_norm": 0.7736100554466248,
      "learning_rate": 7.447013535043091e-06,
      "loss": 0.7263,
      "step": 12394
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 0.8225340247154236,
      "learning_rate": 7.446606613339397e-06,
      "loss": 0.7309,
      "step": 12395
    },
    {
      "epoch": 0.35717167060450644,
      "grad_norm": 0.8145369291305542,
      "learning_rate": 7.446199670328205e-06,
      "loss": 0.7276,
      "step": 12396
    },
    {
      "epoch": 0.35720048406615573,
      "grad_norm": 0.7991620898246765,
      "learning_rate": 7.44579270601306e-06,
      "loss": 0.7501,
      "step": 12397
    },
    {
      "epoch": 0.35722929752780497,
      "grad_norm": 0.7899960875511169,
      "learning_rate": 7.4453857203975056e-06,
      "loss": 0.7578,
      "step": 12398
    },
    {
      "epoch": 0.35725811098945426,
      "grad_norm": 0.7759560942649841,
      "learning_rate": 7.444978713485085e-06,
      "loss": 0.7283,
      "step": 12399
    },
    {
      "epoch": 0.35728692445110355,
      "grad_norm": 0.7859528064727783,
      "learning_rate": 7.444571685279345e-06,
      "loss": 0.7375,
      "step": 12400
    },
    {
      "epoch": 0.35731573791275284,
      "grad_norm": 0.8250030279159546,
      "learning_rate": 7.4441646357838286e-06,
      "loss": 0.7137,
      "step": 12401
    },
    {
      "epoch": 0.35734455137440213,
      "grad_norm": 0.7898454666137695,
      "learning_rate": 7.443757565002081e-06,
      "loss": 0.7567,
      "step": 12402
    },
    {
      "epoch": 0.3573733648360514,
      "grad_norm": 0.7755634784698486,
      "learning_rate": 7.443350472937648e-06,
      "loss": 0.7359,
      "step": 12403
    },
    {
      "epoch": 0.3574021782977007,
      "grad_norm": 0.7775802612304688,
      "learning_rate": 7.442943359594077e-06,
      "loss": 0.7567,
      "step": 12404
    },
    {
      "epoch": 0.35743099175934995,
      "grad_norm": 0.7916503548622131,
      "learning_rate": 7.44253622497491e-06,
      "loss": 0.7432,
      "step": 12405
    },
    {
      "epoch": 0.35745980522099924,
      "grad_norm": 0.7514825463294983,
      "learning_rate": 7.442129069083693e-06,
      "loss": 0.7125,
      "step": 12406
    },
    {
      "epoch": 0.35748861868264853,
      "grad_norm": 0.8029710054397583,
      "learning_rate": 7.441721891923974e-06,
      "loss": 0.7547,
      "step": 12407
    },
    {
      "epoch": 0.3575174321442978,
      "grad_norm": 0.7827309966087341,
      "learning_rate": 7.441314693499299e-06,
      "loss": 0.725,
      "step": 12408
    },
    {
      "epoch": 0.3575462456059471,
      "grad_norm": 0.7812665700912476,
      "learning_rate": 7.440907473813212e-06,
      "loss": 0.7215,
      "step": 12409
    },
    {
      "epoch": 0.3575750590675964,
      "grad_norm": 0.7948166131973267,
      "learning_rate": 7.440500232869262e-06,
      "loss": 0.7349,
      "step": 12410
    },
    {
      "epoch": 0.35760387252924564,
      "grad_norm": 0.7979850769042969,
      "learning_rate": 7.440092970670993e-06,
      "loss": 0.7475,
      "step": 12411
    },
    {
      "epoch": 0.3576326859908949,
      "grad_norm": 0.7619180083274841,
      "learning_rate": 7.439685687221956e-06,
      "loss": 0.7381,
      "step": 12412
    },
    {
      "epoch": 0.3576614994525442,
      "grad_norm": 0.797878623008728,
      "learning_rate": 7.439278382525694e-06,
      "loss": 0.7613,
      "step": 12413
    },
    {
      "epoch": 0.3576903129141935,
      "grad_norm": 0.7528181672096252,
      "learning_rate": 7.438871056585755e-06,
      "loss": 0.7522,
      "step": 12414
    },
    {
      "epoch": 0.3577191263758428,
      "grad_norm": 0.7874255776405334,
      "learning_rate": 7.438463709405688e-06,
      "loss": 0.7353,
      "step": 12415
    },
    {
      "epoch": 0.3577479398374921,
      "grad_norm": 0.8195502161979675,
      "learning_rate": 7.438056340989039e-06,
      "loss": 0.7515,
      "step": 12416
    },
    {
      "epoch": 0.3577767532991414,
      "grad_norm": 0.772638738155365,
      "learning_rate": 7.437648951339358e-06,
      "loss": 0.7453,
      "step": 12417
    },
    {
      "epoch": 0.3578055667607906,
      "grad_norm": 0.7838062644004822,
      "learning_rate": 7.43724154046019e-06,
      "loss": 0.7521,
      "step": 12418
    },
    {
      "epoch": 0.3578343802224399,
      "grad_norm": 0.7885550856590271,
      "learning_rate": 7.436834108355084e-06,
      "loss": 0.7162,
      "step": 12419
    },
    {
      "epoch": 0.3578631936840892,
      "grad_norm": 0.9338173866271973,
      "learning_rate": 7.436426655027589e-06,
      "loss": 0.7383,
      "step": 12420
    },
    {
      "epoch": 0.3578920071457385,
      "grad_norm": 0.7908353805541992,
      "learning_rate": 7.436019180481255e-06,
      "loss": 0.7423,
      "step": 12421
    },
    {
      "epoch": 0.3579208206073878,
      "grad_norm": 0.7741674780845642,
      "learning_rate": 7.435611684719628e-06,
      "loss": 0.7687,
      "step": 12422
    },
    {
      "epoch": 0.35794963406903707,
      "grad_norm": 0.7795880436897278,
      "learning_rate": 7.435204167746257e-06,
      "loss": 0.7506,
      "step": 12423
    },
    {
      "epoch": 0.35797844753068636,
      "grad_norm": 0.8220610618591309,
      "learning_rate": 7.434796629564693e-06,
      "loss": 0.7646,
      "step": 12424
    },
    {
      "epoch": 0.3580072609923356,
      "grad_norm": 0.7612732648849487,
      "learning_rate": 7.434389070178485e-06,
      "loss": 0.7369,
      "step": 12425
    },
    {
      "epoch": 0.3580360744539849,
      "grad_norm": 0.7570117712020874,
      "learning_rate": 7.433981489591181e-06,
      "loss": 0.7366,
      "step": 12426
    },
    {
      "epoch": 0.3580648879156342,
      "grad_norm": 0.7870311737060547,
      "learning_rate": 7.4335738878063314e-06,
      "loss": 0.7489,
      "step": 12427
    },
    {
      "epoch": 0.35809370137728347,
      "grad_norm": 0.8075084090232849,
      "learning_rate": 7.433166264827486e-06,
      "loss": 0.7401,
      "step": 12428
    },
    {
      "epoch": 0.35812251483893276,
      "grad_norm": 0.7523403763771057,
      "learning_rate": 7.432758620658195e-06,
      "loss": 0.7405,
      "step": 12429
    },
    {
      "epoch": 0.35815132830058205,
      "grad_norm": 0.7749588489532471,
      "learning_rate": 7.432350955302008e-06,
      "loss": 0.7282,
      "step": 12430
    },
    {
      "epoch": 0.35818014176223134,
      "grad_norm": 0.7980207800865173,
      "learning_rate": 7.431943268762475e-06,
      "loss": 0.7439,
      "step": 12431
    },
    {
      "epoch": 0.3582089552238806,
      "grad_norm": 0.7880523204803467,
      "learning_rate": 7.4315355610431475e-06,
      "loss": 0.758,
      "step": 12432
    },
    {
      "epoch": 0.35823776868552987,
      "grad_norm": 0.7964421510696411,
      "learning_rate": 7.431127832147575e-06,
      "loss": 0.7468,
      "step": 12433
    },
    {
      "epoch": 0.35826658214717916,
      "grad_norm": 0.7514085173606873,
      "learning_rate": 7.4307200820793116e-06,
      "loss": 0.775,
      "step": 12434
    },
    {
      "epoch": 0.35829539560882845,
      "grad_norm": 0.8096141815185547,
      "learning_rate": 7.430312310841905e-06,
      "loss": 0.7344,
      "step": 12435
    },
    {
      "epoch": 0.35832420907047774,
      "grad_norm": 0.7587864398956299,
      "learning_rate": 7.429904518438907e-06,
      "loss": 0.7437,
      "step": 12436
    },
    {
      "epoch": 0.35835302253212703,
      "grad_norm": 0.7549013495445251,
      "learning_rate": 7.42949670487387e-06,
      "loss": 0.732,
      "step": 12437
    },
    {
      "epoch": 0.35838183599377627,
      "grad_norm": 0.7576326131820679,
      "learning_rate": 7.429088870150345e-06,
      "loss": 0.7137,
      "step": 12438
    },
    {
      "epoch": 0.35841064945542556,
      "grad_norm": 0.7775775194168091,
      "learning_rate": 7.4286810142718845e-06,
      "loss": 0.7216,
      "step": 12439
    },
    {
      "epoch": 0.35843946291707485,
      "grad_norm": 0.7852426171302795,
      "learning_rate": 7.4282731372420405e-06,
      "loss": 0.7326,
      "step": 12440
    },
    {
      "epoch": 0.35846827637872414,
      "grad_norm": 0.7667224407196045,
      "learning_rate": 7.427865239064365e-06,
      "loss": 0.7541,
      "step": 12441
    },
    {
      "epoch": 0.35849708984037343,
      "grad_norm": 0.772609293460846,
      "learning_rate": 7.4274573197424095e-06,
      "loss": 0.7387,
      "step": 12442
    },
    {
      "epoch": 0.3585259033020227,
      "grad_norm": 0.7727214097976685,
      "learning_rate": 7.427049379279728e-06,
      "loss": 0.7233,
      "step": 12443
    },
    {
      "epoch": 0.358554716763672,
      "grad_norm": 0.7646970748901367,
      "learning_rate": 7.426641417679872e-06,
      "loss": 0.7171,
      "step": 12444
    },
    {
      "epoch": 0.35858353022532125,
      "grad_norm": 0.8007053732872009,
      "learning_rate": 7.426233434946395e-06,
      "loss": 0.7299,
      "step": 12445
    },
    {
      "epoch": 0.35861234368697054,
      "grad_norm": 0.7630802392959595,
      "learning_rate": 7.4258254310828506e-06,
      "loss": 0.7199,
      "step": 12446
    },
    {
      "epoch": 0.3586411571486198,
      "grad_norm": 0.7737856507301331,
      "learning_rate": 7.4254174060927915e-06,
      "loss": 0.733,
      "step": 12447
    },
    {
      "epoch": 0.3586699706102691,
      "grad_norm": 0.7752514481544495,
      "learning_rate": 7.425009359979771e-06,
      "loss": 0.7342,
      "step": 12448
    },
    {
      "epoch": 0.3586987840719184,
      "grad_norm": 0.7702576518058777,
      "learning_rate": 7.424601292747344e-06,
      "loss": 0.7542,
      "step": 12449
    },
    {
      "epoch": 0.3587275975335677,
      "grad_norm": 0.7852367162704468,
      "learning_rate": 7.424193204399061e-06,
      "loss": 0.7454,
      "step": 12450
    },
    {
      "epoch": 0.358756410995217,
      "grad_norm": 0.8138161897659302,
      "learning_rate": 7.4237850949384815e-06,
      "loss": 0.7588,
      "step": 12451
    },
    {
      "epoch": 0.3587852244568662,
      "grad_norm": 0.7323285937309265,
      "learning_rate": 7.423376964369155e-06,
      "loss": 0.7477,
      "step": 12452
    },
    {
      "epoch": 0.3588140379185155,
      "grad_norm": 0.8012943863868713,
      "learning_rate": 7.42296881269464e-06,
      "loss": 0.7354,
      "step": 12453
    },
    {
      "epoch": 0.3588428513801648,
      "grad_norm": 0.782137393951416,
      "learning_rate": 7.422560639918486e-06,
      "loss": 0.7327,
      "step": 12454
    },
    {
      "epoch": 0.3588716648418141,
      "grad_norm": 0.7717804312705994,
      "learning_rate": 7.422152446044252e-06,
      "loss": 0.7332,
      "step": 12455
    },
    {
      "epoch": 0.3589004783034634,
      "grad_norm": 0.7691040635108948,
      "learning_rate": 7.42174423107549e-06,
      "loss": 0.753,
      "step": 12456
    },
    {
      "epoch": 0.3589292917651127,
      "grad_norm": 0.9256364703178406,
      "learning_rate": 7.421335995015757e-06,
      "loss": 0.7637,
      "step": 12457
    },
    {
      "epoch": 0.35895810522676197,
      "grad_norm": 0.799561083316803,
      "learning_rate": 7.420927737868608e-06,
      "loss": 0.7546,
      "step": 12458
    },
    {
      "epoch": 0.3589869186884112,
      "grad_norm": 0.7653055787086487,
      "learning_rate": 7.420519459637599e-06,
      "loss": 0.7318,
      "step": 12459
    },
    {
      "epoch": 0.3590157321500605,
      "grad_norm": 0.753657341003418,
      "learning_rate": 7.420111160326286e-06,
      "loss": 0.7442,
      "step": 12460
    },
    {
      "epoch": 0.3590445456117098,
      "grad_norm": 0.7702918648719788,
      "learning_rate": 7.419702839938222e-06,
      "loss": 0.7503,
      "step": 12461
    },
    {
      "epoch": 0.3590733590733591,
      "grad_norm": 0.8062992691993713,
      "learning_rate": 7.4192944984769655e-06,
      "loss": 0.7621,
      "step": 12462
    },
    {
      "epoch": 0.35910217253500837,
      "grad_norm": 0.7386530041694641,
      "learning_rate": 7.418886135946073e-06,
      "loss": 0.7396,
      "step": 12463
    },
    {
      "epoch": 0.35913098599665766,
      "grad_norm": 0.7939289212226868,
      "learning_rate": 7.4184777523491e-06,
      "loss": 0.7405,
      "step": 12464
    },
    {
      "epoch": 0.3591597994583069,
      "grad_norm": 0.7563533186912537,
      "learning_rate": 7.4180693476896035e-06,
      "loss": 0.7407,
      "step": 12465
    },
    {
      "epoch": 0.3591886129199562,
      "grad_norm": 0.7678205966949463,
      "learning_rate": 7.417660921971141e-06,
      "loss": 0.7488,
      "step": 12466
    },
    {
      "epoch": 0.3592174263816055,
      "grad_norm": 0.7764995694160461,
      "learning_rate": 7.417252475197268e-06,
      "loss": 0.7519,
      "step": 12467
    },
    {
      "epoch": 0.35924623984325477,
      "grad_norm": 0.7775485515594482,
      "learning_rate": 7.416844007371541e-06,
      "loss": 0.7524,
      "step": 12468
    },
    {
      "epoch": 0.35927505330490406,
      "grad_norm": 0.7948571443557739,
      "learning_rate": 7.41643551849752e-06,
      "loss": 0.7645,
      "step": 12469
    },
    {
      "epoch": 0.35930386676655335,
      "grad_norm": 0.811663806438446,
      "learning_rate": 7.416027008578761e-06,
      "loss": 0.7497,
      "step": 12470
    },
    {
      "epoch": 0.35933268022820264,
      "grad_norm": 0.7807226181030273,
      "learning_rate": 7.4156184776188226e-06,
      "loss": 0.7407,
      "step": 12471
    },
    {
      "epoch": 0.3593614936898519,
      "grad_norm": 0.7955775856971741,
      "learning_rate": 7.415209925621261e-06,
      "loss": 0.7619,
      "step": 12472
    },
    {
      "epoch": 0.35939030715150116,
      "grad_norm": 0.7717743515968323,
      "learning_rate": 7.414801352589635e-06,
      "loss": 0.7304,
      "step": 12473
    },
    {
      "epoch": 0.35941912061315046,
      "grad_norm": 0.786790668964386,
      "learning_rate": 7.414392758527504e-06,
      "loss": 0.7267,
      "step": 12474
    },
    {
      "epoch": 0.35944793407479975,
      "grad_norm": 0.7806721925735474,
      "learning_rate": 7.413984143438425e-06,
      "loss": 0.7334,
      "step": 12475
    },
    {
      "epoch": 0.35947674753644904,
      "grad_norm": 0.7840075492858887,
      "learning_rate": 7.4135755073259584e-06,
      "loss": 0.7288,
      "step": 12476
    },
    {
      "epoch": 0.35950556099809833,
      "grad_norm": 0.7911686897277832,
      "learning_rate": 7.413166850193659e-06,
      "loss": 0.7368,
      "step": 12477
    },
    {
      "epoch": 0.3595343744597476,
      "grad_norm": 0.7502209544181824,
      "learning_rate": 7.412758172045092e-06,
      "loss": 0.7403,
      "step": 12478
    },
    {
      "epoch": 0.35956318792139685,
      "grad_norm": 0.7518311738967896,
      "learning_rate": 7.412349472883811e-06,
      "loss": 0.7183,
      "step": 12479
    },
    {
      "epoch": 0.35959200138304614,
      "grad_norm": 0.7835520505905151,
      "learning_rate": 7.411940752713378e-06,
      "loss": 0.7529,
      "step": 12480
    },
    {
      "epoch": 0.35962081484469544,
      "grad_norm": 0.7785561680793762,
      "learning_rate": 7.411532011537352e-06,
      "loss": 0.7598,
      "step": 12481
    },
    {
      "epoch": 0.3596496283063447,
      "grad_norm": 0.7839648723602295,
      "learning_rate": 7.411123249359294e-06,
      "loss": 0.7475,
      "step": 12482
    },
    {
      "epoch": 0.359678441767994,
      "grad_norm": 0.7516396641731262,
      "learning_rate": 7.410714466182762e-06,
      "loss": 0.7414,
      "step": 12483
    },
    {
      "epoch": 0.3597072552296433,
      "grad_norm": 0.7723531126976013,
      "learning_rate": 7.410305662011317e-06,
      "loss": 0.7407,
      "step": 12484
    },
    {
      "epoch": 0.3597360686912926,
      "grad_norm": 0.7601457238197327,
      "learning_rate": 7.4098968368485194e-06,
      "loss": 0.7285,
      "step": 12485
    },
    {
      "epoch": 0.35976488215294183,
      "grad_norm": 0.7749766707420349,
      "learning_rate": 7.409487990697928e-06,
      "loss": 0.7539,
      "step": 12486
    },
    {
      "epoch": 0.3597936956145911,
      "grad_norm": 0.7709092497825623,
      "learning_rate": 7.4090791235631055e-06,
      "loss": 0.7197,
      "step": 12487
    },
    {
      "epoch": 0.3598225090762404,
      "grad_norm": 0.7568311095237732,
      "learning_rate": 7.4086702354476125e-06,
      "loss": 0.7417,
      "step": 12488
    },
    {
      "epoch": 0.3598513225378897,
      "grad_norm": 0.7478304505348206,
      "learning_rate": 7.408261326355009e-06,
      "loss": 0.7291,
      "step": 12489
    },
    {
      "epoch": 0.359880135999539,
      "grad_norm": 0.765638530254364,
      "learning_rate": 7.407852396288857e-06,
      "loss": 0.7355,
      "step": 12490
    },
    {
      "epoch": 0.3599089494611883,
      "grad_norm": 0.7973690629005432,
      "learning_rate": 7.407443445252718e-06,
      "loss": 0.7343,
      "step": 12491
    },
    {
      "epoch": 0.3599377629228375,
      "grad_norm": 1.0367443561553955,
      "learning_rate": 7.407034473250152e-06,
      "loss": 0.7163,
      "step": 12492
    },
    {
      "epoch": 0.3599665763844868,
      "grad_norm": 0.7920636534690857,
      "learning_rate": 7.406625480284723e-06,
      "loss": 0.7398,
      "step": 12493
    },
    {
      "epoch": 0.3599953898461361,
      "grad_norm": 0.8121213316917419,
      "learning_rate": 7.406216466359991e-06,
      "loss": 0.7339,
      "step": 12494
    },
    {
      "epoch": 0.3600242033077854,
      "grad_norm": 0.746925950050354,
      "learning_rate": 7.40580743147952e-06,
      "loss": 0.7376,
      "step": 12495
    },
    {
      "epoch": 0.3600530167694347,
      "grad_norm": 0.7986735701560974,
      "learning_rate": 7.40539837564687e-06,
      "loss": 0.7135,
      "step": 12496
    },
    {
      "epoch": 0.360081830231084,
      "grad_norm": 0.8073192238807678,
      "learning_rate": 7.404989298865604e-06,
      "loss": 0.7501,
      "step": 12497
    },
    {
      "epoch": 0.36011064369273327,
      "grad_norm": 0.7739668488502502,
      "learning_rate": 7.404580201139287e-06,
      "loss": 0.7343,
      "step": 12498
    },
    {
      "epoch": 0.3601394571543825,
      "grad_norm": 0.8104111552238464,
      "learning_rate": 7.40417108247148e-06,
      "loss": 0.7347,
      "step": 12499
    },
    {
      "epoch": 0.3601682706160318,
      "grad_norm": 0.8266703486442566,
      "learning_rate": 7.403761942865746e-06,
      "loss": 0.7408,
      "step": 12500
    },
    {
      "epoch": 0.3601970840776811,
      "grad_norm": 0.7636909484863281,
      "learning_rate": 7.403352782325648e-06,
      "loss": 0.7316,
      "step": 12501
    },
    {
      "epoch": 0.3602258975393304,
      "grad_norm": 0.7726298570632935,
      "learning_rate": 7.40294360085475e-06,
      "loss": 0.7369,
      "step": 12502
    },
    {
      "epoch": 0.36025471100097967,
      "grad_norm": 0.7687903642654419,
      "learning_rate": 7.4025343984566155e-06,
      "loss": 0.7116,
      "step": 12503
    },
    {
      "epoch": 0.36028352446262896,
      "grad_norm": 0.809555172920227,
      "learning_rate": 7.402125175134806e-06,
      "loss": 0.7212,
      "step": 12504
    },
    {
      "epoch": 0.36031233792427825,
      "grad_norm": 0.8173814415931702,
      "learning_rate": 7.40171593089289e-06,
      "loss": 0.7332,
      "step": 12505
    },
    {
      "epoch": 0.3603411513859275,
      "grad_norm": 0.7971047163009644,
      "learning_rate": 7.401306665734429e-06,
      "loss": 0.7633,
      "step": 12506
    },
    {
      "epoch": 0.3603699648475768,
      "grad_norm": 0.7519551515579224,
      "learning_rate": 7.400897379662986e-06,
      "loss": 0.7167,
      "step": 12507
    },
    {
      "epoch": 0.36039877830922606,
      "grad_norm": 0.8080171346664429,
      "learning_rate": 7.400488072682128e-06,
      "loss": 0.7458,
      "step": 12508
    },
    {
      "epoch": 0.36042759177087536,
      "grad_norm": 0.8102464079856873,
      "learning_rate": 7.400078744795418e-06,
      "loss": 0.7446,
      "step": 12509
    },
    {
      "epoch": 0.36045640523252465,
      "grad_norm": 0.7790303826332092,
      "learning_rate": 7.399669396006421e-06,
      "loss": 0.7619,
      "step": 12510
    },
    {
      "epoch": 0.36048521869417394,
      "grad_norm": 0.7444841861724854,
      "learning_rate": 7.399260026318702e-06,
      "loss": 0.7235,
      "step": 12511
    },
    {
      "epoch": 0.3605140321558232,
      "grad_norm": 0.7974898815155029,
      "learning_rate": 7.398850635735828e-06,
      "loss": 0.7608,
      "step": 12512
    },
    {
      "epoch": 0.36054284561747246,
      "grad_norm": 0.7859013080596924,
      "learning_rate": 7.398441224261362e-06,
      "loss": 0.7366,
      "step": 12513
    },
    {
      "epoch": 0.36057165907912175,
      "grad_norm": 0.8471567034721375,
      "learning_rate": 7.398031791898872e-06,
      "loss": 0.7342,
      "step": 12514
    },
    {
      "epoch": 0.36060047254077104,
      "grad_norm": 0.8117562532424927,
      "learning_rate": 7.397622338651919e-06,
      "loss": 0.7345,
      "step": 12515
    },
    {
      "epoch": 0.36062928600242034,
      "grad_norm": 0.8226950168609619,
      "learning_rate": 7.397212864524075e-06,
      "loss": 0.751,
      "step": 12516
    },
    {
      "epoch": 0.3606580994640696,
      "grad_norm": 0.7967820763587952,
      "learning_rate": 7.396803369518902e-06,
      "loss": 0.7736,
      "step": 12517
    },
    {
      "epoch": 0.3606869129257189,
      "grad_norm": 0.7676152586936951,
      "learning_rate": 7.396393853639967e-06,
      "loss": 0.7267,
      "step": 12518
    },
    {
      "epoch": 0.36071572638736815,
      "grad_norm": 0.7493687272071838,
      "learning_rate": 7.395984316890838e-06,
      "loss": 0.7359,
      "step": 12519
    },
    {
      "epoch": 0.36074453984901744,
      "grad_norm": 0.8065453767776489,
      "learning_rate": 7.395574759275081e-06,
      "loss": 0.748,
      "step": 12520
    },
    {
      "epoch": 0.36077335331066673,
      "grad_norm": 0.8141460418701172,
      "learning_rate": 7.395165180796263e-06,
      "loss": 0.7343,
      "step": 12521
    },
    {
      "epoch": 0.360802166772316,
      "grad_norm": 0.8001622557640076,
      "learning_rate": 7.394755581457949e-06,
      "loss": 0.7424,
      "step": 12522
    },
    {
      "epoch": 0.3608309802339653,
      "grad_norm": 0.76043701171875,
      "learning_rate": 7.394345961263709e-06,
      "loss": 0.7404,
      "step": 12523
    },
    {
      "epoch": 0.3608597936956146,
      "grad_norm": 0.7580620050430298,
      "learning_rate": 7.393936320217109e-06,
      "loss": 0.7307,
      "step": 12524
    },
    {
      "epoch": 0.3608886071572639,
      "grad_norm": 0.7851080298423767,
      "learning_rate": 7.393526658321717e-06,
      "loss": 0.7264,
      "step": 12525
    },
    {
      "epoch": 0.36091742061891313,
      "grad_norm": 0.7922224998474121,
      "learning_rate": 7.393116975581099e-06,
      "loss": 0.7609,
      "step": 12526
    },
    {
      "epoch": 0.3609462340805624,
      "grad_norm": 0.7493351697921753,
      "learning_rate": 7.392707271998826e-06,
      "loss": 0.7407,
      "step": 12527
    },
    {
      "epoch": 0.3609750475422117,
      "grad_norm": 0.7363767027854919,
      "learning_rate": 7.392297547578463e-06,
      "loss": 0.72,
      "step": 12528
    },
    {
      "epoch": 0.361003861003861,
      "grad_norm": 0.7883221507072449,
      "learning_rate": 7.391887802323581e-06,
      "loss": 0.748,
      "step": 12529
    },
    {
      "epoch": 0.3610326744655103,
      "grad_norm": 0.749475359916687,
      "learning_rate": 7.391478036237747e-06,
      "loss": 0.7308,
      "step": 12530
    },
    {
      "epoch": 0.3610614879271596,
      "grad_norm": 0.7362086176872253,
      "learning_rate": 7.39106824932453e-06,
      "loss": 0.741,
      "step": 12531
    },
    {
      "epoch": 0.3610903013888089,
      "grad_norm": 0.775181770324707,
      "learning_rate": 7.390658441587498e-06,
      "loss": 0.7431,
      "step": 12532
    },
    {
      "epoch": 0.3611191148504581,
      "grad_norm": 0.7526872754096985,
      "learning_rate": 7.390248613030222e-06,
      "loss": 0.7263,
      "step": 12533
    },
    {
      "epoch": 0.3611479283121074,
      "grad_norm": 0.7497518658638,
      "learning_rate": 7.3898387636562685e-06,
      "loss": 0.7319,
      "step": 12534
    },
    {
      "epoch": 0.3611767417737567,
      "grad_norm": 0.7882496118545532,
      "learning_rate": 7.389428893469209e-06,
      "loss": 0.7419,
      "step": 12535
    },
    {
      "epoch": 0.361205555235406,
      "grad_norm": 0.7506888508796692,
      "learning_rate": 7.389019002472611e-06,
      "loss": 0.7198,
      "step": 12536
    },
    {
      "epoch": 0.3612343686970553,
      "grad_norm": 0.7595476508140564,
      "learning_rate": 7.388609090670048e-06,
      "loss": 0.7258,
      "step": 12537
    },
    {
      "epoch": 0.36126318215870457,
      "grad_norm": 0.8177686333656311,
      "learning_rate": 7.388199158065087e-06,
      "loss": 0.7538,
      "step": 12538
    },
    {
      "epoch": 0.36129199562035386,
      "grad_norm": 0.7725651860237122,
      "learning_rate": 7.387789204661297e-06,
      "loss": 0.7688,
      "step": 12539
    },
    {
      "epoch": 0.3613208090820031,
      "grad_norm": 0.7657021880149841,
      "learning_rate": 7.387379230462251e-06,
      "loss": 0.7294,
      "step": 12540
    },
    {
      "epoch": 0.3613496225436524,
      "grad_norm": 0.7750071287155151,
      "learning_rate": 7.38696923547152e-06,
      "loss": 0.756,
      "step": 12541
    },
    {
      "epoch": 0.3613784360053017,
      "grad_norm": 0.7725023627281189,
      "learning_rate": 7.3865592196926715e-06,
      "loss": 0.7228,
      "step": 12542
    },
    {
      "epoch": 0.36140724946695096,
      "grad_norm": 0.8349054455757141,
      "learning_rate": 7.386149183129279e-06,
      "loss": 0.7381,
      "step": 12543
    },
    {
      "epoch": 0.36143606292860025,
      "grad_norm": 0.7772387862205505,
      "learning_rate": 7.3857391257849105e-06,
      "loss": 0.7314,
      "step": 12544
    },
    {
      "epoch": 0.36146487639024955,
      "grad_norm": 0.8256910443305969,
      "learning_rate": 7.38532904766314e-06,
      "loss": 0.762,
      "step": 12545
    },
    {
      "epoch": 0.3614936898518988,
      "grad_norm": 0.7762360572814941,
      "learning_rate": 7.384918948767538e-06,
      "loss": 0.747,
      "step": 12546
    },
    {
      "epoch": 0.36152250331354807,
      "grad_norm": 0.7910135984420776,
      "learning_rate": 7.384508829101675e-06,
      "loss": 0.761,
      "step": 12547
    },
    {
      "epoch": 0.36155131677519736,
      "grad_norm": 0.7967802286148071,
      "learning_rate": 7.384098688669126e-06,
      "loss": 0.741,
      "step": 12548
    },
    {
      "epoch": 0.36158013023684665,
      "grad_norm": 0.7825901508331299,
      "learning_rate": 7.383688527473459e-06,
      "loss": 0.7605,
      "step": 12549
    },
    {
      "epoch": 0.36160894369849594,
      "grad_norm": 0.7733978629112244,
      "learning_rate": 7.3832783455182484e-06,
      "loss": 0.7372,
      "step": 12550
    },
    {
      "epoch": 0.36163775716014523,
      "grad_norm": 0.8271928429603577,
      "learning_rate": 7.382868142807066e-06,
      "loss": 0.7504,
      "step": 12551
    },
    {
      "epoch": 0.3616665706217945,
      "grad_norm": 0.8209702372550964,
      "learning_rate": 7.382457919343482e-06,
      "loss": 0.7317,
      "step": 12552
    },
    {
      "epoch": 0.36169538408344376,
      "grad_norm": 0.783750593662262,
      "learning_rate": 7.382047675131073e-06,
      "loss": 0.7303,
      "step": 12553
    },
    {
      "epoch": 0.36172419754509305,
      "grad_norm": 0.7664333581924438,
      "learning_rate": 7.38163741017341e-06,
      "loss": 0.7427,
      "step": 12554
    },
    {
      "epoch": 0.36175301100674234,
      "grad_norm": 0.7982199192047119,
      "learning_rate": 7.381227124474065e-06,
      "loss": 0.7322,
      "step": 12555
    },
    {
      "epoch": 0.36178182446839163,
      "grad_norm": 0.7868764400482178,
      "learning_rate": 7.380816818036613e-06,
      "loss": 0.7447,
      "step": 12556
    },
    {
      "epoch": 0.3618106379300409,
      "grad_norm": 0.7706965804100037,
      "learning_rate": 7.380406490864627e-06,
      "loss": 0.7363,
      "step": 12557
    },
    {
      "epoch": 0.3618394513916902,
      "grad_norm": 0.7707207202911377,
      "learning_rate": 7.3799961429616774e-06,
      "loss": 0.7286,
      "step": 12558
    },
    {
      "epoch": 0.3618682648533395,
      "grad_norm": 0.7940338850021362,
      "learning_rate": 7.379585774331342e-06,
      "loss": 0.7275,
      "step": 12559
    },
    {
      "epoch": 0.36189707831498874,
      "grad_norm": 0.7506827116012573,
      "learning_rate": 7.379175384977194e-06,
      "loss": 0.7441,
      "step": 12560
    },
    {
      "epoch": 0.36192589177663803,
      "grad_norm": 0.764042854309082,
      "learning_rate": 7.3787649749028064e-06,
      "loss": 0.7319,
      "step": 12561
    },
    {
      "epoch": 0.3619547052382873,
      "grad_norm": 0.7806438207626343,
      "learning_rate": 7.378354544111754e-06,
      "loss": 0.7498,
      "step": 12562
    },
    {
      "epoch": 0.3619835186999366,
      "grad_norm": 0.7904091477394104,
      "learning_rate": 7.377944092607611e-06,
      "loss": 0.7111,
      "step": 12563
    },
    {
      "epoch": 0.3620123321615859,
      "grad_norm": 0.762933611869812,
      "learning_rate": 7.3775336203939504e-06,
      "loss": 0.7548,
      "step": 12564
    },
    {
      "epoch": 0.3620411456232352,
      "grad_norm": 0.7725747227668762,
      "learning_rate": 7.37712312747435e-06,
      "loss": 0.7445,
      "step": 12565
    },
    {
      "epoch": 0.3620699590848845,
      "grad_norm": 0.7784585952758789,
      "learning_rate": 7.376712613852384e-06,
      "loss": 0.7336,
      "step": 12566
    },
    {
      "epoch": 0.3620987725465337,
      "grad_norm": 0.7518300414085388,
      "learning_rate": 7.3763020795316254e-06,
      "loss": 0.7396,
      "step": 12567
    },
    {
      "epoch": 0.362127586008183,
      "grad_norm": 0.760144829750061,
      "learning_rate": 7.375891524515653e-06,
      "loss": 0.7394,
      "step": 12568
    },
    {
      "epoch": 0.3621563994698323,
      "grad_norm": 0.7891154289245605,
      "learning_rate": 7.375480948808038e-06,
      "loss": 0.7428,
      "step": 12569
    },
    {
      "epoch": 0.3621852129314816,
      "grad_norm": 0.769862711429596,
      "learning_rate": 7.37507035241236e-06,
      "loss": 0.7661,
      "step": 12570
    },
    {
      "epoch": 0.3622140263931309,
      "grad_norm": 0.7655613422393799,
      "learning_rate": 7.374659735332194e-06,
      "loss": 0.7394,
      "step": 12571
    },
    {
      "epoch": 0.3622428398547802,
      "grad_norm": 0.7730535268783569,
      "learning_rate": 7.374249097571115e-06,
      "loss": 0.7477,
      "step": 12572
    },
    {
      "epoch": 0.3622716533164294,
      "grad_norm": 0.7616569399833679,
      "learning_rate": 7.3738384391326985e-06,
      "loss": 0.7387,
      "step": 12573
    },
    {
      "epoch": 0.3623004667780787,
      "grad_norm": 0.7594241499900818,
      "learning_rate": 7.373427760020523e-06,
      "loss": 0.7309,
      "step": 12574
    },
    {
      "epoch": 0.362329280239728,
      "grad_norm": 0.7696453332901001,
      "learning_rate": 7.3730170602381645e-06,
      "loss": 0.7358,
      "step": 12575
    },
    {
      "epoch": 0.3623580937013773,
      "grad_norm": 0.7723779082298279,
      "learning_rate": 7.3726063397892e-06,
      "loss": 0.7277,
      "step": 12576
    },
    {
      "epoch": 0.3623869071630266,
      "grad_norm": 0.7839044332504272,
      "learning_rate": 7.372195598677205e-06,
      "loss": 0.7489,
      "step": 12577
    },
    {
      "epoch": 0.36241572062467586,
      "grad_norm": 0.7573975920677185,
      "learning_rate": 7.371784836905758e-06,
      "loss": 0.7557,
      "step": 12578
    },
    {
      "epoch": 0.36244453408632515,
      "grad_norm": 3.5323359966278076,
      "learning_rate": 7.371374054478436e-06,
      "loss": 0.8234,
      "step": 12579
    },
    {
      "epoch": 0.3624733475479744,
      "grad_norm": 0.7539454698562622,
      "learning_rate": 7.370963251398817e-06,
      "loss": 0.7366,
      "step": 12580
    },
    {
      "epoch": 0.3625021610096237,
      "grad_norm": 0.7827955484390259,
      "learning_rate": 7.370552427670479e-06,
      "loss": 0.7328,
      "step": 12581
    },
    {
      "epoch": 0.36253097447127297,
      "grad_norm": 0.7538296580314636,
      "learning_rate": 7.3701415832969966e-06,
      "loss": 0.7423,
      "step": 12582
    },
    {
      "epoch": 0.36255978793292226,
      "grad_norm": 0.8521518111228943,
      "learning_rate": 7.3697307182819525e-06,
      "loss": 0.7861,
      "step": 12583
    },
    {
      "epoch": 0.36258860139457155,
      "grad_norm": 0.7823631763458252,
      "learning_rate": 7.369319832628921e-06,
      "loss": 0.7396,
      "step": 12584
    },
    {
      "epoch": 0.36261741485622084,
      "grad_norm": 0.7498408555984497,
      "learning_rate": 7.368908926341483e-06,
      "loss": 0.7159,
      "step": 12585
    },
    {
      "epoch": 0.36264622831787013,
      "grad_norm": 0.790838360786438,
      "learning_rate": 7.3684979994232165e-06,
      "loss": 0.722,
      "step": 12586
    },
    {
      "epoch": 0.36267504177951937,
      "grad_norm": 0.8165919780731201,
      "learning_rate": 7.3680870518777e-06,
      "loss": 0.74,
      "step": 12587
    },
    {
      "epoch": 0.36270385524116866,
      "grad_norm": 0.8086488246917725,
      "learning_rate": 7.367676083708513e-06,
      "loss": 0.7623,
      "step": 12588
    },
    {
      "epoch": 0.36273266870281795,
      "grad_norm": 0.782007098197937,
      "learning_rate": 7.367265094919233e-06,
      "loss": 0.7545,
      "step": 12589
    },
    {
      "epoch": 0.36276148216446724,
      "grad_norm": 0.7807911038398743,
      "learning_rate": 7.366854085513441e-06,
      "loss": 0.7427,
      "step": 12590
    },
    {
      "epoch": 0.36279029562611653,
      "grad_norm": 0.7722881436347961,
      "learning_rate": 7.366443055494717e-06,
      "loss": 0.7409,
      "step": 12591
    },
    {
      "epoch": 0.3628191090877658,
      "grad_norm": 0.7494285702705383,
      "learning_rate": 7.3660320048666374e-06,
      "loss": 0.7084,
      "step": 12592
    },
    {
      "epoch": 0.3628479225494151,
      "grad_norm": 0.7460600733757019,
      "learning_rate": 7.365620933632784e-06,
      "loss": 0.7312,
      "step": 12593
    },
    {
      "epoch": 0.36287673601106435,
      "grad_norm": 0.7710347771644592,
      "learning_rate": 7.3652098417967385e-06,
      "loss": 0.7411,
      "step": 12594
    },
    {
      "epoch": 0.36290554947271364,
      "grad_norm": 0.7876501083374023,
      "learning_rate": 7.36479872936208e-06,
      "loss": 0.7535,
      "step": 12595
    },
    {
      "epoch": 0.36293436293436293,
      "grad_norm": 0.766454815864563,
      "learning_rate": 7.364387596332387e-06,
      "loss": 0.7341,
      "step": 12596
    },
    {
      "epoch": 0.3629631763960122,
      "grad_norm": 0.8394845128059387,
      "learning_rate": 7.3639764427112426e-06,
      "loss": 0.7443,
      "step": 12597
    },
    {
      "epoch": 0.3629919898576615,
      "grad_norm": 0.7550085186958313,
      "learning_rate": 7.363565268502226e-06,
      "loss": 0.7257,
      "step": 12598
    },
    {
      "epoch": 0.3630208033193108,
      "grad_norm": 0.7651243209838867,
      "learning_rate": 7.363154073708919e-06,
      "loss": 0.7262,
      "step": 12599
    },
    {
      "epoch": 0.36304961678096004,
      "grad_norm": 0.7608970403671265,
      "learning_rate": 7.362742858334902e-06,
      "loss": 0.7368,
      "step": 12600
    },
    {
      "epoch": 0.36307843024260933,
      "grad_norm": 0.7958541512489319,
      "learning_rate": 7.362331622383757e-06,
      "loss": 0.7251,
      "step": 12601
    },
    {
      "epoch": 0.3631072437042586,
      "grad_norm": 0.7904861569404602,
      "learning_rate": 7.361920365859066e-06,
      "loss": 0.7399,
      "step": 12602
    },
    {
      "epoch": 0.3631360571659079,
      "grad_norm": 1.0005412101745605,
      "learning_rate": 7.361509088764407e-06,
      "loss": 0.7093,
      "step": 12603
    },
    {
      "epoch": 0.3631648706275572,
      "grad_norm": 2.901963710784912,
      "learning_rate": 7.361097791103367e-06,
      "loss": 0.7496,
      "step": 12604
    },
    {
      "epoch": 0.3631936840892065,
      "grad_norm": 0.8360681533813477,
      "learning_rate": 7.360686472879525e-06,
      "loss": 0.7458,
      "step": 12605
    },
    {
      "epoch": 0.3632224975508558,
      "grad_norm": 0.7584874629974365,
      "learning_rate": 7.360275134096462e-06,
      "loss": 0.7364,
      "step": 12606
    },
    {
      "epoch": 0.363251311012505,
      "grad_norm": 0.8363709449768066,
      "learning_rate": 7.359863774757763e-06,
      "loss": 0.7445,
      "step": 12607
    },
    {
      "epoch": 0.3632801244741543,
      "grad_norm": 0.8307927250862122,
      "learning_rate": 7.35945239486701e-06,
      "loss": 0.7531,
      "step": 12608
    },
    {
      "epoch": 0.3633089379358036,
      "grad_norm": 0.8141446113586426,
      "learning_rate": 7.359040994427784e-06,
      "loss": 0.7433,
      "step": 12609
    },
    {
      "epoch": 0.3633377513974529,
      "grad_norm": 0.7749306559562683,
      "learning_rate": 7.3586295734436705e-06,
      "loss": 0.7532,
      "step": 12610
    },
    {
      "epoch": 0.3633665648591022,
      "grad_norm": 0.7958052158355713,
      "learning_rate": 7.358218131918251e-06,
      "loss": 0.7337,
      "step": 12611
    },
    {
      "epoch": 0.36339537832075147,
      "grad_norm": 0.7850409746170044,
      "learning_rate": 7.357806669855107e-06,
      "loss": 0.7532,
      "step": 12612
    },
    {
      "epoch": 0.36342419178240076,
      "grad_norm": 0.7906456589698792,
      "learning_rate": 7.357395187257826e-06,
      "loss": 0.7312,
      "step": 12613
    },
    {
      "epoch": 0.36345300524405,
      "grad_norm": 0.7908875346183777,
      "learning_rate": 7.3569836841299905e-06,
      "loss": 0.7533,
      "step": 12614
    },
    {
      "epoch": 0.3634818187056993,
      "grad_norm": 0.8022201061248779,
      "learning_rate": 7.356572160475179e-06,
      "loss": 0.755,
      "step": 12615
    },
    {
      "epoch": 0.3635106321673486,
      "grad_norm": 0.7886639833450317,
      "learning_rate": 7.3561606162969836e-06,
      "loss": 0.7474,
      "step": 12616
    },
    {
      "epoch": 0.36353944562899787,
      "grad_norm": 0.8070919513702393,
      "learning_rate": 7.355749051598982e-06,
      "loss": 0.7406,
      "step": 12617
    },
    {
      "epoch": 0.36356825909064716,
      "grad_norm": 0.7993503212928772,
      "learning_rate": 7.355337466384762e-06,
      "loss": 0.7459,
      "step": 12618
    },
    {
      "epoch": 0.36359707255229645,
      "grad_norm": 0.8039829730987549,
      "learning_rate": 7.354925860657909e-06,
      "loss": 0.7422,
      "step": 12619
    },
    {
      "epoch": 0.36362588601394574,
      "grad_norm": 0.8067542910575867,
      "learning_rate": 7.354514234422003e-06,
      "loss": 0.7526,
      "step": 12620
    },
    {
      "epoch": 0.363654699475595,
      "grad_norm": 0.8027722239494324,
      "learning_rate": 7.354102587680634e-06,
      "loss": 0.7624,
      "step": 12621
    },
    {
      "epoch": 0.36368351293724427,
      "grad_norm": 0.7842795848846436,
      "learning_rate": 7.353690920437384e-06,
      "loss": 0.7577,
      "step": 12622
    },
    {
      "epoch": 0.36371232639889356,
      "grad_norm": 0.7577104568481445,
      "learning_rate": 7.353279232695837e-06,
      "loss": 0.7118,
      "step": 12623
    },
    {
      "epoch": 0.36374113986054285,
      "grad_norm": 0.7732694149017334,
      "learning_rate": 7.352867524459581e-06,
      "loss": 0.7359,
      "step": 12624
    },
    {
      "epoch": 0.36376995332219214,
      "grad_norm": 0.7910149693489075,
      "learning_rate": 7.3524557957322026e-06,
      "loss": 0.7462,
      "step": 12625
    },
    {
      "epoch": 0.36379876678384143,
      "grad_norm": 0.7718191742897034,
      "learning_rate": 7.352044046517286e-06,
      "loss": 0.7352,
      "step": 12626
    },
    {
      "epoch": 0.36382758024549067,
      "grad_norm": 0.7717542052268982,
      "learning_rate": 7.351632276818416e-06,
      "loss": 0.7536,
      "step": 12627
    },
    {
      "epoch": 0.36385639370713996,
      "grad_norm": 0.8049107789993286,
      "learning_rate": 7.351220486639179e-06,
      "loss": 0.7871,
      "step": 12628
    },
    {
      "epoch": 0.36388520716878925,
      "grad_norm": 0.7746100425720215,
      "learning_rate": 7.350808675983162e-06,
      "loss": 0.7642,
      "step": 12629
    },
    {
      "epoch": 0.36391402063043854,
      "grad_norm": 0.798620343208313,
      "learning_rate": 7.350396844853952e-06,
      "loss": 0.7252,
      "step": 12630
    },
    {
      "epoch": 0.36394283409208783,
      "grad_norm": 0.8036872148513794,
      "learning_rate": 7.349984993255135e-06,
      "loss": 0.7759,
      "step": 12631
    },
    {
      "epoch": 0.3639716475537371,
      "grad_norm": 0.7656122446060181,
      "learning_rate": 7.349573121190298e-06,
      "loss": 0.7153,
      "step": 12632
    },
    {
      "epoch": 0.3640004610153864,
      "grad_norm": 0.7873131036758423,
      "learning_rate": 7.349161228663026e-06,
      "loss": 0.7327,
      "step": 12633
    },
    {
      "epoch": 0.36402927447703565,
      "grad_norm": 0.7914145588874817,
      "learning_rate": 7.34874931567691e-06,
      "loss": 0.7222,
      "step": 12634
    },
    {
      "epoch": 0.36405808793868494,
      "grad_norm": 0.7732411623001099,
      "learning_rate": 7.348337382235535e-06,
      "loss": 0.7243,
      "step": 12635
    },
    {
      "epoch": 0.36408690140033423,
      "grad_norm": 0.8372952342033386,
      "learning_rate": 7.347925428342488e-06,
      "loss": 0.7448,
      "step": 12636
    },
    {
      "epoch": 0.3641157148619835,
      "grad_norm": 0.815606951713562,
      "learning_rate": 7.347513454001358e-06,
      "loss": 0.7359,
      "step": 12637
    },
    {
      "epoch": 0.3641445283236328,
      "grad_norm": 0.8282836675643921,
      "learning_rate": 7.347101459215733e-06,
      "loss": 0.7627,
      "step": 12638
    },
    {
      "epoch": 0.3641733417852821,
      "grad_norm": 0.8350638747215271,
      "learning_rate": 7.346689443989201e-06,
      "loss": 0.7666,
      "step": 12639
    },
    {
      "epoch": 0.3642021552469314,
      "grad_norm": 0.7945520877838135,
      "learning_rate": 7.346277408325348e-06,
      "loss": 0.7363,
      "step": 12640
    },
    {
      "epoch": 0.3642309687085806,
      "grad_norm": 0.7952410578727722,
      "learning_rate": 7.345865352227765e-06,
      "loss": 0.7369,
      "step": 12641
    },
    {
      "epoch": 0.3642597821702299,
      "grad_norm": 0.8500712513923645,
      "learning_rate": 7.345453275700041e-06,
      "loss": 0.7392,
      "step": 12642
    },
    {
      "epoch": 0.3642885956318792,
      "grad_norm": 0.7846609950065613,
      "learning_rate": 7.345041178745762e-06,
      "loss": 0.7515,
      "step": 12643
    },
    {
      "epoch": 0.3643174090935285,
      "grad_norm": 0.7631268501281738,
      "learning_rate": 7.34462906136852e-06,
      "loss": 0.746,
      "step": 12644
    },
    {
      "epoch": 0.3643462225551778,
      "grad_norm": 1.9076370000839233,
      "learning_rate": 7.344216923571902e-06,
      "loss": 0.771,
      "step": 12645
    },
    {
      "epoch": 0.3643750360168271,
      "grad_norm": 0.8136571049690247,
      "learning_rate": 7.343804765359497e-06,
      "loss": 0.7515,
      "step": 12646
    },
    {
      "epoch": 0.36440384947847637,
      "grad_norm": 0.8549298048019409,
      "learning_rate": 7.343392586734896e-06,
      "loss": 0.7528,
      "step": 12647
    },
    {
      "epoch": 0.3644326629401256,
      "grad_norm": 0.7519803643226624,
      "learning_rate": 7.342980387701688e-06,
      "loss": 0.7352,
      "step": 12648
    },
    {
      "epoch": 0.3644614764017749,
      "grad_norm": 0.7556874752044678,
      "learning_rate": 7.342568168263464e-06,
      "loss": 0.7419,
      "step": 12649
    },
    {
      "epoch": 0.3644902898634242,
      "grad_norm": 0.7568552494049072,
      "learning_rate": 7.342155928423812e-06,
      "loss": 0.7472,
      "step": 12650
    },
    {
      "epoch": 0.3645191033250735,
      "grad_norm": 0.7606648802757263,
      "learning_rate": 7.341743668186323e-06,
      "loss": 0.735,
      "step": 12651
    },
    {
      "epoch": 0.36454791678672277,
      "grad_norm": 0.7943456768989563,
      "learning_rate": 7.341331387554588e-06,
      "loss": 0.7158,
      "step": 12652
    },
    {
      "epoch": 0.36457673024837206,
      "grad_norm": 0.776254415512085,
      "learning_rate": 7.340919086532197e-06,
      "loss": 0.7488,
      "step": 12653
    },
    {
      "epoch": 0.3646055437100213,
      "grad_norm": 0.7704287767410278,
      "learning_rate": 7.340506765122741e-06,
      "loss": 0.7364,
      "step": 12654
    },
    {
      "epoch": 0.3646343571716706,
      "grad_norm": 0.7743701934814453,
      "learning_rate": 7.340094423329808e-06,
      "loss": 0.7037,
      "step": 12655
    },
    {
      "epoch": 0.3646631706333199,
      "grad_norm": 0.8074947595596313,
      "learning_rate": 7.339682061156996e-06,
      "loss": 0.7452,
      "step": 12656
    },
    {
      "epoch": 0.36469198409496917,
      "grad_norm": 0.7588031888008118,
      "learning_rate": 7.33926967860789e-06,
      "loss": 0.7459,
      "step": 12657
    },
    {
      "epoch": 0.36472079755661846,
      "grad_norm": 0.77749103307724,
      "learning_rate": 7.338857275686084e-06,
      "loss": 0.7493,
      "step": 12658
    },
    {
      "epoch": 0.36474961101826775,
      "grad_norm": 0.7853993773460388,
      "learning_rate": 7.338444852395169e-06,
      "loss": 0.7389,
      "step": 12659
    },
    {
      "epoch": 0.36477842447991704,
      "grad_norm": 0.8028379678726196,
      "learning_rate": 7.338032408738737e-06,
      "loss": 0.7372,
      "step": 12660
    },
    {
      "epoch": 0.3648072379415663,
      "grad_norm": 0.7642970681190491,
      "learning_rate": 7.33761994472038e-06,
      "loss": 0.7336,
      "step": 12661
    },
    {
      "epoch": 0.36483605140321557,
      "grad_norm": 0.7508773803710938,
      "learning_rate": 7.33720746034369e-06,
      "loss": 0.724,
      "step": 12662
    },
    {
      "epoch": 0.36486486486486486,
      "grad_norm": 1.0723037719726562,
      "learning_rate": 7.336794955612259e-06,
      "loss": 0.7238,
      "step": 12663
    },
    {
      "epoch": 0.36489367832651415,
      "grad_norm": 0.7825019359588623,
      "learning_rate": 7.336382430529681e-06,
      "loss": 0.747,
      "step": 12664
    },
    {
      "epoch": 0.36492249178816344,
      "grad_norm": 0.7805911898612976,
      "learning_rate": 7.3359698850995455e-06,
      "loss": 0.7646,
      "step": 12665
    },
    {
      "epoch": 0.36495130524981273,
      "grad_norm": 0.777877688407898,
      "learning_rate": 7.335557319325449e-06,
      "loss": 0.7369,
      "step": 12666
    },
    {
      "epoch": 0.364980118711462,
      "grad_norm": 0.8403513431549072,
      "learning_rate": 7.335144733210983e-06,
      "loss": 0.7475,
      "step": 12667
    },
    {
      "epoch": 0.36500893217311126,
      "grad_norm": 0.8093098402023315,
      "learning_rate": 7.33473212675974e-06,
      "loss": 0.7627,
      "step": 12668
    },
    {
      "epoch": 0.36503774563476055,
      "grad_norm": 0.7345147728919983,
      "learning_rate": 7.334319499975314e-06,
      "loss": 0.7233,
      "step": 12669
    },
    {
      "epoch": 0.36506655909640984,
      "grad_norm": 0.7905108332633972,
      "learning_rate": 7.3339068528613e-06,
      "loss": 0.7425,
      "step": 12670
    },
    {
      "epoch": 0.36509537255805913,
      "grad_norm": 0.8003455996513367,
      "learning_rate": 7.3334941854212885e-06,
      "loss": 0.7229,
      "step": 12671
    },
    {
      "epoch": 0.3651241860197084,
      "grad_norm": 0.8014897108078003,
      "learning_rate": 7.333081497658876e-06,
      "loss": 0.7251,
      "step": 12672
    },
    {
      "epoch": 0.3651529994813577,
      "grad_norm": 0.7840182781219482,
      "learning_rate": 7.3326687895776564e-06,
      "loss": 0.728,
      "step": 12673
    },
    {
      "epoch": 0.365181812943007,
      "grad_norm": 0.810486376285553,
      "learning_rate": 7.3322560611812225e-06,
      "loss": 0.7403,
      "step": 12674
    },
    {
      "epoch": 0.36521062640465624,
      "grad_norm": 0.778663158416748,
      "learning_rate": 7.331843312473171e-06,
      "loss": 0.7227,
      "step": 12675
    },
    {
      "epoch": 0.3652394398663055,
      "grad_norm": 0.7843407988548279,
      "learning_rate": 7.331430543457093e-06,
      "loss": 0.7529,
      "step": 12676
    },
    {
      "epoch": 0.3652682533279548,
      "grad_norm": 0.7560120820999146,
      "learning_rate": 7.331017754136587e-06,
      "loss": 0.7162,
      "step": 12677
    },
    {
      "epoch": 0.3652970667896041,
      "grad_norm": 0.8233844637870789,
      "learning_rate": 7.330604944515246e-06,
      "loss": 0.7331,
      "step": 12678
    },
    {
      "epoch": 0.3653258802512534,
      "grad_norm": 0.7705426812171936,
      "learning_rate": 7.330192114596667e-06,
      "loss": 0.7241,
      "step": 12679
    },
    {
      "epoch": 0.3653546937129027,
      "grad_norm": 0.8630101084709167,
      "learning_rate": 7.3297792643844435e-06,
      "loss": 0.7679,
      "step": 12680
    },
    {
      "epoch": 0.3653835071745519,
      "grad_norm": 0.8062373399734497,
      "learning_rate": 7.329366393882171e-06,
      "loss": 0.7469,
      "step": 12681
    },
    {
      "epoch": 0.3654123206362012,
      "grad_norm": 0.7829160094261169,
      "learning_rate": 7.3289535030934456e-06,
      "loss": 0.7182,
      "step": 12682
    },
    {
      "epoch": 0.3654411340978505,
      "grad_norm": 0.7840810418128967,
      "learning_rate": 7.328540592021864e-06,
      "loss": 0.7264,
      "step": 12683
    },
    {
      "epoch": 0.3654699475594998,
      "grad_norm": 0.7735156416893005,
      "learning_rate": 7.3281276606710205e-06,
      "loss": 0.7573,
      "step": 12684
    },
    {
      "epoch": 0.3654987610211491,
      "grad_norm": 0.7536057829856873,
      "learning_rate": 7.327714709044513e-06,
      "loss": 0.73,
      "step": 12685
    },
    {
      "epoch": 0.3655275744827984,
      "grad_norm": 0.7683953046798706,
      "learning_rate": 7.327301737145938e-06,
      "loss": 0.722,
      "step": 12686
    },
    {
      "epoch": 0.36555638794444767,
      "grad_norm": 0.7706003785133362,
      "learning_rate": 7.3268887449788905e-06,
      "loss": 0.7346,
      "step": 12687
    },
    {
      "epoch": 0.3655852014060969,
      "grad_norm": 0.7834705710411072,
      "learning_rate": 7.3264757325469695e-06,
      "loss": 0.7475,
      "step": 12688
    },
    {
      "epoch": 0.3656140148677462,
      "grad_norm": 0.7271424531936646,
      "learning_rate": 7.326062699853767e-06,
      "loss": 0.6985,
      "step": 12689
    },
    {
      "epoch": 0.3656428283293955,
      "grad_norm": 0.7500882744789124,
      "learning_rate": 7.325649646902887e-06,
      "loss": 0.748,
      "step": 12690
    },
    {
      "epoch": 0.3656716417910448,
      "grad_norm": 0.7516299486160278,
      "learning_rate": 7.325236573697922e-06,
      "loss": 0.7345,
      "step": 12691
    },
    {
      "epoch": 0.36570045525269407,
      "grad_norm": 0.7683053016662598,
      "learning_rate": 7.324823480242471e-06,
      "loss": 0.7388,
      "step": 12692
    },
    {
      "epoch": 0.36572926871434336,
      "grad_norm": 0.772921621799469,
      "learning_rate": 7.324410366540131e-06,
      "loss": 0.7669,
      "step": 12693
    },
    {
      "epoch": 0.36575808217599265,
      "grad_norm": 0.8023581504821777,
      "learning_rate": 7.323997232594501e-06,
      "loss": 0.7576,
      "step": 12694
    },
    {
      "epoch": 0.3657868956376419,
      "grad_norm": 0.7887827157974243,
      "learning_rate": 7.323584078409178e-06,
      "loss": 0.7427,
      "step": 12695
    },
    {
      "epoch": 0.3658157090992912,
      "grad_norm": 0.7783074975013733,
      "learning_rate": 7.3231709039877595e-06,
      "loss": 0.7646,
      "step": 12696
    },
    {
      "epoch": 0.36584452256094047,
      "grad_norm": 0.7914129495620728,
      "learning_rate": 7.3227577093338456e-06,
      "loss": 0.7328,
      "step": 12697
    },
    {
      "epoch": 0.36587333602258976,
      "grad_norm": 0.774924099445343,
      "learning_rate": 7.322344494451035e-06,
      "loss": 0.7485,
      "step": 12698
    },
    {
      "epoch": 0.36590214948423905,
      "grad_norm": 0.753000020980835,
      "learning_rate": 7.321931259342924e-06,
      "loss": 0.7204,
      "step": 12699
    },
    {
      "epoch": 0.36593096294588834,
      "grad_norm": 0.7827401757240295,
      "learning_rate": 7.321518004013113e-06,
      "loss": 0.7294,
      "step": 12700
    },
    {
      "epoch": 0.36595977640753763,
      "grad_norm": 0.7614761590957642,
      "learning_rate": 7.3211047284652e-06,
      "loss": 0.7268,
      "step": 12701
    },
    {
      "epoch": 0.36598858986918686,
      "grad_norm": 0.810624897480011,
      "learning_rate": 7.320691432702787e-06,
      "loss": 0.7269,
      "step": 12702
    },
    {
      "epoch": 0.36601740333083616,
      "grad_norm": 0.7850342392921448,
      "learning_rate": 7.32027811672947e-06,
      "loss": 0.7233,
      "step": 12703
    },
    {
      "epoch": 0.36604621679248545,
      "grad_norm": 0.7929987907409668,
      "learning_rate": 7.319864780548851e-06,
      "loss": 0.7108,
      "step": 12704
    },
    {
      "epoch": 0.36607503025413474,
      "grad_norm": 0.794999897480011,
      "learning_rate": 7.3194514241645285e-06,
      "loss": 0.751,
      "step": 12705
    },
    {
      "epoch": 0.366103843715784,
      "grad_norm": 0.7688424587249756,
      "learning_rate": 7.3190380475801026e-06,
      "loss": 0.7475,
      "step": 12706
    },
    {
      "epoch": 0.3661326571774333,
      "grad_norm": 0.8142318725585938,
      "learning_rate": 7.318624650799173e-06,
      "loss": 0.7575,
      "step": 12707
    },
    {
      "epoch": 0.36616147063908255,
      "grad_norm": 0.8109259605407715,
      "learning_rate": 7.3182112338253405e-06,
      "loss": 0.7533,
      "step": 12708
    },
    {
      "epoch": 0.36619028410073184,
      "grad_norm": 0.7670457363128662,
      "learning_rate": 7.3177977966622075e-06,
      "loss": 0.7477,
      "step": 12709
    },
    {
      "epoch": 0.36621909756238114,
      "grad_norm": 0.7666015625,
      "learning_rate": 7.3173843393133705e-06,
      "loss": 0.7462,
      "step": 12710
    },
    {
      "epoch": 0.3662479110240304,
      "grad_norm": 0.7674766182899475,
      "learning_rate": 7.316970861782433e-06,
      "loss": 0.7294,
      "step": 12711
    },
    {
      "epoch": 0.3662767244856797,
      "grad_norm": 0.7748401761054993,
      "learning_rate": 7.316557364072995e-06,
      "loss": 0.7269,
      "step": 12712
    },
    {
      "epoch": 0.366305537947329,
      "grad_norm": 0.7889994382858276,
      "learning_rate": 7.316143846188658e-06,
      "loss": 0.7461,
      "step": 12713
    },
    {
      "epoch": 0.3663343514089783,
      "grad_norm": 0.7613949775695801,
      "learning_rate": 7.3157303081330235e-06,
      "loss": 0.7556,
      "step": 12714
    },
    {
      "epoch": 0.36636316487062753,
      "grad_norm": 0.7625181078910828,
      "learning_rate": 7.315316749909694e-06,
      "loss": 0.7406,
      "step": 12715
    },
    {
      "epoch": 0.3663919783322768,
      "grad_norm": 0.7823655605316162,
      "learning_rate": 7.3149031715222685e-06,
      "loss": 0.719,
      "step": 12716
    },
    {
      "epoch": 0.3664207917939261,
      "grad_norm": 0.7728776335716248,
      "learning_rate": 7.314489572974351e-06,
      "loss": 0.7367,
      "step": 12717
    },
    {
      "epoch": 0.3664496052555754,
      "grad_norm": 0.7604851722717285,
      "learning_rate": 7.3140759542695435e-06,
      "loss": 0.7307,
      "step": 12718
    },
    {
      "epoch": 0.3664784187172247,
      "grad_norm": 0.7608423829078674,
      "learning_rate": 7.313662315411447e-06,
      "loss": 0.7543,
      "step": 12719
    },
    {
      "epoch": 0.366507232178874,
      "grad_norm": 0.7724249958992004,
      "learning_rate": 7.3132486564036654e-06,
      "loss": 0.7653,
      "step": 12720
    },
    {
      "epoch": 0.3665360456405233,
      "grad_norm": 0.7562296390533447,
      "learning_rate": 7.312834977249801e-06,
      "loss": 0.7441,
      "step": 12721
    },
    {
      "epoch": 0.3665648591021725,
      "grad_norm": 0.7384458780288696,
      "learning_rate": 7.312421277953455e-06,
      "loss": 0.6936,
      "step": 12722
    },
    {
      "epoch": 0.3665936725638218,
      "grad_norm": 0.769664466381073,
      "learning_rate": 7.31200755851823e-06,
      "loss": 0.7447,
      "step": 12723
    },
    {
      "epoch": 0.3666224860254711,
      "grad_norm": 0.7971028685569763,
      "learning_rate": 7.311593818947733e-06,
      "loss": 0.7546,
      "step": 12724
    },
    {
      "epoch": 0.3666512994871204,
      "grad_norm": 0.7790527939796448,
      "learning_rate": 7.311180059245562e-06,
      "loss": 0.7491,
      "step": 12725
    },
    {
      "epoch": 0.3666801129487697,
      "grad_norm": 0.7949868440628052,
      "learning_rate": 7.3107662794153245e-06,
      "loss": 0.7186,
      "step": 12726
    },
    {
      "epoch": 0.36670892641041897,
      "grad_norm": 0.7906820178031921,
      "learning_rate": 7.310352479460622e-06,
      "loss": 0.7242,
      "step": 12727
    },
    {
      "epoch": 0.36673773987206826,
      "grad_norm": 0.8118237257003784,
      "learning_rate": 7.3099386593850585e-06,
      "loss": 0.754,
      "step": 12728
    },
    {
      "epoch": 0.3667665533337175,
      "grad_norm": 0.8005017638206482,
      "learning_rate": 7.30952481919224e-06,
      "loss": 0.7288,
      "step": 12729
    },
    {
      "epoch": 0.3667953667953668,
      "grad_norm": 0.762450635433197,
      "learning_rate": 7.309110958885768e-06,
      "loss": 0.7143,
      "step": 12730
    },
    {
      "epoch": 0.3668241802570161,
      "grad_norm": 0.789707362651825,
      "learning_rate": 7.308697078469247e-06,
      "loss": 0.763,
      "step": 12731
    },
    {
      "epoch": 0.36685299371866537,
      "grad_norm": 0.7765840291976929,
      "learning_rate": 7.3082831779462834e-06,
      "loss": 0.7432,
      "step": 12732
    },
    {
      "epoch": 0.36688180718031466,
      "grad_norm": 0.8113734126091003,
      "learning_rate": 7.307869257320479e-06,
      "loss": 0.7349,
      "step": 12733
    },
    {
      "epoch": 0.36691062064196395,
      "grad_norm": 0.7569311261177063,
      "learning_rate": 7.307455316595442e-06,
      "loss": 0.7284,
      "step": 12734
    },
    {
      "epoch": 0.3669394341036132,
      "grad_norm": 0.7677510976791382,
      "learning_rate": 7.307041355774777e-06,
      "loss": 0.7336,
      "step": 12735
    },
    {
      "epoch": 0.3669682475652625,
      "grad_norm": 0.7669563293457031,
      "learning_rate": 7.3066273748620865e-06,
      "loss": 0.7577,
      "step": 12736
    },
    {
      "epoch": 0.36699706102691176,
      "grad_norm": 0.8070799708366394,
      "learning_rate": 7.3062133738609765e-06,
      "loss": 0.7496,
      "step": 12737
    },
    {
      "epoch": 0.36702587448856105,
      "grad_norm": 0.786372184753418,
      "learning_rate": 7.305799352775054e-06,
      "loss": 0.743,
      "step": 12738
    },
    {
      "epoch": 0.36705468795021035,
      "grad_norm": 0.8069302439689636,
      "learning_rate": 7.305385311607925e-06,
      "loss": 0.7437,
      "step": 12739
    },
    {
      "epoch": 0.36708350141185964,
      "grad_norm": 0.7726650238037109,
      "learning_rate": 7.304971250363194e-06,
      "loss": 0.7427,
      "step": 12740
    },
    {
      "epoch": 0.3671123148735089,
      "grad_norm": 0.7899466156959534,
      "learning_rate": 7.304557169044469e-06,
      "loss": 0.7144,
      "step": 12741
    },
    {
      "epoch": 0.36714112833515816,
      "grad_norm": 0.7596927881240845,
      "learning_rate": 7.3041430676553525e-06,
      "loss": 0.7331,
      "step": 12742
    },
    {
      "epoch": 0.36716994179680745,
      "grad_norm": 0.784072756767273,
      "learning_rate": 7.303728946199453e-06,
      "loss": 0.7467,
      "step": 12743
    },
    {
      "epoch": 0.36719875525845674,
      "grad_norm": 0.7569863200187683,
      "learning_rate": 7.303314804680379e-06,
      "loss": 0.7394,
      "step": 12744
    },
    {
      "epoch": 0.36722756872010603,
      "grad_norm": 0.7826274037361145,
      "learning_rate": 7.3029006431017345e-06,
      "loss": 0.7339,
      "step": 12745
    },
    {
      "epoch": 0.3672563821817553,
      "grad_norm": 0.7748938202857971,
      "learning_rate": 7.3024864614671275e-06,
      "loss": 0.7607,
      "step": 12746
    },
    {
      "epoch": 0.3672851956434046,
      "grad_norm": 0.8109842538833618,
      "learning_rate": 7.302072259780167e-06,
      "loss": 0.7382,
      "step": 12747
    },
    {
      "epoch": 0.3673140091050539,
      "grad_norm": 0.772478461265564,
      "learning_rate": 7.301658038044456e-06,
      "loss": 0.7209,
      "step": 12748
    },
    {
      "epoch": 0.36734282256670314,
      "grad_norm": 0.7758169174194336,
      "learning_rate": 7.301243796263604e-06,
      "loss": 0.7355,
      "step": 12749
    },
    {
      "epoch": 0.36737163602835243,
      "grad_norm": 0.8807239532470703,
      "learning_rate": 7.30082953444122e-06,
      "loss": 0.7063,
      "step": 12750
    },
    {
      "epoch": 0.3674004494900017,
      "grad_norm": 0.7789568901062012,
      "learning_rate": 7.30041525258091e-06,
      "loss": 0.7364,
      "step": 12751
    },
    {
      "epoch": 0.367429262951651,
      "grad_norm": 0.7899745106697083,
      "learning_rate": 7.300000950686284e-06,
      "loss": 0.7367,
      "step": 12752
    },
    {
      "epoch": 0.3674580764133003,
      "grad_norm": 0.7996919751167297,
      "learning_rate": 7.299586628760947e-06,
      "loss": 0.7587,
      "step": 12753
    },
    {
      "epoch": 0.3674868898749496,
      "grad_norm": 0.7837570905685425,
      "learning_rate": 7.2991722868085116e-06,
      "loss": 0.7413,
      "step": 12754
    },
    {
      "epoch": 0.36751570333659883,
      "grad_norm": 0.7753728032112122,
      "learning_rate": 7.2987579248325815e-06,
      "loss": 0.7656,
      "step": 12755
    },
    {
      "epoch": 0.3675445167982481,
      "grad_norm": 0.8268882036209106,
      "learning_rate": 7.2983435428367684e-06,
      "loss": 0.7539,
      "step": 12756
    },
    {
      "epoch": 0.3675733302598974,
      "grad_norm": 1.3023549318313599,
      "learning_rate": 7.297929140824683e-06,
      "loss": 0.7338,
      "step": 12757
    },
    {
      "epoch": 0.3676021437215467,
      "grad_norm": 0.7738940715789795,
      "learning_rate": 7.29751471879993e-06,
      "loss": 0.7226,
      "step": 12758
    },
    {
      "epoch": 0.367630957183196,
      "grad_norm": 0.7684247493743896,
      "learning_rate": 7.297100276766121e-06,
      "loss": 0.7303,
      "step": 12759
    },
    {
      "epoch": 0.3676597706448453,
      "grad_norm": 0.7828477025032043,
      "learning_rate": 7.296685814726863e-06,
      "loss": 0.7602,
      "step": 12760
    },
    {
      "epoch": 0.3676885841064946,
      "grad_norm": 0.7963311076164246,
      "learning_rate": 7.296271332685768e-06,
      "loss": 0.7417,
      "step": 12761
    },
    {
      "epoch": 0.3677173975681438,
      "grad_norm": 0.7692581415176392,
      "learning_rate": 7.2958568306464464e-06,
      "loss": 0.7308,
      "step": 12762
    },
    {
      "epoch": 0.3677462110297931,
      "grad_norm": 0.7721455693244934,
      "learning_rate": 7.295442308612506e-06,
      "loss": 0.7432,
      "step": 12763
    },
    {
      "epoch": 0.3677750244914424,
      "grad_norm": 0.8169425129890442,
      "learning_rate": 7.295027766587559e-06,
      "loss": 0.7463,
      "step": 12764
    },
    {
      "epoch": 0.3678038379530917,
      "grad_norm": 0.7920429110527039,
      "learning_rate": 7.294613204575213e-06,
      "loss": 0.769,
      "step": 12765
    },
    {
      "epoch": 0.367832651414741,
      "grad_norm": 0.8026204109191895,
      "learning_rate": 7.294198622579081e-06,
      "loss": 0.7567,
      "step": 12766
    },
    {
      "epoch": 0.36786146487639027,
      "grad_norm": 0.7837314605712891,
      "learning_rate": 7.29378402060277e-06,
      "loss": 0.7568,
      "step": 12767
    },
    {
      "epoch": 0.36789027833803956,
      "grad_norm": 0.8199963569641113,
      "learning_rate": 7.293369398649894e-06,
      "loss": 0.7451,
      "step": 12768
    },
    {
      "epoch": 0.3679190917996888,
      "grad_norm": 0.7722735404968262,
      "learning_rate": 7.292954756724064e-06,
      "loss": 0.7178,
      "step": 12769
    },
    {
      "epoch": 0.3679479052613381,
      "grad_norm": 0.773000955581665,
      "learning_rate": 7.29254009482889e-06,
      "loss": 0.7609,
      "step": 12770
    },
    {
      "epoch": 0.3679767187229874,
      "grad_norm": 0.757920503616333,
      "learning_rate": 7.292125412967983e-06,
      "loss": 0.7348,
      "step": 12771
    },
    {
      "epoch": 0.36800553218463666,
      "grad_norm": 0.7681493163108826,
      "learning_rate": 7.291710711144955e-06,
      "loss": 0.7216,
      "step": 12772
    },
    {
      "epoch": 0.36803434564628595,
      "grad_norm": 0.7950176000595093,
      "learning_rate": 7.2912959893634175e-06,
      "loss": 0.7446,
      "step": 12773
    },
    {
      "epoch": 0.36806315910793524,
      "grad_norm": 0.8415719866752625,
      "learning_rate": 7.290881247626983e-06,
      "loss": 0.7739,
      "step": 12774
    },
    {
      "epoch": 0.36809197256958454,
      "grad_norm": 0.7975728511810303,
      "learning_rate": 7.290466485939262e-06,
      "loss": 0.722,
      "step": 12775
    },
    {
      "epoch": 0.36812078603123377,
      "grad_norm": 0.8175250887870789,
      "learning_rate": 7.290051704303869e-06,
      "loss": 0.7514,
      "step": 12776
    },
    {
      "epoch": 0.36814959949288306,
      "grad_norm": 0.8779968023300171,
      "learning_rate": 7.289636902724414e-06,
      "loss": 0.7714,
      "step": 12777
    },
    {
      "epoch": 0.36817841295453235,
      "grad_norm": 0.7924428582191467,
      "learning_rate": 7.28922208120451e-06,
      "loss": 0.7371,
      "step": 12778
    },
    {
      "epoch": 0.36820722641618164,
      "grad_norm": 0.7726795077323914,
      "learning_rate": 7.28880723974777e-06,
      "loss": 0.7642,
      "step": 12779
    },
    {
      "epoch": 0.36823603987783093,
      "grad_norm": 0.7658511400222778,
      "learning_rate": 7.288392378357808e-06,
      "loss": 0.7558,
      "step": 12780
    },
    {
      "epoch": 0.3682648533394802,
      "grad_norm": 0.8191518783569336,
      "learning_rate": 7.2879774970382354e-06,
      "loss": 0.7399,
      "step": 12781
    },
    {
      "epoch": 0.36829366680112946,
      "grad_norm": 0.761101245880127,
      "learning_rate": 7.2875625957926656e-06,
      "loss": 0.7495,
      "step": 12782
    },
    {
      "epoch": 0.36832248026277875,
      "grad_norm": 0.7863438129425049,
      "learning_rate": 7.287147674624713e-06,
      "loss": 0.7137,
      "step": 12783
    },
    {
      "epoch": 0.36835129372442804,
      "grad_norm": 0.7505287528038025,
      "learning_rate": 7.28673273353799e-06,
      "loss": 0.7231,
      "step": 12784
    },
    {
      "epoch": 0.36838010718607733,
      "grad_norm": 0.7715272307395935,
      "learning_rate": 7.28631777253611e-06,
      "loss": 0.7123,
      "step": 12785
    },
    {
      "epoch": 0.3684089206477266,
      "grad_norm": 0.9294905662536621,
      "learning_rate": 7.285902791622689e-06,
      "loss": 0.7582,
      "step": 12786
    },
    {
      "epoch": 0.3684377341093759,
      "grad_norm": 0.7689439654350281,
      "learning_rate": 7.285487790801339e-06,
      "loss": 0.7362,
      "step": 12787
    },
    {
      "epoch": 0.3684665475710252,
      "grad_norm": 0.7316070795059204,
      "learning_rate": 7.285072770075675e-06,
      "loss": 0.7457,
      "step": 12788
    },
    {
      "epoch": 0.36849536103267444,
      "grad_norm": 0.7651985883712769,
      "learning_rate": 7.284657729449312e-06,
      "loss": 0.7453,
      "step": 12789
    },
    {
      "epoch": 0.36852417449432373,
      "grad_norm": 0.766254186630249,
      "learning_rate": 7.284242668925863e-06,
      "loss": 0.739,
      "step": 12790
    },
    {
      "epoch": 0.368552987955973,
      "grad_norm": 0.7471922636032104,
      "learning_rate": 7.283827588508945e-06,
      "loss": 0.7352,
      "step": 12791
    },
    {
      "epoch": 0.3685818014176223,
      "grad_norm": 0.7436578869819641,
      "learning_rate": 7.28341248820217e-06,
      "loss": 0.7224,
      "step": 12792
    },
    {
      "epoch": 0.3686106148792716,
      "grad_norm": 0.792259156703949,
      "learning_rate": 7.282997368009156e-06,
      "loss": 0.7296,
      "step": 12793
    },
    {
      "epoch": 0.3686394283409209,
      "grad_norm": 0.7654797434806824,
      "learning_rate": 7.282582227933517e-06,
      "loss": 0.7524,
      "step": 12794
    },
    {
      "epoch": 0.3686682418025702,
      "grad_norm": 0.7782321572303772,
      "learning_rate": 7.282167067978869e-06,
      "loss": 0.7451,
      "step": 12795
    },
    {
      "epoch": 0.3686970552642194,
      "grad_norm": 0.7916568517684937,
      "learning_rate": 7.281751888148827e-06,
      "loss": 0.7685,
      "step": 12796
    },
    {
      "epoch": 0.3687258687258687,
      "grad_norm": 0.7952301502227783,
      "learning_rate": 7.281336688447005e-06,
      "loss": 0.7271,
      "step": 12797
    },
    {
      "epoch": 0.368754682187518,
      "grad_norm": 0.7545308470726013,
      "learning_rate": 7.280921468877023e-06,
      "loss": 0.7166,
      "step": 12798
    },
    {
      "epoch": 0.3687834956491673,
      "grad_norm": 0.7547735571861267,
      "learning_rate": 7.2805062294424956e-06,
      "loss": 0.7262,
      "step": 12799
    },
    {
      "epoch": 0.3688123091108166,
      "grad_norm": 0.767073392868042,
      "learning_rate": 7.280090970147038e-06,
      "loss": 0.7449,
      "step": 12800
    },
    {
      "epoch": 0.3688411225724659,
      "grad_norm": 0.7610882520675659,
      "learning_rate": 7.279675690994266e-06,
      "loss": 0.7628,
      "step": 12801
    },
    {
      "epoch": 0.36886993603411516,
      "grad_norm": 0.7686296701431274,
      "learning_rate": 7.279260391987799e-06,
      "loss": 0.7545,
      "step": 12802
    },
    {
      "epoch": 0.3688987494957644,
      "grad_norm": 0.7714722156524658,
      "learning_rate": 7.278845073131251e-06,
      "loss": 0.7467,
      "step": 12803
    },
    {
      "epoch": 0.3689275629574137,
      "grad_norm": 0.7799524664878845,
      "learning_rate": 7.278429734428241e-06,
      "loss": 0.7434,
      "step": 12804
    },
    {
      "epoch": 0.368956376419063,
      "grad_norm": 0.7945626378059387,
      "learning_rate": 7.278014375882387e-06,
      "loss": 0.724,
      "step": 12805
    },
    {
      "epoch": 0.36898518988071227,
      "grad_norm": 0.783387303352356,
      "learning_rate": 7.277598997497304e-06,
      "loss": 0.7316,
      "step": 12806
    },
    {
      "epoch": 0.36901400334236156,
      "grad_norm": 0.8028669357299805,
      "learning_rate": 7.277183599276611e-06,
      "loss": 0.7628,
      "step": 12807
    },
    {
      "epoch": 0.36904281680401085,
      "grad_norm": 0.7680822610855103,
      "learning_rate": 7.276768181223926e-06,
      "loss": 0.7361,
      "step": 12808
    },
    {
      "epoch": 0.3690716302656601,
      "grad_norm": 0.7691142559051514,
      "learning_rate": 7.2763527433428625e-06,
      "loss": 0.7638,
      "step": 12809
    },
    {
      "epoch": 0.3691004437273094,
      "grad_norm": 0.7806946039199829,
      "learning_rate": 7.275937285637045e-06,
      "loss": 0.7238,
      "step": 12810
    },
    {
      "epoch": 0.36912925718895867,
      "grad_norm": 0.7937992215156555,
      "learning_rate": 7.2755218081100876e-06,
      "loss": 0.7149,
      "step": 12811
    },
    {
      "epoch": 0.36915807065060796,
      "grad_norm": 0.7618255019187927,
      "learning_rate": 7.2751063107656105e-06,
      "loss": 0.7547,
      "step": 12812
    },
    {
      "epoch": 0.36918688411225725,
      "grad_norm": 0.7644799947738647,
      "learning_rate": 7.274690793607232e-06,
      "loss": 0.7324,
      "step": 12813
    },
    {
      "epoch": 0.36921569757390654,
      "grad_norm": 0.782278299331665,
      "learning_rate": 7.27427525663857e-06,
      "loss": 0.7224,
      "step": 12814
    },
    {
      "epoch": 0.36924451103555583,
      "grad_norm": 0.7565146684646606,
      "learning_rate": 7.273859699863244e-06,
      "loss": 0.7247,
      "step": 12815
    },
    {
      "epoch": 0.36927332449720507,
      "grad_norm": 0.8457037806510925,
      "learning_rate": 7.273444123284874e-06,
      "loss": 0.7383,
      "step": 12816
    },
    {
      "epoch": 0.36930213795885436,
      "grad_norm": 0.7723545432090759,
      "learning_rate": 7.273028526907077e-06,
      "loss": 0.735,
      "step": 12817
    },
    {
      "epoch": 0.36933095142050365,
      "grad_norm": 0.7710839509963989,
      "learning_rate": 7.2726129107334756e-06,
      "loss": 0.7363,
      "step": 12818
    },
    {
      "epoch": 0.36935976488215294,
      "grad_norm": 0.7389088273048401,
      "learning_rate": 7.272197274767686e-06,
      "loss": 0.7279,
      "step": 12819
    },
    {
      "epoch": 0.36938857834380223,
      "grad_norm": 0.7598415017127991,
      "learning_rate": 7.27178161901333e-06,
      "loss": 0.7457,
      "step": 12820
    },
    {
      "epoch": 0.3694173918054515,
      "grad_norm": 0.7639048099517822,
      "learning_rate": 7.271365943474026e-06,
      "loss": 0.7437,
      "step": 12821
    },
    {
      "epoch": 0.3694462052671008,
      "grad_norm": 0.7637218236923218,
      "learning_rate": 7.270950248153396e-06,
      "loss": 0.7498,
      "step": 12822
    },
    {
      "epoch": 0.36947501872875005,
      "grad_norm": 0.7824816703796387,
      "learning_rate": 7.2705345330550604e-06,
      "loss": 0.7719,
      "step": 12823
    },
    {
      "epoch": 0.36950383219039934,
      "grad_norm": 0.7432796955108643,
      "learning_rate": 7.270118798182637e-06,
      "loss": 0.7247,
      "step": 12824
    },
    {
      "epoch": 0.36953264565204863,
      "grad_norm": 0.7544249892234802,
      "learning_rate": 7.269703043539749e-06,
      "loss": 0.7421,
      "step": 12825
    },
    {
      "epoch": 0.3695614591136979,
      "grad_norm": 0.7578464150428772,
      "learning_rate": 7.2692872691300165e-06,
      "loss": 0.7034,
      "step": 12826
    },
    {
      "epoch": 0.3695902725753472,
      "grad_norm": 0.7989123463630676,
      "learning_rate": 7.26887147495706e-06,
      "loss": 0.7404,
      "step": 12827
    },
    {
      "epoch": 0.3696190860369965,
      "grad_norm": 0.7508918046951294,
      "learning_rate": 7.2684556610245e-06,
      "loss": 0.7441,
      "step": 12828
    },
    {
      "epoch": 0.3696478994986458,
      "grad_norm": 0.8106495141983032,
      "learning_rate": 7.268039827335961e-06,
      "loss": 0.7437,
      "step": 12829
    },
    {
      "epoch": 0.36967671296029503,
      "grad_norm": 0.772346556186676,
      "learning_rate": 7.267623973895061e-06,
      "loss": 0.7371,
      "step": 12830
    },
    {
      "epoch": 0.3697055264219443,
      "grad_norm": 0.766761302947998,
      "learning_rate": 7.267208100705423e-06,
      "loss": 0.7227,
      "step": 12831
    },
    {
      "epoch": 0.3697343398835936,
      "grad_norm": 0.7449536919593811,
      "learning_rate": 7.26679220777067e-06,
      "loss": 0.7496,
      "step": 12832
    },
    {
      "epoch": 0.3697631533452429,
      "grad_norm": 0.7572663426399231,
      "learning_rate": 7.266376295094421e-06,
      "loss": 0.7211,
      "step": 12833
    },
    {
      "epoch": 0.3697919668068922,
      "grad_norm": 0.8034539818763733,
      "learning_rate": 7.2659603626803e-06,
      "loss": 0.7535,
      "step": 12834
    },
    {
      "epoch": 0.3698207802685415,
      "grad_norm": 0.7606357336044312,
      "learning_rate": 7.265544410531931e-06,
      "loss": 0.7184,
      "step": 12835
    },
    {
      "epoch": 0.3698495937301907,
      "grad_norm": 0.8085760474205017,
      "learning_rate": 7.265128438652934e-06,
      "loss": 0.7344,
      "step": 12836
    },
    {
      "epoch": 0.36987840719184,
      "grad_norm": 0.8181759119033813,
      "learning_rate": 7.264712447046933e-06,
      "loss": 0.7564,
      "step": 12837
    },
    {
      "epoch": 0.3699072206534893,
      "grad_norm": 0.7515325546264648,
      "learning_rate": 7.26429643571755e-06,
      "loss": 0.6983,
      "step": 12838
    },
    {
      "epoch": 0.3699360341151386,
      "grad_norm": 0.771048367023468,
      "learning_rate": 7.263880404668409e-06,
      "loss": 0.7372,
      "step": 12839
    },
    {
      "epoch": 0.3699648475767879,
      "grad_norm": 0.790888786315918,
      "learning_rate": 7.2634643539031325e-06,
      "loss": 0.727,
      "step": 12840
    },
    {
      "epoch": 0.36999366103843717,
      "grad_norm": 0.8080943822860718,
      "learning_rate": 7.263048283425345e-06,
      "loss": 0.7417,
      "step": 12841
    },
    {
      "epoch": 0.37002247450008646,
      "grad_norm": 0.746475100517273,
      "learning_rate": 7.2626321932386676e-06,
      "loss": 0.7096,
      "step": 12842
    },
    {
      "epoch": 0.3700512879617357,
      "grad_norm": 0.7647432684898376,
      "learning_rate": 7.262216083346727e-06,
      "loss": 0.74,
      "step": 12843
    },
    {
      "epoch": 0.370080101423385,
      "grad_norm": 0.8133372068405151,
      "learning_rate": 7.261799953753145e-06,
      "loss": 0.7457,
      "step": 12844
    },
    {
      "epoch": 0.3701089148850343,
      "grad_norm": 0.7848641276359558,
      "learning_rate": 7.261383804461546e-06,
      "loss": 0.7436,
      "step": 12845
    },
    {
      "epoch": 0.37013772834668357,
      "grad_norm": 0.7638106942176819,
      "learning_rate": 7.260967635475554e-06,
      "loss": 0.7497,
      "step": 12846
    },
    {
      "epoch": 0.37016654180833286,
      "grad_norm": 0.7813120484352112,
      "learning_rate": 7.260551446798796e-06,
      "loss": 0.7311,
      "step": 12847
    },
    {
      "epoch": 0.37019535526998215,
      "grad_norm": 0.7636242508888245,
      "learning_rate": 7.260135238434893e-06,
      "loss": 0.7159,
      "step": 12848
    },
    {
      "epoch": 0.37022416873163144,
      "grad_norm": 0.7787232398986816,
      "learning_rate": 7.259719010387471e-06,
      "loss": 0.7348,
      "step": 12849
    },
    {
      "epoch": 0.3702529821932807,
      "grad_norm": 0.7530942559242249,
      "learning_rate": 7.259302762660158e-06,
      "loss": 0.7433,
      "step": 12850
    },
    {
      "epoch": 0.37028179565492997,
      "grad_norm": 0.7441816926002502,
      "learning_rate": 7.258886495256574e-06,
      "loss": 0.7304,
      "step": 12851
    },
    {
      "epoch": 0.37031060911657926,
      "grad_norm": 0.7951955795288086,
      "learning_rate": 7.258470208180347e-06,
      "loss": 0.7382,
      "step": 12852
    },
    {
      "epoch": 0.37033942257822855,
      "grad_norm": 0.797158420085907,
      "learning_rate": 7.258053901435103e-06,
      "loss": 0.7499,
      "step": 12853
    },
    {
      "epoch": 0.37036823603987784,
      "grad_norm": 0.7699688076972961,
      "learning_rate": 7.257637575024465e-06,
      "loss": 0.7432,
      "step": 12854
    },
    {
      "epoch": 0.37039704950152713,
      "grad_norm": 0.7337840795516968,
      "learning_rate": 7.2572212289520626e-06,
      "loss": 0.7455,
      "step": 12855
    },
    {
      "epoch": 0.3704258629631764,
      "grad_norm": 0.786777913570404,
      "learning_rate": 7.2568048632215195e-06,
      "loss": 0.7278,
      "step": 12856
    },
    {
      "epoch": 0.37045467642482566,
      "grad_norm": 0.8002542853355408,
      "learning_rate": 7.25638847783646e-06,
      "loss": 0.728,
      "step": 12857
    },
    {
      "epoch": 0.37048348988647495,
      "grad_norm": 0.766140341758728,
      "learning_rate": 7.255972072800515e-06,
      "loss": 0.735,
      "step": 12858
    },
    {
      "epoch": 0.37051230334812424,
      "grad_norm": 0.7621187567710876,
      "learning_rate": 7.255555648117307e-06,
      "loss": 0.7292,
      "step": 12859
    },
    {
      "epoch": 0.37054111680977353,
      "grad_norm": 0.8019818663597107,
      "learning_rate": 7.255139203790464e-06,
      "loss": 0.7335,
      "step": 12860
    },
    {
      "epoch": 0.3705699302714228,
      "grad_norm": 0.9888969659805298,
      "learning_rate": 7.254722739823614e-06,
      "loss": 0.7221,
      "step": 12861
    },
    {
      "epoch": 0.3705987437330721,
      "grad_norm": 0.7643141746520996,
      "learning_rate": 7.254306256220383e-06,
      "loss": 0.7424,
      "step": 12862
    },
    {
      "epoch": 0.37062755719472135,
      "grad_norm": 0.7846137285232544,
      "learning_rate": 7.253889752984396e-06,
      "loss": 0.7162,
      "step": 12863
    },
    {
      "epoch": 0.37065637065637064,
      "grad_norm": 0.7753503918647766,
      "learning_rate": 7.253473230119283e-06,
      "loss": 0.7338,
      "step": 12864
    },
    {
      "epoch": 0.37068518411801993,
      "grad_norm": 0.7333825826644897,
      "learning_rate": 7.253056687628671e-06,
      "loss": 0.7135,
      "step": 12865
    },
    {
      "epoch": 0.3707139975796692,
      "grad_norm": 0.7504845261573792,
      "learning_rate": 7.252640125516188e-06,
      "loss": 0.7345,
      "step": 12866
    },
    {
      "epoch": 0.3707428110413185,
      "grad_norm": 0.7616289258003235,
      "learning_rate": 7.2522235437854614e-06,
      "loss": 0.7433,
      "step": 12867
    },
    {
      "epoch": 0.3707716245029678,
      "grad_norm": 0.7891879677772522,
      "learning_rate": 7.251806942440117e-06,
      "loss": 0.7459,
      "step": 12868
    },
    {
      "epoch": 0.3708004379646171,
      "grad_norm": 0.761630654335022,
      "learning_rate": 7.251390321483788e-06,
      "loss": 0.7314,
      "step": 12869
    },
    {
      "epoch": 0.3708292514262663,
      "grad_norm": 0.7812952995300293,
      "learning_rate": 7.250973680920098e-06,
      "loss": 0.7348,
      "step": 12870
    },
    {
      "epoch": 0.3708580648879156,
      "grad_norm": 0.7473841905593872,
      "learning_rate": 7.250557020752678e-06,
      "loss": 0.7442,
      "step": 12871
    },
    {
      "epoch": 0.3708868783495649,
      "grad_norm": 0.8166649341583252,
      "learning_rate": 7.250140340985156e-06,
      "loss": 0.7417,
      "step": 12872
    },
    {
      "epoch": 0.3709156918112142,
      "grad_norm": 0.7536835074424744,
      "learning_rate": 7.249723641621161e-06,
      "loss": 0.737,
      "step": 12873
    },
    {
      "epoch": 0.3709445052728635,
      "grad_norm": 0.7510586977005005,
      "learning_rate": 7.249306922664322e-06,
      "loss": 0.7359,
      "step": 12874
    },
    {
      "epoch": 0.3709733187345128,
      "grad_norm": 0.753948450088501,
      "learning_rate": 7.248890184118268e-06,
      "loss": 0.731,
      "step": 12875
    },
    {
      "epoch": 0.37100213219616207,
      "grad_norm": 0.7762637138366699,
      "learning_rate": 7.248473425986629e-06,
      "loss": 0.7484,
      "step": 12876
    },
    {
      "epoch": 0.3710309456578113,
      "grad_norm": 0.7909533977508545,
      "learning_rate": 7.248056648273034e-06,
      "loss": 0.7689,
      "step": 12877
    },
    {
      "epoch": 0.3710597591194606,
      "grad_norm": 0.7769101858139038,
      "learning_rate": 7.247639850981112e-06,
      "loss": 0.7279,
      "step": 12878
    },
    {
      "epoch": 0.3710885725811099,
      "grad_norm": 0.7841305136680603,
      "learning_rate": 7.247223034114495e-06,
      "loss": 0.7539,
      "step": 12879
    },
    {
      "epoch": 0.3711173860427592,
      "grad_norm": 0.7574643492698669,
      "learning_rate": 7.24680619767681e-06,
      "loss": 0.7544,
      "step": 12880
    },
    {
      "epoch": 0.37114619950440847,
      "grad_norm": 0.8172972202301025,
      "learning_rate": 7.246389341671691e-06,
      "loss": 0.7577,
      "step": 12881
    },
    {
      "epoch": 0.37117501296605776,
      "grad_norm": 0.762378990650177,
      "learning_rate": 7.245972466102767e-06,
      "loss": 0.7213,
      "step": 12882
    },
    {
      "epoch": 0.37120382642770705,
      "grad_norm": 0.7561028599739075,
      "learning_rate": 7.245555570973666e-06,
      "loss": 0.7157,
      "step": 12883
    },
    {
      "epoch": 0.3712326398893563,
      "grad_norm": 0.7630577087402344,
      "learning_rate": 7.245138656288021e-06,
      "loss": 0.7263,
      "step": 12884
    },
    {
      "epoch": 0.3712614533510056,
      "grad_norm": 0.7683738470077515,
      "learning_rate": 7.2447217220494634e-06,
      "loss": 0.7359,
      "step": 12885
    },
    {
      "epoch": 0.37129026681265487,
      "grad_norm": 0.7968388199806213,
      "learning_rate": 7.244304768261622e-06,
      "loss": 0.7354,
      "step": 12886
    },
    {
      "epoch": 0.37131908027430416,
      "grad_norm": 0.7899968028068542,
      "learning_rate": 7.24388779492813e-06,
      "loss": 0.7462,
      "step": 12887
    },
    {
      "epoch": 0.37134789373595345,
      "grad_norm": 0.7533883452415466,
      "learning_rate": 7.24347080205262e-06,
      "loss": 0.7299,
      "step": 12888
    },
    {
      "epoch": 0.37137670719760274,
      "grad_norm": 1.9137972593307495,
      "learning_rate": 7.2430537896387206e-06,
      "loss": 0.75,
      "step": 12889
    },
    {
      "epoch": 0.371405520659252,
      "grad_norm": 0.7710535526275635,
      "learning_rate": 7.242636757690064e-06,
      "loss": 0.7395,
      "step": 12890
    },
    {
      "epoch": 0.37143433412090127,
      "grad_norm": 0.7787205576896667,
      "learning_rate": 7.242219706210285e-06,
      "loss": 0.7377,
      "step": 12891
    },
    {
      "epoch": 0.37146314758255056,
      "grad_norm": 0.760084867477417,
      "learning_rate": 7.241802635203013e-06,
      "loss": 0.7052,
      "step": 12892
    },
    {
      "epoch": 0.37149196104419985,
      "grad_norm": 0.7830608487129211,
      "learning_rate": 7.241385544671881e-06,
      "loss": 0.764,
      "step": 12893
    },
    {
      "epoch": 0.37152077450584914,
      "grad_norm": 0.7755436897277832,
      "learning_rate": 7.240968434620521e-06,
      "loss": 0.7556,
      "step": 12894
    },
    {
      "epoch": 0.37154958796749843,
      "grad_norm": 0.7779453992843628,
      "learning_rate": 7.240551305052568e-06,
      "loss": 0.7415,
      "step": 12895
    },
    {
      "epoch": 0.3715784014291477,
      "grad_norm": 0.8081619143486023,
      "learning_rate": 7.240134155971651e-06,
      "loss": 0.765,
      "step": 12896
    },
    {
      "epoch": 0.37160721489079696,
      "grad_norm": 0.7853147983551025,
      "learning_rate": 7.239716987381406e-06,
      "loss": 0.7457,
      "step": 12897
    },
    {
      "epoch": 0.37163602835244625,
      "grad_norm": 0.7664651274681091,
      "learning_rate": 7.239299799285463e-06,
      "loss": 0.7601,
      "step": 12898
    },
    {
      "epoch": 0.37166484181409554,
      "grad_norm": 0.7531177401542664,
      "learning_rate": 7.238882591687458e-06,
      "loss": 0.742,
      "step": 12899
    },
    {
      "epoch": 0.3716936552757448,
      "grad_norm": 0.7561687231063843,
      "learning_rate": 7.238465364591025e-06,
      "loss": 0.7332,
      "step": 12900
    },
    {
      "epoch": 0.3717224687373941,
      "grad_norm": 0.7855347990989685,
      "learning_rate": 7.238048117999795e-06,
      "loss": 0.7281,
      "step": 12901
    },
    {
      "epoch": 0.3717512821990434,
      "grad_norm": 0.7710037231445312,
      "learning_rate": 7.237630851917405e-06,
      "loss": 0.7087,
      "step": 12902
    },
    {
      "epoch": 0.3717800956606927,
      "grad_norm": 0.7636103630065918,
      "learning_rate": 7.237213566347485e-06,
      "loss": 0.7411,
      "step": 12903
    },
    {
      "epoch": 0.37180890912234194,
      "grad_norm": 0.7589457035064697,
      "learning_rate": 7.236796261293671e-06,
      "loss": 0.7356,
      "step": 12904
    },
    {
      "epoch": 0.3718377225839912,
      "grad_norm": 0.7638722658157349,
      "learning_rate": 7.236378936759599e-06,
      "loss": 0.7285,
      "step": 12905
    },
    {
      "epoch": 0.3718665360456405,
      "grad_norm": 0.7735623121261597,
      "learning_rate": 7.235961592748901e-06,
      "loss": 0.7248,
      "step": 12906
    },
    {
      "epoch": 0.3718953495072898,
      "grad_norm": 0.7459738850593567,
      "learning_rate": 7.235544229265214e-06,
      "loss": 0.7295,
      "step": 12907
    },
    {
      "epoch": 0.3719241629689391,
      "grad_norm": 0.7884818911552429,
      "learning_rate": 7.23512684631217e-06,
      "loss": 0.7632,
      "step": 12908
    },
    {
      "epoch": 0.3719529764305884,
      "grad_norm": 0.790768563747406,
      "learning_rate": 7.234709443893406e-06,
      "loss": 0.7496,
      "step": 12909
    },
    {
      "epoch": 0.3719817898922377,
      "grad_norm": 0.7778552174568176,
      "learning_rate": 7.234292022012557e-06,
      "loss": 0.7282,
      "step": 12910
    },
    {
      "epoch": 0.3720106033538869,
      "grad_norm": 0.8514861464500427,
      "learning_rate": 7.2338745806732565e-06,
      "loss": 0.7422,
      "step": 12911
    },
    {
      "epoch": 0.3720394168155362,
      "grad_norm": 0.7779433727264404,
      "learning_rate": 7.233457119879143e-06,
      "loss": 0.7304,
      "step": 12912
    },
    {
      "epoch": 0.3720682302771855,
      "grad_norm": 0.8041344285011292,
      "learning_rate": 7.233039639633851e-06,
      "loss": 0.7522,
      "step": 12913
    },
    {
      "epoch": 0.3720970437388348,
      "grad_norm": 0.7823668122291565,
      "learning_rate": 7.2326221399410154e-06,
      "loss": 0.7474,
      "step": 12914
    },
    {
      "epoch": 0.3721258572004841,
      "grad_norm": 0.8141038417816162,
      "learning_rate": 7.232204620804273e-06,
      "loss": 0.7466,
      "step": 12915
    },
    {
      "epoch": 0.37215467066213337,
      "grad_norm": 0.751179575920105,
      "learning_rate": 7.231787082227258e-06,
      "loss": 0.7319,
      "step": 12916
    },
    {
      "epoch": 0.3721834841237826,
      "grad_norm": 0.7659267783164978,
      "learning_rate": 7.231369524213611e-06,
      "loss": 0.7259,
      "step": 12917
    },
    {
      "epoch": 0.3722122975854319,
      "grad_norm": 1.0100138187408447,
      "learning_rate": 7.2309519467669645e-06,
      "loss": 0.7599,
      "step": 12918
    },
    {
      "epoch": 0.3722411110470812,
      "grad_norm": 0.7759013175964355,
      "learning_rate": 7.230534349890957e-06,
      "loss": 0.7506,
      "step": 12919
    },
    {
      "epoch": 0.3722699245087305,
      "grad_norm": 0.781649649143219,
      "learning_rate": 7.230116733589225e-06,
      "loss": 0.7667,
      "step": 12920
    },
    {
      "epoch": 0.37229873797037977,
      "grad_norm": 0.7730633020401001,
      "learning_rate": 7.229699097865406e-06,
      "loss": 0.7451,
      "step": 12921
    },
    {
      "epoch": 0.37232755143202906,
      "grad_norm": 0.7420182824134827,
      "learning_rate": 7.229281442723136e-06,
      "loss": 0.7275,
      "step": 12922
    },
    {
      "epoch": 0.37235636489367835,
      "grad_norm": 0.7569155693054199,
      "learning_rate": 7.228863768166053e-06,
      "loss": 0.7507,
      "step": 12923
    },
    {
      "epoch": 0.3723851783553276,
      "grad_norm": 0.7634981870651245,
      "learning_rate": 7.228446074197796e-06,
      "loss": 0.7436,
      "step": 12924
    },
    {
      "epoch": 0.3724139918169769,
      "grad_norm": 0.803687572479248,
      "learning_rate": 7.228028360822e-06,
      "loss": 0.7389,
      "step": 12925
    },
    {
      "epoch": 0.37244280527862617,
      "grad_norm": 0.7511393427848816,
      "learning_rate": 7.227610628042305e-06,
      "loss": 0.7348,
      "step": 12926
    },
    {
      "epoch": 0.37247161874027546,
      "grad_norm": 0.7606475949287415,
      "learning_rate": 7.227192875862348e-06,
      "loss": 0.7147,
      "step": 12927
    },
    {
      "epoch": 0.37250043220192475,
      "grad_norm": 0.8075210452079773,
      "learning_rate": 7.226775104285767e-06,
      "loss": 0.7399,
      "step": 12928
    },
    {
      "epoch": 0.37252924566357404,
      "grad_norm": 0.7822110056877136,
      "learning_rate": 7.226357313316201e-06,
      "loss": 0.7314,
      "step": 12929
    },
    {
      "epoch": 0.37255805912522333,
      "grad_norm": 0.7806220650672913,
      "learning_rate": 7.225939502957288e-06,
      "loss": 0.7636,
      "step": 12930
    },
    {
      "epoch": 0.37258687258687256,
      "grad_norm": 0.7662093639373779,
      "learning_rate": 7.225521673212668e-06,
      "loss": 0.7335,
      "step": 12931
    },
    {
      "epoch": 0.37261568604852185,
      "grad_norm": 0.7742823362350464,
      "learning_rate": 7.225103824085977e-06,
      "loss": 0.7305,
      "step": 12932
    },
    {
      "epoch": 0.37264449951017115,
      "grad_norm": 0.7563068270683289,
      "learning_rate": 7.2246859555808575e-06,
      "loss": 0.7195,
      "step": 12933
    },
    {
      "epoch": 0.37267331297182044,
      "grad_norm": 0.8617812395095825,
      "learning_rate": 7.224268067700947e-06,
      "loss": 0.7205,
      "step": 12934
    },
    {
      "epoch": 0.3727021264334697,
      "grad_norm": 0.8118330240249634,
      "learning_rate": 7.2238501604498855e-06,
      "loss": 0.7403,
      "step": 12935
    },
    {
      "epoch": 0.372730939895119,
      "grad_norm": 0.7730581164360046,
      "learning_rate": 7.2234322338313125e-06,
      "loss": 0.7639,
      "step": 12936
    },
    {
      "epoch": 0.3727597533567683,
      "grad_norm": 0.7964685559272766,
      "learning_rate": 7.223014287848866e-06,
      "loss": 0.7371,
      "step": 12937
    },
    {
      "epoch": 0.37278856681841754,
      "grad_norm": 0.8072197437286377,
      "learning_rate": 7.222596322506188e-06,
      "loss": 0.7309,
      "step": 12938
    },
    {
      "epoch": 0.37281738028006683,
      "grad_norm": 0.7723722457885742,
      "learning_rate": 7.222178337806918e-06,
      "loss": 0.7383,
      "step": 12939
    },
    {
      "epoch": 0.3728461937417161,
      "grad_norm": 0.7337687015533447,
      "learning_rate": 7.2217603337546945e-06,
      "loss": 0.7358,
      "step": 12940
    },
    {
      "epoch": 0.3728750072033654,
      "grad_norm": 0.7414670586585999,
      "learning_rate": 7.221342310353161e-06,
      "loss": 0.7353,
      "step": 12941
    },
    {
      "epoch": 0.3729038206650147,
      "grad_norm": 0.845198392868042,
      "learning_rate": 7.220924267605958e-06,
      "loss": 0.7386,
      "step": 12942
    },
    {
      "epoch": 0.372932634126664,
      "grad_norm": 0.7622140049934387,
      "learning_rate": 7.2205062055167215e-06,
      "loss": 0.7273,
      "step": 12943
    },
    {
      "epoch": 0.37296144758831323,
      "grad_norm": 0.7678223252296448,
      "learning_rate": 7.220088124089097e-06,
      "loss": 0.7592,
      "step": 12944
    },
    {
      "epoch": 0.3729902610499625,
      "grad_norm": 0.7864711284637451,
      "learning_rate": 7.219670023326726e-06,
      "loss": 0.7408,
      "step": 12945
    },
    {
      "epoch": 0.3730190745116118,
      "grad_norm": 0.8031368255615234,
      "learning_rate": 7.219251903233245e-06,
      "loss": 0.742,
      "step": 12946
    },
    {
      "epoch": 0.3730478879732611,
      "grad_norm": 0.7798982262611389,
      "learning_rate": 7.2188337638122995e-06,
      "loss": 0.7256,
      "step": 12947
    },
    {
      "epoch": 0.3730767014349104,
      "grad_norm": 0.7334528565406799,
      "learning_rate": 7.2184156050675305e-06,
      "loss": 0.7558,
      "step": 12948
    },
    {
      "epoch": 0.3731055148965597,
      "grad_norm": 0.8065120577812195,
      "learning_rate": 7.217997427002578e-06,
      "loss": 0.7435,
      "step": 12949
    },
    {
      "epoch": 0.373134328358209,
      "grad_norm": 0.7576470971107483,
      "learning_rate": 7.217579229621085e-06,
      "loss": 0.7445,
      "step": 12950
    },
    {
      "epoch": 0.3731631418198582,
      "grad_norm": 0.7553635835647583,
      "learning_rate": 7.217161012926695e-06,
      "loss": 0.7712,
      "step": 12951
    },
    {
      "epoch": 0.3731919552815075,
      "grad_norm": 0.740827202796936,
      "learning_rate": 7.216742776923048e-06,
      "loss": 0.7101,
      "step": 12952
    },
    {
      "epoch": 0.3732207687431568,
      "grad_norm": 0.7637404203414917,
      "learning_rate": 7.216324521613789e-06,
      "loss": 0.7389,
      "step": 12953
    },
    {
      "epoch": 0.3732495822048061,
      "grad_norm": 0.7408134341239929,
      "learning_rate": 7.215906247002557e-06,
      "loss": 0.7366,
      "step": 12954
    },
    {
      "epoch": 0.3732783956664554,
      "grad_norm": 0.7469652891159058,
      "learning_rate": 7.215487953092998e-06,
      "loss": 0.7581,
      "step": 12955
    },
    {
      "epoch": 0.37330720912810467,
      "grad_norm": 0.7330608367919922,
      "learning_rate": 7.215069639888753e-06,
      "loss": 0.7147,
      "step": 12956
    },
    {
      "epoch": 0.37333602258975396,
      "grad_norm": 0.7558363676071167,
      "learning_rate": 7.214651307393466e-06,
      "loss": 0.7341,
      "step": 12957
    },
    {
      "epoch": 0.3733648360514032,
      "grad_norm": 0.7714803218841553,
      "learning_rate": 7.214232955610779e-06,
      "loss": 0.738,
      "step": 12958
    },
    {
      "epoch": 0.3733936495130525,
      "grad_norm": 0.7593832015991211,
      "learning_rate": 7.213814584544337e-06,
      "loss": 0.7341,
      "step": 12959
    },
    {
      "epoch": 0.3734224629747018,
      "grad_norm": 0.7914369106292725,
      "learning_rate": 7.213396194197784e-06,
      "loss": 0.7409,
      "step": 12960
    },
    {
      "epoch": 0.37345127643635107,
      "grad_norm": 0.7874299883842468,
      "learning_rate": 7.212977784574762e-06,
      "loss": 0.7229,
      "step": 12961
    },
    {
      "epoch": 0.37348008989800036,
      "grad_norm": 0.7720276117324829,
      "learning_rate": 7.212559355678915e-06,
      "loss": 0.7617,
      "step": 12962
    },
    {
      "epoch": 0.37350890335964965,
      "grad_norm": 0.7551112174987793,
      "learning_rate": 7.212140907513888e-06,
      "loss": 0.728,
      "step": 12963
    },
    {
      "epoch": 0.37353771682129894,
      "grad_norm": 0.7498865127563477,
      "learning_rate": 7.211722440083324e-06,
      "loss": 0.7366,
      "step": 12964
    },
    {
      "epoch": 0.3735665302829482,
      "grad_norm": 0.793864369392395,
      "learning_rate": 7.211303953390871e-06,
      "loss": 0.7467,
      "step": 12965
    },
    {
      "epoch": 0.37359534374459746,
      "grad_norm": 0.7875111699104309,
      "learning_rate": 7.21088544744017e-06,
      "loss": 0.7519,
      "step": 12966
    },
    {
      "epoch": 0.37362415720624675,
      "grad_norm": 0.7864309549331665,
      "learning_rate": 7.210466922234867e-06,
      "loss": 0.7636,
      "step": 12967
    },
    {
      "epoch": 0.37365297066789604,
      "grad_norm": 0.7951882481575012,
      "learning_rate": 7.210048377778605e-06,
      "loss": 0.7433,
      "step": 12968
    },
    {
      "epoch": 0.37368178412954534,
      "grad_norm": 0.7626104950904846,
      "learning_rate": 7.209629814075032e-06,
      "loss": 0.7479,
      "step": 12969
    },
    {
      "epoch": 0.3737105975911946,
      "grad_norm": 0.8197329640388489,
      "learning_rate": 7.209211231127792e-06,
      "loss": 0.7407,
      "step": 12970
    },
    {
      "epoch": 0.37373941105284386,
      "grad_norm": 1.030278205871582,
      "learning_rate": 7.20879262894053e-06,
      "loss": 0.7196,
      "step": 12971
    },
    {
      "epoch": 0.37376822451449315,
      "grad_norm": 0.7693921327590942,
      "learning_rate": 7.208374007516894e-06,
      "loss": 0.7689,
      "step": 12972
    },
    {
      "epoch": 0.37379703797614244,
      "grad_norm": 0.770916223526001,
      "learning_rate": 7.207955366860527e-06,
      "loss": 0.7145,
      "step": 12973
    },
    {
      "epoch": 0.37382585143779173,
      "grad_norm": 0.7680028676986694,
      "learning_rate": 7.207536706975077e-06,
      "loss": 0.7366,
      "step": 12974
    },
    {
      "epoch": 0.373854664899441,
      "grad_norm": 0.7868092060089111,
      "learning_rate": 7.207118027864188e-06,
      "loss": 0.7484,
      "step": 12975
    },
    {
      "epoch": 0.3738834783610903,
      "grad_norm": 0.7429144382476807,
      "learning_rate": 7.206699329531506e-06,
      "loss": 0.737,
      "step": 12976
    },
    {
      "epoch": 0.3739122918227396,
      "grad_norm": 0.7854464650154114,
      "learning_rate": 7.20628061198068e-06,
      "loss": 0.737,
      "step": 12977
    },
    {
      "epoch": 0.37394110528438884,
      "grad_norm": 0.7169559597969055,
      "learning_rate": 7.205861875215356e-06,
      "loss": 0.6951,
      "step": 12978
    },
    {
      "epoch": 0.37396991874603813,
      "grad_norm": 0.7691394686698914,
      "learning_rate": 7.205443119239179e-06,
      "loss": 0.723,
      "step": 12979
    },
    {
      "epoch": 0.3739987322076874,
      "grad_norm": 0.7918962240219116,
      "learning_rate": 7.205024344055797e-06,
      "loss": 0.7614,
      "step": 12980
    },
    {
      "epoch": 0.3740275456693367,
      "grad_norm": 0.7489340901374817,
      "learning_rate": 7.20460554966886e-06,
      "loss": 0.7329,
      "step": 12981
    },
    {
      "epoch": 0.374056359130986,
      "grad_norm": 0.7739359140396118,
      "learning_rate": 7.204186736082009e-06,
      "loss": 0.7389,
      "step": 12982
    },
    {
      "epoch": 0.3740851725926353,
      "grad_norm": 0.7575268745422363,
      "learning_rate": 7.2037679032988964e-06,
      "loss": 0.7514,
      "step": 12983
    },
    {
      "epoch": 0.3741139860542846,
      "grad_norm": 0.760890781879425,
      "learning_rate": 7.203349051323167e-06,
      "loss": 0.7394,
      "step": 12984
    },
    {
      "epoch": 0.3741427995159338,
      "grad_norm": 0.7855082750320435,
      "learning_rate": 7.202930180158471e-06,
      "loss": 0.7429,
      "step": 12985
    },
    {
      "epoch": 0.3741716129775831,
      "grad_norm": 0.7681463360786438,
      "learning_rate": 7.2025112898084555e-06,
      "loss": 0.7313,
      "step": 12986
    },
    {
      "epoch": 0.3742004264392324,
      "grad_norm": 0.7657806873321533,
      "learning_rate": 7.202092380276769e-06,
      "loss": 0.7436,
      "step": 12987
    },
    {
      "epoch": 0.3742292399008817,
      "grad_norm": 0.7845662236213684,
      "learning_rate": 7.201673451567058e-06,
      "loss": 0.7313,
      "step": 12988
    },
    {
      "epoch": 0.374258053362531,
      "grad_norm": 0.8000510334968567,
      "learning_rate": 7.201254503682971e-06,
      "loss": 0.7416,
      "step": 12989
    },
    {
      "epoch": 0.3742868668241803,
      "grad_norm": 0.7746292948722839,
      "learning_rate": 7.200835536628159e-06,
      "loss": 0.715,
      "step": 12990
    },
    {
      "epoch": 0.37431568028582957,
      "grad_norm": 0.8226252794265747,
      "learning_rate": 7.20041655040627e-06,
      "loss": 0.7379,
      "step": 12991
    },
    {
      "epoch": 0.3743444937474788,
      "grad_norm": 0.8024334907531738,
      "learning_rate": 7.199997545020952e-06,
      "loss": 0.7577,
      "step": 12992
    },
    {
      "epoch": 0.3743733072091281,
      "grad_norm": 0.7640321254730225,
      "learning_rate": 7.199578520475854e-06,
      "loss": 0.7253,
      "step": 12993
    },
    {
      "epoch": 0.3744021206707774,
      "grad_norm": 0.7829971313476562,
      "learning_rate": 7.1991594767746265e-06,
      "loss": 0.7527,
      "step": 12994
    },
    {
      "epoch": 0.3744309341324267,
      "grad_norm": 0.7890623807907104,
      "learning_rate": 7.198740413920918e-06,
      "loss": 0.7266,
      "step": 12995
    },
    {
      "epoch": 0.37445974759407596,
      "grad_norm": 0.7598782181739807,
      "learning_rate": 7.198321331918378e-06,
      "loss": 0.7367,
      "step": 12996
    },
    {
      "epoch": 0.37448856105572526,
      "grad_norm": 0.7496626377105713,
      "learning_rate": 7.197902230770658e-06,
      "loss": 0.7202,
      "step": 12997
    },
    {
      "epoch": 0.3745173745173745,
      "grad_norm": 0.7647709250450134,
      "learning_rate": 7.197483110481407e-06,
      "loss": 0.75,
      "step": 12998
    },
    {
      "epoch": 0.3745461879790238,
      "grad_norm": 0.7564080953598022,
      "learning_rate": 7.197063971054274e-06,
      "loss": 0.7128,
      "step": 12999
    },
    {
      "epoch": 0.37457500144067307,
      "grad_norm": 0.7531507611274719,
      "learning_rate": 7.196644812492909e-06,
      "loss": 0.7466,
      "step": 13000
    },
    {
      "epoch": 0.37460381490232236,
      "grad_norm": 0.7473645806312561,
      "learning_rate": 7.196225634800964e-06,
      "loss": 0.731,
      "step": 13001
    },
    {
      "epoch": 0.37463262836397165,
      "grad_norm": 0.814231812953949,
      "learning_rate": 7.19580643798209e-06,
      "loss": 0.7383,
      "step": 13002
    },
    {
      "epoch": 0.37466144182562094,
      "grad_norm": 0.7832647562026978,
      "learning_rate": 7.195387222039937e-06,
      "loss": 0.7249,
      "step": 13003
    },
    {
      "epoch": 0.37469025528727024,
      "grad_norm": 0.7858816385269165,
      "learning_rate": 7.194967986978156e-06,
      "loss": 0.7548,
      "step": 13004
    },
    {
      "epoch": 0.37471906874891947,
      "grad_norm": 0.7644627690315247,
      "learning_rate": 7.1945487328003975e-06,
      "loss": 0.7361,
      "step": 13005
    },
    {
      "epoch": 0.37474788221056876,
      "grad_norm": 0.7455390095710754,
      "learning_rate": 7.194129459510312e-06,
      "loss": 0.7199,
      "step": 13006
    },
    {
      "epoch": 0.37477669567221805,
      "grad_norm": 0.7702011466026306,
      "learning_rate": 7.193710167111554e-06,
      "loss": 0.7477,
      "step": 13007
    },
    {
      "epoch": 0.37480550913386734,
      "grad_norm": 0.7770794630050659,
      "learning_rate": 7.1932908556077705e-06,
      "loss": 0.7229,
      "step": 13008
    },
    {
      "epoch": 0.37483432259551663,
      "grad_norm": 0.7796345949172974,
      "learning_rate": 7.192871525002619e-06,
      "loss": 0.7719,
      "step": 13009
    },
    {
      "epoch": 0.3748631360571659,
      "grad_norm": 0.8315966725349426,
      "learning_rate": 7.192452175299749e-06,
      "loss": 0.7341,
      "step": 13010
    },
    {
      "epoch": 0.3748919495188152,
      "grad_norm": 0.8770843744277954,
      "learning_rate": 7.192032806502811e-06,
      "loss": 0.7512,
      "step": 13011
    },
    {
      "epoch": 0.37492076298046445,
      "grad_norm": 0.7770740389823914,
      "learning_rate": 7.191613418615458e-06,
      "loss": 0.7432,
      "step": 13012
    },
    {
      "epoch": 0.37494957644211374,
      "grad_norm": 0.7424713969230652,
      "learning_rate": 7.191194011641344e-06,
      "loss": 0.7249,
      "step": 13013
    },
    {
      "epoch": 0.37497838990376303,
      "grad_norm": 0.755680501461029,
      "learning_rate": 7.190774585584119e-06,
      "loss": 0.7313,
      "step": 13014
    },
    {
      "epoch": 0.3750072033654123,
      "grad_norm": 0.8062906265258789,
      "learning_rate": 7.190355140447439e-06,
      "loss": 0.7372,
      "step": 13015
    },
    {
      "epoch": 0.3750360168270616,
      "grad_norm": 0.7710446715354919,
      "learning_rate": 7.189935676234956e-06,
      "loss": 0.7503,
      "step": 13016
    },
    {
      "epoch": 0.3750648302887109,
      "grad_norm": 0.7576496601104736,
      "learning_rate": 7.189516192950321e-06,
      "loss": 0.7108,
      "step": 13017
    },
    {
      "epoch": 0.3750936437503602,
      "grad_norm": 0.764728844165802,
      "learning_rate": 7.189096690597188e-06,
      "loss": 0.7357,
      "step": 13018
    },
    {
      "epoch": 0.37512245721200943,
      "grad_norm": 0.7361440062522888,
      "learning_rate": 7.188677169179213e-06,
      "loss": 0.7629,
      "step": 13019
    },
    {
      "epoch": 0.3751512706736587,
      "grad_norm": 0.7812852263450623,
      "learning_rate": 7.188257628700046e-06,
      "loss": 0.7424,
      "step": 13020
    },
    {
      "epoch": 0.375180084135308,
      "grad_norm": 0.7361230254173279,
      "learning_rate": 7.1878380691633435e-06,
      "loss": 0.7221,
      "step": 13021
    },
    {
      "epoch": 0.3752088975969573,
      "grad_norm": 0.7421243786811829,
      "learning_rate": 7.1874184905727585e-06,
      "loss": 0.7399,
      "step": 13022
    },
    {
      "epoch": 0.3752377110586066,
      "grad_norm": 0.7625911235809326,
      "learning_rate": 7.186998892931944e-06,
      "loss": 0.7348,
      "step": 13023
    },
    {
      "epoch": 0.3752665245202559,
      "grad_norm": 0.7681647539138794,
      "learning_rate": 7.186579276244556e-06,
      "loss": 0.757,
      "step": 13024
    },
    {
      "epoch": 0.3752953379819051,
      "grad_norm": 0.7814731001853943,
      "learning_rate": 7.186159640514249e-06,
      "loss": 0.7597,
      "step": 13025
    },
    {
      "epoch": 0.3753241514435544,
      "grad_norm": 0.8093010783195496,
      "learning_rate": 7.185739985744675e-06,
      "loss": 0.7291,
      "step": 13026
    },
    {
      "epoch": 0.3753529649052037,
      "grad_norm": 0.7854681015014648,
      "learning_rate": 7.185320311939492e-06,
      "loss": 0.7258,
      "step": 13027
    },
    {
      "epoch": 0.375381778366853,
      "grad_norm": 0.7429960370063782,
      "learning_rate": 7.184900619102353e-06,
      "loss": 0.7353,
      "step": 13028
    },
    {
      "epoch": 0.3754105918285023,
      "grad_norm": 0.7550674676895142,
      "learning_rate": 7.184480907236915e-06,
      "loss": 0.7378,
      "step": 13029
    },
    {
      "epoch": 0.3754394052901516,
      "grad_norm": 0.7394548654556274,
      "learning_rate": 7.1840611763468305e-06,
      "loss": 0.7423,
      "step": 13030
    },
    {
      "epoch": 0.37546821875180086,
      "grad_norm": 0.7431045770645142,
      "learning_rate": 7.183641426435756e-06,
      "loss": 0.7259,
      "step": 13031
    },
    {
      "epoch": 0.3754970322134501,
      "grad_norm": 0.7603760361671448,
      "learning_rate": 7.18322165750735e-06,
      "loss": 0.7205,
      "step": 13032
    },
    {
      "epoch": 0.3755258456750994,
      "grad_norm": 0.7438333630561829,
      "learning_rate": 7.182801869565265e-06,
      "loss": 0.7354,
      "step": 13033
    },
    {
      "epoch": 0.3755546591367487,
      "grad_norm": 0.7679675817489624,
      "learning_rate": 7.1823820626131565e-06,
      "loss": 0.7453,
      "step": 13034
    },
    {
      "epoch": 0.37558347259839797,
      "grad_norm": 0.7786396145820618,
      "learning_rate": 7.181962236654683e-06,
      "loss": 0.7329,
      "step": 13035
    },
    {
      "epoch": 0.37561228606004726,
      "grad_norm": 0.7655700445175171,
      "learning_rate": 7.181542391693498e-06,
      "loss": 0.7267,
      "step": 13036
    },
    {
      "epoch": 0.37564109952169655,
      "grad_norm": 0.7624366879463196,
      "learning_rate": 7.181122527733261e-06,
      "loss": 0.7335,
      "step": 13037
    },
    {
      "epoch": 0.37566991298334584,
      "grad_norm": 0.7735892534255981,
      "learning_rate": 7.180702644777627e-06,
      "loss": 0.7382,
      "step": 13038
    },
    {
      "epoch": 0.3756987264449951,
      "grad_norm": 0.7252569198608398,
      "learning_rate": 7.180282742830252e-06,
      "loss": 0.7121,
      "step": 13039
    },
    {
      "epoch": 0.37572753990664437,
      "grad_norm": 0.7612737417221069,
      "learning_rate": 7.179862821894796e-06,
      "loss": 0.7686,
      "step": 13040
    },
    {
      "epoch": 0.37575635336829366,
      "grad_norm": 0.7744590640068054,
      "learning_rate": 7.1794428819749115e-06,
      "loss": 0.7587,
      "step": 13041
    },
    {
      "epoch": 0.37578516682994295,
      "grad_norm": 0.7674844861030579,
      "learning_rate": 7.179022923074258e-06,
      "loss": 0.7504,
      "step": 13042
    },
    {
      "epoch": 0.37581398029159224,
      "grad_norm": 0.7501981258392334,
      "learning_rate": 7.178602945196493e-06,
      "loss": 0.7069,
      "step": 13043
    },
    {
      "epoch": 0.37584279375324153,
      "grad_norm": 0.8184661269187927,
      "learning_rate": 7.178182948345276e-06,
      "loss": 0.7315,
      "step": 13044
    },
    {
      "epoch": 0.3758716072148908,
      "grad_norm": 0.7674031257629395,
      "learning_rate": 7.1777629325242625e-06,
      "loss": 0.7382,
      "step": 13045
    },
    {
      "epoch": 0.37590042067654006,
      "grad_norm": 0.737325131893158,
      "learning_rate": 7.177342897737109e-06,
      "loss": 0.7545,
      "step": 13046
    },
    {
      "epoch": 0.37592923413818935,
      "grad_norm": 0.7867429852485657,
      "learning_rate": 7.176922843987478e-06,
      "loss": 0.7534,
      "step": 13047
    },
    {
      "epoch": 0.37595804759983864,
      "grad_norm": 0.7417970895767212,
      "learning_rate": 7.1765027712790235e-06,
      "loss": 0.6976,
      "step": 13048
    },
    {
      "epoch": 0.37598686106148793,
      "grad_norm": 0.7720916867256165,
      "learning_rate": 7.176082679615405e-06,
      "loss": 0.7196,
      "step": 13049
    },
    {
      "epoch": 0.3760156745231372,
      "grad_norm": 0.7712863087654114,
      "learning_rate": 7.175662569000283e-06,
      "loss": 0.7446,
      "step": 13050
    },
    {
      "epoch": 0.3760444879847865,
      "grad_norm": 0.7498886585235596,
      "learning_rate": 7.1752424394373145e-06,
      "loss": 0.7196,
      "step": 13051
    },
    {
      "epoch": 0.37607330144643575,
      "grad_norm": 0.8101080060005188,
      "learning_rate": 7.174822290930159e-06,
      "loss": 0.7284,
      "step": 13052
    },
    {
      "epoch": 0.37610211490808504,
      "grad_norm": 0.7546941041946411,
      "learning_rate": 7.174402123482475e-06,
      "loss": 0.7298,
      "step": 13053
    },
    {
      "epoch": 0.37613092836973433,
      "grad_norm": 0.7768452763557434,
      "learning_rate": 7.1739819370979215e-06,
      "loss": 0.737,
      "step": 13054
    },
    {
      "epoch": 0.3761597418313836,
      "grad_norm": 0.7682976126670837,
      "learning_rate": 7.1735617317801584e-06,
      "loss": 0.7473,
      "step": 13055
    },
    {
      "epoch": 0.3761885552930329,
      "grad_norm": 0.7806897163391113,
      "learning_rate": 7.173141507532847e-06,
      "loss": 0.7112,
      "step": 13056
    },
    {
      "epoch": 0.3762173687546822,
      "grad_norm": 0.8033234477043152,
      "learning_rate": 7.172721264359646e-06,
      "loss": 0.7553,
      "step": 13057
    },
    {
      "epoch": 0.3762461822163315,
      "grad_norm": 0.7928246259689331,
      "learning_rate": 7.172301002264212e-06,
      "loss": 0.7422,
      "step": 13058
    },
    {
      "epoch": 0.37627499567798073,
      "grad_norm": 0.7834072113037109,
      "learning_rate": 7.171880721250211e-06,
      "loss": 0.7264,
      "step": 13059
    },
    {
      "epoch": 0.37630380913963,
      "grad_norm": 0.8376430869102478,
      "learning_rate": 7.171460421321297e-06,
      "loss": 0.7529,
      "step": 13060
    },
    {
      "epoch": 0.3763326226012793,
      "grad_norm": 0.7631802558898926,
      "learning_rate": 7.171040102481134e-06,
      "loss": 0.732,
      "step": 13061
    },
    {
      "epoch": 0.3763614360629286,
      "grad_norm": 0.8361687064170837,
      "learning_rate": 7.170619764733383e-06,
      "loss": 0.7516,
      "step": 13062
    },
    {
      "epoch": 0.3763902495245779,
      "grad_norm": 0.7920535802841187,
      "learning_rate": 7.170199408081703e-06,
      "loss": 0.7607,
      "step": 13063
    },
    {
      "epoch": 0.3764190629862272,
      "grad_norm": 0.7645507454872131,
      "learning_rate": 7.169779032529755e-06,
      "loss": 0.7037,
      "step": 13064
    },
    {
      "epoch": 0.3764478764478765,
      "grad_norm": 0.7308569550514221,
      "learning_rate": 7.169358638081202e-06,
      "loss": 0.7237,
      "step": 13065
    },
    {
      "epoch": 0.3764766899095257,
      "grad_norm": 0.7562464475631714,
      "learning_rate": 7.1689382247397035e-06,
      "loss": 0.7464,
      "step": 13066
    },
    {
      "epoch": 0.376505503371175,
      "grad_norm": 0.7887590527534485,
      "learning_rate": 7.16851779250892e-06,
      "loss": 0.7175,
      "step": 13067
    },
    {
      "epoch": 0.3765343168328243,
      "grad_norm": 0.7932326197624207,
      "learning_rate": 7.168097341392515e-06,
      "loss": 0.7653,
      "step": 13068
    },
    {
      "epoch": 0.3765631302944736,
      "grad_norm": 0.762687623500824,
      "learning_rate": 7.1676768713941495e-06,
      "loss": 0.7164,
      "step": 13069
    },
    {
      "epoch": 0.37659194375612287,
      "grad_norm": 0.7827132344245911,
      "learning_rate": 7.167256382517486e-06,
      "loss": 0.7566,
      "step": 13070
    },
    {
      "epoch": 0.37662075721777216,
      "grad_norm": 0.7758330702781677,
      "learning_rate": 7.1668358747661845e-06,
      "loss": 0.7427,
      "step": 13071
    },
    {
      "epoch": 0.37664957067942145,
      "grad_norm": 0.7823618650436401,
      "learning_rate": 7.166415348143909e-06,
      "loss": 0.7628,
      "step": 13072
    },
    {
      "epoch": 0.3766783841410707,
      "grad_norm": 0.8055115938186646,
      "learning_rate": 7.165994802654323e-06,
      "loss": 0.7807,
      "step": 13073
    },
    {
      "epoch": 0.37670719760272,
      "grad_norm": 0.7908453941345215,
      "learning_rate": 7.165574238301085e-06,
      "loss": 0.7202,
      "step": 13074
    },
    {
      "epoch": 0.37673601106436927,
      "grad_norm": 0.7597435116767883,
      "learning_rate": 7.165153655087862e-06,
      "loss": 0.744,
      "step": 13075
    },
    {
      "epoch": 0.37676482452601856,
      "grad_norm": 0.7846789956092834,
      "learning_rate": 7.164733053018314e-06,
      "loss": 0.727,
      "step": 13076
    },
    {
      "epoch": 0.37679363798766785,
      "grad_norm": 0.7563250660896301,
      "learning_rate": 7.1643124320961064e-06,
      "loss": 0.7232,
      "step": 13077
    },
    {
      "epoch": 0.37682245144931714,
      "grad_norm": 0.7510699033737183,
      "learning_rate": 7.1638917923249e-06,
      "loss": 0.7329,
      "step": 13078
    },
    {
      "epoch": 0.3768512649109664,
      "grad_norm": 0.7557149529457092,
      "learning_rate": 7.1634711337083596e-06,
      "loss": 0.7243,
      "step": 13079
    },
    {
      "epoch": 0.37688007837261567,
      "grad_norm": 0.7868647575378418,
      "learning_rate": 7.163050456250148e-06,
      "loss": 0.7541,
      "step": 13080
    },
    {
      "epoch": 0.37690889183426496,
      "grad_norm": 1.6033551692962646,
      "learning_rate": 7.162629759953929e-06,
      "loss": 0.7541,
      "step": 13081
    },
    {
      "epoch": 0.37693770529591425,
      "grad_norm": 0.766011655330658,
      "learning_rate": 7.162209044823367e-06,
      "loss": 0.7446,
      "step": 13082
    },
    {
      "epoch": 0.37696651875756354,
      "grad_norm": 0.8151608109474182,
      "learning_rate": 7.161788310862125e-06,
      "loss": 0.717,
      "step": 13083
    },
    {
      "epoch": 0.37699533221921283,
      "grad_norm": 0.7748329043388367,
      "learning_rate": 7.161367558073868e-06,
      "loss": 0.7177,
      "step": 13084
    },
    {
      "epoch": 0.3770241456808621,
      "grad_norm": 0.762808620929718,
      "learning_rate": 7.160946786462262e-06,
      "loss": 0.7327,
      "step": 13085
    },
    {
      "epoch": 0.37705295914251136,
      "grad_norm": 0.7827445864677429,
      "learning_rate": 7.160525996030967e-06,
      "loss": 0.7388,
      "step": 13086
    },
    {
      "epoch": 0.37708177260416065,
      "grad_norm": 0.8154492378234863,
      "learning_rate": 7.160105186783651e-06,
      "loss": 0.7298,
      "step": 13087
    },
    {
      "epoch": 0.37711058606580994,
      "grad_norm": 0.8154090642929077,
      "learning_rate": 7.159684358723978e-06,
      "loss": 0.7601,
      "step": 13088
    },
    {
      "epoch": 0.37713939952745923,
      "grad_norm": 0.7501934766769409,
      "learning_rate": 7.159263511855614e-06,
      "loss": 0.7525,
      "step": 13089
    },
    {
      "epoch": 0.3771682129891085,
      "grad_norm": 0.8016427755355835,
      "learning_rate": 7.158842646182222e-06,
      "loss": 0.7175,
      "step": 13090
    },
    {
      "epoch": 0.3771970264507578,
      "grad_norm": 0.8063896298408508,
      "learning_rate": 7.15842176170747e-06,
      "loss": 0.7311,
      "step": 13091
    },
    {
      "epoch": 0.3772258399124071,
      "grad_norm": 0.7950237989425659,
      "learning_rate": 7.158000858435023e-06,
      "loss": 0.7533,
      "step": 13092
    },
    {
      "epoch": 0.37725465337405634,
      "grad_norm": 0.8332023620605469,
      "learning_rate": 7.157579936368544e-06,
      "loss": 0.7516,
      "step": 13093
    },
    {
      "epoch": 0.3772834668357056,
      "grad_norm": 0.7966944575309753,
      "learning_rate": 7.157158995511701e-06,
      "loss": 0.7514,
      "step": 13094
    },
    {
      "epoch": 0.3773122802973549,
      "grad_norm": 0.8009659647941589,
      "learning_rate": 7.15673803586816e-06,
      "loss": 0.7482,
      "step": 13095
    },
    {
      "epoch": 0.3773410937590042,
      "grad_norm": 0.7498139142990112,
      "learning_rate": 7.156317057441585e-06,
      "loss": 0.7082,
      "step": 13096
    },
    {
      "epoch": 0.3773699072206535,
      "grad_norm": 0.7611503601074219,
      "learning_rate": 7.1558960602356455e-06,
      "loss": 0.7277,
      "step": 13097
    },
    {
      "epoch": 0.3773987206823028,
      "grad_norm": 1.0645549297332764,
      "learning_rate": 7.155475044254006e-06,
      "loss": 0.7345,
      "step": 13098
    },
    {
      "epoch": 0.3774275341439521,
      "grad_norm": 0.7815979719161987,
      "learning_rate": 7.155054009500334e-06,
      "loss": 0.7359,
      "step": 13099
    },
    {
      "epoch": 0.3774563476056013,
      "grad_norm": 0.802670419216156,
      "learning_rate": 7.154632955978295e-06,
      "loss": 0.7566,
      "step": 13100
    },
    {
      "epoch": 0.3774851610672506,
      "grad_norm": 0.7574096322059631,
      "learning_rate": 7.1542118836915576e-06,
      "loss": 0.7495,
      "step": 13101
    },
    {
      "epoch": 0.3775139745288999,
      "grad_norm": 0.7809908390045166,
      "learning_rate": 7.153790792643787e-06,
      "loss": 0.7254,
      "step": 13102
    },
    {
      "epoch": 0.3775427879905492,
      "grad_norm": 0.8527669310569763,
      "learning_rate": 7.153369682838652e-06,
      "loss": 0.7356,
      "step": 13103
    },
    {
      "epoch": 0.3775716014521985,
      "grad_norm": 0.7726868391036987,
      "learning_rate": 7.15294855427982e-06,
      "loss": 0.7305,
      "step": 13104
    },
    {
      "epoch": 0.37760041491384777,
      "grad_norm": 0.7706466317176819,
      "learning_rate": 7.152527406970958e-06,
      "loss": 0.7245,
      "step": 13105
    },
    {
      "epoch": 0.377629228375497,
      "grad_norm": 0.8232429027557373,
      "learning_rate": 7.152106240915735e-06,
      "loss": 0.7377,
      "step": 13106
    },
    {
      "epoch": 0.3776580418371463,
      "grad_norm": 0.7859817147254944,
      "learning_rate": 7.151685056117816e-06,
      "loss": 0.7435,
      "step": 13107
    },
    {
      "epoch": 0.3776868552987956,
      "grad_norm": 0.7483980655670166,
      "learning_rate": 7.151263852580872e-06,
      "loss": 0.7478,
      "step": 13108
    },
    {
      "epoch": 0.3777156687604449,
      "grad_norm": 0.7807617783546448,
      "learning_rate": 7.150842630308571e-06,
      "loss": 0.7562,
      "step": 13109
    },
    {
      "epoch": 0.37774448222209417,
      "grad_norm": 0.7801811099052429,
      "learning_rate": 7.15042138930458e-06,
      "loss": 0.7436,
      "step": 13110
    },
    {
      "epoch": 0.37777329568374346,
      "grad_norm": 0.749826192855835,
      "learning_rate": 7.15000012957257e-06,
      "loss": 0.6988,
      "step": 13111
    },
    {
      "epoch": 0.37780210914539275,
      "grad_norm": 0.812320351600647,
      "learning_rate": 7.149578851116206e-06,
      "loss": 0.7309,
      "step": 13112
    },
    {
      "epoch": 0.377830922607042,
      "grad_norm": 0.7750247716903687,
      "learning_rate": 7.14915755393916e-06,
      "loss": 0.7397,
      "step": 13113
    },
    {
      "epoch": 0.3778597360686913,
      "grad_norm": 1.240710973739624,
      "learning_rate": 7.148736238045099e-06,
      "loss": 0.7285,
      "step": 13114
    },
    {
      "epoch": 0.37788854953034057,
      "grad_norm": 0.7867616415023804,
      "learning_rate": 7.148314903437694e-06,
      "loss": 0.7434,
      "step": 13115
    },
    {
      "epoch": 0.37791736299198986,
      "grad_norm": 0.7909741401672363,
      "learning_rate": 7.147893550120614e-06,
      "loss": 0.7402,
      "step": 13116
    },
    {
      "epoch": 0.37794617645363915,
      "grad_norm": 0.7971020936965942,
      "learning_rate": 7.1474721780975285e-06,
      "loss": 0.7247,
      "step": 13117
    },
    {
      "epoch": 0.37797498991528844,
      "grad_norm": 0.7600317001342773,
      "learning_rate": 7.1470507873721075e-06,
      "loss": 0.739,
      "step": 13118
    },
    {
      "epoch": 0.37800380337693773,
      "grad_norm": 0.745133101940155,
      "learning_rate": 7.146629377948019e-06,
      "loss": 0.7289,
      "step": 13119
    },
    {
      "epoch": 0.37803261683858697,
      "grad_norm": 0.7645416855812073,
      "learning_rate": 7.146207949828934e-06,
      "loss": 0.7508,
      "step": 13120
    },
    {
      "epoch": 0.37806143030023626,
      "grad_norm": 0.7349223494529724,
      "learning_rate": 7.145786503018525e-06,
      "loss": 0.7435,
      "step": 13121
    },
    {
      "epoch": 0.37809024376188555,
      "grad_norm": 0.7660600543022156,
      "learning_rate": 7.14536503752046e-06,
      "loss": 0.7652,
      "step": 13122
    },
    {
      "epoch": 0.37811905722353484,
      "grad_norm": 0.7506208419799805,
      "learning_rate": 7.144943553338411e-06,
      "loss": 0.7498,
      "step": 13123
    },
    {
      "epoch": 0.37814787068518413,
      "grad_norm": 0.7348117232322693,
      "learning_rate": 7.1445220504760445e-06,
      "loss": 0.7254,
      "step": 13124
    },
    {
      "epoch": 0.3781766841468334,
      "grad_norm": 0.7961315512657166,
      "learning_rate": 7.144100528937037e-06,
      "loss": 0.7329,
      "step": 13125
    },
    {
      "epoch": 0.3782054976084827,
      "grad_norm": 0.7779295444488525,
      "learning_rate": 7.143678988725058e-06,
      "loss": 0.7384,
      "step": 13126
    },
    {
      "epoch": 0.37823431107013195,
      "grad_norm": 0.7551276087760925,
      "learning_rate": 7.143257429843775e-06,
      "loss": 0.7427,
      "step": 13127
    },
    {
      "epoch": 0.37826312453178124,
      "grad_norm": 0.7864892482757568,
      "learning_rate": 7.142835852296865e-06,
      "loss": 0.7631,
      "step": 13128
    },
    {
      "epoch": 0.3782919379934305,
      "grad_norm": 2.8585731983184814,
      "learning_rate": 7.142414256087996e-06,
      "loss": 0.8434,
      "step": 13129
    },
    {
      "epoch": 0.3783207514550798,
      "grad_norm": 0.7682065963745117,
      "learning_rate": 7.14199264122084e-06,
      "loss": 0.7347,
      "step": 13130
    },
    {
      "epoch": 0.3783495649167291,
      "grad_norm": 0.7631239295005798,
      "learning_rate": 7.141571007699071e-06,
      "loss": 0.7356,
      "step": 13131
    },
    {
      "epoch": 0.3783783783783784,
      "grad_norm": 0.7862837910652161,
      "learning_rate": 7.141149355526357e-06,
      "loss": 0.7325,
      "step": 13132
    },
    {
      "epoch": 0.37840719184002763,
      "grad_norm": 0.7951736450195312,
      "learning_rate": 7.140727684706373e-06,
      "loss": 0.7211,
      "step": 13133
    },
    {
      "epoch": 0.3784360053016769,
      "grad_norm": 0.7676878571510315,
      "learning_rate": 7.1403059952427926e-06,
      "loss": 0.7294,
      "step": 13134
    },
    {
      "epoch": 0.3784648187633262,
      "grad_norm": 0.8045559525489807,
      "learning_rate": 7.139884287139285e-06,
      "loss": 0.7624,
      "step": 13135
    },
    {
      "epoch": 0.3784936322249755,
      "grad_norm": 0.8248607516288757,
      "learning_rate": 7.139462560399525e-06,
      "loss": 0.7213,
      "step": 13136
    },
    {
      "epoch": 0.3785224456866248,
      "grad_norm": 0.8119875192642212,
      "learning_rate": 7.139040815027185e-06,
      "loss": 0.7226,
      "step": 13137
    },
    {
      "epoch": 0.3785512591482741,
      "grad_norm": 0.8054575324058533,
      "learning_rate": 7.138619051025936e-06,
      "loss": 0.731,
      "step": 13138
    },
    {
      "epoch": 0.3785800726099234,
      "grad_norm": 2.6573121547698975,
      "learning_rate": 7.1381972683994535e-06,
      "loss": 0.7581,
      "step": 13139
    },
    {
      "epoch": 0.3786088860715726,
      "grad_norm": 0.7912265658378601,
      "learning_rate": 7.137775467151411e-06,
      "loss": 0.7242,
      "step": 13140
    },
    {
      "epoch": 0.3786376995332219,
      "grad_norm": 0.7906114459037781,
      "learning_rate": 7.137353647285481e-06,
      "loss": 0.7468,
      "step": 13141
    },
    {
      "epoch": 0.3786665129948712,
      "grad_norm": 0.7430979609489441,
      "learning_rate": 7.136931808805338e-06,
      "loss": 0.7265,
      "step": 13142
    },
    {
      "epoch": 0.3786953264565205,
      "grad_norm": 0.7599256634712219,
      "learning_rate": 7.136509951714654e-06,
      "loss": 0.7328,
      "step": 13143
    },
    {
      "epoch": 0.3787241399181698,
      "grad_norm": 0.7740522027015686,
      "learning_rate": 7.136088076017105e-06,
      "loss": 0.7346,
      "step": 13144
    },
    {
      "epoch": 0.37875295337981907,
      "grad_norm": 0.8082802891731262,
      "learning_rate": 7.135666181716363e-06,
      "loss": 0.7164,
      "step": 13145
    },
    {
      "epoch": 0.37878176684146836,
      "grad_norm": 0.8434605598449707,
      "learning_rate": 7.135244268816103e-06,
      "loss": 0.7444,
      "step": 13146
    },
    {
      "epoch": 0.3788105803031176,
      "grad_norm": 0.7976865768432617,
      "learning_rate": 7.134822337320001e-06,
      "loss": 0.752,
      "step": 13147
    },
    {
      "epoch": 0.3788393937647669,
      "grad_norm": 0.8268904089927673,
      "learning_rate": 7.13440038723173e-06,
      "loss": 0.7533,
      "step": 13148
    },
    {
      "epoch": 0.3788682072264162,
      "grad_norm": 0.7901270389556885,
      "learning_rate": 7.133978418554965e-06,
      "loss": 0.7498,
      "step": 13149
    },
    {
      "epoch": 0.37889702068806547,
      "grad_norm": 0.7754499316215515,
      "learning_rate": 7.133556431293382e-06,
      "loss": 0.7551,
      "step": 13150
    },
    {
      "epoch": 0.37892583414971476,
      "grad_norm": 0.785685658454895,
      "learning_rate": 7.1331344254506565e-06,
      "loss": 0.7231,
      "step": 13151
    },
    {
      "epoch": 0.37895464761136405,
      "grad_norm": 0.8199127912521362,
      "learning_rate": 7.132712401030461e-06,
      "loss": 0.7518,
      "step": 13152
    },
    {
      "epoch": 0.37898346107301334,
      "grad_norm": 0.7655794024467468,
      "learning_rate": 7.132290358036472e-06,
      "loss": 0.7486,
      "step": 13153
    },
    {
      "epoch": 0.3790122745346626,
      "grad_norm": 0.7613972425460815,
      "learning_rate": 7.131868296472366e-06,
      "loss": 0.7531,
      "step": 13154
    },
    {
      "epoch": 0.37904108799631187,
      "grad_norm": 0.7928082346916199,
      "learning_rate": 7.1314462163418176e-06,
      "loss": 0.7402,
      "step": 13155
    },
    {
      "epoch": 0.37906990145796116,
      "grad_norm": 0.7626936435699463,
      "learning_rate": 7.131024117648504e-06,
      "loss": 0.7168,
      "step": 13156
    },
    {
      "epoch": 0.37909871491961045,
      "grad_norm": 0.7567980289459229,
      "learning_rate": 7.130602000396101e-06,
      "loss": 0.7232,
      "step": 13157
    },
    {
      "epoch": 0.37912752838125974,
      "grad_norm": 0.8267423510551453,
      "learning_rate": 7.130179864588284e-06,
      "loss": 0.7642,
      "step": 13158
    },
    {
      "epoch": 0.37915634184290903,
      "grad_norm": 0.8319969177246094,
      "learning_rate": 7.129757710228729e-06,
      "loss": 0.741,
      "step": 13159
    },
    {
      "epoch": 0.37918515530455826,
      "grad_norm": 0.7491408586502075,
      "learning_rate": 7.1293355373211146e-06,
      "loss": 0.7399,
      "step": 13160
    },
    {
      "epoch": 0.37921396876620755,
      "grad_norm": 0.7243315577507019,
      "learning_rate": 7.128913345869115e-06,
      "loss": 0.7219,
      "step": 13161
    },
    {
      "epoch": 0.37924278222785685,
      "grad_norm": 0.7465955018997192,
      "learning_rate": 7.12849113587641e-06,
      "loss": 0.7528,
      "step": 13162
    },
    {
      "epoch": 0.37927159568950614,
      "grad_norm": 0.7857682704925537,
      "learning_rate": 7.128068907346672e-06,
      "loss": 0.7269,
      "step": 13163
    },
    {
      "epoch": 0.3793004091511554,
      "grad_norm": 0.764539897441864,
      "learning_rate": 7.1276466602835845e-06,
      "loss": 0.7279,
      "step": 13164
    },
    {
      "epoch": 0.3793292226128047,
      "grad_norm": 0.7414140105247498,
      "learning_rate": 7.127224394690819e-06,
      "loss": 0.7332,
      "step": 13165
    },
    {
      "epoch": 0.379358036074454,
      "grad_norm": 0.7789487242698669,
      "learning_rate": 7.126802110572057e-06,
      "loss": 0.7555,
      "step": 13166
    },
    {
      "epoch": 0.37938684953610324,
      "grad_norm": 0.7704275250434875,
      "learning_rate": 7.126379807930972e-06,
      "loss": 0.7395,
      "step": 13167
    },
    {
      "epoch": 0.37941566299775253,
      "grad_norm": 0.7718679308891296,
      "learning_rate": 7.1259574867712466e-06,
      "loss": 0.7399,
      "step": 13168
    },
    {
      "epoch": 0.3794444764594018,
      "grad_norm": 0.8074487447738647,
      "learning_rate": 7.125535147096557e-06,
      "loss": 0.7388,
      "step": 13169
    },
    {
      "epoch": 0.3794732899210511,
      "grad_norm": 0.7749435901641846,
      "learning_rate": 7.125112788910581e-06,
      "loss": 0.764,
      "step": 13170
    },
    {
      "epoch": 0.3795021033827004,
      "grad_norm": 0.7698729038238525,
      "learning_rate": 7.124690412216997e-06,
      "loss": 0.7287,
      "step": 13171
    },
    {
      "epoch": 0.3795309168443497,
      "grad_norm": 0.758610725402832,
      "learning_rate": 7.124268017019483e-06,
      "loss": 0.7386,
      "step": 13172
    },
    {
      "epoch": 0.379559730305999,
      "grad_norm": 0.7617526650428772,
      "learning_rate": 7.123845603321717e-06,
      "loss": 0.7291,
      "step": 13173
    },
    {
      "epoch": 0.3795885437676482,
      "grad_norm": 0.7951836585998535,
      "learning_rate": 7.123423171127381e-06,
      "loss": 0.7515,
      "step": 13174
    },
    {
      "epoch": 0.3796173572292975,
      "grad_norm": 0.7494876384735107,
      "learning_rate": 7.12300072044015e-06,
      "loss": 0.717,
      "step": 13175
    },
    {
      "epoch": 0.3796461706909468,
      "grad_norm": 0.7592732906341553,
      "learning_rate": 7.122578251263705e-06,
      "loss": 0.7299,
      "step": 13176
    },
    {
      "epoch": 0.3796749841525961,
      "grad_norm": 0.7940019369125366,
      "learning_rate": 7.122155763601727e-06,
      "loss": 0.7333,
      "step": 13177
    },
    {
      "epoch": 0.3797037976142454,
      "grad_norm": 0.8047069311141968,
      "learning_rate": 7.121733257457892e-06,
      "loss": 0.7319,
      "step": 13178
    },
    {
      "epoch": 0.3797326110758947,
      "grad_norm": 0.7894076108932495,
      "learning_rate": 7.121310732835882e-06,
      "loss": 0.709,
      "step": 13179
    },
    {
      "epoch": 0.37976142453754397,
      "grad_norm": 0.7831112146377563,
      "learning_rate": 7.120888189739375e-06,
      "loss": 0.7378,
      "step": 13180
    },
    {
      "epoch": 0.3797902379991932,
      "grad_norm": 0.7723489999771118,
      "learning_rate": 7.120465628172053e-06,
      "loss": 0.7237,
      "step": 13181
    },
    {
      "epoch": 0.3798190514608425,
      "grad_norm": 0.8654604554176331,
      "learning_rate": 7.120043048137596e-06,
      "loss": 0.7397,
      "step": 13182
    },
    {
      "epoch": 0.3798478649224918,
      "grad_norm": 0.8640220165252686,
      "learning_rate": 7.119620449639682e-06,
      "loss": 0.7647,
      "step": 13183
    },
    {
      "epoch": 0.3798766783841411,
      "grad_norm": 0.7953008413314819,
      "learning_rate": 7.119197832681992e-06,
      "loss": 0.7437,
      "step": 13184
    },
    {
      "epoch": 0.37990549184579037,
      "grad_norm": 0.7500513195991516,
      "learning_rate": 7.118775197268209e-06,
      "loss": 0.71,
      "step": 13185
    },
    {
      "epoch": 0.37993430530743966,
      "grad_norm": 0.7972039580345154,
      "learning_rate": 7.118352543402011e-06,
      "loss": 0.7443,
      "step": 13186
    },
    {
      "epoch": 0.3799631187690889,
      "grad_norm": 0.8701677322387695,
      "learning_rate": 7.117929871087079e-06,
      "loss": 0.7423,
      "step": 13187
    },
    {
      "epoch": 0.3799919322307382,
      "grad_norm": 0.8027120232582092,
      "learning_rate": 7.1175071803270964e-06,
      "loss": 0.7382,
      "step": 13188
    },
    {
      "epoch": 0.3800207456923875,
      "grad_norm": 0.7874975204467773,
      "learning_rate": 7.117084471125744e-06,
      "loss": 0.7517,
      "step": 13189
    },
    {
      "epoch": 0.38004955915403676,
      "grad_norm": 0.7801333069801331,
      "learning_rate": 7.1166617434867015e-06,
      "loss": 0.7468,
      "step": 13190
    },
    {
      "epoch": 0.38007837261568606,
      "grad_norm": 0.7434768676757812,
      "learning_rate": 7.116238997413649e-06,
      "loss": 0.7389,
      "step": 13191
    },
    {
      "epoch": 0.38010718607733535,
      "grad_norm": 0.8037508130073547,
      "learning_rate": 7.115816232910272e-06,
      "loss": 0.7482,
      "step": 13192
    },
    {
      "epoch": 0.38013599953898464,
      "grad_norm": 0.7859622240066528,
      "learning_rate": 7.115393449980252e-06,
      "loss": 0.7387,
      "step": 13193
    },
    {
      "epoch": 0.38016481300063387,
      "grad_norm": 0.7765452265739441,
      "learning_rate": 7.114970648627268e-06,
      "loss": 0.7327,
      "step": 13194
    },
    {
      "epoch": 0.38019362646228316,
      "grad_norm": 0.8934324979782104,
      "learning_rate": 7.114547828855003e-06,
      "loss": 0.738,
      "step": 13195
    },
    {
      "epoch": 0.38022243992393245,
      "grad_norm": 0.8035918474197388,
      "learning_rate": 7.114124990667141e-06,
      "loss": 0.7232,
      "step": 13196
    },
    {
      "epoch": 0.38025125338558174,
      "grad_norm": 1.8725234270095825,
      "learning_rate": 7.113702134067364e-06,
      "loss": 0.7795,
      "step": 13197
    },
    {
      "epoch": 0.38028006684723104,
      "grad_norm": 0.850597083568573,
      "learning_rate": 7.1132792590593545e-06,
      "loss": 0.7421,
      "step": 13198
    },
    {
      "epoch": 0.3803088803088803,
      "grad_norm": 0.8224457502365112,
      "learning_rate": 7.112856365646795e-06,
      "loss": 0.7381,
      "step": 13199
    },
    {
      "epoch": 0.3803376937705296,
      "grad_norm": 0.8155181407928467,
      "learning_rate": 7.112433453833369e-06,
      "loss": 0.736,
      "step": 13200
    },
    {
      "epoch": 0.38036650723217885,
      "grad_norm": 0.789824366569519,
      "learning_rate": 7.112010523622759e-06,
      "loss": 0.7265,
      "step": 13201
    },
    {
      "epoch": 0.38039532069382814,
      "grad_norm": 0.8005701899528503,
      "learning_rate": 7.1115875750186484e-06,
      "loss": 0.7262,
      "step": 13202
    },
    {
      "epoch": 0.38042413415547743,
      "grad_norm": 0.7570990324020386,
      "learning_rate": 7.11116460802472e-06,
      "loss": 0.7257,
      "step": 13203
    },
    {
      "epoch": 0.3804529476171267,
      "grad_norm": 0.8066978454589844,
      "learning_rate": 7.110741622644659e-06,
      "loss": 0.7428,
      "step": 13204
    },
    {
      "epoch": 0.380481761078776,
      "grad_norm": 0.7969775199890137,
      "learning_rate": 7.110318618882148e-06,
      "loss": 0.7248,
      "step": 13205
    },
    {
      "epoch": 0.3805105745404253,
      "grad_norm": 0.7392417788505554,
      "learning_rate": 7.109895596740873e-06,
      "loss": 0.7149,
      "step": 13206
    },
    {
      "epoch": 0.3805393880020746,
      "grad_norm": 0.7530917525291443,
      "learning_rate": 7.109472556224515e-06,
      "loss": 0.7282,
      "step": 13207
    },
    {
      "epoch": 0.38056820146372383,
      "grad_norm": 0.8923720717430115,
      "learning_rate": 7.1090494973367595e-06,
      "loss": 0.7373,
      "step": 13208
    },
    {
      "epoch": 0.3805970149253731,
      "grad_norm": 0.7676225900650024,
      "learning_rate": 7.108626420081291e-06,
      "loss": 0.7508,
      "step": 13209
    },
    {
      "epoch": 0.3806258283870224,
      "grad_norm": 0.8067589402198792,
      "learning_rate": 7.108203324461795e-06,
      "loss": 0.7407,
      "step": 13210
    },
    {
      "epoch": 0.3806546418486717,
      "grad_norm": 0.7667214870452881,
      "learning_rate": 7.107780210481956e-06,
      "loss": 0.7601,
      "step": 13211
    },
    {
      "epoch": 0.380683455310321,
      "grad_norm": 0.7930948734283447,
      "learning_rate": 7.107357078145458e-06,
      "loss": 0.7413,
      "step": 13212
    },
    {
      "epoch": 0.3807122687719703,
      "grad_norm": 0.8183894753456116,
      "learning_rate": 7.106933927455987e-06,
      "loss": 0.731,
      "step": 13213
    },
    {
      "epoch": 0.3807410822336195,
      "grad_norm": 0.7460421919822693,
      "learning_rate": 7.106510758417226e-06,
      "loss": 0.7355,
      "step": 13214
    },
    {
      "epoch": 0.3807698956952688,
      "grad_norm": 0.7872521877288818,
      "learning_rate": 7.106087571032863e-06,
      "loss": 0.7466,
      "step": 13215
    },
    {
      "epoch": 0.3807987091569181,
      "grad_norm": 0.7960558533668518,
      "learning_rate": 7.1056643653065825e-06,
      "loss": 0.7339,
      "step": 13216
    },
    {
      "epoch": 0.3808275226185674,
      "grad_norm": 0.8217551708221436,
      "learning_rate": 7.10524114124207e-06,
      "loss": 0.7084,
      "step": 13217
    },
    {
      "epoch": 0.3808563360802167,
      "grad_norm": 0.8096862435340881,
      "learning_rate": 7.1048178988430126e-06,
      "loss": 0.7274,
      "step": 13218
    },
    {
      "epoch": 0.380885149541866,
      "grad_norm": 0.8039559721946716,
      "learning_rate": 7.1043946381130945e-06,
      "loss": 0.7455,
      "step": 13219
    },
    {
      "epoch": 0.38091396300351527,
      "grad_norm": 0.7683417201042175,
      "learning_rate": 7.103971359056003e-06,
      "loss": 0.7142,
      "step": 13220
    },
    {
      "epoch": 0.3809427764651645,
      "grad_norm": 0.812969446182251,
      "learning_rate": 7.103548061675424e-06,
      "loss": 0.7522,
      "step": 13221
    },
    {
      "epoch": 0.3809715899268138,
      "grad_norm": 0.8123406767845154,
      "learning_rate": 7.103124745975045e-06,
      "loss": 0.7471,
      "step": 13222
    },
    {
      "epoch": 0.3810004033884631,
      "grad_norm": 0.7669793963432312,
      "learning_rate": 7.10270141195855e-06,
      "loss": 0.7156,
      "step": 13223
    },
    {
      "epoch": 0.3810292168501124,
      "grad_norm": 0.7412782907485962,
      "learning_rate": 7.102278059629628e-06,
      "loss": 0.7473,
      "step": 13224
    },
    {
      "epoch": 0.38105803031176166,
      "grad_norm": 0.7994743585586548,
      "learning_rate": 7.101854688991967e-06,
      "loss": 0.7755,
      "step": 13225
    },
    {
      "epoch": 0.38108684377341095,
      "grad_norm": 0.7787079215049744,
      "learning_rate": 7.101431300049253e-06,
      "loss": 0.7246,
      "step": 13226
    },
    {
      "epoch": 0.38111565723506025,
      "grad_norm": 0.7449604868888855,
      "learning_rate": 7.1010078928051706e-06,
      "loss": 0.7164,
      "step": 13227
    },
    {
      "epoch": 0.3811444706967095,
      "grad_norm": 0.7510868906974792,
      "learning_rate": 7.100584467263412e-06,
      "loss": 0.726,
      "step": 13228
    },
    {
      "epoch": 0.38117328415835877,
      "grad_norm": 0.7638269662857056,
      "learning_rate": 7.100161023427661e-06,
      "loss": 0.736,
      "step": 13229
    },
    {
      "epoch": 0.38120209762000806,
      "grad_norm": 0.7786377668380737,
      "learning_rate": 7.099737561301608e-06,
      "loss": 0.7453,
      "step": 13230
    },
    {
      "epoch": 0.38123091108165735,
      "grad_norm": 0.8149323463439941,
      "learning_rate": 7.099314080888939e-06,
      "loss": 0.7532,
      "step": 13231
    },
    {
      "epoch": 0.38125972454330664,
      "grad_norm": 1.6771000623703003,
      "learning_rate": 7.0988905821933444e-06,
      "loss": 0.753,
      "step": 13232
    },
    {
      "epoch": 0.38128853800495593,
      "grad_norm": 0.7505082488059998,
      "learning_rate": 7.098467065218508e-06,
      "loss": 0.7428,
      "step": 13233
    },
    {
      "epoch": 0.3813173514666052,
      "grad_norm": 0.7797925472259521,
      "learning_rate": 7.098043529968124e-06,
      "loss": 0.7511,
      "step": 13234
    },
    {
      "epoch": 0.38134616492825446,
      "grad_norm": 0.7698721885681152,
      "learning_rate": 7.097619976445877e-06,
      "loss": 0.71,
      "step": 13235
    },
    {
      "epoch": 0.38137497838990375,
      "grad_norm": 0.7555069327354431,
      "learning_rate": 7.0971964046554575e-06,
      "loss": 0.7071,
      "step": 13236
    },
    {
      "epoch": 0.38140379185155304,
      "grad_norm": 0.7631537318229675,
      "learning_rate": 7.096772814600553e-06,
      "loss": 0.7336,
      "step": 13237
    },
    {
      "epoch": 0.38143260531320233,
      "grad_norm": 0.7525351047515869,
      "learning_rate": 7.096349206284854e-06,
      "loss": 0.7532,
      "step": 13238
    },
    {
      "epoch": 0.3814614187748516,
      "grad_norm": 0.7459418773651123,
      "learning_rate": 7.095925579712049e-06,
      "loss": 0.7448,
      "step": 13239
    },
    {
      "epoch": 0.3814902322365009,
      "grad_norm": 0.7699084877967834,
      "learning_rate": 7.095501934885826e-06,
      "loss": 0.7235,
      "step": 13240
    },
    {
      "epoch": 0.38151904569815015,
      "grad_norm": 0.7695544958114624,
      "learning_rate": 7.095078271809877e-06,
      "loss": 0.7372,
      "step": 13241
    },
    {
      "epoch": 0.38154785915979944,
      "grad_norm": 0.7247528433799744,
      "learning_rate": 7.094654590487891e-06,
      "loss": 0.6976,
      "step": 13242
    },
    {
      "epoch": 0.38157667262144873,
      "grad_norm": 0.7423374056816101,
      "learning_rate": 7.094230890923555e-06,
      "loss": 0.7391,
      "step": 13243
    },
    {
      "epoch": 0.381605486083098,
      "grad_norm": 0.7756534218788147,
      "learning_rate": 7.0938071731205635e-06,
      "loss": 0.7577,
      "step": 13244
    },
    {
      "epoch": 0.3816342995447473,
      "grad_norm": 0.7970913052558899,
      "learning_rate": 7.093383437082605e-06,
      "loss": 0.7523,
      "step": 13245
    },
    {
      "epoch": 0.3816631130063966,
      "grad_norm": 0.7880094051361084,
      "learning_rate": 7.092959682813367e-06,
      "loss": 0.7149,
      "step": 13246
    },
    {
      "epoch": 0.3816919264680459,
      "grad_norm": 0.7705859541893005,
      "learning_rate": 7.092535910316545e-06,
      "loss": 0.7336,
      "step": 13247
    },
    {
      "epoch": 0.38172073992969513,
      "grad_norm": 0.7569336891174316,
      "learning_rate": 7.092112119595826e-06,
      "loss": 0.748,
      "step": 13248
    },
    {
      "epoch": 0.3817495533913444,
      "grad_norm": 0.8138129115104675,
      "learning_rate": 7.091688310654902e-06,
      "loss": 0.743,
      "step": 13249
    },
    {
      "epoch": 0.3817783668529937,
      "grad_norm": 0.7641441822052002,
      "learning_rate": 7.0912644834974635e-06,
      "loss": 0.7211,
      "step": 13250
    },
    {
      "epoch": 0.381807180314643,
      "grad_norm": 0.7543177008628845,
      "learning_rate": 7.090840638127201e-06,
      "loss": 0.7304,
      "step": 13251
    },
    {
      "epoch": 0.3818359937762923,
      "grad_norm": 0.8107726573944092,
      "learning_rate": 7.090416774547807e-06,
      "loss": 0.7021,
      "step": 13252
    },
    {
      "epoch": 0.3818648072379416,
      "grad_norm": 0.7944405674934387,
      "learning_rate": 7.089992892762973e-06,
      "loss": 0.7306,
      "step": 13253
    },
    {
      "epoch": 0.3818936206995909,
      "grad_norm": 0.7912240028381348,
      "learning_rate": 7.08956899277639e-06,
      "loss": 0.7367,
      "step": 13254
    },
    {
      "epoch": 0.3819224341612401,
      "grad_norm": 0.7587143778800964,
      "learning_rate": 7.089145074591751e-06,
      "loss": 0.7197,
      "step": 13255
    },
    {
      "epoch": 0.3819512476228894,
      "grad_norm": 0.7740282416343689,
      "learning_rate": 7.0887211382127465e-06,
      "loss": 0.725,
      "step": 13256
    },
    {
      "epoch": 0.3819800610845387,
      "grad_norm": 0.7529832124710083,
      "learning_rate": 7.088297183643067e-06,
      "loss": 0.7493,
      "step": 13257
    },
    {
      "epoch": 0.382008874546188,
      "grad_norm": 0.7635983228683472,
      "learning_rate": 7.087873210886407e-06,
      "loss": 0.7564,
      "step": 13258
    },
    {
      "epoch": 0.3820376880078373,
      "grad_norm": 0.7679519057273865,
      "learning_rate": 7.087449219946459e-06,
      "loss": 0.7298,
      "step": 13259
    },
    {
      "epoch": 0.38206650146948656,
      "grad_norm": 0.7825263738632202,
      "learning_rate": 7.087025210826916e-06,
      "loss": 0.7602,
      "step": 13260
    },
    {
      "epoch": 0.38209531493113585,
      "grad_norm": 0.8492825627326965,
      "learning_rate": 7.086601183531469e-06,
      "loss": 0.7544,
      "step": 13261
    },
    {
      "epoch": 0.3821241283927851,
      "grad_norm": 0.7487807273864746,
      "learning_rate": 7.08617713806381e-06,
      "loss": 0.7291,
      "step": 13262
    },
    {
      "epoch": 0.3821529418544344,
      "grad_norm": 0.762776255607605,
      "learning_rate": 7.0857530744276346e-06,
      "loss": 0.7149,
      "step": 13263
    },
    {
      "epoch": 0.38218175531608367,
      "grad_norm": 0.7735134959220886,
      "learning_rate": 7.085328992626635e-06,
      "loss": 0.7243,
      "step": 13264
    },
    {
      "epoch": 0.38221056877773296,
      "grad_norm": 0.7422505021095276,
      "learning_rate": 7.084904892664505e-06,
      "loss": 0.7358,
      "step": 13265
    },
    {
      "epoch": 0.38223938223938225,
      "grad_norm": 0.7621713876724243,
      "learning_rate": 7.084480774544937e-06,
      "loss": 0.7416,
      "step": 13266
    },
    {
      "epoch": 0.38226819570103154,
      "grad_norm": 0.7414082884788513,
      "learning_rate": 7.084056638271626e-06,
      "loss": 0.7343,
      "step": 13267
    },
    {
      "epoch": 0.3822970091626808,
      "grad_norm": 0.7942991852760315,
      "learning_rate": 7.083632483848264e-06,
      "loss": 0.7686,
      "step": 13268
    },
    {
      "epoch": 0.38232582262433007,
      "grad_norm": 0.7688164710998535,
      "learning_rate": 7.0832083112785455e-06,
      "loss": 0.7173,
      "step": 13269
    },
    {
      "epoch": 0.38235463608597936,
      "grad_norm": 0.7852559685707092,
      "learning_rate": 7.082784120566166e-06,
      "loss": 0.727,
      "step": 13270
    },
    {
      "epoch": 0.38238344954762865,
      "grad_norm": 0.73174649477005,
      "learning_rate": 7.082359911714819e-06,
      "loss": 0.7507,
      "step": 13271
    },
    {
      "epoch": 0.38241226300927794,
      "grad_norm": 0.7492059469223022,
      "learning_rate": 7.081935684728199e-06,
      "loss": 0.7302,
      "step": 13272
    },
    {
      "epoch": 0.38244107647092723,
      "grad_norm": 0.767081081867218,
      "learning_rate": 7.08151143961e-06,
      "loss": 0.7287,
      "step": 13273
    },
    {
      "epoch": 0.3824698899325765,
      "grad_norm": 0.7709833383560181,
      "learning_rate": 7.081087176363917e-06,
      "loss": 0.7437,
      "step": 13274
    },
    {
      "epoch": 0.38249870339422576,
      "grad_norm": 0.7566392421722412,
      "learning_rate": 7.080662894993643e-06,
      "loss": 0.727,
      "step": 13275
    },
    {
      "epoch": 0.38252751685587505,
      "grad_norm": 0.7387537956237793,
      "learning_rate": 7.0802385955028774e-06,
      "loss": 0.7273,
      "step": 13276
    },
    {
      "epoch": 0.38255633031752434,
      "grad_norm": 0.7477244734764099,
      "learning_rate": 7.0798142778953125e-06,
      "loss": 0.7563,
      "step": 13277
    },
    {
      "epoch": 0.38258514377917363,
      "grad_norm": 0.7906595468521118,
      "learning_rate": 7.079389942174645e-06,
      "loss": 0.7309,
      "step": 13278
    },
    {
      "epoch": 0.3826139572408229,
      "grad_norm": 0.7481029629707336,
      "learning_rate": 7.078965588344568e-06,
      "loss": 0.7146,
      "step": 13279
    },
    {
      "epoch": 0.3826427707024722,
      "grad_norm": 0.7555207014083862,
      "learning_rate": 7.078541216408781e-06,
      "loss": 0.7455,
      "step": 13280
    },
    {
      "epoch": 0.3826715841641215,
      "grad_norm": 0.7881113886833191,
      "learning_rate": 7.078116826370977e-06,
      "loss": 0.7334,
      "step": 13281
    },
    {
      "epoch": 0.38270039762577074,
      "grad_norm": 0.7911213040351868,
      "learning_rate": 7.077692418234852e-06,
      "loss": 0.7437,
      "step": 13282
    },
    {
      "epoch": 0.38272921108742003,
      "grad_norm": 0.75885409116745,
      "learning_rate": 7.077267992004103e-06,
      "loss": 0.7528,
      "step": 13283
    },
    {
      "epoch": 0.3827580245490693,
      "grad_norm": 0.7848898768424988,
      "learning_rate": 7.076843547682427e-06,
      "loss": 0.7533,
      "step": 13284
    },
    {
      "epoch": 0.3827868380107186,
      "grad_norm": 0.7735159993171692,
      "learning_rate": 7.076419085273521e-06,
      "loss": 0.7436,
      "step": 13285
    },
    {
      "epoch": 0.3828156514723679,
      "grad_norm": 0.776881754398346,
      "learning_rate": 7.075994604781078e-06,
      "loss": 0.7486,
      "step": 13286
    },
    {
      "epoch": 0.3828444649340172,
      "grad_norm": 0.7713169455528259,
      "learning_rate": 7.075570106208798e-06,
      "loss": 0.7542,
      "step": 13287
    },
    {
      "epoch": 0.3828732783956664,
      "grad_norm": 0.7438216209411621,
      "learning_rate": 7.075145589560377e-06,
      "loss": 0.7327,
      "step": 13288
    },
    {
      "epoch": 0.3829020918573157,
      "grad_norm": 0.7632834911346436,
      "learning_rate": 7.074721054839514e-06,
      "loss": 0.7417,
      "step": 13289
    },
    {
      "epoch": 0.382930905318965,
      "grad_norm": 0.7669516801834106,
      "learning_rate": 7.074296502049903e-06,
      "loss": 0.7177,
      "step": 13290
    },
    {
      "epoch": 0.3829597187806143,
      "grad_norm": 0.7996196746826172,
      "learning_rate": 7.0738719311952446e-06,
      "loss": 0.7478,
      "step": 13291
    },
    {
      "epoch": 0.3829885322422636,
      "grad_norm": 0.7941195368766785,
      "learning_rate": 7.073447342279234e-06,
      "loss": 0.7393,
      "step": 13292
    },
    {
      "epoch": 0.3830173457039129,
      "grad_norm": 0.7421728372573853,
      "learning_rate": 7.073022735305568e-06,
      "loss": 0.7449,
      "step": 13293
    },
    {
      "epoch": 0.3830461591655622,
      "grad_norm": 0.7579863667488098,
      "learning_rate": 7.072598110277947e-06,
      "loss": 0.7196,
      "step": 13294
    },
    {
      "epoch": 0.3830749726272114,
      "grad_norm": 0.7174354791641235,
      "learning_rate": 7.072173467200069e-06,
      "loss": 0.7114,
      "step": 13295
    },
    {
      "epoch": 0.3831037860888607,
      "grad_norm": 0.7555776238441467,
      "learning_rate": 7.071748806075632e-06,
      "loss": 0.729,
      "step": 13296
    },
    {
      "epoch": 0.38313259955051,
      "grad_norm": 0.7664331197738647,
      "learning_rate": 7.0713241269083344e-06,
      "loss": 0.7471,
      "step": 13297
    },
    {
      "epoch": 0.3831614130121593,
      "grad_norm": 0.7479694485664368,
      "learning_rate": 7.0708994297018725e-06,
      "loss": 0.7276,
      "step": 13298
    },
    {
      "epoch": 0.38319022647380857,
      "grad_norm": 0.7711441516876221,
      "learning_rate": 7.070474714459947e-06,
      "loss": 0.7253,
      "step": 13299
    },
    {
      "epoch": 0.38321903993545786,
      "grad_norm": 1.6596155166625977,
      "learning_rate": 7.070049981186257e-06,
      "loss": 0.7346,
      "step": 13300
    },
    {
      "epoch": 0.38324785339710715,
      "grad_norm": 0.7696949243545532,
      "learning_rate": 7.069625229884501e-06,
      "loss": 0.7085,
      "step": 13301
    },
    {
      "epoch": 0.3832766668587564,
      "grad_norm": 0.8053165078163147,
      "learning_rate": 7.069200460558376e-06,
      "loss": 0.7679,
      "step": 13302
    },
    {
      "epoch": 0.3833054803204057,
      "grad_norm": 2.0677244663238525,
      "learning_rate": 7.0687756732115864e-06,
      "loss": 0.7312,
      "step": 13303
    },
    {
      "epoch": 0.38333429378205497,
      "grad_norm": 0.7883168458938599,
      "learning_rate": 7.068350867847827e-06,
      "loss": 0.7109,
      "step": 13304
    },
    {
      "epoch": 0.38336310724370426,
      "grad_norm": 0.7855682969093323,
      "learning_rate": 7.0679260444708e-06,
      "loss": 0.7333,
      "step": 13305
    },
    {
      "epoch": 0.38339192070535355,
      "grad_norm": 0.7687048316001892,
      "learning_rate": 7.0675012030842035e-06,
      "loss": 0.755,
      "step": 13306
    },
    {
      "epoch": 0.38342073416700284,
      "grad_norm": 0.7693661451339722,
      "learning_rate": 7.067076343691739e-06,
      "loss": 0.7246,
      "step": 13307
    },
    {
      "epoch": 0.38344954762865213,
      "grad_norm": 0.7447250485420227,
      "learning_rate": 7.0666514662971065e-06,
      "loss": 0.7306,
      "step": 13308
    },
    {
      "epoch": 0.38347836109030137,
      "grad_norm": 0.765826940536499,
      "learning_rate": 7.066226570904004e-06,
      "loss": 0.7215,
      "step": 13309
    },
    {
      "epoch": 0.38350717455195066,
      "grad_norm": 0.7484069466590881,
      "learning_rate": 7.065801657516136e-06,
      "loss": 0.7231,
      "step": 13310
    },
    {
      "epoch": 0.38353598801359995,
      "grad_norm": 0.7768756747245789,
      "learning_rate": 7.065376726137197e-06,
      "loss": 0.74,
      "step": 13311
    },
    {
      "epoch": 0.38356480147524924,
      "grad_norm": 0.8073358535766602,
      "learning_rate": 7.0649517767708945e-06,
      "loss": 0.7463,
      "step": 13312
    },
    {
      "epoch": 0.38359361493689853,
      "grad_norm": 0.7633518576622009,
      "learning_rate": 7.064526809420924e-06,
      "loss": 0.7093,
      "step": 13313
    },
    {
      "epoch": 0.3836224283985478,
      "grad_norm": 0.7491855025291443,
      "learning_rate": 7.0641018240909905e-06,
      "loss": 0.7183,
      "step": 13314
    },
    {
      "epoch": 0.38365124186019706,
      "grad_norm": 0.7962595820426941,
      "learning_rate": 7.063676820784793e-06,
      "loss": 0.7224,
      "step": 13315
    },
    {
      "epoch": 0.38368005532184635,
      "grad_norm": 0.7534985542297363,
      "learning_rate": 7.0632517995060335e-06,
      "loss": 0.7114,
      "step": 13316
    },
    {
      "epoch": 0.38370886878349564,
      "grad_norm": 0.7611765265464783,
      "learning_rate": 7.062826760258412e-06,
      "loss": 0.7356,
      "step": 13317
    },
    {
      "epoch": 0.38373768224514493,
      "grad_norm": 0.7701900601387024,
      "learning_rate": 7.0624017030456325e-06,
      "loss": 0.7571,
      "step": 13318
    },
    {
      "epoch": 0.3837664957067942,
      "grad_norm": 0.7979373335838318,
      "learning_rate": 7.0619766278713965e-06,
      "loss": 0.7413,
      "step": 13319
    },
    {
      "epoch": 0.3837953091684435,
      "grad_norm": 0.7955649495124817,
      "learning_rate": 7.061551534739404e-06,
      "loss": 0.7232,
      "step": 13320
    },
    {
      "epoch": 0.3838241226300928,
      "grad_norm": 0.775236189365387,
      "learning_rate": 7.061126423653359e-06,
      "loss": 0.7219,
      "step": 13321
    },
    {
      "epoch": 0.38385293609174204,
      "grad_norm": 0.7608160376548767,
      "learning_rate": 7.0607012946169625e-06,
      "loss": 0.7309,
      "step": 13322
    },
    {
      "epoch": 0.3838817495533913,
      "grad_norm": 0.7604827284812927,
      "learning_rate": 7.060276147633918e-06,
      "loss": 0.7345,
      "step": 13323
    },
    {
      "epoch": 0.3839105630150406,
      "grad_norm": 0.7343815565109253,
      "learning_rate": 7.0598509827079275e-06,
      "loss": 0.7336,
      "step": 13324
    },
    {
      "epoch": 0.3839393764766899,
      "grad_norm": 0.782738447189331,
      "learning_rate": 7.059425799842695e-06,
      "loss": 0.7249,
      "step": 13325
    },
    {
      "epoch": 0.3839681899383392,
      "grad_norm": 0.7748640179634094,
      "learning_rate": 7.059000599041922e-06,
      "loss": 0.7227,
      "step": 13326
    },
    {
      "epoch": 0.3839970033999885,
      "grad_norm": 0.7902774214744568,
      "learning_rate": 7.058575380309312e-06,
      "loss": 0.7184,
      "step": 13327
    },
    {
      "epoch": 0.3840258168616378,
      "grad_norm": 0.7809252142906189,
      "learning_rate": 7.058150143648569e-06,
      "loss": 0.742,
      "step": 13328
    },
    {
      "epoch": 0.384054630323287,
      "grad_norm": 0.761084258556366,
      "learning_rate": 7.057724889063395e-06,
      "loss": 0.7437,
      "step": 13329
    },
    {
      "epoch": 0.3840834437849363,
      "grad_norm": 0.7817826271057129,
      "learning_rate": 7.057299616557494e-06,
      "loss": 0.752,
      "step": 13330
    },
    {
      "epoch": 0.3841122572465856,
      "grad_norm": 0.805786669254303,
      "learning_rate": 7.056874326134569e-06,
      "loss": 0.7565,
      "step": 13331
    },
    {
      "epoch": 0.3841410707082349,
      "grad_norm": 0.797986626625061,
      "learning_rate": 7.056449017798327e-06,
      "loss": 0.7354,
      "step": 13332
    },
    {
      "epoch": 0.3841698841698842,
      "grad_norm": 0.7589388489723206,
      "learning_rate": 7.056023691552468e-06,
      "loss": 0.736,
      "step": 13333
    },
    {
      "epoch": 0.38419869763153347,
      "grad_norm": 0.8068683743476868,
      "learning_rate": 7.0555983474007e-06,
      "loss": 0.7296,
      "step": 13334
    },
    {
      "epoch": 0.38422751109318276,
      "grad_norm": 0.8003888726234436,
      "learning_rate": 7.055172985346722e-06,
      "loss": 0.7324,
      "step": 13335
    },
    {
      "epoch": 0.384256324554832,
      "grad_norm": 0.7934733033180237,
      "learning_rate": 7.054747605394243e-06,
      "loss": 0.7247,
      "step": 13336
    },
    {
      "epoch": 0.3842851380164813,
      "grad_norm": 0.7731021046638489,
      "learning_rate": 7.0543222075469674e-06,
      "loss": 0.749,
      "step": 13337
    },
    {
      "epoch": 0.3843139514781306,
      "grad_norm": 0.7763952612876892,
      "learning_rate": 7.053896791808599e-06,
      "loss": 0.7353,
      "step": 13338
    },
    {
      "epoch": 0.38434276493977987,
      "grad_norm": 0.7283328175544739,
      "learning_rate": 7.053471358182842e-06,
      "loss": 0.74,
      "step": 13339
    },
    {
      "epoch": 0.38437157840142916,
      "grad_norm": 0.7329838871955872,
      "learning_rate": 7.053045906673402e-06,
      "loss": 0.7557,
      "step": 13340
    },
    {
      "epoch": 0.38440039186307845,
      "grad_norm": 0.8807564973831177,
      "learning_rate": 7.052620437283984e-06,
      "loss": 0.7584,
      "step": 13341
    },
    {
      "epoch": 0.3844292053247277,
      "grad_norm": 0.7948534488677979,
      "learning_rate": 7.052194950018295e-06,
      "loss": 0.7253,
      "step": 13342
    },
    {
      "epoch": 0.384458018786377,
      "grad_norm": 0.7806177735328674,
      "learning_rate": 7.051769444880039e-06,
      "loss": 0.7457,
      "step": 13343
    },
    {
      "epoch": 0.38448683224802627,
      "grad_norm": 0.7624856233596802,
      "learning_rate": 7.051343921872922e-06,
      "loss": 0.7417,
      "step": 13344
    },
    {
      "epoch": 0.38451564570967556,
      "grad_norm": 0.8057020306587219,
      "learning_rate": 7.05091838100065e-06,
      "loss": 0.7647,
      "step": 13345
    },
    {
      "epoch": 0.38454445917132485,
      "grad_norm": 0.7595983147621155,
      "learning_rate": 7.050492822266929e-06,
      "loss": 0.7396,
      "step": 13346
    },
    {
      "epoch": 0.38457327263297414,
      "grad_norm": 0.7703843116760254,
      "learning_rate": 7.050067245675465e-06,
      "loss": 0.7495,
      "step": 13347
    },
    {
      "epoch": 0.38460208609462343,
      "grad_norm": 0.8018490076065063,
      "learning_rate": 7.049641651229966e-06,
      "loss": 0.7352,
      "step": 13348
    },
    {
      "epoch": 0.38463089955627267,
      "grad_norm": 0.7835618257522583,
      "learning_rate": 7.049216038934136e-06,
      "loss": 0.7333,
      "step": 13349
    },
    {
      "epoch": 0.38465971301792196,
      "grad_norm": 0.7647236585617065,
      "learning_rate": 7.048790408791683e-06,
      "loss": 0.7223,
      "step": 13350
    },
    {
      "epoch": 0.38468852647957125,
      "grad_norm": 0.7665435671806335,
      "learning_rate": 7.048364760806313e-06,
      "loss": 0.7292,
      "step": 13351
    },
    {
      "epoch": 0.38471733994122054,
      "grad_norm": 0.7851794958114624,
      "learning_rate": 7.047939094981735e-06,
      "loss": 0.7409,
      "step": 13352
    },
    {
      "epoch": 0.38474615340286983,
      "grad_norm": 0.7857627868652344,
      "learning_rate": 7.047513411321653e-06,
      "loss": 0.7466,
      "step": 13353
    },
    {
      "epoch": 0.3847749668645191,
      "grad_norm": 0.7575058937072754,
      "learning_rate": 7.047087709829776e-06,
      "loss": 0.7449,
      "step": 13354
    },
    {
      "epoch": 0.3848037803261684,
      "grad_norm": 0.7643128037452698,
      "learning_rate": 7.046661990509813e-06,
      "loss": 0.7336,
      "step": 13355
    },
    {
      "epoch": 0.38483259378781765,
      "grad_norm": 0.7845592498779297,
      "learning_rate": 7.046236253365469e-06,
      "loss": 0.74,
      "step": 13356
    },
    {
      "epoch": 0.38486140724946694,
      "grad_norm": 0.7556120157241821,
      "learning_rate": 7.045810498400454e-06,
      "loss": 0.7217,
      "step": 13357
    },
    {
      "epoch": 0.3848902207111162,
      "grad_norm": 0.776157796382904,
      "learning_rate": 7.045384725618473e-06,
      "loss": 0.7306,
      "step": 13358
    },
    {
      "epoch": 0.3849190341727655,
      "grad_norm": 0.7725396156311035,
      "learning_rate": 7.044958935023235e-06,
      "loss": 0.7585,
      "step": 13359
    },
    {
      "epoch": 0.3849478476344148,
      "grad_norm": 0.7695205211639404,
      "learning_rate": 7.04453312661845e-06,
      "loss": 0.7428,
      "step": 13360
    },
    {
      "epoch": 0.3849766610960641,
      "grad_norm": 0.743716835975647,
      "learning_rate": 7.044107300407825e-06,
      "loss": 0.7188,
      "step": 13361
    },
    {
      "epoch": 0.3850054745577134,
      "grad_norm": 0.7496239542961121,
      "learning_rate": 7.0436814563950685e-06,
      "loss": 0.7295,
      "step": 13362
    },
    {
      "epoch": 0.3850342880193626,
      "grad_norm": 0.7539101839065552,
      "learning_rate": 7.043255594583891e-06,
      "loss": 0.7399,
      "step": 13363
    },
    {
      "epoch": 0.3850631014810119,
      "grad_norm": 0.7379118204116821,
      "learning_rate": 7.042829714977999e-06,
      "loss": 0.7096,
      "step": 13364
    },
    {
      "epoch": 0.3850919149426612,
      "grad_norm": 0.7420912981033325,
      "learning_rate": 7.0424038175811005e-06,
      "loss": 0.7718,
      "step": 13365
    },
    {
      "epoch": 0.3851207284043105,
      "grad_norm": 0.7617236375808716,
      "learning_rate": 7.041977902396908e-06,
      "loss": 0.7264,
      "step": 13366
    },
    {
      "epoch": 0.3851495418659598,
      "grad_norm": 0.7266903519630432,
      "learning_rate": 7.04155196942913e-06,
      "loss": 0.7162,
      "step": 13367
    },
    {
      "epoch": 0.3851783553276091,
      "grad_norm": 0.762427031993866,
      "learning_rate": 7.041126018681475e-06,
      "loss": 0.7424,
      "step": 13368
    },
    {
      "epoch": 0.3852071687892583,
      "grad_norm": 0.7604819536209106,
      "learning_rate": 7.040700050157652e-06,
      "loss": 0.7386,
      "step": 13369
    },
    {
      "epoch": 0.3852359822509076,
      "grad_norm": 0.7749103307723999,
      "learning_rate": 7.040274063861371e-06,
      "loss": 0.7301,
      "step": 13370
    },
    {
      "epoch": 0.3852647957125569,
      "grad_norm": 0.7570476531982422,
      "learning_rate": 7.039848059796341e-06,
      "loss": 0.7383,
      "step": 13371
    },
    {
      "epoch": 0.3852936091742062,
      "grad_norm": 0.7662016749382019,
      "learning_rate": 7.039422037966277e-06,
      "loss": 0.7351,
      "step": 13372
    },
    {
      "epoch": 0.3853224226358555,
      "grad_norm": 0.7450062036514282,
      "learning_rate": 7.038995998374885e-06,
      "loss": 0.7132,
      "step": 13373
    },
    {
      "epoch": 0.38535123609750477,
      "grad_norm": 0.8087038993835449,
      "learning_rate": 7.038569941025874e-06,
      "loss": 0.7281,
      "step": 13374
    },
    {
      "epoch": 0.38538004955915406,
      "grad_norm": 0.744335949420929,
      "learning_rate": 7.038143865922958e-06,
      "loss": 0.7259,
      "step": 13375
    },
    {
      "epoch": 0.3854088630208033,
      "grad_norm": 0.7423966526985168,
      "learning_rate": 7.0377177730698476e-06,
      "loss": 0.7451,
      "step": 13376
    },
    {
      "epoch": 0.3854376764824526,
      "grad_norm": 0.8167110681533813,
      "learning_rate": 7.037291662470249e-06,
      "loss": 0.7411,
      "step": 13377
    },
    {
      "epoch": 0.3854664899441019,
      "grad_norm": 0.7643268704414368,
      "learning_rate": 7.036865534127879e-06,
      "loss": 0.736,
      "step": 13378
    },
    {
      "epoch": 0.38549530340575117,
      "grad_norm": 0.7488981485366821,
      "learning_rate": 7.0364393880464464e-06,
      "loss": 0.762,
      "step": 13379
    },
    {
      "epoch": 0.38552411686740046,
      "grad_norm": 0.7488610148429871,
      "learning_rate": 7.036013224229662e-06,
      "loss": 0.7487,
      "step": 13380
    },
    {
      "epoch": 0.38555293032904975,
      "grad_norm": 0.7930134534835815,
      "learning_rate": 7.035587042681237e-06,
      "loss": 0.7471,
      "step": 13381
    },
    {
      "epoch": 0.38558174379069904,
      "grad_norm": 0.763251781463623,
      "learning_rate": 7.035160843404885e-06,
      "loss": 0.7335,
      "step": 13382
    },
    {
      "epoch": 0.3856105572523483,
      "grad_norm": 0.7631905674934387,
      "learning_rate": 7.034734626404315e-06,
      "loss": 0.7408,
      "step": 13383
    },
    {
      "epoch": 0.38563937071399756,
      "grad_norm": 0.7632477283477783,
      "learning_rate": 7.034308391683241e-06,
      "loss": 0.7261,
      "step": 13384
    },
    {
      "epoch": 0.38566818417564686,
      "grad_norm": 0.7243921160697937,
      "learning_rate": 7.033882139245375e-06,
      "loss": 0.7082,
      "step": 13385
    },
    {
      "epoch": 0.38569699763729615,
      "grad_norm": 0.7478147149085999,
      "learning_rate": 7.033455869094428e-06,
      "loss": 0.7503,
      "step": 13386
    },
    {
      "epoch": 0.38572581109894544,
      "grad_norm": 0.7526957988739014,
      "learning_rate": 7.033029581234114e-06,
      "loss": 0.7766,
      "step": 13387
    },
    {
      "epoch": 0.3857546245605947,
      "grad_norm": 0.7685060501098633,
      "learning_rate": 7.032603275668144e-06,
      "loss": 0.7263,
      "step": 13388
    },
    {
      "epoch": 0.385783438022244,
      "grad_norm": 0.7394968867301941,
      "learning_rate": 7.032176952400231e-06,
      "loss": 0.7328,
      "step": 13389
    },
    {
      "epoch": 0.38581225148389325,
      "grad_norm": 0.7703744769096375,
      "learning_rate": 7.031750611434088e-06,
      "loss": 0.752,
      "step": 13390
    },
    {
      "epoch": 0.38584106494554254,
      "grad_norm": 0.7549912333488464,
      "learning_rate": 7.031324252773429e-06,
      "loss": 0.7071,
      "step": 13391
    },
    {
      "epoch": 0.38586987840719184,
      "grad_norm": 0.7843210101127625,
      "learning_rate": 7.0308978764219675e-06,
      "loss": 0.7148,
      "step": 13392
    },
    {
      "epoch": 0.3858986918688411,
      "grad_norm": 0.7637206315994263,
      "learning_rate": 7.030471482383414e-06,
      "loss": 0.758,
      "step": 13393
    },
    {
      "epoch": 0.3859275053304904,
      "grad_norm": 0.7403991222381592,
      "learning_rate": 7.030045070661484e-06,
      "loss": 0.7506,
      "step": 13394
    },
    {
      "epoch": 0.3859563187921397,
      "grad_norm": 0.803901731967926,
      "learning_rate": 7.02961864125989e-06,
      "loss": 0.7349,
      "step": 13395
    },
    {
      "epoch": 0.38598513225378894,
      "grad_norm": 0.7465941905975342,
      "learning_rate": 7.029192194182347e-06,
      "loss": 0.7389,
      "step": 13396
    },
    {
      "epoch": 0.38601394571543823,
      "grad_norm": 0.7553490400314331,
      "learning_rate": 7.028765729432569e-06,
      "loss": 0.7345,
      "step": 13397
    },
    {
      "epoch": 0.3860427591770875,
      "grad_norm": 0.7769975662231445,
      "learning_rate": 7.028339247014269e-06,
      "loss": 0.7403,
      "step": 13398
    },
    {
      "epoch": 0.3860715726387368,
      "grad_norm": 0.757041871547699,
      "learning_rate": 7.027912746931162e-06,
      "loss": 0.7592,
      "step": 13399
    },
    {
      "epoch": 0.3861003861003861,
      "grad_norm": 0.8102391958236694,
      "learning_rate": 7.027486229186961e-06,
      "loss": 0.7441,
      "step": 13400
    },
    {
      "epoch": 0.3861291995620354,
      "grad_norm": 0.7820296287536621,
      "learning_rate": 7.027059693785383e-06,
      "loss": 0.7224,
      "step": 13401
    },
    {
      "epoch": 0.3861580130236847,
      "grad_norm": 0.774726927280426,
      "learning_rate": 7.02663314073014e-06,
      "loss": 0.7441,
      "step": 13402
    },
    {
      "epoch": 0.3861868264853339,
      "grad_norm": 0.7753574252128601,
      "learning_rate": 7.026206570024949e-06,
      "loss": 0.7715,
      "step": 13403
    },
    {
      "epoch": 0.3862156399469832,
      "grad_norm": 0.7414454221725464,
      "learning_rate": 7.0257799816735266e-06,
      "loss": 0.7091,
      "step": 13404
    },
    {
      "epoch": 0.3862444534086325,
      "grad_norm": 0.7130387425422668,
      "learning_rate": 7.025353375679582e-06,
      "loss": 0.7333,
      "step": 13405
    },
    {
      "epoch": 0.3862732668702818,
      "grad_norm": 0.7822401523590088,
      "learning_rate": 7.024926752046836e-06,
      "loss": 0.7368,
      "step": 13406
    },
    {
      "epoch": 0.3863020803319311,
      "grad_norm": 0.7749193906784058,
      "learning_rate": 7.024500110779001e-06,
      "loss": 0.74,
      "step": 13407
    },
    {
      "epoch": 0.3863308937935804,
      "grad_norm": 0.757901668548584,
      "learning_rate": 7.024073451879795e-06,
      "loss": 0.7151,
      "step": 13408
    },
    {
      "epoch": 0.38635970725522967,
      "grad_norm": 0.7601590156555176,
      "learning_rate": 7.023646775352932e-06,
      "loss": 0.7265,
      "step": 13409
    },
    {
      "epoch": 0.3863885207168789,
      "grad_norm": 0.7792931199073792,
      "learning_rate": 7.023220081202129e-06,
      "loss": 0.7354,
      "step": 13410
    },
    {
      "epoch": 0.3864173341785282,
      "grad_norm": 0.742027997970581,
      "learning_rate": 7.022793369431102e-06,
      "loss": 0.7284,
      "step": 13411
    },
    {
      "epoch": 0.3864461476401775,
      "grad_norm": 0.7830774784088135,
      "learning_rate": 7.0223666400435654e-06,
      "loss": 0.7287,
      "step": 13412
    },
    {
      "epoch": 0.3864749611018268,
      "grad_norm": 0.7806115746498108,
      "learning_rate": 7.021939893043237e-06,
      "loss": 0.7427,
      "step": 13413
    },
    {
      "epoch": 0.38650377456347607,
      "grad_norm": 1.4213805198669434,
      "learning_rate": 7.021513128433835e-06,
      "loss": 0.7288,
      "step": 13414
    },
    {
      "epoch": 0.38653258802512536,
      "grad_norm": 0.7667146325111389,
      "learning_rate": 7.021086346219074e-06,
      "loss": 0.7454,
      "step": 13415
    },
    {
      "epoch": 0.38656140148677465,
      "grad_norm": 0.7607215642929077,
      "learning_rate": 7.020659546402671e-06,
      "loss": 0.7341,
      "step": 13416
    },
    {
      "epoch": 0.3865902149484239,
      "grad_norm": 0.8077625036239624,
      "learning_rate": 7.020232728988342e-06,
      "loss": 0.7369,
      "step": 13417
    },
    {
      "epoch": 0.3866190284100732,
      "grad_norm": 0.8011221885681152,
      "learning_rate": 7.0198058939798056e-06,
      "loss": 0.7253,
      "step": 13418
    },
    {
      "epoch": 0.38664784187172246,
      "grad_norm": 0.7775962948799133,
      "learning_rate": 7.019379041380781e-06,
      "loss": 0.7349,
      "step": 13419
    },
    {
      "epoch": 0.38667665533337175,
      "grad_norm": 0.7928162813186646,
      "learning_rate": 7.018952171194981e-06,
      "loss": 0.7304,
      "step": 13420
    },
    {
      "epoch": 0.38670546879502105,
      "grad_norm": 0.8076457977294922,
      "learning_rate": 7.018525283426127e-06,
      "loss": 0.7311,
      "step": 13421
    },
    {
      "epoch": 0.38673428225667034,
      "grad_norm": 0.789971113204956,
      "learning_rate": 7.018098378077935e-06,
      "loss": 0.7314,
      "step": 13422
    },
    {
      "epoch": 0.38676309571831957,
      "grad_norm": 0.8788828253746033,
      "learning_rate": 7.0176714551541245e-06,
      "loss": 0.7337,
      "step": 13423
    },
    {
      "epoch": 0.38679190917996886,
      "grad_norm": 0.826739490032196,
      "learning_rate": 7.017244514658412e-06,
      "loss": 0.7309,
      "step": 13424
    },
    {
      "epoch": 0.38682072264161815,
      "grad_norm": 0.813884973526001,
      "learning_rate": 7.016817556594516e-06,
      "loss": 0.7319,
      "step": 13425
    },
    {
      "epoch": 0.38684953610326744,
      "grad_norm": 0.7819615602493286,
      "learning_rate": 7.016390580966157e-06,
      "loss": 0.7501,
      "step": 13426
    },
    {
      "epoch": 0.38687834956491673,
      "grad_norm": 0.8401505947113037,
      "learning_rate": 7.015963587777051e-06,
      "loss": 0.7309,
      "step": 13427
    },
    {
      "epoch": 0.386907163026566,
      "grad_norm": 0.8156852126121521,
      "learning_rate": 7.015536577030916e-06,
      "loss": 0.7395,
      "step": 13428
    },
    {
      "epoch": 0.3869359764882153,
      "grad_norm": 0.8190972208976746,
      "learning_rate": 7.015109548731473e-06,
      "loss": 0.699,
      "step": 13429
    },
    {
      "epoch": 0.38696478994986455,
      "grad_norm": 0.8008633255958557,
      "learning_rate": 7.014682502882439e-06,
      "loss": 0.7201,
      "step": 13430
    },
    {
      "epoch": 0.38699360341151384,
      "grad_norm": 0.8128229975700378,
      "learning_rate": 7.014255439487536e-06,
      "loss": 0.7245,
      "step": 13431
    },
    {
      "epoch": 0.38702241687316313,
      "grad_norm": 0.780552327632904,
      "learning_rate": 7.013828358550481e-06,
      "loss": 0.7345,
      "step": 13432
    },
    {
      "epoch": 0.3870512303348124,
      "grad_norm": 0.7887643575668335,
      "learning_rate": 7.013401260074994e-06,
      "loss": 0.7413,
      "step": 13433
    },
    {
      "epoch": 0.3870800437964617,
      "grad_norm": 0.8067939281463623,
      "learning_rate": 7.012974144064796e-06,
      "loss": 0.767,
      "step": 13434
    },
    {
      "epoch": 0.387108857258111,
      "grad_norm": 0.7548166513442993,
      "learning_rate": 7.012547010523605e-06,
      "loss": 0.736,
      "step": 13435
    },
    {
      "epoch": 0.3871376707197603,
      "grad_norm": 0.8456559181213379,
      "learning_rate": 7.01211985945514e-06,
      "loss": 0.7577,
      "step": 13436
    },
    {
      "epoch": 0.38716648418140953,
      "grad_norm": 0.7786241769790649,
      "learning_rate": 7.011692690863122e-06,
      "loss": 0.7236,
      "step": 13437
    },
    {
      "epoch": 0.3871952976430588,
      "grad_norm": 0.7812225222587585,
      "learning_rate": 7.011265504751274e-06,
      "loss": 0.7409,
      "step": 13438
    },
    {
      "epoch": 0.3872241111047081,
      "grad_norm": 0.8107535243034363,
      "learning_rate": 7.0108383011233125e-06,
      "loss": 0.7396,
      "step": 13439
    },
    {
      "epoch": 0.3872529245663574,
      "grad_norm": 0.7841975092887878,
      "learning_rate": 7.010411079982959e-06,
      "loss": 0.753,
      "step": 13440
    },
    {
      "epoch": 0.3872817380280067,
      "grad_norm": 0.7622168660163879,
      "learning_rate": 7.009983841333935e-06,
      "loss": 0.7502,
      "step": 13441
    },
    {
      "epoch": 0.387310551489656,
      "grad_norm": 0.771993100643158,
      "learning_rate": 7.009556585179961e-06,
      "loss": 0.7328,
      "step": 13442
    },
    {
      "epoch": 0.3873393649513053,
      "grad_norm": 0.7699947953224182,
      "learning_rate": 7.009129311524759e-06,
      "loss": 0.7243,
      "step": 13443
    },
    {
      "epoch": 0.3873681784129545,
      "grad_norm": 0.7327153086662292,
      "learning_rate": 7.008702020372048e-06,
      "loss": 0.7154,
      "step": 13444
    },
    {
      "epoch": 0.3873969918746038,
      "grad_norm": 0.7945801019668579,
      "learning_rate": 7.008274711725551e-06,
      "loss": 0.7416,
      "step": 13445
    },
    {
      "epoch": 0.3874258053362531,
      "grad_norm": 0.7926563024520874,
      "learning_rate": 7.007847385588988e-06,
      "loss": 0.7422,
      "step": 13446
    },
    {
      "epoch": 0.3874546187979024,
      "grad_norm": 0.7769578099250793,
      "learning_rate": 7.007420041966083e-06,
      "loss": 0.727,
      "step": 13447
    },
    {
      "epoch": 0.3874834322595517,
      "grad_norm": 0.7666195034980774,
      "learning_rate": 7.006992680860554e-06,
      "loss": 0.7252,
      "step": 13448
    },
    {
      "epoch": 0.38751224572120097,
      "grad_norm": 0.7843568921089172,
      "learning_rate": 7.006565302276126e-06,
      "loss": 0.7425,
      "step": 13449
    },
    {
      "epoch": 0.3875410591828502,
      "grad_norm": 0.816440999507904,
      "learning_rate": 7.00613790621652e-06,
      "loss": 0.7224,
      "step": 13450
    },
    {
      "epoch": 0.3875698726444995,
      "grad_norm": 0.7765073776245117,
      "learning_rate": 7.005710492685459e-06,
      "loss": 0.7191,
      "step": 13451
    },
    {
      "epoch": 0.3875986861061488,
      "grad_norm": 0.7528331279754639,
      "learning_rate": 7.005283061686664e-06,
      "loss": 0.7216,
      "step": 13452
    },
    {
      "epoch": 0.3876274995677981,
      "grad_norm": 0.8004789352416992,
      "learning_rate": 7.004855613223858e-06,
      "loss": 0.7457,
      "step": 13453
    },
    {
      "epoch": 0.38765631302944736,
      "grad_norm": 1.3634320497512817,
      "learning_rate": 7.0044281473007636e-06,
      "loss": 0.756,
      "step": 13454
    },
    {
      "epoch": 0.38768512649109665,
      "grad_norm": 0.7765142917633057,
      "learning_rate": 7.004000663921104e-06,
      "loss": 0.7322,
      "step": 13455
    },
    {
      "epoch": 0.38771393995274595,
      "grad_norm": 0.7624093294143677,
      "learning_rate": 7.003573163088602e-06,
      "loss": 0.7203,
      "step": 13456
    },
    {
      "epoch": 0.3877427534143952,
      "grad_norm": 0.7367938756942749,
      "learning_rate": 7.003145644806981e-06,
      "loss": 0.7185,
      "step": 13457
    },
    {
      "epoch": 0.38777156687604447,
      "grad_norm": 0.7808826565742493,
      "learning_rate": 7.0027181090799635e-06,
      "loss": 0.7424,
      "step": 13458
    },
    {
      "epoch": 0.38780038033769376,
      "grad_norm": 0.7541369795799255,
      "learning_rate": 7.002290555911272e-06,
      "loss": 0.7444,
      "step": 13459
    },
    {
      "epoch": 0.38782919379934305,
      "grad_norm": 0.7574939131736755,
      "learning_rate": 7.0018629853046326e-06,
      "loss": 0.7151,
      "step": 13460
    },
    {
      "epoch": 0.38785800726099234,
      "grad_norm": 0.8358484506607056,
      "learning_rate": 7.001435397263766e-06,
      "loss": 0.7442,
      "step": 13461
    },
    {
      "epoch": 0.38788682072264163,
      "grad_norm": 0.7970345616340637,
      "learning_rate": 7.0010077917924006e-06,
      "loss": 0.7752,
      "step": 13462
    },
    {
      "epoch": 0.3879156341842909,
      "grad_norm": 0.7644673585891724,
      "learning_rate": 7.0005801688942555e-06,
      "loss": 0.741,
      "step": 13463
    },
    {
      "epoch": 0.38794444764594016,
      "grad_norm": 0.7657188177108765,
      "learning_rate": 7.000152528573058e-06,
      "loss": 0.735,
      "step": 13464
    },
    {
      "epoch": 0.38797326110758945,
      "grad_norm": 0.7439926862716675,
      "learning_rate": 6.999724870832532e-06,
      "loss": 0.7237,
      "step": 13465
    },
    {
      "epoch": 0.38800207456923874,
      "grad_norm": 0.7545333504676819,
      "learning_rate": 6.999297195676399e-06,
      "loss": 0.7069,
      "step": 13466
    },
    {
      "epoch": 0.38803088803088803,
      "grad_norm": 0.7588849067687988,
      "learning_rate": 6.998869503108387e-06,
      "loss": 0.7537,
      "step": 13467
    },
    {
      "epoch": 0.3880597014925373,
      "grad_norm": 0.772977352142334,
      "learning_rate": 6.998441793132221e-06,
      "loss": 0.7251,
      "step": 13468
    },
    {
      "epoch": 0.3880885149541866,
      "grad_norm": 0.798004686832428,
      "learning_rate": 6.998014065751623e-06,
      "loss": 0.7412,
      "step": 13469
    },
    {
      "epoch": 0.3881173284158359,
      "grad_norm": 0.7415132522583008,
      "learning_rate": 6.997586320970322e-06,
      "loss": 0.7157,
      "step": 13470
    },
    {
      "epoch": 0.38814614187748514,
      "grad_norm": 0.7710871696472168,
      "learning_rate": 6.99715855879204e-06,
      "loss": 0.7372,
      "step": 13471
    },
    {
      "epoch": 0.38817495533913443,
      "grad_norm": 0.7833796143531799,
      "learning_rate": 6.9967307792205015e-06,
      "loss": 0.7654,
      "step": 13472
    },
    {
      "epoch": 0.3882037688007837,
      "grad_norm": 0.761448323726654,
      "learning_rate": 6.996302982259436e-06,
      "loss": 0.7304,
      "step": 13473
    },
    {
      "epoch": 0.388232582262433,
      "grad_norm": 0.7553778290748596,
      "learning_rate": 6.995875167912567e-06,
      "loss": 0.7457,
      "step": 13474
    },
    {
      "epoch": 0.3882613957240823,
      "grad_norm": 0.7569868564605713,
      "learning_rate": 6.99544733618362e-06,
      "loss": 0.7236,
      "step": 13475
    },
    {
      "epoch": 0.3882902091857316,
      "grad_norm": 0.7775837182998657,
      "learning_rate": 6.995019487076322e-06,
      "loss": 0.7345,
      "step": 13476
    },
    {
      "epoch": 0.38831902264738083,
      "grad_norm": 0.7744821310043335,
      "learning_rate": 6.994591620594397e-06,
      "loss": 0.7393,
      "step": 13477
    },
    {
      "epoch": 0.3883478361090301,
      "grad_norm": 0.7161028981208801,
      "learning_rate": 6.994163736741574e-06,
      "loss": 0.7291,
      "step": 13478
    },
    {
      "epoch": 0.3883766495706794,
      "grad_norm": 0.7329187989234924,
      "learning_rate": 6.993735835521578e-06,
      "loss": 0.7261,
      "step": 13479
    },
    {
      "epoch": 0.3884054630323287,
      "grad_norm": 0.7614918947219849,
      "learning_rate": 6.9933079169381355e-06,
      "loss": 0.7448,
      "step": 13480
    },
    {
      "epoch": 0.388434276493978,
      "grad_norm": 0.7563287019729614,
      "learning_rate": 6.992879980994974e-06,
      "loss": 0.7189,
      "step": 13481
    },
    {
      "epoch": 0.3884630899556273,
      "grad_norm": 0.7518003582954407,
      "learning_rate": 6.992452027695821e-06,
      "loss": 0.7401,
      "step": 13482
    },
    {
      "epoch": 0.3884919034172766,
      "grad_norm": 0.7530829906463623,
      "learning_rate": 6.992024057044402e-06,
      "loss": 0.7527,
      "step": 13483
    },
    {
      "epoch": 0.3885207168789258,
      "grad_norm": 0.7409534454345703,
      "learning_rate": 6.991596069044444e-06,
      "loss": 0.7183,
      "step": 13484
    },
    {
      "epoch": 0.3885495303405751,
      "grad_norm": 0.7647305727005005,
      "learning_rate": 6.991168063699675e-06,
      "loss": 0.7307,
      "step": 13485
    },
    {
      "epoch": 0.3885783438022244,
      "grad_norm": 0.7455063462257385,
      "learning_rate": 6.990740041013824e-06,
      "loss": 0.7168,
      "step": 13486
    },
    {
      "epoch": 0.3886071572638737,
      "grad_norm": 0.8014045357704163,
      "learning_rate": 6.990312000990615e-06,
      "loss": 0.7467,
      "step": 13487
    },
    {
      "epoch": 0.388635970725523,
      "grad_norm": 0.7878272533416748,
      "learning_rate": 6.989883943633781e-06,
      "loss": 0.731,
      "step": 13488
    },
    {
      "epoch": 0.38866478418717226,
      "grad_norm": 0.7483944892883301,
      "learning_rate": 6.989455868947046e-06,
      "loss": 0.7313,
      "step": 13489
    },
    {
      "epoch": 0.38869359764882155,
      "grad_norm": 0.7684438228607178,
      "learning_rate": 6.989027776934138e-06,
      "loss": 0.7307,
      "step": 13490
    },
    {
      "epoch": 0.3887224111104708,
      "grad_norm": 0.7617771029472351,
      "learning_rate": 6.988599667598786e-06,
      "loss": 0.7513,
      "step": 13491
    },
    {
      "epoch": 0.3887512245721201,
      "grad_norm": 0.8100094795227051,
      "learning_rate": 6.98817154094472e-06,
      "loss": 0.7463,
      "step": 13492
    },
    {
      "epoch": 0.38878003803376937,
      "grad_norm": 0.7830343246459961,
      "learning_rate": 6.987743396975668e-06,
      "loss": 0.7385,
      "step": 13493
    },
    {
      "epoch": 0.38880885149541866,
      "grad_norm": 0.7550753951072693,
      "learning_rate": 6.987315235695357e-06,
      "loss": 0.7286,
      "step": 13494
    },
    {
      "epoch": 0.38883766495706795,
      "grad_norm": 0.7685143351554871,
      "learning_rate": 6.986887057107516e-06,
      "loss": 0.7484,
      "step": 13495
    },
    {
      "epoch": 0.38886647841871724,
      "grad_norm": 0.7383798360824585,
      "learning_rate": 6.986458861215875e-06,
      "loss": 0.7269,
      "step": 13496
    },
    {
      "epoch": 0.38889529188036653,
      "grad_norm": 0.7511153221130371,
      "learning_rate": 6.986030648024165e-06,
      "loss": 0.7321,
      "step": 13497
    },
    {
      "epoch": 0.38892410534201577,
      "grad_norm": 0.7738615274429321,
      "learning_rate": 6.985602417536113e-06,
      "loss": 0.723,
      "step": 13498
    },
    {
      "epoch": 0.38895291880366506,
      "grad_norm": 0.7458723187446594,
      "learning_rate": 6.985174169755446e-06,
      "loss": 0.7062,
      "step": 13499
    },
    {
      "epoch": 0.38898173226531435,
      "grad_norm": 0.7586478590965271,
      "learning_rate": 6.984745904685899e-06,
      "loss": 0.721,
      "step": 13500
    },
    {
      "epoch": 0.38901054572696364,
      "grad_norm": 0.7743328213691711,
      "learning_rate": 6.984317622331199e-06,
      "loss": 0.7261,
      "step": 13501
    },
    {
      "epoch": 0.38903935918861293,
      "grad_norm": 0.7868052124977112,
      "learning_rate": 6.983889322695075e-06,
      "loss": 0.7161,
      "step": 13502
    },
    {
      "epoch": 0.3890681726502622,
      "grad_norm": 0.7356621026992798,
      "learning_rate": 6.98346100578126e-06,
      "loss": 0.7144,
      "step": 13503
    },
    {
      "epoch": 0.38909698611191146,
      "grad_norm": 0.7632468342781067,
      "learning_rate": 6.983032671593482e-06,
      "loss": 0.7399,
      "step": 13504
    },
    {
      "epoch": 0.38912579957356075,
      "grad_norm": 0.7584823966026306,
      "learning_rate": 6.982604320135471e-06,
      "loss": 0.7163,
      "step": 13505
    },
    {
      "epoch": 0.38915461303521004,
      "grad_norm": 0.7700862288475037,
      "learning_rate": 6.982175951410958e-06,
      "loss": 0.7266,
      "step": 13506
    },
    {
      "epoch": 0.38918342649685933,
      "grad_norm": 0.7519564032554626,
      "learning_rate": 6.981747565423674e-06,
      "loss": 0.7148,
      "step": 13507
    },
    {
      "epoch": 0.3892122399585086,
      "grad_norm": 0.7695292234420776,
      "learning_rate": 6.9813191621773495e-06,
      "loss": 0.7245,
      "step": 13508
    },
    {
      "epoch": 0.3892410534201579,
      "grad_norm": 0.7576800584793091,
      "learning_rate": 6.980890741675717e-06,
      "loss": 0.7444,
      "step": 13509
    },
    {
      "epoch": 0.3892698668818072,
      "grad_norm": 0.7507606148719788,
      "learning_rate": 6.980462303922505e-06,
      "loss": 0.7077,
      "step": 13510
    },
    {
      "epoch": 0.38929868034345644,
      "grad_norm": 0.7456789016723633,
      "learning_rate": 6.980033848921447e-06,
      "loss": 0.7438,
      "step": 13511
    },
    {
      "epoch": 0.38932749380510573,
      "grad_norm": 0.7795290350914001,
      "learning_rate": 6.979605376676273e-06,
      "loss": 0.7468,
      "step": 13512
    },
    {
      "epoch": 0.389356307266755,
      "grad_norm": 0.7636303305625916,
      "learning_rate": 6.979176887190715e-06,
      "loss": 0.7289,
      "step": 13513
    },
    {
      "epoch": 0.3893851207284043,
      "grad_norm": 0.7406308650970459,
      "learning_rate": 6.9787483804685045e-06,
      "loss": 0.7342,
      "step": 13514
    },
    {
      "epoch": 0.3894139341900536,
      "grad_norm": 0.7558963894844055,
      "learning_rate": 6.978319856513374e-06,
      "loss": 0.7211,
      "step": 13515
    },
    {
      "epoch": 0.3894427476517029,
      "grad_norm": 0.7579062581062317,
      "learning_rate": 6.977891315329055e-06,
      "loss": 0.7088,
      "step": 13516
    },
    {
      "epoch": 0.3894715611133522,
      "grad_norm": 0.7735475897789001,
      "learning_rate": 6.9774627569192795e-06,
      "loss": 0.6995,
      "step": 13517
    },
    {
      "epoch": 0.3895003745750014,
      "grad_norm": 0.7709737420082092,
      "learning_rate": 6.977034181287781e-06,
      "loss": 0.7071,
      "step": 13518
    },
    {
      "epoch": 0.3895291880366507,
      "grad_norm": 0.7485153079032898,
      "learning_rate": 6.97660558843829e-06,
      "loss": 0.7111,
      "step": 13519
    },
    {
      "epoch": 0.3895580014983,
      "grad_norm": 0.7803723216056824,
      "learning_rate": 6.976176978374541e-06,
      "loss": 0.7076,
      "step": 13520
    },
    {
      "epoch": 0.3895868149599493,
      "grad_norm": 0.7612702250480652,
      "learning_rate": 6.9757483511002655e-06,
      "loss": 0.7412,
      "step": 13521
    },
    {
      "epoch": 0.3896156284215986,
      "grad_norm": 0.7721981406211853,
      "learning_rate": 6.975319706619197e-06,
      "loss": 0.7344,
      "step": 13522
    },
    {
      "epoch": 0.38964444188324787,
      "grad_norm": 0.7457444071769714,
      "learning_rate": 6.974891044935069e-06,
      "loss": 0.7205,
      "step": 13523
    },
    {
      "epoch": 0.38967325534489716,
      "grad_norm": 0.772022008895874,
      "learning_rate": 6.974462366051614e-06,
      "loss": 0.7432,
      "step": 13524
    },
    {
      "epoch": 0.3897020688065464,
      "grad_norm": 0.7545737624168396,
      "learning_rate": 6.974033669972566e-06,
      "loss": 0.7359,
      "step": 13525
    },
    {
      "epoch": 0.3897308822681957,
      "grad_norm": 0.7608352303504944,
      "learning_rate": 6.973604956701656e-06,
      "loss": 0.7372,
      "step": 13526
    },
    {
      "epoch": 0.389759695729845,
      "grad_norm": 0.72576504945755,
      "learning_rate": 6.973176226242622e-06,
      "loss": 0.7272,
      "step": 13527
    },
    {
      "epoch": 0.38978850919149427,
      "grad_norm": 0.8019658923149109,
      "learning_rate": 6.972747478599194e-06,
      "loss": 0.7382,
      "step": 13528
    },
    {
      "epoch": 0.38981732265314356,
      "grad_norm": 0.7719135284423828,
      "learning_rate": 6.972318713775109e-06,
      "loss": 0.7704,
      "step": 13529
    },
    {
      "epoch": 0.38984613611479285,
      "grad_norm": 0.7934250235557556,
      "learning_rate": 6.9718899317740984e-06,
      "loss": 0.7078,
      "step": 13530
    },
    {
      "epoch": 0.3898749495764421,
      "grad_norm": 0.7805318832397461,
      "learning_rate": 6.971461132599899e-06,
      "loss": 0.7485,
      "step": 13531
    },
    {
      "epoch": 0.3899037630380914,
      "grad_norm": 0.7355784773826599,
      "learning_rate": 6.9710323162562425e-06,
      "loss": 0.7224,
      "step": 13532
    },
    {
      "epoch": 0.38993257649974067,
      "grad_norm": 0.7322406768798828,
      "learning_rate": 6.970603482746865e-06,
      "loss": 0.7076,
      "step": 13533
    },
    {
      "epoch": 0.38996138996138996,
      "grad_norm": 0.7455061078071594,
      "learning_rate": 6.970174632075501e-06,
      "loss": 0.7397,
      "step": 13534
    },
    {
      "epoch": 0.38999020342303925,
      "grad_norm": 0.7585722208023071,
      "learning_rate": 6.969745764245885e-06,
      "loss": 0.7651,
      "step": 13535
    },
    {
      "epoch": 0.39001901688468854,
      "grad_norm": 0.7145497798919678,
      "learning_rate": 6.969316879261754e-06,
      "loss": 0.7343,
      "step": 13536
    },
    {
      "epoch": 0.39004783034633783,
      "grad_norm": 0.7634356021881104,
      "learning_rate": 6.96888797712684e-06,
      "loss": 0.7365,
      "step": 13537
    },
    {
      "epoch": 0.39007664380798707,
      "grad_norm": 0.7612000703811646,
      "learning_rate": 6.968459057844881e-06,
      "loss": 0.712,
      "step": 13538
    },
    {
      "epoch": 0.39010545726963636,
      "grad_norm": 0.7789143323898315,
      "learning_rate": 6.968030121419611e-06,
      "loss": 0.7282,
      "step": 13539
    },
    {
      "epoch": 0.39013427073128565,
      "grad_norm": 0.7268484234809875,
      "learning_rate": 6.967601167854765e-06,
      "loss": 0.7453,
      "step": 13540
    },
    {
      "epoch": 0.39016308419293494,
      "grad_norm": 0.7566686272621155,
      "learning_rate": 6.967172197154081e-06,
      "loss": 0.7401,
      "step": 13541
    },
    {
      "epoch": 0.39019189765458423,
      "grad_norm": 0.7566941380500793,
      "learning_rate": 6.966743209321292e-06,
      "loss": 0.726,
      "step": 13542
    },
    {
      "epoch": 0.3902207111162335,
      "grad_norm": 0.8174333572387695,
      "learning_rate": 6.966314204360138e-06,
      "loss": 0.7343,
      "step": 13543
    },
    {
      "epoch": 0.3902495245778828,
      "grad_norm": 0.8012925386428833,
      "learning_rate": 6.965885182274351e-06,
      "loss": 0.7292,
      "step": 13544
    },
    {
      "epoch": 0.39027833803953205,
      "grad_norm": 0.7705973982810974,
      "learning_rate": 6.965456143067671e-06,
      "loss": 0.7389,
      "step": 13545
    },
    {
      "epoch": 0.39030715150118134,
      "grad_norm": 0.7698975801467896,
      "learning_rate": 6.965027086743831e-06,
      "loss": 0.716,
      "step": 13546
    },
    {
      "epoch": 0.39033596496283063,
      "grad_norm": 0.7808730006217957,
      "learning_rate": 6.9645980133065705e-06,
      "loss": 0.7387,
      "step": 13547
    },
    {
      "epoch": 0.3903647784244799,
      "grad_norm": 0.7788817286491394,
      "learning_rate": 6.964168922759624e-06,
      "loss": 0.7445,
      "step": 13548
    },
    {
      "epoch": 0.3903935918861292,
      "grad_norm": 0.727451503276825,
      "learning_rate": 6.963739815106731e-06,
      "loss": 0.7182,
      "step": 13549
    },
    {
      "epoch": 0.3904224053477785,
      "grad_norm": 0.7707175612449646,
      "learning_rate": 6.963310690351626e-06,
      "loss": 0.7447,
      "step": 13550
    },
    {
      "epoch": 0.3904512188094278,
      "grad_norm": 0.7655211687088013,
      "learning_rate": 6.962881548498048e-06,
      "loss": 0.7387,
      "step": 13551
    },
    {
      "epoch": 0.390480032271077,
      "grad_norm": 0.7526623606681824,
      "learning_rate": 6.962452389549734e-06,
      "loss": 0.7438,
      "step": 13552
    },
    {
      "epoch": 0.3905088457327263,
      "grad_norm": 0.7488507628440857,
      "learning_rate": 6.962023213510423e-06,
      "loss": 0.719,
      "step": 13553
    },
    {
      "epoch": 0.3905376591943756,
      "grad_norm": 0.7568005323410034,
      "learning_rate": 6.9615940203838485e-06,
      "loss": 0.732,
      "step": 13554
    },
    {
      "epoch": 0.3905664726560249,
      "grad_norm": 0.7392094135284424,
      "learning_rate": 6.961164810173753e-06,
      "loss": 0.7295,
      "step": 13555
    },
    {
      "epoch": 0.3905952861176742,
      "grad_norm": 0.7552598118782043,
      "learning_rate": 6.960735582883871e-06,
      "loss": 0.7583,
      "step": 13556
    },
    {
      "epoch": 0.3906240995793235,
      "grad_norm": 0.7434862852096558,
      "learning_rate": 6.960306338517941e-06,
      "loss": 0.7286,
      "step": 13557
    },
    {
      "epoch": 0.3906529130409727,
      "grad_norm": 0.767288863658905,
      "learning_rate": 6.959877077079704e-06,
      "loss": 0.7453,
      "step": 13558
    },
    {
      "epoch": 0.390681726502622,
      "grad_norm": 0.7559803128242493,
      "learning_rate": 6.959447798572897e-06,
      "loss": 0.7205,
      "step": 13559
    },
    {
      "epoch": 0.3907105399642713,
      "grad_norm": 0.7190377712249756,
      "learning_rate": 6.959018503001258e-06,
      "loss": 0.7319,
      "step": 13560
    },
    {
      "epoch": 0.3907393534259206,
      "grad_norm": 0.7926653027534485,
      "learning_rate": 6.958589190368527e-06,
      "loss": 0.7487,
      "step": 13561
    },
    {
      "epoch": 0.3907681668875699,
      "grad_norm": 0.7500717043876648,
      "learning_rate": 6.95815986067844e-06,
      "loss": 0.7209,
      "step": 13562
    },
    {
      "epoch": 0.39079698034921917,
      "grad_norm": 0.7930117249488831,
      "learning_rate": 6.957730513934739e-06,
      "loss": 0.7191,
      "step": 13563
    },
    {
      "epoch": 0.39082579381086846,
      "grad_norm": 0.7743133306503296,
      "learning_rate": 6.957301150141163e-06,
      "loss": 0.7548,
      "step": 13564
    },
    {
      "epoch": 0.3908546072725177,
      "grad_norm": 0.7705294489860535,
      "learning_rate": 6.9568717693014495e-06,
      "loss": 0.7344,
      "step": 13565
    },
    {
      "epoch": 0.390883420734167,
      "grad_norm": 0.8097782135009766,
      "learning_rate": 6.956442371419341e-06,
      "loss": 0.7518,
      "step": 13566
    },
    {
      "epoch": 0.3909122341958163,
      "grad_norm": 0.7592378854751587,
      "learning_rate": 6.9560129564985746e-06,
      "loss": 0.7139,
      "step": 13567
    },
    {
      "epoch": 0.39094104765746557,
      "grad_norm": 0.7654987573623657,
      "learning_rate": 6.955583524542888e-06,
      "loss": 0.7603,
      "step": 13568
    },
    {
      "epoch": 0.39096986111911486,
      "grad_norm": 0.7761741876602173,
      "learning_rate": 6.955154075556027e-06,
      "loss": 0.7186,
      "step": 13569
    },
    {
      "epoch": 0.39099867458076415,
      "grad_norm": 0.7582441568374634,
      "learning_rate": 6.954724609541726e-06,
      "loss": 0.7385,
      "step": 13570
    },
    {
      "epoch": 0.39102748804241344,
      "grad_norm": 0.7650005221366882,
      "learning_rate": 6.95429512650373e-06,
      "loss": 0.7226,
      "step": 13571
    },
    {
      "epoch": 0.3910563015040627,
      "grad_norm": 0.7781538367271423,
      "learning_rate": 6.953865626445775e-06,
      "loss": 0.7329,
      "step": 13572
    },
    {
      "epoch": 0.39108511496571197,
      "grad_norm": 0.7556276917457581,
      "learning_rate": 6.953436109371605e-06,
      "loss": 0.7277,
      "step": 13573
    },
    {
      "epoch": 0.39111392842736126,
      "grad_norm": 0.7452958822250366,
      "learning_rate": 6.953006575284957e-06,
      "loss": 0.7381,
      "step": 13574
    },
    {
      "epoch": 0.39114274188901055,
      "grad_norm": 0.7537205219268799,
      "learning_rate": 6.952577024189576e-06,
      "loss": 0.7473,
      "step": 13575
    },
    {
      "epoch": 0.39117155535065984,
      "grad_norm": 0.7580601572990417,
      "learning_rate": 6.952147456089201e-06,
      "loss": 0.7306,
      "step": 13576
    },
    {
      "epoch": 0.39120036881230913,
      "grad_norm": 0.7503858804702759,
      "learning_rate": 6.951717870987573e-06,
      "loss": 0.7158,
      "step": 13577
    },
    {
      "epoch": 0.3912291822739584,
      "grad_norm": 0.7618153691291809,
      "learning_rate": 6.951288268888432e-06,
      "loss": 0.7465,
      "step": 13578
    },
    {
      "epoch": 0.39125799573560766,
      "grad_norm": 0.772315502166748,
      "learning_rate": 6.950858649795522e-06,
      "loss": 0.738,
      "step": 13579
    },
    {
      "epoch": 0.39128680919725695,
      "grad_norm": 0.7670373320579529,
      "learning_rate": 6.950429013712583e-06,
      "loss": 0.731,
      "step": 13580
    },
    {
      "epoch": 0.39131562265890624,
      "grad_norm": 0.7670548558235168,
      "learning_rate": 6.949999360643357e-06,
      "loss": 0.7403,
      "step": 13581
    },
    {
      "epoch": 0.3913444361205555,
      "grad_norm": 0.7318572998046875,
      "learning_rate": 6.949569690591586e-06,
      "loss": 0.7406,
      "step": 13582
    },
    {
      "epoch": 0.3913732495822048,
      "grad_norm": 0.7283187508583069,
      "learning_rate": 6.9491400035610116e-06,
      "loss": 0.7439,
      "step": 13583
    },
    {
      "epoch": 0.3914020630438541,
      "grad_norm": 0.7548932433128357,
      "learning_rate": 6.948710299555377e-06,
      "loss": 0.7713,
      "step": 13584
    },
    {
      "epoch": 0.39143087650550334,
      "grad_norm": 0.8087356090545654,
      "learning_rate": 6.948280578578422e-06,
      "loss": 0.7489,
      "step": 13585
    },
    {
      "epoch": 0.39145968996715264,
      "grad_norm": 0.7437154650688171,
      "learning_rate": 6.947850840633892e-06,
      "loss": 0.7012,
      "step": 13586
    },
    {
      "epoch": 0.3914885034288019,
      "grad_norm": 0.8114485144615173,
      "learning_rate": 6.947421085725528e-06,
      "loss": 0.7513,
      "step": 13587
    },
    {
      "epoch": 0.3915173168904512,
      "grad_norm": 0.7656626105308533,
      "learning_rate": 6.9469913138570735e-06,
      "loss": 0.7384,
      "step": 13588
    },
    {
      "epoch": 0.3915461303521005,
      "grad_norm": 0.7793855667114258,
      "learning_rate": 6.94656152503227e-06,
      "loss": 0.7268,
      "step": 13589
    },
    {
      "epoch": 0.3915749438137498,
      "grad_norm": 0.7328400611877441,
      "learning_rate": 6.946131719254863e-06,
      "loss": 0.7637,
      "step": 13590
    },
    {
      "epoch": 0.3916037572753991,
      "grad_norm": 0.7683141827583313,
      "learning_rate": 6.945701896528594e-06,
      "loss": 0.7212,
      "step": 13591
    },
    {
      "epoch": 0.3916325707370483,
      "grad_norm": 0.7525392770767212,
      "learning_rate": 6.945272056857204e-06,
      "loss": 0.7242,
      "step": 13592
    },
    {
      "epoch": 0.3916613841986976,
      "grad_norm": 0.7808974981307983,
      "learning_rate": 6.944842200244441e-06,
      "loss": 0.7384,
      "step": 13593
    },
    {
      "epoch": 0.3916901976603469,
      "grad_norm": 0.7504081726074219,
      "learning_rate": 6.944412326694045e-06,
      "loss": 0.7183,
      "step": 13594
    },
    {
      "epoch": 0.3917190111219962,
      "grad_norm": 0.7673866748809814,
      "learning_rate": 6.9439824362097645e-06,
      "loss": 0.7095,
      "step": 13595
    },
    {
      "epoch": 0.3917478245836455,
      "grad_norm": 0.7638444900512695,
      "learning_rate": 6.9435525287953365e-06,
      "loss": 0.7124,
      "step": 13596
    },
    {
      "epoch": 0.3917766380452948,
      "grad_norm": 0.7250419855117798,
      "learning_rate": 6.943122604454511e-06,
      "loss": 0.7305,
      "step": 13597
    },
    {
      "epoch": 0.39180545150694407,
      "grad_norm": 0.7417078018188477,
      "learning_rate": 6.9426926631910294e-06,
      "loss": 0.7333,
      "step": 13598
    },
    {
      "epoch": 0.3918342649685933,
      "grad_norm": 0.7433469295501709,
      "learning_rate": 6.9422627050086365e-06,
      "loss": 0.7467,
      "step": 13599
    },
    {
      "epoch": 0.3918630784302426,
      "grad_norm": 0.7852305769920349,
      "learning_rate": 6.941832729911078e-06,
      "loss": 0.7724,
      "step": 13600
    },
    {
      "epoch": 0.3918918918918919,
      "grad_norm": 0.7380136847496033,
      "learning_rate": 6.941402737902096e-06,
      "loss": 0.7338,
      "step": 13601
    },
    {
      "epoch": 0.3919207053535412,
      "grad_norm": 1.0342776775360107,
      "learning_rate": 6.940972728985439e-06,
      "loss": 0.7439,
      "step": 13602
    },
    {
      "epoch": 0.39194951881519047,
      "grad_norm": 0.7421815991401672,
      "learning_rate": 6.940542703164849e-06,
      "loss": 0.7502,
      "step": 13603
    },
    {
      "epoch": 0.39197833227683976,
      "grad_norm": 0.7428436279296875,
      "learning_rate": 6.940112660444069e-06,
      "loss": 0.7514,
      "step": 13604
    },
    {
      "epoch": 0.39200714573848905,
      "grad_norm": 0.7558738589286804,
      "learning_rate": 6.939682600826852e-06,
      "loss": 0.7564,
      "step": 13605
    },
    {
      "epoch": 0.3920359592001383,
      "grad_norm": 0.7484170198440552,
      "learning_rate": 6.939252524316935e-06,
      "loss": 0.751,
      "step": 13606
    },
    {
      "epoch": 0.3920647726617876,
      "grad_norm": 0.7396354675292969,
      "learning_rate": 6.938822430918068e-06,
      "loss": 0.719,
      "step": 13607
    },
    {
      "epoch": 0.39209358612343687,
      "grad_norm": 0.7507046461105347,
      "learning_rate": 6.938392320633996e-06,
      "loss": 0.6985,
      "step": 13608
    },
    {
      "epoch": 0.39212239958508616,
      "grad_norm": 0.7274302244186401,
      "learning_rate": 6.937962193468465e-06,
      "loss": 0.7224,
      "step": 13609
    },
    {
      "epoch": 0.39215121304673545,
      "grad_norm": 0.7275493741035461,
      "learning_rate": 6.93753204942522e-06,
      "loss": 0.7135,
      "step": 13610
    },
    {
      "epoch": 0.39218002650838474,
      "grad_norm": 0.7835533022880554,
      "learning_rate": 6.9371018885080064e-06,
      "loss": 0.742,
      "step": 13611
    },
    {
      "epoch": 0.392208839970034,
      "grad_norm": 0.8001426458358765,
      "learning_rate": 6.9366717107205736e-06,
      "loss": 0.7211,
      "step": 13612
    },
    {
      "epoch": 0.39223765343168326,
      "grad_norm": 0.7343330979347229,
      "learning_rate": 6.936241516066667e-06,
      "loss": 0.7288,
      "step": 13613
    },
    {
      "epoch": 0.39226646689333255,
      "grad_norm": 0.7350882291793823,
      "learning_rate": 6.935811304550031e-06,
      "loss": 0.7316,
      "step": 13614
    },
    {
      "epoch": 0.39229528035498185,
      "grad_norm": 1.600974202156067,
      "learning_rate": 6.935381076174413e-06,
      "loss": 0.7418,
      "step": 13615
    },
    {
      "epoch": 0.39232409381663114,
      "grad_norm": 0.7391193509101868,
      "learning_rate": 6.934950830943561e-06,
      "loss": 0.7292,
      "step": 13616
    },
    {
      "epoch": 0.3923529072782804,
      "grad_norm": 0.7609961032867432,
      "learning_rate": 6.93452056886122e-06,
      "loss": 0.7331,
      "step": 13617
    },
    {
      "epoch": 0.3923817207399297,
      "grad_norm": 0.7622455358505249,
      "learning_rate": 6.93409028993114e-06,
      "loss": 0.7339,
      "step": 13618
    },
    {
      "epoch": 0.39241053420157895,
      "grad_norm": 0.779733419418335,
      "learning_rate": 6.9336599941570684e-06,
      "loss": 0.7246,
      "step": 13619
    },
    {
      "epoch": 0.39243934766322824,
      "grad_norm": 0.7492197155952454,
      "learning_rate": 6.93322968154275e-06,
      "loss": 0.7243,
      "step": 13620
    },
    {
      "epoch": 0.39246816112487753,
      "grad_norm": 0.7398223876953125,
      "learning_rate": 6.932799352091934e-06,
      "loss": 0.7361,
      "step": 13621
    },
    {
      "epoch": 0.3924969745865268,
      "grad_norm": 0.7444704174995422,
      "learning_rate": 6.932369005808366e-06,
      "loss": 0.7459,
      "step": 13622
    },
    {
      "epoch": 0.3925257880481761,
      "grad_norm": 0.7496070861816406,
      "learning_rate": 6.931938642695797e-06,
      "loss": 0.7213,
      "step": 13623
    },
    {
      "epoch": 0.3925546015098254,
      "grad_norm": 0.761983335018158,
      "learning_rate": 6.9315082627579735e-06,
      "loss": 0.7547,
      "step": 13624
    },
    {
      "epoch": 0.3925834149714747,
      "grad_norm": 0.7422485947608948,
      "learning_rate": 6.931077865998644e-06,
      "loss": 0.7343,
      "step": 13625
    },
    {
      "epoch": 0.39261222843312393,
      "grad_norm": 0.7545139193534851,
      "learning_rate": 6.930647452421557e-06,
      "loss": 0.7555,
      "step": 13626
    },
    {
      "epoch": 0.3926410418947732,
      "grad_norm": 0.7401263117790222,
      "learning_rate": 6.930217022030461e-06,
      "loss": 0.7317,
      "step": 13627
    },
    {
      "epoch": 0.3926698553564225,
      "grad_norm": 0.7494049668312073,
      "learning_rate": 6.929786574829103e-06,
      "loss": 0.7303,
      "step": 13628
    },
    {
      "epoch": 0.3926986688180718,
      "grad_norm": 0.7551189661026001,
      "learning_rate": 6.929356110821233e-06,
      "loss": 0.7522,
      "step": 13629
    },
    {
      "epoch": 0.3927274822797211,
      "grad_norm": 0.7584486603736877,
      "learning_rate": 6.9289256300106e-06,
      "loss": 0.7598,
      "step": 13630
    },
    {
      "epoch": 0.3927562957413704,
      "grad_norm": 0.7341799139976501,
      "learning_rate": 6.9284951324009544e-06,
      "loss": 0.7302,
      "step": 13631
    },
    {
      "epoch": 0.3927851092030197,
      "grad_norm": 0.7491787672042847,
      "learning_rate": 6.928064617996042e-06,
      "loss": 0.7384,
      "step": 13632
    },
    {
      "epoch": 0.3928139226646689,
      "grad_norm": 0.7394197583198547,
      "learning_rate": 6.927634086799616e-06,
      "loss": 0.7203,
      "step": 13633
    },
    {
      "epoch": 0.3928427361263182,
      "grad_norm": 0.760225236415863,
      "learning_rate": 6.927203538815423e-06,
      "loss": 0.7379,
      "step": 13634
    },
    {
      "epoch": 0.3928715495879675,
      "grad_norm": 5.341920375823975,
      "learning_rate": 6.9267729740472135e-06,
      "loss": 0.7135,
      "step": 13635
    },
    {
      "epoch": 0.3929003630496168,
      "grad_norm": 0.7592067122459412,
      "learning_rate": 6.926342392498737e-06,
      "loss": 0.7328,
      "step": 13636
    },
    {
      "epoch": 0.3929291765112661,
      "grad_norm": 0.7667611241340637,
      "learning_rate": 6.925911794173745e-06,
      "loss": 0.7383,
      "step": 13637
    },
    {
      "epoch": 0.39295798997291537,
      "grad_norm": 0.7733917236328125,
      "learning_rate": 6.925481179075986e-06,
      "loss": 0.7187,
      "step": 13638
    },
    {
      "epoch": 0.3929868034345646,
      "grad_norm": 0.7820752263069153,
      "learning_rate": 6.9250505472092115e-06,
      "loss": 0.7445,
      "step": 13639
    },
    {
      "epoch": 0.3930156168962139,
      "grad_norm": 0.7940263152122498,
      "learning_rate": 6.92461989857717e-06,
      "loss": 0.7133,
      "step": 13640
    },
    {
      "epoch": 0.3930444303578632,
      "grad_norm": 0.7541308999061584,
      "learning_rate": 6.924189233183612e-06,
      "loss": 0.7001,
      "step": 13641
    },
    {
      "epoch": 0.3930732438195125,
      "grad_norm": 0.7885478734970093,
      "learning_rate": 6.923758551032291e-06,
      "loss": 0.7509,
      "step": 13642
    },
    {
      "epoch": 0.39310205728116177,
      "grad_norm": 0.769711971282959,
      "learning_rate": 6.923327852126958e-06,
      "loss": 0.7394,
      "step": 13643
    },
    {
      "epoch": 0.39313087074281106,
      "grad_norm": 0.8059483170509338,
      "learning_rate": 6.9228971364713595e-06,
      "loss": 0.7503,
      "step": 13644
    },
    {
      "epoch": 0.39315968420446035,
      "grad_norm": 0.9511551260948181,
      "learning_rate": 6.922466404069249e-06,
      "loss": 0.7836,
      "step": 13645
    },
    {
      "epoch": 0.3931884976661096,
      "grad_norm": 0.8135656714439392,
      "learning_rate": 6.922035654924377e-06,
      "loss": 0.7656,
      "step": 13646
    },
    {
      "epoch": 0.3932173111277589,
      "grad_norm": 0.8049529790878296,
      "learning_rate": 6.9216048890404985e-06,
      "loss": 0.7362,
      "step": 13647
    },
    {
      "epoch": 0.39324612458940816,
      "grad_norm": 0.7826334238052368,
      "learning_rate": 6.921174106421361e-06,
      "loss": 0.7421,
      "step": 13648
    },
    {
      "epoch": 0.39327493805105745,
      "grad_norm": 0.8023945093154907,
      "learning_rate": 6.920743307070717e-06,
      "loss": 0.7291,
      "step": 13649
    },
    {
      "epoch": 0.39330375151270675,
      "grad_norm": 0.7905008792877197,
      "learning_rate": 6.92031249099232e-06,
      "loss": 0.7318,
      "step": 13650
    },
    {
      "epoch": 0.39333256497435604,
      "grad_norm": 0.9342812299728394,
      "learning_rate": 6.919881658189921e-06,
      "loss": 0.7829,
      "step": 13651
    },
    {
      "epoch": 0.3933613784360053,
      "grad_norm": 0.7944200038909912,
      "learning_rate": 6.9194508086672695e-06,
      "loss": 0.7449,
      "step": 13652
    },
    {
      "epoch": 0.39339019189765456,
      "grad_norm": 0.7525548338890076,
      "learning_rate": 6.9190199424281225e-06,
      "loss": 0.722,
      "step": 13653
    },
    {
      "epoch": 0.39341900535930385,
      "grad_norm": 0.7942398190498352,
      "learning_rate": 6.918589059476229e-06,
      "loss": 0.7516,
      "step": 13654
    },
    {
      "epoch": 0.39344781882095314,
      "grad_norm": 0.8214628100395203,
      "learning_rate": 6.918158159815343e-06,
      "loss": 0.7411,
      "step": 13655
    },
    {
      "epoch": 0.39347663228260243,
      "grad_norm": 0.7738909125328064,
      "learning_rate": 6.917727243449217e-06,
      "loss": 0.742,
      "step": 13656
    },
    {
      "epoch": 0.3935054457442517,
      "grad_norm": 1.2779383659362793,
      "learning_rate": 6.917296310381604e-06,
      "loss": 0.7591,
      "step": 13657
    },
    {
      "epoch": 0.393534259205901,
      "grad_norm": 0.7725021243095398,
      "learning_rate": 6.916865360616256e-06,
      "loss": 0.7309,
      "step": 13658
    },
    {
      "epoch": 0.3935630726675503,
      "grad_norm": 0.760139524936676,
      "learning_rate": 6.916434394156927e-06,
      "loss": 0.7451,
      "step": 13659
    },
    {
      "epoch": 0.39359188612919954,
      "grad_norm": 0.839628279209137,
      "learning_rate": 6.91600341100737e-06,
      "loss": 0.7342,
      "step": 13660
    },
    {
      "epoch": 0.39362069959084883,
      "grad_norm": 0.7925708293914795,
      "learning_rate": 6.915572411171341e-06,
      "loss": 0.7658,
      "step": 13661
    },
    {
      "epoch": 0.3936495130524981,
      "grad_norm": 0.7694087624549866,
      "learning_rate": 6.915141394652589e-06,
      "loss": 0.7306,
      "step": 13662
    },
    {
      "epoch": 0.3936783265141474,
      "grad_norm": 0.8040652871131897,
      "learning_rate": 6.914710361454871e-06,
      "loss": 0.7435,
      "step": 13663
    },
    {
      "epoch": 0.3937071399757967,
      "grad_norm": 0.740056574344635,
      "learning_rate": 6.914279311581937e-06,
      "loss": 0.7191,
      "step": 13664
    },
    {
      "epoch": 0.393735953437446,
      "grad_norm": 0.7619879245758057,
      "learning_rate": 6.913848245037546e-06,
      "loss": 0.7188,
      "step": 13665
    },
    {
      "epoch": 0.39376476689909523,
      "grad_norm": 0.7488957643508911,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.7203,
      "step": 13666
    },
    {
      "epoch": 0.3937935803607445,
      "grad_norm": 0.7767561674118042,
      "learning_rate": 6.912986061949403e-06,
      "loss": 0.7414,
      "step": 13667
    },
    {
      "epoch": 0.3938223938223938,
      "grad_norm": 0.808869481086731,
      "learning_rate": 6.912554945413159e-06,
      "loss": 0.7194,
      "step": 13668
    },
    {
      "epoch": 0.3938512072840431,
      "grad_norm": 0.7264196276664734,
      "learning_rate": 6.912123812220474e-06,
      "loss": 0.7137,
      "step": 13669
    },
    {
      "epoch": 0.3938800207456924,
      "grad_norm": 0.7514216303825378,
      "learning_rate": 6.911692662375101e-06,
      "loss": 0.7207,
      "step": 13670
    },
    {
      "epoch": 0.3939088342073417,
      "grad_norm": 0.7933518290519714,
      "learning_rate": 6.911261495880797e-06,
      "loss": 0.7493,
      "step": 13671
    },
    {
      "epoch": 0.393937647668991,
      "grad_norm": 0.8210514783859253,
      "learning_rate": 6.910830312741317e-06,
      "loss": 0.7619,
      "step": 13672
    },
    {
      "epoch": 0.3939664611306402,
      "grad_norm": 0.7564484477043152,
      "learning_rate": 6.9103991129604135e-06,
      "loss": 0.7235,
      "step": 13673
    },
    {
      "epoch": 0.3939952745922895,
      "grad_norm": 0.7724443674087524,
      "learning_rate": 6.909967896541844e-06,
      "loss": 0.7399,
      "step": 13674
    },
    {
      "epoch": 0.3940240880539388,
      "grad_norm": 0.733241856098175,
      "learning_rate": 6.9095366634893625e-06,
      "loss": 0.7358,
      "step": 13675
    },
    {
      "epoch": 0.3940529015155881,
      "grad_norm": 0.7686418294906616,
      "learning_rate": 6.909105413806726e-06,
      "loss": 0.734,
      "step": 13676
    },
    {
      "epoch": 0.3940817149772374,
      "grad_norm": 0.746892511844635,
      "learning_rate": 6.908674147497689e-06,
      "loss": 0.746,
      "step": 13677
    },
    {
      "epoch": 0.39411052843888666,
      "grad_norm": 0.7563686966896057,
      "learning_rate": 6.908242864566009e-06,
      "loss": 0.7394,
      "step": 13678
    },
    {
      "epoch": 0.39413934190053596,
      "grad_norm": 0.7789371013641357,
      "learning_rate": 6.907811565015441e-06,
      "loss": 0.7371,
      "step": 13679
    },
    {
      "epoch": 0.3941681553621852,
      "grad_norm": 0.7676272988319397,
      "learning_rate": 6.907380248849742e-06,
      "loss": 0.7524,
      "step": 13680
    },
    {
      "epoch": 0.3941969688238345,
      "grad_norm": 0.7655232548713684,
      "learning_rate": 6.906948916072667e-06,
      "loss": 0.7603,
      "step": 13681
    },
    {
      "epoch": 0.3942257822854838,
      "grad_norm": 0.7664503455162048,
      "learning_rate": 6.906517566687974e-06,
      "loss": 0.7323,
      "step": 13682
    },
    {
      "epoch": 0.39425459574713306,
      "grad_norm": 0.7682721614837646,
      "learning_rate": 6.906086200699418e-06,
      "loss": 0.7126,
      "step": 13683
    },
    {
      "epoch": 0.39428340920878235,
      "grad_norm": 0.7676952481269836,
      "learning_rate": 6.905654818110756e-06,
      "loss": 0.7159,
      "step": 13684
    },
    {
      "epoch": 0.39431222267043164,
      "grad_norm": 0.777401864528656,
      "learning_rate": 6.905223418925747e-06,
      "loss": 0.7499,
      "step": 13685
    },
    {
      "epoch": 0.39434103613208094,
      "grad_norm": 0.7453166246414185,
      "learning_rate": 6.904792003148146e-06,
      "loss": 0.7322,
      "step": 13686
    },
    {
      "epoch": 0.39436984959373017,
      "grad_norm": 0.762321412563324,
      "learning_rate": 6.904360570781711e-06,
      "loss": 0.7338,
      "step": 13687
    },
    {
      "epoch": 0.39439866305537946,
      "grad_norm": 0.7495461702346802,
      "learning_rate": 6.903929121830198e-06,
      "loss": 0.7338,
      "step": 13688
    },
    {
      "epoch": 0.39442747651702875,
      "grad_norm": 0.7973958849906921,
      "learning_rate": 6.903497656297366e-06,
      "loss": 0.732,
      "step": 13689
    },
    {
      "epoch": 0.39445628997867804,
      "grad_norm": 0.7458484768867493,
      "learning_rate": 6.903066174186974e-06,
      "loss": 0.7073,
      "step": 13690
    },
    {
      "epoch": 0.39448510344032733,
      "grad_norm": 0.7573550343513489,
      "learning_rate": 6.902634675502776e-06,
      "loss": 0.7466,
      "step": 13691
    },
    {
      "epoch": 0.3945139169019766,
      "grad_norm": 0.7582619190216064,
      "learning_rate": 6.902203160248533e-06,
      "loss": 0.7446,
      "step": 13692
    },
    {
      "epoch": 0.39454273036362586,
      "grad_norm": 0.7596327662467957,
      "learning_rate": 6.901771628428001e-06,
      "loss": 0.7435,
      "step": 13693
    },
    {
      "epoch": 0.39457154382527515,
      "grad_norm": 0.755450963973999,
      "learning_rate": 6.901340080044939e-06,
      "loss": 0.742,
      "step": 13694
    },
    {
      "epoch": 0.39460035728692444,
      "grad_norm": 0.7921367883682251,
      "learning_rate": 6.9009085151031076e-06,
      "loss": 0.7644,
      "step": 13695
    },
    {
      "epoch": 0.39462917074857373,
      "grad_norm": 0.7733742594718933,
      "learning_rate": 6.900476933606261e-06,
      "loss": 0.724,
      "step": 13696
    },
    {
      "epoch": 0.394657984210223,
      "grad_norm": 0.7610011100769043,
      "learning_rate": 6.900045335558161e-06,
      "loss": 0.7418,
      "step": 13697
    },
    {
      "epoch": 0.3946867976718723,
      "grad_norm": 0.7649070620536804,
      "learning_rate": 6.8996137209625655e-06,
      "loss": 0.7362,
      "step": 13698
    },
    {
      "epoch": 0.3947156111335216,
      "grad_norm": 0.795117199420929,
      "learning_rate": 6.899182089823232e-06,
      "loss": 0.7256,
      "step": 13699
    },
    {
      "epoch": 0.39474442459517084,
      "grad_norm": 0.7433562278747559,
      "learning_rate": 6.898750442143922e-06,
      "loss": 0.7343,
      "step": 13700
    },
    {
      "epoch": 0.39477323805682013,
      "grad_norm": 0.733026921749115,
      "learning_rate": 6.898318777928394e-06,
      "loss": 0.7344,
      "step": 13701
    },
    {
      "epoch": 0.3948020515184694,
      "grad_norm": 0.8198468089103699,
      "learning_rate": 6.897887097180408e-06,
      "loss": 0.7485,
      "step": 13702
    },
    {
      "epoch": 0.3948308649801187,
      "grad_norm": 0.7676005959510803,
      "learning_rate": 6.897455399903722e-06,
      "loss": 0.7563,
      "step": 13703
    },
    {
      "epoch": 0.394859678441768,
      "grad_norm": 0.7490749359130859,
      "learning_rate": 6.897023686102095e-06,
      "loss": 0.7247,
      "step": 13704
    },
    {
      "epoch": 0.3948884919034173,
      "grad_norm": 0.7764137387275696,
      "learning_rate": 6.89659195577929e-06,
      "loss": 0.7383,
      "step": 13705
    },
    {
      "epoch": 0.3949173053650666,
      "grad_norm": 0.7923038601875305,
      "learning_rate": 6.896160208939064e-06,
      "loss": 0.7324,
      "step": 13706
    },
    {
      "epoch": 0.3949461188267158,
      "grad_norm": 0.7528685927391052,
      "learning_rate": 6.895728445585177e-06,
      "loss": 0.7358,
      "step": 13707
    },
    {
      "epoch": 0.3949749322883651,
      "grad_norm": 0.772514283657074,
      "learning_rate": 6.895296665721392e-06,
      "loss": 0.7272,
      "step": 13708
    },
    {
      "epoch": 0.3950037457500144,
      "grad_norm": 0.7696088552474976,
      "learning_rate": 6.894864869351466e-06,
      "loss": 0.7314,
      "step": 13709
    },
    {
      "epoch": 0.3950325592116637,
      "grad_norm": 0.7586928009986877,
      "learning_rate": 6.894433056479163e-06,
      "loss": 0.7264,
      "step": 13710
    },
    {
      "epoch": 0.395061372673313,
      "grad_norm": 0.8243218064308167,
      "learning_rate": 6.8940012271082405e-06,
      "loss": 0.7218,
      "step": 13711
    },
    {
      "epoch": 0.3950901861349623,
      "grad_norm": 0.8840034604072571,
      "learning_rate": 6.893569381242461e-06,
      "loss": 0.7242,
      "step": 13712
    },
    {
      "epoch": 0.39511899959661156,
      "grad_norm": 0.7497971057891846,
      "learning_rate": 6.893137518885584e-06,
      "loss": 0.7473,
      "step": 13713
    },
    {
      "epoch": 0.3951478130582608,
      "grad_norm": 0.7587982416152954,
      "learning_rate": 6.8927056400413735e-06,
      "loss": 0.731,
      "step": 13714
    },
    {
      "epoch": 0.3951766265199101,
      "grad_norm": 0.7761501669883728,
      "learning_rate": 6.892273744713587e-06,
      "loss": 0.7353,
      "step": 13715
    },
    {
      "epoch": 0.3952054399815594,
      "grad_norm": 0.744129478931427,
      "learning_rate": 6.891841832905989e-06,
      "loss": 0.7423,
      "step": 13716
    },
    {
      "epoch": 0.39523425344320867,
      "grad_norm": 0.8008683919906616,
      "learning_rate": 6.891409904622341e-06,
      "loss": 0.7422,
      "step": 13717
    },
    {
      "epoch": 0.39526306690485796,
      "grad_norm": 0.775619387626648,
      "learning_rate": 6.8909779598664e-06,
      "loss": 0.7457,
      "step": 13718
    },
    {
      "epoch": 0.39529188036650725,
      "grad_norm": 0.7783471345901489,
      "learning_rate": 6.890545998641933e-06,
      "loss": 0.764,
      "step": 13719
    },
    {
      "epoch": 0.3953206938281565,
      "grad_norm": 0.788543164730072,
      "learning_rate": 6.890114020952702e-06,
      "loss": 0.7149,
      "step": 13720
    },
    {
      "epoch": 0.3953495072898058,
      "grad_norm": 0.7741332650184631,
      "learning_rate": 6.8896820268024656e-06,
      "loss": 0.7384,
      "step": 13721
    },
    {
      "epoch": 0.39537832075145507,
      "grad_norm": 0.7880368828773499,
      "learning_rate": 6.889250016194988e-06,
      "loss": 0.7249,
      "step": 13722
    },
    {
      "epoch": 0.39540713421310436,
      "grad_norm": 0.7593222260475159,
      "learning_rate": 6.88881798913403e-06,
      "loss": 0.7374,
      "step": 13723
    },
    {
      "epoch": 0.39543594767475365,
      "grad_norm": 0.7351554036140442,
      "learning_rate": 6.888385945623357e-06,
      "loss": 0.7195,
      "step": 13724
    },
    {
      "epoch": 0.39546476113640294,
      "grad_norm": 0.749724268913269,
      "learning_rate": 6.88795388566673e-06,
      "loss": 0.7234,
      "step": 13725
    },
    {
      "epoch": 0.39549357459805223,
      "grad_norm": 0.7988089323043823,
      "learning_rate": 6.887521809267912e-06,
      "loss": 0.7469,
      "step": 13726
    },
    {
      "epoch": 0.39552238805970147,
      "grad_norm": 0.7748697996139526,
      "learning_rate": 6.887089716430666e-06,
      "loss": 0.7591,
      "step": 13727
    },
    {
      "epoch": 0.39555120152135076,
      "grad_norm": 0.7622685432434082,
      "learning_rate": 6.886657607158754e-06,
      "loss": 0.7419,
      "step": 13728
    },
    {
      "epoch": 0.39558001498300005,
      "grad_norm": 0.7462995052337646,
      "learning_rate": 6.886225481455941e-06,
      "loss": 0.7392,
      "step": 13729
    },
    {
      "epoch": 0.39560882844464934,
      "grad_norm": 0.7873317003250122,
      "learning_rate": 6.885793339325989e-06,
      "loss": 0.7221,
      "step": 13730
    },
    {
      "epoch": 0.39563764190629863,
      "grad_norm": 0.7992826104164124,
      "learning_rate": 6.885361180772663e-06,
      "loss": 0.7108,
      "step": 13731
    },
    {
      "epoch": 0.3956664553679479,
      "grad_norm": 0.8144090175628662,
      "learning_rate": 6.884929005799725e-06,
      "loss": 0.7265,
      "step": 13732
    },
    {
      "epoch": 0.3956952688295972,
      "grad_norm": 0.7509315609931946,
      "learning_rate": 6.88449681441094e-06,
      "loss": 0.7269,
      "step": 13733
    },
    {
      "epoch": 0.39572408229124645,
      "grad_norm": 0.764607310295105,
      "learning_rate": 6.884064606610072e-06,
      "loss": 0.7461,
      "step": 13734
    },
    {
      "epoch": 0.39575289575289574,
      "grad_norm": 0.7514469027519226,
      "learning_rate": 6.883632382400883e-06,
      "loss": 0.7058,
      "step": 13735
    },
    {
      "epoch": 0.39578170921454503,
      "grad_norm": 0.7897694110870361,
      "learning_rate": 6.883200141787141e-06,
      "loss": 0.7358,
      "step": 13736
    },
    {
      "epoch": 0.3958105226761943,
      "grad_norm": 0.7715076208114624,
      "learning_rate": 6.882767884772605e-06,
      "loss": 0.7481,
      "step": 13737
    },
    {
      "epoch": 0.3958393361378436,
      "grad_norm": 0.7599079608917236,
      "learning_rate": 6.882335611361046e-06,
      "loss": 0.7535,
      "step": 13738
    },
    {
      "epoch": 0.3958681495994929,
      "grad_norm": 0.7422821521759033,
      "learning_rate": 6.881903321556225e-06,
      "loss": 0.7347,
      "step": 13739
    },
    {
      "epoch": 0.3958969630611422,
      "grad_norm": 0.7877892255783081,
      "learning_rate": 6.881471015361907e-06,
      "loss": 0.7425,
      "step": 13740
    },
    {
      "epoch": 0.39592577652279143,
      "grad_norm": 0.7748306393623352,
      "learning_rate": 6.881038692781856e-06,
      "loss": 0.7296,
      "step": 13741
    },
    {
      "epoch": 0.3959545899844407,
      "grad_norm": 0.7675625085830688,
      "learning_rate": 6.88060635381984e-06,
      "loss": 0.7316,
      "step": 13742
    },
    {
      "epoch": 0.39598340344609,
      "grad_norm": 0.7634471654891968,
      "learning_rate": 6.880173998479622e-06,
      "loss": 0.7452,
      "step": 13743
    },
    {
      "epoch": 0.3960122169077393,
      "grad_norm": 0.7722693085670471,
      "learning_rate": 6.879741626764968e-06,
      "loss": 0.7318,
      "step": 13744
    },
    {
      "epoch": 0.3960410303693886,
      "grad_norm": 0.773980438709259,
      "learning_rate": 6.879309238679643e-06,
      "loss": 0.7404,
      "step": 13745
    },
    {
      "epoch": 0.3960698438310379,
      "grad_norm": 0.9131379127502441,
      "learning_rate": 6.878876834227414e-06,
      "loss": 0.7254,
      "step": 13746
    },
    {
      "epoch": 0.3960986572926871,
      "grad_norm": 0.7728403806686401,
      "learning_rate": 6.878444413412044e-06,
      "loss": 0.7313,
      "step": 13747
    },
    {
      "epoch": 0.3961274707543364,
      "grad_norm": 0.7667877674102783,
      "learning_rate": 6.878011976237303e-06,
      "loss": 0.7286,
      "step": 13748
    },
    {
      "epoch": 0.3961562842159857,
      "grad_norm": 0.7616279125213623,
      "learning_rate": 6.877579522706955e-06,
      "loss": 0.7321,
      "step": 13749
    },
    {
      "epoch": 0.396185097677635,
      "grad_norm": 0.8259749412536621,
      "learning_rate": 6.877147052824767e-06,
      "loss": 0.7525,
      "step": 13750
    },
    {
      "epoch": 0.3962139111392843,
      "grad_norm": 0.7854470610618591,
      "learning_rate": 6.876714566594502e-06,
      "loss": 0.7289,
      "step": 13751
    },
    {
      "epoch": 0.39624272460093357,
      "grad_norm": 0.7466049790382385,
      "learning_rate": 6.876282064019931e-06,
      "loss": 0.7218,
      "step": 13752
    },
    {
      "epoch": 0.39627153806258286,
      "grad_norm": 0.7603347897529602,
      "learning_rate": 6.875849545104818e-06,
      "loss": 0.7613,
      "step": 13753
    },
    {
      "epoch": 0.3963003515242321,
      "grad_norm": 0.7719988822937012,
      "learning_rate": 6.8754170098529304e-06,
      "loss": 0.7459,
      "step": 13754
    },
    {
      "epoch": 0.3963291649858814,
      "grad_norm": 0.7624863386154175,
      "learning_rate": 6.874984458268035e-06,
      "loss": 0.7341,
      "step": 13755
    },
    {
      "epoch": 0.3963579784475307,
      "grad_norm": 0.7582473754882812,
      "learning_rate": 6.874551890353901e-06,
      "loss": 0.7505,
      "step": 13756
    },
    {
      "epoch": 0.39638679190917997,
      "grad_norm": 0.7555980682373047,
      "learning_rate": 6.874119306114293e-06,
      "loss": 0.7596,
      "step": 13757
    },
    {
      "epoch": 0.39641560537082926,
      "grad_norm": 0.7357990741729736,
      "learning_rate": 6.87368670555298e-06,
      "loss": 0.7328,
      "step": 13758
    },
    {
      "epoch": 0.39644441883247855,
      "grad_norm": 0.7684670090675354,
      "learning_rate": 6.873254088673727e-06,
      "loss": 0.7175,
      "step": 13759
    },
    {
      "epoch": 0.39647323229412784,
      "grad_norm": 0.7616785168647766,
      "learning_rate": 6.872821455480305e-06,
      "loss": 0.7423,
      "step": 13760
    },
    {
      "epoch": 0.3965020457557771,
      "grad_norm": 0.755226731300354,
      "learning_rate": 6.8723888059764795e-06,
      "loss": 0.7526,
      "step": 13761
    },
    {
      "epoch": 0.39653085921742637,
      "grad_norm": 0.7481119632720947,
      "learning_rate": 6.8719561401660195e-06,
      "loss": 0.7322,
      "step": 13762
    },
    {
      "epoch": 0.39655967267907566,
      "grad_norm": 0.7457230687141418,
      "learning_rate": 6.8715234580526934e-06,
      "loss": 0.7167,
      "step": 13763
    },
    {
      "epoch": 0.39658848614072495,
      "grad_norm": 0.7722078561782837,
      "learning_rate": 6.871090759640269e-06,
      "loss": 0.7392,
      "step": 13764
    },
    {
      "epoch": 0.39661729960237424,
      "grad_norm": 0.7637680768966675,
      "learning_rate": 6.870658044932512e-06,
      "loss": 0.7286,
      "step": 13765
    },
    {
      "epoch": 0.39664611306402353,
      "grad_norm": 0.7962849140167236,
      "learning_rate": 6.870225313933195e-06,
      "loss": 0.7419,
      "step": 13766
    },
    {
      "epoch": 0.3966749265256728,
      "grad_norm": 0.7548282742500305,
      "learning_rate": 6.8697925666460864e-06,
      "loss": 0.721,
      "step": 13767
    },
    {
      "epoch": 0.39670373998732206,
      "grad_norm": 0.788074254989624,
      "learning_rate": 6.869359803074952e-06,
      "loss": 0.7195,
      "step": 13768
    },
    {
      "epoch": 0.39673255344897135,
      "grad_norm": 0.735577404499054,
      "learning_rate": 6.868927023223564e-06,
      "loss": 0.7242,
      "step": 13769
    },
    {
      "epoch": 0.39676136691062064,
      "grad_norm": 0.757122814655304,
      "learning_rate": 6.8684942270956875e-06,
      "loss": 0.7146,
      "step": 13770
    },
    {
      "epoch": 0.39679018037226993,
      "grad_norm": 0.7731769680976868,
      "learning_rate": 6.868061414695095e-06,
      "loss": 0.7239,
      "step": 13771
    },
    {
      "epoch": 0.3968189938339192,
      "grad_norm": 0.7691024541854858,
      "learning_rate": 6.867628586025555e-06,
      "loss": 0.7064,
      "step": 13772
    },
    {
      "epoch": 0.3968478072955685,
      "grad_norm": 0.7826371192932129,
      "learning_rate": 6.867195741090838e-06,
      "loss": 0.7301,
      "step": 13773
    },
    {
      "epoch": 0.39687662075721775,
      "grad_norm": 0.7683568000793457,
      "learning_rate": 6.8667628798947105e-06,
      "loss": 0.7694,
      "step": 13774
    },
    {
      "epoch": 0.39690543421886704,
      "grad_norm": 0.8329829573631287,
      "learning_rate": 6.866330002440946e-06,
      "loss": 0.7511,
      "step": 13775
    },
    {
      "epoch": 0.39693424768051633,
      "grad_norm": 0.7905882596969604,
      "learning_rate": 6.8658971087333136e-06,
      "loss": 0.7514,
      "step": 13776
    },
    {
      "epoch": 0.3969630611421656,
      "grad_norm": 0.7617536187171936,
      "learning_rate": 6.86546419877558e-06,
      "loss": 0.7149,
      "step": 13777
    },
    {
      "epoch": 0.3969918746038149,
      "grad_norm": 0.7540045380592346,
      "learning_rate": 6.8650312725715205e-06,
      "loss": 0.7073,
      "step": 13778
    },
    {
      "epoch": 0.3970206880654642,
      "grad_norm": 0.7511802911758423,
      "learning_rate": 6.864598330124902e-06,
      "loss": 0.7216,
      "step": 13779
    },
    {
      "epoch": 0.3970495015271135,
      "grad_norm": 0.7885406613349915,
      "learning_rate": 6.864165371439496e-06,
      "loss": 0.7226,
      "step": 13780
    },
    {
      "epoch": 0.3970783149887627,
      "grad_norm": 0.7607347965240479,
      "learning_rate": 6.863732396519074e-06,
      "loss": 0.7316,
      "step": 13781
    },
    {
      "epoch": 0.397107128450412,
      "grad_norm": 0.7553650736808777,
      "learning_rate": 6.863299405367405e-06,
      "loss": 0.7536,
      "step": 13782
    },
    {
      "epoch": 0.3971359419120613,
      "grad_norm": 0.7571521401405334,
      "learning_rate": 6.8628663979882595e-06,
      "loss": 0.7485,
      "step": 13783
    },
    {
      "epoch": 0.3971647553737106,
      "grad_norm": 0.7589356899261475,
      "learning_rate": 6.862433374385411e-06,
      "loss": 0.7228,
      "step": 13784
    },
    {
      "epoch": 0.3971935688353599,
      "grad_norm": 0.7535999417304993,
      "learning_rate": 6.862000334562631e-06,
      "loss": 0.7048,
      "step": 13785
    },
    {
      "epoch": 0.3972223822970092,
      "grad_norm": 0.7631509304046631,
      "learning_rate": 6.861567278523687e-06,
      "loss": 0.7384,
      "step": 13786
    },
    {
      "epoch": 0.39725119575865847,
      "grad_norm": 0.7836354374885559,
      "learning_rate": 6.861134206272355e-06,
      "loss": 0.7584,
      "step": 13787
    },
    {
      "epoch": 0.3972800092203077,
      "grad_norm": 0.758626401424408,
      "learning_rate": 6.860701117812403e-06,
      "loss": 0.7503,
      "step": 13788
    },
    {
      "epoch": 0.397308822681957,
      "grad_norm": 0.8155133128166199,
      "learning_rate": 6.860268013147604e-06,
      "loss": 0.7162,
      "step": 13789
    },
    {
      "epoch": 0.3973376361436063,
      "grad_norm": 0.737069845199585,
      "learning_rate": 6.8598348922817315e-06,
      "loss": 0.7236,
      "step": 13790
    },
    {
      "epoch": 0.3973664496052556,
      "grad_norm": 0.7453553676605225,
      "learning_rate": 6.859401755218555e-06,
      "loss": 0.7389,
      "step": 13791
    },
    {
      "epoch": 0.39739526306690487,
      "grad_norm": 0.7327378988265991,
      "learning_rate": 6.85896860196185e-06,
      "loss": 0.6945,
      "step": 13792
    },
    {
      "epoch": 0.39742407652855416,
      "grad_norm": 0.7656872868537903,
      "learning_rate": 6.858535432515384e-06,
      "loss": 0.7477,
      "step": 13793
    },
    {
      "epoch": 0.39745288999020345,
      "grad_norm": 0.7616003155708313,
      "learning_rate": 6.858102246882934e-06,
      "loss": 0.7332,
      "step": 13794
    },
    {
      "epoch": 0.3974817034518527,
      "grad_norm": 0.7280584573745728,
      "learning_rate": 6.8576690450682715e-06,
      "loss": 0.7272,
      "step": 13795
    },
    {
      "epoch": 0.397510516913502,
      "grad_norm": 0.7555036544799805,
      "learning_rate": 6.857235827075166e-06,
      "loss": 0.7357,
      "step": 13796
    },
    {
      "epoch": 0.39753933037515127,
      "grad_norm": 0.7381149530410767,
      "learning_rate": 6.8568025929073944e-06,
      "loss": 0.7474,
      "step": 13797
    },
    {
      "epoch": 0.39756814383680056,
      "grad_norm": 0.7617025375366211,
      "learning_rate": 6.856369342568729e-06,
      "loss": 0.7319,
      "step": 13798
    },
    {
      "epoch": 0.39759695729844985,
      "grad_norm": 0.7633859515190125,
      "learning_rate": 6.855936076062942e-06,
      "loss": 0.7386,
      "step": 13799
    },
    {
      "epoch": 0.39762577076009914,
      "grad_norm": 0.7378416061401367,
      "learning_rate": 6.855502793393808e-06,
      "loss": 0.7313,
      "step": 13800
    },
    {
      "epoch": 0.3976545842217484,
      "grad_norm": 0.7263152599334717,
      "learning_rate": 6.855069494565097e-06,
      "loss": 0.741,
      "step": 13801
    },
    {
      "epoch": 0.39768339768339767,
      "grad_norm": 0.7446420192718506,
      "learning_rate": 6.854636179580587e-06,
      "loss": 0.7225,
      "step": 13802
    },
    {
      "epoch": 0.39771221114504696,
      "grad_norm": 0.7451476454734802,
      "learning_rate": 6.854202848444049e-06,
      "loss": 0.7051,
      "step": 13803
    },
    {
      "epoch": 0.39774102460669625,
      "grad_norm": 0.7394868731498718,
      "learning_rate": 6.853769501159259e-06,
      "loss": 0.7456,
      "step": 13804
    },
    {
      "epoch": 0.39776983806834554,
      "grad_norm": 0.7398494482040405,
      "learning_rate": 6.853336137729989e-06,
      "loss": 0.7422,
      "step": 13805
    },
    {
      "epoch": 0.39779865152999483,
      "grad_norm": 0.7583437561988831,
      "learning_rate": 6.852902758160014e-06,
      "loss": 0.7428,
      "step": 13806
    },
    {
      "epoch": 0.3978274649916441,
      "grad_norm": 0.747229814529419,
      "learning_rate": 6.852469362453107e-06,
      "loss": 0.7302,
      "step": 13807
    },
    {
      "epoch": 0.39785627845329335,
      "grad_norm": 0.731737494468689,
      "learning_rate": 6.852035950613044e-06,
      "loss": 0.7124,
      "step": 13808
    },
    {
      "epoch": 0.39788509191494265,
      "grad_norm": 0.738952100276947,
      "learning_rate": 6.8516025226436e-06,
      "loss": 0.7207,
      "step": 13809
    },
    {
      "epoch": 0.39791390537659194,
      "grad_norm": 0.7764305472373962,
      "learning_rate": 6.85116907854855e-06,
      "loss": 0.7132,
      "step": 13810
    },
    {
      "epoch": 0.3979427188382412,
      "grad_norm": 0.7622454762458801,
      "learning_rate": 6.850735618331666e-06,
      "loss": 0.7349,
      "step": 13811
    },
    {
      "epoch": 0.3979715322998905,
      "grad_norm": 0.769554615020752,
      "learning_rate": 6.850302141996724e-06,
      "loss": 0.7414,
      "step": 13812
    },
    {
      "epoch": 0.3980003457615398,
      "grad_norm": 0.7618257403373718,
      "learning_rate": 6.849868649547502e-06,
      "loss": 0.7178,
      "step": 13813
    },
    {
      "epoch": 0.3980291592231891,
      "grad_norm": 0.7948417663574219,
      "learning_rate": 6.8494351409877726e-06,
      "loss": 0.7412,
      "step": 13814
    },
    {
      "epoch": 0.39805797268483833,
      "grad_norm": 0.7449653744697571,
      "learning_rate": 6.849001616321312e-06,
      "loss": 0.7168,
      "step": 13815
    },
    {
      "epoch": 0.3980867861464876,
      "grad_norm": 0.7446622252464294,
      "learning_rate": 6.848568075551895e-06,
      "loss": 0.7525,
      "step": 13816
    },
    {
      "epoch": 0.3981155996081369,
      "grad_norm": 0.7283443808555603,
      "learning_rate": 6.848134518683298e-06,
      "loss": 0.7163,
      "step": 13817
    },
    {
      "epoch": 0.3981444130697862,
      "grad_norm": 0.8456035852432251,
      "learning_rate": 6.847700945719298e-06,
      "loss": 0.7495,
      "step": 13818
    },
    {
      "epoch": 0.3981732265314355,
      "grad_norm": 0.7481231093406677,
      "learning_rate": 6.8472673566636675e-06,
      "loss": 0.7338,
      "step": 13819
    },
    {
      "epoch": 0.3982020399930848,
      "grad_norm": 0.7583329677581787,
      "learning_rate": 6.846833751520186e-06,
      "loss": 0.7477,
      "step": 13820
    },
    {
      "epoch": 0.3982308534547341,
      "grad_norm": 0.752494215965271,
      "learning_rate": 6.84640013029263e-06,
      "loss": 0.7231,
      "step": 13821
    },
    {
      "epoch": 0.3982596669163833,
      "grad_norm": 0.7736232280731201,
      "learning_rate": 6.845966492984775e-06,
      "loss": 0.7336,
      "step": 13822
    },
    {
      "epoch": 0.3982884803780326,
      "grad_norm": 0.7735109925270081,
      "learning_rate": 6.845532839600395e-06,
      "loss": 0.7103,
      "step": 13823
    },
    {
      "epoch": 0.3983172938396819,
      "grad_norm": 0.7495059967041016,
      "learning_rate": 6.84509917014327e-06,
      "loss": 0.7479,
      "step": 13824
    },
    {
      "epoch": 0.3983461073013312,
      "grad_norm": 0.7569525837898254,
      "learning_rate": 6.844665484617175e-06,
      "loss": 0.743,
      "step": 13825
    },
    {
      "epoch": 0.3983749207629805,
      "grad_norm": 0.732342004776001,
      "learning_rate": 6.844231783025887e-06,
      "loss": 0.7314,
      "step": 13826
    },
    {
      "epoch": 0.39840373422462977,
      "grad_norm": 0.7594707608222961,
      "learning_rate": 6.843798065373186e-06,
      "loss": 0.717,
      "step": 13827
    },
    {
      "epoch": 0.398432547686279,
      "grad_norm": 0.7551814913749695,
      "learning_rate": 6.843364331662845e-06,
      "loss": 0.7232,
      "step": 13828
    },
    {
      "epoch": 0.3984613611479283,
      "grad_norm": 0.7594829797744751,
      "learning_rate": 6.842930581898644e-06,
      "loss": 0.7335,
      "step": 13829
    },
    {
      "epoch": 0.3984901746095776,
      "grad_norm": 0.739265501499176,
      "learning_rate": 6.842496816084361e-06,
      "loss": 0.7229,
      "step": 13830
    },
    {
      "epoch": 0.3985189880712269,
      "grad_norm": 0.7644234895706177,
      "learning_rate": 6.8420630342237705e-06,
      "loss": 0.7357,
      "step": 13831
    },
    {
      "epoch": 0.39854780153287617,
      "grad_norm": 0.762273907661438,
      "learning_rate": 6.841629236320654e-06,
      "loss": 0.7328,
      "step": 13832
    },
    {
      "epoch": 0.39857661499452546,
      "grad_norm": 0.7468553185462952,
      "learning_rate": 6.841195422378787e-06,
      "loss": 0.7513,
      "step": 13833
    },
    {
      "epoch": 0.39860542845617475,
      "grad_norm": 0.7757792472839355,
      "learning_rate": 6.840761592401948e-06,
      "loss": 0.736,
      "step": 13834
    },
    {
      "epoch": 0.398634241917824,
      "grad_norm": 0.734500527381897,
      "learning_rate": 6.840327746393915e-06,
      "loss": 0.7225,
      "step": 13835
    },
    {
      "epoch": 0.3986630553794733,
      "grad_norm": 0.7672228813171387,
      "learning_rate": 6.839893884358469e-06,
      "loss": 0.7404,
      "step": 13836
    },
    {
      "epoch": 0.39869186884112257,
      "grad_norm": 0.7543317675590515,
      "learning_rate": 6.839460006299386e-06,
      "loss": 0.7455,
      "step": 13837
    },
    {
      "epoch": 0.39872068230277186,
      "grad_norm": 0.7377316355705261,
      "learning_rate": 6.839026112220444e-06,
      "loss": 0.7366,
      "step": 13838
    },
    {
      "epoch": 0.39874949576442115,
      "grad_norm": 0.7749660611152649,
      "learning_rate": 6.838592202125424e-06,
      "loss": 0.7425,
      "step": 13839
    },
    {
      "epoch": 0.39877830922607044,
      "grad_norm": 0.7499288320541382,
      "learning_rate": 6.8381582760181035e-06,
      "loss": 0.7216,
      "step": 13840
    },
    {
      "epoch": 0.39880712268771973,
      "grad_norm": 0.7610665559768677,
      "learning_rate": 6.837724333902263e-06,
      "loss": 0.7264,
      "step": 13841
    },
    {
      "epoch": 0.39883593614936896,
      "grad_norm": 0.7353660464286804,
      "learning_rate": 6.8372903757816785e-06,
      "loss": 0.7585,
      "step": 13842
    },
    {
      "epoch": 0.39886474961101825,
      "grad_norm": 0.7467086911201477,
      "learning_rate": 6.836856401660132e-06,
      "loss": 0.7685,
      "step": 13843
    },
    {
      "epoch": 0.39889356307266755,
      "grad_norm": 0.7591050267219543,
      "learning_rate": 6.836422411541403e-06,
      "loss": 0.7144,
      "step": 13844
    },
    {
      "epoch": 0.39892237653431684,
      "grad_norm": 0.7489648461341858,
      "learning_rate": 6.83598840542927e-06,
      "loss": 0.7378,
      "step": 13845
    },
    {
      "epoch": 0.3989511899959661,
      "grad_norm": 0.7724493741989136,
      "learning_rate": 6.835554383327514e-06,
      "loss": 0.7146,
      "step": 13846
    },
    {
      "epoch": 0.3989800034576154,
      "grad_norm": 0.7628548741340637,
      "learning_rate": 6.835120345239914e-06,
      "loss": 0.7304,
      "step": 13847
    },
    {
      "epoch": 0.39900881691926465,
      "grad_norm": 0.7400162220001221,
      "learning_rate": 6.8346862911702506e-06,
      "loss": 0.7382,
      "step": 13848
    },
    {
      "epoch": 0.39903763038091394,
      "grad_norm": 0.7721336483955383,
      "learning_rate": 6.834252221122302e-06,
      "loss": 0.7434,
      "step": 13849
    },
    {
      "epoch": 0.39906644384256323,
      "grad_norm": 0.7514805793762207,
      "learning_rate": 6.833818135099852e-06,
      "loss": 0.7472,
      "step": 13850
    },
    {
      "epoch": 0.3990952573042125,
      "grad_norm": 0.7269894480705261,
      "learning_rate": 6.833384033106679e-06,
      "loss": 0.7201,
      "step": 13851
    },
    {
      "epoch": 0.3991240707658618,
      "grad_norm": 0.7267375588417053,
      "learning_rate": 6.832949915146563e-06,
      "loss": 0.7221,
      "step": 13852
    },
    {
      "epoch": 0.3991528842275111,
      "grad_norm": 0.7561330795288086,
      "learning_rate": 6.832515781223285e-06,
      "loss": 0.7333,
      "step": 13853
    },
    {
      "epoch": 0.3991816976891604,
      "grad_norm": 0.755845844745636,
      "learning_rate": 6.832081631340628e-06,
      "loss": 0.7236,
      "step": 13854
    },
    {
      "epoch": 0.39921051115080963,
      "grad_norm": 0.7388278841972351,
      "learning_rate": 6.831647465502369e-06,
      "loss": 0.7234,
      "step": 13855
    },
    {
      "epoch": 0.3992393246124589,
      "grad_norm": 0.750384509563446,
      "learning_rate": 6.831213283712294e-06,
      "loss": 0.7303,
      "step": 13856
    },
    {
      "epoch": 0.3992681380741082,
      "grad_norm": 0.7591168880462646,
      "learning_rate": 6.83077908597418e-06,
      "loss": 0.752,
      "step": 13857
    },
    {
      "epoch": 0.3992969515357575,
      "grad_norm": 0.7458991408348083,
      "learning_rate": 6.830344872291812e-06,
      "loss": 0.707,
      "step": 13858
    },
    {
      "epoch": 0.3993257649974068,
      "grad_norm": 0.7351931929588318,
      "learning_rate": 6.8299106426689696e-06,
      "loss": 0.7235,
      "step": 13859
    },
    {
      "epoch": 0.3993545784590561,
      "grad_norm": 0.73829585313797,
      "learning_rate": 6.829476397109434e-06,
      "loss": 0.7211,
      "step": 13860
    },
    {
      "epoch": 0.3993833919207054,
      "grad_norm": 0.743499219417572,
      "learning_rate": 6.8290421356169875e-06,
      "loss": 0.7597,
      "step": 13861
    },
    {
      "epoch": 0.3994122053823546,
      "grad_norm": 0.7863559126853943,
      "learning_rate": 6.828607858195413e-06,
      "loss": 0.7417,
      "step": 13862
    },
    {
      "epoch": 0.3994410188440039,
      "grad_norm": 0.7476334571838379,
      "learning_rate": 6.828173564848493e-06,
      "loss": 0.7319,
      "step": 13863
    },
    {
      "epoch": 0.3994698323056532,
      "grad_norm": 0.720197319984436,
      "learning_rate": 6.827739255580007e-06,
      "loss": 0.7311,
      "step": 13864
    },
    {
      "epoch": 0.3994986457673025,
      "grad_norm": 0.7655223608016968,
      "learning_rate": 6.827304930393741e-06,
      "loss": 0.7126,
      "step": 13865
    },
    {
      "epoch": 0.3995274592289518,
      "grad_norm": 0.7461180686950684,
      "learning_rate": 6.826870589293474e-06,
      "loss": 0.7262,
      "step": 13866
    },
    {
      "epoch": 0.39955627269060107,
      "grad_norm": 0.7592753767967224,
      "learning_rate": 6.82643623228299e-06,
      "loss": 0.7328,
      "step": 13867
    },
    {
      "epoch": 0.39958508615225036,
      "grad_norm": 0.7329624891281128,
      "learning_rate": 6.826001859366074e-06,
      "loss": 0.7288,
      "step": 13868
    },
    {
      "epoch": 0.3996138996138996,
      "grad_norm": 0.7555491924285889,
      "learning_rate": 6.825567470546506e-06,
      "loss": 0.7162,
      "step": 13869
    },
    {
      "epoch": 0.3996427130755489,
      "grad_norm": 0.7861951589584351,
      "learning_rate": 6.8251330658280705e-06,
      "loss": 0.7725,
      "step": 13870
    },
    {
      "epoch": 0.3996715265371982,
      "grad_norm": 0.7598845958709717,
      "learning_rate": 6.82469864521455e-06,
      "loss": 0.7209,
      "step": 13871
    },
    {
      "epoch": 0.39970033999884746,
      "grad_norm": 0.7441072463989258,
      "learning_rate": 6.824264208709729e-06,
      "loss": 0.7375,
      "step": 13872
    },
    {
      "epoch": 0.39972915346049676,
      "grad_norm": 0.7594054341316223,
      "learning_rate": 6.82382975631739e-06,
      "loss": 0.7403,
      "step": 13873
    },
    {
      "epoch": 0.39975796692214605,
      "grad_norm": 0.7401418089866638,
      "learning_rate": 6.823395288041315e-06,
      "loss": 0.7229,
      "step": 13874
    },
    {
      "epoch": 0.3997867803837953,
      "grad_norm": 0.7376258373260498,
      "learning_rate": 6.8229608038852925e-06,
      "loss": 0.7355,
      "step": 13875
    },
    {
      "epoch": 0.3998155938454446,
      "grad_norm": 0.7681481838226318,
      "learning_rate": 6.822526303853103e-06,
      "loss": 0.7314,
      "step": 13876
    },
    {
      "epoch": 0.39984440730709386,
      "grad_norm": 0.7496481537818909,
      "learning_rate": 6.822091787948532e-06,
      "loss": 0.7271,
      "step": 13877
    },
    {
      "epoch": 0.39987322076874315,
      "grad_norm": 0.7409397959709167,
      "learning_rate": 6.821657256175362e-06,
      "loss": 0.7308,
      "step": 13878
    },
    {
      "epoch": 0.39990203423039244,
      "grad_norm": 0.7351797819137573,
      "learning_rate": 6.821222708537379e-06,
      "loss": 0.7249,
      "step": 13879
    },
    {
      "epoch": 0.39993084769204174,
      "grad_norm": 0.7823394536972046,
      "learning_rate": 6.820788145038366e-06,
      "loss": 0.7427,
      "step": 13880
    },
    {
      "epoch": 0.399959661153691,
      "grad_norm": 0.7640260457992554,
      "learning_rate": 6.820353565682109e-06,
      "loss": 0.709,
      "step": 13881
    },
    {
      "epoch": 0.39998847461534026,
      "grad_norm": 1.028047800064087,
      "learning_rate": 6.819918970472391e-06,
      "loss": 0.7273,
      "step": 13882
    },
    {
      "epoch": 0.40001728807698955,
      "grad_norm": 0.7623040080070496,
      "learning_rate": 6.819484359413e-06,
      "loss": 0.7508,
      "step": 13883
    },
    {
      "epoch": 0.40004610153863884,
      "grad_norm": 0.7710664868354797,
      "learning_rate": 6.819049732507717e-06,
      "loss": 0.7259,
      "step": 13884
    },
    {
      "epoch": 0.40007491500028813,
      "grad_norm": 0.7448649406433105,
      "learning_rate": 6.81861508976033e-06,
      "loss": 0.7457,
      "step": 13885
    },
    {
      "epoch": 0.4001037284619374,
      "grad_norm": 0.7724429965019226,
      "learning_rate": 6.8181804311746235e-06,
      "loss": 0.7442,
      "step": 13886
    },
    {
      "epoch": 0.4001325419235867,
      "grad_norm": 0.7308474183082581,
      "learning_rate": 6.817745756754384e-06,
      "loss": 0.7101,
      "step": 13887
    },
    {
      "epoch": 0.400161355385236,
      "grad_norm": 0.7304876446723938,
      "learning_rate": 6.817311066503395e-06,
      "loss": 0.7167,
      "step": 13888
    },
    {
      "epoch": 0.40019016884688524,
      "grad_norm": 0.7566004395484924,
      "learning_rate": 6.816876360425443e-06,
      "loss": 0.7301,
      "step": 13889
    },
    {
      "epoch": 0.40021898230853453,
      "grad_norm": 0.7309519052505493,
      "learning_rate": 6.816441638524315e-06,
      "loss": 0.7278,
      "step": 13890
    },
    {
      "epoch": 0.4002477957701838,
      "grad_norm": 0.7901450395584106,
      "learning_rate": 6.816006900803794e-06,
      "loss": 0.7706,
      "step": 13891
    },
    {
      "epoch": 0.4002766092318331,
      "grad_norm": 0.7358590960502625,
      "learning_rate": 6.81557214726767e-06,
      "loss": 0.7129,
      "step": 13892
    },
    {
      "epoch": 0.4003054226934824,
      "grad_norm": 0.7774516940116882,
      "learning_rate": 6.815137377919728e-06,
      "loss": 0.7481,
      "step": 13893
    },
    {
      "epoch": 0.4003342361551317,
      "grad_norm": 0.7558368444442749,
      "learning_rate": 6.814702592763752e-06,
      "loss": 0.7315,
      "step": 13894
    },
    {
      "epoch": 0.400363049616781,
      "grad_norm": 0.7879504561424255,
      "learning_rate": 6.814267791803533e-06,
      "loss": 0.7387,
      "step": 13895
    },
    {
      "epoch": 0.4003918630784302,
      "grad_norm": 0.7288538813591003,
      "learning_rate": 6.813832975042854e-06,
      "loss": 0.7439,
      "step": 13896
    },
    {
      "epoch": 0.4004206765400795,
      "grad_norm": 0.8352261185646057,
      "learning_rate": 6.813398142485501e-06,
      "loss": 0.7116,
      "step": 13897
    },
    {
      "epoch": 0.4004494900017288,
      "grad_norm": 0.7624921202659607,
      "learning_rate": 6.812963294135265e-06,
      "loss": 0.7385,
      "step": 13898
    },
    {
      "epoch": 0.4004783034633781,
      "grad_norm": 1.2149969339370728,
      "learning_rate": 6.8125284299959304e-06,
      "loss": 0.7397,
      "step": 13899
    },
    {
      "epoch": 0.4005071169250274,
      "grad_norm": 0.7557981014251709,
      "learning_rate": 6.8120935500712855e-06,
      "loss": 0.7334,
      "step": 13900
    },
    {
      "epoch": 0.4005359303866767,
      "grad_norm": 0.7277815937995911,
      "learning_rate": 6.8116586543651175e-06,
      "loss": 0.7196,
      "step": 13901
    },
    {
      "epoch": 0.4005647438483259,
      "grad_norm": 0.770733654499054,
      "learning_rate": 6.811223742881213e-06,
      "loss": 0.7212,
      "step": 13902
    },
    {
      "epoch": 0.4005935573099752,
      "grad_norm": 0.76931232213974,
      "learning_rate": 6.81078881562336e-06,
      "loss": 0.7155,
      "step": 13903
    },
    {
      "epoch": 0.4006223707716245,
      "grad_norm": 0.7627962231636047,
      "learning_rate": 6.810353872595348e-06,
      "loss": 0.7498,
      "step": 13904
    },
    {
      "epoch": 0.4006511842332738,
      "grad_norm": 0.7601114511489868,
      "learning_rate": 6.809918913800961e-06,
      "loss": 0.7389,
      "step": 13905
    },
    {
      "epoch": 0.4006799976949231,
      "grad_norm": 0.7462021112442017,
      "learning_rate": 6.809483939243991e-06,
      "loss": 0.7027,
      "step": 13906
    },
    {
      "epoch": 0.40070881115657236,
      "grad_norm": 0.7964787483215332,
      "learning_rate": 6.809048948928225e-06,
      "loss": 0.7495,
      "step": 13907
    },
    {
      "epoch": 0.40073762461822166,
      "grad_norm": 0.7707760334014893,
      "learning_rate": 6.808613942857452e-06,
      "loss": 0.7383,
      "step": 13908
    },
    {
      "epoch": 0.4007664380798709,
      "grad_norm": 0.7839051485061646,
      "learning_rate": 6.808178921035456e-06,
      "loss": 0.7233,
      "step": 13909
    },
    {
      "epoch": 0.4007952515415202,
      "grad_norm": 0.7656719088554382,
      "learning_rate": 6.8077438834660326e-06,
      "loss": 0.7347,
      "step": 13910
    },
    {
      "epoch": 0.40082406500316947,
      "grad_norm": 0.7562004923820496,
      "learning_rate": 6.807308830152966e-06,
      "loss": 0.7232,
      "step": 13911
    },
    {
      "epoch": 0.40085287846481876,
      "grad_norm": 0.7295008301734924,
      "learning_rate": 6.806873761100045e-06,
      "loss": 0.7243,
      "step": 13912
    },
    {
      "epoch": 0.40088169192646805,
      "grad_norm": 0.7531232833862305,
      "learning_rate": 6.80643867631106e-06,
      "loss": 0.7247,
      "step": 13913
    },
    {
      "epoch": 0.40091050538811734,
      "grad_norm": 0.7504108548164368,
      "learning_rate": 6.806003575789801e-06,
      "loss": 0.737,
      "step": 13914
    },
    {
      "epoch": 0.40093931884976663,
      "grad_norm": 0.7398231029510498,
      "learning_rate": 6.805568459540055e-06,
      "loss": 0.7152,
      "step": 13915
    },
    {
      "epoch": 0.40096813231141587,
      "grad_norm": 0.7221415042877197,
      "learning_rate": 6.805133327565614e-06,
      "loss": 0.703,
      "step": 13916
    },
    {
      "epoch": 0.40099694577306516,
      "grad_norm": 0.7422804832458496,
      "learning_rate": 6.804698179870266e-06,
      "loss": 0.7078,
      "step": 13917
    },
    {
      "epoch": 0.40102575923471445,
      "grad_norm": 0.7346204519271851,
      "learning_rate": 6.804263016457801e-06,
      "loss": 0.7274,
      "step": 13918
    },
    {
      "epoch": 0.40105457269636374,
      "grad_norm": 0.7398753762245178,
      "learning_rate": 6.803827837332007e-06,
      "loss": 0.718,
      "step": 13919
    },
    {
      "epoch": 0.40108338615801303,
      "grad_norm": 0.7854095697402954,
      "learning_rate": 6.803392642496678e-06,
      "loss": 0.7478,
      "step": 13920
    },
    {
      "epoch": 0.4011121996196623,
      "grad_norm": 0.7886198163032532,
      "learning_rate": 6.802957431955599e-06,
      "loss": 0.7493,
      "step": 13921
    },
    {
      "epoch": 0.4011410130813116,
      "grad_norm": 2.840467929840088,
      "learning_rate": 6.802522205712565e-06,
      "loss": 0.7885,
      "step": 13922
    },
    {
      "epoch": 0.40116982654296085,
      "grad_norm": 0.7558680176734924,
      "learning_rate": 6.802086963771365e-06,
      "loss": 0.7146,
      "step": 13923
    },
    {
      "epoch": 0.40119864000461014,
      "grad_norm": 0.7344669699668884,
      "learning_rate": 6.801651706135788e-06,
      "loss": 0.7148,
      "step": 13924
    },
    {
      "epoch": 0.40122745346625943,
      "grad_norm": 0.7725070118904114,
      "learning_rate": 6.801216432809627e-06,
      "loss": 0.7311,
      "step": 13925
    },
    {
      "epoch": 0.4012562669279087,
      "grad_norm": 0.7843835949897766,
      "learning_rate": 6.80078114379667e-06,
      "loss": 0.7414,
      "step": 13926
    },
    {
      "epoch": 0.401285080389558,
      "grad_norm": 0.7355837225914001,
      "learning_rate": 6.800345839100708e-06,
      "loss": 0.7275,
      "step": 13927
    },
    {
      "epoch": 0.4013138938512073,
      "grad_norm": 0.7203369140625,
      "learning_rate": 6.799910518725536e-06,
      "loss": 0.7336,
      "step": 13928
    },
    {
      "epoch": 0.40134270731285654,
      "grad_norm": 0.7738032937049866,
      "learning_rate": 6.799475182674942e-06,
      "loss": 0.7431,
      "step": 13929
    },
    {
      "epoch": 0.40137152077450583,
      "grad_norm": 0.7857947945594788,
      "learning_rate": 6.799039830952718e-06,
      "loss": 0.7482,
      "step": 13930
    },
    {
      "epoch": 0.4014003342361551,
      "grad_norm": 0.795666515827179,
      "learning_rate": 6.7986044635626546e-06,
      "loss": 0.7633,
      "step": 13931
    },
    {
      "epoch": 0.4014291476978044,
      "grad_norm": 0.7762463688850403,
      "learning_rate": 6.798169080508544e-06,
      "loss": 0.7386,
      "step": 13932
    },
    {
      "epoch": 0.4014579611594537,
      "grad_norm": 0.7831991910934448,
      "learning_rate": 6.797733681794179e-06,
      "loss": 0.7391,
      "step": 13933
    },
    {
      "epoch": 0.401486774621103,
      "grad_norm": 0.7918998599052429,
      "learning_rate": 6.79729826742335e-06,
      "loss": 0.7165,
      "step": 13934
    },
    {
      "epoch": 0.4015155880827523,
      "grad_norm": 0.7778486013412476,
      "learning_rate": 6.796862837399851e-06,
      "loss": 0.7324,
      "step": 13935
    },
    {
      "epoch": 0.4015444015444015,
      "grad_norm": 2.8010592460632324,
      "learning_rate": 6.796427391727472e-06,
      "loss": 0.765,
      "step": 13936
    },
    {
      "epoch": 0.4015732150060508,
      "grad_norm": 0.7498674392700195,
      "learning_rate": 6.795991930410007e-06,
      "loss": 0.7111,
      "step": 13937
    },
    {
      "epoch": 0.4016020284677001,
      "grad_norm": 0.7610108852386475,
      "learning_rate": 6.795556453451247e-06,
      "loss": 0.7415,
      "step": 13938
    },
    {
      "epoch": 0.4016308419293494,
      "grad_norm": 0.7530759572982788,
      "learning_rate": 6.795120960854985e-06,
      "loss": 0.7383,
      "step": 13939
    },
    {
      "epoch": 0.4016596553909987,
      "grad_norm": 0.8382225632667542,
      "learning_rate": 6.794685452625015e-06,
      "loss": 0.7607,
      "step": 13940
    },
    {
      "epoch": 0.401688468852648,
      "grad_norm": 0.7805237770080566,
      "learning_rate": 6.794249928765127e-06,
      "loss": 0.7614,
      "step": 13941
    },
    {
      "epoch": 0.40171728231429726,
      "grad_norm": 0.7571566104888916,
      "learning_rate": 6.793814389279117e-06,
      "loss": 0.7278,
      "step": 13942
    },
    {
      "epoch": 0.4017460957759465,
      "grad_norm": 0.7429221868515015,
      "learning_rate": 6.793378834170777e-06,
      "loss": 0.728,
      "step": 13943
    },
    {
      "epoch": 0.4017749092375958,
      "grad_norm": 0.7815713882446289,
      "learning_rate": 6.792943263443898e-06,
      "loss": 0.7385,
      "step": 13944
    },
    {
      "epoch": 0.4018037226992451,
      "grad_norm": 0.7481291890144348,
      "learning_rate": 6.792507677102278e-06,
      "loss": 0.7296,
      "step": 13945
    },
    {
      "epoch": 0.40183253616089437,
      "grad_norm": 0.7543662190437317,
      "learning_rate": 6.792072075149707e-06,
      "loss": 0.718,
      "step": 13946
    },
    {
      "epoch": 0.40186134962254366,
      "grad_norm": 0.7961996793746948,
      "learning_rate": 6.79163645758998e-06,
      "loss": 0.7296,
      "step": 13947
    },
    {
      "epoch": 0.40189016308419295,
      "grad_norm": 0.7748412489891052,
      "learning_rate": 6.791200824426891e-06,
      "loss": 0.7466,
      "step": 13948
    },
    {
      "epoch": 0.40191897654584224,
      "grad_norm": 0.753455638885498,
      "learning_rate": 6.790765175664232e-06,
      "loss": 0.7191,
      "step": 13949
    },
    {
      "epoch": 0.4019477900074915,
      "grad_norm": 0.740777313709259,
      "learning_rate": 6.790329511305798e-06,
      "loss": 0.7315,
      "step": 13950
    },
    {
      "epoch": 0.40197660346914077,
      "grad_norm": 0.7276339530944824,
      "learning_rate": 6.7898938313553855e-06,
      "loss": 0.7285,
      "step": 13951
    },
    {
      "epoch": 0.40200541693079006,
      "grad_norm": 1.0789248943328857,
      "learning_rate": 6.789458135816784e-06,
      "loss": 0.7477,
      "step": 13952
    },
    {
      "epoch": 0.40203423039243935,
      "grad_norm": 0.7740891575813293,
      "learning_rate": 6.789022424693794e-06,
      "loss": 0.7412,
      "step": 13953
    },
    {
      "epoch": 0.40206304385408864,
      "grad_norm": 0.7221028804779053,
      "learning_rate": 6.788586697990206e-06,
      "loss": 0.7038,
      "step": 13954
    },
    {
      "epoch": 0.40209185731573793,
      "grad_norm": 0.7901386618614197,
      "learning_rate": 6.788150955709816e-06,
      "loss": 0.7122,
      "step": 13955
    },
    {
      "epoch": 0.40212067077738717,
      "grad_norm": 0.7449027895927429,
      "learning_rate": 6.7877151978564174e-06,
      "loss": 0.699,
      "step": 13956
    },
    {
      "epoch": 0.40214948423903646,
      "grad_norm": 0.7322900891304016,
      "learning_rate": 6.787279424433808e-06,
      "loss": 0.7255,
      "step": 13957
    },
    {
      "epoch": 0.40217829770068575,
      "grad_norm": 0.7543475031852722,
      "learning_rate": 6.7868436354457815e-06,
      "loss": 0.7311,
      "step": 13958
    },
    {
      "epoch": 0.40220711116233504,
      "grad_norm": 0.7589486837387085,
      "learning_rate": 6.786407830896133e-06,
      "loss": 0.7136,
      "step": 13959
    },
    {
      "epoch": 0.40223592462398433,
      "grad_norm": 0.7811024188995361,
      "learning_rate": 6.785972010788657e-06,
      "loss": 0.7421,
      "step": 13960
    },
    {
      "epoch": 0.4022647380856336,
      "grad_norm": 0.7628427147865295,
      "learning_rate": 6.785536175127152e-06,
      "loss": 0.7195,
      "step": 13961
    },
    {
      "epoch": 0.4022935515472829,
      "grad_norm": 0.732218325138092,
      "learning_rate": 6.785100323915409e-06,
      "loss": 0.7402,
      "step": 13962
    },
    {
      "epoch": 0.40232236500893215,
      "grad_norm": 0.7482765316963196,
      "learning_rate": 6.7846644571572285e-06,
      "loss": 0.7152,
      "step": 13963
    },
    {
      "epoch": 0.40235117847058144,
      "grad_norm": 0.8266645669937134,
      "learning_rate": 6.784228574856404e-06,
      "loss": 0.7147,
      "step": 13964
    },
    {
      "epoch": 0.40237999193223073,
      "grad_norm": 0.7343252897262573,
      "learning_rate": 6.783792677016733e-06,
      "loss": 0.7303,
      "step": 13965
    },
    {
      "epoch": 0.40240880539388,
      "grad_norm": 0.7409564256668091,
      "learning_rate": 6.78335676364201e-06,
      "loss": 0.7292,
      "step": 13966
    },
    {
      "epoch": 0.4024376188555293,
      "grad_norm": 0.7586785554885864,
      "learning_rate": 6.782920834736033e-06,
      "loss": 0.7397,
      "step": 13967
    },
    {
      "epoch": 0.4024664323171786,
      "grad_norm": 0.7271866202354431,
      "learning_rate": 6.782484890302597e-06,
      "loss": 0.7233,
      "step": 13968
    },
    {
      "epoch": 0.4024952457788279,
      "grad_norm": 0.7646732926368713,
      "learning_rate": 6.7820489303455e-06,
      "loss": 0.7397,
      "step": 13969
    },
    {
      "epoch": 0.40252405924047713,
      "grad_norm": 0.7262097001075745,
      "learning_rate": 6.781612954868538e-06,
      "loss": 0.7541,
      "step": 13970
    },
    {
      "epoch": 0.4025528727021264,
      "grad_norm": 0.7571394443511963,
      "learning_rate": 6.781176963875508e-06,
      "loss": 0.7368,
      "step": 13971
    },
    {
      "epoch": 0.4025816861637757,
      "grad_norm": 0.7356496453285217,
      "learning_rate": 6.780740957370208e-06,
      "loss": 0.7405,
      "step": 13972
    },
    {
      "epoch": 0.402610499625425,
      "grad_norm": 0.7470766305923462,
      "learning_rate": 6.780304935356433e-06,
      "loss": 0.7227,
      "step": 13973
    },
    {
      "epoch": 0.4026393130870743,
      "grad_norm": 0.764542281627655,
      "learning_rate": 6.779868897837983e-06,
      "loss": 0.7326,
      "step": 13974
    },
    {
      "epoch": 0.4026681265487236,
      "grad_norm": 0.7563154101371765,
      "learning_rate": 6.779432844818653e-06,
      "loss": 0.7357,
      "step": 13975
    },
    {
      "epoch": 0.4026969400103729,
      "grad_norm": 0.7277926206588745,
      "learning_rate": 6.7789967763022425e-06,
      "loss": 0.6879,
      "step": 13976
    },
    {
      "epoch": 0.4027257534720221,
      "grad_norm": 0.7396674156188965,
      "learning_rate": 6.778560692292548e-06,
      "loss": 0.7416,
      "step": 13977
    },
    {
      "epoch": 0.4027545669336714,
      "grad_norm": 0.764132022857666,
      "learning_rate": 6.778124592793368e-06,
      "loss": 0.7216,
      "step": 13978
    },
    {
      "epoch": 0.4027833803953207,
      "grad_norm": 0.7761433124542236,
      "learning_rate": 6.7776884778085005e-06,
      "loss": 0.7448,
      "step": 13979
    },
    {
      "epoch": 0.40281219385697,
      "grad_norm": 0.7204237580299377,
      "learning_rate": 6.7772523473417426e-06,
      "loss": 0.7088,
      "step": 13980
    },
    {
      "epoch": 0.40284100731861927,
      "grad_norm": 0.7370696663856506,
      "learning_rate": 6.776816201396894e-06,
      "loss": 0.7309,
      "step": 13981
    },
    {
      "epoch": 0.40286982078026856,
      "grad_norm": 0.7179544568061829,
      "learning_rate": 6.776380039977752e-06,
      "loss": 0.7369,
      "step": 13982
    },
    {
      "epoch": 0.4028986342419178,
      "grad_norm": 0.778170108795166,
      "learning_rate": 6.775943863088117e-06,
      "loss": 0.7303,
      "step": 13983
    },
    {
      "epoch": 0.4029274477035671,
      "grad_norm": 0.7775899767875671,
      "learning_rate": 6.775507670731785e-06,
      "loss": 0.7275,
      "step": 13984
    },
    {
      "epoch": 0.4029562611652164,
      "grad_norm": 0.7669978141784668,
      "learning_rate": 6.775071462912557e-06,
      "loss": 0.7288,
      "step": 13985
    },
    {
      "epoch": 0.40298507462686567,
      "grad_norm": 0.7433903217315674,
      "learning_rate": 6.7746352396342286e-06,
      "loss": 0.7204,
      "step": 13986
    },
    {
      "epoch": 0.40301388808851496,
      "grad_norm": 0.7374637722969055,
      "learning_rate": 6.774199000900603e-06,
      "loss": 0.7443,
      "step": 13987
    },
    {
      "epoch": 0.40304270155016425,
      "grad_norm": 0.7331851124763489,
      "learning_rate": 6.773762746715477e-06,
      "loss": 0.7076,
      "step": 13988
    },
    {
      "epoch": 0.40307151501181354,
      "grad_norm": 0.7592601180076599,
      "learning_rate": 6.773326477082652e-06,
      "loss": 0.7412,
      "step": 13989
    },
    {
      "epoch": 0.4031003284734628,
      "grad_norm": 0.7571284174919128,
      "learning_rate": 6.772890192005924e-06,
      "loss": 0.752,
      "step": 13990
    },
    {
      "epoch": 0.40312914193511207,
      "grad_norm": 0.7451076507568359,
      "learning_rate": 6.772453891489097e-06,
      "loss": 0.7203,
      "step": 13991
    },
    {
      "epoch": 0.40315795539676136,
      "grad_norm": 0.7405235171318054,
      "learning_rate": 6.772017575535967e-06,
      "loss": 0.739,
      "step": 13992
    },
    {
      "epoch": 0.40318676885841065,
      "grad_norm": 0.7454937696456909,
      "learning_rate": 6.771581244150335e-06,
      "loss": 0.7281,
      "step": 13993
    },
    {
      "epoch": 0.40321558232005994,
      "grad_norm": 0.7693125009536743,
      "learning_rate": 6.771144897336002e-06,
      "loss": 0.751,
      "step": 13994
    },
    {
      "epoch": 0.40324439578170923,
      "grad_norm": 0.7459592223167419,
      "learning_rate": 6.770708535096767e-06,
      "loss": 0.7445,
      "step": 13995
    },
    {
      "epoch": 0.4032732092433585,
      "grad_norm": 0.7605525255203247,
      "learning_rate": 6.770272157436432e-06,
      "loss": 0.7377,
      "step": 13996
    },
    {
      "epoch": 0.40330202270500776,
      "grad_norm": 0.7519412636756897,
      "learning_rate": 6.769835764358794e-06,
      "loss": 0.7269,
      "step": 13997
    },
    {
      "epoch": 0.40333083616665705,
      "grad_norm": 0.7353032231330872,
      "learning_rate": 6.769399355867656e-06,
      "loss": 0.7252,
      "step": 13998
    },
    {
      "epoch": 0.40335964962830634,
      "grad_norm": 0.7554631233215332,
      "learning_rate": 6.7689629319668206e-06,
      "loss": 0.7201,
      "step": 13999
    },
    {
      "epoch": 0.40338846308995563,
      "grad_norm": 0.8062451481819153,
      "learning_rate": 6.768526492660085e-06,
      "loss": 0.7265,
      "step": 14000
    },
    {
      "epoch": 0.4034172765516049,
      "grad_norm": 0.7597629427909851,
      "learning_rate": 6.768090037951252e-06,
      "loss": 0.7493,
      "step": 14001
    },
    {
      "epoch": 0.4034460900132542,
      "grad_norm": 0.7534950971603394,
      "learning_rate": 6.767653567844121e-06,
      "loss": 0.7452,
      "step": 14002
    },
    {
      "epoch": 0.4034749034749035,
      "grad_norm": 0.7388479113578796,
      "learning_rate": 6.767217082342495e-06,
      "loss": 0.7105,
      "step": 14003
    },
    {
      "epoch": 0.40350371693655274,
      "grad_norm": 0.7337167263031006,
      "learning_rate": 6.766780581450174e-06,
      "loss": 0.7267,
      "step": 14004
    },
    {
      "epoch": 0.403532530398202,
      "grad_norm": 0.7736970782279968,
      "learning_rate": 6.766344065170961e-06,
      "loss": 0.7206,
      "step": 14005
    },
    {
      "epoch": 0.4035613438598513,
      "grad_norm": 0.7692937850952148,
      "learning_rate": 6.765907533508658e-06,
      "loss": 0.7312,
      "step": 14006
    },
    {
      "epoch": 0.4035901573215006,
      "grad_norm": 0.7303367257118225,
      "learning_rate": 6.765470986467064e-06,
      "loss": 0.7236,
      "step": 14007
    },
    {
      "epoch": 0.4036189707831499,
      "grad_norm": 0.748474657535553,
      "learning_rate": 6.765034424049984e-06,
      "loss": 0.7609,
      "step": 14008
    },
    {
      "epoch": 0.4036477842447992,
      "grad_norm": 0.8036729693412781,
      "learning_rate": 6.764597846261217e-06,
      "loss": 0.7202,
      "step": 14009
    },
    {
      "epoch": 0.4036765977064484,
      "grad_norm": 0.7841523885726929,
      "learning_rate": 6.764161253104567e-06,
      "loss": 0.7158,
      "step": 14010
    },
    {
      "epoch": 0.4037054111680977,
      "grad_norm": 0.7749262452125549,
      "learning_rate": 6.763724644583838e-06,
      "loss": 0.7343,
      "step": 14011
    },
    {
      "epoch": 0.403734224629747,
      "grad_norm": 0.7391152381896973,
      "learning_rate": 6.7632880207028276e-06,
      "loss": 0.7169,
      "step": 14012
    },
    {
      "epoch": 0.4037630380913963,
      "grad_norm": 0.75925213098526,
      "learning_rate": 6.762851381465343e-06,
      "loss": 0.7348,
      "step": 14013
    },
    {
      "epoch": 0.4037918515530456,
      "grad_norm": 0.7584971189498901,
      "learning_rate": 6.762414726875184e-06,
      "loss": 0.7451,
      "step": 14014
    },
    {
      "epoch": 0.4038206650146949,
      "grad_norm": 0.7940345406532288,
      "learning_rate": 6.761978056936156e-06,
      "loss": 0.7059,
      "step": 14015
    },
    {
      "epoch": 0.40384947847634417,
      "grad_norm": 0.7754139304161072,
      "learning_rate": 6.7615413716520585e-06,
      "loss": 0.75,
      "step": 14016
    },
    {
      "epoch": 0.4038782919379934,
      "grad_norm": 0.7534303665161133,
      "learning_rate": 6.761104671026697e-06,
      "loss": 0.7387,
      "step": 14017
    },
    {
      "epoch": 0.4039071053996427,
      "grad_norm": 0.7428169250488281,
      "learning_rate": 6.760667955063875e-06,
      "loss": 0.7339,
      "step": 14018
    },
    {
      "epoch": 0.403935918861292,
      "grad_norm": 0.7413035035133362,
      "learning_rate": 6.760231223767395e-06,
      "loss": 0.7417,
      "step": 14019
    },
    {
      "epoch": 0.4039647323229413,
      "grad_norm": 0.77325439453125,
      "learning_rate": 6.759794477141062e-06,
      "loss": 0.7337,
      "step": 14020
    },
    {
      "epoch": 0.40399354578459057,
      "grad_norm": 0.753341794013977,
      "learning_rate": 6.759357715188677e-06,
      "loss": 0.741,
      "step": 14021
    },
    {
      "epoch": 0.40402235924623986,
      "grad_norm": 0.7603271007537842,
      "learning_rate": 6.758920937914044e-06,
      "loss": 0.7435,
      "step": 14022
    },
    {
      "epoch": 0.40405117270788915,
      "grad_norm": 0.7830179333686829,
      "learning_rate": 6.758484145320968e-06,
      "loss": 0.7451,
      "step": 14023
    },
    {
      "epoch": 0.4040799861695384,
      "grad_norm": 0.79144686460495,
      "learning_rate": 6.758047337413253e-06,
      "loss": 0.7471,
      "step": 14024
    },
    {
      "epoch": 0.4041087996311877,
      "grad_norm": 0.7567646503448486,
      "learning_rate": 6.757610514194706e-06,
      "loss": 0.7252,
      "step": 14025
    },
    {
      "epoch": 0.40413761309283697,
      "grad_norm": 0.7536794543266296,
      "learning_rate": 6.757173675669126e-06,
      "loss": 0.7354,
      "step": 14026
    },
    {
      "epoch": 0.40416642655448626,
      "grad_norm": 0.79564368724823,
      "learning_rate": 6.75673682184032e-06,
      "loss": 0.7157,
      "step": 14027
    },
    {
      "epoch": 0.40419524001613555,
      "grad_norm": 0.7696869969367981,
      "learning_rate": 6.756299952712091e-06,
      "loss": 0.7443,
      "step": 14028
    },
    {
      "epoch": 0.40422405347778484,
      "grad_norm": 0.7583709359169006,
      "learning_rate": 6.755863068288247e-06,
      "loss": 0.7659,
      "step": 14029
    },
    {
      "epoch": 0.40425286693943413,
      "grad_norm": 0.7297559976577759,
      "learning_rate": 6.755426168572591e-06,
      "loss": 0.7129,
      "step": 14030
    },
    {
      "epoch": 0.40428168040108337,
      "grad_norm": 0.7621325254440308,
      "learning_rate": 6.754989253568926e-06,
      "loss": 0.7154,
      "step": 14031
    },
    {
      "epoch": 0.40431049386273266,
      "grad_norm": 0.7279840111732483,
      "learning_rate": 6.75455232328106e-06,
      "loss": 0.7389,
      "step": 14032
    },
    {
      "epoch": 0.40433930732438195,
      "grad_norm": 0.7282305359840393,
      "learning_rate": 6.754115377712797e-06,
      "loss": 0.733,
      "step": 14033
    },
    {
      "epoch": 0.40436812078603124,
      "grad_norm": 0.7297706604003906,
      "learning_rate": 6.7536784168679435e-06,
      "loss": 0.7164,
      "step": 14034
    },
    {
      "epoch": 0.40439693424768053,
      "grad_norm": 0.7452936172485352,
      "learning_rate": 6.753241440750303e-06,
      "loss": 0.7236,
      "step": 14035
    },
    {
      "epoch": 0.4044257477093298,
      "grad_norm": 0.7545789480209351,
      "learning_rate": 6.752804449363683e-06,
      "loss": 0.7427,
      "step": 14036
    },
    {
      "epoch": 0.40445456117097905,
      "grad_norm": 0.7371099591255188,
      "learning_rate": 6.752367442711888e-06,
      "loss": 0.7279,
      "step": 14037
    },
    {
      "epoch": 0.40448337463262835,
      "grad_norm": 0.7362310886383057,
      "learning_rate": 6.7519304207987245e-06,
      "loss": 0.726,
      "step": 14038
    },
    {
      "epoch": 0.40451218809427764,
      "grad_norm": 0.7563129663467407,
      "learning_rate": 6.751493383627999e-06,
      "loss": 0.7372,
      "step": 14039
    },
    {
      "epoch": 0.4045410015559269,
      "grad_norm": 0.7474780678749084,
      "learning_rate": 6.751056331203515e-06,
      "loss": 0.7355,
      "step": 14040
    },
    {
      "epoch": 0.4045698150175762,
      "grad_norm": 0.7386165857315063,
      "learning_rate": 6.750619263529082e-06,
      "loss": 0.7162,
      "step": 14041
    },
    {
      "epoch": 0.4045986284792255,
      "grad_norm": 0.7433596849441528,
      "learning_rate": 6.7501821806085046e-06,
      "loss": 0.7237,
      "step": 14042
    },
    {
      "epoch": 0.4046274419408748,
      "grad_norm": 0.7467425465583801,
      "learning_rate": 6.749745082445592e-06,
      "loss": 0.7361,
      "step": 14043
    },
    {
      "epoch": 0.40465625540252403,
      "grad_norm": 0.7391537427902222,
      "learning_rate": 6.749307969044147e-06,
      "loss": 0.7278,
      "step": 14044
    },
    {
      "epoch": 0.4046850688641733,
      "grad_norm": 0.720696210861206,
      "learning_rate": 6.748870840407978e-06,
      "loss": 0.7433,
      "step": 14045
    },
    {
      "epoch": 0.4047138823258226,
      "grad_norm": 0.7626859545707703,
      "learning_rate": 6.748433696540891e-06,
      "loss": 0.7385,
      "step": 14046
    },
    {
      "epoch": 0.4047426957874719,
      "grad_norm": 0.7075384855270386,
      "learning_rate": 6.747996537446697e-06,
      "loss": 0.7223,
      "step": 14047
    },
    {
      "epoch": 0.4047715092491212,
      "grad_norm": 0.7452930212020874,
      "learning_rate": 6.747559363129199e-06,
      "loss": 0.747,
      "step": 14048
    },
    {
      "epoch": 0.4048003227107705,
      "grad_norm": 0.7638458013534546,
      "learning_rate": 6.7471221735922065e-06,
      "loss": 0.721,
      "step": 14049
    },
    {
      "epoch": 0.4048291361724198,
      "grad_norm": 0.7718395590782166,
      "learning_rate": 6.7466849688395255e-06,
      "loss": 0.7314,
      "step": 14050
    },
    {
      "epoch": 0.404857949634069,
      "grad_norm": 0.7378211617469788,
      "learning_rate": 6.7462477488749654e-06,
      "loss": 0.7218,
      "step": 14051
    },
    {
      "epoch": 0.4048867630957183,
      "grad_norm": 0.7467851638793945,
      "learning_rate": 6.745810513702332e-06,
      "loss": 0.7514,
      "step": 14052
    },
    {
      "epoch": 0.4049155765573676,
      "grad_norm": 0.8358282446861267,
      "learning_rate": 6.745373263325435e-06,
      "loss": 0.7157,
      "step": 14053
    },
    {
      "epoch": 0.4049443900190169,
      "grad_norm": 0.7375636100769043,
      "learning_rate": 6.744935997748082e-06,
      "loss": 0.7318,
      "step": 14054
    },
    {
      "epoch": 0.4049732034806662,
      "grad_norm": 0.7306642532348633,
      "learning_rate": 6.744498716974081e-06,
      "loss": 0.7182,
      "step": 14055
    },
    {
      "epoch": 0.40500201694231547,
      "grad_norm": 0.754840075969696,
      "learning_rate": 6.74406142100724e-06,
      "loss": 0.7269,
      "step": 14056
    },
    {
      "epoch": 0.40503083040396476,
      "grad_norm": 0.7387733459472656,
      "learning_rate": 6.743624109851367e-06,
      "loss": 0.7273,
      "step": 14057
    },
    {
      "epoch": 0.405059643865614,
      "grad_norm": 0.7342123985290527,
      "learning_rate": 6.74318678351027e-06,
      "loss": 0.7349,
      "step": 14058
    },
    {
      "epoch": 0.4050884573272633,
      "grad_norm": 0.7430001497268677,
      "learning_rate": 6.742749441987759e-06,
      "loss": 0.7408,
      "step": 14059
    },
    {
      "epoch": 0.4051172707889126,
      "grad_norm": 0.7685555219650269,
      "learning_rate": 6.742312085287643e-06,
      "loss": 0.7471,
      "step": 14060
    },
    {
      "epoch": 0.40514608425056187,
      "grad_norm": 0.7328569889068604,
      "learning_rate": 6.7418747134137305e-06,
      "loss": 0.7395,
      "step": 14061
    },
    {
      "epoch": 0.40517489771221116,
      "grad_norm": 0.7605501413345337,
      "learning_rate": 6.741437326369831e-06,
      "loss": 0.7207,
      "step": 14062
    },
    {
      "epoch": 0.40520371117386045,
      "grad_norm": 0.7381850481033325,
      "learning_rate": 6.740999924159753e-06,
      "loss": 0.7099,
      "step": 14063
    },
    {
      "epoch": 0.4052325246355097,
      "grad_norm": 0.7653622627258301,
      "learning_rate": 6.740562506787305e-06,
      "loss": 0.7351,
      "step": 14064
    },
    {
      "epoch": 0.405261338097159,
      "grad_norm": 0.7363989949226379,
      "learning_rate": 6.740125074256297e-06,
      "loss": 0.731,
      "step": 14065
    },
    {
      "epoch": 0.40529015155880826,
      "grad_norm": 0.7672138810157776,
      "learning_rate": 6.739687626570542e-06,
      "loss": 0.7619,
      "step": 14066
    },
    {
      "epoch": 0.40531896502045756,
      "grad_norm": 0.7546598315238953,
      "learning_rate": 6.739250163733844e-06,
      "loss": 0.7468,
      "step": 14067
    },
    {
      "epoch": 0.40534777848210685,
      "grad_norm": 0.76655513048172,
      "learning_rate": 6.738812685750017e-06,
      "loss": 0.7419,
      "step": 14068
    },
    {
      "epoch": 0.40537659194375614,
      "grad_norm": 0.7558541893959045,
      "learning_rate": 6.738375192622868e-06,
      "loss": 0.7138,
      "step": 14069
    },
    {
      "epoch": 0.40540540540540543,
      "grad_norm": 0.8163595795631409,
      "learning_rate": 6.737937684356211e-06,
      "loss": 0.7532,
      "step": 14070
    },
    {
      "epoch": 0.40543421886705466,
      "grad_norm": 0.7783345580101013,
      "learning_rate": 6.737500160953852e-06,
      "loss": 0.7486,
      "step": 14071
    },
    {
      "epoch": 0.40546303232870395,
      "grad_norm": 0.7670953869819641,
      "learning_rate": 6.737062622419605e-06,
      "loss": 0.7166,
      "step": 14072
    },
    {
      "epoch": 0.40549184579035324,
      "grad_norm": 0.7638422846794128,
      "learning_rate": 6.736625068757279e-06,
      "loss": 0.738,
      "step": 14073
    },
    {
      "epoch": 0.40552065925200254,
      "grad_norm": 0.786703884601593,
      "learning_rate": 6.736187499970684e-06,
      "loss": 0.7374,
      "step": 14074
    },
    {
      "epoch": 0.4055494727136518,
      "grad_norm": 0.7334575057029724,
      "learning_rate": 6.735749916063633e-06,
      "loss": 0.7296,
      "step": 14075
    },
    {
      "epoch": 0.4055782861753011,
      "grad_norm": 0.7504628896713257,
      "learning_rate": 6.735312317039933e-06,
      "loss": 0.7301,
      "step": 14076
    },
    {
      "epoch": 0.4056070996369504,
      "grad_norm": 0.7491046190261841,
      "learning_rate": 6.734874702903399e-06,
      "loss": 0.712,
      "step": 14077
    },
    {
      "epoch": 0.40563591309859964,
      "grad_norm": 0.7368943691253662,
      "learning_rate": 6.73443707365784e-06,
      "loss": 0.7105,
      "step": 14078
    },
    {
      "epoch": 0.40566472656024893,
      "grad_norm": 0.7388572096824646,
      "learning_rate": 6.733999429307068e-06,
      "loss": 0.7275,
      "step": 14079
    },
    {
      "epoch": 0.4056935400218982,
      "grad_norm": 0.7316092848777771,
      "learning_rate": 6.733561769854895e-06,
      "loss": 0.7348,
      "step": 14080
    },
    {
      "epoch": 0.4057223534835475,
      "grad_norm": 0.7515270709991455,
      "learning_rate": 6.7331240953051316e-06,
      "loss": 0.7227,
      "step": 14081
    },
    {
      "epoch": 0.4057511669451968,
      "grad_norm": 0.7497867345809937,
      "learning_rate": 6.732686405661588e-06,
      "loss": 0.7156,
      "step": 14082
    },
    {
      "epoch": 0.4057799804068461,
      "grad_norm": 0.7623345255851746,
      "learning_rate": 6.732248700928079e-06,
      "loss": 0.7053,
      "step": 14083
    },
    {
      "epoch": 0.4058087938684954,
      "grad_norm": 0.7734067440032959,
      "learning_rate": 6.731810981108416e-06,
      "loss": 0.7521,
      "step": 14084
    },
    {
      "epoch": 0.4058376073301446,
      "grad_norm": 0.7359962463378906,
      "learning_rate": 6.731373246206411e-06,
      "loss": 0.741,
      "step": 14085
    },
    {
      "epoch": 0.4058664207917939,
      "grad_norm": 0.7491630911827087,
      "learning_rate": 6.730935496225874e-06,
      "loss": 0.7174,
      "step": 14086
    },
    {
      "epoch": 0.4058952342534432,
      "grad_norm": 0.756737470626831,
      "learning_rate": 6.73049773117062e-06,
      "loss": 0.7495,
      "step": 14087
    },
    {
      "epoch": 0.4059240477150925,
      "grad_norm": 0.7599438428878784,
      "learning_rate": 6.730059951044461e-06,
      "loss": 0.7359,
      "step": 14088
    },
    {
      "epoch": 0.4059528611767418,
      "grad_norm": 0.7484210133552551,
      "learning_rate": 6.72962215585121e-06,
      "loss": 0.6984,
      "step": 14089
    },
    {
      "epoch": 0.4059816746383911,
      "grad_norm": 0.7349493503570557,
      "learning_rate": 6.7291843455946774e-06,
      "loss": 0.745,
      "step": 14090
    },
    {
      "epoch": 0.4060104881000403,
      "grad_norm": 0.7240884304046631,
      "learning_rate": 6.728746520278677e-06,
      "loss": 0.735,
      "step": 14091
    },
    {
      "epoch": 0.4060393015616896,
      "grad_norm": 0.746278703212738,
      "learning_rate": 6.728308679907025e-06,
      "loss": 0.7525,
      "step": 14092
    },
    {
      "epoch": 0.4060681150233389,
      "grad_norm": 0.7399089932441711,
      "learning_rate": 6.727870824483531e-06,
      "loss": 0.7208,
      "step": 14093
    },
    {
      "epoch": 0.4060969284849882,
      "grad_norm": 0.7360291481018066,
      "learning_rate": 6.7274329540120085e-06,
      "loss": 0.7101,
      "step": 14094
    },
    {
      "epoch": 0.4061257419466375,
      "grad_norm": 0.7335733771324158,
      "learning_rate": 6.726995068496273e-06,
      "loss": 0.7294,
      "step": 14095
    },
    {
      "epoch": 0.40615455540828677,
      "grad_norm": 0.7535902261734009,
      "learning_rate": 6.7265571679401365e-06,
      "loss": 0.7277,
      "step": 14096
    },
    {
      "epoch": 0.40618336886993606,
      "grad_norm": 0.7434768080711365,
      "learning_rate": 6.726119252347413e-06,
      "loss": 0.727,
      "step": 14097
    },
    {
      "epoch": 0.4062121823315853,
      "grad_norm": 0.7419313788414001,
      "learning_rate": 6.725681321721916e-06,
      "loss": 0.725,
      "step": 14098
    },
    {
      "epoch": 0.4062409957932346,
      "grad_norm": 0.7752344012260437,
      "learning_rate": 6.7252433760674605e-06,
      "loss": 0.7472,
      "step": 14099
    },
    {
      "epoch": 0.4062698092548839,
      "grad_norm": 0.7451915740966797,
      "learning_rate": 6.724805415387858e-06,
      "loss": 0.7144,
      "step": 14100
    },
    {
      "epoch": 0.40629862271653316,
      "grad_norm": 0.7282819151878357,
      "learning_rate": 6.7243674396869264e-06,
      "loss": 0.7095,
      "step": 14101
    },
    {
      "epoch": 0.40632743617818246,
      "grad_norm": 0.813463032245636,
      "learning_rate": 6.723929448968477e-06,
      "loss": 0.7392,
      "step": 14102
    },
    {
      "epoch": 0.40635624963983175,
      "grad_norm": 0.7801231741905212,
      "learning_rate": 6.723491443236326e-06,
      "loss": 0.725,
      "step": 14103
    },
    {
      "epoch": 0.40638506310148104,
      "grad_norm": 0.7576268315315247,
      "learning_rate": 6.723053422494288e-06,
      "loss": 0.7389,
      "step": 14104
    },
    {
      "epoch": 0.40641387656313027,
      "grad_norm": 0.7823492288589478,
      "learning_rate": 6.722615386746176e-06,
      "loss": 0.7327,
      "step": 14105
    },
    {
      "epoch": 0.40644269002477956,
      "grad_norm": 0.8197482228279114,
      "learning_rate": 6.722177335995805e-06,
      "loss": 0.7443,
      "step": 14106
    },
    {
      "epoch": 0.40647150348642885,
      "grad_norm": 0.7510470747947693,
      "learning_rate": 6.721739270246993e-06,
      "loss": 0.7427,
      "step": 14107
    },
    {
      "epoch": 0.40650031694807814,
      "grad_norm": 0.7565300464630127,
      "learning_rate": 6.721301189503553e-06,
      "loss": 0.7124,
      "step": 14108
    },
    {
      "epoch": 0.40652913040972743,
      "grad_norm": 0.7823172211647034,
      "learning_rate": 6.720863093769298e-06,
      "loss": 0.7461,
      "step": 14109
    },
    {
      "epoch": 0.4065579438713767,
      "grad_norm": 0.7282228469848633,
      "learning_rate": 6.7204249830480485e-06,
      "loss": 0.7117,
      "step": 14110
    },
    {
      "epoch": 0.406586757333026,
      "grad_norm": 0.755691409111023,
      "learning_rate": 6.719986857343617e-06,
      "loss": 0.7502,
      "step": 14111
    },
    {
      "epoch": 0.40661557079467525,
      "grad_norm": 0.7410200238227844,
      "learning_rate": 6.7195487166598165e-06,
      "loss": 0.723,
      "step": 14112
    },
    {
      "epoch": 0.40664438425632454,
      "grad_norm": 0.7507827281951904,
      "learning_rate": 6.719110561000468e-06,
      "loss": 0.7147,
      "step": 14113
    },
    {
      "epoch": 0.40667319771797383,
      "grad_norm": 0.7575891017913818,
      "learning_rate": 6.718672390369384e-06,
      "loss": 0.7175,
      "step": 14114
    },
    {
      "epoch": 0.4067020111796231,
      "grad_norm": 0.7734874486923218,
      "learning_rate": 6.718234204770382e-06,
      "loss": 0.7656,
      "step": 14115
    },
    {
      "epoch": 0.4067308246412724,
      "grad_norm": 0.7725403904914856,
      "learning_rate": 6.717796004207279e-06,
      "loss": 0.7229,
      "step": 14116
    },
    {
      "epoch": 0.4067596381029217,
      "grad_norm": 0.797510027885437,
      "learning_rate": 6.7173577886838885e-06,
      "loss": 0.7082,
      "step": 14117
    },
    {
      "epoch": 0.40678845156457094,
      "grad_norm": 0.7333446145057678,
      "learning_rate": 6.716919558204027e-06,
      "loss": 0.7184,
      "step": 14118
    },
    {
      "epoch": 0.40681726502622023,
      "grad_norm": 0.7535864114761353,
      "learning_rate": 6.716481312771515e-06,
      "loss": 0.7484,
      "step": 14119
    },
    {
      "epoch": 0.4068460784878695,
      "grad_norm": 0.7565536499023438,
      "learning_rate": 6.716043052390165e-06,
      "loss": 0.7461,
      "step": 14120
    },
    {
      "epoch": 0.4068748919495188,
      "grad_norm": 0.7406027317047119,
      "learning_rate": 6.715604777063796e-06,
      "loss": 0.7058,
      "step": 14121
    },
    {
      "epoch": 0.4069037054111681,
      "grad_norm": 0.7326602935791016,
      "learning_rate": 6.715166486796225e-06,
      "loss": 0.7288,
      "step": 14122
    },
    {
      "epoch": 0.4069325188728174,
      "grad_norm": 0.7644399404525757,
      "learning_rate": 6.714728181591268e-06,
      "loss": 0.7455,
      "step": 14123
    },
    {
      "epoch": 0.4069613323344667,
      "grad_norm": 0.7686924934387207,
      "learning_rate": 6.714289861452742e-06,
      "loss": 0.7297,
      "step": 14124
    },
    {
      "epoch": 0.4069901457961159,
      "grad_norm": 0.7401422262191772,
      "learning_rate": 6.713851526384465e-06,
      "loss": 0.7357,
      "step": 14125
    },
    {
      "epoch": 0.4070189592577652,
      "grad_norm": 0.7687620520591736,
      "learning_rate": 6.7134131763902555e-06,
      "loss": 0.7346,
      "step": 14126
    },
    {
      "epoch": 0.4070477727194145,
      "grad_norm": 0.7821947336196899,
      "learning_rate": 6.712974811473929e-06,
      "loss": 0.746,
      "step": 14127
    },
    {
      "epoch": 0.4070765861810638,
      "grad_norm": 0.7466185688972473,
      "learning_rate": 6.712536431639305e-06,
      "loss": 0.7546,
      "step": 14128
    },
    {
      "epoch": 0.4071053996427131,
      "grad_norm": 0.7539228200912476,
      "learning_rate": 6.7120980368902e-06,
      "loss": 0.7311,
      "step": 14129
    },
    {
      "epoch": 0.4071342131043624,
      "grad_norm": 0.7518174052238464,
      "learning_rate": 6.711659627230431e-06,
      "loss": 0.7155,
      "step": 14130
    },
    {
      "epoch": 0.40716302656601167,
      "grad_norm": 0.760629415512085,
      "learning_rate": 6.71122120266382e-06,
      "loss": 0.7269,
      "step": 14131
    },
    {
      "epoch": 0.4071918400276609,
      "grad_norm": 0.7542800903320312,
      "learning_rate": 6.710782763194183e-06,
      "loss": 0.7201,
      "step": 14132
    },
    {
      "epoch": 0.4072206534893102,
      "grad_norm": 0.7861917614936829,
      "learning_rate": 6.710344308825338e-06,
      "loss": 0.7204,
      "step": 14133
    },
    {
      "epoch": 0.4072494669509595,
      "grad_norm": 0.71820068359375,
      "learning_rate": 6.709905839561104e-06,
      "loss": 0.7142,
      "step": 14134
    },
    {
      "epoch": 0.4072782804126088,
      "grad_norm": 0.7377345561981201,
      "learning_rate": 6.709467355405298e-06,
      "loss": 0.7066,
      "step": 14135
    },
    {
      "epoch": 0.40730709387425806,
      "grad_norm": 0.7625753879547119,
      "learning_rate": 6.709028856361741e-06,
      "loss": 0.7048,
      "step": 14136
    },
    {
      "epoch": 0.40733590733590735,
      "grad_norm": 0.7604210376739502,
      "learning_rate": 6.708590342434251e-06,
      "loss": 0.7228,
      "step": 14137
    },
    {
      "epoch": 0.40736472079755665,
      "grad_norm": 0.7421254515647888,
      "learning_rate": 6.708151813626648e-06,
      "loss": 0.716,
      "step": 14138
    },
    {
      "epoch": 0.4073935342592059,
      "grad_norm": 0.7290027141571045,
      "learning_rate": 6.7077132699427484e-06,
      "loss": 0.7111,
      "step": 14139
    },
    {
      "epoch": 0.40742234772085517,
      "grad_norm": 0.7654015421867371,
      "learning_rate": 6.707274711386375e-06,
      "loss": 0.7334,
      "step": 14140
    },
    {
      "epoch": 0.40745116118250446,
      "grad_norm": 0.7494015097618103,
      "learning_rate": 6.706836137961346e-06,
      "loss": 0.7187,
      "step": 14141
    },
    {
      "epoch": 0.40747997464415375,
      "grad_norm": 0.7849544286727905,
      "learning_rate": 6.706397549671477e-06,
      "loss": 0.7422,
      "step": 14142
    },
    {
      "epoch": 0.40750878810580304,
      "grad_norm": 0.7601641416549683,
      "learning_rate": 6.705958946520595e-06,
      "loss": 0.7254,
      "step": 14143
    },
    {
      "epoch": 0.40753760156745233,
      "grad_norm": 0.7448112964630127,
      "learning_rate": 6.705520328512514e-06,
      "loss": 0.7372,
      "step": 14144
    },
    {
      "epoch": 0.40756641502910157,
      "grad_norm": 0.7673789262771606,
      "learning_rate": 6.705081695651056e-06,
      "loss": 0.751,
      "step": 14145
    },
    {
      "epoch": 0.40759522849075086,
      "grad_norm": 0.7193508744239807,
      "learning_rate": 6.7046430479400415e-06,
      "loss": 0.7041,
      "step": 14146
    },
    {
      "epoch": 0.40762404195240015,
      "grad_norm": 0.7647351026535034,
      "learning_rate": 6.704204385383289e-06,
      "loss": 0.7166,
      "step": 14147
    },
    {
      "epoch": 0.40765285541404944,
      "grad_norm": 0.7393225431442261,
      "learning_rate": 6.703765707984619e-06,
      "loss": 0.7164,
      "step": 14148
    },
    {
      "epoch": 0.40768166887569873,
      "grad_norm": 0.7323169112205505,
      "learning_rate": 6.703327015747855e-06,
      "loss": 0.7248,
      "step": 14149
    },
    {
      "epoch": 0.407710482337348,
      "grad_norm": 0.7365840077400208,
      "learning_rate": 6.702888308676815e-06,
      "loss": 0.7187,
      "step": 14150
    },
    {
      "epoch": 0.4077392957989973,
      "grad_norm": 0.7527682781219482,
      "learning_rate": 6.702449586775319e-06,
      "loss": 0.7358,
      "step": 14151
    },
    {
      "epoch": 0.40776810926064655,
      "grad_norm": 0.7544325590133667,
      "learning_rate": 6.7020108500471905e-06,
      "loss": 0.7248,
      "step": 14152
    },
    {
      "epoch": 0.40779692272229584,
      "grad_norm": 0.7495025992393494,
      "learning_rate": 6.701572098496248e-06,
      "loss": 0.7174,
      "step": 14153
    },
    {
      "epoch": 0.40782573618394513,
      "grad_norm": 0.745873212814331,
      "learning_rate": 6.701133332126312e-06,
      "loss": 0.725,
      "step": 14154
    },
    {
      "epoch": 0.4078545496455944,
      "grad_norm": 0.75608229637146,
      "learning_rate": 6.700694550941208e-06,
      "loss": 0.7419,
      "step": 14155
    },
    {
      "epoch": 0.4078833631072437,
      "grad_norm": 0.75030517578125,
      "learning_rate": 6.700255754944754e-06,
      "loss": 0.7199,
      "step": 14156
    },
    {
      "epoch": 0.407912176568893,
      "grad_norm": 0.7323379516601562,
      "learning_rate": 6.699816944140771e-06,
      "loss": 0.7007,
      "step": 14157
    },
    {
      "epoch": 0.4079409900305423,
      "grad_norm": 0.7683362364768982,
      "learning_rate": 6.699378118533082e-06,
      "loss": 0.7597,
      "step": 14158
    },
    {
      "epoch": 0.40796980349219153,
      "grad_norm": 0.7444737553596497,
      "learning_rate": 6.698939278125509e-06,
      "loss": 0.7498,
      "step": 14159
    },
    {
      "epoch": 0.4079986169538408,
      "grad_norm": 0.7410433888435364,
      "learning_rate": 6.698500422921872e-06,
      "loss": 0.7413,
      "step": 14160
    },
    {
      "epoch": 0.4080274304154901,
      "grad_norm": 0.7263144850730896,
      "learning_rate": 6.698061552925994e-06,
      "loss": 0.7243,
      "step": 14161
    },
    {
      "epoch": 0.4080562438771394,
      "grad_norm": 0.755642831325531,
      "learning_rate": 6.697622668141699e-06,
      "loss": 0.7651,
      "step": 14162
    },
    {
      "epoch": 0.4080850573387887,
      "grad_norm": 0.7525361776351929,
      "learning_rate": 6.697183768572806e-06,
      "loss": 0.7324,
      "step": 14163
    },
    {
      "epoch": 0.408113870800438,
      "grad_norm": 0.7185506820678711,
      "learning_rate": 6.696744854223141e-06,
      "loss": 0.7411,
      "step": 14164
    },
    {
      "epoch": 0.4081426842620873,
      "grad_norm": 0.7449265122413635,
      "learning_rate": 6.696305925096523e-06,
      "loss": 0.739,
      "step": 14165
    },
    {
      "epoch": 0.4081714977237365,
      "grad_norm": 0.7543044686317444,
      "learning_rate": 6.695866981196774e-06,
      "loss": 0.7264,
      "step": 14166
    },
    {
      "epoch": 0.4082003111853858,
      "grad_norm": 0.761467695236206,
      "learning_rate": 6.6954280225277215e-06,
      "loss": 0.7304,
      "step": 14167
    },
    {
      "epoch": 0.4082291246470351,
      "grad_norm": 0.7453007102012634,
      "learning_rate": 6.6949890490931845e-06,
      "loss": 0.7285,
      "step": 14168
    },
    {
      "epoch": 0.4082579381086844,
      "grad_norm": 0.7523971796035767,
      "learning_rate": 6.6945500608969875e-06,
      "loss": 0.7134,
      "step": 14169
    },
    {
      "epoch": 0.4082867515703337,
      "grad_norm": 0.7454303503036499,
      "learning_rate": 6.694111057942953e-06,
      "loss": 0.7277,
      "step": 14170
    },
    {
      "epoch": 0.40831556503198296,
      "grad_norm": 0.7550442218780518,
      "learning_rate": 6.693672040234906e-06,
      "loss": 0.7324,
      "step": 14171
    },
    {
      "epoch": 0.4083443784936322,
      "grad_norm": 0.7844300866127014,
      "learning_rate": 6.693233007776668e-06,
      "loss": 0.7244,
      "step": 14172
    },
    {
      "epoch": 0.4083731919552815,
      "grad_norm": 0.7764277458190918,
      "learning_rate": 6.692793960572064e-06,
      "loss": 0.726,
      "step": 14173
    },
    {
      "epoch": 0.4084020054169308,
      "grad_norm": 0.7635419368743896,
      "learning_rate": 6.692354898624916e-06,
      "loss": 0.72,
      "step": 14174
    },
    {
      "epoch": 0.40843081887858007,
      "grad_norm": 0.7435584664344788,
      "learning_rate": 6.691915821939048e-06,
      "loss": 0.7446,
      "step": 14175
    },
    {
      "epoch": 0.40845963234022936,
      "grad_norm": 0.7374675869941711,
      "learning_rate": 6.6914767305182845e-06,
      "loss": 0.7255,
      "step": 14176
    },
    {
      "epoch": 0.40848844580187865,
      "grad_norm": 0.7616056799888611,
      "learning_rate": 6.691037624366451e-06,
      "loss": 0.705,
      "step": 14177
    },
    {
      "epoch": 0.40851725926352794,
      "grad_norm": 0.8072970509529114,
      "learning_rate": 6.6905985034873685e-06,
      "loss": 0.7268,
      "step": 14178
    },
    {
      "epoch": 0.4085460727251772,
      "grad_norm": 0.7516577839851379,
      "learning_rate": 6.690159367884863e-06,
      "loss": 0.7298,
      "step": 14179
    },
    {
      "epoch": 0.40857488618682647,
      "grad_norm": 0.717341423034668,
      "learning_rate": 6.689720217562761e-06,
      "loss": 0.7223,
      "step": 14180
    },
    {
      "epoch": 0.40860369964847576,
      "grad_norm": 0.8071351647377014,
      "learning_rate": 6.689281052524883e-06,
      "loss": 0.761,
      "step": 14181
    },
    {
      "epoch": 0.40863251311012505,
      "grad_norm": 0.7681382298469543,
      "learning_rate": 6.688841872775057e-06,
      "loss": 0.7232,
      "step": 14182
    },
    {
      "epoch": 0.40866132657177434,
      "grad_norm": 0.7589249014854431,
      "learning_rate": 6.688402678317106e-06,
      "loss": 0.7431,
      "step": 14183
    },
    {
      "epoch": 0.40869014003342363,
      "grad_norm": 0.7811243534088135,
      "learning_rate": 6.687963469154854e-06,
      "loss": 0.7227,
      "step": 14184
    },
    {
      "epoch": 0.4087189534950729,
      "grad_norm": 0.7724524736404419,
      "learning_rate": 6.6875242452921295e-06,
      "loss": 0.7333,
      "step": 14185
    },
    {
      "epoch": 0.40874776695672216,
      "grad_norm": 0.7858082056045532,
      "learning_rate": 6.687085006732755e-06,
      "loss": 0.7378,
      "step": 14186
    },
    {
      "epoch": 0.40877658041837145,
      "grad_norm": 0.7559976577758789,
      "learning_rate": 6.6866457534805564e-06,
      "loss": 0.7453,
      "step": 14187
    },
    {
      "epoch": 0.40880539388002074,
      "grad_norm": 0.776553213596344,
      "learning_rate": 6.686206485539359e-06,
      "loss": 0.7377,
      "step": 14188
    },
    {
      "epoch": 0.40883420734167003,
      "grad_norm": 0.7563828229904175,
      "learning_rate": 6.6857672029129895e-06,
      "loss": 0.7305,
      "step": 14189
    },
    {
      "epoch": 0.4088630208033193,
      "grad_norm": 0.770618200302124,
      "learning_rate": 6.685327905605272e-06,
      "loss": 0.7213,
      "step": 14190
    },
    {
      "epoch": 0.4088918342649686,
      "grad_norm": 0.7949202060699463,
      "learning_rate": 6.684888593620034e-06,
      "loss": 0.7345,
      "step": 14191
    },
    {
      "epoch": 0.4089206477266179,
      "grad_norm": 0.7523308992385864,
      "learning_rate": 6.684449266961101e-06,
      "loss": 0.7428,
      "step": 14192
    },
    {
      "epoch": 0.40894946118826714,
      "grad_norm": 0.7643951177597046,
      "learning_rate": 6.684009925632298e-06,
      "loss": 0.7295,
      "step": 14193
    },
    {
      "epoch": 0.40897827464991643,
      "grad_norm": 0.7666624188423157,
      "learning_rate": 6.683570569637452e-06,
      "loss": 0.7009,
      "step": 14194
    },
    {
      "epoch": 0.4090070881115657,
      "grad_norm": 0.7611402869224548,
      "learning_rate": 6.6831311989803884e-06,
      "loss": 0.7316,
      "step": 14195
    },
    {
      "epoch": 0.409035901573215,
      "grad_norm": 0.7449886798858643,
      "learning_rate": 6.682691813664936e-06,
      "loss": 0.721,
      "step": 14196
    },
    {
      "epoch": 0.4090647150348643,
      "grad_norm": 0.7311615347862244,
      "learning_rate": 6.68225241369492e-06,
      "loss": 0.7396,
      "step": 14197
    },
    {
      "epoch": 0.4090935284965136,
      "grad_norm": 0.736498236656189,
      "learning_rate": 6.681812999074167e-06,
      "loss": 0.7305,
      "step": 14198
    },
    {
      "epoch": 0.4091223419581628,
      "grad_norm": 0.806958019733429,
      "learning_rate": 6.681373569806503e-06,
      "loss": 0.7377,
      "step": 14199
    },
    {
      "epoch": 0.4091511554198121,
      "grad_norm": 0.7596305012702942,
      "learning_rate": 6.6809341258957575e-06,
      "loss": 0.7266,
      "step": 14200
    },
    {
      "epoch": 0.4091799688814614,
      "grad_norm": 0.7298586368560791,
      "learning_rate": 6.680494667345754e-06,
      "loss": 0.7297,
      "step": 14201
    },
    {
      "epoch": 0.4092087823431107,
      "grad_norm": 0.7643237709999084,
      "learning_rate": 6.6800551941603235e-06,
      "loss": 0.7169,
      "step": 14202
    },
    {
      "epoch": 0.40923759580476,
      "grad_norm": 0.7991321682929993,
      "learning_rate": 6.679615706343291e-06,
      "loss": 0.7469,
      "step": 14203
    },
    {
      "epoch": 0.4092664092664093,
      "grad_norm": 0.7540565133094788,
      "learning_rate": 6.679176203898484e-06,
      "loss": 0.7339,
      "step": 14204
    },
    {
      "epoch": 0.40929522272805857,
      "grad_norm": 0.7469000220298767,
      "learning_rate": 6.678736686829732e-06,
      "loss": 0.7361,
      "step": 14205
    },
    {
      "epoch": 0.4093240361897078,
      "grad_norm": 0.7931962609291077,
      "learning_rate": 6.678297155140861e-06,
      "loss": 0.7203,
      "step": 14206
    },
    {
      "epoch": 0.4093528496513571,
      "grad_norm": 0.7667528390884399,
      "learning_rate": 6.677857608835699e-06,
      "loss": 0.7384,
      "step": 14207
    },
    {
      "epoch": 0.4093816631130064,
      "grad_norm": 0.7554669380187988,
      "learning_rate": 6.677418047918076e-06,
      "loss": 0.736,
      "step": 14208
    },
    {
      "epoch": 0.4094104765746557,
      "grad_norm": 0.7797155380249023,
      "learning_rate": 6.6769784723918176e-06,
      "loss": 0.7332,
      "step": 14209
    },
    {
      "epoch": 0.40943929003630497,
      "grad_norm": 0.7524963617324829,
      "learning_rate": 6.676538882260752e-06,
      "loss": 0.7523,
      "step": 14210
    },
    {
      "epoch": 0.40946810349795426,
      "grad_norm": 0.7511873245239258,
      "learning_rate": 6.676099277528711e-06,
      "loss": 0.735,
      "step": 14211
    },
    {
      "epoch": 0.40949691695960355,
      "grad_norm": 0.7478708028793335,
      "learning_rate": 6.675659658199521e-06,
      "loss": 0.7179,
      "step": 14212
    },
    {
      "epoch": 0.4095257304212528,
      "grad_norm": 0.7448198795318604,
      "learning_rate": 6.675220024277007e-06,
      "loss": 0.7303,
      "step": 14213
    },
    {
      "epoch": 0.4095545438829021,
      "grad_norm": 0.7366493344306946,
      "learning_rate": 6.674780375765003e-06,
      "loss": 0.7179,
      "step": 14214
    },
    {
      "epoch": 0.40958335734455137,
      "grad_norm": 0.7266343832015991,
      "learning_rate": 6.6743407126673375e-06,
      "loss": 0.7428,
      "step": 14215
    },
    {
      "epoch": 0.40961217080620066,
      "grad_norm": 0.7359840273857117,
      "learning_rate": 6.673901034987838e-06,
      "loss": 0.7314,
      "step": 14216
    },
    {
      "epoch": 0.40964098426784995,
      "grad_norm": 0.7341649532318115,
      "learning_rate": 6.673461342730332e-06,
      "loss": 0.7373,
      "step": 14217
    },
    {
      "epoch": 0.40966979772949924,
      "grad_norm": 0.7331247329711914,
      "learning_rate": 6.673021635898651e-06,
      "loss": 0.7357,
      "step": 14218
    },
    {
      "epoch": 0.40969861119114853,
      "grad_norm": 0.7369255423545837,
      "learning_rate": 6.672581914496624e-06,
      "loss": 0.7327,
      "step": 14219
    },
    {
      "epoch": 0.40972742465279777,
      "grad_norm": 0.7363700270652771,
      "learning_rate": 6.6721421785280805e-06,
      "loss": 0.7381,
      "step": 14220
    },
    {
      "epoch": 0.40975623811444706,
      "grad_norm": 0.7529113292694092,
      "learning_rate": 6.6717024279968515e-06,
      "loss": 0.7379,
      "step": 14221
    },
    {
      "epoch": 0.40978505157609635,
      "grad_norm": 0.7352333664894104,
      "learning_rate": 6.6712626629067634e-06,
      "loss": 0.7289,
      "step": 14222
    },
    {
      "epoch": 0.40981386503774564,
      "grad_norm": 0.7315628528594971,
      "learning_rate": 6.670822883261649e-06,
      "loss": 0.7202,
      "step": 14223
    },
    {
      "epoch": 0.40984267849939493,
      "grad_norm": 0.742908775806427,
      "learning_rate": 6.670383089065339e-06,
      "loss": 0.7444,
      "step": 14224
    },
    {
      "epoch": 0.4098714919610442,
      "grad_norm": 0.7632289528846741,
      "learning_rate": 6.6699432803216605e-06,
      "loss": 0.6994,
      "step": 14225
    },
    {
      "epoch": 0.40990030542269346,
      "grad_norm": 0.7399792075157166,
      "learning_rate": 6.669503457034445e-06,
      "loss": 0.7274,
      "step": 14226
    },
    {
      "epoch": 0.40992911888434275,
      "grad_norm": 0.7370651364326477,
      "learning_rate": 6.6690636192075244e-06,
      "loss": 0.7168,
      "step": 14227
    },
    {
      "epoch": 0.40995793234599204,
      "grad_norm": 0.7574241161346436,
      "learning_rate": 6.6686237668447265e-06,
      "loss": 0.7512,
      "step": 14228
    },
    {
      "epoch": 0.40998674580764133,
      "grad_norm": 0.7710551023483276,
      "learning_rate": 6.668183899949885e-06,
      "loss": 0.7251,
      "step": 14229
    },
    {
      "epoch": 0.4100155592692906,
      "grad_norm": 0.7494319677352905,
      "learning_rate": 6.667744018526829e-06,
      "loss": 0.7308,
      "step": 14230
    },
    {
      "epoch": 0.4100443727309399,
      "grad_norm": 0.7282792925834656,
      "learning_rate": 6.667304122579389e-06,
      "loss": 0.7392,
      "step": 14231
    },
    {
      "epoch": 0.4100731861925892,
      "grad_norm": 0.7337425947189331,
      "learning_rate": 6.666864212111398e-06,
      "loss": 0.7486,
      "step": 14232
    },
    {
      "epoch": 0.41010199965423844,
      "grad_norm": 0.7542322874069214,
      "learning_rate": 6.666424287126685e-06,
      "loss": 0.7233,
      "step": 14233
    },
    {
      "epoch": 0.4101308131158877,
      "grad_norm": 0.7706727385520935,
      "learning_rate": 6.665984347629083e-06,
      "loss": 0.7485,
      "step": 14234
    },
    {
      "epoch": 0.410159626577537,
      "grad_norm": 0.7311323881149292,
      "learning_rate": 6.665544393622423e-06,
      "loss": 0.7255,
      "step": 14235
    },
    {
      "epoch": 0.4101884400391863,
      "grad_norm": 0.7547802925109863,
      "learning_rate": 6.6651044251105355e-06,
      "loss": 0.7275,
      "step": 14236
    },
    {
      "epoch": 0.4102172535008356,
      "grad_norm": 0.7452728152275085,
      "learning_rate": 6.664664442097253e-06,
      "loss": 0.7166,
      "step": 14237
    },
    {
      "epoch": 0.4102460669624849,
      "grad_norm": 0.7782124280929565,
      "learning_rate": 6.664224444586407e-06,
      "loss": 0.7443,
      "step": 14238
    },
    {
      "epoch": 0.4102748804241342,
      "grad_norm": 0.7194359302520752,
      "learning_rate": 6.66378443258183e-06,
      "loss": 0.718,
      "step": 14239
    },
    {
      "epoch": 0.4103036938857834,
      "grad_norm": 0.7598550915718079,
      "learning_rate": 6.663344406087355e-06,
      "loss": 0.7429,
      "step": 14240
    },
    {
      "epoch": 0.4103325073474327,
      "grad_norm": 0.743251383304596,
      "learning_rate": 6.6629043651068115e-06,
      "loss": 0.742,
      "step": 14241
    },
    {
      "epoch": 0.410361320809082,
      "grad_norm": 0.7503008246421814,
      "learning_rate": 6.662464309644034e-06,
      "loss": 0.7286,
      "step": 14242
    },
    {
      "epoch": 0.4103901342707313,
      "grad_norm": 0.7370084524154663,
      "learning_rate": 6.662024239702853e-06,
      "loss": 0.7215,
      "step": 14243
    },
    {
      "epoch": 0.4104189477323806,
      "grad_norm": 0.7606528997421265,
      "learning_rate": 6.6615841552871045e-06,
      "loss": 0.7281,
      "step": 14244
    },
    {
      "epoch": 0.41044776119402987,
      "grad_norm": 0.7569265961647034,
      "learning_rate": 6.661144056400617e-06,
      "loss": 0.7294,
      "step": 14245
    },
    {
      "epoch": 0.41047657465567916,
      "grad_norm": 0.7403921484947205,
      "learning_rate": 6.6607039430472255e-06,
      "loss": 0.7606,
      "step": 14246
    },
    {
      "epoch": 0.4105053881173284,
      "grad_norm": 0.7477627992630005,
      "learning_rate": 6.660263815230763e-06,
      "loss": 0.7276,
      "step": 14247
    },
    {
      "epoch": 0.4105342015789777,
      "grad_norm": 0.7352302670478821,
      "learning_rate": 6.659823672955062e-06,
      "loss": 0.6969,
      "step": 14248
    },
    {
      "epoch": 0.410563015040627,
      "grad_norm": 0.7521499991416931,
      "learning_rate": 6.659383516223956e-06,
      "loss": 0.7275,
      "step": 14249
    },
    {
      "epoch": 0.41059182850227627,
      "grad_norm": 0.7488097548484802,
      "learning_rate": 6.65894334504128e-06,
      "loss": 0.7294,
      "step": 14250
    },
    {
      "epoch": 0.41062064196392556,
      "grad_norm": 0.748032808303833,
      "learning_rate": 6.658503159410864e-06,
      "loss": 0.742,
      "step": 14251
    },
    {
      "epoch": 0.41064945542557485,
      "grad_norm": 0.7719808220863342,
      "learning_rate": 6.658062959336545e-06,
      "loss": 0.7311,
      "step": 14252
    },
    {
      "epoch": 0.4106782688872241,
      "grad_norm": 0.7528712153434753,
      "learning_rate": 6.657622744822154e-06,
      "loss": 0.74,
      "step": 14253
    },
    {
      "epoch": 0.4107070823488734,
      "grad_norm": 0.7600452303886414,
      "learning_rate": 6.657182515871526e-06,
      "loss": 0.7421,
      "step": 14254
    },
    {
      "epoch": 0.41073589581052267,
      "grad_norm": 0.7522567510604858,
      "learning_rate": 6.656742272488494e-06,
      "loss": 0.7322,
      "step": 14255
    },
    {
      "epoch": 0.41076470927217196,
      "grad_norm": 0.7484437227249146,
      "learning_rate": 6.6563020146768945e-06,
      "loss": 0.7356,
      "step": 14256
    },
    {
      "epoch": 0.41079352273382125,
      "grad_norm": 0.7427611947059631,
      "learning_rate": 6.655861742440559e-06,
      "loss": 0.743,
      "step": 14257
    },
    {
      "epoch": 0.41082233619547054,
      "grad_norm": 0.7431588768959045,
      "learning_rate": 6.6554214557833245e-06,
      "loss": 0.7226,
      "step": 14258
    },
    {
      "epoch": 0.41085114965711983,
      "grad_norm": 0.7524802088737488,
      "learning_rate": 6.654981154709022e-06,
      "loss": 0.7224,
      "step": 14259
    },
    {
      "epoch": 0.41087996311876906,
      "grad_norm": 0.7494732141494751,
      "learning_rate": 6.654540839221491e-06,
      "loss": 0.723,
      "step": 14260
    },
    {
      "epoch": 0.41090877658041836,
      "grad_norm": 0.7577069997787476,
      "learning_rate": 6.654100509324559e-06,
      "loss": 0.7113,
      "step": 14261
    },
    {
      "epoch": 0.41093759004206765,
      "grad_norm": 0.7819788455963135,
      "learning_rate": 6.6536601650220675e-06,
      "loss": 0.7301,
      "step": 14262
    },
    {
      "epoch": 0.41096640350371694,
      "grad_norm": 0.7279804944992065,
      "learning_rate": 6.65321980631785e-06,
      "loss": 0.7117,
      "step": 14263
    },
    {
      "epoch": 0.41099521696536623,
      "grad_norm": 0.7504709362983704,
      "learning_rate": 6.652779433215739e-06,
      "loss": 0.7179,
      "step": 14264
    },
    {
      "epoch": 0.4110240304270155,
      "grad_norm": 0.7667988538742065,
      "learning_rate": 6.652339045719571e-06,
      "loss": 0.7225,
      "step": 14265
    },
    {
      "epoch": 0.4110528438886648,
      "grad_norm": 0.7319972515106201,
      "learning_rate": 6.651898643833182e-06,
      "loss": 0.7377,
      "step": 14266
    },
    {
      "epoch": 0.41108165735031404,
      "grad_norm": 0.7212993502616882,
      "learning_rate": 6.651458227560408e-06,
      "loss": 0.7359,
      "step": 14267
    },
    {
      "epoch": 0.41111047081196334,
      "grad_norm": 0.7376965284347534,
      "learning_rate": 6.651017796905082e-06,
      "loss": 0.7407,
      "step": 14268
    },
    {
      "epoch": 0.4111392842736126,
      "grad_norm": 0.7434485554695129,
      "learning_rate": 6.650577351871043e-06,
      "loss": 0.7238,
      "step": 14269
    },
    {
      "epoch": 0.4111680977352619,
      "grad_norm": 0.7747595906257629,
      "learning_rate": 6.650136892462125e-06,
      "loss": 0.718,
      "step": 14270
    },
    {
      "epoch": 0.4111969111969112,
      "grad_norm": 0.7494572401046753,
      "learning_rate": 6.6496964186821635e-06,
      "loss": 0.7636,
      "step": 14271
    },
    {
      "epoch": 0.4112257246585605,
      "grad_norm": 0.771290123462677,
      "learning_rate": 6.649255930534997e-06,
      "loss": 0.7258,
      "step": 14272
    },
    {
      "epoch": 0.4112545381202098,
      "grad_norm": 0.7475396394729614,
      "learning_rate": 6.6488154280244575e-06,
      "loss": 0.7265,
      "step": 14273
    },
    {
      "epoch": 0.411283351581859,
      "grad_norm": 0.7488657236099243,
      "learning_rate": 6.648374911154385e-06,
      "loss": 0.7557,
      "step": 14274
    },
    {
      "epoch": 0.4113121650435083,
      "grad_norm": 0.7520209550857544,
      "learning_rate": 6.6479343799286156e-06,
      "loss": 0.7263,
      "step": 14275
    },
    {
      "epoch": 0.4113409785051576,
      "grad_norm": 0.736788272857666,
      "learning_rate": 6.647493834350984e-06,
      "loss": 0.7361,
      "step": 14276
    },
    {
      "epoch": 0.4113697919668069,
      "grad_norm": 0.7491538524627686,
      "learning_rate": 6.647053274425328e-06,
      "loss": 0.7153,
      "step": 14277
    },
    {
      "epoch": 0.4113986054284562,
      "grad_norm": 0.7349960803985596,
      "learning_rate": 6.6466127001554855e-06,
      "loss": 0.7283,
      "step": 14278
    },
    {
      "epoch": 0.4114274188901055,
      "grad_norm": 0.7484182715415955,
      "learning_rate": 6.64617211154529e-06,
      "loss": 0.739,
      "step": 14279
    },
    {
      "epoch": 0.4114562323517547,
      "grad_norm": 0.931310772895813,
      "learning_rate": 6.645731508598584e-06,
      "loss": 0.7258,
      "step": 14280
    },
    {
      "epoch": 0.411485045813404,
      "grad_norm": 0.7563157677650452,
      "learning_rate": 6.645290891319201e-06,
      "loss": 0.7403,
      "step": 14281
    },
    {
      "epoch": 0.4115138592750533,
      "grad_norm": 0.7245189547538757,
      "learning_rate": 6.644850259710978e-06,
      "loss": 0.7337,
      "step": 14282
    },
    {
      "epoch": 0.4115426727367026,
      "grad_norm": 0.9522839784622192,
      "learning_rate": 6.644409613777754e-06,
      "loss": 0.7247,
      "step": 14283
    },
    {
      "epoch": 0.4115714861983519,
      "grad_norm": 0.7467026710510254,
      "learning_rate": 6.643968953523365e-06,
      "loss": 0.7447,
      "step": 14284
    },
    {
      "epoch": 0.41160029966000117,
      "grad_norm": 0.7264918684959412,
      "learning_rate": 6.643528278951651e-06,
      "loss": 0.729,
      "step": 14285
    },
    {
      "epoch": 0.41162911312165046,
      "grad_norm": 0.7481515407562256,
      "learning_rate": 6.643087590066448e-06,
      "loss": 0.7196,
      "step": 14286
    },
    {
      "epoch": 0.4116579265832997,
      "grad_norm": 0.7198364734649658,
      "learning_rate": 6.642646886871594e-06,
      "loss": 0.7094,
      "step": 14287
    },
    {
      "epoch": 0.411686740044949,
      "grad_norm": 0.7507749199867249,
      "learning_rate": 6.64220616937093e-06,
      "loss": 0.7257,
      "step": 14288
    },
    {
      "epoch": 0.4117155535065983,
      "grad_norm": 0.7591798901557922,
      "learning_rate": 6.641765437568291e-06,
      "loss": 0.7334,
      "step": 14289
    },
    {
      "epoch": 0.41174436696824757,
      "grad_norm": 0.743860125541687,
      "learning_rate": 6.641324691467514e-06,
      "loss": 0.7152,
      "step": 14290
    },
    {
      "epoch": 0.41177318042989686,
      "grad_norm": 0.74799644947052,
      "learning_rate": 6.640883931072441e-06,
      "loss": 0.7399,
      "step": 14291
    },
    {
      "epoch": 0.41180199389154615,
      "grad_norm": 0.7850226759910583,
      "learning_rate": 6.64044315638691e-06,
      "loss": 0.7504,
      "step": 14292
    },
    {
      "epoch": 0.41183080735319544,
      "grad_norm": 0.7719393968582153,
      "learning_rate": 6.640002367414759e-06,
      "loss": 0.7309,
      "step": 14293
    },
    {
      "epoch": 0.4118596208148447,
      "grad_norm": 0.7567669153213501,
      "learning_rate": 6.6395615641598265e-06,
      "loss": 0.713,
      "step": 14294
    },
    {
      "epoch": 0.41188843427649396,
      "grad_norm": 0.7324767112731934,
      "learning_rate": 6.639120746625951e-06,
      "loss": 0.712,
      "step": 14295
    },
    {
      "epoch": 0.41191724773814326,
      "grad_norm": 0.7518460750579834,
      "learning_rate": 6.638679914816974e-06,
      "loss": 0.751,
      "step": 14296
    },
    {
      "epoch": 0.41194606119979255,
      "grad_norm": 0.759214460849762,
      "learning_rate": 6.638239068736731e-06,
      "loss": 0.738,
      "step": 14297
    },
    {
      "epoch": 0.41197487466144184,
      "grad_norm": 0.7570059895515442,
      "learning_rate": 6.637798208389064e-06,
      "loss": 0.7254,
      "step": 14298
    },
    {
      "epoch": 0.4120036881230911,
      "grad_norm": 0.7419716715812683,
      "learning_rate": 6.637357333777812e-06,
      "loss": 0.7068,
      "step": 14299
    },
    {
      "epoch": 0.4120325015847404,
      "grad_norm": 0.7372866868972778,
      "learning_rate": 6.636916444906815e-06,
      "loss": 0.7389,
      "step": 14300
    },
    {
      "epoch": 0.41206131504638965,
      "grad_norm": 0.7820755243301392,
      "learning_rate": 6.636475541779911e-06,
      "loss": 0.7331,
      "step": 14301
    },
    {
      "epoch": 0.41209012850803894,
      "grad_norm": 0.8066464066505432,
      "learning_rate": 6.636034624400942e-06,
      "loss": 0.677,
      "step": 14302
    },
    {
      "epoch": 0.41211894196968824,
      "grad_norm": 0.7777602672576904,
      "learning_rate": 6.635593692773745e-06,
      "loss": 0.739,
      "step": 14303
    },
    {
      "epoch": 0.4121477554313375,
      "grad_norm": 0.7538105249404907,
      "learning_rate": 6.635152746902164e-06,
      "loss": 0.7316,
      "step": 14304
    },
    {
      "epoch": 0.4121765688929868,
      "grad_norm": 0.7778418660163879,
      "learning_rate": 6.634711786790036e-06,
      "loss": 0.7475,
      "step": 14305
    },
    {
      "epoch": 0.4122053823546361,
      "grad_norm": 0.7504875063896179,
      "learning_rate": 6.634270812441201e-06,
      "loss": 0.7463,
      "step": 14306
    },
    {
      "epoch": 0.41223419581628534,
      "grad_norm": 0.7373106479644775,
      "learning_rate": 6.633829823859503e-06,
      "loss": 0.7463,
      "step": 14307
    },
    {
      "epoch": 0.41226300927793463,
      "grad_norm": 0.748153030872345,
      "learning_rate": 6.6333888210487805e-06,
      "loss": 0.7342,
      "step": 14308
    },
    {
      "epoch": 0.4122918227395839,
      "grad_norm": 0.7357035875320435,
      "learning_rate": 6.632947804012874e-06,
      "loss": 0.7227,
      "step": 14309
    },
    {
      "epoch": 0.4123206362012332,
      "grad_norm": 0.7301989793777466,
      "learning_rate": 6.632506772755624e-06,
      "loss": 0.7012,
      "step": 14310
    },
    {
      "epoch": 0.4123494496628825,
      "grad_norm": 0.7500131726264954,
      "learning_rate": 6.632065727280872e-06,
      "loss": 0.742,
      "step": 14311
    },
    {
      "epoch": 0.4123782631245318,
      "grad_norm": 0.7305372357368469,
      "learning_rate": 6.63162466759246e-06,
      "loss": 0.7085,
      "step": 14312
    },
    {
      "epoch": 0.4124070765861811,
      "grad_norm": 0.7287654280662537,
      "learning_rate": 6.631183593694227e-06,
      "loss": 0.709,
      "step": 14313
    },
    {
      "epoch": 0.4124358900478303,
      "grad_norm": 0.7281307578086853,
      "learning_rate": 6.6307425055900154e-06,
      "loss": 0.7336,
      "step": 14314
    },
    {
      "epoch": 0.4124647035094796,
      "grad_norm": 0.7332679629325867,
      "learning_rate": 6.630301403283666e-06,
      "loss": 0.6956,
      "step": 14315
    },
    {
      "epoch": 0.4124935169711289,
      "grad_norm": 0.7541736364364624,
      "learning_rate": 6.629860286779023e-06,
      "loss": 0.7474,
      "step": 14316
    },
    {
      "epoch": 0.4125223304327782,
      "grad_norm": 0.7545510530471802,
      "learning_rate": 6.629419156079926e-06,
      "loss": 0.7265,
      "step": 14317
    },
    {
      "epoch": 0.4125511438944275,
      "grad_norm": 0.7382469177246094,
      "learning_rate": 6.628978011190217e-06,
      "loss": 0.7388,
      "step": 14318
    },
    {
      "epoch": 0.4125799573560768,
      "grad_norm": 0.783026933670044,
      "learning_rate": 6.6285368521137375e-06,
      "loss": 0.729,
      "step": 14319
    },
    {
      "epoch": 0.41260877081772607,
      "grad_norm": 0.7795729637145996,
      "learning_rate": 6.628095678854329e-06,
      "loss": 0.7571,
      "step": 14320
    },
    {
      "epoch": 0.4126375842793753,
      "grad_norm": 0.7473739385604858,
      "learning_rate": 6.627654491415835e-06,
      "loss": 0.755,
      "step": 14321
    },
    {
      "epoch": 0.4126663977410246,
      "grad_norm": 0.7494299411773682,
      "learning_rate": 6.627213289802098e-06,
      "loss": 0.717,
      "step": 14322
    },
    {
      "epoch": 0.4126952112026739,
      "grad_norm": 0.7658362984657288,
      "learning_rate": 6.626772074016959e-06,
      "loss": 0.7251,
      "step": 14323
    },
    {
      "epoch": 0.4127240246643232,
      "grad_norm": 0.7848718762397766,
      "learning_rate": 6.626330844064263e-06,
      "loss": 0.7313,
      "step": 14324
    },
    {
      "epoch": 0.41275283812597247,
      "grad_norm": 0.7610113620758057,
      "learning_rate": 6.625889599947849e-06,
      "loss": 0.7264,
      "step": 14325
    },
    {
      "epoch": 0.41278165158762176,
      "grad_norm": 0.7733773589134216,
      "learning_rate": 6.625448341671563e-06,
      "loss": 0.7551,
      "step": 14326
    },
    {
      "epoch": 0.41281046504927105,
      "grad_norm": 0.7482337355613708,
      "learning_rate": 6.625007069239246e-06,
      "loss": 0.7705,
      "step": 14327
    },
    {
      "epoch": 0.4128392785109203,
      "grad_norm": 0.7430883646011353,
      "learning_rate": 6.624565782654744e-06,
      "loss": 0.7519,
      "step": 14328
    },
    {
      "epoch": 0.4128680919725696,
      "grad_norm": 0.7367554306983948,
      "learning_rate": 6.624124481921896e-06,
      "loss": 0.7377,
      "step": 14329
    },
    {
      "epoch": 0.41289690543421886,
      "grad_norm": 0.7343851923942566,
      "learning_rate": 6.623683167044548e-06,
      "loss": 0.7088,
      "step": 14330
    },
    {
      "epoch": 0.41292571889586815,
      "grad_norm": 0.749998152256012,
      "learning_rate": 6.6232418380265425e-06,
      "loss": 0.7278,
      "step": 14331
    },
    {
      "epoch": 0.41295453235751745,
      "grad_norm": 0.7698357105255127,
      "learning_rate": 6.622800494871724e-06,
      "loss": 0.7366,
      "step": 14332
    },
    {
      "epoch": 0.41298334581916674,
      "grad_norm": 0.7485611438751221,
      "learning_rate": 6.622359137583933e-06,
      "loss": 0.7522,
      "step": 14333
    },
    {
      "epoch": 0.41301215928081597,
      "grad_norm": 0.7408976554870605,
      "learning_rate": 6.621917766167018e-06,
      "loss": 0.7164,
      "step": 14334
    },
    {
      "epoch": 0.41304097274246526,
      "grad_norm": 0.7601205110549927,
      "learning_rate": 6.621476380624821e-06,
      "loss": 0.7496,
      "step": 14335
    },
    {
      "epoch": 0.41306978620411455,
      "grad_norm": 0.7758787274360657,
      "learning_rate": 6.621034980961185e-06,
      "loss": 0.7414,
      "step": 14336
    },
    {
      "epoch": 0.41309859966576384,
      "grad_norm": 0.7604936361312866,
      "learning_rate": 6.6205935671799545e-06,
      "loss": 0.7232,
      "step": 14337
    },
    {
      "epoch": 0.41312741312741313,
      "grad_norm": 0.7914555072784424,
      "learning_rate": 6.620152139284974e-06,
      "loss": 0.7446,
      "step": 14338
    },
    {
      "epoch": 0.4131562265890624,
      "grad_norm": 0.7901769280433655,
      "learning_rate": 6.619710697280087e-06,
      "loss": 0.7518,
      "step": 14339
    },
    {
      "epoch": 0.4131850400507117,
      "grad_norm": 0.75748211145401,
      "learning_rate": 6.619269241169141e-06,
      "loss": 0.7165,
      "step": 14340
    },
    {
      "epoch": 0.41321385351236095,
      "grad_norm": 0.7423606514930725,
      "learning_rate": 6.618827770955976e-06,
      "loss": 0.7117,
      "step": 14341
    },
    {
      "epoch": 0.41324266697401024,
      "grad_norm": 0.7574731707572937,
      "learning_rate": 6.618386286644442e-06,
      "loss": 0.752,
      "step": 14342
    },
    {
      "epoch": 0.41327148043565953,
      "grad_norm": 0.7685872912406921,
      "learning_rate": 6.61794478823838e-06,
      "loss": 0.7347,
      "step": 14343
    },
    {
      "epoch": 0.4133002938973088,
      "grad_norm": 0.7526795268058777,
      "learning_rate": 6.617503275741636e-06,
      "loss": 0.7204,
      "step": 14344
    },
    {
      "epoch": 0.4133291073589581,
      "grad_norm": 0.7575008273124695,
      "learning_rate": 6.617061749158056e-06,
      "loss": 0.7264,
      "step": 14345
    },
    {
      "epoch": 0.4133579208206074,
      "grad_norm": 0.7402541041374207,
      "learning_rate": 6.616620208491482e-06,
      "loss": 0.7426,
      "step": 14346
    },
    {
      "epoch": 0.4133867342822567,
      "grad_norm": 0.7774802446365356,
      "learning_rate": 6.616178653745765e-06,
      "loss": 0.7071,
      "step": 14347
    },
    {
      "epoch": 0.41341554774390593,
      "grad_norm": 0.7498378753662109,
      "learning_rate": 6.615737084924747e-06,
      "loss": 0.7628,
      "step": 14348
    },
    {
      "epoch": 0.4134443612055552,
      "grad_norm": 0.7534478306770325,
      "learning_rate": 6.615295502032274e-06,
      "loss": 0.7372,
      "step": 14349
    },
    {
      "epoch": 0.4134731746672045,
      "grad_norm": 0.7426822185516357,
      "learning_rate": 6.614853905072192e-06,
      "loss": 0.7499,
      "step": 14350
    },
    {
      "epoch": 0.4135019881288538,
      "grad_norm": 0.7388256192207336,
      "learning_rate": 6.614412294048346e-06,
      "loss": 0.7326,
      "step": 14351
    },
    {
      "epoch": 0.4135308015905031,
      "grad_norm": 0.7472388744354248,
      "learning_rate": 6.613970668964583e-06,
      "loss": 0.7411,
      "step": 14352
    },
    {
      "epoch": 0.4135596150521524,
      "grad_norm": 0.7753364443778992,
      "learning_rate": 6.6135290298247505e-06,
      "loss": 0.713,
      "step": 14353
    },
    {
      "epoch": 0.4135884285138017,
      "grad_norm": 0.7465027570724487,
      "learning_rate": 6.613087376632691e-06,
      "loss": 0.7288,
      "step": 14354
    },
    {
      "epoch": 0.4136172419754509,
      "grad_norm": 0.7494297027587891,
      "learning_rate": 6.612645709392254e-06,
      "loss": 0.7363,
      "step": 14355
    },
    {
      "epoch": 0.4136460554371002,
      "grad_norm": 0.7465082406997681,
      "learning_rate": 6.612204028107285e-06,
      "loss": 0.7398,
      "step": 14356
    },
    {
      "epoch": 0.4136748688987495,
      "grad_norm": 0.7940166592597961,
      "learning_rate": 6.611762332781629e-06,
      "loss": 0.7344,
      "step": 14357
    },
    {
      "epoch": 0.4137036823603988,
      "grad_norm": 0.7167472839355469,
      "learning_rate": 6.611320623419135e-06,
      "loss": 0.7154,
      "step": 14358
    },
    {
      "epoch": 0.4137324958220481,
      "grad_norm": 0.7392259836196899,
      "learning_rate": 6.610878900023651e-06,
      "loss": 0.7387,
      "step": 14359
    },
    {
      "epoch": 0.41376130928369736,
      "grad_norm": 0.7789989709854126,
      "learning_rate": 6.61043716259902e-06,
      "loss": 0.724,
      "step": 14360
    },
    {
      "epoch": 0.4137901227453466,
      "grad_norm": 0.8118203282356262,
      "learning_rate": 6.609995411149093e-06,
      "loss": 0.7385,
      "step": 14361
    },
    {
      "epoch": 0.4138189362069959,
      "grad_norm": 0.7710491418838501,
      "learning_rate": 6.609553645677713e-06,
      "loss": 0.7358,
      "step": 14362
    },
    {
      "epoch": 0.4138477496686452,
      "grad_norm": 0.7756072282791138,
      "learning_rate": 6.609111866188731e-06,
      "loss": 0.7433,
      "step": 14363
    },
    {
      "epoch": 0.4138765631302945,
      "grad_norm": 0.7656488418579102,
      "learning_rate": 6.608670072685992e-06,
      "loss": 0.7378,
      "step": 14364
    },
    {
      "epoch": 0.41390537659194376,
      "grad_norm": 0.72762531042099,
      "learning_rate": 6.608228265173345e-06,
      "loss": 0.7316,
      "step": 14365
    },
    {
      "epoch": 0.41393419005359305,
      "grad_norm": 0.7465712428092957,
      "learning_rate": 6.607786443654639e-06,
      "loss": 0.7531,
      "step": 14366
    },
    {
      "epoch": 0.41396300351524234,
      "grad_norm": 0.7359187006950378,
      "learning_rate": 6.607344608133719e-06,
      "loss": 0.7375,
      "step": 14367
    },
    {
      "epoch": 0.4139918169768916,
      "grad_norm": 0.7417243123054504,
      "learning_rate": 6.6069027586144354e-06,
      "loss": 0.7153,
      "step": 14368
    },
    {
      "epoch": 0.41402063043854087,
      "grad_norm": 0.7336434125900269,
      "learning_rate": 6.606460895100634e-06,
      "loss": 0.7417,
      "step": 14369
    },
    {
      "epoch": 0.41404944390019016,
      "grad_norm": 0.7397353053092957,
      "learning_rate": 6.606019017596164e-06,
      "loss": 0.7512,
      "step": 14370
    },
    {
      "epoch": 0.41407825736183945,
      "grad_norm": 0.762209951877594,
      "learning_rate": 6.605577126104874e-06,
      "loss": 0.7332,
      "step": 14371
    },
    {
      "epoch": 0.41410707082348874,
      "grad_norm": 0.7858898043632507,
      "learning_rate": 6.605135220630613e-06,
      "loss": 0.7698,
      "step": 14372
    },
    {
      "epoch": 0.41413588428513803,
      "grad_norm": 0.7762240767478943,
      "learning_rate": 6.604693301177229e-06,
      "loss": 0.7575,
      "step": 14373
    },
    {
      "epoch": 0.4141646977467873,
      "grad_norm": 0.7642592787742615,
      "learning_rate": 6.6042513677485695e-06,
      "loss": 0.7452,
      "step": 14374
    },
    {
      "epoch": 0.41419351120843656,
      "grad_norm": 0.7413358688354492,
      "learning_rate": 6.603809420348484e-06,
      "loss": 0.7505,
      "step": 14375
    },
    {
      "epoch": 0.41422232467008585,
      "grad_norm": 0.764811635017395,
      "learning_rate": 6.6033674589808214e-06,
      "loss": 0.7273,
      "step": 14376
    },
    {
      "epoch": 0.41425113813173514,
      "grad_norm": 0.7592869400978088,
      "learning_rate": 6.602925483649433e-06,
      "loss": 0.7157,
      "step": 14377
    },
    {
      "epoch": 0.41427995159338443,
      "grad_norm": 0.7713355422019958,
      "learning_rate": 6.602483494358164e-06,
      "loss": 0.7421,
      "step": 14378
    },
    {
      "epoch": 0.4143087650550337,
      "grad_norm": 0.7457348108291626,
      "learning_rate": 6.602041491110867e-06,
      "loss": 0.7326,
      "step": 14379
    },
    {
      "epoch": 0.414337578516683,
      "grad_norm": 0.768741250038147,
      "learning_rate": 6.601599473911389e-06,
      "loss": 0.7382,
      "step": 14380
    },
    {
      "epoch": 0.41436639197833225,
      "grad_norm": 0.7734686732292175,
      "learning_rate": 6.601157442763581e-06,
      "loss": 0.7189,
      "step": 14381
    },
    {
      "epoch": 0.41439520543998154,
      "grad_norm": 0.770772397518158,
      "learning_rate": 6.600715397671292e-06,
      "loss": 0.7314,
      "step": 14382
    },
    {
      "epoch": 0.41442401890163083,
      "grad_norm": 0.76553875207901,
      "learning_rate": 6.600273338638372e-06,
      "loss": 0.7253,
      "step": 14383
    },
    {
      "epoch": 0.4144528323632801,
      "grad_norm": 0.7574799060821533,
      "learning_rate": 6.59983126566867e-06,
      "loss": 0.7129,
      "step": 14384
    },
    {
      "epoch": 0.4144816458249294,
      "grad_norm": 0.7582066655158997,
      "learning_rate": 6.599389178766039e-06,
      "loss": 0.7172,
      "step": 14385
    },
    {
      "epoch": 0.4145104592865787,
      "grad_norm": 0.7769907116889954,
      "learning_rate": 6.5989470779343265e-06,
      "loss": 0.7276,
      "step": 14386
    },
    {
      "epoch": 0.414539272748228,
      "grad_norm": 0.7935575842857361,
      "learning_rate": 6.598504963177382e-06,
      "loss": 0.7618,
      "step": 14387
    },
    {
      "epoch": 0.41456808620987723,
      "grad_norm": 0.7232626676559448,
      "learning_rate": 6.598062834499058e-06,
      "loss": 0.7341,
      "step": 14388
    },
    {
      "epoch": 0.4145968996715265,
      "grad_norm": 0.7769969701766968,
      "learning_rate": 6.597620691903205e-06,
      "loss": 0.7378,
      "step": 14389
    },
    {
      "epoch": 0.4146257131331758,
      "grad_norm": 0.7859870195388794,
      "learning_rate": 6.597178535393672e-06,
      "loss": 0.6981,
      "step": 14390
    },
    {
      "epoch": 0.4146545265948251,
      "grad_norm": 0.7631061673164368,
      "learning_rate": 6.596736364974311e-06,
      "loss": 0.7102,
      "step": 14391
    },
    {
      "epoch": 0.4146833400564744,
      "grad_norm": 0.7906966805458069,
      "learning_rate": 6.596294180648972e-06,
      "loss": 0.7285,
      "step": 14392
    },
    {
      "epoch": 0.4147121535181237,
      "grad_norm": 0.7591965198516846,
      "learning_rate": 6.595851982421506e-06,
      "loss": 0.7361,
      "step": 14393
    },
    {
      "epoch": 0.414740966979773,
      "grad_norm": 0.7382796406745911,
      "learning_rate": 6.595409770295765e-06,
      "loss": 0.7196,
      "step": 14394
    },
    {
      "epoch": 0.4147697804414222,
      "grad_norm": 0.7558622360229492,
      "learning_rate": 6.594967544275601e-06,
      "loss": 0.7173,
      "step": 14395
    },
    {
      "epoch": 0.4147985939030715,
      "grad_norm": 0.7359570264816284,
      "learning_rate": 6.5945253043648624e-06,
      "loss": 0.7463,
      "step": 14396
    },
    {
      "epoch": 0.4148274073647208,
      "grad_norm": 0.7820096015930176,
      "learning_rate": 6.594083050567404e-06,
      "loss": 0.7267,
      "step": 14397
    },
    {
      "epoch": 0.4148562208263701,
      "grad_norm": 0.7596745491027832,
      "learning_rate": 6.5936407828870745e-06,
      "loss": 0.749,
      "step": 14398
    },
    {
      "epoch": 0.41488503428801937,
      "grad_norm": 0.7265925407409668,
      "learning_rate": 6.593198501327726e-06,
      "loss": 0.7426,
      "step": 14399
    },
    {
      "epoch": 0.41491384774966866,
      "grad_norm": 0.7543421983718872,
      "learning_rate": 6.5927562058932116e-06,
      "loss": 0.7272,
      "step": 14400
    },
    {
      "epoch": 0.41494266121131795,
      "grad_norm": 0.7587100267410278,
      "learning_rate": 6.5923138965873834e-06,
      "loss": 0.7457,
      "step": 14401
    },
    {
      "epoch": 0.4149714746729672,
      "grad_norm": 0.7437349557876587,
      "learning_rate": 6.591871573414093e-06,
      "loss": 0.7213,
      "step": 14402
    },
    {
      "epoch": 0.4150002881346165,
      "grad_norm": 0.7414727807044983,
      "learning_rate": 6.591429236377191e-06,
      "loss": 0.7258,
      "step": 14403
    },
    {
      "epoch": 0.41502910159626577,
      "grad_norm": 0.7511314153671265,
      "learning_rate": 6.590986885480533e-06,
      "loss": 0.7203,
      "step": 14404
    },
    {
      "epoch": 0.41505791505791506,
      "grad_norm": 0.7090073227882385,
      "learning_rate": 6.5905445207279685e-06,
      "loss": 0.6955,
      "step": 14405
    },
    {
      "epoch": 0.41508672851956435,
      "grad_norm": 0.7624993920326233,
      "learning_rate": 6.590102142123351e-06,
      "loss": 0.7265,
      "step": 14406
    },
    {
      "epoch": 0.41511554198121364,
      "grad_norm": 0.7931844592094421,
      "learning_rate": 6.589659749670535e-06,
      "loss": 0.7305,
      "step": 14407
    },
    {
      "epoch": 0.4151443554428629,
      "grad_norm": 0.7706127762794495,
      "learning_rate": 6.589217343373371e-06,
      "loss": 0.7458,
      "step": 14408
    },
    {
      "epoch": 0.41517316890451217,
      "grad_norm": 0.7233623266220093,
      "learning_rate": 6.5887749232357124e-06,
      "loss": 0.7337,
      "step": 14409
    },
    {
      "epoch": 0.41520198236616146,
      "grad_norm": 0.7674185037612915,
      "learning_rate": 6.588332489261412e-06,
      "loss": 0.7344,
      "step": 14410
    },
    {
      "epoch": 0.41523079582781075,
      "grad_norm": 0.7303096055984497,
      "learning_rate": 6.587890041454324e-06,
      "loss": 0.7477,
      "step": 14411
    },
    {
      "epoch": 0.41525960928946004,
      "grad_norm": 0.7291093468666077,
      "learning_rate": 6.5874475798183005e-06,
      "loss": 0.6979,
      "step": 14412
    },
    {
      "epoch": 0.41528842275110933,
      "grad_norm": 0.75597083568573,
      "learning_rate": 6.587005104357196e-06,
      "loss": 0.7426,
      "step": 14413
    },
    {
      "epoch": 0.4153172362127586,
      "grad_norm": 0.7487912774085999,
      "learning_rate": 6.586562615074863e-06,
      "loss": 0.7151,
      "step": 14414
    },
    {
      "epoch": 0.41534604967440786,
      "grad_norm": 0.7680459022521973,
      "learning_rate": 6.586120111975156e-06,
      "loss": 0.7401,
      "step": 14415
    },
    {
      "epoch": 0.41537486313605715,
      "grad_norm": 0.7722229361534119,
      "learning_rate": 6.585677595061928e-06,
      "loss": 0.712,
      "step": 14416
    },
    {
      "epoch": 0.41540367659770644,
      "grad_norm": 0.7674531936645508,
      "learning_rate": 6.585235064339033e-06,
      "loss": 0.7304,
      "step": 14417
    },
    {
      "epoch": 0.41543249005935573,
      "grad_norm": 0.7680970430374146,
      "learning_rate": 6.584792519810326e-06,
      "loss": 0.7297,
      "step": 14418
    },
    {
      "epoch": 0.415461303521005,
      "grad_norm": 0.7324626445770264,
      "learning_rate": 6.584349961479659e-06,
      "loss": 0.7004,
      "step": 14419
    },
    {
      "epoch": 0.4154901169826543,
      "grad_norm": 0.7926222085952759,
      "learning_rate": 6.5839073893508895e-06,
      "loss": 0.7415,
      "step": 14420
    },
    {
      "epoch": 0.4155189304443036,
      "grad_norm": 0.7481798529624939,
      "learning_rate": 6.5834648034278694e-06,
      "loss": 0.7089,
      "step": 14421
    },
    {
      "epoch": 0.41554774390595284,
      "grad_norm": 0.7461360692977905,
      "learning_rate": 6.583022203714452e-06,
      "loss": 0.7607,
      "step": 14422
    },
    {
      "epoch": 0.41557655736760213,
      "grad_norm": 0.7455697059631348,
      "learning_rate": 6.582579590214495e-06,
      "loss": 0.7436,
      "step": 14423
    },
    {
      "epoch": 0.4156053708292514,
      "grad_norm": 0.7637297511100769,
      "learning_rate": 6.582136962931851e-06,
      "loss": 0.7336,
      "step": 14424
    },
    {
      "epoch": 0.4156341842909007,
      "grad_norm": 0.737547755241394,
      "learning_rate": 6.581694321870378e-06,
      "loss": 0.7344,
      "step": 14425
    },
    {
      "epoch": 0.41566299775255,
      "grad_norm": 0.7545158267021179,
      "learning_rate": 6.581251667033926e-06,
      "loss": 0.7524,
      "step": 14426
    },
    {
      "epoch": 0.4156918112141993,
      "grad_norm": 0.7729477286338806,
      "learning_rate": 6.580808998426354e-06,
      "loss": 0.7012,
      "step": 14427
    },
    {
      "epoch": 0.4157206246758486,
      "grad_norm": 0.7386178374290466,
      "learning_rate": 6.5803663160515165e-06,
      "loss": 0.7416,
      "step": 14428
    },
    {
      "epoch": 0.4157494381374978,
      "grad_norm": 0.7341049909591675,
      "learning_rate": 6.5799236199132655e-06,
      "loss": 0.7189,
      "step": 14429
    },
    {
      "epoch": 0.4157782515991471,
      "grad_norm": 0.7803229689598083,
      "learning_rate": 6.579480910015461e-06,
      "loss": 0.7333,
      "step": 14430
    },
    {
      "epoch": 0.4158070650607964,
      "grad_norm": 0.7359032034873962,
      "learning_rate": 6.579038186361957e-06,
      "loss": 0.7332,
      "step": 14431
    },
    {
      "epoch": 0.4158358785224457,
      "grad_norm": 0.7462955117225647,
      "learning_rate": 6.578595448956608e-06,
      "loss": 0.7164,
      "step": 14432
    },
    {
      "epoch": 0.415864691984095,
      "grad_norm": 0.7955107688903809,
      "learning_rate": 6.578152697803272e-06,
      "loss": 0.7517,
      "step": 14433
    },
    {
      "epoch": 0.41589350544574427,
      "grad_norm": 0.7647842764854431,
      "learning_rate": 6.577709932905802e-06,
      "loss": 0.7523,
      "step": 14434
    },
    {
      "epoch": 0.4159223189073935,
      "grad_norm": 0.7327582836151123,
      "learning_rate": 6.577267154268055e-06,
      "loss": 0.7271,
      "step": 14435
    },
    {
      "epoch": 0.4159511323690428,
      "grad_norm": 0.7348340749740601,
      "learning_rate": 6.57682436189389e-06,
      "loss": 0.7156,
      "step": 14436
    },
    {
      "epoch": 0.4159799458306921,
      "grad_norm": 0.7369200587272644,
      "learning_rate": 6.57638155578716e-06,
      "loss": 0.7222,
      "step": 14437
    },
    {
      "epoch": 0.4160087592923414,
      "grad_norm": 0.7429260015487671,
      "learning_rate": 6.5759387359517224e-06,
      "loss": 0.755,
      "step": 14438
    },
    {
      "epoch": 0.41603757275399067,
      "grad_norm": 0.7587293386459351,
      "learning_rate": 6.575495902391435e-06,
      "loss": 0.7277,
      "step": 14439
    },
    {
      "epoch": 0.41606638621563996,
      "grad_norm": 0.7415571212768555,
      "learning_rate": 6.575053055110152e-06,
      "loss": 0.7218,
      "step": 14440
    },
    {
      "epoch": 0.41609519967728925,
      "grad_norm": 0.765501856803894,
      "learning_rate": 6.5746101941117305e-06,
      "loss": 0.7293,
      "step": 14441
    },
    {
      "epoch": 0.4161240131389385,
      "grad_norm": 0.7369219660758972,
      "learning_rate": 6.5741673194000285e-06,
      "loss": 0.753,
      "step": 14442
    },
    {
      "epoch": 0.4161528266005878,
      "grad_norm": 0.744978129863739,
      "learning_rate": 6.573724430978903e-06,
      "loss": 0.7462,
      "step": 14443
    },
    {
      "epoch": 0.41618164006223707,
      "grad_norm": 0.757867693901062,
      "learning_rate": 6.573281528852211e-06,
      "loss": 0.7435,
      "step": 14444
    },
    {
      "epoch": 0.41621045352388636,
      "grad_norm": 0.7818431258201599,
      "learning_rate": 6.572838613023811e-06,
      "loss": 0.7271,
      "step": 14445
    },
    {
      "epoch": 0.41623926698553565,
      "grad_norm": 0.7431662082672119,
      "learning_rate": 6.572395683497558e-06,
      "loss": 0.7312,
      "step": 14446
    },
    {
      "epoch": 0.41626808044718494,
      "grad_norm": 0.7407742738723755,
      "learning_rate": 6.5719527402773084e-06,
      "loss": 0.7327,
      "step": 14447
    },
    {
      "epoch": 0.41629689390883423,
      "grad_norm": 0.7451844215393066,
      "learning_rate": 6.571509783366925e-06,
      "loss": 0.7154,
      "step": 14448
    },
    {
      "epoch": 0.41632570737048347,
      "grad_norm": 0.7604103684425354,
      "learning_rate": 6.57106681277026e-06,
      "loss": 0.7521,
      "step": 14449
    },
    {
      "epoch": 0.41635452083213276,
      "grad_norm": 0.7614752650260925,
      "learning_rate": 6.570623828491175e-06,
      "loss": 0.7477,
      "step": 14450
    },
    {
      "epoch": 0.41638333429378205,
      "grad_norm": 0.7524944543838501,
      "learning_rate": 6.570180830533526e-06,
      "loss": 0.7392,
      "step": 14451
    },
    {
      "epoch": 0.41641214775543134,
      "grad_norm": 0.7492449283599854,
      "learning_rate": 6.569737818901172e-06,
      "loss": 0.7289,
      "step": 14452
    },
    {
      "epoch": 0.41644096121708063,
      "grad_norm": 0.7552488446235657,
      "learning_rate": 6.569294793597969e-06,
      "loss": 0.6925,
      "step": 14453
    },
    {
      "epoch": 0.4164697746787299,
      "grad_norm": 0.7591588497161865,
      "learning_rate": 6.5688517546277785e-06,
      "loss": 0.7207,
      "step": 14454
    },
    {
      "epoch": 0.4164985881403792,
      "grad_norm": 0.73388671875,
      "learning_rate": 6.568408701994459e-06,
      "loss": 0.7378,
      "step": 14455
    },
    {
      "epoch": 0.41652740160202845,
      "grad_norm": 0.7422863245010376,
      "learning_rate": 6.5679656357018655e-06,
      "loss": 0.7318,
      "step": 14456
    },
    {
      "epoch": 0.41655621506367774,
      "grad_norm": 0.7573275566101074,
      "learning_rate": 6.567522555753859e-06,
      "loss": 0.7318,
      "step": 14457
    },
    {
      "epoch": 0.41658502852532703,
      "grad_norm": 0.7329188585281372,
      "learning_rate": 6.567079462154298e-06,
      "loss": 0.7264,
      "step": 14458
    },
    {
      "epoch": 0.4166138419869763,
      "grad_norm": 0.7339499592781067,
      "learning_rate": 6.566636354907041e-06,
      "loss": 0.6971,
      "step": 14459
    },
    {
      "epoch": 0.4166426554486256,
      "grad_norm": 0.738288402557373,
      "learning_rate": 6.5661932340159485e-06,
      "loss": 0.7034,
      "step": 14460
    },
    {
      "epoch": 0.4166714689102749,
      "grad_norm": 0.7756058573722839,
      "learning_rate": 6.5657500994848775e-06,
      "loss": 0.7297,
      "step": 14461
    },
    {
      "epoch": 0.41670028237192414,
      "grad_norm": 0.7477059364318848,
      "learning_rate": 6.565306951317688e-06,
      "loss": 0.7094,
      "step": 14462
    },
    {
      "epoch": 0.4167290958335734,
      "grad_norm": 0.7690911889076233,
      "learning_rate": 6.564863789518241e-06,
      "loss": 0.7149,
      "step": 14463
    },
    {
      "epoch": 0.4167579092952227,
      "grad_norm": 0.7852935791015625,
      "learning_rate": 6.564420614090394e-06,
      "loss": 0.7343,
      "step": 14464
    },
    {
      "epoch": 0.416786722756872,
      "grad_norm": 0.7499577403068542,
      "learning_rate": 6.563977425038007e-06,
      "loss": 0.7329,
      "step": 14465
    },
    {
      "epoch": 0.4168155362185213,
      "grad_norm": 0.7836015820503235,
      "learning_rate": 6.563534222364941e-06,
      "loss": 0.7463,
      "step": 14466
    },
    {
      "epoch": 0.4168443496801706,
      "grad_norm": 0.7461283802986145,
      "learning_rate": 6.563091006075054e-06,
      "loss": 0.7173,
      "step": 14467
    },
    {
      "epoch": 0.4168731631418199,
      "grad_norm": 0.7236834764480591,
      "learning_rate": 6.562647776172209e-06,
      "loss": 0.7385,
      "step": 14468
    },
    {
      "epoch": 0.4169019766034691,
      "grad_norm": 0.7328738570213318,
      "learning_rate": 6.562204532660261e-06,
      "loss": 0.7523,
      "step": 14469
    },
    {
      "epoch": 0.4169307900651184,
      "grad_norm": 0.7320482730865479,
      "learning_rate": 6.561761275543074e-06,
      "loss": 0.7141,
      "step": 14470
    },
    {
      "epoch": 0.4169596035267677,
      "grad_norm": 0.7453826069831848,
      "learning_rate": 6.561318004824508e-06,
      "loss": 0.7265,
      "step": 14471
    },
    {
      "epoch": 0.416988416988417,
      "grad_norm": 0.7465415000915527,
      "learning_rate": 6.560874720508423e-06,
      "loss": 0.7386,
      "step": 14472
    },
    {
      "epoch": 0.4170172304500663,
      "grad_norm": 0.7428856492042542,
      "learning_rate": 6.560431422598678e-06,
      "loss": 0.7359,
      "step": 14473
    },
    {
      "epoch": 0.41704604391171557,
      "grad_norm": 0.7228876948356628,
      "learning_rate": 6.559988111099138e-06,
      "loss": 0.7127,
      "step": 14474
    },
    {
      "epoch": 0.41707485737336486,
      "grad_norm": 0.772635281085968,
      "learning_rate": 6.559544786013659e-06,
      "loss": 0.7556,
      "step": 14475
    },
    {
      "epoch": 0.4171036708350141,
      "grad_norm": 0.7150520086288452,
      "learning_rate": 6.559101447346103e-06,
      "loss": 0.736,
      "step": 14476
    },
    {
      "epoch": 0.4171324842966634,
      "grad_norm": 0.7504207491874695,
      "learning_rate": 6.558658095100334e-06,
      "loss": 0.7115,
      "step": 14477
    },
    {
      "epoch": 0.4171612977583127,
      "grad_norm": 0.7234876155853271,
      "learning_rate": 6.55821472928021e-06,
      "loss": 0.7287,
      "step": 14478
    },
    {
      "epoch": 0.41719011121996197,
      "grad_norm": 0.7212435007095337,
      "learning_rate": 6.557771349889595e-06,
      "loss": 0.7204,
      "step": 14479
    },
    {
      "epoch": 0.41721892468161126,
      "grad_norm": 0.7157378196716309,
      "learning_rate": 6.557327956932347e-06,
      "loss": 0.7001,
      "step": 14480
    },
    {
      "epoch": 0.41724773814326055,
      "grad_norm": 0.7652549743652344,
      "learning_rate": 6.556884550412328e-06,
      "loss": 0.7339,
      "step": 14481
    },
    {
      "epoch": 0.41727655160490984,
      "grad_norm": 0.7535755038261414,
      "learning_rate": 6.556441130333403e-06,
      "loss": 0.7359,
      "step": 14482
    },
    {
      "epoch": 0.4173053650665591,
      "grad_norm": 0.7647430896759033,
      "learning_rate": 6.555997696699432e-06,
      "loss": 0.7077,
      "step": 14483
    },
    {
      "epoch": 0.41733417852820837,
      "grad_norm": 0.7705483436584473,
      "learning_rate": 6.555554249514274e-06,
      "loss": 0.7432,
      "step": 14484
    },
    {
      "epoch": 0.41736299198985766,
      "grad_norm": 0.7469948530197144,
      "learning_rate": 6.555110788781795e-06,
      "loss": 0.727,
      "step": 14485
    },
    {
      "epoch": 0.41739180545150695,
      "grad_norm": 0.7510130405426025,
      "learning_rate": 6.554667314505856e-06,
      "loss": 0.7246,
      "step": 14486
    },
    {
      "epoch": 0.41742061891315624,
      "grad_norm": 0.7356817126274109,
      "learning_rate": 6.554223826690319e-06,
      "loss": 0.7174,
      "step": 14487
    },
    {
      "epoch": 0.41744943237480553,
      "grad_norm": 0.7247133851051331,
      "learning_rate": 6.553780325339046e-06,
      "loss": 0.7283,
      "step": 14488
    },
    {
      "epoch": 0.41747824583645476,
      "grad_norm": 0.7722261548042297,
      "learning_rate": 6.5533368104558985e-06,
      "loss": 0.7197,
      "step": 14489
    },
    {
      "epoch": 0.41750705929810406,
      "grad_norm": 0.7532077431678772,
      "learning_rate": 6.552893282044741e-06,
      "loss": 0.7354,
      "step": 14490
    },
    {
      "epoch": 0.41753587275975335,
      "grad_norm": 0.7954533696174622,
      "learning_rate": 6.552449740109435e-06,
      "loss": 0.748,
      "step": 14491
    },
    {
      "epoch": 0.41756468622140264,
      "grad_norm": 0.7589174509048462,
      "learning_rate": 6.552006184653844e-06,
      "loss": 0.7453,
      "step": 14492
    },
    {
      "epoch": 0.4175934996830519,
      "grad_norm": 0.743971586227417,
      "learning_rate": 6.55156261568183e-06,
      "loss": 0.7194,
      "step": 14493
    },
    {
      "epoch": 0.4176223131447012,
      "grad_norm": 0.7626279592514038,
      "learning_rate": 6.551119033197256e-06,
      "loss": 0.7188,
      "step": 14494
    },
    {
      "epoch": 0.4176511266063505,
      "grad_norm": 0.7380303740501404,
      "learning_rate": 6.550675437203987e-06,
      "loss": 0.707,
      "step": 14495
    },
    {
      "epoch": 0.41767994006799974,
      "grad_norm": 0.7458621263504028,
      "learning_rate": 6.550231827705885e-06,
      "loss": 0.7247,
      "step": 14496
    },
    {
      "epoch": 0.41770875352964904,
      "grad_norm": 0.7330793738365173,
      "learning_rate": 6.549788204706814e-06,
      "loss": 0.7119,
      "step": 14497
    },
    {
      "epoch": 0.4177375669912983,
      "grad_norm": 0.7385437488555908,
      "learning_rate": 6.549344568210636e-06,
      "loss": 0.7162,
      "step": 14498
    },
    {
      "epoch": 0.4177663804529476,
      "grad_norm": 0.7350939512252808,
      "learning_rate": 6.548900918221216e-06,
      "loss": 0.7321,
      "step": 14499
    },
    {
      "epoch": 0.4177951939145969,
      "grad_norm": 0.7308743000030518,
      "learning_rate": 6.548457254742416e-06,
      "loss": 0.7193,
      "step": 14500
    },
    {
      "epoch": 0.4178240073762462,
      "grad_norm": 0.724012017250061,
      "learning_rate": 6.5480135777781025e-06,
      "loss": 0.7433,
      "step": 14501
    },
    {
      "epoch": 0.4178528208378955,
      "grad_norm": 0.7365670204162598,
      "learning_rate": 6.547569887332139e-06,
      "loss": 0.7453,
      "step": 14502
    },
    {
      "epoch": 0.4178816342995447,
      "grad_norm": 0.7541968822479248,
      "learning_rate": 6.547126183408387e-06,
      "loss": 0.7262,
      "step": 14503
    },
    {
      "epoch": 0.417910447761194,
      "grad_norm": 0.7883062362670898,
      "learning_rate": 6.546682466010713e-06,
      "loss": 0.7441,
      "step": 14504
    },
    {
      "epoch": 0.4179392612228433,
      "grad_norm": 0.7543039321899414,
      "learning_rate": 6.546238735142981e-06,
      "loss": 0.7244,
      "step": 14505
    },
    {
      "epoch": 0.4179680746844926,
      "grad_norm": 0.7333993911743164,
      "learning_rate": 6.545794990809055e-06,
      "loss": 0.6955,
      "step": 14506
    },
    {
      "epoch": 0.4179968881461419,
      "grad_norm": 0.7628651261329651,
      "learning_rate": 6.5453512330128e-06,
      "loss": 0.7189,
      "step": 14507
    },
    {
      "epoch": 0.4180257016077912,
      "grad_norm": 0.7765212059020996,
      "learning_rate": 6.544907461758081e-06,
      "loss": 0.7427,
      "step": 14508
    },
    {
      "epoch": 0.41805451506944047,
      "grad_norm": 0.7626785635948181,
      "learning_rate": 6.544463677048762e-06,
      "loss": 0.7262,
      "step": 14509
    },
    {
      "epoch": 0.4180833285310897,
      "grad_norm": 0.7702701091766357,
      "learning_rate": 6.54401987888871e-06,
      "loss": 0.7016,
      "step": 14510
    },
    {
      "epoch": 0.418112141992739,
      "grad_norm": 0.7828769087791443,
      "learning_rate": 6.543576067281786e-06,
      "loss": 0.7332,
      "step": 14511
    },
    {
      "epoch": 0.4181409554543883,
      "grad_norm": 0.7450779676437378,
      "learning_rate": 6.543132242231857e-06,
      "loss": 0.728,
      "step": 14512
    },
    {
      "epoch": 0.4181697689160376,
      "grad_norm": 0.7337931394577026,
      "learning_rate": 6.542688403742791e-06,
      "loss": 0.7211,
      "step": 14513
    },
    {
      "epoch": 0.41819858237768687,
      "grad_norm": 0.742231011390686,
      "learning_rate": 6.542244551818451e-06,
      "loss": 0.7303,
      "step": 14514
    },
    {
      "epoch": 0.41822739583933616,
      "grad_norm": 0.7751911878585815,
      "learning_rate": 6.541800686462702e-06,
      "loss": 0.7233,
      "step": 14515
    },
    {
      "epoch": 0.4182562093009854,
      "grad_norm": 0.7535369992256165,
      "learning_rate": 6.541356807679409e-06,
      "loss": 0.7092,
      "step": 14516
    },
    {
      "epoch": 0.4182850227626347,
      "grad_norm": 0.7294137477874756,
      "learning_rate": 6.54091291547244e-06,
      "loss": 0.7079,
      "step": 14517
    },
    {
      "epoch": 0.418313836224284,
      "grad_norm": 0.7493801116943359,
      "learning_rate": 6.540469009845659e-06,
      "loss": 0.6841,
      "step": 14518
    },
    {
      "epoch": 0.41834264968593327,
      "grad_norm": 0.7887111902236938,
      "learning_rate": 6.540025090802934e-06,
      "loss": 0.7548,
      "step": 14519
    },
    {
      "epoch": 0.41837146314758256,
      "grad_norm": 0.7234043478965759,
      "learning_rate": 6.539581158348129e-06,
      "loss": 0.7045,
      "step": 14520
    },
    {
      "epoch": 0.41840027660923185,
      "grad_norm": 0.7605888247489929,
      "learning_rate": 6.539137212485112e-06,
      "loss": 0.7675,
      "step": 14521
    },
    {
      "epoch": 0.41842909007088114,
      "grad_norm": 0.7495982050895691,
      "learning_rate": 6.538693253217748e-06,
      "loss": 0.7082,
      "step": 14522
    },
    {
      "epoch": 0.4184579035325304,
      "grad_norm": 0.7609519958496094,
      "learning_rate": 6.538249280549904e-06,
      "loss": 0.7683,
      "step": 14523
    },
    {
      "epoch": 0.41848671699417966,
      "grad_norm": 0.7701473236083984,
      "learning_rate": 6.537805294485447e-06,
      "loss": 0.7572,
      "step": 14524
    },
    {
      "epoch": 0.41851553045582895,
      "grad_norm": 0.7352463006973267,
      "learning_rate": 6.537361295028242e-06,
      "loss": 0.7128,
      "step": 14525
    },
    {
      "epoch": 0.41854434391747825,
      "grad_norm": 0.7446797490119934,
      "learning_rate": 6.536917282182158e-06,
      "loss": 0.7189,
      "step": 14526
    },
    {
      "epoch": 0.41857315737912754,
      "grad_norm": 0.7386152744293213,
      "learning_rate": 6.5364732559510605e-06,
      "loss": 0.7308,
      "step": 14527
    },
    {
      "epoch": 0.4186019708407768,
      "grad_norm": 0.7373806238174438,
      "learning_rate": 6.536029216338817e-06,
      "loss": 0.7344,
      "step": 14528
    },
    {
      "epoch": 0.4186307843024261,
      "grad_norm": 0.740878164768219,
      "learning_rate": 6.535585163349294e-06,
      "loss": 0.7036,
      "step": 14529
    },
    {
      "epoch": 0.41865959776407535,
      "grad_norm": 0.7892038822174072,
      "learning_rate": 6.535141096986359e-06,
      "loss": 0.734,
      "step": 14530
    },
    {
      "epoch": 0.41868841122572464,
      "grad_norm": 0.7785128951072693,
      "learning_rate": 6.53469701725388e-06,
      "loss": 0.7229,
      "step": 14531
    },
    {
      "epoch": 0.41871722468737393,
      "grad_norm": 0.7450493574142456,
      "learning_rate": 6.534252924155724e-06,
      "loss": 0.717,
      "step": 14532
    },
    {
      "epoch": 0.4187460381490232,
      "grad_norm": 0.7745087742805481,
      "learning_rate": 6.533808817695759e-06,
      "loss": 0.7496,
      "step": 14533
    },
    {
      "epoch": 0.4187748516106725,
      "grad_norm": 0.7984237670898438,
      "learning_rate": 6.533364697877853e-06,
      "loss": 0.7319,
      "step": 14534
    },
    {
      "epoch": 0.4188036650723218,
      "grad_norm": 0.7819334864616394,
      "learning_rate": 6.5329205647058734e-06,
      "loss": 0.7216,
      "step": 14535
    },
    {
      "epoch": 0.4188324785339711,
      "grad_norm": 0.729034423828125,
      "learning_rate": 6.532476418183686e-06,
      "loss": 0.7352,
      "step": 14536
    },
    {
      "epoch": 0.41886129199562033,
      "grad_norm": 0.7657025456428528,
      "learning_rate": 6.532032258315162e-06,
      "loss": 0.7335,
      "step": 14537
    },
    {
      "epoch": 0.4188901054572696,
      "grad_norm": 0.7885342240333557,
      "learning_rate": 6.531588085104169e-06,
      "loss": 0.7517,
      "step": 14538
    },
    {
      "epoch": 0.4189189189189189,
      "grad_norm": 0.773557186126709,
      "learning_rate": 6.531143898554575e-06,
      "loss": 0.715,
      "step": 14539
    },
    {
      "epoch": 0.4189477323805682,
      "grad_norm": 0.7962628602981567,
      "learning_rate": 6.530699698670249e-06,
      "loss": 0.7252,
      "step": 14540
    },
    {
      "epoch": 0.4189765458422175,
      "grad_norm": 0.8053373098373413,
      "learning_rate": 6.530255485455056e-06,
      "loss": 0.7393,
      "step": 14541
    },
    {
      "epoch": 0.4190053593038668,
      "grad_norm": 0.7543562054634094,
      "learning_rate": 6.529811258912869e-06,
      "loss": 0.726,
      "step": 14542
    },
    {
      "epoch": 0.419034172765516,
      "grad_norm": 0.7918773293495178,
      "learning_rate": 6.5293670190475546e-06,
      "loss": 0.7248,
      "step": 14543
    },
    {
      "epoch": 0.4190629862271653,
      "grad_norm": 0.7622688412666321,
      "learning_rate": 6.528922765862983e-06,
      "loss": 0.7245,
      "step": 14544
    },
    {
      "epoch": 0.4190917996888146,
      "grad_norm": 0.7850104570388794,
      "learning_rate": 6.5284784993630225e-06,
      "loss": 0.7283,
      "step": 14545
    },
    {
      "epoch": 0.4191206131504639,
      "grad_norm": 0.7325103282928467,
      "learning_rate": 6.528034219551542e-06,
      "loss": 0.7217,
      "step": 14546
    },
    {
      "epoch": 0.4191494266121132,
      "grad_norm": 0.751262366771698,
      "learning_rate": 6.52758992643241e-06,
      "loss": 0.7096,
      "step": 14547
    },
    {
      "epoch": 0.4191782400737625,
      "grad_norm": 0.7557001709938049,
      "learning_rate": 6.527145620009497e-06,
      "loss": 0.7252,
      "step": 14548
    },
    {
      "epoch": 0.41920705353541177,
      "grad_norm": 0.7490007877349854,
      "learning_rate": 6.526701300286673e-06,
      "loss": 0.7026,
      "step": 14549
    },
    {
      "epoch": 0.419235866997061,
      "grad_norm": 0.7720834612846375,
      "learning_rate": 6.526256967267808e-06,
      "loss": 0.7144,
      "step": 14550
    },
    {
      "epoch": 0.4192646804587103,
      "grad_norm": 0.7804278135299683,
      "learning_rate": 6.5258126209567685e-06,
      "loss": 0.7341,
      "step": 14551
    },
    {
      "epoch": 0.4192934939203596,
      "grad_norm": 0.7656057476997375,
      "learning_rate": 6.525368261357428e-06,
      "loss": 0.7465,
      "step": 14552
    },
    {
      "epoch": 0.4193223073820089,
      "grad_norm": 0.7693214416503906,
      "learning_rate": 6.5249238884736534e-06,
      "loss": 0.7468,
      "step": 14553
    },
    {
      "epoch": 0.41935112084365816,
      "grad_norm": 0.7591188549995422,
      "learning_rate": 6.524479502309316e-06,
      "loss": 0.7509,
      "step": 14554
    },
    {
      "epoch": 0.41937993430530746,
      "grad_norm": 0.7793684601783752,
      "learning_rate": 6.524035102868287e-06,
      "loss": 0.7249,
      "step": 14555
    },
    {
      "epoch": 0.41940874776695675,
      "grad_norm": 0.7552603483200073,
      "learning_rate": 6.523590690154436e-06,
      "loss": 0.7489,
      "step": 14556
    },
    {
      "epoch": 0.419437561228606,
      "grad_norm": 0.7282139658927917,
      "learning_rate": 6.523146264171631e-06,
      "loss": 0.7323,
      "step": 14557
    },
    {
      "epoch": 0.4194663746902553,
      "grad_norm": 0.7403602600097656,
      "learning_rate": 6.522701824923747e-06,
      "loss": 0.7093,
      "step": 14558
    },
    {
      "epoch": 0.41949518815190456,
      "grad_norm": 0.7637699246406555,
      "learning_rate": 6.522257372414651e-06,
      "loss": 0.7328,
      "step": 14559
    },
    {
      "epoch": 0.41952400161355385,
      "grad_norm": 0.7416549324989319,
      "learning_rate": 6.521812906648215e-06,
      "loss": 0.7282,
      "step": 14560
    },
    {
      "epoch": 0.41955281507520314,
      "grad_norm": 0.7765712141990662,
      "learning_rate": 6.521368427628309e-06,
      "loss": 0.7422,
      "step": 14561
    },
    {
      "epoch": 0.41958162853685244,
      "grad_norm": 0.724388837814331,
      "learning_rate": 6.520923935358806e-06,
      "loss": 0.7112,
      "step": 14562
    },
    {
      "epoch": 0.4196104419985017,
      "grad_norm": 0.7544626593589783,
      "learning_rate": 6.520479429843576e-06,
      "loss": 0.7339,
      "step": 14563
    },
    {
      "epoch": 0.41963925546015096,
      "grad_norm": 0.7710082530975342,
      "learning_rate": 6.52003491108649e-06,
      "loss": 0.741,
      "step": 14564
    },
    {
      "epoch": 0.41966806892180025,
      "grad_norm": 0.7594892978668213,
      "learning_rate": 6.51959037909142e-06,
      "loss": 0.7242,
      "step": 14565
    },
    {
      "epoch": 0.41969688238344954,
      "grad_norm": 0.7904366850852966,
      "learning_rate": 6.519145833862235e-06,
      "loss": 0.7457,
      "step": 14566
    },
    {
      "epoch": 0.41972569584509883,
      "grad_norm": 0.7835882902145386,
      "learning_rate": 6.51870127540281e-06,
      "loss": 0.7248,
      "step": 14567
    },
    {
      "epoch": 0.4197545093067481,
      "grad_norm": 0.7818170189857483,
      "learning_rate": 6.518256703717014e-06,
      "loss": 0.7385,
      "step": 14568
    },
    {
      "epoch": 0.4197833227683974,
      "grad_norm": 0.7327814698219299,
      "learning_rate": 6.517812118808721e-06,
      "loss": 0.7488,
      "step": 14569
    },
    {
      "epoch": 0.41981213623004665,
      "grad_norm": 0.7807397246360779,
      "learning_rate": 6.517367520681801e-06,
      "loss": 0.7307,
      "step": 14570
    },
    {
      "epoch": 0.41984094969169594,
      "grad_norm": 0.7955352663993835,
      "learning_rate": 6.516922909340128e-06,
      "loss": 0.7171,
      "step": 14571
    },
    {
      "epoch": 0.41986976315334523,
      "grad_norm": 0.7450814247131348,
      "learning_rate": 6.516478284787569e-06,
      "loss": 0.7464,
      "step": 14572
    },
    {
      "epoch": 0.4198985766149945,
      "grad_norm": 0.7443754076957703,
      "learning_rate": 6.5160336470280035e-06,
      "loss": 0.7306,
      "step": 14573
    },
    {
      "epoch": 0.4199273900766438,
      "grad_norm": 0.7685709595680237,
      "learning_rate": 6.5155889960653006e-06,
      "loss": 0.7365,
      "step": 14574
    },
    {
      "epoch": 0.4199562035382931,
      "grad_norm": 0.750731885433197,
      "learning_rate": 6.5151443319033315e-06,
      "loss": 0.7399,
      "step": 14575
    },
    {
      "epoch": 0.4199850169999424,
      "grad_norm": 0.7600321173667908,
      "learning_rate": 6.51469965454597e-06,
      "loss": 0.7421,
      "step": 14576
    },
    {
      "epoch": 0.42001383046159163,
      "grad_norm": 0.7343473434448242,
      "learning_rate": 6.514254963997089e-06,
      "loss": 0.7265,
      "step": 14577
    },
    {
      "epoch": 0.4200426439232409,
      "grad_norm": 0.7165581583976746,
      "learning_rate": 6.513810260260559e-06,
      "loss": 0.7381,
      "step": 14578
    },
    {
      "epoch": 0.4200714573848902,
      "grad_norm": 0.7695158123970032,
      "learning_rate": 6.513365543340257e-06,
      "loss": 0.7329,
      "step": 14579
    },
    {
      "epoch": 0.4201002708465395,
      "grad_norm": 0.7708241939544678,
      "learning_rate": 6.512920813240053e-06,
      "loss": 0.7585,
      "step": 14580
    },
    {
      "epoch": 0.4201290843081888,
      "grad_norm": 0.7829635143280029,
      "learning_rate": 6.512476069963821e-06,
      "loss": 0.7366,
      "step": 14581
    },
    {
      "epoch": 0.4201578977698381,
      "grad_norm": 0.7373799681663513,
      "learning_rate": 6.512031313515435e-06,
      "loss": 0.7276,
      "step": 14582
    },
    {
      "epoch": 0.4201867112314874,
      "grad_norm": 0.7179800868034363,
      "learning_rate": 6.511586543898768e-06,
      "loss": 0.7252,
      "step": 14583
    },
    {
      "epoch": 0.4202155246931366,
      "grad_norm": 0.7814826369285583,
      "learning_rate": 6.511141761117692e-06,
      "loss": 0.7256,
      "step": 14584
    },
    {
      "epoch": 0.4202443381547859,
      "grad_norm": 0.7509123682975769,
      "learning_rate": 6.510696965176082e-06,
      "loss": 0.7419,
      "step": 14585
    },
    {
      "epoch": 0.4202731516164352,
      "grad_norm": 1.9895633459091187,
      "learning_rate": 6.510252156077813e-06,
      "loss": 0.7223,
      "step": 14586
    },
    {
      "epoch": 0.4203019650780845,
      "grad_norm": 0.7432743906974792,
      "learning_rate": 6.509807333826757e-06,
      "loss": 0.7223,
      "step": 14587
    },
    {
      "epoch": 0.4203307785397338,
      "grad_norm": 0.8022254109382629,
      "learning_rate": 6.509362498426788e-06,
      "loss": 0.7404,
      "step": 14588
    },
    {
      "epoch": 0.42035959200138306,
      "grad_norm": 0.7511430382728577,
      "learning_rate": 6.50891764988178e-06,
      "loss": 0.7272,
      "step": 14589
    },
    {
      "epoch": 0.42038840546303236,
      "grad_norm": 0.7456490397453308,
      "learning_rate": 6.5084727881956074e-06,
      "loss": 0.7131,
      "step": 14590
    },
    {
      "epoch": 0.4204172189246816,
      "grad_norm": 0.7625361084938049,
      "learning_rate": 6.508027913372146e-06,
      "loss": 0.7574,
      "step": 14591
    },
    {
      "epoch": 0.4204460323863309,
      "grad_norm": 0.7895218133926392,
      "learning_rate": 6.507583025415268e-06,
      "loss": 0.7171,
      "step": 14592
    },
    {
      "epoch": 0.42047484584798017,
      "grad_norm": 0.7827379107475281,
      "learning_rate": 6.50713812432885e-06,
      "loss": 0.7316,
      "step": 14593
    },
    {
      "epoch": 0.42050365930962946,
      "grad_norm": 0.7598482966423035,
      "learning_rate": 6.506693210116765e-06,
      "loss": 0.7359,
      "step": 14594
    },
    {
      "epoch": 0.42053247277127875,
      "grad_norm": 0.7547750473022461,
      "learning_rate": 6.506248282782888e-06,
      "loss": 0.7174,
      "step": 14595
    },
    {
      "epoch": 0.42056128623292804,
      "grad_norm": 0.7675867676734924,
      "learning_rate": 6.5058033423310926e-06,
      "loss": 0.7161,
      "step": 14596
    },
    {
      "epoch": 0.4205900996945773,
      "grad_norm": 0.7688458561897278,
      "learning_rate": 6.505358388765257e-06,
      "loss": 0.7292,
      "step": 14597
    },
    {
      "epoch": 0.42061891315622657,
      "grad_norm": 0.7738723754882812,
      "learning_rate": 6.504913422089256e-06,
      "loss": 0.7164,
      "step": 14598
    },
    {
      "epoch": 0.42064772661787586,
      "grad_norm": 0.7353576421737671,
      "learning_rate": 6.504468442306961e-06,
      "loss": 0.7138,
      "step": 14599
    },
    {
      "epoch": 0.42067654007952515,
      "grad_norm": 0.7458694577217102,
      "learning_rate": 6.504023449422251e-06,
      "loss": 0.7341,
      "step": 14600
    },
    {
      "epoch": 0.42070535354117444,
      "grad_norm": 0.764142632484436,
      "learning_rate": 6.503578443439e-06,
      "loss": 0.7558,
      "step": 14601
    },
    {
      "epoch": 0.42073416700282373,
      "grad_norm": 0.7425388693809509,
      "learning_rate": 6.503133424361083e-06,
      "loss": 0.7489,
      "step": 14602
    },
    {
      "epoch": 0.420762980464473,
      "grad_norm": 0.7604843378067017,
      "learning_rate": 6.502688392192378e-06,
      "loss": 0.7373,
      "step": 14603
    },
    {
      "epoch": 0.42079179392612226,
      "grad_norm": 0.743228018283844,
      "learning_rate": 6.502243346936759e-06,
      "loss": 0.7281,
      "step": 14604
    },
    {
      "epoch": 0.42082060738777155,
      "grad_norm": 0.7240402698516846,
      "learning_rate": 6.501798288598101e-06,
      "loss": 0.7278,
      "step": 14605
    },
    {
      "epoch": 0.42084942084942084,
      "grad_norm": 0.7442803382873535,
      "learning_rate": 6.501353217180282e-06,
      "loss": 0.7167,
      "step": 14606
    },
    {
      "epoch": 0.42087823431107013,
      "grad_norm": 0.781667947769165,
      "learning_rate": 6.500908132687177e-06,
      "loss": 0.7398,
      "step": 14607
    },
    {
      "epoch": 0.4209070477727194,
      "grad_norm": 0.7398574948310852,
      "learning_rate": 6.500463035122662e-06,
      "loss": 0.7189,
      "step": 14608
    },
    {
      "epoch": 0.4209358612343687,
      "grad_norm": 0.8967156410217285,
      "learning_rate": 6.500017924490615e-06,
      "loss": 0.7419,
      "step": 14609
    },
    {
      "epoch": 0.420964674696018,
      "grad_norm": 0.7639592289924622,
      "learning_rate": 6.499572800794911e-06,
      "loss": 0.7239,
      "step": 14610
    },
    {
      "epoch": 0.42099348815766724,
      "grad_norm": 0.7415057420730591,
      "learning_rate": 6.4991276640394266e-06,
      "loss": 0.7297,
      "step": 14611
    },
    {
      "epoch": 0.42102230161931653,
      "grad_norm": 0.7418016791343689,
      "learning_rate": 6.49868251422804e-06,
      "loss": 0.7372,
      "step": 14612
    },
    {
      "epoch": 0.4210511150809658,
      "grad_norm": 0.7641198635101318,
      "learning_rate": 6.498237351364626e-06,
      "loss": 0.7219,
      "step": 14613
    },
    {
      "epoch": 0.4210799285426151,
      "grad_norm": 0.7699970602989197,
      "learning_rate": 6.497792175453062e-06,
      "loss": 0.7511,
      "step": 14614
    },
    {
      "epoch": 0.4211087420042644,
      "grad_norm": 0.7478667497634888,
      "learning_rate": 6.497346986497225e-06,
      "loss": 0.7135,
      "step": 14615
    },
    {
      "epoch": 0.4211375554659137,
      "grad_norm": 0.75236976146698,
      "learning_rate": 6.496901784500993e-06,
      "loss": 0.727,
      "step": 14616
    },
    {
      "epoch": 0.421166368927563,
      "grad_norm": 0.7241716980934143,
      "learning_rate": 6.496456569468244e-06,
      "loss": 0.7105,
      "step": 14617
    },
    {
      "epoch": 0.4211951823892122,
      "grad_norm": 0.7258962392807007,
      "learning_rate": 6.496011341402853e-06,
      "loss": 0.7176,
      "step": 14618
    },
    {
      "epoch": 0.4212239958508615,
      "grad_norm": 1.7500625848770142,
      "learning_rate": 6.495566100308698e-06,
      "loss": 0.7401,
      "step": 14619
    },
    {
      "epoch": 0.4212528093125108,
      "grad_norm": 0.7623234391212463,
      "learning_rate": 6.495120846189658e-06,
      "loss": 0.7185,
      "step": 14620
    },
    {
      "epoch": 0.4212816227741601,
      "grad_norm": 0.7639103531837463,
      "learning_rate": 6.494675579049609e-06,
      "loss": 0.7332,
      "step": 14621
    },
    {
      "epoch": 0.4213104362358094,
      "grad_norm": 0.7593166828155518,
      "learning_rate": 6.494230298892431e-06,
      "loss": 0.7421,
      "step": 14622
    },
    {
      "epoch": 0.4213392496974587,
      "grad_norm": 0.7541190981864929,
      "learning_rate": 6.493785005722001e-06,
      "loss": 0.7316,
      "step": 14623
    },
    {
      "epoch": 0.4213680631591079,
      "grad_norm": 0.766276478767395,
      "learning_rate": 6.493339699542197e-06,
      "loss": 0.7699,
      "step": 14624
    },
    {
      "epoch": 0.4213968766207572,
      "grad_norm": 0.7387105226516724,
      "learning_rate": 6.492894380356897e-06,
      "loss": 0.712,
      "step": 14625
    },
    {
      "epoch": 0.4214256900824065,
      "grad_norm": 4.664315700531006,
      "learning_rate": 6.492449048169977e-06,
      "loss": 0.7391,
      "step": 14626
    },
    {
      "epoch": 0.4214545035440558,
      "grad_norm": 0.7352954149246216,
      "learning_rate": 6.492003702985321e-06,
      "loss": 0.7348,
      "step": 14627
    },
    {
      "epoch": 0.42148331700570507,
      "grad_norm": 0.7627981901168823,
      "learning_rate": 6.491558344806802e-06,
      "loss": 0.7185,
      "step": 14628
    },
    {
      "epoch": 0.42151213046735436,
      "grad_norm": 0.8329259753227234,
      "learning_rate": 6.491112973638303e-06,
      "loss": 0.7429,
      "step": 14629
    },
    {
      "epoch": 0.42154094392900365,
      "grad_norm": 0.7544717788696289,
      "learning_rate": 6.490667589483699e-06,
      "loss": 0.7279,
      "step": 14630
    },
    {
      "epoch": 0.4215697573906529,
      "grad_norm": 0.7995635271072388,
      "learning_rate": 6.49022219234687e-06,
      "loss": 0.7585,
      "step": 14631
    },
    {
      "epoch": 0.4215985708523022,
      "grad_norm": 0.7622101306915283,
      "learning_rate": 6.489776782231696e-06,
      "loss": 0.7567,
      "step": 14632
    },
    {
      "epoch": 0.42162738431395147,
      "grad_norm": 0.7971873879432678,
      "learning_rate": 6.489331359142055e-06,
      "loss": 0.7484,
      "step": 14633
    },
    {
      "epoch": 0.42165619777560076,
      "grad_norm": 0.7831166982650757,
      "learning_rate": 6.4888859230818265e-06,
      "loss": 0.7417,
      "step": 14634
    },
    {
      "epoch": 0.42168501123725005,
      "grad_norm": 0.7492903470993042,
      "learning_rate": 6.488440474054891e-06,
      "loss": 0.7394,
      "step": 14635
    },
    {
      "epoch": 0.42171382469889934,
      "grad_norm": 0.7577508687973022,
      "learning_rate": 6.487995012065127e-06,
      "loss": 0.7372,
      "step": 14636
    },
    {
      "epoch": 0.42174263816054863,
      "grad_norm": 0.8011428713798523,
      "learning_rate": 6.487549537116413e-06,
      "loss": 0.737,
      "step": 14637
    },
    {
      "epoch": 0.42177145162219787,
      "grad_norm": 0.751054584980011,
      "learning_rate": 6.487104049212629e-06,
      "loss": 0.7496,
      "step": 14638
    },
    {
      "epoch": 0.42180026508384716,
      "grad_norm": 0.7607036232948303,
      "learning_rate": 6.486658548357657e-06,
      "loss": 0.7334,
      "step": 14639
    },
    {
      "epoch": 0.42182907854549645,
      "grad_norm": 0.7447863817214966,
      "learning_rate": 6.486213034555373e-06,
      "loss": 0.7353,
      "step": 14640
    },
    {
      "epoch": 0.42185789200714574,
      "grad_norm": 0.7982693314552307,
      "learning_rate": 6.485767507809662e-06,
      "loss": 0.7318,
      "step": 14641
    },
    {
      "epoch": 0.42188670546879503,
      "grad_norm": 0.7808882594108582,
      "learning_rate": 6.485321968124398e-06,
      "loss": 0.7336,
      "step": 14642
    },
    {
      "epoch": 0.4219155189304443,
      "grad_norm": 0.7715519666671753,
      "learning_rate": 6.484876415503467e-06,
      "loss": 0.6965,
      "step": 14643
    },
    {
      "epoch": 0.4219443323920936,
      "grad_norm": 0.7324472069740295,
      "learning_rate": 6.484430849950745e-06,
      "loss": 0.7116,
      "step": 14644
    },
    {
      "epoch": 0.42197314585374285,
      "grad_norm": 0.789878249168396,
      "learning_rate": 6.483985271470116e-06,
      "loss": 0.7438,
      "step": 14645
    },
    {
      "epoch": 0.42200195931539214,
      "grad_norm": 0.7778681516647339,
      "learning_rate": 6.483539680065457e-06,
      "loss": 0.7301,
      "step": 14646
    },
    {
      "epoch": 0.42203077277704143,
      "grad_norm": 0.7594836354255676,
      "learning_rate": 6.483094075740651e-06,
      "loss": 0.7196,
      "step": 14647
    },
    {
      "epoch": 0.4220595862386907,
      "grad_norm": 0.7478933334350586,
      "learning_rate": 6.482648458499579e-06,
      "loss": 0.7488,
      "step": 14648
    },
    {
      "epoch": 0.42208839970034,
      "grad_norm": 0.7621017694473267,
      "learning_rate": 6.482202828346121e-06,
      "loss": 0.7287,
      "step": 14649
    },
    {
      "epoch": 0.4221172131619893,
      "grad_norm": 0.7561421990394592,
      "learning_rate": 6.481757185284156e-06,
      "loss": 0.733,
      "step": 14650
    },
    {
      "epoch": 0.42214602662363854,
      "grad_norm": 0.7462197542190552,
      "learning_rate": 6.481311529317569e-06,
      "loss": 0.7452,
      "step": 14651
    },
    {
      "epoch": 0.42217484008528783,
      "grad_norm": 0.7642948627471924,
      "learning_rate": 6.480865860450238e-06,
      "loss": 0.7311,
      "step": 14652
    },
    {
      "epoch": 0.4222036535469371,
      "grad_norm": 0.7496647238731384,
      "learning_rate": 6.4804201786860475e-06,
      "loss": 0.7359,
      "step": 14653
    },
    {
      "epoch": 0.4222324670085864,
      "grad_norm": 0.7650327086448669,
      "learning_rate": 6.479974484028876e-06,
      "loss": 0.7189,
      "step": 14654
    },
    {
      "epoch": 0.4222612804702357,
      "grad_norm": 0.7673014998435974,
      "learning_rate": 6.479528776482606e-06,
      "loss": 0.731,
      "step": 14655
    },
    {
      "epoch": 0.422290093931885,
      "grad_norm": 0.7411987781524658,
      "learning_rate": 6.47908305605112e-06,
      "loss": 0.7344,
      "step": 14656
    },
    {
      "epoch": 0.4223189073935343,
      "grad_norm": 0.7460326552391052,
      "learning_rate": 6.478637322738298e-06,
      "loss": 0.7121,
      "step": 14657
    },
    {
      "epoch": 0.4223477208551835,
      "grad_norm": 0.7586778402328491,
      "learning_rate": 6.478191576548024e-06,
      "loss": 0.7185,
      "step": 14658
    },
    {
      "epoch": 0.4223765343168328,
      "grad_norm": 0.7327917814254761,
      "learning_rate": 6.477745817484178e-06,
      "loss": 0.7336,
      "step": 14659
    },
    {
      "epoch": 0.4224053477784821,
      "grad_norm": 0.7330420017242432,
      "learning_rate": 6.477300045550645e-06,
      "loss": 0.7323,
      "step": 14660
    },
    {
      "epoch": 0.4224341612401314,
      "grad_norm": 0.7726958394050598,
      "learning_rate": 6.476854260751304e-06,
      "loss": 0.7283,
      "step": 14661
    },
    {
      "epoch": 0.4224629747017807,
      "grad_norm": 0.7554693222045898,
      "learning_rate": 6.476408463090038e-06,
      "loss": 0.7316,
      "step": 14662
    },
    {
      "epoch": 0.42249178816342997,
      "grad_norm": 0.7352268695831299,
      "learning_rate": 6.475962652570731e-06,
      "loss": 0.7225,
      "step": 14663
    },
    {
      "epoch": 0.42252060162507926,
      "grad_norm": 0.7656000852584839,
      "learning_rate": 6.4755168291972655e-06,
      "loss": 0.7316,
      "step": 14664
    },
    {
      "epoch": 0.4225494150867285,
      "grad_norm": 0.7416296005249023,
      "learning_rate": 6.475070992973524e-06,
      "loss": 0.7315,
      "step": 14665
    },
    {
      "epoch": 0.4225782285483778,
      "grad_norm": 0.7569639682769775,
      "learning_rate": 6.474625143903387e-06,
      "loss": 0.7175,
      "step": 14666
    },
    {
      "epoch": 0.4226070420100271,
      "grad_norm": 0.7543926239013672,
      "learning_rate": 6.47417928199074e-06,
      "loss": 0.7655,
      "step": 14667
    },
    {
      "epoch": 0.42263585547167637,
      "grad_norm": 0.7692974805831909,
      "learning_rate": 6.473733407239463e-06,
      "loss": 0.7347,
      "step": 14668
    },
    {
      "epoch": 0.42266466893332566,
      "grad_norm": 0.7626014351844788,
      "learning_rate": 6.473287519653444e-06,
      "loss": 0.749,
      "step": 14669
    },
    {
      "epoch": 0.42269348239497495,
      "grad_norm": 0.7420811057090759,
      "learning_rate": 6.4728416192365615e-06,
      "loss": 0.7487,
      "step": 14670
    },
    {
      "epoch": 0.42272229585662424,
      "grad_norm": 0.7290780544281006,
      "learning_rate": 6.472395705992702e-06,
      "loss": 0.7398,
      "step": 14671
    },
    {
      "epoch": 0.4227511093182735,
      "grad_norm": 0.7557438611984253,
      "learning_rate": 6.471949779925748e-06,
      "loss": 0.7353,
      "step": 14672
    },
    {
      "epoch": 0.42277992277992277,
      "grad_norm": 0.7622619867324829,
      "learning_rate": 6.4715038410395815e-06,
      "loss": 0.7329,
      "step": 14673
    },
    {
      "epoch": 0.42280873624157206,
      "grad_norm": 0.7205125689506531,
      "learning_rate": 6.471057889338088e-06,
      "loss": 0.7283,
      "step": 14674
    },
    {
      "epoch": 0.42283754970322135,
      "grad_norm": 0.7537312507629395,
      "learning_rate": 6.470611924825151e-06,
      "loss": 0.7138,
      "step": 14675
    },
    {
      "epoch": 0.42286636316487064,
      "grad_norm": 0.7403140664100647,
      "learning_rate": 6.470165947504654e-06,
      "loss": 0.7146,
      "step": 14676
    },
    {
      "epoch": 0.42289517662651993,
      "grad_norm": 0.7862645387649536,
      "learning_rate": 6.4697199573804825e-06,
      "loss": 0.7342,
      "step": 14677
    },
    {
      "epoch": 0.42292399008816917,
      "grad_norm": 0.7523658871650696,
      "learning_rate": 6.469273954456516e-06,
      "loss": 0.722,
      "step": 14678
    },
    {
      "epoch": 0.42295280354981846,
      "grad_norm": 0.7588355541229248,
      "learning_rate": 6.468827938736646e-06,
      "loss": 0.7353,
      "step": 14679
    },
    {
      "epoch": 0.42298161701146775,
      "grad_norm": 0.7672023177146912,
      "learning_rate": 6.468381910224751e-06,
      "loss": 0.7637,
      "step": 14680
    },
    {
      "epoch": 0.42301043047311704,
      "grad_norm": 0.7544943690299988,
      "learning_rate": 6.467935868924717e-06,
      "loss": 0.7354,
      "step": 14681
    },
    {
      "epoch": 0.42303924393476633,
      "grad_norm": 0.7468137741088867,
      "learning_rate": 6.4674898148404295e-06,
      "loss": 0.7169,
      "step": 14682
    },
    {
      "epoch": 0.4230680573964156,
      "grad_norm": 0.764240562915802,
      "learning_rate": 6.467043747975773e-06,
      "loss": 0.7231,
      "step": 14683
    },
    {
      "epoch": 0.4230968708580649,
      "grad_norm": 0.7783116102218628,
      "learning_rate": 6.466597668334632e-06,
      "loss": 0.7225,
      "step": 14684
    },
    {
      "epoch": 0.42312568431971415,
      "grad_norm": 0.7595934271812439,
      "learning_rate": 6.466151575920892e-06,
      "loss": 0.7407,
      "step": 14685
    },
    {
      "epoch": 0.42315449778136344,
      "grad_norm": 0.7723734974861145,
      "learning_rate": 6.465705470738436e-06,
      "loss": 0.7522,
      "step": 14686
    },
    {
      "epoch": 0.4231833112430127,
      "grad_norm": 0.7522721290588379,
      "learning_rate": 6.465259352791151e-06,
      "loss": 0.7304,
      "step": 14687
    },
    {
      "epoch": 0.423212124704662,
      "grad_norm": 0.7427682280540466,
      "learning_rate": 6.464813222082922e-06,
      "loss": 0.7247,
      "step": 14688
    },
    {
      "epoch": 0.4232409381663113,
      "grad_norm": 0.7791905403137207,
      "learning_rate": 6.464367078617634e-06,
      "loss": 0.7215,
      "step": 14689
    },
    {
      "epoch": 0.4232697516279606,
      "grad_norm": 0.8087671995162964,
      "learning_rate": 6.463920922399173e-06,
      "loss": 0.756,
      "step": 14690
    },
    {
      "epoch": 0.4232985650896099,
      "grad_norm": 0.738970935344696,
      "learning_rate": 6.463474753431424e-06,
      "loss": 0.74,
      "step": 14691
    },
    {
      "epoch": 0.4233273785512591,
      "grad_norm": 0.7728750109672546,
      "learning_rate": 6.4630285717182714e-06,
      "loss": 0.7188,
      "step": 14692
    },
    {
      "epoch": 0.4233561920129084,
      "grad_norm": 0.7488132119178772,
      "learning_rate": 6.462582377263604e-06,
      "loss": 0.729,
      "step": 14693
    },
    {
      "epoch": 0.4233850054745577,
      "grad_norm": 0.7561067342758179,
      "learning_rate": 6.462136170071306e-06,
      "loss": 0.7488,
      "step": 14694
    },
    {
      "epoch": 0.423413818936207,
      "grad_norm": 0.7560498118400574,
      "learning_rate": 6.461689950145263e-06,
      "loss": 0.7512,
      "step": 14695
    },
    {
      "epoch": 0.4234426323978563,
      "grad_norm": 0.7283270359039307,
      "learning_rate": 6.4612437174893624e-06,
      "loss": 0.7369,
      "step": 14696
    },
    {
      "epoch": 0.4234714458595056,
      "grad_norm": 0.7646386027336121,
      "learning_rate": 6.460797472107489e-06,
      "loss": 0.6902,
      "step": 14697
    },
    {
      "epoch": 0.42350025932115487,
      "grad_norm": 0.7634744048118591,
      "learning_rate": 6.46035121400353e-06,
      "loss": 0.751,
      "step": 14698
    },
    {
      "epoch": 0.4235290727828041,
      "grad_norm": 0.7835397720336914,
      "learning_rate": 6.459904943181372e-06,
      "loss": 0.7517,
      "step": 14699
    },
    {
      "epoch": 0.4235578862444534,
      "grad_norm": 0.7563515305519104,
      "learning_rate": 6.4594586596449e-06,
      "loss": 0.7331,
      "step": 14700
    },
    {
      "epoch": 0.4235866997061027,
      "grad_norm": 0.7983858585357666,
      "learning_rate": 6.459012363398004e-06,
      "loss": 0.7247,
      "step": 14701
    },
    {
      "epoch": 0.423615513167752,
      "grad_norm": 0.7529875040054321,
      "learning_rate": 6.458566054444568e-06,
      "loss": 0.747,
      "step": 14702
    },
    {
      "epoch": 0.42364432662940127,
      "grad_norm": 0.7482179999351501,
      "learning_rate": 6.45811973278848e-06,
      "loss": 0.7396,
      "step": 14703
    },
    {
      "epoch": 0.42367314009105056,
      "grad_norm": 0.7557046413421631,
      "learning_rate": 6.457673398433626e-06,
      "loss": 0.7229,
      "step": 14704
    },
    {
      "epoch": 0.4237019535526998,
      "grad_norm": 0.7530228495597839,
      "learning_rate": 6.4572270513838946e-06,
      "loss": 0.7303,
      "step": 14705
    },
    {
      "epoch": 0.4237307670143491,
      "grad_norm": 0.7343759536743164,
      "learning_rate": 6.456780691643172e-06,
      "loss": 0.7172,
      "step": 14706
    },
    {
      "epoch": 0.4237595804759984,
      "grad_norm": 0.7461184859275818,
      "learning_rate": 6.456334319215346e-06,
      "loss": 0.7222,
      "step": 14707
    },
    {
      "epoch": 0.42378839393764767,
      "grad_norm": 0.7286520004272461,
      "learning_rate": 6.455887934104304e-06,
      "loss": 0.712,
      "step": 14708
    },
    {
      "epoch": 0.42381720739929696,
      "grad_norm": 0.7844378352165222,
      "learning_rate": 6.455441536313933e-06,
      "loss": 0.7506,
      "step": 14709
    },
    {
      "epoch": 0.42384602086094625,
      "grad_norm": 0.7734556794166565,
      "learning_rate": 6.4549951258481206e-06,
      "loss": 0.7293,
      "step": 14710
    },
    {
      "epoch": 0.42387483432259554,
      "grad_norm": 0.7739797830581665,
      "learning_rate": 6.454548702710756e-06,
      "loss": 0.7325,
      "step": 14711
    },
    {
      "epoch": 0.4239036477842448,
      "grad_norm": 0.7681077122688293,
      "learning_rate": 6.4541022669057265e-06,
      "loss": 0.722,
      "step": 14712
    },
    {
      "epoch": 0.42393246124589407,
      "grad_norm": 0.7575719356536865,
      "learning_rate": 6.453655818436919e-06,
      "loss": 0.7305,
      "step": 14713
    },
    {
      "epoch": 0.42396127470754336,
      "grad_norm": 0.7680052518844604,
      "learning_rate": 6.453209357308224e-06,
      "loss": 0.7256,
      "step": 14714
    },
    {
      "epoch": 0.42399008816919265,
      "grad_norm": 0.7768230438232422,
      "learning_rate": 6.452762883523525e-06,
      "loss": 0.7231,
      "step": 14715
    },
    {
      "epoch": 0.42401890163084194,
      "grad_norm": 0.7700117230415344,
      "learning_rate": 6.452316397086715e-06,
      "loss": 0.7261,
      "step": 14716
    },
    {
      "epoch": 0.42404771509249123,
      "grad_norm": 0.74945467710495,
      "learning_rate": 6.451869898001681e-06,
      "loss": 0.7485,
      "step": 14717
    },
    {
      "epoch": 0.4240765285541405,
      "grad_norm": 0.7529350519180298,
      "learning_rate": 6.451423386272312e-06,
      "loss": 0.7178,
      "step": 14718
    },
    {
      "epoch": 0.42410534201578975,
      "grad_norm": 0.8015321493148804,
      "learning_rate": 6.450976861902496e-06,
      "loss": 0.7402,
      "step": 14719
    },
    {
      "epoch": 0.42413415547743905,
      "grad_norm": 0.7821657061576843,
      "learning_rate": 6.4505303248961225e-06,
      "loss": 0.7309,
      "step": 14720
    },
    {
      "epoch": 0.42416296893908834,
      "grad_norm": 0.7730380892753601,
      "learning_rate": 6.450083775257079e-06,
      "loss": 0.7291,
      "step": 14721
    },
    {
      "epoch": 0.4241917824007376,
      "grad_norm": 0.7640557289123535,
      "learning_rate": 6.449637212989256e-06,
      "loss": 0.7428,
      "step": 14722
    },
    {
      "epoch": 0.4242205958623869,
      "grad_norm": 0.7482209205627441,
      "learning_rate": 6.449190638096542e-06,
      "loss": 0.7144,
      "step": 14723
    },
    {
      "epoch": 0.4242494093240362,
      "grad_norm": 0.7604823112487793,
      "learning_rate": 6.448744050582826e-06,
      "loss": 0.7336,
      "step": 14724
    },
    {
      "epoch": 0.4242782227856855,
      "grad_norm": 0.7538583874702454,
      "learning_rate": 6.4482974504519976e-06,
      "loss": 0.7384,
      "step": 14725
    },
    {
      "epoch": 0.42430703624733473,
      "grad_norm": 0.7773265242576599,
      "learning_rate": 6.447850837707946e-06,
      "loss": 0.7313,
      "step": 14726
    },
    {
      "epoch": 0.424335849708984,
      "grad_norm": 0.7573897242546082,
      "learning_rate": 6.447404212354561e-06,
      "loss": 0.7099,
      "step": 14727
    },
    {
      "epoch": 0.4243646631706333,
      "grad_norm": 0.7574650645256042,
      "learning_rate": 6.446957574395733e-06,
      "loss": 0.7397,
      "step": 14728
    },
    {
      "epoch": 0.4243934766322826,
      "grad_norm": 0.7468056082725525,
      "learning_rate": 6.44651092383535e-06,
      "loss": 0.7088,
      "step": 14729
    },
    {
      "epoch": 0.4244222900939319,
      "grad_norm": 0.7696532607078552,
      "learning_rate": 6.446064260677302e-06,
      "loss": 0.7443,
      "step": 14730
    },
    {
      "epoch": 0.4244511035555812,
      "grad_norm": 0.7712768912315369,
      "learning_rate": 6.445617584925482e-06,
      "loss": 0.738,
      "step": 14731
    },
    {
      "epoch": 0.4244799170172304,
      "grad_norm": 0.7908291816711426,
      "learning_rate": 6.445170896583777e-06,
      "loss": 0.7501,
      "step": 14732
    },
    {
      "epoch": 0.4245087304788797,
      "grad_norm": 0.7351630330085754,
      "learning_rate": 6.444724195656077e-06,
      "loss": 0.711,
      "step": 14733
    },
    {
      "epoch": 0.424537543940529,
      "grad_norm": 0.7364716529846191,
      "learning_rate": 6.444277482146275e-06,
      "loss": 0.7278,
      "step": 14734
    },
    {
      "epoch": 0.4245663574021783,
      "grad_norm": 0.7381463646888733,
      "learning_rate": 6.443830756058259e-06,
      "loss": 0.7091,
      "step": 14735
    },
    {
      "epoch": 0.4245951708638276,
      "grad_norm": 0.7610281109809875,
      "learning_rate": 6.443384017395921e-06,
      "loss": 0.7316,
      "step": 14736
    },
    {
      "epoch": 0.4246239843254769,
      "grad_norm": 0.7523400187492371,
      "learning_rate": 6.44293726616315e-06,
      "loss": 0.7388,
      "step": 14737
    },
    {
      "epoch": 0.42465279778712617,
      "grad_norm": 0.7117711901664734,
      "learning_rate": 6.442490502363838e-06,
      "loss": 0.7443,
      "step": 14738
    },
    {
      "epoch": 0.4246816112487754,
      "grad_norm": 0.7649595141410828,
      "learning_rate": 6.442043726001876e-06,
      "loss": 0.7352,
      "step": 14739
    },
    {
      "epoch": 0.4247104247104247,
      "grad_norm": 0.7585142850875854,
      "learning_rate": 6.441596937081155e-06,
      "loss": 0.7139,
      "step": 14740
    },
    {
      "epoch": 0.424739238172074,
      "grad_norm": 0.7658657431602478,
      "learning_rate": 6.441150135605566e-06,
      "loss": 0.7252,
      "step": 14741
    },
    {
      "epoch": 0.4247680516337233,
      "grad_norm": 0.7498206496238708,
      "learning_rate": 6.440703321578999e-06,
      "loss": 0.707,
      "step": 14742
    },
    {
      "epoch": 0.42479686509537257,
      "grad_norm": 0.7844424247741699,
      "learning_rate": 6.440256495005348e-06,
      "loss": 0.7491,
      "step": 14743
    },
    {
      "epoch": 0.42482567855702186,
      "grad_norm": 0.7632704973220825,
      "learning_rate": 6.439809655888502e-06,
      "loss": 0.7358,
      "step": 14744
    },
    {
      "epoch": 0.42485449201867115,
      "grad_norm": 0.7733750939369202,
      "learning_rate": 6.439362804232352e-06,
      "loss": 0.7371,
      "step": 14745
    },
    {
      "epoch": 0.4248833054803204,
      "grad_norm": 0.7573368549346924,
      "learning_rate": 6.438915940040791e-06,
      "loss": 0.7154,
      "step": 14746
    },
    {
      "epoch": 0.4249121189419697,
      "grad_norm": 0.7578839659690857,
      "learning_rate": 6.4384690633177114e-06,
      "loss": 0.7358,
      "step": 14747
    },
    {
      "epoch": 0.42494093240361897,
      "grad_norm": 0.7622101902961731,
      "learning_rate": 6.438022174067003e-06,
      "loss": 0.7364,
      "step": 14748
    },
    {
      "epoch": 0.42496974586526826,
      "grad_norm": 0.7254498600959778,
      "learning_rate": 6.437575272292561e-06,
      "loss": 0.706,
      "step": 14749
    },
    {
      "epoch": 0.42499855932691755,
      "grad_norm": 0.8303992748260498,
      "learning_rate": 6.437128357998275e-06,
      "loss": 0.7387,
      "step": 14750
    },
    {
      "epoch": 0.42502737278856684,
      "grad_norm": 0.7569587230682373,
      "learning_rate": 6.436681431188035e-06,
      "loss": 0.7292,
      "step": 14751
    },
    {
      "epoch": 0.42505618625021613,
      "grad_norm": 0.7510020732879639,
      "learning_rate": 6.436234491865738e-06,
      "loss": 0.7246,
      "step": 14752
    },
    {
      "epoch": 0.42508499971186536,
      "grad_norm": 0.7741932272911072,
      "learning_rate": 6.435787540035274e-06,
      "loss": 0.7471,
      "step": 14753
    },
    {
      "epoch": 0.42511381317351465,
      "grad_norm": 0.7393023371696472,
      "learning_rate": 6.435340575700536e-06,
      "loss": 0.7191,
      "step": 14754
    },
    {
      "epoch": 0.42514262663516394,
      "grad_norm": 0.7348744869232178,
      "learning_rate": 6.434893598865417e-06,
      "loss": 0.7099,
      "step": 14755
    },
    {
      "epoch": 0.42517144009681324,
      "grad_norm": 0.7763263583183289,
      "learning_rate": 6.4344466095338075e-06,
      "loss": 0.7509,
      "step": 14756
    },
    {
      "epoch": 0.4252002535584625,
      "grad_norm": 0.7632871270179749,
      "learning_rate": 6.433999607709602e-06,
      "loss": 0.7214,
      "step": 14757
    },
    {
      "epoch": 0.4252290670201118,
      "grad_norm": 0.7654014825820923,
      "learning_rate": 6.433552593396694e-06,
      "loss": 0.7243,
      "step": 14758
    },
    {
      "epoch": 0.42525788048176105,
      "grad_norm": 0.7899818420410156,
      "learning_rate": 6.433105566598976e-06,
      "loss": 0.7417,
      "step": 14759
    },
    {
      "epoch": 0.42528669394341034,
      "grad_norm": 0.778091549873352,
      "learning_rate": 6.432658527320341e-06,
      "loss": 0.7196,
      "step": 14760
    },
    {
      "epoch": 0.42531550740505963,
      "grad_norm": 0.779940664768219,
      "learning_rate": 6.432211475564683e-06,
      "loss": 0.7312,
      "step": 14761
    },
    {
      "epoch": 0.4253443208667089,
      "grad_norm": 0.7476904392242432,
      "learning_rate": 6.431764411335895e-06,
      "loss": 0.7123,
      "step": 14762
    },
    {
      "epoch": 0.4253731343283582,
      "grad_norm": 0.7757673263549805,
      "learning_rate": 6.4313173346378685e-06,
      "loss": 0.74,
      "step": 14763
    },
    {
      "epoch": 0.4254019477900075,
      "grad_norm": 0.7707495093345642,
      "learning_rate": 6.4308702454744995e-06,
      "loss": 0.7314,
      "step": 14764
    },
    {
      "epoch": 0.4254307612516568,
      "grad_norm": 0.7487444877624512,
      "learning_rate": 6.4304231438496825e-06,
      "loss": 0.735,
      "step": 14765
    },
    {
      "epoch": 0.42545957471330603,
      "grad_norm": 0.7080214023590088,
      "learning_rate": 6.429976029767309e-06,
      "loss": 0.7109,
      "step": 14766
    },
    {
      "epoch": 0.4254883881749553,
      "grad_norm": 0.9010277390480042,
      "learning_rate": 6.4295289032312725e-06,
      "loss": 0.7288,
      "step": 14767
    },
    {
      "epoch": 0.4255172016366046,
      "grad_norm": 0.7937707901000977,
      "learning_rate": 6.4290817642454705e-06,
      "loss": 0.7387,
      "step": 14768
    },
    {
      "epoch": 0.4255460150982539,
      "grad_norm": 0.8017603158950806,
      "learning_rate": 6.428634612813792e-06,
      "loss": 0.731,
      "step": 14769
    },
    {
      "epoch": 0.4255748285599032,
      "grad_norm": 0.7348268032073975,
      "learning_rate": 6.428187448940136e-06,
      "loss": 0.7261,
      "step": 14770
    },
    {
      "epoch": 0.4256036420215525,
      "grad_norm": 0.7850468754768372,
      "learning_rate": 6.427740272628396e-06,
      "loss": 0.7203,
      "step": 14771
    },
    {
      "epoch": 0.4256324554832018,
      "grad_norm": 0.741317868232727,
      "learning_rate": 6.427293083882463e-06,
      "loss": 0.7377,
      "step": 14772
    },
    {
      "epoch": 0.425661268944851,
      "grad_norm": 0.7784586548805237,
      "learning_rate": 6.426845882706235e-06,
      "loss": 0.7351,
      "step": 14773
    },
    {
      "epoch": 0.4256900824065003,
      "grad_norm": 0.7569825053215027,
      "learning_rate": 6.426398669103606e-06,
      "loss": 0.7101,
      "step": 14774
    },
    {
      "epoch": 0.4257188958681496,
      "grad_norm": 0.7674072980880737,
      "learning_rate": 6.42595144307847e-06,
      "loss": 0.744,
      "step": 14775
    },
    {
      "epoch": 0.4257477093297989,
      "grad_norm": 0.7377945184707642,
      "learning_rate": 6.425504204634723e-06,
      "loss": 0.7203,
      "step": 14776
    },
    {
      "epoch": 0.4257765227914482,
      "grad_norm": 0.7734999060630798,
      "learning_rate": 6.425056953776259e-06,
      "loss": 0.7135,
      "step": 14777
    },
    {
      "epoch": 0.42580533625309747,
      "grad_norm": 0.777328372001648,
      "learning_rate": 6.424609690506972e-06,
      "loss": 0.703,
      "step": 14778
    },
    {
      "epoch": 0.42583414971474676,
      "grad_norm": 0.7382118105888367,
      "learning_rate": 6.424162414830761e-06,
      "loss": 0.718,
      "step": 14779
    },
    {
      "epoch": 0.425862963176396,
      "grad_norm": 0.7377414703369141,
      "learning_rate": 6.423715126751517e-06,
      "loss": 0.7587,
      "step": 14780
    },
    {
      "epoch": 0.4258917766380453,
      "grad_norm": 0.7924081087112427,
      "learning_rate": 6.4232678262731384e-06,
      "loss": 0.7368,
      "step": 14781
    },
    {
      "epoch": 0.4259205900996946,
      "grad_norm": 0.7442958354949951,
      "learning_rate": 6.42282051339952e-06,
      "loss": 0.7291,
      "step": 14782
    },
    {
      "epoch": 0.42594940356134386,
      "grad_norm": 0.8149773478507996,
      "learning_rate": 6.422373188134558e-06,
      "loss": 0.7503,
      "step": 14783
    },
    {
      "epoch": 0.42597821702299316,
      "grad_norm": 0.7278854846954346,
      "learning_rate": 6.421925850482147e-06,
      "loss": 0.6975,
      "step": 14784
    },
    {
      "epoch": 0.42600703048464245,
      "grad_norm": 0.7508648037910461,
      "learning_rate": 6.421478500446183e-06,
      "loss": 0.7235,
      "step": 14785
    },
    {
      "epoch": 0.4260358439462917,
      "grad_norm": 0.7442710995674133,
      "learning_rate": 6.421031138030563e-06,
      "loss": 0.7242,
      "step": 14786
    },
    {
      "epoch": 0.42606465740794097,
      "grad_norm": 0.7981765270233154,
      "learning_rate": 6.42058376323918e-06,
      "loss": 0.7244,
      "step": 14787
    },
    {
      "epoch": 0.42609347086959026,
      "grad_norm": 0.7811549305915833,
      "learning_rate": 6.4201363760759335e-06,
      "loss": 0.7257,
      "step": 14788
    },
    {
      "epoch": 0.42612228433123955,
      "grad_norm": 0.7856372594833374,
      "learning_rate": 6.419688976544719e-06,
      "loss": 0.7465,
      "step": 14789
    },
    {
      "epoch": 0.42615109779288884,
      "grad_norm": 0.8064274787902832,
      "learning_rate": 6.419241564649434e-06,
      "loss": 0.7412,
      "step": 14790
    },
    {
      "epoch": 0.42617991125453814,
      "grad_norm": 0.720162034034729,
      "learning_rate": 6.418794140393973e-06,
      "loss": 0.7213,
      "step": 14791
    },
    {
      "epoch": 0.4262087247161874,
      "grad_norm": 0.7746598720550537,
      "learning_rate": 6.418346703782232e-06,
      "loss": 0.7265,
      "step": 14792
    },
    {
      "epoch": 0.42623753817783666,
      "grad_norm": 4.16434907913208,
      "learning_rate": 6.417899254818109e-06,
      "loss": 0.7946,
      "step": 14793
    },
    {
      "epoch": 0.42626635163948595,
      "grad_norm": 0.7635318636894226,
      "learning_rate": 6.417451793505502e-06,
      "loss": 0.7178,
      "step": 14794
    },
    {
      "epoch": 0.42629516510113524,
      "grad_norm": 0.7481290102005005,
      "learning_rate": 6.417004319848307e-06,
      "loss": 0.7157,
      "step": 14795
    },
    {
      "epoch": 0.42632397856278453,
      "grad_norm": 0.7566304206848145,
      "learning_rate": 6.41655683385042e-06,
      "loss": 0.7237,
      "step": 14796
    },
    {
      "epoch": 0.4263527920244338,
      "grad_norm": 0.8163439035415649,
      "learning_rate": 6.416109335515739e-06,
      "loss": 0.7423,
      "step": 14797
    },
    {
      "epoch": 0.4263816054860831,
      "grad_norm": 0.7593400478363037,
      "learning_rate": 6.415661824848162e-06,
      "loss": 0.7374,
      "step": 14798
    },
    {
      "epoch": 0.4264104189477324,
      "grad_norm": 0.7496384382247925,
      "learning_rate": 6.415214301851584e-06,
      "loss": 0.7272,
      "step": 14799
    },
    {
      "epoch": 0.42643923240938164,
      "grad_norm": 0.7862865328788757,
      "learning_rate": 6.414766766529905e-06,
      "loss": 0.7037,
      "step": 14800
    },
    {
      "epoch": 0.42646804587103093,
      "grad_norm": 0.8425211310386658,
      "learning_rate": 6.414319218887023e-06,
      "loss": 0.7423,
      "step": 14801
    },
    {
      "epoch": 0.4264968593326802,
      "grad_norm": 0.7560219168663025,
      "learning_rate": 6.413871658926834e-06,
      "loss": 0.72,
      "step": 14802
    },
    {
      "epoch": 0.4265256727943295,
      "grad_norm": 0.7384374737739563,
      "learning_rate": 6.413424086653236e-06,
      "loss": 0.7184,
      "step": 14803
    },
    {
      "epoch": 0.4265544862559788,
      "grad_norm": 0.745919406414032,
      "learning_rate": 6.4129765020701275e-06,
      "loss": 0.704,
      "step": 14804
    },
    {
      "epoch": 0.4265832997176281,
      "grad_norm": 0.7693890929222107,
      "learning_rate": 6.412528905181404e-06,
      "loss": 0.712,
      "step": 14805
    },
    {
      "epoch": 0.4266121131792774,
      "grad_norm": 0.7733645439147949,
      "learning_rate": 6.412081295990967e-06,
      "loss": 0.7076,
      "step": 14806
    },
    {
      "epoch": 0.4266409266409266,
      "grad_norm": 0.7805889248847961,
      "learning_rate": 6.411633674502714e-06,
      "loss": 0.7447,
      "step": 14807
    },
    {
      "epoch": 0.4266697401025759,
      "grad_norm": 0.8413772583007812,
      "learning_rate": 6.411186040720543e-06,
      "loss": 0.7479,
      "step": 14808
    },
    {
      "epoch": 0.4266985535642252,
      "grad_norm": 0.7698777318000793,
      "learning_rate": 6.4107383946483525e-06,
      "loss": 0.7024,
      "step": 14809
    },
    {
      "epoch": 0.4267273670258745,
      "grad_norm": 0.7572746872901917,
      "learning_rate": 6.41029073629004e-06,
      "loss": 0.7368,
      "step": 14810
    },
    {
      "epoch": 0.4267561804875238,
      "grad_norm": 0.7608927488327026,
      "learning_rate": 6.409843065649504e-06,
      "loss": 0.7379,
      "step": 14811
    },
    {
      "epoch": 0.4267849939491731,
      "grad_norm": 0.7515569925308228,
      "learning_rate": 6.409395382730645e-06,
      "loss": 0.734,
      "step": 14812
    },
    {
      "epoch": 0.4268138074108223,
      "grad_norm": 0.7720732688903809,
      "learning_rate": 6.408947687537362e-06,
      "loss": 0.7575,
      "step": 14813
    },
    {
      "epoch": 0.4268426208724716,
      "grad_norm": 0.7949859499931335,
      "learning_rate": 6.408499980073552e-06,
      "loss": 0.7304,
      "step": 14814
    },
    {
      "epoch": 0.4268714343341209,
      "grad_norm": 0.7622661590576172,
      "learning_rate": 6.408052260343116e-06,
      "loss": 0.7462,
      "step": 14815
    },
    {
      "epoch": 0.4269002477957702,
      "grad_norm": 0.7928668856620789,
      "learning_rate": 6.407604528349952e-06,
      "loss": 0.7215,
      "step": 14816
    },
    {
      "epoch": 0.4269290612574195,
      "grad_norm": 0.7645274996757507,
      "learning_rate": 6.4071567840979595e-06,
      "loss": 0.7271,
      "step": 14817
    },
    {
      "epoch": 0.42695787471906876,
      "grad_norm": 0.7426854372024536,
      "learning_rate": 6.406709027591038e-06,
      "loss": 0.7297,
      "step": 14818
    },
    {
      "epoch": 0.42698668818071805,
      "grad_norm": 0.778925895690918,
      "learning_rate": 6.406261258833086e-06,
      "loss": 0.7115,
      "step": 14819
    },
    {
      "epoch": 0.4270155016423673,
      "grad_norm": 0.7493165731430054,
      "learning_rate": 6.405813477828007e-06,
      "loss": 0.724,
      "step": 14820
    },
    {
      "epoch": 0.4270443151040166,
      "grad_norm": 0.7755562663078308,
      "learning_rate": 6.405365684579696e-06,
      "loss": 0.7412,
      "step": 14821
    },
    {
      "epoch": 0.42707312856566587,
      "grad_norm": 0.7614396810531616,
      "learning_rate": 6.404917879092055e-06,
      "loss": 0.7251,
      "step": 14822
    },
    {
      "epoch": 0.42710194202731516,
      "grad_norm": 0.7756987810134888,
      "learning_rate": 6.404470061368984e-06,
      "loss": 0.7539,
      "step": 14823
    },
    {
      "epoch": 0.42713075548896445,
      "grad_norm": 0.7385587096214294,
      "learning_rate": 6.404022231414382e-06,
      "loss": 0.7469,
      "step": 14824
    },
    {
      "epoch": 0.42715956895061374,
      "grad_norm": 3.915477752685547,
      "learning_rate": 6.40357438923215e-06,
      "loss": 0.7523,
      "step": 14825
    },
    {
      "epoch": 0.42718838241226303,
      "grad_norm": 0.7682848572731018,
      "learning_rate": 6.40312653482619e-06,
      "loss": 0.7498,
      "step": 14826
    },
    {
      "epoch": 0.42721719587391227,
      "grad_norm": 0.8024984002113342,
      "learning_rate": 6.4026786682003985e-06,
      "loss": 0.71,
      "step": 14827
    },
    {
      "epoch": 0.42724600933556156,
      "grad_norm": 0.734535276889801,
      "learning_rate": 6.402230789358679e-06,
      "loss": 0.7233,
      "step": 14828
    },
    {
      "epoch": 0.42727482279721085,
      "grad_norm": 0.7473489046096802,
      "learning_rate": 6.401782898304929e-06,
      "loss": 0.7428,
      "step": 14829
    },
    {
      "epoch": 0.42730363625886014,
      "grad_norm": 0.7603120803833008,
      "learning_rate": 6.401334995043052e-06,
      "loss": 0.7392,
      "step": 14830
    },
    {
      "epoch": 0.42733244972050943,
      "grad_norm": 0.7981840372085571,
      "learning_rate": 6.40088707957695e-06,
      "loss": 0.7451,
      "step": 14831
    },
    {
      "epoch": 0.4273612631821587,
      "grad_norm": 0.7890772223472595,
      "learning_rate": 6.400439151910521e-06,
      "loss": 0.7564,
      "step": 14832
    },
    {
      "epoch": 0.427390076643808,
      "grad_norm": 0.7437437772750854,
      "learning_rate": 6.399991212047667e-06,
      "loss": 0.7576,
      "step": 14833
    },
    {
      "epoch": 0.42741889010545725,
      "grad_norm": 2.5040621757507324,
      "learning_rate": 6.399543259992288e-06,
      "loss": 0.7326,
      "step": 14834
    },
    {
      "epoch": 0.42744770356710654,
      "grad_norm": 0.7641273736953735,
      "learning_rate": 6.399095295748286e-06,
      "loss": 0.7346,
      "step": 14835
    },
    {
      "epoch": 0.42747651702875583,
      "grad_norm": 0.7515361309051514,
      "learning_rate": 6.398647319319564e-06,
      "loss": 0.7069,
      "step": 14836
    },
    {
      "epoch": 0.4275053304904051,
      "grad_norm": 0.7402758598327637,
      "learning_rate": 6.39819933071002e-06,
      "loss": 0.7318,
      "step": 14837
    },
    {
      "epoch": 0.4275341439520544,
      "grad_norm": 0.7656605839729309,
      "learning_rate": 6.39775132992356e-06,
      "loss": 0.7104,
      "step": 14838
    },
    {
      "epoch": 0.4275629574137037,
      "grad_norm": 0.7522373795509338,
      "learning_rate": 6.3973033169640805e-06,
      "loss": 0.7231,
      "step": 14839
    },
    {
      "epoch": 0.42759177087535294,
      "grad_norm": 0.7934535145759583,
      "learning_rate": 6.396855291835489e-06,
      "loss": 0.719,
      "step": 14840
    },
    {
      "epoch": 0.42762058433700223,
      "grad_norm": 0.7693396210670471,
      "learning_rate": 6.3964072545416815e-06,
      "loss": 0.76,
      "step": 14841
    },
    {
      "epoch": 0.4276493977986515,
      "grad_norm": 0.7326162457466125,
      "learning_rate": 6.395959205086564e-06,
      "loss": 0.7172,
      "step": 14842
    },
    {
      "epoch": 0.4276782112603008,
      "grad_norm": 0.7576520442962646,
      "learning_rate": 6.395511143474037e-06,
      "loss": 0.727,
      "step": 14843
    },
    {
      "epoch": 0.4277070247219501,
      "grad_norm": 0.7459388971328735,
      "learning_rate": 6.395063069708003e-06,
      "loss": 0.7158,
      "step": 14844
    },
    {
      "epoch": 0.4277358381835994,
      "grad_norm": 0.7303087711334229,
      "learning_rate": 6.394614983792364e-06,
      "loss": 0.7093,
      "step": 14845
    },
    {
      "epoch": 0.4277646516452487,
      "grad_norm": 0.7771157622337341,
      "learning_rate": 6.394166885731024e-06,
      "loss": 0.7458,
      "step": 14846
    },
    {
      "epoch": 0.4277934651068979,
      "grad_norm": 0.7161285281181335,
      "learning_rate": 6.393718775527881e-06,
      "loss": 0.7106,
      "step": 14847
    },
    {
      "epoch": 0.4278222785685472,
      "grad_norm": 0.7330728769302368,
      "learning_rate": 6.393270653186844e-06,
      "loss": 0.7101,
      "step": 14848
    },
    {
      "epoch": 0.4278510920301965,
      "grad_norm": 0.7470408082008362,
      "learning_rate": 6.392822518711811e-06,
      "loss": 0.7259,
      "step": 14849
    },
    {
      "epoch": 0.4278799054918458,
      "grad_norm": 0.7306782007217407,
      "learning_rate": 6.392374372106687e-06,
      "loss": 0.6936,
      "step": 14850
    },
    {
      "epoch": 0.4279087189534951,
      "grad_norm": 0.7676200270652771,
      "learning_rate": 6.3919262133753735e-06,
      "loss": 0.7401,
      "step": 14851
    },
    {
      "epoch": 0.4279375324151444,
      "grad_norm": 0.7231429219245911,
      "learning_rate": 6.3914780425217735e-06,
      "loss": 0.7162,
      "step": 14852
    },
    {
      "epoch": 0.42796634587679366,
      "grad_norm": 0.7435022592544556,
      "learning_rate": 6.391029859549791e-06,
      "loss": 0.7161,
      "step": 14853
    },
    {
      "epoch": 0.4279951593384429,
      "grad_norm": 0.7485002279281616,
      "learning_rate": 6.3905816644633285e-06,
      "loss": 0.748,
      "step": 14854
    },
    {
      "epoch": 0.4280239728000922,
      "grad_norm": 0.7248014807701111,
      "learning_rate": 6.390133457266291e-06,
      "loss": 0.7315,
      "step": 14855
    },
    {
      "epoch": 0.4280527862617415,
      "grad_norm": 0.7650557160377502,
      "learning_rate": 6.38968523796258e-06,
      "loss": 0.7453,
      "step": 14856
    },
    {
      "epoch": 0.42808159972339077,
      "grad_norm": 0.7392159700393677,
      "learning_rate": 6.3892370065560995e-06,
      "loss": 0.6983,
      "step": 14857
    },
    {
      "epoch": 0.42811041318504006,
      "grad_norm": 0.7591854333877563,
      "learning_rate": 6.388788763050753e-06,
      "loss": 0.7273,
      "step": 14858
    },
    {
      "epoch": 0.42813922664668935,
      "grad_norm": 0.7532356977462769,
      "learning_rate": 6.3883405074504445e-06,
      "loss": 0.7316,
      "step": 14859
    },
    {
      "epoch": 0.42816804010833864,
      "grad_norm": 0.7471120953559875,
      "learning_rate": 6.387892239759079e-06,
      "loss": 0.7328,
      "step": 14860
    },
    {
      "epoch": 0.4281968535699879,
      "grad_norm": 0.7649021744728088,
      "learning_rate": 6.387443959980559e-06,
      "loss": 0.7383,
      "step": 14861
    },
    {
      "epoch": 0.42822566703163717,
      "grad_norm": 0.7533407211303711,
      "learning_rate": 6.38699566811879e-06,
      "loss": 0.7357,
      "step": 14862
    },
    {
      "epoch": 0.42825448049328646,
      "grad_norm": 0.7464460730552673,
      "learning_rate": 6.386547364177675e-06,
      "loss": 0.7338,
      "step": 14863
    },
    {
      "epoch": 0.42828329395493575,
      "grad_norm": 0.7481254935264587,
      "learning_rate": 6.386099048161119e-06,
      "loss": 0.7257,
      "step": 14864
    },
    {
      "epoch": 0.42831210741658504,
      "grad_norm": 0.7443568110466003,
      "learning_rate": 6.385650720073023e-06,
      "loss": 0.7034,
      "step": 14865
    },
    {
      "epoch": 0.42834092087823433,
      "grad_norm": 0.7493339776992798,
      "learning_rate": 6.385202379917297e-06,
      "loss": 0.7205,
      "step": 14866
    },
    {
      "epoch": 0.42836973433988357,
      "grad_norm": 0.753596842288971,
      "learning_rate": 6.384754027697843e-06,
      "loss": 0.7353,
      "step": 14867
    },
    {
      "epoch": 0.42839854780153286,
      "grad_norm": 0.7346935272216797,
      "learning_rate": 6.3843056634185655e-06,
      "loss": 0.6849,
      "step": 14868
    },
    {
      "epoch": 0.42842736126318215,
      "grad_norm": 0.7786039710044861,
      "learning_rate": 6.383857287083368e-06,
      "loss": 0.7533,
      "step": 14869
    },
    {
      "epoch": 0.42845617472483144,
      "grad_norm": 0.7781940698623657,
      "learning_rate": 6.383408898696158e-06,
      "loss": 0.7486,
      "step": 14870
    },
    {
      "epoch": 0.42848498818648073,
      "grad_norm": 0.7311420440673828,
      "learning_rate": 6.38296049826084e-06,
      "loss": 0.7386,
      "step": 14871
    },
    {
      "epoch": 0.42851380164813,
      "grad_norm": 0.7507746815681458,
      "learning_rate": 6.382512085781318e-06,
      "loss": 0.7273,
      "step": 14872
    },
    {
      "epoch": 0.4285426151097793,
      "grad_norm": 0.7630894184112549,
      "learning_rate": 6.3820636612614965e-06,
      "loss": 0.719,
      "step": 14873
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.7375978231430054,
      "learning_rate": 6.381615224705283e-06,
      "loss": 0.7192,
      "step": 14874
    },
    {
      "epoch": 0.42860024203307784,
      "grad_norm": 0.7615840435028076,
      "learning_rate": 6.381166776116581e-06,
      "loss": 0.7299,
      "step": 14875
    },
    {
      "epoch": 0.42862905549472713,
      "grad_norm": 0.7282421588897705,
      "learning_rate": 6.380718315499299e-06,
      "loss": 0.6991,
      "step": 14876
    },
    {
      "epoch": 0.4286578689563764,
      "grad_norm": 0.747689962387085,
      "learning_rate": 6.380269842857338e-06,
      "loss": 0.717,
      "step": 14877
    },
    {
      "epoch": 0.4286866824180257,
      "grad_norm": 0.7381802797317505,
      "learning_rate": 6.379821358194609e-06,
      "loss": 0.7074,
      "step": 14878
    },
    {
      "epoch": 0.428715495879675,
      "grad_norm": 0.7716342806816101,
      "learning_rate": 6.379372861515014e-06,
      "loss": 0.7259,
      "step": 14879
    },
    {
      "epoch": 0.4287443093413243,
      "grad_norm": 0.7680320143699646,
      "learning_rate": 6.378924352822462e-06,
      "loss": 0.7247,
      "step": 14880
    },
    {
      "epoch": 0.4287731228029735,
      "grad_norm": 0.7778452038764954,
      "learning_rate": 6.378475832120856e-06,
      "loss": 0.7262,
      "step": 14881
    },
    {
      "epoch": 0.4288019362646228,
      "grad_norm": 0.7157253623008728,
      "learning_rate": 6.378027299414103e-06,
      "loss": 0.6987,
      "step": 14882
    },
    {
      "epoch": 0.4288307497262721,
      "grad_norm": 0.8042288422584534,
      "learning_rate": 6.37757875470611e-06,
      "loss": 0.7303,
      "step": 14883
    },
    {
      "epoch": 0.4288595631879214,
      "grad_norm": 0.7065703272819519,
      "learning_rate": 6.3771301980007835e-06,
      "loss": 0.7197,
      "step": 14884
    },
    {
      "epoch": 0.4288883766495707,
      "grad_norm": 0.7436899542808533,
      "learning_rate": 6.3766816293020295e-06,
      "loss": 0.7282,
      "step": 14885
    },
    {
      "epoch": 0.42891719011122,
      "grad_norm": 0.7638518810272217,
      "learning_rate": 6.3762330486137555e-06,
      "loss": 0.744,
      "step": 14886
    },
    {
      "epoch": 0.42894600357286927,
      "grad_norm": 0.7677706480026245,
      "learning_rate": 6.375784455939866e-06,
      "loss": 0.7259,
      "step": 14887
    },
    {
      "epoch": 0.4289748170345185,
      "grad_norm": 0.7473864555358887,
      "learning_rate": 6.37533585128427e-06,
      "loss": 0.7583,
      "step": 14888
    },
    {
      "epoch": 0.4290036304961678,
      "grad_norm": 0.7407164573669434,
      "learning_rate": 6.374887234650873e-06,
      "loss": 0.7362,
      "step": 14889
    },
    {
      "epoch": 0.4290324439578171,
      "grad_norm": 0.741328775882721,
      "learning_rate": 6.374438606043582e-06,
      "loss": 0.7272,
      "step": 14890
    },
    {
      "epoch": 0.4290612574194664,
      "grad_norm": 0.7463662028312683,
      "learning_rate": 6.373989965466306e-06,
      "loss": 0.7424,
      "step": 14891
    },
    {
      "epoch": 0.42909007088111567,
      "grad_norm": 0.7529745101928711,
      "learning_rate": 6.373541312922951e-06,
      "loss": 0.7436,
      "step": 14892
    },
    {
      "epoch": 0.42911888434276496,
      "grad_norm": 0.7645828127861023,
      "learning_rate": 6.373092648417424e-06,
      "loss": 0.7393,
      "step": 14893
    },
    {
      "epoch": 0.4291476978044142,
      "grad_norm": 0.7521879076957703,
      "learning_rate": 6.372643971953631e-06,
      "loss": 0.7308,
      "step": 14894
    },
    {
      "epoch": 0.4291765112660635,
      "grad_norm": 0.744902491569519,
      "learning_rate": 6.372195283535483e-06,
      "loss": 0.7307,
      "step": 14895
    },
    {
      "epoch": 0.4292053247277128,
      "grad_norm": 0.7386431097984314,
      "learning_rate": 6.3717465831668846e-06,
      "loss": 0.7351,
      "step": 14896
    },
    {
      "epoch": 0.42923413818936207,
      "grad_norm": 0.7603155970573425,
      "learning_rate": 6.3712978708517446e-06,
      "loss": 0.7189,
      "step": 14897
    },
    {
      "epoch": 0.42926295165101136,
      "grad_norm": 0.7938945889472961,
      "learning_rate": 6.370849146593973e-06,
      "loss": 0.7373,
      "step": 14898
    },
    {
      "epoch": 0.42929176511266065,
      "grad_norm": 0.7365541458129883,
      "learning_rate": 6.370400410397474e-06,
      "loss": 0.7275,
      "step": 14899
    },
    {
      "epoch": 0.42932057857430994,
      "grad_norm": 0.7577306032180786,
      "learning_rate": 6.369951662266158e-06,
      "loss": 0.742,
      "step": 14900
    },
    {
      "epoch": 0.4293493920359592,
      "grad_norm": 0.7379314303398132,
      "learning_rate": 6.369502902203931e-06,
      "loss": 0.7143,
      "step": 14901
    },
    {
      "epoch": 0.42937820549760847,
      "grad_norm": 0.7522093057632446,
      "learning_rate": 6.369054130214705e-06,
      "loss": 0.7457,
      "step": 14902
    },
    {
      "epoch": 0.42940701895925776,
      "grad_norm": 0.759132444858551,
      "learning_rate": 6.368605346302385e-06,
      "loss": 0.7209,
      "step": 14903
    },
    {
      "epoch": 0.42943583242090705,
      "grad_norm": 0.7414523363113403,
      "learning_rate": 6.36815655047088e-06,
      "loss": 0.7439,
      "step": 14904
    },
    {
      "epoch": 0.42946464588255634,
      "grad_norm": 0.7581375241279602,
      "learning_rate": 6.3677077427241e-06,
      "loss": 0.7378,
      "step": 14905
    },
    {
      "epoch": 0.42949345934420563,
      "grad_norm": 0.7573219537734985,
      "learning_rate": 6.367258923065952e-06,
      "loss": 0.7326,
      "step": 14906
    },
    {
      "epoch": 0.4295222728058549,
      "grad_norm": 0.7615821361541748,
      "learning_rate": 6.366810091500345e-06,
      "loss": 0.7188,
      "step": 14907
    },
    {
      "epoch": 0.42955108626750416,
      "grad_norm": 0.7463173270225525,
      "learning_rate": 6.36636124803119e-06,
      "loss": 0.7279,
      "step": 14908
    },
    {
      "epoch": 0.42957989972915345,
      "grad_norm": 0.7531698942184448,
      "learning_rate": 6.365912392662394e-06,
      "loss": 0.7596,
      "step": 14909
    },
    {
      "epoch": 0.42960871319080274,
      "grad_norm": 0.731533408164978,
      "learning_rate": 6.365463525397866e-06,
      "loss": 0.6994,
      "step": 14910
    },
    {
      "epoch": 0.42963752665245203,
      "grad_norm": 0.7440407276153564,
      "learning_rate": 6.365014646241515e-06,
      "loss": 0.7258,
      "step": 14911
    },
    {
      "epoch": 0.4296663401141013,
      "grad_norm": 0.7800418138504028,
      "learning_rate": 6.364565755197253e-06,
      "loss": 0.7263,
      "step": 14912
    },
    {
      "epoch": 0.4296951535757506,
      "grad_norm": 0.7292817234992981,
      "learning_rate": 6.364116852268984e-06,
      "loss": 0.7247,
      "step": 14913
    },
    {
      "epoch": 0.42972396703739985,
      "grad_norm": 0.7297229170799255,
      "learning_rate": 6.363667937460624e-06,
      "loss": 0.7404,
      "step": 14914
    },
    {
      "epoch": 0.42975278049904914,
      "grad_norm": 0.7547641396522522,
      "learning_rate": 6.3632190107760775e-06,
      "loss": 0.7073,
      "step": 14915
    },
    {
      "epoch": 0.4297815939606984,
      "grad_norm": 0.7408095002174377,
      "learning_rate": 6.362770072219257e-06,
      "loss": 0.7405,
      "step": 14916
    },
    {
      "epoch": 0.4298104074223477,
      "grad_norm": 0.7197651863098145,
      "learning_rate": 6.362321121794071e-06,
      "loss": 0.7273,
      "step": 14917
    },
    {
      "epoch": 0.429839220883997,
      "grad_norm": 0.7796767354011536,
      "learning_rate": 6.3618721595044305e-06,
      "loss": 0.7245,
      "step": 14918
    },
    {
      "epoch": 0.4298680343456463,
      "grad_norm": 0.7515218257904053,
      "learning_rate": 6.361423185354244e-06,
      "loss": 0.7216,
      "step": 14919
    },
    {
      "epoch": 0.4298968478072956,
      "grad_norm": 0.7499987483024597,
      "learning_rate": 6.360974199347424e-06,
      "loss": 0.7543,
      "step": 14920
    },
    {
      "epoch": 0.4299256612689448,
      "grad_norm": 0.7459003925323486,
      "learning_rate": 6.360525201487878e-06,
      "loss": 0.7313,
      "step": 14921
    },
    {
      "epoch": 0.4299544747305941,
      "grad_norm": 0.7131929993629456,
      "learning_rate": 6.3600761917795185e-06,
      "loss": 0.6967,
      "step": 14922
    },
    {
      "epoch": 0.4299832881922434,
      "grad_norm": 0.7510853409767151,
      "learning_rate": 6.3596271702262545e-06,
      "loss": 0.7045,
      "step": 14923
    },
    {
      "epoch": 0.4300121016538927,
      "grad_norm": 0.7565658092498779,
      "learning_rate": 6.359178136831996e-06,
      "loss": 0.7376,
      "step": 14924
    },
    {
      "epoch": 0.430040915115542,
      "grad_norm": 0.7463688850402832,
      "learning_rate": 6.358729091600655e-06,
      "loss": 0.7043,
      "step": 14925
    },
    {
      "epoch": 0.4300697285771913,
      "grad_norm": 0.7313911318778992,
      "learning_rate": 6.358280034536141e-06,
      "loss": 0.7033,
      "step": 14926
    },
    {
      "epoch": 0.43009854203884057,
      "grad_norm": 0.7408479452133179,
      "learning_rate": 6.3578309656423685e-06,
      "loss": 0.728,
      "step": 14927
    },
    {
      "epoch": 0.4301273555004898,
      "grad_norm": 0.7770610451698303,
      "learning_rate": 6.357381884923244e-06,
      "loss": 0.7259,
      "step": 14928
    },
    {
      "epoch": 0.4301561689621391,
      "grad_norm": 0.7173165082931519,
      "learning_rate": 6.356932792382681e-06,
      "loss": 0.7219,
      "step": 14929
    },
    {
      "epoch": 0.4301849824237884,
      "grad_norm": 0.7524070739746094,
      "learning_rate": 6.356483688024589e-06,
      "loss": 0.7004,
      "step": 14930
    },
    {
      "epoch": 0.4302137958854377,
      "grad_norm": 0.7659949660301208,
      "learning_rate": 6.356034571852879e-06,
      "loss": 0.7249,
      "step": 14931
    },
    {
      "epoch": 0.43024260934708697,
      "grad_norm": 0.749024510383606,
      "learning_rate": 6.355585443871465e-06,
      "loss": 0.7281,
      "step": 14932
    },
    {
      "epoch": 0.43027142280873626,
      "grad_norm": 0.7784853577613831,
      "learning_rate": 6.355136304084256e-06,
      "loss": 0.7353,
      "step": 14933
    },
    {
      "epoch": 0.43030023627038555,
      "grad_norm": 0.7335349321365356,
      "learning_rate": 6.354687152495165e-06,
      "loss": 0.7325,
      "step": 14934
    },
    {
      "epoch": 0.4303290497320348,
      "grad_norm": 0.7355932593345642,
      "learning_rate": 6.3542379891081014e-06,
      "loss": 0.7101,
      "step": 14935
    },
    {
      "epoch": 0.4303578631936841,
      "grad_norm": 0.7420572638511658,
      "learning_rate": 6.353788813926981e-06,
      "loss": 0.697,
      "step": 14936
    },
    {
      "epoch": 0.43038667665533337,
      "grad_norm": 0.754419207572937,
      "learning_rate": 6.3533396269557125e-06,
      "loss": 0.7552,
      "step": 14937
    },
    {
      "epoch": 0.43041549011698266,
      "grad_norm": 0.7337173819541931,
      "learning_rate": 6.3528904281982085e-06,
      "loss": 0.6845,
      "step": 14938
    },
    {
      "epoch": 0.43044430357863195,
      "grad_norm": 0.7220969200134277,
      "learning_rate": 6.35244121765838e-06,
      "loss": 0.737,
      "step": 14939
    },
    {
      "epoch": 0.43047311704028124,
      "grad_norm": 0.7331954836845398,
      "learning_rate": 6.351991995340143e-06,
      "loss": 0.7079,
      "step": 14940
    },
    {
      "epoch": 0.4305019305019305,
      "grad_norm": 0.7328211665153503,
      "learning_rate": 6.351542761247407e-06,
      "loss": 0.7194,
      "step": 14941
    },
    {
      "epoch": 0.43053074396357977,
      "grad_norm": 0.7668646574020386,
      "learning_rate": 6.351093515384083e-06,
      "loss": 0.7397,
      "step": 14942
    },
    {
      "epoch": 0.43055955742522906,
      "grad_norm": 0.798556387424469,
      "learning_rate": 6.350644257754085e-06,
      "loss": 0.736,
      "step": 14943
    },
    {
      "epoch": 0.43058837088687835,
      "grad_norm": 0.7545557022094727,
      "learning_rate": 6.350194988361327e-06,
      "loss": 0.7457,
      "step": 14944
    },
    {
      "epoch": 0.43061718434852764,
      "grad_norm": 0.7283338904380798,
      "learning_rate": 6.3497457072097205e-06,
      "loss": 0.6904,
      "step": 14945
    },
    {
      "epoch": 0.43064599781017693,
      "grad_norm": 0.7609022259712219,
      "learning_rate": 6.349296414303176e-06,
      "loss": 0.7363,
      "step": 14946
    },
    {
      "epoch": 0.4306748112718262,
      "grad_norm": 0.7418211102485657,
      "learning_rate": 6.348847109645611e-06,
      "loss": 0.7365,
      "step": 14947
    },
    {
      "epoch": 0.43070362473347545,
      "grad_norm": 0.7320227026939392,
      "learning_rate": 6.348397793240934e-06,
      "loss": 0.7033,
      "step": 14948
    },
    {
      "epoch": 0.43073243819512474,
      "grad_norm": 0.7562380433082581,
      "learning_rate": 6.347948465093062e-06,
      "loss": 0.7455,
      "step": 14949
    },
    {
      "epoch": 0.43076125165677404,
      "grad_norm": 0.7486295700073242,
      "learning_rate": 6.347499125205905e-06,
      "loss": 0.7385,
      "step": 14950
    },
    {
      "epoch": 0.4307900651184233,
      "grad_norm": 0.7702192068099976,
      "learning_rate": 6.347049773583378e-06,
      "loss": 0.728,
      "step": 14951
    },
    {
      "epoch": 0.4308188785800726,
      "grad_norm": 0.7443729043006897,
      "learning_rate": 6.346600410229394e-06,
      "loss": 0.7327,
      "step": 14952
    },
    {
      "epoch": 0.4308476920417219,
      "grad_norm": 0.7576589584350586,
      "learning_rate": 6.346151035147866e-06,
      "loss": 0.7517,
      "step": 14953
    },
    {
      "epoch": 0.4308765055033712,
      "grad_norm": 0.7557307481765747,
      "learning_rate": 6.345701648342709e-06,
      "loss": 0.7367,
      "step": 14954
    },
    {
      "epoch": 0.43090531896502043,
      "grad_norm": 0.776191234588623,
      "learning_rate": 6.345252249817836e-06,
      "loss": 0.7376,
      "step": 14955
    },
    {
      "epoch": 0.4309341324266697,
      "grad_norm": 0.807257890701294,
      "learning_rate": 6.34480283957716e-06,
      "loss": 0.7626,
      "step": 14956
    },
    {
      "epoch": 0.430962945888319,
      "grad_norm": 0.7450563907623291,
      "learning_rate": 6.344353417624596e-06,
      "loss": 0.7222,
      "step": 14957
    },
    {
      "epoch": 0.4309917593499683,
      "grad_norm": 0.7409473061561584,
      "learning_rate": 6.343903983964057e-06,
      "loss": 0.7299,
      "step": 14958
    },
    {
      "epoch": 0.4310205728116176,
      "grad_norm": 0.7368948459625244,
      "learning_rate": 6.343454538599458e-06,
      "loss": 0.7328,
      "step": 14959
    },
    {
      "epoch": 0.4310493862732669,
      "grad_norm": 0.7984203696250916,
      "learning_rate": 6.343005081534714e-06,
      "loss": 0.7438,
      "step": 14960
    },
    {
      "epoch": 0.4310781997349162,
      "grad_norm": 0.7711316347122192,
      "learning_rate": 6.342555612773736e-06,
      "loss": 0.7334,
      "step": 14961
    },
    {
      "epoch": 0.4311070131965654,
      "grad_norm": 0.7398523092269897,
      "learning_rate": 6.342106132320442e-06,
      "loss": 0.7212,
      "step": 14962
    },
    {
      "epoch": 0.4311358266582147,
      "grad_norm": 0.7490386366844177,
      "learning_rate": 6.341656640178746e-06,
      "loss": 0.7336,
      "step": 14963
    },
    {
      "epoch": 0.431164640119864,
      "grad_norm": 0.7798791527748108,
      "learning_rate": 6.3412071363525605e-06,
      "loss": 0.7376,
      "step": 14964
    },
    {
      "epoch": 0.4311934535815133,
      "grad_norm": 0.7441942095756531,
      "learning_rate": 6.340757620845803e-06,
      "loss": 0.7421,
      "step": 14965
    },
    {
      "epoch": 0.4312222670431626,
      "grad_norm": 0.7601637244224548,
      "learning_rate": 6.340308093662386e-06,
      "loss": 0.7327,
      "step": 14966
    },
    {
      "epoch": 0.43125108050481187,
      "grad_norm": 0.7670607566833496,
      "learning_rate": 6.339858554806224e-06,
      "loss": 0.7274,
      "step": 14967
    },
    {
      "epoch": 0.4312798939664611,
      "grad_norm": 0.773454487323761,
      "learning_rate": 6.339409004281236e-06,
      "loss": 0.7229,
      "step": 14968
    },
    {
      "epoch": 0.4313087074281104,
      "grad_norm": 0.7152255177497864,
      "learning_rate": 6.338959442091333e-06,
      "loss": 0.7156,
      "step": 14969
    },
    {
      "epoch": 0.4313375208897597,
      "grad_norm": 0.7585921883583069,
      "learning_rate": 6.338509868240432e-06,
      "loss": 0.7041,
      "step": 14970
    },
    {
      "epoch": 0.431366334351409,
      "grad_norm": 0.7575915455818176,
      "learning_rate": 6.338060282732448e-06,
      "loss": 0.7069,
      "step": 14971
    },
    {
      "epoch": 0.43139514781305827,
      "grad_norm": 0.722586452960968,
      "learning_rate": 6.337610685571296e-06,
      "loss": 0.7194,
      "step": 14972
    },
    {
      "epoch": 0.43142396127470756,
      "grad_norm": 0.7305258512496948,
      "learning_rate": 6.337161076760893e-06,
      "loss": 0.7288,
      "step": 14973
    },
    {
      "epoch": 0.43145277473635685,
      "grad_norm": 0.7429549694061279,
      "learning_rate": 6.336711456305154e-06,
      "loss": 0.7178,
      "step": 14974
    },
    {
      "epoch": 0.4314815881980061,
      "grad_norm": 0.7453470826148987,
      "learning_rate": 6.336261824207992e-06,
      "loss": 0.7208,
      "step": 14975
    },
    {
      "epoch": 0.4315104016596554,
      "grad_norm": 0.7419823408126831,
      "learning_rate": 6.335812180473326e-06,
      "loss": 0.714,
      "step": 14976
    },
    {
      "epoch": 0.43153921512130466,
      "grad_norm": 0.7521888017654419,
      "learning_rate": 6.3353625251050734e-06,
      "loss": 0.7129,
      "step": 14977
    },
    {
      "epoch": 0.43156802858295396,
      "grad_norm": 0.7226569652557373,
      "learning_rate": 6.334912858107147e-06,
      "loss": 0.7498,
      "step": 14978
    },
    {
      "epoch": 0.43159684204460325,
      "grad_norm": 0.7733092308044434,
      "learning_rate": 6.334463179483463e-06,
      "loss": 0.7477,
      "step": 14979
    },
    {
      "epoch": 0.43162565550625254,
      "grad_norm": 0.7396924495697021,
      "learning_rate": 6.33401348923794e-06,
      "loss": 0.7638,
      "step": 14980
    },
    {
      "epoch": 0.4316544689679018,
      "grad_norm": 0.734826385974884,
      "learning_rate": 6.333563787374493e-06,
      "loss": 0.7332,
      "step": 14981
    },
    {
      "epoch": 0.43168328242955106,
      "grad_norm": 0.7816147208213806,
      "learning_rate": 6.333114073897038e-06,
      "loss": 0.7288,
      "step": 14982
    },
    {
      "epoch": 0.43171209589120035,
      "grad_norm": 0.7176358699798584,
      "learning_rate": 6.332664348809492e-06,
      "loss": 0.7181,
      "step": 14983
    },
    {
      "epoch": 0.43174090935284964,
      "grad_norm": 0.7184991240501404,
      "learning_rate": 6.332214612115771e-06,
      "loss": 0.7238,
      "step": 14984
    },
    {
      "epoch": 0.43176972281449894,
      "grad_norm": 0.7509245872497559,
      "learning_rate": 6.3317648638197934e-06,
      "loss": 0.7454,
      "step": 14985
    },
    {
      "epoch": 0.4317985362761482,
      "grad_norm": 0.7359778881072998,
      "learning_rate": 6.331315103925475e-06,
      "loss": 0.7285,
      "step": 14986
    },
    {
      "epoch": 0.4318273497377975,
      "grad_norm": 0.7334901690483093,
      "learning_rate": 6.330865332436733e-06,
      "loss": 0.7234,
      "step": 14987
    },
    {
      "epoch": 0.4318561631994468,
      "grad_norm": 0.7315958738327026,
      "learning_rate": 6.330415549357485e-06,
      "loss": 0.7125,
      "step": 14988
    },
    {
      "epoch": 0.43188497666109604,
      "grad_norm": 0.7308953404426575,
      "learning_rate": 6.329965754691647e-06,
      "loss": 0.7385,
      "step": 14989
    },
    {
      "epoch": 0.43191379012274533,
      "grad_norm": 0.7366459369659424,
      "learning_rate": 6.329515948443135e-06,
      "loss": 0.7198,
      "step": 14990
    },
    {
      "epoch": 0.4319426035843946,
      "grad_norm": 0.7958762049674988,
      "learning_rate": 6.329066130615871e-06,
      "loss": 0.747,
      "step": 14991
    },
    {
      "epoch": 0.4319714170460439,
      "grad_norm": 0.7572961449623108,
      "learning_rate": 6.328616301213768e-06,
      "loss": 0.7264,
      "step": 14992
    },
    {
      "epoch": 0.4320002305076932,
      "grad_norm": 0.7532553672790527,
      "learning_rate": 6.328166460240746e-06,
      "loss": 0.7318,
      "step": 14993
    },
    {
      "epoch": 0.4320290439693425,
      "grad_norm": 0.7745027542114258,
      "learning_rate": 6.327716607700719e-06,
      "loss": 0.7381,
      "step": 14994
    },
    {
      "epoch": 0.43205785743099173,
      "grad_norm": 0.7706671357154846,
      "learning_rate": 6.32726674359761e-06,
      "loss": 0.7425,
      "step": 14995
    },
    {
      "epoch": 0.432086670892641,
      "grad_norm": 0.7464802265167236,
      "learning_rate": 6.326816867935334e-06,
      "loss": 0.7244,
      "step": 14996
    },
    {
      "epoch": 0.4321154843542903,
      "grad_norm": 0.7630820870399475,
      "learning_rate": 6.326366980717809e-06,
      "loss": 0.7116,
      "step": 14997
    },
    {
      "epoch": 0.4321442978159396,
      "grad_norm": 0.7487971782684326,
      "learning_rate": 6.325917081948954e-06,
      "loss": 0.7126,
      "step": 14998
    },
    {
      "epoch": 0.4321731112775889,
      "grad_norm": 0.7678527235984802,
      "learning_rate": 6.3254671716326865e-06,
      "loss": 0.7194,
      "step": 14999
    },
    {
      "epoch": 0.4322019247392382,
      "grad_norm": 0.7542545795440674,
      "learning_rate": 6.325017249772926e-06,
      "loss": 0.7181,
      "step": 15000
    },
    {
      "epoch": 0.4322307382008875,
      "grad_norm": 0.7274553775787354,
      "learning_rate": 6.324567316373589e-06,
      "loss": 0.7334,
      "step": 15001
    },
    {
      "epoch": 0.4322595516625367,
      "grad_norm": 0.7312504649162292,
      "learning_rate": 6.324117371438594e-06,
      "loss": 0.7364,
      "step": 15002
    },
    {
      "epoch": 0.432288365124186,
      "grad_norm": 0.7194334864616394,
      "learning_rate": 6.323667414971861e-06,
      "loss": 0.6982,
      "step": 15003
    },
    {
      "epoch": 0.4323171785858353,
      "grad_norm": 0.7127925753593445,
      "learning_rate": 6.323217446977308e-06,
      "loss": 0.7205,
      "step": 15004
    },
    {
      "epoch": 0.4323459920474846,
      "grad_norm": 0.7242520451545715,
      "learning_rate": 6.322767467458854e-06,
      "loss": 0.7194,
      "step": 15005
    },
    {
      "epoch": 0.4323748055091339,
      "grad_norm": 0.7598992586135864,
      "learning_rate": 6.322317476420418e-06,
      "loss": 0.7215,
      "step": 15006
    },
    {
      "epoch": 0.43240361897078317,
      "grad_norm": 0.7206487655639648,
      "learning_rate": 6.321867473865918e-06,
      "loss": 0.727,
      "step": 15007
    },
    {
      "epoch": 0.43243243243243246,
      "grad_norm": 0.7555259466171265,
      "learning_rate": 6.321417459799273e-06,
      "loss": 0.7397,
      "step": 15008
    },
    {
      "epoch": 0.4324612458940817,
      "grad_norm": 0.7263649106025696,
      "learning_rate": 6.320967434224403e-06,
      "loss": 0.7194,
      "step": 15009
    },
    {
      "epoch": 0.432490059355731,
      "grad_norm": 0.7380584478378296,
      "learning_rate": 6.320517397145228e-06,
      "loss": 0.7501,
      "step": 15010
    },
    {
      "epoch": 0.4325188728173803,
      "grad_norm": 0.7424178123474121,
      "learning_rate": 6.320067348565666e-06,
      "loss": 0.7211,
      "step": 15011
    },
    {
      "epoch": 0.43254768627902956,
      "grad_norm": 0.7583703994750977,
      "learning_rate": 6.319617288489638e-06,
      "loss": 0.7465,
      "step": 15012
    },
    {
      "epoch": 0.43257649974067885,
      "grad_norm": 0.7849952578544617,
      "learning_rate": 6.31916721692106e-06,
      "loss": 0.7338,
      "step": 15013
    },
    {
      "epoch": 0.43260531320232815,
      "grad_norm": 0.773768424987793,
      "learning_rate": 6.318717133863856e-06,
      "loss": 0.7494,
      "step": 15014
    },
    {
      "epoch": 0.43263412666397744,
      "grad_norm": 0.7630691528320312,
      "learning_rate": 6.318267039321943e-06,
      "loss": 0.7357,
      "step": 15015
    },
    {
      "epoch": 0.43266294012562667,
      "grad_norm": 0.7539356350898743,
      "learning_rate": 6.317816933299243e-06,
      "loss": 0.7475,
      "step": 15016
    },
    {
      "epoch": 0.43269175358727596,
      "grad_norm": 0.7435582280158997,
      "learning_rate": 6.317366815799676e-06,
      "loss": 0.7098,
      "step": 15017
    },
    {
      "epoch": 0.43272056704892525,
      "grad_norm": 0.7744205594062805,
      "learning_rate": 6.316916686827159e-06,
      "loss": 0.7205,
      "step": 15018
    },
    {
      "epoch": 0.43274938051057454,
      "grad_norm": 0.7277206182479858,
      "learning_rate": 6.316466546385616e-06,
      "loss": 0.745,
      "step": 15019
    },
    {
      "epoch": 0.43277819397222383,
      "grad_norm": 0.7476612329483032,
      "learning_rate": 6.316016394478963e-06,
      "loss": 0.7238,
      "step": 15020
    },
    {
      "epoch": 0.4328070074338731,
      "grad_norm": 0.7860918045043945,
      "learning_rate": 6.315566231111123e-06,
      "loss": 0.752,
      "step": 15021
    },
    {
      "epoch": 0.43283582089552236,
      "grad_norm": 0.7444726228713989,
      "learning_rate": 6.315116056286017e-06,
      "loss": 0.7442,
      "step": 15022
    },
    {
      "epoch": 0.43286463435717165,
      "grad_norm": 0.8012540936470032,
      "learning_rate": 6.314665870007566e-06,
      "loss": 0.7341,
      "step": 15023
    },
    {
      "epoch": 0.43289344781882094,
      "grad_norm": 0.7398999929428101,
      "learning_rate": 6.3142156722796875e-06,
      "loss": 0.711,
      "step": 15024
    },
    {
      "epoch": 0.43292226128047023,
      "grad_norm": 0.7260816097259521,
      "learning_rate": 6.313765463106305e-06,
      "loss": 0.7064,
      "step": 15025
    },
    {
      "epoch": 0.4329510747421195,
      "grad_norm": 0.7598549723625183,
      "learning_rate": 6.313315242491338e-06,
      "loss": 0.735,
      "step": 15026
    },
    {
      "epoch": 0.4329798882037688,
      "grad_norm": 0.8145784735679626,
      "learning_rate": 6.312865010438709e-06,
      "loss": 0.7484,
      "step": 15027
    },
    {
      "epoch": 0.4330087016654181,
      "grad_norm": 0.7369046807289124,
      "learning_rate": 6.312414766952338e-06,
      "loss": 0.7273,
      "step": 15028
    },
    {
      "epoch": 0.43303751512706734,
      "grad_norm": 0.7499142289161682,
      "learning_rate": 6.311964512036146e-06,
      "loss": 0.7427,
      "step": 15029
    },
    {
      "epoch": 0.43306632858871663,
      "grad_norm": 0.7792626619338989,
      "learning_rate": 6.311514245694055e-06,
      "loss": 0.7289,
      "step": 15030
    },
    {
      "epoch": 0.4330951420503659,
      "grad_norm": 2.234104871749878,
      "learning_rate": 6.311063967929985e-06,
      "loss": 0.7454,
      "step": 15031
    },
    {
      "epoch": 0.4331239555120152,
      "grad_norm": 0.7232475280761719,
      "learning_rate": 6.310613678747859e-06,
      "loss": 0.7158,
      "step": 15032
    },
    {
      "epoch": 0.4331527689736645,
      "grad_norm": 0.7588796019554138,
      "learning_rate": 6.310163378151598e-06,
      "loss": 0.7219,
      "step": 15033
    },
    {
      "epoch": 0.4331815824353138,
      "grad_norm": 0.7518603205680847,
      "learning_rate": 6.3097130661451235e-06,
      "loss": 0.7118,
      "step": 15034
    },
    {
      "epoch": 0.4332103958969631,
      "grad_norm": 0.7775900959968567,
      "learning_rate": 6.309262742732357e-06,
      "loss": 0.7624,
      "step": 15035
    },
    {
      "epoch": 0.4332392093586123,
      "grad_norm": 0.7579478621482849,
      "learning_rate": 6.308812407917222e-06,
      "loss": 0.7016,
      "step": 15036
    },
    {
      "epoch": 0.4332680228202616,
      "grad_norm": 0.7816409468650818,
      "learning_rate": 6.308362061703639e-06,
      "loss": 0.7355,
      "step": 15037
    },
    {
      "epoch": 0.4332968362819109,
      "grad_norm": 0.7418741583824158,
      "learning_rate": 6.3079117040955295e-06,
      "loss": 0.7015,
      "step": 15038
    },
    {
      "epoch": 0.4333256497435602,
      "grad_norm": 0.7492384910583496,
      "learning_rate": 6.307461335096817e-06,
      "loss": 0.7272,
      "step": 15039
    },
    {
      "epoch": 0.4333544632052095,
      "grad_norm": 0.7279199361801147,
      "learning_rate": 6.3070109547114235e-06,
      "loss": 0.7075,
      "step": 15040
    },
    {
      "epoch": 0.4333832766668588,
      "grad_norm": 0.7485374808311462,
      "learning_rate": 6.306560562943272e-06,
      "loss": 0.7252,
      "step": 15041
    },
    {
      "epoch": 0.43341209012850807,
      "grad_norm": 0.7382106184959412,
      "learning_rate": 6.306110159796282e-06,
      "loss": 0.7023,
      "step": 15042
    },
    {
      "epoch": 0.4334409035901573,
      "grad_norm": 0.7598785161972046,
      "learning_rate": 6.30565974527438e-06,
      "loss": 0.7639,
      "step": 15043
    },
    {
      "epoch": 0.4334697170518066,
      "grad_norm": 0.7663024663925171,
      "learning_rate": 6.305209319381485e-06,
      "loss": 0.7334,
      "step": 15044
    },
    {
      "epoch": 0.4334985305134559,
      "grad_norm": 2.9959919452667236,
      "learning_rate": 6.304758882121522e-06,
      "loss": 0.7265,
      "step": 15045
    },
    {
      "epoch": 0.4335273439751052,
      "grad_norm": 0.7522894740104675,
      "learning_rate": 6.304308433498414e-06,
      "loss": 0.7358,
      "step": 15046
    },
    {
      "epoch": 0.43355615743675446,
      "grad_norm": 0.777973473072052,
      "learning_rate": 6.3038579735160834e-06,
      "loss": 0.7267,
      "step": 15047
    },
    {
      "epoch": 0.43358497089840375,
      "grad_norm": 0.7616091966629028,
      "learning_rate": 6.303407502178454e-06,
      "loss": 0.7405,
      "step": 15048
    },
    {
      "epoch": 0.433613784360053,
      "grad_norm": 0.7228960990905762,
      "learning_rate": 6.302957019489447e-06,
      "loss": 0.7234,
      "step": 15049
    },
    {
      "epoch": 0.4336425978217023,
      "grad_norm": 0.7622433304786682,
      "learning_rate": 6.302506525452986e-06,
      "loss": 0.7273,
      "step": 15050
    },
    {
      "epoch": 0.43367141128335157,
      "grad_norm": 0.7798157930374146,
      "learning_rate": 6.302056020072996e-06,
      "loss": 0.7305,
      "step": 15051
    },
    {
      "epoch": 0.43370022474500086,
      "grad_norm": 0.7500669360160828,
      "learning_rate": 6.301605503353401e-06,
      "loss": 0.711,
      "step": 15052
    },
    {
      "epoch": 0.43372903820665015,
      "grad_norm": 0.7355537414550781,
      "learning_rate": 6.301154975298121e-06,
      "loss": 0.747,
      "step": 15053
    },
    {
      "epoch": 0.43375785166829944,
      "grad_norm": 0.7535845041275024,
      "learning_rate": 6.300704435911083e-06,
      "loss": 0.7413,
      "step": 15054
    },
    {
      "epoch": 0.43378666512994873,
      "grad_norm": 0.7530983090400696,
      "learning_rate": 6.30025388519621e-06,
      "loss": 0.7139,
      "step": 15055
    },
    {
      "epoch": 0.43381547859159797,
      "grad_norm": 0.7473952770233154,
      "learning_rate": 6.299803323157424e-06,
      "loss": 0.7345,
      "step": 15056
    },
    {
      "epoch": 0.43384429205324726,
      "grad_norm": 0.7605932950973511,
      "learning_rate": 6.299352749798652e-06,
      "loss": 0.733,
      "step": 15057
    },
    {
      "epoch": 0.43387310551489655,
      "grad_norm": 0.7272268533706665,
      "learning_rate": 6.298902165123815e-06,
      "loss": 0.722,
      "step": 15058
    },
    {
      "epoch": 0.43390191897654584,
      "grad_norm": 0.7466461658477783,
      "learning_rate": 6.29845156913684e-06,
      "loss": 0.7117,
      "step": 15059
    },
    {
      "epoch": 0.43393073243819513,
      "grad_norm": 0.7321277856826782,
      "learning_rate": 6.298000961841649e-06,
      "loss": 0.7237,
      "step": 15060
    },
    {
      "epoch": 0.4339595458998444,
      "grad_norm": 0.7532580494880676,
      "learning_rate": 6.297550343242168e-06,
      "loss": 0.7257,
      "step": 15061
    },
    {
      "epoch": 0.4339883593614937,
      "grad_norm": 0.7596437931060791,
      "learning_rate": 6.2970997133423186e-06,
      "loss": 0.7269,
      "step": 15062
    },
    {
      "epoch": 0.43401717282314295,
      "grad_norm": 0.7509198784828186,
      "learning_rate": 6.2966490721460294e-06,
      "loss": 0.7196,
      "step": 15063
    },
    {
      "epoch": 0.43404598628479224,
      "grad_norm": 0.7436909675598145,
      "learning_rate": 6.2961984196572224e-06,
      "loss": 0.7293,
      "step": 15064
    },
    {
      "epoch": 0.43407479974644153,
      "grad_norm": 0.7660581469535828,
      "learning_rate": 6.295747755879823e-06,
      "loss": 0.7366,
      "step": 15065
    },
    {
      "epoch": 0.4341036132080908,
      "grad_norm": 0.7370564341545105,
      "learning_rate": 6.2952970808177545e-06,
      "loss": 0.7258,
      "step": 15066
    },
    {
      "epoch": 0.4341324266697401,
      "grad_norm": 0.748960018157959,
      "learning_rate": 6.294846394474945e-06,
      "loss": 0.7103,
      "step": 15067
    },
    {
      "epoch": 0.4341612401313894,
      "grad_norm": 0.7631029486656189,
      "learning_rate": 6.294395696855317e-06,
      "loss": 0.7155,
      "step": 15068
    },
    {
      "epoch": 0.4341900535930387,
      "grad_norm": 0.7353963255882263,
      "learning_rate": 6.293944987962797e-06,
      "loss": 0.7375,
      "step": 15069
    },
    {
      "epoch": 0.43421886705468793,
      "grad_norm": 0.758220374584198,
      "learning_rate": 6.293494267801309e-06,
      "loss": 0.7414,
      "step": 15070
    },
    {
      "epoch": 0.4342476805163372,
      "grad_norm": 0.7307944297790527,
      "learning_rate": 6.293043536374779e-06,
      "loss": 0.7305,
      "step": 15071
    },
    {
      "epoch": 0.4342764939779865,
      "grad_norm": 0.7511945366859436,
      "learning_rate": 6.292592793687131e-06,
      "loss": 0.7148,
      "step": 15072
    },
    {
      "epoch": 0.4343053074396358,
      "grad_norm": 0.7112933397293091,
      "learning_rate": 6.292142039742295e-06,
      "loss": 0.7215,
      "step": 15073
    },
    {
      "epoch": 0.4343341209012851,
      "grad_norm": 0.7210623621940613,
      "learning_rate": 6.2916912745441904e-06,
      "loss": 0.7077,
      "step": 15074
    },
    {
      "epoch": 0.4343629343629344,
      "grad_norm": 0.7485143542289734,
      "learning_rate": 6.291240498096748e-06,
      "loss": 0.7372,
      "step": 15075
    },
    {
      "epoch": 0.4343917478245836,
      "grad_norm": 0.7268690466880798,
      "learning_rate": 6.290789710403891e-06,
      "loss": 0.7162,
      "step": 15076
    },
    {
      "epoch": 0.4344205612862329,
      "grad_norm": 0.7277371883392334,
      "learning_rate": 6.290338911469547e-06,
      "loss": 0.7194,
      "step": 15077
    },
    {
      "epoch": 0.4344493747478822,
      "grad_norm": 0.7403417229652405,
      "learning_rate": 6.2898881012976385e-06,
      "loss": 0.7408,
      "step": 15078
    },
    {
      "epoch": 0.4344781882095315,
      "grad_norm": 0.7720884084701538,
      "learning_rate": 6.2894372798920965e-06,
      "loss": 0.753,
      "step": 15079
    },
    {
      "epoch": 0.4345070016711808,
      "grad_norm": 0.7228409647941589,
      "learning_rate": 6.288986447256843e-06,
      "loss": 0.7314,
      "step": 15080
    },
    {
      "epoch": 0.43453581513283007,
      "grad_norm": 0.7200401425361633,
      "learning_rate": 6.288535603395806e-06,
      "loss": 0.7028,
      "step": 15081
    },
    {
      "epoch": 0.43456462859447936,
      "grad_norm": 0.7195661067962646,
      "learning_rate": 6.2880847483129144e-06,
      "loss": 0.7154,
      "step": 15082
    },
    {
      "epoch": 0.4345934420561286,
      "grad_norm": 0.756522536277771,
      "learning_rate": 6.287633882012091e-06,
      "loss": 0.6941,
      "step": 15083
    },
    {
      "epoch": 0.4346222555177779,
      "grad_norm": 0.7440406084060669,
      "learning_rate": 6.287183004497264e-06,
      "loss": 0.7476,
      "step": 15084
    },
    {
      "epoch": 0.4346510689794272,
      "grad_norm": 0.7840383648872375,
      "learning_rate": 6.2867321157723585e-06,
      "loss": 0.7512,
      "step": 15085
    },
    {
      "epoch": 0.43467988244107647,
      "grad_norm": 0.7331530451774597,
      "learning_rate": 6.286281215841302e-06,
      "loss": 0.718,
      "step": 15086
    },
    {
      "epoch": 0.43470869590272576,
      "grad_norm": 1.6364833116531372,
      "learning_rate": 6.285830304708024e-06,
      "loss": 0.7352,
      "step": 15087
    },
    {
      "epoch": 0.43473750936437505,
      "grad_norm": 0.7768648862838745,
      "learning_rate": 6.2853793823764485e-06,
      "loss": 0.7421,
      "step": 15088
    },
    {
      "epoch": 0.43476632282602434,
      "grad_norm": 0.7664570808410645,
      "learning_rate": 6.284928448850504e-06,
      "loss": 0.7095,
      "step": 15089
    },
    {
      "epoch": 0.4347951362876736,
      "grad_norm": 0.7419267296791077,
      "learning_rate": 6.284477504134117e-06,
      "loss": 0.7245,
      "step": 15090
    },
    {
      "epoch": 0.43482394974932287,
      "grad_norm": 0.7298779487609863,
      "learning_rate": 6.2840265482312125e-06,
      "loss": 0.7208,
      "step": 15091
    },
    {
      "epoch": 0.43485276321097216,
      "grad_norm": 0.7685499787330627,
      "learning_rate": 6.283575581145722e-06,
      "loss": 0.738,
      "step": 15092
    },
    {
      "epoch": 0.43488157667262145,
      "grad_norm": 0.7701641321182251,
      "learning_rate": 6.283124602881572e-06,
      "loss": 0.7139,
      "step": 15093
    },
    {
      "epoch": 0.43491039013427074,
      "grad_norm": 0.7707163691520691,
      "learning_rate": 6.282673613442689e-06,
      "loss": 0.7345,
      "step": 15094
    },
    {
      "epoch": 0.43493920359592003,
      "grad_norm": 0.747524619102478,
      "learning_rate": 6.282222612833001e-06,
      "loss": 0.7436,
      "step": 15095
    },
    {
      "epoch": 0.4349680170575693,
      "grad_norm": 0.7429303526878357,
      "learning_rate": 6.2817716010564365e-06,
      "loss": 0.7335,
      "step": 15096
    },
    {
      "epoch": 0.43499683051921856,
      "grad_norm": 0.7499467730522156,
      "learning_rate": 6.281320578116922e-06,
      "loss": 0.7351,
      "step": 15097
    },
    {
      "epoch": 0.43502564398086785,
      "grad_norm": 0.7507138848304749,
      "learning_rate": 6.280869544018386e-06,
      "loss": 0.7122,
      "step": 15098
    },
    {
      "epoch": 0.43505445744251714,
      "grad_norm": 0.7434881925582886,
      "learning_rate": 6.2804184987647556e-06,
      "loss": 0.7314,
      "step": 15099
    },
    {
      "epoch": 0.43508327090416643,
      "grad_norm": 0.7217179536819458,
      "learning_rate": 6.279967442359962e-06,
      "loss": 0.7025,
      "step": 15100
    },
    {
      "epoch": 0.4351120843658157,
      "grad_norm": 0.7323155999183655,
      "learning_rate": 6.279516374807931e-06,
      "loss": 0.7263,
      "step": 15101
    },
    {
      "epoch": 0.435140897827465,
      "grad_norm": 0.7792452573776245,
      "learning_rate": 6.279065296112591e-06,
      "loss": 0.7504,
      "step": 15102
    },
    {
      "epoch": 0.43516971128911425,
      "grad_norm": 0.7390832901000977,
      "learning_rate": 6.278614206277871e-06,
      "loss": 0.7412,
      "step": 15103
    },
    {
      "epoch": 0.43519852475076354,
      "grad_norm": 0.768287181854248,
      "learning_rate": 6.278163105307699e-06,
      "loss": 0.7148,
      "step": 15104
    },
    {
      "epoch": 0.43522733821241283,
      "grad_norm": 0.7429295778274536,
      "learning_rate": 6.277711993206004e-06,
      "loss": 0.7198,
      "step": 15105
    },
    {
      "epoch": 0.4352561516740621,
      "grad_norm": 0.7363647818565369,
      "learning_rate": 6.277260869976716e-06,
      "loss": 0.7232,
      "step": 15106
    },
    {
      "epoch": 0.4352849651357114,
      "grad_norm": 0.7316232323646545,
      "learning_rate": 6.276809735623762e-06,
      "loss": 0.7253,
      "step": 15107
    },
    {
      "epoch": 0.4353137785973607,
      "grad_norm": 0.7041569948196411,
      "learning_rate": 6.276358590151071e-06,
      "loss": 0.711,
      "step": 15108
    },
    {
      "epoch": 0.43534259205901,
      "grad_norm": 0.7868814468383789,
      "learning_rate": 6.2759074335625735e-06,
      "loss": 0.725,
      "step": 15109
    },
    {
      "epoch": 0.4353714055206592,
      "grad_norm": 0.7616071701049805,
      "learning_rate": 6.275456265862196e-06,
      "loss": 0.7565,
      "step": 15110
    },
    {
      "epoch": 0.4354002189823085,
      "grad_norm": 0.7199476361274719,
      "learning_rate": 6.275005087053872e-06,
      "loss": 0.7129,
      "step": 15111
    },
    {
      "epoch": 0.4354290324439578,
      "grad_norm": 0.750660240650177,
      "learning_rate": 6.274553897141526e-06,
      "loss": 0.7135,
      "step": 15112
    },
    {
      "epoch": 0.4354578459056071,
      "grad_norm": 0.7711654305458069,
      "learning_rate": 6.27410269612909e-06,
      "loss": 0.7389,
      "step": 15113
    },
    {
      "epoch": 0.4354866593672564,
      "grad_norm": 0.7406135201454163,
      "learning_rate": 6.2736514840204935e-06,
      "loss": 0.7253,
      "step": 15114
    },
    {
      "epoch": 0.4355154728289057,
      "grad_norm": 0.7589178085327148,
      "learning_rate": 6.273200260819666e-06,
      "loss": 0.7476,
      "step": 15115
    },
    {
      "epoch": 0.43554428629055497,
      "grad_norm": 0.7349554300308228,
      "learning_rate": 6.2727490265305356e-06,
      "loss": 0.7143,
      "step": 15116
    },
    {
      "epoch": 0.4355730997522042,
      "grad_norm": 0.7820210456848145,
      "learning_rate": 6.272297781157035e-06,
      "loss": 0.7423,
      "step": 15117
    },
    {
      "epoch": 0.4356019132138535,
      "grad_norm": 0.7410609722137451,
      "learning_rate": 6.271846524703092e-06,
      "loss": 0.746,
      "step": 15118
    },
    {
      "epoch": 0.4356307266755028,
      "grad_norm": 0.7631399035453796,
      "learning_rate": 6.271395257172639e-06,
      "loss": 0.7473,
      "step": 15119
    },
    {
      "epoch": 0.4356595401371521,
      "grad_norm": 0.725403368473053,
      "learning_rate": 6.270943978569602e-06,
      "loss": 0.6998,
      "step": 15120
    },
    {
      "epoch": 0.43568835359880137,
      "grad_norm": 0.7442708015441895,
      "learning_rate": 6.270492688897914e-06,
      "loss": 0.7462,
      "step": 15121
    },
    {
      "epoch": 0.43571716706045066,
      "grad_norm": 0.7305096983909607,
      "learning_rate": 6.2700413881615045e-06,
      "loss": 0.7205,
      "step": 15122
    },
    {
      "epoch": 0.43574598052209995,
      "grad_norm": 0.749198853969574,
      "learning_rate": 6.269590076364305e-06,
      "loss": 0.7306,
      "step": 15123
    },
    {
      "epoch": 0.4357747939837492,
      "grad_norm": 0.7882585525512695,
      "learning_rate": 6.269138753510244e-06,
      "loss": 0.7237,
      "step": 15124
    },
    {
      "epoch": 0.4358036074453985,
      "grad_norm": 0.7701780796051025,
      "learning_rate": 6.268687419603254e-06,
      "loss": 0.717,
      "step": 15125
    },
    {
      "epoch": 0.43583242090704777,
      "grad_norm": 0.763511598110199,
      "learning_rate": 6.268236074647265e-06,
      "loss": 0.7202,
      "step": 15126
    },
    {
      "epoch": 0.43586123436869706,
      "grad_norm": 0.761695146560669,
      "learning_rate": 6.267784718646206e-06,
      "loss": 0.7135,
      "step": 15127
    },
    {
      "epoch": 0.43589004783034635,
      "grad_norm": 0.7844553589820862,
      "learning_rate": 6.26733335160401e-06,
      "loss": 0.7496,
      "step": 15128
    },
    {
      "epoch": 0.43591886129199564,
      "grad_norm": 0.8028609752655029,
      "learning_rate": 6.266881973524608e-06,
      "loss": 0.7355,
      "step": 15129
    },
    {
      "epoch": 0.4359476747536449,
      "grad_norm": 0.7839823365211487,
      "learning_rate": 6.26643058441193e-06,
      "loss": 0.735,
      "step": 15130
    },
    {
      "epoch": 0.43597648821529417,
      "grad_norm": 0.760722815990448,
      "learning_rate": 6.265979184269907e-06,
      "loss": 0.7051,
      "step": 15131
    },
    {
      "epoch": 0.43600530167694346,
      "grad_norm": 0.7435877323150635,
      "learning_rate": 6.265527773102471e-06,
      "loss": 0.7263,
      "step": 15132
    },
    {
      "epoch": 0.43603411513859275,
      "grad_norm": 0.7836940884590149,
      "learning_rate": 6.265076350913555e-06,
      "loss": 0.7346,
      "step": 15133
    },
    {
      "epoch": 0.43606292860024204,
      "grad_norm": 0.7412467002868652,
      "learning_rate": 6.264624917707086e-06,
      "loss": 0.7238,
      "step": 15134
    },
    {
      "epoch": 0.43609174206189133,
      "grad_norm": 0.7454859614372253,
      "learning_rate": 6.264173473487e-06,
      "loss": 0.7178,
      "step": 15135
    },
    {
      "epoch": 0.4361205555235406,
      "grad_norm": 0.7573898434638977,
      "learning_rate": 6.263722018257226e-06,
      "loss": 0.7349,
      "step": 15136
    },
    {
      "epoch": 0.43614936898518986,
      "grad_norm": 0.7402186989784241,
      "learning_rate": 6.2632705520216965e-06,
      "loss": 0.7073,
      "step": 15137
    },
    {
      "epoch": 0.43617818244683915,
      "grad_norm": 0.7619134187698364,
      "learning_rate": 6.262819074784344e-06,
      "loss": 0.7216,
      "step": 15138
    },
    {
      "epoch": 0.43620699590848844,
      "grad_norm": 0.740086555480957,
      "learning_rate": 6.2623675865491e-06,
      "loss": 0.7203,
      "step": 15139
    },
    {
      "epoch": 0.43623580937013773,
      "grad_norm": 0.7455657720565796,
      "learning_rate": 6.261916087319894e-06,
      "loss": 0.7396,
      "step": 15140
    },
    {
      "epoch": 0.436264622831787,
      "grad_norm": 0.7789521217346191,
      "learning_rate": 6.261464577100662e-06,
      "loss": 0.776,
      "step": 15141
    },
    {
      "epoch": 0.4362934362934363,
      "grad_norm": 0.7548785209655762,
      "learning_rate": 6.2610130558953345e-06,
      "loss": 0.7317,
      "step": 15142
    },
    {
      "epoch": 0.4363222497550856,
      "grad_norm": 0.6991636753082275,
      "learning_rate": 6.2605615237078435e-06,
      "loss": 0.7041,
      "step": 15143
    },
    {
      "epoch": 0.43635106321673484,
      "grad_norm": 0.7194889783859253,
      "learning_rate": 6.260109980542122e-06,
      "loss": 0.7436,
      "step": 15144
    },
    {
      "epoch": 0.4363798766783841,
      "grad_norm": 0.7272387742996216,
      "learning_rate": 6.259658426402102e-06,
      "loss": 0.7144,
      "step": 15145
    },
    {
      "epoch": 0.4364086901400334,
      "grad_norm": 0.7397661209106445,
      "learning_rate": 6.259206861291716e-06,
      "loss": 0.729,
      "step": 15146
    },
    {
      "epoch": 0.4364375036016827,
      "grad_norm": 0.741352379322052,
      "learning_rate": 6.258755285214897e-06,
      "loss": 0.7304,
      "step": 15147
    },
    {
      "epoch": 0.436466317063332,
      "grad_norm": 0.752251923084259,
      "learning_rate": 6.2583036981755784e-06,
      "loss": 0.7247,
      "step": 15148
    },
    {
      "epoch": 0.4364951305249813,
      "grad_norm": 0.7192510962486267,
      "learning_rate": 6.257852100177692e-06,
      "loss": 0.6991,
      "step": 15149
    },
    {
      "epoch": 0.4365239439866306,
      "grad_norm": 0.7403880953788757,
      "learning_rate": 6.257400491225171e-06,
      "loss": 0.7113,
      "step": 15150
    },
    {
      "epoch": 0.4365527574482798,
      "grad_norm": 1.8628876209259033,
      "learning_rate": 6.25694887132195e-06,
      "loss": 0.7254,
      "step": 15151
    },
    {
      "epoch": 0.4365815709099291,
      "grad_norm": 0.7662656307220459,
      "learning_rate": 6.2564972404719594e-06,
      "loss": 0.7171,
      "step": 15152
    },
    {
      "epoch": 0.4366103843715784,
      "grad_norm": 0.7093855738639832,
      "learning_rate": 6.256045598679134e-06,
      "loss": 0.717,
      "step": 15153
    },
    {
      "epoch": 0.4366391978332277,
      "grad_norm": 0.7241555452346802,
      "learning_rate": 6.255593945947407e-06,
      "loss": 0.7199,
      "step": 15154
    },
    {
      "epoch": 0.436668011294877,
      "grad_norm": 0.7840353846549988,
      "learning_rate": 6.255142282280712e-06,
      "loss": 0.7535,
      "step": 15155
    },
    {
      "epoch": 0.43669682475652627,
      "grad_norm": 0.7367922067642212,
      "learning_rate": 6.254690607682983e-06,
      "loss": 0.7357,
      "step": 15156
    },
    {
      "epoch": 0.4367256382181755,
      "grad_norm": 0.7358642220497131,
      "learning_rate": 6.254238922158153e-06,
      "loss": 0.7521,
      "step": 15157
    },
    {
      "epoch": 0.4367544516798248,
      "grad_norm": 0.7426034212112427,
      "learning_rate": 6.253787225710154e-06,
      "loss": 0.7028,
      "step": 15158
    },
    {
      "epoch": 0.4367832651414741,
      "grad_norm": 0.7536616921424866,
      "learning_rate": 6.253335518342923e-06,
      "loss": 0.7446,
      "step": 15159
    },
    {
      "epoch": 0.4368120786031234,
      "grad_norm": 0.7819907665252686,
      "learning_rate": 6.252883800060393e-06,
      "loss": 0.709,
      "step": 15160
    },
    {
      "epoch": 0.43684089206477267,
      "grad_norm": 0.7644881010055542,
      "learning_rate": 6.252432070866498e-06,
      "loss": 0.7229,
      "step": 15161
    },
    {
      "epoch": 0.43686970552642196,
      "grad_norm": 0.7616397738456726,
      "learning_rate": 6.251980330765171e-06,
      "loss": 0.7204,
      "step": 15162
    },
    {
      "epoch": 0.43689851898807125,
      "grad_norm": 1.740898847579956,
      "learning_rate": 6.2515285797603465e-06,
      "loss": 0.7398,
      "step": 15163
    },
    {
      "epoch": 0.4369273324497205,
      "grad_norm": 0.7607806324958801,
      "learning_rate": 6.2510768178559575e-06,
      "loss": 0.7257,
      "step": 15164
    },
    {
      "epoch": 0.4369561459113698,
      "grad_norm": 0.7531924843788147,
      "learning_rate": 6.250625045055942e-06,
      "loss": 0.7226,
      "step": 15165
    },
    {
      "epoch": 0.43698495937301907,
      "grad_norm": 0.7789646983146667,
      "learning_rate": 6.250173261364232e-06,
      "loss": 0.729,
      "step": 15166
    },
    {
      "epoch": 0.43701377283466836,
      "grad_norm": 0.7635838985443115,
      "learning_rate": 6.2497214667847635e-06,
      "loss": 0.7201,
      "step": 15167
    },
    {
      "epoch": 0.43704258629631765,
      "grad_norm": 0.7812569737434387,
      "learning_rate": 6.24926966132147e-06,
      "loss": 0.7524,
      "step": 15168
    },
    {
      "epoch": 0.43707139975796694,
      "grad_norm": 0.7548714280128479,
      "learning_rate": 6.248817844978285e-06,
      "loss": 0.7221,
      "step": 15169
    },
    {
      "epoch": 0.43710021321961623,
      "grad_norm": 0.7353742718696594,
      "learning_rate": 6.2483660177591465e-06,
      "loss": 0.7047,
      "step": 15170
    },
    {
      "epoch": 0.43712902668126546,
      "grad_norm": 0.7907098531723022,
      "learning_rate": 6.247914179667988e-06,
      "loss": 0.733,
      "step": 15171
    },
    {
      "epoch": 0.43715784014291476,
      "grad_norm": 0.742530345916748,
      "learning_rate": 6.247462330708742e-06,
      "loss": 0.7432,
      "step": 15172
    },
    {
      "epoch": 0.43718665360456405,
      "grad_norm": 0.7422332763671875,
      "learning_rate": 6.247010470885348e-06,
      "loss": 0.7294,
      "step": 15173
    },
    {
      "epoch": 0.43721546706621334,
      "grad_norm": 0.7523233890533447,
      "learning_rate": 6.24655860020174e-06,
      "loss": 0.7478,
      "step": 15174
    },
    {
      "epoch": 0.4372442805278626,
      "grad_norm": 0.7646411657333374,
      "learning_rate": 6.246106718661853e-06,
      "loss": 0.7387,
      "step": 15175
    },
    {
      "epoch": 0.4372730939895119,
      "grad_norm": 0.723703920841217,
      "learning_rate": 6.24565482626962e-06,
      "loss": 0.7039,
      "step": 15176
    },
    {
      "epoch": 0.4373019074511612,
      "grad_norm": 0.7614469528198242,
      "learning_rate": 6.24520292302898e-06,
      "loss": 0.7092,
      "step": 15177
    },
    {
      "epoch": 0.43733072091281044,
      "grad_norm": 0.7195727229118347,
      "learning_rate": 6.2447510089438675e-06,
      "loss": 0.7222,
      "step": 15178
    },
    {
      "epoch": 0.43735953437445974,
      "grad_norm": 0.761772871017456,
      "learning_rate": 6.244299084018217e-06,
      "loss": 0.7143,
      "step": 15179
    },
    {
      "epoch": 0.437388347836109,
      "grad_norm": 0.7792402505874634,
      "learning_rate": 6.243847148255967e-06,
      "loss": 0.7357,
      "step": 15180
    },
    {
      "epoch": 0.4374171612977583,
      "grad_norm": 0.7368561029434204,
      "learning_rate": 6.243395201661051e-06,
      "loss": 0.718,
      "step": 15181
    },
    {
      "epoch": 0.4374459747594076,
      "grad_norm": 0.7642274498939514,
      "learning_rate": 6.242943244237405e-06,
      "loss": 0.7451,
      "step": 15182
    },
    {
      "epoch": 0.4374747882210569,
      "grad_norm": 0.7699384093284607,
      "learning_rate": 6.242491275988966e-06,
      "loss": 0.7464,
      "step": 15183
    },
    {
      "epoch": 0.43750360168270613,
      "grad_norm": 0.7921477556228638,
      "learning_rate": 6.242039296919671e-06,
      "loss": 0.749,
      "step": 15184
    },
    {
      "epoch": 0.4375324151443554,
      "grad_norm": 0.76450115442276,
      "learning_rate": 6.241587307033454e-06,
      "loss": 0.7572,
      "step": 15185
    },
    {
      "epoch": 0.4375612286060047,
      "grad_norm": 0.7626978158950806,
      "learning_rate": 6.241135306334254e-06,
      "loss": 0.7281,
      "step": 15186
    },
    {
      "epoch": 0.437590042067654,
      "grad_norm": 0.7260051965713501,
      "learning_rate": 6.240683294826005e-06,
      "loss": 0.7286,
      "step": 15187
    },
    {
      "epoch": 0.4376188555293033,
      "grad_norm": 0.7391970753669739,
      "learning_rate": 6.240231272512645e-06,
      "loss": 0.7335,
      "step": 15188
    },
    {
      "epoch": 0.4376476689909526,
      "grad_norm": 0.7417942881584167,
      "learning_rate": 6.23977923939811e-06,
      "loss": 0.7313,
      "step": 15189
    },
    {
      "epoch": 0.4376764824526019,
      "grad_norm": 0.8014991283416748,
      "learning_rate": 6.2393271954863386e-06,
      "loss": 0.7313,
      "step": 15190
    },
    {
      "epoch": 0.4377052959142511,
      "grad_norm": 0.7426168918609619,
      "learning_rate": 6.238875140781264e-06,
      "loss": 0.7325,
      "step": 15191
    },
    {
      "epoch": 0.4377341093759004,
      "grad_norm": 0.7384558320045471,
      "learning_rate": 6.238423075286827e-06,
      "loss": 0.6948,
      "step": 15192
    },
    {
      "epoch": 0.4377629228375497,
      "grad_norm": 0.7541776299476624,
      "learning_rate": 6.237970999006964e-06,
      "loss": 0.7232,
      "step": 15193
    },
    {
      "epoch": 0.437791736299199,
      "grad_norm": 0.7710739970207214,
      "learning_rate": 6.237518911945608e-06,
      "loss": 0.7293,
      "step": 15194
    },
    {
      "epoch": 0.4378205497608483,
      "grad_norm": 0.7656593918800354,
      "learning_rate": 6.237066814106702e-06,
      "loss": 0.7368,
      "step": 15195
    },
    {
      "epoch": 0.43784936322249757,
      "grad_norm": 0.7585647702217102,
      "learning_rate": 6.236614705494179e-06,
      "loss": 0.7261,
      "step": 15196
    },
    {
      "epoch": 0.43787817668414686,
      "grad_norm": 0.7517191767692566,
      "learning_rate": 6.236162586111979e-06,
      "loss": 0.7277,
      "step": 15197
    },
    {
      "epoch": 0.4379069901457961,
      "grad_norm": 0.7512376308441162,
      "learning_rate": 6.235710455964039e-06,
      "loss": 0.7067,
      "step": 15198
    },
    {
      "epoch": 0.4379358036074454,
      "grad_norm": 0.7806180119514465,
      "learning_rate": 6.235258315054296e-06,
      "loss": 0.7294,
      "step": 15199
    },
    {
      "epoch": 0.4379646170690947,
      "grad_norm": 0.7388423681259155,
      "learning_rate": 6.234806163386686e-06,
      "loss": 0.7264,
      "step": 15200
    },
    {
      "epoch": 0.43799343053074397,
      "grad_norm": 0.7685832381248474,
      "learning_rate": 6.234354000965151e-06,
      "loss": 0.6922,
      "step": 15201
    },
    {
      "epoch": 0.43802224399239326,
      "grad_norm": 0.797859787940979,
      "learning_rate": 6.2339018277936255e-06,
      "loss": 0.7259,
      "step": 15202
    },
    {
      "epoch": 0.43805105745404255,
      "grad_norm": 0.7601922750473022,
      "learning_rate": 6.23344964387605e-06,
      "loss": 0.7084,
      "step": 15203
    },
    {
      "epoch": 0.43807987091569184,
      "grad_norm": 0.810004711151123,
      "learning_rate": 6.232997449216359e-06,
      "loss": 0.7397,
      "step": 15204
    },
    {
      "epoch": 0.4381086843773411,
      "grad_norm": 0.7511412501335144,
      "learning_rate": 6.232545243818494e-06,
      "loss": 0.7325,
      "step": 15205
    },
    {
      "epoch": 0.43813749783899036,
      "grad_norm": 0.7395210266113281,
      "learning_rate": 6.232093027686391e-06,
      "loss": 0.7183,
      "step": 15206
    },
    {
      "epoch": 0.43816631130063965,
      "grad_norm": 0.7446989417076111,
      "learning_rate": 6.231640800823991e-06,
      "loss": 0.7402,
      "step": 15207
    },
    {
      "epoch": 0.43819512476228895,
      "grad_norm": 0.7732528448104858,
      "learning_rate": 6.23118856323523e-06,
      "loss": 0.726,
      "step": 15208
    },
    {
      "epoch": 0.43822393822393824,
      "grad_norm": 0.7378426194190979,
      "learning_rate": 6.230736314924047e-06,
      "loss": 0.7152,
      "step": 15209
    },
    {
      "epoch": 0.4382527516855875,
      "grad_norm": 0.7652280330657959,
      "learning_rate": 6.23028405589438e-06,
      "loss": 0.741,
      "step": 15210
    },
    {
      "epoch": 0.43828156514723676,
      "grad_norm": 0.7175522446632385,
      "learning_rate": 6.22983178615017e-06,
      "loss": 0.7009,
      "step": 15211
    },
    {
      "epoch": 0.43831037860888605,
      "grad_norm": 0.7463091611862183,
      "learning_rate": 6.2293795056953544e-06,
      "loss": 0.7177,
      "step": 15212
    },
    {
      "epoch": 0.43833919207053534,
      "grad_norm": 0.753535270690918,
      "learning_rate": 6.228927214533872e-06,
      "loss": 0.718,
      "step": 15213
    },
    {
      "epoch": 0.43836800553218463,
      "grad_norm": 0.7388637661933899,
      "learning_rate": 6.228474912669662e-06,
      "loss": 0.7259,
      "step": 15214
    },
    {
      "epoch": 0.4383968189938339,
      "grad_norm": 0.739133894443512,
      "learning_rate": 6.2280226001066645e-06,
      "loss": 0.726,
      "step": 15215
    },
    {
      "epoch": 0.4384256324554832,
      "grad_norm": 0.73780757188797,
      "learning_rate": 6.227570276848818e-06,
      "loss": 0.7324,
      "step": 15216
    },
    {
      "epoch": 0.4384544459171325,
      "grad_norm": 0.775407612323761,
      "learning_rate": 6.227117942900061e-06,
      "loss": 0.716,
      "step": 15217
    },
    {
      "epoch": 0.43848325937878174,
      "grad_norm": 0.7262284755706787,
      "learning_rate": 6.226665598264331e-06,
      "loss": 0.7041,
      "step": 15218
    },
    {
      "epoch": 0.43851207284043103,
      "grad_norm": 0.7353664636611938,
      "learning_rate": 6.226213242945573e-06,
      "loss": 0.7295,
      "step": 15219
    },
    {
      "epoch": 0.4385408863020803,
      "grad_norm": 0.7478204369544983,
      "learning_rate": 6.225760876947722e-06,
      "loss": 0.7132,
      "step": 15220
    },
    {
      "epoch": 0.4385696997637296,
      "grad_norm": 0.7626323699951172,
      "learning_rate": 6.225308500274719e-06,
      "loss": 0.7262,
      "step": 15221
    },
    {
      "epoch": 0.4385985132253789,
      "grad_norm": 0.7746855616569519,
      "learning_rate": 6.224856112930504e-06,
      "loss": 0.7386,
      "step": 15222
    },
    {
      "epoch": 0.4386273266870282,
      "grad_norm": 0.7807922959327698,
      "learning_rate": 6.224403714919017e-06,
      "loss": 0.7475,
      "step": 15223
    },
    {
      "epoch": 0.4386561401486775,
      "grad_norm": 0.7733880281448364,
      "learning_rate": 6.223951306244197e-06,
      "loss": 0.7153,
      "step": 15224
    },
    {
      "epoch": 0.4386849536103267,
      "grad_norm": 0.8058910369873047,
      "learning_rate": 6.2234988869099855e-06,
      "loss": 0.7315,
      "step": 15225
    },
    {
      "epoch": 0.438713767071976,
      "grad_norm": 0.7492385506629944,
      "learning_rate": 6.223046456920321e-06,
      "loss": 0.7399,
      "step": 15226
    },
    {
      "epoch": 0.4387425805336253,
      "grad_norm": 0.7849544286727905,
      "learning_rate": 6.2225940162791455e-06,
      "loss": 0.7325,
      "step": 15227
    },
    {
      "epoch": 0.4387713939952746,
      "grad_norm": 0.7721318006515503,
      "learning_rate": 6.222141564990396e-06,
      "loss": 0.733,
      "step": 15228
    },
    {
      "epoch": 0.4388002074569239,
      "grad_norm": 0.7333948612213135,
      "learning_rate": 6.2216891030580176e-06,
      "loss": 0.7021,
      "step": 15229
    },
    {
      "epoch": 0.4388290209185732,
      "grad_norm": 0.7524639964103699,
      "learning_rate": 6.221236630485947e-06,
      "loss": 0.7137,
      "step": 15230
    },
    {
      "epoch": 0.43885783438022247,
      "grad_norm": 0.7799622416496277,
      "learning_rate": 6.220784147278126e-06,
      "loss": 0.6884,
      "step": 15231
    },
    {
      "epoch": 0.4388866478418717,
      "grad_norm": 0.7957830429077148,
      "learning_rate": 6.220331653438496e-06,
      "loss": 0.7176,
      "step": 15232
    },
    {
      "epoch": 0.438915461303521,
      "grad_norm": 0.7409564852714539,
      "learning_rate": 6.219879148970998e-06,
      "loss": 0.7156,
      "step": 15233
    },
    {
      "epoch": 0.4389442747651703,
      "grad_norm": 0.7502449154853821,
      "learning_rate": 6.2194266338795705e-06,
      "loss": 0.7274,
      "step": 15234
    },
    {
      "epoch": 0.4389730882268196,
      "grad_norm": 0.7735193967819214,
      "learning_rate": 6.218974108168156e-06,
      "loss": 0.7285,
      "step": 15235
    },
    {
      "epoch": 0.43900190168846887,
      "grad_norm": 0.8121298551559448,
      "learning_rate": 6.2185215718406966e-06,
      "loss": 0.7215,
      "step": 15236
    },
    {
      "epoch": 0.43903071515011816,
      "grad_norm": 0.7551196217536926,
      "learning_rate": 6.218069024901132e-06,
      "loss": 0.7088,
      "step": 15237
    },
    {
      "epoch": 0.4390595286117674,
      "grad_norm": 0.807294487953186,
      "learning_rate": 6.217616467353404e-06,
      "loss": 0.7154,
      "step": 15238
    },
    {
      "epoch": 0.4390883420734167,
      "grad_norm": 0.7368510365486145,
      "learning_rate": 6.217163899201454e-06,
      "loss": 0.6946,
      "step": 15239
    },
    {
      "epoch": 0.439117155535066,
      "grad_norm": 0.8352342844009399,
      "learning_rate": 6.216711320449223e-06,
      "loss": 0.7502,
      "step": 15240
    },
    {
      "epoch": 0.43914596899671526,
      "grad_norm": 0.756565272808075,
      "learning_rate": 6.2162587311006525e-06,
      "loss": 0.7432,
      "step": 15241
    },
    {
      "epoch": 0.43917478245836455,
      "grad_norm": 0.7344276905059814,
      "learning_rate": 6.215806131159683e-06,
      "loss": 0.732,
      "step": 15242
    },
    {
      "epoch": 0.43920359592001385,
      "grad_norm": 0.719190239906311,
      "learning_rate": 6.215353520630258e-06,
      "loss": 0.7243,
      "step": 15243
    },
    {
      "epoch": 0.43923240938166314,
      "grad_norm": 0.7442511320114136,
      "learning_rate": 6.21490089951632e-06,
      "loss": 0.7084,
      "step": 15244
    },
    {
      "epoch": 0.43926122284331237,
      "grad_norm": 0.7436389327049255,
      "learning_rate": 6.21444826782181e-06,
      "loss": 0.734,
      "step": 15245
    },
    {
      "epoch": 0.43929003630496166,
      "grad_norm": 0.7492524981498718,
      "learning_rate": 6.213995625550669e-06,
      "loss": 0.7203,
      "step": 15246
    },
    {
      "epoch": 0.43931884976661095,
      "grad_norm": 0.7658203840255737,
      "learning_rate": 6.2135429727068375e-06,
      "loss": 0.7289,
      "step": 15247
    },
    {
      "epoch": 0.43934766322826024,
      "grad_norm": 0.7471435070037842,
      "learning_rate": 6.213090309294262e-06,
      "loss": 0.7483,
      "step": 15248
    },
    {
      "epoch": 0.43937647668990953,
      "grad_norm": 0.7297144532203674,
      "learning_rate": 6.2126376353168825e-06,
      "loss": 0.7367,
      "step": 15249
    },
    {
      "epoch": 0.4394052901515588,
      "grad_norm": 0.7449117302894592,
      "learning_rate": 6.21218495077864e-06,
      "loss": 0.7368,
      "step": 15250
    },
    {
      "epoch": 0.4394341036132081,
      "grad_norm": 0.7505099177360535,
      "learning_rate": 6.211732255683479e-06,
      "loss": 0.7029,
      "step": 15251
    },
    {
      "epoch": 0.43946291707485735,
      "grad_norm": 0.7263779044151306,
      "learning_rate": 6.211279550035343e-06,
      "loss": 0.717,
      "step": 15252
    },
    {
      "epoch": 0.43949173053650664,
      "grad_norm": 0.7822989821434021,
      "learning_rate": 6.210826833838171e-06,
      "loss": 0.7372,
      "step": 15253
    },
    {
      "epoch": 0.43952054399815593,
      "grad_norm": 0.7391130924224854,
      "learning_rate": 6.210374107095908e-06,
      "loss": 0.6992,
      "step": 15254
    },
    {
      "epoch": 0.4395493574598052,
      "grad_norm": 0.7411826848983765,
      "learning_rate": 6.209921369812496e-06,
      "loss": 0.7388,
      "step": 15255
    },
    {
      "epoch": 0.4395781709214545,
      "grad_norm": 0.7225351333618164,
      "learning_rate": 6.20946862199188e-06,
      "loss": 0.731,
      "step": 15256
    },
    {
      "epoch": 0.4396069843831038,
      "grad_norm": 0.7570134401321411,
      "learning_rate": 6.209015863638e-06,
      "loss": 0.7015,
      "step": 15257
    },
    {
      "epoch": 0.4396357978447531,
      "grad_norm": 0.7863545417785645,
      "learning_rate": 6.2085630947548015e-06,
      "loss": 0.7063,
      "step": 15258
    },
    {
      "epoch": 0.43966461130640233,
      "grad_norm": 0.7408368587493896,
      "learning_rate": 6.208110315346225e-06,
      "loss": 0.6943,
      "step": 15259
    },
    {
      "epoch": 0.4396934247680516,
      "grad_norm": 0.7824837565422058,
      "learning_rate": 6.207657525416217e-06,
      "loss": 0.7508,
      "step": 15260
    },
    {
      "epoch": 0.4397222382297009,
      "grad_norm": 0.7697440981864929,
      "learning_rate": 6.207204724968717e-06,
      "loss": 0.7446,
      "step": 15261
    },
    {
      "epoch": 0.4397510516913502,
      "grad_norm": 0.7887468338012695,
      "learning_rate": 6.206751914007673e-06,
      "loss": 0.7451,
      "step": 15262
    },
    {
      "epoch": 0.4397798651529995,
      "grad_norm": 0.7763252854347229,
      "learning_rate": 6.206299092537025e-06,
      "loss": 0.7236,
      "step": 15263
    },
    {
      "epoch": 0.4398086786146488,
      "grad_norm": 0.7573784589767456,
      "learning_rate": 6.2058462605607185e-06,
      "loss": 0.7499,
      "step": 15264
    },
    {
      "epoch": 0.439837492076298,
      "grad_norm": 0.7475622296333313,
      "learning_rate": 6.2053934180826945e-06,
      "loss": 0.7362,
      "step": 15265
    },
    {
      "epoch": 0.4398663055379473,
      "grad_norm": 0.7423236966133118,
      "learning_rate": 6.204940565106901e-06,
      "loss": 0.7128,
      "step": 15266
    },
    {
      "epoch": 0.4398951189995966,
      "grad_norm": 0.7680919170379639,
      "learning_rate": 6.204487701637279e-06,
      "loss": 0.7312,
      "step": 15267
    },
    {
      "epoch": 0.4399239324612459,
      "grad_norm": 0.7434064745903015,
      "learning_rate": 6.2040348276777725e-06,
      "loss": 0.7087,
      "step": 15268
    },
    {
      "epoch": 0.4399527459228952,
      "grad_norm": 0.738036572933197,
      "learning_rate": 6.203581943232326e-06,
      "loss": 0.7206,
      "step": 15269
    },
    {
      "epoch": 0.4399815593845445,
      "grad_norm": 0.7356757521629333,
      "learning_rate": 6.203129048304885e-06,
      "loss": 0.7109,
      "step": 15270
    },
    {
      "epoch": 0.44001037284619376,
      "grad_norm": 0.7898030877113342,
      "learning_rate": 6.202676142899391e-06,
      "loss": 0.7431,
      "step": 15271
    },
    {
      "epoch": 0.440039186307843,
      "grad_norm": 0.7285081744194031,
      "learning_rate": 6.202223227019791e-06,
      "loss": 0.729,
      "step": 15272
    },
    {
      "epoch": 0.4400679997694923,
      "grad_norm": 0.7828373312950134,
      "learning_rate": 6.201770300670029e-06,
      "loss": 0.7305,
      "step": 15273
    },
    {
      "epoch": 0.4400968132311416,
      "grad_norm": 0.7321763634681702,
      "learning_rate": 6.20131736385405e-06,
      "loss": 0.7036,
      "step": 15274
    },
    {
      "epoch": 0.44012562669279087,
      "grad_norm": 0.7442046999931335,
      "learning_rate": 6.200864416575796e-06,
      "loss": 0.721,
      "step": 15275
    },
    {
      "epoch": 0.44015444015444016,
      "grad_norm": 0.7160236835479736,
      "learning_rate": 6.200411458839215e-06,
      "loss": 0.7361,
      "step": 15276
    },
    {
      "epoch": 0.44018325361608945,
      "grad_norm": 0.7609564661979675,
      "learning_rate": 6.199958490648248e-06,
      "loss": 0.7226,
      "step": 15277
    },
    {
      "epoch": 0.44021206707773874,
      "grad_norm": 0.7471807599067688,
      "learning_rate": 6.1995055120068425e-06,
      "loss": 0.7399,
      "step": 15278
    },
    {
      "epoch": 0.440240880539388,
      "grad_norm": 0.7546960115432739,
      "learning_rate": 6.199052522918944e-06,
      "loss": 0.7248,
      "step": 15279
    },
    {
      "epoch": 0.44026969400103727,
      "grad_norm": 0.7304674983024597,
      "learning_rate": 6.198599523388496e-06,
      "loss": 0.7119,
      "step": 15280
    },
    {
      "epoch": 0.44029850746268656,
      "grad_norm": 0.7283404469490051,
      "learning_rate": 6.198146513419445e-06,
      "loss": 0.7259,
      "step": 15281
    },
    {
      "epoch": 0.44032732092433585,
      "grad_norm": 0.7562888860702515,
      "learning_rate": 6.197693493015735e-06,
      "loss": 0.7328,
      "step": 15282
    },
    {
      "epoch": 0.44035613438598514,
      "grad_norm": 0.7769057750701904,
      "learning_rate": 6.197240462181312e-06,
      "loss": 0.7493,
      "step": 15283
    },
    {
      "epoch": 0.44038494784763443,
      "grad_norm": 0.7290241718292236,
      "learning_rate": 6.196787420920122e-06,
      "loss": 0.6959,
      "step": 15284
    },
    {
      "epoch": 0.4404137613092837,
      "grad_norm": 0.7468346953392029,
      "learning_rate": 6.196334369236109e-06,
      "loss": 0.7296,
      "step": 15285
    },
    {
      "epoch": 0.44044257477093296,
      "grad_norm": 0.7509096264839172,
      "learning_rate": 6.19588130713322e-06,
      "loss": 0.7215,
      "step": 15286
    },
    {
      "epoch": 0.44047138823258225,
      "grad_norm": 0.7818681597709656,
      "learning_rate": 6.1954282346154005e-06,
      "loss": 0.7422,
      "step": 15287
    },
    {
      "epoch": 0.44050020169423154,
      "grad_norm": 0.7216747999191284,
      "learning_rate": 6.194975151686595e-06,
      "loss": 0.7147,
      "step": 15288
    },
    {
      "epoch": 0.44052901515588083,
      "grad_norm": 0.7899408936500549,
      "learning_rate": 6.194522058350751e-06,
      "loss": 0.7269,
      "step": 15289
    },
    {
      "epoch": 0.4405578286175301,
      "grad_norm": 0.7265111804008484,
      "learning_rate": 6.194068954611814e-06,
      "loss": 0.7264,
      "step": 15290
    },
    {
      "epoch": 0.4405866420791794,
      "grad_norm": 0.816444993019104,
      "learning_rate": 6.1936158404737304e-06,
      "loss": 0.7021,
      "step": 15291
    },
    {
      "epoch": 0.44061545554082865,
      "grad_norm": 0.7526209354400635,
      "learning_rate": 6.193162715940445e-06,
      "loss": 0.7176,
      "step": 15292
    },
    {
      "epoch": 0.44064426900247794,
      "grad_norm": 0.7259186506271362,
      "learning_rate": 6.192709581015906e-06,
      "loss": 0.7482,
      "step": 15293
    },
    {
      "epoch": 0.44067308246412723,
      "grad_norm": 0.7527778148651123,
      "learning_rate": 6.192256435704059e-06,
      "loss": 0.7256,
      "step": 15294
    },
    {
      "epoch": 0.4407018959257765,
      "grad_norm": 0.7676270008087158,
      "learning_rate": 6.191803280008849e-06,
      "loss": 0.7059,
      "step": 15295
    },
    {
      "epoch": 0.4407307093874258,
      "grad_norm": 0.7599726915359497,
      "learning_rate": 6.191350113934226e-06,
      "loss": 0.7116,
      "step": 15296
    },
    {
      "epoch": 0.4407595228490751,
      "grad_norm": 0.7797514200210571,
      "learning_rate": 6.190896937484133e-06,
      "loss": 0.7368,
      "step": 15297
    },
    {
      "epoch": 0.4407883363107244,
      "grad_norm": 0.7431712746620178,
      "learning_rate": 6.190443750662518e-06,
      "loss": 0.7121,
      "step": 15298
    },
    {
      "epoch": 0.44081714977237363,
      "grad_norm": 0.82438725233078,
      "learning_rate": 6.189990553473329e-06,
      "loss": 0.7009,
      "step": 15299
    },
    {
      "epoch": 0.4408459632340229,
      "grad_norm": 0.774603009223938,
      "learning_rate": 6.189537345920511e-06,
      "loss": 0.7243,
      "step": 15300
    },
    {
      "epoch": 0.4408747766956722,
      "grad_norm": 0.8387705087661743,
      "learning_rate": 6.189084128008012e-06,
      "loss": 0.737,
      "step": 15301
    },
    {
      "epoch": 0.4409035901573215,
      "grad_norm": 0.7802671790122986,
      "learning_rate": 6.188630899739778e-06,
      "loss": 0.732,
      "step": 15302
    },
    {
      "epoch": 0.4409324036189708,
      "grad_norm": 0.7652105093002319,
      "learning_rate": 6.188177661119758e-06,
      "loss": 0.7363,
      "step": 15303
    },
    {
      "epoch": 0.4409612170806201,
      "grad_norm": 0.7520017027854919,
      "learning_rate": 6.1877244121518985e-06,
      "loss": 0.7413,
      "step": 15304
    },
    {
      "epoch": 0.4409900305422694,
      "grad_norm": 0.7761738300323486,
      "learning_rate": 6.187271152840146e-06,
      "loss": 0.7348,
      "step": 15305
    },
    {
      "epoch": 0.4410188440039186,
      "grad_norm": 0.7552495002746582,
      "learning_rate": 6.18681788318845e-06,
      "loss": 0.7185,
      "step": 15306
    },
    {
      "epoch": 0.4410476574655679,
      "grad_norm": 0.7378146052360535,
      "learning_rate": 6.186364603200754e-06,
      "loss": 0.7187,
      "step": 15307
    },
    {
      "epoch": 0.4410764709272172,
      "grad_norm": 0.7688295245170593,
      "learning_rate": 6.18591131288101e-06,
      "loss": 0.7358,
      "step": 15308
    },
    {
      "epoch": 0.4411052843888665,
      "grad_norm": 0.7413017749786377,
      "learning_rate": 6.185458012233163e-06,
      "loss": 0.7124,
      "step": 15309
    },
    {
      "epoch": 0.44113409785051577,
      "grad_norm": 0.7422394156455994,
      "learning_rate": 6.185004701261162e-06,
      "loss": 0.7316,
      "step": 15310
    },
    {
      "epoch": 0.44116291131216506,
      "grad_norm": 0.7301048040390015,
      "learning_rate": 6.184551379968956e-06,
      "loss": 0.7173,
      "step": 15311
    },
    {
      "epoch": 0.44119172477381435,
      "grad_norm": 0.7647614479064941,
      "learning_rate": 6.184098048360491e-06,
      "loss": 0.7513,
      "step": 15312
    },
    {
      "epoch": 0.4412205382354636,
      "grad_norm": 0.7342881560325623,
      "learning_rate": 6.183644706439714e-06,
      "loss": 0.7393,
      "step": 15313
    },
    {
      "epoch": 0.4412493516971129,
      "grad_norm": 0.7144263386726379,
      "learning_rate": 6.183191354210577e-06,
      "loss": 0.7169,
      "step": 15314
    },
    {
      "epoch": 0.44127816515876217,
      "grad_norm": 0.7606735229492188,
      "learning_rate": 6.1827379916770245e-06,
      "loss": 0.7154,
      "step": 15315
    },
    {
      "epoch": 0.44130697862041146,
      "grad_norm": 0.7162841558456421,
      "learning_rate": 6.182284618843008e-06,
      "loss": 0.7203,
      "step": 15316
    },
    {
      "epoch": 0.44133579208206075,
      "grad_norm": 0.7500267624855042,
      "learning_rate": 6.181831235712473e-06,
      "loss": 0.7239,
      "step": 15317
    },
    {
      "epoch": 0.44136460554371004,
      "grad_norm": 0.7660659551620483,
      "learning_rate": 6.181377842289371e-06,
      "loss": 0.7494,
      "step": 15318
    },
    {
      "epoch": 0.4413934190053593,
      "grad_norm": 0.7548140287399292,
      "learning_rate": 6.180924438577647e-06,
      "loss": 0.7397,
      "step": 15319
    },
    {
      "epoch": 0.44142223246700857,
      "grad_norm": 0.7517357468605042,
      "learning_rate": 6.180471024581251e-06,
      "loss": 0.7175,
      "step": 15320
    },
    {
      "epoch": 0.44145104592865786,
      "grad_norm": 0.7245396375656128,
      "learning_rate": 6.180017600304134e-06,
      "loss": 0.7166,
      "step": 15321
    },
    {
      "epoch": 0.44147985939030715,
      "grad_norm": 0.7591111063957214,
      "learning_rate": 6.179564165750245e-06,
      "loss": 0.7503,
      "step": 15322
    },
    {
      "epoch": 0.44150867285195644,
      "grad_norm": 0.7287992238998413,
      "learning_rate": 6.179110720923529e-06,
      "loss": 0.7153,
      "step": 15323
    },
    {
      "epoch": 0.44153748631360573,
      "grad_norm": 0.7842256426811218,
      "learning_rate": 6.1786572658279376e-06,
      "loss": 0.7296,
      "step": 15324
    },
    {
      "epoch": 0.441566299775255,
      "grad_norm": 0.7568082809448242,
      "learning_rate": 6.178203800467418e-06,
      "loss": 0.7234,
      "step": 15325
    },
    {
      "epoch": 0.44159511323690426,
      "grad_norm": 0.7906664609909058,
      "learning_rate": 6.1777503248459235e-06,
      "loss": 0.7227,
      "step": 15326
    },
    {
      "epoch": 0.44162392669855355,
      "grad_norm": 0.7539035081863403,
      "learning_rate": 6.1772968389674004e-06,
      "loss": 0.7257,
      "step": 15327
    },
    {
      "epoch": 0.44165274016020284,
      "grad_norm": 0.7591491937637329,
      "learning_rate": 6.176843342835798e-06,
      "loss": 0.7494,
      "step": 15328
    },
    {
      "epoch": 0.44168155362185213,
      "grad_norm": 0.7762150764465332,
      "learning_rate": 6.176389836455067e-06,
      "loss": 0.7209,
      "step": 15329
    },
    {
      "epoch": 0.4417103670835014,
      "grad_norm": 0.7898803949356079,
      "learning_rate": 6.175936319829157e-06,
      "loss": 0.7189,
      "step": 15330
    },
    {
      "epoch": 0.4417391805451507,
      "grad_norm": 0.7311180830001831,
      "learning_rate": 6.175482792962015e-06,
      "loss": 0.7362,
      "step": 15331
    },
    {
      "epoch": 0.4417679940068,
      "grad_norm": 0.737220048904419,
      "learning_rate": 6.1750292558575955e-06,
      "loss": 0.722,
      "step": 15332
    },
    {
      "epoch": 0.44179680746844924,
      "grad_norm": 0.7497318387031555,
      "learning_rate": 6.174575708519844e-06,
      "loss": 0.733,
      "step": 15333
    },
    {
      "epoch": 0.44182562093009853,
      "grad_norm": 0.7584882378578186,
      "learning_rate": 6.174122150952714e-06,
      "loss": 0.6982,
      "step": 15334
    },
    {
      "epoch": 0.4418544343917478,
      "grad_norm": 0.7425776124000549,
      "learning_rate": 6.173668583160153e-06,
      "loss": 0.6914,
      "step": 15335
    },
    {
      "epoch": 0.4418832478533971,
      "grad_norm": 0.739646315574646,
      "learning_rate": 6.173215005146111e-06,
      "loss": 0.7386,
      "step": 15336
    },
    {
      "epoch": 0.4419120613150464,
      "grad_norm": 0.7596250176429749,
      "learning_rate": 6.17276141691454e-06,
      "loss": 0.7317,
      "step": 15337
    },
    {
      "epoch": 0.4419408747766957,
      "grad_norm": 0.7734020352363586,
      "learning_rate": 6.17230781846939e-06,
      "loss": 0.7496,
      "step": 15338
    },
    {
      "epoch": 0.441969688238345,
      "grad_norm": 0.7745952606201172,
      "learning_rate": 6.17185420981461e-06,
      "loss": 0.7411,
      "step": 15339
    },
    {
      "epoch": 0.4419985016999942,
      "grad_norm": 0.7307266592979431,
      "learning_rate": 6.171400590954151e-06,
      "loss": 0.7365,
      "step": 15340
    },
    {
      "epoch": 0.4420273151616435,
      "grad_norm": 0.7706065773963928,
      "learning_rate": 6.170946961891964e-06,
      "loss": 0.7166,
      "step": 15341
    },
    {
      "epoch": 0.4420561286232928,
      "grad_norm": 0.7396848201751709,
      "learning_rate": 6.170493322632e-06,
      "loss": 0.7068,
      "step": 15342
    },
    {
      "epoch": 0.4420849420849421,
      "grad_norm": 0.7426074147224426,
      "learning_rate": 6.170039673178208e-06,
      "loss": 0.7291,
      "step": 15343
    },
    {
      "epoch": 0.4421137555465914,
      "grad_norm": 0.7874369025230408,
      "learning_rate": 6.169586013534541e-06,
      "loss": 0.7355,
      "step": 15344
    },
    {
      "epoch": 0.44214256900824067,
      "grad_norm": 0.727537214756012,
      "learning_rate": 6.169132343704949e-06,
      "loss": 0.7003,
      "step": 15345
    },
    {
      "epoch": 0.4421713824698899,
      "grad_norm": 0.7700155973434448,
      "learning_rate": 6.168678663693383e-06,
      "loss": 0.7103,
      "step": 15346
    },
    {
      "epoch": 0.4422001959315392,
      "grad_norm": 0.7385095953941345,
      "learning_rate": 6.168224973503793e-06,
      "loss": 0.7525,
      "step": 15347
    },
    {
      "epoch": 0.4422290093931885,
      "grad_norm": 0.7041547298431396,
      "learning_rate": 6.167771273140132e-06,
      "loss": 0.7142,
      "step": 15348
    },
    {
      "epoch": 0.4422578228548378,
      "grad_norm": 0.7225295901298523,
      "learning_rate": 6.16731756260635e-06,
      "loss": 0.7256,
      "step": 15349
    },
    {
      "epoch": 0.44228663631648707,
      "grad_norm": 0.7530107498168945,
      "learning_rate": 6.1668638419064e-06,
      "loss": 0.726,
      "step": 15350
    },
    {
      "epoch": 0.44231544977813636,
      "grad_norm": 0.7558289766311646,
      "learning_rate": 6.166410111044232e-06,
      "loss": 0.7323,
      "step": 15351
    },
    {
      "epoch": 0.44234426323978565,
      "grad_norm": 0.72817462682724,
      "learning_rate": 6.165956370023798e-06,
      "loss": 0.7359,
      "step": 15352
    },
    {
      "epoch": 0.4423730767014349,
      "grad_norm": 0.7325452566146851,
      "learning_rate": 6.165502618849049e-06,
      "loss": 0.7635,
      "step": 15353
    },
    {
      "epoch": 0.4424018901630842,
      "grad_norm": 0.7563667893409729,
      "learning_rate": 6.165048857523939e-06,
      "loss": 0.7294,
      "step": 15354
    },
    {
      "epoch": 0.44243070362473347,
      "grad_norm": 0.7344677448272705,
      "learning_rate": 6.1645950860524165e-06,
      "loss": 0.7373,
      "step": 15355
    },
    {
      "epoch": 0.44245951708638276,
      "grad_norm": 0.7364979982376099,
      "learning_rate": 6.164141304438434e-06,
      "loss": 0.7394,
      "step": 15356
    },
    {
      "epoch": 0.44248833054803205,
      "grad_norm": 0.7379264235496521,
      "learning_rate": 6.163687512685946e-06,
      "loss": 0.7174,
      "step": 15357
    },
    {
      "epoch": 0.44251714400968134,
      "grad_norm": 0.7682889699935913,
      "learning_rate": 6.163233710798904e-06,
      "loss": 0.7317,
      "step": 15358
    },
    {
      "epoch": 0.44254595747133063,
      "grad_norm": 0.7525418400764465,
      "learning_rate": 6.162779898781257e-06,
      "loss": 0.7256,
      "step": 15359
    },
    {
      "epoch": 0.44257477093297987,
      "grad_norm": 0.7164143919944763,
      "learning_rate": 6.162326076636961e-06,
      "loss": 0.7319,
      "step": 15360
    },
    {
      "epoch": 0.44260358439462916,
      "grad_norm": 0.7337435483932495,
      "learning_rate": 6.161872244369965e-06,
      "loss": 0.7042,
      "step": 15361
    },
    {
      "epoch": 0.44263239785627845,
      "grad_norm": 0.7563109993934631,
      "learning_rate": 6.161418401984225e-06,
      "loss": 0.7156,
      "step": 15362
    },
    {
      "epoch": 0.44266121131792774,
      "grad_norm": 0.7442519664764404,
      "learning_rate": 6.160964549483691e-06,
      "loss": 0.7183,
      "step": 15363
    },
    {
      "epoch": 0.44269002477957703,
      "grad_norm": 0.7397215366363525,
      "learning_rate": 6.160510686872317e-06,
      "loss": 0.7036,
      "step": 15364
    },
    {
      "epoch": 0.4427188382412263,
      "grad_norm": 0.7612791657447815,
      "learning_rate": 6.160056814154054e-06,
      "loss": 0.7386,
      "step": 15365
    },
    {
      "epoch": 0.4427476517028756,
      "grad_norm": 0.752701997756958,
      "learning_rate": 6.159602931332855e-06,
      "loss": 0.7222,
      "step": 15366
    },
    {
      "epoch": 0.44277646516452485,
      "grad_norm": 0.7714448571205139,
      "learning_rate": 6.159149038412676e-06,
      "loss": 0.7574,
      "step": 15367
    },
    {
      "epoch": 0.44280527862617414,
      "grad_norm": 0.7597922682762146,
      "learning_rate": 6.158695135397466e-06,
      "loss": 0.7201,
      "step": 15368
    },
    {
      "epoch": 0.4428340920878234,
      "grad_norm": 0.791650116443634,
      "learning_rate": 6.158241222291179e-06,
      "loss": 0.7279,
      "step": 15369
    },
    {
      "epoch": 0.4428629055494727,
      "grad_norm": 0.7580758929252625,
      "learning_rate": 6.157787299097771e-06,
      "loss": 0.7576,
      "step": 15370
    },
    {
      "epoch": 0.442891719011122,
      "grad_norm": 0.7193683385848999,
      "learning_rate": 6.157333365821192e-06,
      "loss": 0.7286,
      "step": 15371
    },
    {
      "epoch": 0.4429205324727713,
      "grad_norm": 0.7619051933288574,
      "learning_rate": 6.156879422465396e-06,
      "loss": 0.7262,
      "step": 15372
    },
    {
      "epoch": 0.44294934593442054,
      "grad_norm": 0.7769802808761597,
      "learning_rate": 6.156425469034336e-06,
      "loss": 0.7383,
      "step": 15373
    },
    {
      "epoch": 0.4429781593960698,
      "grad_norm": 0.7841039896011353,
      "learning_rate": 6.155971505531967e-06,
      "loss": 0.736,
      "step": 15374
    },
    {
      "epoch": 0.4430069728577191,
      "grad_norm": 0.7706829905509949,
      "learning_rate": 6.155517531962242e-06,
      "loss": 0.7353,
      "step": 15375
    },
    {
      "epoch": 0.4430357863193684,
      "grad_norm": 0.732950747013092,
      "learning_rate": 6.155063548329115e-06,
      "loss": 0.7099,
      "step": 15376
    },
    {
      "epoch": 0.4430645997810177,
      "grad_norm": 0.7466595768928528,
      "learning_rate": 6.154609554636538e-06,
      "loss": 0.7105,
      "step": 15377
    },
    {
      "epoch": 0.443093413242667,
      "grad_norm": 0.7695491909980774,
      "learning_rate": 6.154155550888467e-06,
      "loss": 0.7177,
      "step": 15378
    },
    {
      "epoch": 0.4431222267043163,
      "grad_norm": 0.7295702695846558,
      "learning_rate": 6.153701537088852e-06,
      "loss": 0.7471,
      "step": 15379
    },
    {
      "epoch": 0.4431510401659655,
      "grad_norm": 0.7410559058189392,
      "learning_rate": 6.153247513241652e-06,
      "loss": 0.7231,
      "step": 15380
    },
    {
      "epoch": 0.4431798536276148,
      "grad_norm": 0.7222655415534973,
      "learning_rate": 6.152793479350819e-06,
      "loss": 0.7276,
      "step": 15381
    },
    {
      "epoch": 0.4432086670892641,
      "grad_norm": 0.7480906248092651,
      "learning_rate": 6.152339435420308e-06,
      "loss": 0.6957,
      "step": 15382
    },
    {
      "epoch": 0.4432374805509134,
      "grad_norm": 0.731249988079071,
      "learning_rate": 6.151885381454071e-06,
      "loss": 0.7245,
      "step": 15383
    },
    {
      "epoch": 0.4432662940125627,
      "grad_norm": 0.7640938758850098,
      "learning_rate": 6.151431317456065e-06,
      "loss": 0.746,
      "step": 15384
    },
    {
      "epoch": 0.44329510747421197,
      "grad_norm": 0.7560901641845703,
      "learning_rate": 6.150977243430242e-06,
      "loss": 0.7208,
      "step": 15385
    },
    {
      "epoch": 0.44332392093586126,
      "grad_norm": 0.7780019640922546,
      "learning_rate": 6.150523159380558e-06,
      "loss": 0.7477,
      "step": 15386
    },
    {
      "epoch": 0.4433527343975105,
      "grad_norm": 0.7782517671585083,
      "learning_rate": 6.1500690653109675e-06,
      "loss": 0.7439,
      "step": 15387
    },
    {
      "epoch": 0.4433815478591598,
      "grad_norm": 0.7497801184654236,
      "learning_rate": 6.149614961225425e-06,
      "loss": 0.7094,
      "step": 15388
    },
    {
      "epoch": 0.4434103613208091,
      "grad_norm": 0.7350432872772217,
      "learning_rate": 6.149160847127886e-06,
      "loss": 0.7045,
      "step": 15389
    },
    {
      "epoch": 0.44343917478245837,
      "grad_norm": 0.7845161557197571,
      "learning_rate": 6.148706723022305e-06,
      "loss": 0.7339,
      "step": 15390
    },
    {
      "epoch": 0.44346798824410766,
      "grad_norm": 0.796843945980072,
      "learning_rate": 6.148252588912635e-06,
      "loss": 0.7241,
      "step": 15391
    },
    {
      "epoch": 0.44349680170575695,
      "grad_norm": 0.7995924353599548,
      "learning_rate": 6.147798444802835e-06,
      "loss": 0.745,
      "step": 15392
    },
    {
      "epoch": 0.44352561516740624,
      "grad_norm": 0.7572984099388123,
      "learning_rate": 6.1473442906968565e-06,
      "loss": 0.7184,
      "step": 15393
    },
    {
      "epoch": 0.4435544286290555,
      "grad_norm": 0.7218140959739685,
      "learning_rate": 6.146890126598657e-06,
      "loss": 0.7193,
      "step": 15394
    },
    {
      "epoch": 0.44358324209070477,
      "grad_norm": 0.7468966841697693,
      "learning_rate": 6.146435952512191e-06,
      "loss": 0.7252,
      "step": 15395
    },
    {
      "epoch": 0.44361205555235406,
      "grad_norm": 0.8445331454277039,
      "learning_rate": 6.1459817684414135e-06,
      "loss": 0.7219,
      "step": 15396
    },
    {
      "epoch": 0.44364086901400335,
      "grad_norm": 0.8068002462387085,
      "learning_rate": 6.145527574390279e-06,
      "loss": 0.723,
      "step": 15397
    },
    {
      "epoch": 0.44366968247565264,
      "grad_norm": 0.7628556489944458,
      "learning_rate": 6.1450733703627465e-06,
      "loss": 0.7172,
      "step": 15398
    },
    {
      "epoch": 0.44369849593730193,
      "grad_norm": 0.710496723651886,
      "learning_rate": 6.144619156362768e-06,
      "loss": 0.7145,
      "step": 15399
    },
    {
      "epoch": 0.44372730939895116,
      "grad_norm": 0.7714911103248596,
      "learning_rate": 6.144164932394302e-06,
      "loss": 0.738,
      "step": 15400
    },
    {
      "epoch": 0.44375612286060045,
      "grad_norm": 0.7681055665016174,
      "learning_rate": 6.143710698461302e-06,
      "loss": 0.7273,
      "step": 15401
    },
    {
      "epoch": 0.44378493632224975,
      "grad_norm": 0.7685766816139221,
      "learning_rate": 6.143256454567728e-06,
      "loss": 0.7466,
      "step": 15402
    },
    {
      "epoch": 0.44381374978389904,
      "grad_norm": 0.7381992936134338,
      "learning_rate": 6.1428022007175295e-06,
      "loss": 0.7333,
      "step": 15403
    },
    {
      "epoch": 0.4438425632455483,
      "grad_norm": 0.7503634095191956,
      "learning_rate": 6.1423479369146675e-06,
      "loss": 0.7159,
      "step": 15404
    },
    {
      "epoch": 0.4438713767071976,
      "grad_norm": 0.7490809559822083,
      "learning_rate": 6.141893663163098e-06,
      "loss": 0.7158,
      "step": 15405
    },
    {
      "epoch": 0.4439001901688469,
      "grad_norm": 0.7399235963821411,
      "learning_rate": 6.141439379466774e-06,
      "loss": 0.7055,
      "step": 15406
    },
    {
      "epoch": 0.44392900363049614,
      "grad_norm": 0.7726225256919861,
      "learning_rate": 6.140985085829656e-06,
      "loss": 0.7317,
      "step": 15407
    },
    {
      "epoch": 0.44395781709214543,
      "grad_norm": 0.7478116154670715,
      "learning_rate": 6.140530782255698e-06,
      "loss": 0.7299,
      "step": 15408
    },
    {
      "epoch": 0.4439866305537947,
      "grad_norm": 0.7178694605827332,
      "learning_rate": 6.1400764687488555e-06,
      "loss": 0.7051,
      "step": 15409
    },
    {
      "epoch": 0.444015444015444,
      "grad_norm": 0.7593804597854614,
      "learning_rate": 6.139622145313089e-06,
      "loss": 0.7182,
      "step": 15410
    },
    {
      "epoch": 0.4440442574770933,
      "grad_norm": 0.736320972442627,
      "learning_rate": 6.139167811952351e-06,
      "loss": 0.7161,
      "step": 15411
    },
    {
      "epoch": 0.4440730709387426,
      "grad_norm": 0.7876901030540466,
      "learning_rate": 6.138713468670601e-06,
      "loss": 0.7374,
      "step": 15412
    },
    {
      "epoch": 0.4441018844003919,
      "grad_norm": 0.6987826228141785,
      "learning_rate": 6.138259115471795e-06,
      "loss": 0.6995,
      "step": 15413
    },
    {
      "epoch": 0.4441306978620411,
      "grad_norm": 0.7650513052940369,
      "learning_rate": 6.13780475235989e-06,
      "loss": 0.7439,
      "step": 15414
    },
    {
      "epoch": 0.4441595113236904,
      "grad_norm": 0.7649845480918884,
      "learning_rate": 6.137350379338842e-06,
      "loss": 0.7108,
      "step": 15415
    },
    {
      "epoch": 0.4441883247853397,
      "grad_norm": 0.9030205607414246,
      "learning_rate": 6.136895996412611e-06,
      "loss": 0.707,
      "step": 15416
    },
    {
      "epoch": 0.444217138246989,
      "grad_norm": 0.7130142450332642,
      "learning_rate": 6.13644160358515e-06,
      "loss": 0.7125,
      "step": 15417
    },
    {
      "epoch": 0.4442459517086383,
      "grad_norm": 0.7306005954742432,
      "learning_rate": 6.1359872008604215e-06,
      "loss": 0.7272,
      "step": 15418
    },
    {
      "epoch": 0.4442747651702876,
      "grad_norm": 0.7361530661582947,
      "learning_rate": 6.1355327882423775e-06,
      "loss": 0.7276,
      "step": 15419
    },
    {
      "epoch": 0.44430357863193687,
      "grad_norm": 0.777225911617279,
      "learning_rate": 6.13507836573498e-06,
      "loss": 0.7487,
      "step": 15420
    },
    {
      "epoch": 0.4443323920935861,
      "grad_norm": 0.7523325085639954,
      "learning_rate": 6.1346239333421835e-06,
      "loss": 0.7516,
      "step": 15421
    },
    {
      "epoch": 0.4443612055552354,
      "grad_norm": 0.7145194411277771,
      "learning_rate": 6.1341694910679465e-06,
      "loss": 0.7209,
      "step": 15422
    },
    {
      "epoch": 0.4443900190168847,
      "grad_norm": 0.7154605984687805,
      "learning_rate": 6.133715038916227e-06,
      "loss": 0.712,
      "step": 15423
    },
    {
      "epoch": 0.444418832478534,
      "grad_norm": 0.7619231939315796,
      "learning_rate": 6.133260576890983e-06,
      "loss": 0.7137,
      "step": 15424
    },
    {
      "epoch": 0.44444764594018327,
      "grad_norm": 0.7270449995994568,
      "learning_rate": 6.132806104996172e-06,
      "loss": 0.7255,
      "step": 15425
    },
    {
      "epoch": 0.44447645940183256,
      "grad_norm": 0.7316169738769531,
      "learning_rate": 6.1323516232357525e-06,
      "loss": 0.7118,
      "step": 15426
    },
    {
      "epoch": 0.4445052728634818,
      "grad_norm": 0.7680772542953491,
      "learning_rate": 6.131897131613683e-06,
      "loss": 0.732,
      "step": 15427
    },
    {
      "epoch": 0.4445340863251311,
      "grad_norm": 0.7359438538551331,
      "learning_rate": 6.1314426301339194e-06,
      "loss": 0.7251,
      "step": 15428
    },
    {
      "epoch": 0.4445628997867804,
      "grad_norm": 0.7615799903869629,
      "learning_rate": 6.130988118800423e-06,
      "loss": 0.7237,
      "step": 15429
    },
    {
      "epoch": 0.44459171324842967,
      "grad_norm": 0.7509126663208008,
      "learning_rate": 6.13053359761715e-06,
      "loss": 0.7336,
      "step": 15430
    },
    {
      "epoch": 0.44462052671007896,
      "grad_norm": 0.740966796875,
      "learning_rate": 6.13007906658806e-06,
      "loss": 0.7412,
      "step": 15431
    },
    {
      "epoch": 0.44464934017172825,
      "grad_norm": 0.727592408657074,
      "learning_rate": 6.1296245257171104e-06,
      "loss": 0.7062,
      "step": 15432
    },
    {
      "epoch": 0.44467815363337754,
      "grad_norm": 0.757945716381073,
      "learning_rate": 6.12916997500826e-06,
      "loss": 0.7065,
      "step": 15433
    },
    {
      "epoch": 0.4447069670950268,
      "grad_norm": 0.7393312454223633,
      "learning_rate": 6.128715414465469e-06,
      "loss": 0.756,
      "step": 15434
    },
    {
      "epoch": 0.44473578055667606,
      "grad_norm": 0.7681131362915039,
      "learning_rate": 6.128260844092695e-06,
      "loss": 0.7091,
      "step": 15435
    },
    {
      "epoch": 0.44476459401832535,
      "grad_norm": 0.7528800964355469,
      "learning_rate": 6.127806263893896e-06,
      "loss": 0.7217,
      "step": 15436
    },
    {
      "epoch": 0.44479340747997465,
      "grad_norm": 0.7391312718391418,
      "learning_rate": 6.127351673873033e-06,
      "loss": 0.7155,
      "step": 15437
    },
    {
      "epoch": 0.44482222094162394,
      "grad_norm": 0.7296426892280579,
      "learning_rate": 6.1268970740340625e-06,
      "loss": 0.7211,
      "step": 15438
    },
    {
      "epoch": 0.4448510344032732,
      "grad_norm": 0.7669895887374878,
      "learning_rate": 6.126442464380945e-06,
      "loss": 0.7186,
      "step": 15439
    },
    {
      "epoch": 0.4448798478649225,
      "grad_norm": 0.756522536277771,
      "learning_rate": 6.125987844917641e-06,
      "loss": 0.7627,
      "step": 15440
    },
    {
      "epoch": 0.44490866132657175,
      "grad_norm": 0.772100031375885,
      "learning_rate": 6.125533215648107e-06,
      "loss": 0.7489,
      "step": 15441
    },
    {
      "epoch": 0.44493747478822104,
      "grad_norm": 0.7447546124458313,
      "learning_rate": 6.125078576576305e-06,
      "loss": 0.7266,
      "step": 15442
    },
    {
      "epoch": 0.44496628824987033,
      "grad_norm": 0.7554000020027161,
      "learning_rate": 6.124623927706193e-06,
      "loss": 0.7539,
      "step": 15443
    },
    {
      "epoch": 0.4449951017115196,
      "grad_norm": 0.7886604070663452,
      "learning_rate": 6.12416926904173e-06,
      "loss": 0.7234,
      "step": 15444
    },
    {
      "epoch": 0.4450239151731689,
      "grad_norm": 0.7310032248497009,
      "learning_rate": 6.123714600586878e-06,
      "loss": 0.7219,
      "step": 15445
    },
    {
      "epoch": 0.4450527286348182,
      "grad_norm": 0.740385890007019,
      "learning_rate": 6.123259922345593e-06,
      "loss": 0.7394,
      "step": 15446
    },
    {
      "epoch": 0.4450815420964675,
      "grad_norm": 0.8619644045829773,
      "learning_rate": 6.122805234321838e-06,
      "loss": 0.7446,
      "step": 15447
    },
    {
      "epoch": 0.44511035555811673,
      "grad_norm": 0.7151124477386475,
      "learning_rate": 6.122350536519573e-06,
      "loss": 0.718,
      "step": 15448
    },
    {
      "epoch": 0.445139169019766,
      "grad_norm": 0.7114518880844116,
      "learning_rate": 6.121895828942755e-06,
      "loss": 0.7228,
      "step": 15449
    },
    {
      "epoch": 0.4451679824814153,
      "grad_norm": 0.7419223189353943,
      "learning_rate": 6.1214411115953465e-06,
      "loss": 0.7106,
      "step": 15450
    },
    {
      "epoch": 0.4451967959430646,
      "grad_norm": 0.7543535828590393,
      "learning_rate": 6.120986384481307e-06,
      "loss": 0.7263,
      "step": 15451
    },
    {
      "epoch": 0.4452256094047139,
      "grad_norm": 0.7460193634033203,
      "learning_rate": 6.120531647604596e-06,
      "loss": 0.7399,
      "step": 15452
    },
    {
      "epoch": 0.4452544228663632,
      "grad_norm": 0.7248448133468628,
      "learning_rate": 6.120076900969175e-06,
      "loss": 0.7145,
      "step": 15453
    },
    {
      "epoch": 0.4452832363280124,
      "grad_norm": 0.7625817656517029,
      "learning_rate": 6.1196221445790034e-06,
      "loss": 0.7114,
      "step": 15454
    },
    {
      "epoch": 0.4453120497896617,
      "grad_norm": 0.7536250352859497,
      "learning_rate": 6.119167378438044e-06,
      "loss": 0.7054,
      "step": 15455
    },
    {
      "epoch": 0.445340863251311,
      "grad_norm": 0.7622250318527222,
      "learning_rate": 6.118712602550254e-06,
      "loss": 0.7184,
      "step": 15456
    },
    {
      "epoch": 0.4453696767129603,
      "grad_norm": 0.7575435638427734,
      "learning_rate": 6.118257816919595e-06,
      "loss": 0.7165,
      "step": 15457
    },
    {
      "epoch": 0.4453984901746096,
      "grad_norm": 0.7175551056861877,
      "learning_rate": 6.117803021550028e-06,
      "loss": 0.7351,
      "step": 15458
    },
    {
      "epoch": 0.4454273036362589,
      "grad_norm": 0.7581770420074463,
      "learning_rate": 6.117348216445516e-06,
      "loss": 0.7173,
      "step": 15459
    },
    {
      "epoch": 0.44545611709790817,
      "grad_norm": 0.7682700753211975,
      "learning_rate": 6.116893401610017e-06,
      "loss": 0.7268,
      "step": 15460
    },
    {
      "epoch": 0.4454849305595574,
      "grad_norm": 0.7332660555839539,
      "learning_rate": 6.116438577047493e-06,
      "loss": 0.7304,
      "step": 15461
    },
    {
      "epoch": 0.4455137440212067,
      "grad_norm": 0.736323893070221,
      "learning_rate": 6.115983742761905e-06,
      "loss": 0.7275,
      "step": 15462
    },
    {
      "epoch": 0.445542557482856,
      "grad_norm": 0.7176517248153687,
      "learning_rate": 6.115528898757213e-06,
      "loss": 0.7101,
      "step": 15463
    },
    {
      "epoch": 0.4455713709445053,
      "grad_norm": 0.746425986289978,
      "learning_rate": 6.115074045037381e-06,
      "loss": 0.7169,
      "step": 15464
    },
    {
      "epoch": 0.44560018440615456,
      "grad_norm": 0.7410021424293518,
      "learning_rate": 6.11461918160637e-06,
      "loss": 0.7446,
      "step": 15465
    },
    {
      "epoch": 0.44562899786780386,
      "grad_norm": 0.7252499461174011,
      "learning_rate": 6.114164308468137e-06,
      "loss": 0.7145,
      "step": 15466
    },
    {
      "epoch": 0.44565781132945315,
      "grad_norm": 0.747466504573822,
      "learning_rate": 6.113709425626648e-06,
      "loss": 0.7122,
      "step": 15467
    },
    {
      "epoch": 0.4456866247911024,
      "grad_norm": 0.7335594296455383,
      "learning_rate": 6.113254533085864e-06,
      "loss": 0.7202,
      "step": 15468
    },
    {
      "epoch": 0.4457154382527517,
      "grad_norm": 0.7379734516143799,
      "learning_rate": 6.112799630849744e-06,
      "loss": 0.7149,
      "step": 15469
    },
    {
      "epoch": 0.44574425171440096,
      "grad_norm": 0.7228883504867554,
      "learning_rate": 6.1123447189222526e-06,
      "loss": 0.7165,
      "step": 15470
    },
    {
      "epoch": 0.44577306517605025,
      "grad_norm": 0.734038233757019,
      "learning_rate": 6.111889797307351e-06,
      "loss": 0.7449,
      "step": 15471
    },
    {
      "epoch": 0.44580187863769954,
      "grad_norm": 0.7478382587432861,
      "learning_rate": 6.111434866009e-06,
      "loss": 0.7354,
      "step": 15472
    },
    {
      "epoch": 0.44583069209934884,
      "grad_norm": 0.7313913702964783,
      "learning_rate": 6.110979925031164e-06,
      "loss": 0.7226,
      "step": 15473
    },
    {
      "epoch": 0.44585950556099807,
      "grad_norm": 0.7473092675209045,
      "learning_rate": 6.110524974377802e-06,
      "loss": 0.7409,
      "step": 15474
    },
    {
      "epoch": 0.44588831902264736,
      "grad_norm": 0.7483137845993042,
      "learning_rate": 6.110070014052877e-06,
      "loss": 0.7337,
      "step": 15475
    },
    {
      "epoch": 0.44591713248429665,
      "grad_norm": 0.727321982383728,
      "learning_rate": 6.109615044060353e-06,
      "loss": 0.732,
      "step": 15476
    },
    {
      "epoch": 0.44594594594594594,
      "grad_norm": 0.7444713115692139,
      "learning_rate": 6.109160064404191e-06,
      "loss": 0.7319,
      "step": 15477
    },
    {
      "epoch": 0.44597475940759523,
      "grad_norm": 0.7155830264091492,
      "learning_rate": 6.1087050750883535e-06,
      "loss": 0.724,
      "step": 15478
    },
    {
      "epoch": 0.4460035728692445,
      "grad_norm": 0.7354269623756409,
      "learning_rate": 6.108250076116804e-06,
      "loss": 0.7294,
      "step": 15479
    },
    {
      "epoch": 0.4460323863308938,
      "grad_norm": 0.7361624240875244,
      "learning_rate": 6.107795067493503e-06,
      "loss": 0.7123,
      "step": 15480
    },
    {
      "epoch": 0.44606119979254305,
      "grad_norm": 0.7487308382987976,
      "learning_rate": 6.107340049222413e-06,
      "loss": 0.7099,
      "step": 15481
    },
    {
      "epoch": 0.44609001325419234,
      "grad_norm": 0.7332214713096619,
      "learning_rate": 6.106885021307501e-06,
      "loss": 0.7325,
      "step": 15482
    },
    {
      "epoch": 0.44611882671584163,
      "grad_norm": 0.7202698588371277,
      "learning_rate": 6.106429983752725e-06,
      "loss": 0.7044,
      "step": 15483
    },
    {
      "epoch": 0.4461476401774909,
      "grad_norm": 0.7642117738723755,
      "learning_rate": 6.1059749365620515e-06,
      "loss": 0.7022,
      "step": 15484
    },
    {
      "epoch": 0.4461764536391402,
      "grad_norm": 0.7164748907089233,
      "learning_rate": 6.10551987973944e-06,
      "loss": 0.7035,
      "step": 15485
    },
    {
      "epoch": 0.4462052671007895,
      "grad_norm": 0.7184696793556213,
      "learning_rate": 6.105064813288856e-06,
      "loss": 0.7202,
      "step": 15486
    },
    {
      "epoch": 0.4462340805624388,
      "grad_norm": 0.7528743743896484,
      "learning_rate": 6.104609737214262e-06,
      "loss": 0.7406,
      "step": 15487
    },
    {
      "epoch": 0.44626289402408803,
      "grad_norm": 0.7754421830177307,
      "learning_rate": 6.104154651519622e-06,
      "loss": 0.7264,
      "step": 15488
    },
    {
      "epoch": 0.4462917074857373,
      "grad_norm": 0.7979010343551636,
      "learning_rate": 6.103699556208899e-06,
      "loss": 0.727,
      "step": 15489
    },
    {
      "epoch": 0.4463205209473866,
      "grad_norm": 0.7407504916191101,
      "learning_rate": 6.103244451286056e-06,
      "loss": 0.7062,
      "step": 15490
    },
    {
      "epoch": 0.4463493344090359,
      "grad_norm": 0.7449102401733398,
      "learning_rate": 6.1027893367550576e-06,
      "loss": 0.7367,
      "step": 15491
    },
    {
      "epoch": 0.4463781478706852,
      "grad_norm": 0.735660970211029,
      "learning_rate": 6.102334212619865e-06,
      "loss": 0.7168,
      "step": 15492
    },
    {
      "epoch": 0.4464069613323345,
      "grad_norm": 0.7626729011535645,
      "learning_rate": 6.101879078884443e-06,
      "loss": 0.7289,
      "step": 15493
    },
    {
      "epoch": 0.4464357747939838,
      "grad_norm": 0.7296010255813599,
      "learning_rate": 6.101423935552756e-06,
      "loss": 0.7505,
      "step": 15494
    },
    {
      "epoch": 0.446464588255633,
      "grad_norm": 0.7389209270477295,
      "learning_rate": 6.100968782628768e-06,
      "loss": 0.7383,
      "step": 15495
    },
    {
      "epoch": 0.4464934017172823,
      "grad_norm": 0.7369561791419983,
      "learning_rate": 6.100513620116443e-06,
      "loss": 0.7067,
      "step": 15496
    },
    {
      "epoch": 0.4465222151789316,
      "grad_norm": 0.7815985679626465,
      "learning_rate": 6.100058448019743e-06,
      "loss": 0.7382,
      "step": 15497
    },
    {
      "epoch": 0.4465510286405809,
      "grad_norm": 0.7466460466384888,
      "learning_rate": 6.099603266342636e-06,
      "loss": 0.7432,
      "step": 15498
    },
    {
      "epoch": 0.4465798421022302,
      "grad_norm": 0.719864010810852,
      "learning_rate": 6.0991480750890805e-06,
      "loss": 0.7176,
      "step": 15499
    },
    {
      "epoch": 0.44660865556387946,
      "grad_norm": 0.7892400622367859,
      "learning_rate": 6.098692874263046e-06,
      "loss": 0.7512,
      "step": 15500
    },
    {
      "epoch": 0.4466374690255287,
      "grad_norm": 0.7181862592697144,
      "learning_rate": 6.0982376638684945e-06,
      "loss": 0.7314,
      "step": 15501
    },
    {
      "epoch": 0.446666282487178,
      "grad_norm": 0.7226002216339111,
      "learning_rate": 6.097782443909392e-06,
      "loss": 0.7323,
      "step": 15502
    },
    {
      "epoch": 0.4466950959488273,
      "grad_norm": 0.75860595703125,
      "learning_rate": 6.0973272143897e-06,
      "loss": 0.7235,
      "step": 15503
    },
    {
      "epoch": 0.44672390941047657,
      "grad_norm": 0.7454968094825745,
      "learning_rate": 6.096871975313384e-06,
      "loss": 0.7363,
      "step": 15504
    },
    {
      "epoch": 0.44675272287212586,
      "grad_norm": 0.729597270488739,
      "learning_rate": 6.0964167266844125e-06,
      "loss": 0.7467,
      "step": 15505
    },
    {
      "epoch": 0.44678153633377515,
      "grad_norm": 0.7181832790374756,
      "learning_rate": 6.095961468506744e-06,
      "loss": 0.7414,
      "step": 15506
    },
    {
      "epoch": 0.44681034979542444,
      "grad_norm": 0.7224687337875366,
      "learning_rate": 6.095506200784349e-06,
      "loss": 0.6999,
      "step": 15507
    },
    {
      "epoch": 0.4468391632570737,
      "grad_norm": 0.7174885869026184,
      "learning_rate": 6.095050923521191e-06,
      "loss": 0.7339,
      "step": 15508
    },
    {
      "epoch": 0.44686797671872297,
      "grad_norm": 0.7640635371208191,
      "learning_rate": 6.094595636721232e-06,
      "loss": 0.7408,
      "step": 15509
    },
    {
      "epoch": 0.44689679018037226,
      "grad_norm": 0.7971625328063965,
      "learning_rate": 6.094140340388439e-06,
      "loss": 0.7425,
      "step": 15510
    },
    {
      "epoch": 0.44692560364202155,
      "grad_norm": 0.7869690656661987,
      "learning_rate": 6.093685034526779e-06,
      "loss": 0.7473,
      "step": 15511
    },
    {
      "epoch": 0.44695441710367084,
      "grad_norm": 0.7714492678642273,
      "learning_rate": 6.093229719140215e-06,
      "loss": 0.7583,
      "step": 15512
    },
    {
      "epoch": 0.44698323056532013,
      "grad_norm": 0.7203275561332703,
      "learning_rate": 6.092774394232713e-06,
      "loss": 0.7224,
      "step": 15513
    },
    {
      "epoch": 0.4470120440269694,
      "grad_norm": 0.7717308402061462,
      "learning_rate": 6.092319059808238e-06,
      "loss": 0.7156,
      "step": 15514
    },
    {
      "epoch": 0.44704085748861866,
      "grad_norm": 0.7762143015861511,
      "learning_rate": 6.091863715870757e-06,
      "loss": 0.7385,
      "step": 15515
    },
    {
      "epoch": 0.44706967095026795,
      "grad_norm": 0.7368711829185486,
      "learning_rate": 6.091408362424233e-06,
      "loss": 0.7174,
      "step": 15516
    },
    {
      "epoch": 0.44709848441191724,
      "grad_norm": 0.7602149248123169,
      "learning_rate": 6.090952999472634e-06,
      "loss": 0.7178,
      "step": 15517
    },
    {
      "epoch": 0.44712729787356653,
      "grad_norm": 0.7393301129341125,
      "learning_rate": 6.090497627019926e-06,
      "loss": 0.6983,
      "step": 15518
    },
    {
      "epoch": 0.4471561113352158,
      "grad_norm": 0.7444227337837219,
      "learning_rate": 6.0900422450700725e-06,
      "loss": 0.7337,
      "step": 15519
    },
    {
      "epoch": 0.4471849247968651,
      "grad_norm": 0.7589403986930847,
      "learning_rate": 6.0895868536270404e-06,
      "loss": 0.7127,
      "step": 15520
    },
    {
      "epoch": 0.4472137382585144,
      "grad_norm": 0.7349547147750854,
      "learning_rate": 6.089131452694797e-06,
      "loss": 0.7276,
      "step": 15521
    },
    {
      "epoch": 0.44724255172016364,
      "grad_norm": 0.740061342716217,
      "learning_rate": 6.088676042277306e-06,
      "loss": 0.7449,
      "step": 15522
    },
    {
      "epoch": 0.44727136518181293,
      "grad_norm": 0.7263198494911194,
      "learning_rate": 6.088220622378536e-06,
      "loss": 0.7198,
      "step": 15523
    },
    {
      "epoch": 0.4473001786434622,
      "grad_norm": 0.75332111120224,
      "learning_rate": 6.087765193002452e-06,
      "loss": 0.7069,
      "step": 15524
    },
    {
      "epoch": 0.4473289921051115,
      "grad_norm": 0.742156982421875,
      "learning_rate": 6.087309754153019e-06,
      "loss": 0.7248,
      "step": 15525
    },
    {
      "epoch": 0.4473578055667608,
      "grad_norm": 0.7456270456314087,
      "learning_rate": 6.086854305834207e-06,
      "loss": 0.7627,
      "step": 15526
    },
    {
      "epoch": 0.4473866190284101,
      "grad_norm": 0.7364377379417419,
      "learning_rate": 6.08639884804998e-06,
      "loss": 0.7101,
      "step": 15527
    },
    {
      "epoch": 0.44741543249005933,
      "grad_norm": 0.7437418699264526,
      "learning_rate": 6.085943380804304e-06,
      "loss": 0.748,
      "step": 15528
    },
    {
      "epoch": 0.4474442459517086,
      "grad_norm": 0.7374807596206665,
      "learning_rate": 6.085487904101148e-06,
      "loss": 0.7003,
      "step": 15529
    },
    {
      "epoch": 0.4474730594133579,
      "grad_norm": 0.7174702882766724,
      "learning_rate": 6.085032417944478e-06,
      "loss": 0.7113,
      "step": 15530
    },
    {
      "epoch": 0.4475018728750072,
      "grad_norm": 0.7178665399551392,
      "learning_rate": 6.084576922338259e-06,
      "loss": 0.7273,
      "step": 15531
    },
    {
      "epoch": 0.4475306863366565,
      "grad_norm": 0.7434943318367004,
      "learning_rate": 6.08412141728646e-06,
      "loss": 0.7215,
      "step": 15532
    },
    {
      "epoch": 0.4475594997983058,
      "grad_norm": 0.7258363366127014,
      "learning_rate": 6.0836659027930465e-06,
      "loss": 0.7335,
      "step": 15533
    },
    {
      "epoch": 0.4475883132599551,
      "grad_norm": 0.7907829880714417,
      "learning_rate": 6.083210378861985e-06,
      "loss": 0.7529,
      "step": 15534
    },
    {
      "epoch": 0.4476171267216043,
      "grad_norm": 0.7585773468017578,
      "learning_rate": 6.0827548454972465e-06,
      "loss": 0.7271,
      "step": 15535
    },
    {
      "epoch": 0.4476459401832536,
      "grad_norm": 0.7332708835601807,
      "learning_rate": 6.082299302702795e-06,
      "loss": 0.7178,
      "step": 15536
    },
    {
      "epoch": 0.4476747536449029,
      "grad_norm": 0.7351890802383423,
      "learning_rate": 6.0818437504825975e-06,
      "loss": 0.7032,
      "step": 15537
    },
    {
      "epoch": 0.4477035671065522,
      "grad_norm": 0.7425558567047119,
      "learning_rate": 6.081388188840623e-06,
      "loss": 0.7208,
      "step": 15538
    },
    {
      "epoch": 0.44773238056820147,
      "grad_norm": 0.7430572509765625,
      "learning_rate": 6.08093261778084e-06,
      "loss": 0.7189,
      "step": 15539
    },
    {
      "epoch": 0.44776119402985076,
      "grad_norm": 0.7503519654273987,
      "learning_rate": 6.0804770373072105e-06,
      "loss": 0.7069,
      "step": 15540
    },
    {
      "epoch": 0.44779000749150005,
      "grad_norm": 0.7716618180274963,
      "learning_rate": 6.0800214474237085e-06,
      "loss": 0.7281,
      "step": 15541
    },
    {
      "epoch": 0.4478188209531493,
      "grad_norm": 0.749349057674408,
      "learning_rate": 6.0795658481343004e-06,
      "loss": 0.723,
      "step": 15542
    },
    {
      "epoch": 0.4478476344147986,
      "grad_norm": 0.7683305144309998,
      "learning_rate": 6.079110239442951e-06,
      "loss": 0.715,
      "step": 15543
    },
    {
      "epoch": 0.44787644787644787,
      "grad_norm": 0.8010706305503845,
      "learning_rate": 6.07865462135363e-06,
      "loss": 0.7168,
      "step": 15544
    },
    {
      "epoch": 0.44790526133809716,
      "grad_norm": 0.7740085124969482,
      "learning_rate": 6.078198993870307e-06,
      "loss": 0.7113,
      "step": 15545
    },
    {
      "epoch": 0.44793407479974645,
      "grad_norm": 0.7316869497299194,
      "learning_rate": 6.077743356996947e-06,
      "loss": 0.7156,
      "step": 15546
    },
    {
      "epoch": 0.44796288826139574,
      "grad_norm": 0.7379708290100098,
      "learning_rate": 6.077287710737521e-06,
      "loss": 0.7342,
      "step": 15547
    },
    {
      "epoch": 0.44799170172304503,
      "grad_norm": 0.741089403629303,
      "learning_rate": 6.076832055095995e-06,
      "loss": 0.7099,
      "step": 15548
    },
    {
      "epoch": 0.44802051518469427,
      "grad_norm": 0.7420641183853149,
      "learning_rate": 6.076376390076338e-06,
      "loss": 0.7186,
      "step": 15549
    },
    {
      "epoch": 0.44804932864634356,
      "grad_norm": 0.7116845846176147,
      "learning_rate": 6.075920715682521e-06,
      "loss": 0.7118,
      "step": 15550
    },
    {
      "epoch": 0.44807814210799285,
      "grad_norm": 0.7416692972183228,
      "learning_rate": 6.0754650319185085e-06,
      "loss": 0.7285,
      "step": 15551
    },
    {
      "epoch": 0.44810695556964214,
      "grad_norm": 0.7726588845252991,
      "learning_rate": 6.075009338788269e-06,
      "loss": 0.7295,
      "step": 15552
    },
    {
      "epoch": 0.44813576903129143,
      "grad_norm": 0.7491762042045593,
      "learning_rate": 6.074553636295775e-06,
      "loss": 0.7418,
      "step": 15553
    },
    {
      "epoch": 0.4481645824929407,
      "grad_norm": 0.7242811918258667,
      "learning_rate": 6.074097924444993e-06,
      "loss": 0.7104,
      "step": 15554
    },
    {
      "epoch": 0.44819339595458996,
      "grad_norm": 0.7459506392478943,
      "learning_rate": 6.073642203239891e-06,
      "loss": 0.7244,
      "step": 15555
    },
    {
      "epoch": 0.44822220941623925,
      "grad_norm": 0.7799716591835022,
      "learning_rate": 6.073186472684438e-06,
      "loss": 0.7351,
      "step": 15556
    },
    {
      "epoch": 0.44825102287788854,
      "grad_norm": 0.7218397855758667,
      "learning_rate": 6.0727307327826055e-06,
      "loss": 0.6993,
      "step": 15557
    },
    {
      "epoch": 0.44827983633953783,
      "grad_norm": 0.7172562479972839,
      "learning_rate": 6.072274983538359e-06,
      "loss": 0.7202,
      "step": 15558
    },
    {
      "epoch": 0.4483086498011871,
      "grad_norm": 0.7649515867233276,
      "learning_rate": 6.0718192249556695e-06,
      "loss": 0.7429,
      "step": 15559
    },
    {
      "epoch": 0.4483374632628364,
      "grad_norm": 0.7303295135498047,
      "learning_rate": 6.071363457038506e-06,
      "loss": 0.7152,
      "step": 15560
    },
    {
      "epoch": 0.4483662767244857,
      "grad_norm": 0.7196844816207886,
      "learning_rate": 6.0709076797908384e-06,
      "loss": 0.7401,
      "step": 15561
    },
    {
      "epoch": 0.44839509018613494,
      "grad_norm": 0.734734058380127,
      "learning_rate": 6.070451893216636e-06,
      "loss": 0.7025,
      "step": 15562
    },
    {
      "epoch": 0.4484239036477842,
      "grad_norm": 0.7749587893486023,
      "learning_rate": 6.069996097319866e-06,
      "loss": 0.7349,
      "step": 15563
    },
    {
      "epoch": 0.4484527171094335,
      "grad_norm": 0.7382546663284302,
      "learning_rate": 6.0695402921045e-06,
      "loss": 0.7056,
      "step": 15564
    },
    {
      "epoch": 0.4484815305710828,
      "grad_norm": 0.6970280408859253,
      "learning_rate": 6.069084477574507e-06,
      "loss": 0.719,
      "step": 15565
    },
    {
      "epoch": 0.4485103440327321,
      "grad_norm": 0.7688168883323669,
      "learning_rate": 6.068628653733858e-06,
      "loss": 0.7432,
      "step": 15566
    },
    {
      "epoch": 0.4485391574943814,
      "grad_norm": 0.7487704157829285,
      "learning_rate": 6.068172820586521e-06,
      "loss": 0.7232,
      "step": 15567
    },
    {
      "epoch": 0.4485679709560307,
      "grad_norm": 0.7400180697441101,
      "learning_rate": 6.067716978136466e-06,
      "loss": 0.7196,
      "step": 15568
    },
    {
      "epoch": 0.4485967844176799,
      "grad_norm": 0.7498291730880737,
      "learning_rate": 6.067261126387664e-06,
      "loss": 0.7153,
      "step": 15569
    },
    {
      "epoch": 0.4486255978793292,
      "grad_norm": 0.7721036076545715,
      "learning_rate": 6.066805265344084e-06,
      "loss": 0.7431,
      "step": 15570
    },
    {
      "epoch": 0.4486544113409785,
      "grad_norm": 0.7421286702156067,
      "learning_rate": 6.066349395009699e-06,
      "loss": 0.7167,
      "step": 15571
    },
    {
      "epoch": 0.4486832248026278,
      "grad_norm": 0.7480964660644531,
      "learning_rate": 6.065893515388475e-06,
      "loss": 0.7236,
      "step": 15572
    },
    {
      "epoch": 0.4487120382642771,
      "grad_norm": 0.719143807888031,
      "learning_rate": 6.065437626484383e-06,
      "loss": 0.713,
      "step": 15573
    },
    {
      "epoch": 0.44874085172592637,
      "grad_norm": 0.7390130758285522,
      "learning_rate": 6.064981728301396e-06,
      "loss": 0.7542,
      "step": 15574
    },
    {
      "epoch": 0.44876966518757566,
      "grad_norm": 0.7200917601585388,
      "learning_rate": 6.064525820843482e-06,
      "loss": 0.7223,
      "step": 15575
    },
    {
      "epoch": 0.4487984786492249,
      "grad_norm": 0.7540034651756287,
      "learning_rate": 6.064069904114611e-06,
      "loss": 0.724,
      "step": 15576
    },
    {
      "epoch": 0.4488272921108742,
      "grad_norm": 0.7382283806800842,
      "learning_rate": 6.063613978118756e-06,
      "loss": 0.7335,
      "step": 15577
    },
    {
      "epoch": 0.4488561055725235,
      "grad_norm": 0.7366286516189575,
      "learning_rate": 6.0631580428598865e-06,
      "loss": 0.7329,
      "step": 15578
    },
    {
      "epoch": 0.44888491903417277,
      "grad_norm": 0.7450399994850159,
      "learning_rate": 6.062702098341974e-06,
      "loss": 0.7285,
      "step": 15579
    },
    {
      "epoch": 0.44891373249582206,
      "grad_norm": 0.7540349364280701,
      "learning_rate": 6.0622461445689874e-06,
      "loss": 0.7407,
      "step": 15580
    },
    {
      "epoch": 0.44894254595747135,
      "grad_norm": 2.5308144092559814,
      "learning_rate": 6.061790181544899e-06,
      "loss": 0.7442,
      "step": 15581
    },
    {
      "epoch": 0.4489713594191206,
      "grad_norm": 0.7219765186309814,
      "learning_rate": 6.0613342092736795e-06,
      "loss": 0.7273,
      "step": 15582
    },
    {
      "epoch": 0.4490001728807699,
      "grad_norm": 0.7121964693069458,
      "learning_rate": 6.0608782277592996e-06,
      "loss": 0.6867,
      "step": 15583
    },
    {
      "epoch": 0.44902898634241917,
      "grad_norm": 0.7560808658599854,
      "learning_rate": 6.06042223700573e-06,
      "loss": 0.7253,
      "step": 15584
    },
    {
      "epoch": 0.44905779980406846,
      "grad_norm": 0.7401188015937805,
      "learning_rate": 6.059966237016945e-06,
      "loss": 0.7533,
      "step": 15585
    },
    {
      "epoch": 0.44908661326571775,
      "grad_norm": 0.732262372970581,
      "learning_rate": 6.0595102277969124e-06,
      "loss": 0.7083,
      "step": 15586
    },
    {
      "epoch": 0.44911542672736704,
      "grad_norm": 0.7389309406280518,
      "learning_rate": 6.059054209349606e-06,
      "loss": 0.7364,
      "step": 15587
    },
    {
      "epoch": 0.44914424018901633,
      "grad_norm": 0.7358714938163757,
      "learning_rate": 6.058598181678994e-06,
      "loss": 0.7442,
      "step": 15588
    },
    {
      "epoch": 0.44917305365066557,
      "grad_norm": 0.7483800053596497,
      "learning_rate": 6.058142144789051e-06,
      "loss": 0.7284,
      "step": 15589
    },
    {
      "epoch": 0.44920186711231486,
      "grad_norm": 0.7218446731567383,
      "learning_rate": 6.0576860986837485e-06,
      "loss": 0.7417,
      "step": 15590
    },
    {
      "epoch": 0.44923068057396415,
      "grad_norm": 0.7437112331390381,
      "learning_rate": 6.057230043367056e-06,
      "loss": 0.707,
      "step": 15591
    },
    {
      "epoch": 0.44925949403561344,
      "grad_norm": 0.7414839863777161,
      "learning_rate": 6.056773978842948e-06,
      "loss": 0.7206,
      "step": 15592
    },
    {
      "epoch": 0.44928830749726273,
      "grad_norm": 0.7505842447280884,
      "learning_rate": 6.0563179051153955e-06,
      "loss": 0.7505,
      "step": 15593
    },
    {
      "epoch": 0.449317120958912,
      "grad_norm": 0.7159397006034851,
      "learning_rate": 6.055861822188367e-06,
      "loss": 0.7367,
      "step": 15594
    },
    {
      "epoch": 0.4493459344205613,
      "grad_norm": 0.7394070625305176,
      "learning_rate": 6.05540573006584e-06,
      "loss": 0.7175,
      "step": 15595
    },
    {
      "epoch": 0.44937474788221055,
      "grad_norm": 0.7418857216835022,
      "learning_rate": 6.0549496287517835e-06,
      "loss": 0.7397,
      "step": 15596
    },
    {
      "epoch": 0.44940356134385984,
      "grad_norm": 0.7206522226333618,
      "learning_rate": 6.054493518250171e-06,
      "loss": 0.7269,
      "step": 15597
    },
    {
      "epoch": 0.4494323748055091,
      "grad_norm": 0.7158924341201782,
      "learning_rate": 6.054037398564974e-06,
      "loss": 0.72,
      "step": 15598
    },
    {
      "epoch": 0.4494611882671584,
      "grad_norm": 0.717442512512207,
      "learning_rate": 6.053581269700164e-06,
      "loss": 0.7241,
      "step": 15599
    },
    {
      "epoch": 0.4494900017288077,
      "grad_norm": 0.7302351593971252,
      "learning_rate": 6.053125131659714e-06,
      "loss": 0.7146,
      "step": 15600
    },
    {
      "epoch": 0.449518815190457,
      "grad_norm": 0.7329717874526978,
      "learning_rate": 6.0526689844475974e-06,
      "loss": 0.7207,
      "step": 15601
    },
    {
      "epoch": 0.4495476286521063,
      "grad_norm": 0.7449902892112732,
      "learning_rate": 6.052212828067787e-06,
      "loss": 0.7345,
      "step": 15602
    },
    {
      "epoch": 0.4495764421137555,
      "grad_norm": 0.7339549660682678,
      "learning_rate": 6.051756662524253e-06,
      "loss": 0.7344,
      "step": 15603
    },
    {
      "epoch": 0.4496052555754048,
      "grad_norm": 0.7408739328384399,
      "learning_rate": 6.051300487820971e-06,
      "loss": 0.72,
      "step": 15604
    },
    {
      "epoch": 0.4496340690370541,
      "grad_norm": 0.7195303440093994,
      "learning_rate": 6.050844303961913e-06,
      "loss": 0.725,
      "step": 15605
    },
    {
      "epoch": 0.4496628824987034,
      "grad_norm": 0.7326838970184326,
      "learning_rate": 6.050388110951049e-06,
      "loss": 0.7208,
      "step": 15606
    },
    {
      "epoch": 0.4496916959603527,
      "grad_norm": 0.7377586960792542,
      "learning_rate": 6.049931908792358e-06,
      "loss": 0.7157,
      "step": 15607
    },
    {
      "epoch": 0.449720509422002,
      "grad_norm": 0.7517102956771851,
      "learning_rate": 6.049475697489808e-06,
      "loss": 0.7525,
      "step": 15608
    },
    {
      "epoch": 0.4497493228836512,
      "grad_norm": 0.7509217858314514,
      "learning_rate": 6.049019477047374e-06,
      "loss": 0.7503,
      "step": 15609
    },
    {
      "epoch": 0.4497781363453005,
      "grad_norm": 0.763481855392456,
      "learning_rate": 6.04856324746903e-06,
      "loss": 0.7207,
      "step": 15610
    },
    {
      "epoch": 0.4498069498069498,
      "grad_norm": 0.7882217764854431,
      "learning_rate": 6.048107008758748e-06,
      "loss": 0.7606,
      "step": 15611
    },
    {
      "epoch": 0.4498357632685991,
      "grad_norm": 0.7204503417015076,
      "learning_rate": 6.047650760920499e-06,
      "loss": 0.715,
      "step": 15612
    },
    {
      "epoch": 0.4498645767302484,
      "grad_norm": 0.6965982913970947,
      "learning_rate": 6.047194503958262e-06,
      "loss": 0.7109,
      "step": 15613
    },
    {
      "epoch": 0.44989339019189767,
      "grad_norm": 0.7574352622032166,
      "learning_rate": 6.046738237876007e-06,
      "loss": 0.7202,
      "step": 15614
    },
    {
      "epoch": 0.44992220365354696,
      "grad_norm": 0.7382513880729675,
      "learning_rate": 6.046281962677708e-06,
      "loss": 0.7297,
      "step": 15615
    },
    {
      "epoch": 0.4499510171151962,
      "grad_norm": 0.7082535028457642,
      "learning_rate": 6.0458256783673395e-06,
      "loss": 0.7269,
      "step": 15616
    },
    {
      "epoch": 0.4499798305768455,
      "grad_norm": 0.742998480796814,
      "learning_rate": 6.045369384948875e-06,
      "loss": 0.7162,
      "step": 15617
    },
    {
      "epoch": 0.4500086440384948,
      "grad_norm": 0.7303020358085632,
      "learning_rate": 6.044913082426287e-06,
      "loss": 0.7067,
      "step": 15618
    },
    {
      "epoch": 0.45003745750014407,
      "grad_norm": 0.788821816444397,
      "learning_rate": 6.0444567708035505e-06,
      "loss": 0.7176,
      "step": 15619
    },
    {
      "epoch": 0.45006627096179336,
      "grad_norm": 0.744990348815918,
      "learning_rate": 6.044000450084641e-06,
      "loss": 0.7195,
      "step": 15620
    },
    {
      "epoch": 0.45009508442344265,
      "grad_norm": 0.7397294044494629,
      "learning_rate": 6.043544120273531e-06,
      "loss": 0.7268,
      "step": 15621
    },
    {
      "epoch": 0.45012389788509194,
      "grad_norm": 0.7410881519317627,
      "learning_rate": 6.043087781374193e-06,
      "loss": 0.7189,
      "step": 15622
    },
    {
      "epoch": 0.4501527113467412,
      "grad_norm": 0.7679278254508972,
      "learning_rate": 6.0426314333906045e-06,
      "loss": 0.7265,
      "step": 15623
    },
    {
      "epoch": 0.45018152480839047,
      "grad_norm": 0.7289451360702515,
      "learning_rate": 6.04217507632674e-06,
      "loss": 0.6852,
      "step": 15624
    },
    {
      "epoch": 0.45021033827003976,
      "grad_norm": 0.7111722230911255,
      "learning_rate": 6.04171871018657e-06,
      "loss": 0.7137,
      "step": 15625
    },
    {
      "epoch": 0.45023915173168905,
      "grad_norm": 0.7553772926330566,
      "learning_rate": 6.041262334974072e-06,
      "loss": 0.7094,
      "step": 15626
    },
    {
      "epoch": 0.45026796519333834,
      "grad_norm": 0.7684621214866638,
      "learning_rate": 6.040805950693219e-06,
      "loss": 0.7279,
      "step": 15627
    },
    {
      "epoch": 0.45029677865498763,
      "grad_norm": 0.7333987355232239,
      "learning_rate": 6.040349557347989e-06,
      "loss": 0.7101,
      "step": 15628
    },
    {
      "epoch": 0.4503255921166369,
      "grad_norm": 0.7653083205223083,
      "learning_rate": 6.039893154942354e-06,
      "loss": 0.7321,
      "step": 15629
    },
    {
      "epoch": 0.45035440557828615,
      "grad_norm": 0.7202820777893066,
      "learning_rate": 6.039436743480286e-06,
      "loss": 0.7107,
      "step": 15630
    },
    {
      "epoch": 0.45038321903993545,
      "grad_norm": 0.7457996010780334,
      "learning_rate": 6.038980322965766e-06,
      "loss": 0.7487,
      "step": 15631
    },
    {
      "epoch": 0.45041203250158474,
      "grad_norm": 0.7185271978378296,
      "learning_rate": 6.038523893402766e-06,
      "loss": 0.7264,
      "step": 15632
    },
    {
      "epoch": 0.450440845963234,
      "grad_norm": 0.7416335344314575,
      "learning_rate": 6.03806745479526e-06,
      "loss": 0.7225,
      "step": 15633
    },
    {
      "epoch": 0.4504696594248833,
      "grad_norm": 0.7731525301933289,
      "learning_rate": 6.037611007147224e-06,
      "loss": 0.6932,
      "step": 15634
    },
    {
      "epoch": 0.4504984728865326,
      "grad_norm": 0.7530256509780884,
      "learning_rate": 6.037154550462634e-06,
      "loss": 0.7154,
      "step": 15635
    },
    {
      "epoch": 0.45052728634818184,
      "grad_norm": 0.7287603616714478,
      "learning_rate": 6.036698084745464e-06,
      "loss": 0.7379,
      "step": 15636
    },
    {
      "epoch": 0.45055609980983113,
      "grad_norm": 0.7393726110458374,
      "learning_rate": 6.03624160999969e-06,
      "loss": 0.7322,
      "step": 15637
    },
    {
      "epoch": 0.4505849132714804,
      "grad_norm": 0.7561525702476501,
      "learning_rate": 6.035785126229288e-06,
      "loss": 0.7296,
      "step": 15638
    },
    {
      "epoch": 0.4506137267331297,
      "grad_norm": 0.7325783967971802,
      "learning_rate": 6.035328633438233e-06,
      "loss": 0.7032,
      "step": 15639
    },
    {
      "epoch": 0.450642540194779,
      "grad_norm": 0.7415780425071716,
      "learning_rate": 6.0348721316305e-06,
      "loss": 0.7259,
      "step": 15640
    },
    {
      "epoch": 0.4506713536564283,
      "grad_norm": 0.736998975276947,
      "learning_rate": 6.0344156208100656e-06,
      "loss": 0.7164,
      "step": 15641
    },
    {
      "epoch": 0.4507001671180776,
      "grad_norm": 0.7299202680587769,
      "learning_rate": 6.033959100980905e-06,
      "loss": 0.7326,
      "step": 15642
    },
    {
      "epoch": 0.4507289805797268,
      "grad_norm": 0.7022631764411926,
      "learning_rate": 6.0335025721469944e-06,
      "loss": 0.711,
      "step": 15643
    },
    {
      "epoch": 0.4507577940413761,
      "grad_norm": 0.7511224150657654,
      "learning_rate": 6.033046034312309e-06,
      "loss": 0.7472,
      "step": 15644
    },
    {
      "epoch": 0.4507866075030254,
      "grad_norm": 0.7568235993385315,
      "learning_rate": 6.032589487480825e-06,
      "loss": 0.7705,
      "step": 15645
    },
    {
      "epoch": 0.4508154209646747,
      "grad_norm": 0.7569099068641663,
      "learning_rate": 6.03213293165652e-06,
      "loss": 0.6991,
      "step": 15646
    },
    {
      "epoch": 0.450844234426324,
      "grad_norm": 0.7105903029441833,
      "learning_rate": 6.03167636684337e-06,
      "loss": 0.7236,
      "step": 15647
    },
    {
      "epoch": 0.4508730478879733,
      "grad_norm": 0.7346863746643066,
      "learning_rate": 6.031219793045348e-06,
      "loss": 0.7201,
      "step": 15648
    },
    {
      "epoch": 0.45090186134962257,
      "grad_norm": 0.7363259196281433,
      "learning_rate": 6.030763210266433e-06,
      "loss": 0.728,
      "step": 15649
    },
    {
      "epoch": 0.4509306748112718,
      "grad_norm": 0.942347526550293,
      "learning_rate": 6.0303066185106005e-06,
      "loss": 0.7286,
      "step": 15650
    },
    {
      "epoch": 0.4509594882729211,
      "grad_norm": 0.7457442283630371,
      "learning_rate": 6.029850017781827e-06,
      "loss": 0.7467,
      "step": 15651
    },
    {
      "epoch": 0.4509883017345704,
      "grad_norm": 0.735348641872406,
      "learning_rate": 6.029393408084091e-06,
      "loss": 0.7211,
      "step": 15652
    },
    {
      "epoch": 0.4510171151962197,
      "grad_norm": 0.7314090132713318,
      "learning_rate": 6.028936789421367e-06,
      "loss": 0.7076,
      "step": 15653
    },
    {
      "epoch": 0.45104592865786897,
      "grad_norm": 0.7301682233810425,
      "learning_rate": 6.028480161797631e-06,
      "loss": 0.7097,
      "step": 15654
    },
    {
      "epoch": 0.45107474211951826,
      "grad_norm": 0.7629924416542053,
      "learning_rate": 6.028023525216861e-06,
      "loss": 0.7266,
      "step": 15655
    },
    {
      "epoch": 0.45110355558116755,
      "grad_norm": 0.7696409821510315,
      "learning_rate": 6.027566879683034e-06,
      "loss": 0.7484,
      "step": 15656
    },
    {
      "epoch": 0.4511323690428168,
      "grad_norm": 0.7193523049354553,
      "learning_rate": 6.027110225200128e-06,
      "loss": 0.7217,
      "step": 15657
    },
    {
      "epoch": 0.4511611825044661,
      "grad_norm": 0.7508053779602051,
      "learning_rate": 6.026653561772118e-06,
      "loss": 0.7355,
      "step": 15658
    },
    {
      "epoch": 0.45118999596611536,
      "grad_norm": 0.7834644317626953,
      "learning_rate": 6.026196889402981e-06,
      "loss": 0.7129,
      "step": 15659
    },
    {
      "epoch": 0.45121880942776466,
      "grad_norm": 0.7475160360336304,
      "learning_rate": 6.025740208096695e-06,
      "loss": 0.7311,
      "step": 15660
    },
    {
      "epoch": 0.45124762288941395,
      "grad_norm": 0.732720136642456,
      "learning_rate": 6.025283517857236e-06,
      "loss": 0.7024,
      "step": 15661
    },
    {
      "epoch": 0.45127643635106324,
      "grad_norm": 0.7303975820541382,
      "learning_rate": 6.024826818688585e-06,
      "loss": 0.7294,
      "step": 15662
    },
    {
      "epoch": 0.4513052498127125,
      "grad_norm": 0.7808976173400879,
      "learning_rate": 6.024370110594715e-06,
      "loss": 0.7075,
      "step": 15663
    },
    {
      "epoch": 0.45133406327436176,
      "grad_norm": 0.7377644777297974,
      "learning_rate": 6.023913393579606e-06,
      "loss": 0.7161,
      "step": 15664
    },
    {
      "epoch": 0.45136287673601105,
      "grad_norm": 0.7406570911407471,
      "learning_rate": 6.023456667647235e-06,
      "loss": 0.7467,
      "step": 15665
    },
    {
      "epoch": 0.45139169019766034,
      "grad_norm": 0.7672085165977478,
      "learning_rate": 6.0229999328015786e-06,
      "loss": 0.7599,
      "step": 15666
    },
    {
      "epoch": 0.45142050365930964,
      "grad_norm": 0.6947684288024902,
      "learning_rate": 6.022543189046616e-06,
      "loss": 0.7116,
      "step": 15667
    },
    {
      "epoch": 0.4514493171209589,
      "grad_norm": 0.7345786094665527,
      "learning_rate": 6.022086436386325e-06,
      "loss": 0.717,
      "step": 15668
    },
    {
      "epoch": 0.4514781305826082,
      "grad_norm": 0.7390628457069397,
      "learning_rate": 6.021629674824683e-06,
      "loss": 0.7084,
      "step": 15669
    },
    {
      "epoch": 0.45150694404425745,
      "grad_norm": 0.770431637763977,
      "learning_rate": 6.021172904365667e-06,
      "loss": 0.7345,
      "step": 15670
    },
    {
      "epoch": 0.45153575750590674,
      "grad_norm": 0.750197172164917,
      "learning_rate": 6.020716125013256e-06,
      "loss": 0.7192,
      "step": 15671
    },
    {
      "epoch": 0.45156457096755603,
      "grad_norm": 0.7759861946105957,
      "learning_rate": 6.0202593367714264e-06,
      "loss": 0.7392,
      "step": 15672
    },
    {
      "epoch": 0.4515933844292053,
      "grad_norm": 0.798180878162384,
      "learning_rate": 6.0198025396441605e-06,
      "loss": 0.74,
      "step": 15673
    },
    {
      "epoch": 0.4516221978908546,
      "grad_norm": 0.744706928730011,
      "learning_rate": 6.019345733635433e-06,
      "loss": 0.7247,
      "step": 15674
    },
    {
      "epoch": 0.4516510113525039,
      "grad_norm": 0.7291055917739868,
      "learning_rate": 6.018888918749223e-06,
      "loss": 0.7255,
      "step": 15675
    },
    {
      "epoch": 0.4516798248141532,
      "grad_norm": 0.7531571984291077,
      "learning_rate": 6.018432094989509e-06,
      "loss": 0.7184,
      "step": 15676
    },
    {
      "epoch": 0.45170863827580243,
      "grad_norm": 0.7551710605621338,
      "learning_rate": 6.0179752623602694e-06,
      "loss": 0.7273,
      "step": 15677
    },
    {
      "epoch": 0.4517374517374517,
      "grad_norm": 0.7191562056541443,
      "learning_rate": 6.017518420865482e-06,
      "loss": 0.7152,
      "step": 15678
    },
    {
      "epoch": 0.451766265199101,
      "grad_norm": 0.7320390939712524,
      "learning_rate": 6.017061570509128e-06,
      "loss": 0.696,
      "step": 15679
    },
    {
      "epoch": 0.4517950786607503,
      "grad_norm": 0.8076924085617065,
      "learning_rate": 6.016604711295184e-06,
      "loss": 0.7329,
      "step": 15680
    },
    {
      "epoch": 0.4518238921223996,
      "grad_norm": 0.7281904220581055,
      "learning_rate": 6.016147843227629e-06,
      "loss": 0.7146,
      "step": 15681
    },
    {
      "epoch": 0.4518527055840489,
      "grad_norm": 0.7482620477676392,
      "learning_rate": 6.015690966310441e-06,
      "loss": 0.7473,
      "step": 15682
    },
    {
      "epoch": 0.4518815190456982,
      "grad_norm": 0.7492658495903015,
      "learning_rate": 6.015234080547602e-06,
      "loss": 0.7298,
      "step": 15683
    },
    {
      "epoch": 0.4519103325073474,
      "grad_norm": 0.7683690190315247,
      "learning_rate": 6.014777185943086e-06,
      "loss": 0.7006,
      "step": 15684
    },
    {
      "epoch": 0.4519391459689967,
      "grad_norm": 0.7333506345748901,
      "learning_rate": 6.0143202825008775e-06,
      "loss": 0.7283,
      "step": 15685
    },
    {
      "epoch": 0.451967959430646,
      "grad_norm": 0.7133269309997559,
      "learning_rate": 6.0138633702249525e-06,
      "loss": 0.6918,
      "step": 15686
    },
    {
      "epoch": 0.4519967728922953,
      "grad_norm": 0.7229185104370117,
      "learning_rate": 6.013406449119293e-06,
      "loss": 0.7209,
      "step": 15687
    },
    {
      "epoch": 0.4520255863539446,
      "grad_norm": 0.7547323703765869,
      "learning_rate": 6.012949519187874e-06,
      "loss": 0.7153,
      "step": 15688
    },
    {
      "epoch": 0.45205439981559387,
      "grad_norm": 0.7310751676559448,
      "learning_rate": 6.012492580434679e-06,
      "loss": 0.7208,
      "step": 15689
    },
    {
      "epoch": 0.4520832132772431,
      "grad_norm": 0.7415522336959839,
      "learning_rate": 6.012035632863683e-06,
      "loss": 0.7054,
      "step": 15690
    },
    {
      "epoch": 0.4521120267388924,
      "grad_norm": 0.730228841304779,
      "learning_rate": 6.011578676478871e-06,
      "loss": 0.7023,
      "step": 15691
    },
    {
      "epoch": 0.4521408402005417,
      "grad_norm": 0.7189375162124634,
      "learning_rate": 6.011121711284219e-06,
      "loss": 0.7045,
      "step": 15692
    },
    {
      "epoch": 0.452169653662191,
      "grad_norm": 0.7300407886505127,
      "learning_rate": 6.010664737283708e-06,
      "loss": 0.7188,
      "step": 15693
    },
    {
      "epoch": 0.45219846712384026,
      "grad_norm": 0.7414857149124146,
      "learning_rate": 6.010207754481317e-06,
      "loss": 0.7441,
      "step": 15694
    },
    {
      "epoch": 0.45222728058548955,
      "grad_norm": 0.7147143483161926,
      "learning_rate": 6.009750762881027e-06,
      "loss": 0.694,
      "step": 15695
    },
    {
      "epoch": 0.45225609404713885,
      "grad_norm": 0.7144348621368408,
      "learning_rate": 6.0092937624868155e-06,
      "loss": 0.7404,
      "step": 15696
    },
    {
      "epoch": 0.4522849075087881,
      "grad_norm": 0.7305715680122375,
      "learning_rate": 6.008836753302666e-06,
      "loss": 0.7316,
      "step": 15697
    },
    {
      "epoch": 0.45231372097043737,
      "grad_norm": 1.2767695188522339,
      "learning_rate": 6.008379735332556e-06,
      "loss": 0.7096,
      "step": 15698
    },
    {
      "epoch": 0.45234253443208666,
      "grad_norm": 0.7450024485588074,
      "learning_rate": 6.007922708580467e-06,
      "loss": 0.7126,
      "step": 15699
    },
    {
      "epoch": 0.45237134789373595,
      "grad_norm": 0.7537642121315002,
      "learning_rate": 6.007465673050378e-06,
      "loss": 0.7237,
      "step": 15700
    },
    {
      "epoch": 0.45240016135538524,
      "grad_norm": 0.7293379902839661,
      "learning_rate": 6.0070086287462705e-06,
      "loss": 0.7252,
      "step": 15701
    },
    {
      "epoch": 0.45242897481703453,
      "grad_norm": 0.7239987850189209,
      "learning_rate": 6.0065515756721245e-06,
      "loss": 0.705,
      "step": 15702
    },
    {
      "epoch": 0.4524577882786838,
      "grad_norm": 0.7689626812934875,
      "learning_rate": 6.006094513831919e-06,
      "loss": 0.76,
      "step": 15703
    },
    {
      "epoch": 0.45248660174033306,
      "grad_norm": 0.7292937636375427,
      "learning_rate": 6.005637443229638e-06,
      "loss": 0.7152,
      "step": 15704
    },
    {
      "epoch": 0.45251541520198235,
      "grad_norm": 0.7506252527236938,
      "learning_rate": 6.0051803638692605e-06,
      "loss": 0.7459,
      "step": 15705
    },
    {
      "epoch": 0.45254422866363164,
      "grad_norm": 0.7253707051277161,
      "learning_rate": 6.004723275754766e-06,
      "loss": 0.7199,
      "step": 15706
    },
    {
      "epoch": 0.45257304212528093,
      "grad_norm": 0.7375288605690002,
      "learning_rate": 6.004266178890137e-06,
      "loss": 0.7365,
      "step": 15707
    },
    {
      "epoch": 0.4526018555869302,
      "grad_norm": 0.7459708452224731,
      "learning_rate": 6.003809073279351e-06,
      "loss": 0.7,
      "step": 15708
    },
    {
      "epoch": 0.4526306690485795,
      "grad_norm": 0.7616977691650391,
      "learning_rate": 6.0033519589263936e-06,
      "loss": 0.7535,
      "step": 15709
    },
    {
      "epoch": 0.4526594825102288,
      "grad_norm": 0.7364082336425781,
      "learning_rate": 6.002894835835244e-06,
      "loss": 0.7252,
      "step": 15710
    },
    {
      "epoch": 0.45268829597187804,
      "grad_norm": 0.7719966769218445,
      "learning_rate": 6.002437704009881e-06,
      "loss": 0.7361,
      "step": 15711
    },
    {
      "epoch": 0.45271710943352733,
      "grad_norm": 0.7533991932868958,
      "learning_rate": 6.001980563454289e-06,
      "loss": 0.7471,
      "step": 15712
    },
    {
      "epoch": 0.4527459228951766,
      "grad_norm": 0.7425944805145264,
      "learning_rate": 6.001523414172448e-06,
      "loss": 0.7261,
      "step": 15713
    },
    {
      "epoch": 0.4527747363568259,
      "grad_norm": 0.7598413825035095,
      "learning_rate": 6.001066256168337e-06,
      "loss": 0.7408,
      "step": 15714
    },
    {
      "epoch": 0.4528035498184752,
      "grad_norm": 0.7215869426727295,
      "learning_rate": 6.0006090894459425e-06,
      "loss": 0.727,
      "step": 15715
    },
    {
      "epoch": 0.4528323632801245,
      "grad_norm": 0.7172523140907288,
      "learning_rate": 6.000151914009242e-06,
      "loss": 0.7161,
      "step": 15716
    },
    {
      "epoch": 0.45286117674177373,
      "grad_norm": 0.7320870757102966,
      "learning_rate": 5.999694729862218e-06,
      "loss": 0.7301,
      "step": 15717
    },
    {
      "epoch": 0.452889990203423,
      "grad_norm": 0.757877767086029,
      "learning_rate": 5.9992375370088525e-06,
      "loss": 0.7224,
      "step": 15718
    },
    {
      "epoch": 0.4529188036650723,
      "grad_norm": 0.8299986124038696,
      "learning_rate": 5.998780335453126e-06,
      "loss": 0.7327,
      "step": 15719
    },
    {
      "epoch": 0.4529476171267216,
      "grad_norm": 0.713117778301239,
      "learning_rate": 5.998323125199021e-06,
      "loss": 0.7086,
      "step": 15720
    },
    {
      "epoch": 0.4529764305883709,
      "grad_norm": 0.7246496081352234,
      "learning_rate": 5.997865906250521e-06,
      "loss": 0.7213,
      "step": 15721
    },
    {
      "epoch": 0.4530052440500202,
      "grad_norm": 0.7359387278556824,
      "learning_rate": 5.997408678611606e-06,
      "loss": 0.7216,
      "step": 15722
    },
    {
      "epoch": 0.4530340575116695,
      "grad_norm": 0.7463909387588501,
      "learning_rate": 5.996951442286258e-06,
      "loss": 0.7272,
      "step": 15723
    },
    {
      "epoch": 0.4530628709733187,
      "grad_norm": 0.7446713447570801,
      "learning_rate": 5.9964941972784605e-06,
      "loss": 0.709,
      "step": 15724
    },
    {
      "epoch": 0.453091684434968,
      "grad_norm": 0.7920011281967163,
      "learning_rate": 5.9960369435921935e-06,
      "loss": 0.7384,
      "step": 15725
    },
    {
      "epoch": 0.4531204978966173,
      "grad_norm": 0.8212590217590332,
      "learning_rate": 5.99557968123144e-06,
      "loss": 0.7335,
      "step": 15726
    },
    {
      "epoch": 0.4531493113582666,
      "grad_norm": 0.7786679267883301,
      "learning_rate": 5.9951224102001846e-06,
      "loss": 0.7326,
      "step": 15727
    },
    {
      "epoch": 0.4531781248199159,
      "grad_norm": 0.7361161112785339,
      "learning_rate": 5.994665130502407e-06,
      "loss": 0.7239,
      "step": 15728
    },
    {
      "epoch": 0.45320693828156516,
      "grad_norm": 0.733446478843689,
      "learning_rate": 5.99420784214209e-06,
      "loss": 0.729,
      "step": 15729
    },
    {
      "epoch": 0.45323575174321445,
      "grad_norm": 0.7388532161712646,
      "learning_rate": 5.993750545123217e-06,
      "loss": 0.7243,
      "step": 15730
    },
    {
      "epoch": 0.4532645652048637,
      "grad_norm": 0.7369421720504761,
      "learning_rate": 5.99329323944977e-06,
      "loss": 0.7338,
      "step": 15731
    },
    {
      "epoch": 0.453293378666513,
      "grad_norm": 0.7357411980628967,
      "learning_rate": 5.992835925125732e-06,
      "loss": 0.7245,
      "step": 15732
    },
    {
      "epoch": 0.45332219212816227,
      "grad_norm": 0.7476458549499512,
      "learning_rate": 5.992378602155085e-06,
      "loss": 0.7373,
      "step": 15733
    },
    {
      "epoch": 0.45335100558981156,
      "grad_norm": 0.7592787146568298,
      "learning_rate": 5.991921270541813e-06,
      "loss": 0.7194,
      "step": 15734
    },
    {
      "epoch": 0.45337981905146085,
      "grad_norm": 0.7340019345283508,
      "learning_rate": 5.991463930289898e-06,
      "loss": 0.7004,
      "step": 15735
    },
    {
      "epoch": 0.45340863251311014,
      "grad_norm": 0.7706626057624817,
      "learning_rate": 5.991006581403324e-06,
      "loss": 0.706,
      "step": 15736
    },
    {
      "epoch": 0.45343744597475943,
      "grad_norm": 0.7193279266357422,
      "learning_rate": 5.990549223886072e-06,
      "loss": 0.6986,
      "step": 15737
    },
    {
      "epoch": 0.45346625943640867,
      "grad_norm": 0.7688961625099182,
      "learning_rate": 5.990091857742126e-06,
      "loss": 0.7139,
      "step": 15738
    },
    {
      "epoch": 0.45349507289805796,
      "grad_norm": 0.8026692867279053,
      "learning_rate": 5.989634482975471e-06,
      "loss": 0.749,
      "step": 15739
    },
    {
      "epoch": 0.45352388635970725,
      "grad_norm": 0.7526742219924927,
      "learning_rate": 5.989177099590089e-06,
      "loss": 0.735,
      "step": 15740
    },
    {
      "epoch": 0.45355269982135654,
      "grad_norm": 0.7703985571861267,
      "learning_rate": 5.988719707589961e-06,
      "loss": 0.7229,
      "step": 15741
    },
    {
      "epoch": 0.45358151328300583,
      "grad_norm": 0.7434118390083313,
      "learning_rate": 5.988262306979075e-06,
      "loss": 0.7292,
      "step": 15742
    },
    {
      "epoch": 0.4536103267446551,
      "grad_norm": 0.7411415576934814,
      "learning_rate": 5.987804897761412e-06,
      "loss": 0.7287,
      "step": 15743
    },
    {
      "epoch": 0.45363914020630436,
      "grad_norm": 0.7517611384391785,
      "learning_rate": 5.987347479940955e-06,
      "loss": 0.7449,
      "step": 15744
    },
    {
      "epoch": 0.45366795366795365,
      "grad_norm": 0.7414214015007019,
      "learning_rate": 5.986890053521689e-06,
      "loss": 0.7207,
      "step": 15745
    },
    {
      "epoch": 0.45369676712960294,
      "grad_norm": 0.7552695274353027,
      "learning_rate": 5.986432618507598e-06,
      "loss": 0.715,
      "step": 15746
    },
    {
      "epoch": 0.45372558059125223,
      "grad_norm": 0.7507020831108093,
      "learning_rate": 5.985975174902664e-06,
      "loss": 0.747,
      "step": 15747
    },
    {
      "epoch": 0.4537543940529015,
      "grad_norm": 0.7407980561256409,
      "learning_rate": 5.985517722710872e-06,
      "loss": 0.7405,
      "step": 15748
    },
    {
      "epoch": 0.4537832075145508,
      "grad_norm": 0.7159649133682251,
      "learning_rate": 5.985060261936206e-06,
      "loss": 0.7074,
      "step": 15749
    },
    {
      "epoch": 0.4538120209762001,
      "grad_norm": 0.7841942310333252,
      "learning_rate": 5.984602792582649e-06,
      "loss": 0.7198,
      "step": 15750
    },
    {
      "epoch": 0.45384083443784934,
      "grad_norm": 0.7623614072799683,
      "learning_rate": 5.984145314654186e-06,
      "loss": 0.7361,
      "step": 15751
    },
    {
      "epoch": 0.45386964789949863,
      "grad_norm": 0.7522637248039246,
      "learning_rate": 5.983687828154802e-06,
      "loss": 0.7341,
      "step": 15752
    },
    {
      "epoch": 0.4538984613611479,
      "grad_norm": 0.713721513748169,
      "learning_rate": 5.98323033308848e-06,
      "loss": 0.7322,
      "step": 15753
    },
    {
      "epoch": 0.4539272748227972,
      "grad_norm": 0.7471556663513184,
      "learning_rate": 5.982772829459205e-06,
      "loss": 0.7432,
      "step": 15754
    },
    {
      "epoch": 0.4539560882844465,
      "grad_norm": 0.7196092009544373,
      "learning_rate": 5.9823153172709615e-06,
      "loss": 0.7144,
      "step": 15755
    },
    {
      "epoch": 0.4539849017460958,
      "grad_norm": 0.7427512407302856,
      "learning_rate": 5.981857796527731e-06,
      "loss": 0.7339,
      "step": 15756
    },
    {
      "epoch": 0.4540137152077451,
      "grad_norm": 0.7224912047386169,
      "learning_rate": 5.981400267233503e-06,
      "loss": 0.695,
      "step": 15757
    },
    {
      "epoch": 0.4540425286693943,
      "grad_norm": 0.7509170174598694,
      "learning_rate": 5.9809427293922585e-06,
      "loss": 0.7251,
      "step": 15758
    },
    {
      "epoch": 0.4540713421310436,
      "grad_norm": 0.7717222571372986,
      "learning_rate": 5.980485183007985e-06,
      "loss": 0.7473,
      "step": 15759
    },
    {
      "epoch": 0.4541001555926929,
      "grad_norm": 0.7422093749046326,
      "learning_rate": 5.980027628084664e-06,
      "loss": 0.7301,
      "step": 15760
    },
    {
      "epoch": 0.4541289690543422,
      "grad_norm": 0.7564943432807922,
      "learning_rate": 5.979570064626282e-06,
      "loss": 0.7209,
      "step": 15761
    },
    {
      "epoch": 0.4541577825159915,
      "grad_norm": 0.7443110346794128,
      "learning_rate": 5.979112492636824e-06,
      "loss": 0.7297,
      "step": 15762
    },
    {
      "epoch": 0.4541865959776408,
      "grad_norm": 0.755811870098114,
      "learning_rate": 5.978654912120275e-06,
      "loss": 0.7344,
      "step": 15763
    },
    {
      "epoch": 0.45421540943929006,
      "grad_norm": 0.7628932595252991,
      "learning_rate": 5.978197323080621e-06,
      "loss": 0.7048,
      "step": 15764
    },
    {
      "epoch": 0.4542442229009393,
      "grad_norm": 0.7186676263809204,
      "learning_rate": 5.977739725521846e-06,
      "loss": 0.7194,
      "step": 15765
    },
    {
      "epoch": 0.4542730363625886,
      "grad_norm": 0.7630026936531067,
      "learning_rate": 5.977282119447936e-06,
      "loss": 0.7204,
      "step": 15766
    },
    {
      "epoch": 0.4543018498242379,
      "grad_norm": 0.7738118767738342,
      "learning_rate": 5.976824504862875e-06,
      "loss": 0.7479,
      "step": 15767
    },
    {
      "epoch": 0.45433066328588717,
      "grad_norm": 0.7548545002937317,
      "learning_rate": 5.976366881770648e-06,
      "loss": 0.7184,
      "step": 15768
    },
    {
      "epoch": 0.45435947674753646,
      "grad_norm": 0.7544317245483398,
      "learning_rate": 5.975909250175241e-06,
      "loss": 0.7178,
      "step": 15769
    },
    {
      "epoch": 0.45438829020918575,
      "grad_norm": 0.7342791557312012,
      "learning_rate": 5.975451610080643e-06,
      "loss": 0.7302,
      "step": 15770
    },
    {
      "epoch": 0.454417103670835,
      "grad_norm": 0.7373170852661133,
      "learning_rate": 5.9749939614908335e-06,
      "loss": 0.7354,
      "step": 15771
    },
    {
      "epoch": 0.4544459171324843,
      "grad_norm": 0.7814970016479492,
      "learning_rate": 5.9745363044098036e-06,
      "loss": 0.7108,
      "step": 15772
    },
    {
      "epoch": 0.45447473059413357,
      "grad_norm": 0.7252787947654724,
      "learning_rate": 5.974078638841534e-06,
      "loss": 0.7229,
      "step": 15773
    },
    {
      "epoch": 0.45450354405578286,
      "grad_norm": 0.7577654123306274,
      "learning_rate": 5.973620964790016e-06,
      "loss": 0.7118,
      "step": 15774
    },
    {
      "epoch": 0.45453235751743215,
      "grad_norm": 0.7463682889938354,
      "learning_rate": 5.973163282259231e-06,
      "loss": 0.7269,
      "step": 15775
    },
    {
      "epoch": 0.45456117097908144,
      "grad_norm": 0.7380702495574951,
      "learning_rate": 5.972705591253167e-06,
      "loss": 0.7052,
      "step": 15776
    },
    {
      "epoch": 0.45458998444073073,
      "grad_norm": 0.7279001474380493,
      "learning_rate": 5.97224789177581e-06,
      "loss": 0.7084,
      "step": 15777
    },
    {
      "epoch": 0.45461879790237997,
      "grad_norm": 0.73750901222229,
      "learning_rate": 5.971790183831146e-06,
      "loss": 0.7076,
      "step": 15778
    },
    {
      "epoch": 0.45464761136402926,
      "grad_norm": 0.7575610876083374,
      "learning_rate": 5.971332467423159e-06,
      "loss": 0.7332,
      "step": 15779
    },
    {
      "epoch": 0.45467642482567855,
      "grad_norm": 0.7908751368522644,
      "learning_rate": 5.970874742555839e-06,
      "loss": 0.7444,
      "step": 15780
    },
    {
      "epoch": 0.45470523828732784,
      "grad_norm": 0.7388196587562561,
      "learning_rate": 5.970417009233169e-06,
      "loss": 0.7254,
      "step": 15781
    },
    {
      "epoch": 0.45473405174897713,
      "grad_norm": 0.7404540181159973,
      "learning_rate": 5.969959267459138e-06,
      "loss": 0.7419,
      "step": 15782
    },
    {
      "epoch": 0.4547628652106264,
      "grad_norm": 0.7278289794921875,
      "learning_rate": 5.969501517237732e-06,
      "loss": 0.7314,
      "step": 15783
    },
    {
      "epoch": 0.4547916786722757,
      "grad_norm": 0.8631079196929932,
      "learning_rate": 5.969043758572937e-06,
      "loss": 0.7266,
      "step": 15784
    },
    {
      "epoch": 0.45482049213392495,
      "grad_norm": 0.7478541135787964,
      "learning_rate": 5.9685859914687375e-06,
      "loss": 0.7321,
      "step": 15785
    },
    {
      "epoch": 0.45484930559557424,
      "grad_norm": 0.7568060755729675,
      "learning_rate": 5.968128215929123e-06,
      "loss": 0.71,
      "step": 15786
    },
    {
      "epoch": 0.45487811905722353,
      "grad_norm": 0.7397012710571289,
      "learning_rate": 5.967670431958081e-06,
      "loss": 0.7148,
      "step": 15787
    },
    {
      "epoch": 0.4549069325188728,
      "grad_norm": 0.7356055974960327,
      "learning_rate": 5.9672126395595966e-06,
      "loss": 0.7194,
      "step": 15788
    },
    {
      "epoch": 0.4549357459805221,
      "grad_norm": 0.7317219376564026,
      "learning_rate": 5.9667548387376565e-06,
      "loss": 0.7183,
      "step": 15789
    },
    {
      "epoch": 0.4549645594421714,
      "grad_norm": 0.7838772535324097,
      "learning_rate": 5.966297029496247e-06,
      "loss": 0.6878,
      "step": 15790
    },
    {
      "epoch": 0.4549933729038207,
      "grad_norm": 0.7959295511245728,
      "learning_rate": 5.965839211839357e-06,
      "loss": 0.7397,
      "step": 15791
    },
    {
      "epoch": 0.4550221863654699,
      "grad_norm": 0.7400234341621399,
      "learning_rate": 5.965381385770974e-06,
      "loss": 0.7185,
      "step": 15792
    },
    {
      "epoch": 0.4550509998271192,
      "grad_norm": 0.7427038550376892,
      "learning_rate": 5.964923551295083e-06,
      "loss": 0.7081,
      "step": 15793
    },
    {
      "epoch": 0.4550798132887685,
      "grad_norm": 0.7267093658447266,
      "learning_rate": 5.964465708415673e-06,
      "loss": 0.7287,
      "step": 15794
    },
    {
      "epoch": 0.4551086267504178,
      "grad_norm": 0.7092856764793396,
      "learning_rate": 5.964007857136731e-06,
      "loss": 0.7193,
      "step": 15795
    },
    {
      "epoch": 0.4551374402120671,
      "grad_norm": 0.7450470328330994,
      "learning_rate": 5.963549997462244e-06,
      "loss": 0.7565,
      "step": 15796
    },
    {
      "epoch": 0.4551662536737164,
      "grad_norm": 0.7814900279045105,
      "learning_rate": 5.963092129396198e-06,
      "loss": 0.7197,
      "step": 15797
    },
    {
      "epoch": 0.4551950671353656,
      "grad_norm": 0.7586331367492676,
      "learning_rate": 5.9626342529425855e-06,
      "loss": 0.7114,
      "step": 15798
    },
    {
      "epoch": 0.4552238805970149,
      "grad_norm": 0.7305636405944824,
      "learning_rate": 5.962176368105389e-06,
      "loss": 0.7215,
      "step": 15799
    },
    {
      "epoch": 0.4552526940586642,
      "grad_norm": 0.771137535572052,
      "learning_rate": 5.9617184748885985e-06,
      "loss": 0.7637,
      "step": 15800
    },
    {
      "epoch": 0.4552815075203135,
      "grad_norm": 0.7538104057312012,
      "learning_rate": 5.961260573296201e-06,
      "loss": 0.7285,
      "step": 15801
    },
    {
      "epoch": 0.4553103209819628,
      "grad_norm": 0.7528659701347351,
      "learning_rate": 5.960802663332185e-06,
      "loss": 0.7338,
      "step": 15802
    },
    {
      "epoch": 0.45533913444361207,
      "grad_norm": 0.7609459161758423,
      "learning_rate": 5.960344745000538e-06,
      "loss": 0.7196,
      "step": 15803
    },
    {
      "epoch": 0.45536794790526136,
      "grad_norm": 0.7550565600395203,
      "learning_rate": 5.959886818305248e-06,
      "loss": 0.735,
      "step": 15804
    },
    {
      "epoch": 0.4553967613669106,
      "grad_norm": 0.7509769201278687,
      "learning_rate": 5.959428883250304e-06,
      "loss": 0.7323,
      "step": 15805
    },
    {
      "epoch": 0.4554255748285599,
      "grad_norm": 0.7640590071678162,
      "learning_rate": 5.958970939839695e-06,
      "loss": 0.7221,
      "step": 15806
    },
    {
      "epoch": 0.4554543882902092,
      "grad_norm": 0.718815267086029,
      "learning_rate": 5.958512988077406e-06,
      "loss": 0.7236,
      "step": 15807
    },
    {
      "epoch": 0.45548320175185847,
      "grad_norm": 0.7484660148620605,
      "learning_rate": 5.958055027967428e-06,
      "loss": 0.72,
      "step": 15808
    },
    {
      "epoch": 0.45551201521350776,
      "grad_norm": 0.7299633622169495,
      "learning_rate": 5.957597059513747e-06,
      "loss": 0.7181,
      "step": 15809
    },
    {
      "epoch": 0.45554082867515705,
      "grad_norm": 0.7030128836631775,
      "learning_rate": 5.9571390827203534e-06,
      "loss": 0.7118,
      "step": 15810
    },
    {
      "epoch": 0.45556964213680634,
      "grad_norm": 0.7659476399421692,
      "learning_rate": 5.956681097591236e-06,
      "loss": 0.735,
      "step": 15811
    },
    {
      "epoch": 0.4555984555984556,
      "grad_norm": 0.7344136238098145,
      "learning_rate": 5.956223104130382e-06,
      "loss": 0.7248,
      "step": 15812
    },
    {
      "epoch": 0.45562726906010487,
      "grad_norm": 0.6970354914665222,
      "learning_rate": 5.955765102341781e-06,
      "loss": 0.7303,
      "step": 15813
    },
    {
      "epoch": 0.45565608252175416,
      "grad_norm": 0.7471344470977783,
      "learning_rate": 5.955307092229421e-06,
      "loss": 0.7109,
      "step": 15814
    },
    {
      "epoch": 0.45568489598340345,
      "grad_norm": 0.7446432709693909,
      "learning_rate": 5.954849073797291e-06,
      "loss": 0.7214,
      "step": 15815
    },
    {
      "epoch": 0.45571370944505274,
      "grad_norm": 0.7497261762619019,
      "learning_rate": 5.954391047049381e-06,
      "loss": 0.7276,
      "step": 15816
    },
    {
      "epoch": 0.45574252290670203,
      "grad_norm": 0.7243994474411011,
      "learning_rate": 5.953933011989678e-06,
      "loss": 0.7393,
      "step": 15817
    },
    {
      "epoch": 0.4557713363683513,
      "grad_norm": 0.7481703162193298,
      "learning_rate": 5.953474968622172e-06,
      "loss": 0.7418,
      "step": 15818
    },
    {
      "epoch": 0.45580014983000056,
      "grad_norm": 0.747757613658905,
      "learning_rate": 5.953016916950851e-06,
      "loss": 0.7064,
      "step": 15819
    },
    {
      "epoch": 0.45582896329164985,
      "grad_norm": 0.7423036694526672,
      "learning_rate": 5.952558856979707e-06,
      "loss": 0.7155,
      "step": 15820
    },
    {
      "epoch": 0.45585777675329914,
      "grad_norm": 0.7095452547073364,
      "learning_rate": 5.952100788712728e-06,
      "loss": 0.6909,
      "step": 15821
    },
    {
      "epoch": 0.45588659021494843,
      "grad_norm": 0.7292914986610413,
      "learning_rate": 5.9516427121539015e-06,
      "loss": 0.7317,
      "step": 15822
    },
    {
      "epoch": 0.4559154036765977,
      "grad_norm": 0.8324432969093323,
      "learning_rate": 5.951184627307219e-06,
      "loss": 0.7405,
      "step": 15823
    },
    {
      "epoch": 0.455944217138247,
      "grad_norm": 0.7945167422294617,
      "learning_rate": 5.950726534176669e-06,
      "loss": 0.7537,
      "step": 15824
    },
    {
      "epoch": 0.45597303059989625,
      "grad_norm": 0.707093358039856,
      "learning_rate": 5.9502684327662415e-06,
      "loss": 0.7204,
      "step": 15825
    },
    {
      "epoch": 0.45600184406154554,
      "grad_norm": 0.7635951042175293,
      "learning_rate": 5.949810323079927e-06,
      "loss": 0.7233,
      "step": 15826
    },
    {
      "epoch": 0.4560306575231948,
      "grad_norm": 0.8201340436935425,
      "learning_rate": 5.949352205121712e-06,
      "loss": 0.7193,
      "step": 15827
    },
    {
      "epoch": 0.4560594709848441,
      "grad_norm": 0.75358647108078,
      "learning_rate": 5.94889407889559e-06,
      "loss": 0.7039,
      "step": 15828
    },
    {
      "epoch": 0.4560882844464934,
      "grad_norm": 0.7840701937675476,
      "learning_rate": 5.948435944405548e-06,
      "loss": 0.7733,
      "step": 15829
    },
    {
      "epoch": 0.4561170979081427,
      "grad_norm": 0.7406386137008667,
      "learning_rate": 5.947977801655578e-06,
      "loss": 0.7186,
      "step": 15830
    },
    {
      "epoch": 0.456145911369792,
      "grad_norm": 0.7760697603225708,
      "learning_rate": 5.947519650649668e-06,
      "loss": 0.7328,
      "step": 15831
    },
    {
      "epoch": 0.4561747248314412,
      "grad_norm": 0.7435947060585022,
      "learning_rate": 5.94706149139181e-06,
      "loss": 0.6941,
      "step": 15832
    },
    {
      "epoch": 0.4562035382930905,
      "grad_norm": 0.7552011609077454,
      "learning_rate": 5.946603323885991e-06,
      "loss": 0.7076,
      "step": 15833
    },
    {
      "epoch": 0.4562323517547398,
      "grad_norm": 0.7303442358970642,
      "learning_rate": 5.946145148136205e-06,
      "loss": 0.7309,
      "step": 15834
    },
    {
      "epoch": 0.4562611652163891,
      "grad_norm": 0.7329015135765076,
      "learning_rate": 5.945686964146441e-06,
      "loss": 0.7193,
      "step": 15835
    },
    {
      "epoch": 0.4562899786780384,
      "grad_norm": 0.7316609621047974,
      "learning_rate": 5.945228771920688e-06,
      "loss": 0.7174,
      "step": 15836
    },
    {
      "epoch": 0.4563187921396877,
      "grad_norm": 0.7448671460151672,
      "learning_rate": 5.944770571462938e-06,
      "loss": 0.7141,
      "step": 15837
    },
    {
      "epoch": 0.45634760560133697,
      "grad_norm": 0.7523444294929504,
      "learning_rate": 5.9443123627771796e-06,
      "loss": 0.6991,
      "step": 15838
    },
    {
      "epoch": 0.4563764190629862,
      "grad_norm": 0.7376376986503601,
      "learning_rate": 5.943854145867405e-06,
      "loss": 0.703,
      "step": 15839
    },
    {
      "epoch": 0.4564052325246355,
      "grad_norm": 0.7397398352622986,
      "learning_rate": 5.943395920737605e-06,
      "loss": 0.7199,
      "step": 15840
    },
    {
      "epoch": 0.4564340459862848,
      "grad_norm": 0.7641570568084717,
      "learning_rate": 5.9429376873917675e-06,
      "loss": 0.7398,
      "step": 15841
    },
    {
      "epoch": 0.4564628594479341,
      "grad_norm": 0.735123872756958,
      "learning_rate": 5.942479445833887e-06,
      "loss": 0.7127,
      "step": 15842
    },
    {
      "epoch": 0.45649167290958337,
      "grad_norm": 0.7295388579368591,
      "learning_rate": 5.942021196067953e-06,
      "loss": 0.7229,
      "step": 15843
    },
    {
      "epoch": 0.45652048637123266,
      "grad_norm": 0.7327266335487366,
      "learning_rate": 5.941562938097956e-06,
      "loss": 0.7333,
      "step": 15844
    },
    {
      "epoch": 0.45654929983288195,
      "grad_norm": 0.7497517466545105,
      "learning_rate": 5.941104671927887e-06,
      "loss": 0.7418,
      "step": 15845
    },
    {
      "epoch": 0.4565781132945312,
      "grad_norm": 0.7732550501823425,
      "learning_rate": 5.940646397561735e-06,
      "loss": 0.7056,
      "step": 15846
    },
    {
      "epoch": 0.4566069267561805,
      "grad_norm": 0.7154529690742493,
      "learning_rate": 5.940188115003495e-06,
      "loss": 0.7169,
      "step": 15847
    },
    {
      "epoch": 0.45663574021782977,
      "grad_norm": 0.7415459156036377,
      "learning_rate": 5.939729824257157e-06,
      "loss": 0.7027,
      "step": 15848
    },
    {
      "epoch": 0.45666455367947906,
      "grad_norm": 0.7539465427398682,
      "learning_rate": 5.939271525326711e-06,
      "loss": 0.7137,
      "step": 15849
    },
    {
      "epoch": 0.45669336714112835,
      "grad_norm": 0.7599228024482727,
      "learning_rate": 5.938813218216149e-06,
      "loss": 0.7145,
      "step": 15850
    },
    {
      "epoch": 0.45672218060277764,
      "grad_norm": 0.7250373363494873,
      "learning_rate": 5.938354902929462e-06,
      "loss": 0.7187,
      "step": 15851
    },
    {
      "epoch": 0.4567509940644269,
      "grad_norm": 0.7320865392684937,
      "learning_rate": 5.937896579470641e-06,
      "loss": 0.7332,
      "step": 15852
    },
    {
      "epoch": 0.45677980752607616,
      "grad_norm": 0.7259392142295837,
      "learning_rate": 5.93743824784368e-06,
      "loss": 0.7271,
      "step": 15853
    },
    {
      "epoch": 0.45680862098772546,
      "grad_norm": 0.7279067635536194,
      "learning_rate": 5.936979908052569e-06,
      "loss": 0.7012,
      "step": 15854
    },
    {
      "epoch": 0.45683743444937475,
      "grad_norm": 0.7553988099098206,
      "learning_rate": 5.9365215601012985e-06,
      "loss": 0.7565,
      "step": 15855
    },
    {
      "epoch": 0.45686624791102404,
      "grad_norm": 0.7241990566253662,
      "learning_rate": 5.936063203993863e-06,
      "loss": 0.7343,
      "step": 15856
    },
    {
      "epoch": 0.45689506137267333,
      "grad_norm": 0.7548038363456726,
      "learning_rate": 5.93560483973425e-06,
      "loss": 0.7322,
      "step": 15857
    },
    {
      "epoch": 0.4569238748343226,
      "grad_norm": 0.7442339062690735,
      "learning_rate": 5.935146467326456e-06,
      "loss": 0.7196,
      "step": 15858
    },
    {
      "epoch": 0.45695268829597185,
      "grad_norm": 0.761381983757019,
      "learning_rate": 5.93468808677447e-06,
      "loss": 0.7302,
      "step": 15859
    },
    {
      "epoch": 0.45698150175762114,
      "grad_norm": 0.740362286567688,
      "learning_rate": 5.934229698082286e-06,
      "loss": 0.7114,
      "step": 15860
    },
    {
      "epoch": 0.45701031521927044,
      "grad_norm": 0.7314867377281189,
      "learning_rate": 5.9337713012538945e-06,
      "loss": 0.7097,
      "step": 15861
    },
    {
      "epoch": 0.4570391286809197,
      "grad_norm": 0.7824556827545166,
      "learning_rate": 5.933312896293288e-06,
      "loss": 0.7337,
      "step": 15862
    },
    {
      "epoch": 0.457067942142569,
      "grad_norm": 0.7408084869384766,
      "learning_rate": 5.93285448320446e-06,
      "loss": 0.7194,
      "step": 15863
    },
    {
      "epoch": 0.4570967556042183,
      "grad_norm": 0.7440304160118103,
      "learning_rate": 5.932396061991402e-06,
      "loss": 0.728,
      "step": 15864
    },
    {
      "epoch": 0.4571255690658676,
      "grad_norm": 0.7912642359733582,
      "learning_rate": 5.9319376326581065e-06,
      "loss": 0.7153,
      "step": 15865
    },
    {
      "epoch": 0.45715438252751683,
      "grad_norm": 0.747555673122406,
      "learning_rate": 5.931479195208566e-06,
      "loss": 0.7054,
      "step": 15866
    },
    {
      "epoch": 0.4571831959891661,
      "grad_norm": 0.7724409699440002,
      "learning_rate": 5.9310207496467734e-06,
      "loss": 0.7106,
      "step": 15867
    },
    {
      "epoch": 0.4572120094508154,
      "grad_norm": 0.7283397316932678,
      "learning_rate": 5.930562295976719e-06,
      "loss": 0.7095,
      "step": 15868
    },
    {
      "epoch": 0.4572408229124647,
      "grad_norm": 0.7362477779388428,
      "learning_rate": 5.930103834202397e-06,
      "loss": 0.7241,
      "step": 15869
    },
    {
      "epoch": 0.457269636374114,
      "grad_norm": 0.754420816898346,
      "learning_rate": 5.929645364327803e-06,
      "loss": 0.7145,
      "step": 15870
    },
    {
      "epoch": 0.4572984498357633,
      "grad_norm": 0.7441608309745789,
      "learning_rate": 5.929186886356926e-06,
      "loss": 0.7165,
      "step": 15871
    },
    {
      "epoch": 0.4573272632974126,
      "grad_norm": 0.7041699886322021,
      "learning_rate": 5.92872840029376e-06,
      "loss": 0.7014,
      "step": 15872
    },
    {
      "epoch": 0.4573560767590618,
      "grad_norm": 0.7788078188896179,
      "learning_rate": 5.928269906142297e-06,
      "loss": 0.7419,
      "step": 15873
    },
    {
      "epoch": 0.4573848902207111,
      "grad_norm": 0.7620367407798767,
      "learning_rate": 5.927811403906531e-06,
      "loss": 0.7521,
      "step": 15874
    },
    {
      "epoch": 0.4574137036823604,
      "grad_norm": 0.7222951650619507,
      "learning_rate": 5.927352893590457e-06,
      "loss": 0.691,
      "step": 15875
    },
    {
      "epoch": 0.4574425171440097,
      "grad_norm": 0.7470213770866394,
      "learning_rate": 5.926894375198065e-06,
      "loss": 0.7133,
      "step": 15876
    },
    {
      "epoch": 0.457471330605659,
      "grad_norm": 0.7409847974777222,
      "learning_rate": 5.926435848733349e-06,
      "loss": 0.7119,
      "step": 15877
    },
    {
      "epoch": 0.45750014406730827,
      "grad_norm": 0.7303250432014465,
      "learning_rate": 5.925977314200303e-06,
      "loss": 0.7335,
      "step": 15878
    },
    {
      "epoch": 0.4575289575289575,
      "grad_norm": 0.7653803825378418,
      "learning_rate": 5.925518771602921e-06,
      "loss": 0.7346,
      "step": 15879
    },
    {
      "epoch": 0.4575577709906068,
      "grad_norm": 0.7398756742477417,
      "learning_rate": 5.925060220945196e-06,
      "loss": 0.729,
      "step": 15880
    },
    {
      "epoch": 0.4575865844522561,
      "grad_norm": 0.7418298125267029,
      "learning_rate": 5.92460166223112e-06,
      "loss": 0.71,
      "step": 15881
    },
    {
      "epoch": 0.4576153979139054,
      "grad_norm": 0.7322677373886108,
      "learning_rate": 5.924143095464688e-06,
      "loss": 0.7087,
      "step": 15882
    },
    {
      "epoch": 0.45764421137555467,
      "grad_norm": 0.727523922920227,
      "learning_rate": 5.923684520649895e-06,
      "loss": 0.7182,
      "step": 15883
    },
    {
      "epoch": 0.45767302483720396,
      "grad_norm": 0.7295402884483337,
      "learning_rate": 5.923225937790732e-06,
      "loss": 0.731,
      "step": 15884
    },
    {
      "epoch": 0.45770183829885325,
      "grad_norm": 0.7429417371749878,
      "learning_rate": 5.922767346891194e-06,
      "loss": 0.7127,
      "step": 15885
    },
    {
      "epoch": 0.4577306517605025,
      "grad_norm": 0.8086495995521545,
      "learning_rate": 5.922308747955276e-06,
      "loss": 0.7616,
      "step": 15886
    },
    {
      "epoch": 0.4577594652221518,
      "grad_norm": 0.7571387887001038,
      "learning_rate": 5.921850140986969e-06,
      "loss": 0.7297,
      "step": 15887
    },
    {
      "epoch": 0.45778827868380106,
      "grad_norm": 0.7236172556877136,
      "learning_rate": 5.921391525990269e-06,
      "loss": 0.7203,
      "step": 15888
    },
    {
      "epoch": 0.45781709214545036,
      "grad_norm": 0.7504367232322693,
      "learning_rate": 5.9209329029691705e-06,
      "loss": 0.7308,
      "step": 15889
    },
    {
      "epoch": 0.45784590560709965,
      "grad_norm": 0.7241341471672058,
      "learning_rate": 5.920474271927668e-06,
      "loss": 0.7429,
      "step": 15890
    },
    {
      "epoch": 0.45787471906874894,
      "grad_norm": 0.7033592462539673,
      "learning_rate": 5.920015632869753e-06,
      "loss": 0.7002,
      "step": 15891
    },
    {
      "epoch": 0.4579035325303982,
      "grad_norm": 0.7167545557022095,
      "learning_rate": 5.919556985799423e-06,
      "loss": 0.707,
      "step": 15892
    },
    {
      "epoch": 0.45793234599204746,
      "grad_norm": 0.7204049229621887,
      "learning_rate": 5.919098330720669e-06,
      "loss": 0.7368,
      "step": 15893
    },
    {
      "epoch": 0.45796115945369675,
      "grad_norm": 0.7445470094680786,
      "learning_rate": 5.918639667637488e-06,
      "loss": 0.7437,
      "step": 15894
    },
    {
      "epoch": 0.45798997291534604,
      "grad_norm": 0.740896999835968,
      "learning_rate": 5.918180996553874e-06,
      "loss": 0.7203,
      "step": 15895
    },
    {
      "epoch": 0.45801878637699533,
      "grad_norm": 0.7211903929710388,
      "learning_rate": 5.917722317473822e-06,
      "loss": 0.733,
      "step": 15896
    },
    {
      "epoch": 0.4580475998386446,
      "grad_norm": 0.7228211760520935,
      "learning_rate": 5.917263630401324e-06,
      "loss": 0.7324,
      "step": 15897
    },
    {
      "epoch": 0.4580764133002939,
      "grad_norm": 0.7475596070289612,
      "learning_rate": 5.916804935340379e-06,
      "loss": 0.7165,
      "step": 15898
    },
    {
      "epoch": 0.4581052267619432,
      "grad_norm": 0.7192298769950867,
      "learning_rate": 5.916346232294978e-06,
      "loss": 0.7336,
      "step": 15899
    },
    {
      "epoch": 0.45813404022359244,
      "grad_norm": 0.7118078470230103,
      "learning_rate": 5.915887521269118e-06,
      "loss": 0.7172,
      "step": 15900
    },
    {
      "epoch": 0.45816285368524173,
      "grad_norm": 0.7261578440666199,
      "learning_rate": 5.915428802266794e-06,
      "loss": 0.7397,
      "step": 15901
    },
    {
      "epoch": 0.458191667146891,
      "grad_norm": 0.7237961888313293,
      "learning_rate": 5.9149700752919995e-06,
      "loss": 0.7322,
      "step": 15902
    },
    {
      "epoch": 0.4582204806085403,
      "grad_norm": 0.723392903804779,
      "learning_rate": 5.91451134034873e-06,
      "loss": 0.7162,
      "step": 15903
    },
    {
      "epoch": 0.4582492940701896,
      "grad_norm": 0.7614875435829163,
      "learning_rate": 5.914052597440981e-06,
      "loss": 0.7332,
      "step": 15904
    },
    {
      "epoch": 0.4582781075318389,
      "grad_norm": 0.7093752026557922,
      "learning_rate": 5.913593846572747e-06,
      "loss": 0.7147,
      "step": 15905
    },
    {
      "epoch": 0.45830692099348813,
      "grad_norm": 0.7387365102767944,
      "learning_rate": 5.913135087748025e-06,
      "loss": 0.7101,
      "step": 15906
    },
    {
      "epoch": 0.4583357344551374,
      "grad_norm": 0.7395095825195312,
      "learning_rate": 5.91267632097081e-06,
      "loss": 0.7195,
      "step": 15907
    },
    {
      "epoch": 0.4583645479167867,
      "grad_norm": 0.7169559597969055,
      "learning_rate": 5.912217546245095e-06,
      "loss": 0.7099,
      "step": 15908
    },
    {
      "epoch": 0.458393361378436,
      "grad_norm": 0.7402868866920471,
      "learning_rate": 5.911758763574877e-06,
      "loss": 0.7251,
      "step": 15909
    },
    {
      "epoch": 0.4584221748400853,
      "grad_norm": 0.7546412944793701,
      "learning_rate": 5.911299972964152e-06,
      "loss": 0.7286,
      "step": 15910
    },
    {
      "epoch": 0.4584509883017346,
      "grad_norm": 0.7915189862251282,
      "learning_rate": 5.9108411744169146e-06,
      "loss": 0.7313,
      "step": 15911
    },
    {
      "epoch": 0.4584798017633839,
      "grad_norm": 0.7812353372573853,
      "learning_rate": 5.910382367937162e-06,
      "loss": 0.7148,
      "step": 15912
    },
    {
      "epoch": 0.4585086152250331,
      "grad_norm": 0.7361043691635132,
      "learning_rate": 5.909923553528889e-06,
      "loss": 0.7039,
      "step": 15913
    },
    {
      "epoch": 0.4585374286866824,
      "grad_norm": 0.7422569990158081,
      "learning_rate": 5.909464731196091e-06,
      "loss": 0.7244,
      "step": 15914
    },
    {
      "epoch": 0.4585662421483317,
      "grad_norm": 0.7495401501655579,
      "learning_rate": 5.909005900942765e-06,
      "loss": 0.7075,
      "step": 15915
    },
    {
      "epoch": 0.458595055609981,
      "grad_norm": 0.7710012793540955,
      "learning_rate": 5.908547062772905e-06,
      "loss": 0.7421,
      "step": 15916
    },
    {
      "epoch": 0.4586238690716303,
      "grad_norm": 0.7415044903755188,
      "learning_rate": 5.908088216690509e-06,
      "loss": 0.722,
      "step": 15917
    },
    {
      "epoch": 0.45865268253327957,
      "grad_norm": 0.7248898148536682,
      "learning_rate": 5.9076293626995725e-06,
      "loss": 0.7284,
      "step": 15918
    },
    {
      "epoch": 0.45868149599492886,
      "grad_norm": 0.7424614429473877,
      "learning_rate": 5.90717050080409e-06,
      "loss": 0.7325,
      "step": 15919
    },
    {
      "epoch": 0.4587103094565781,
      "grad_norm": 0.7543795704841614,
      "learning_rate": 5.906711631008062e-06,
      "loss": 0.7285,
      "step": 15920
    },
    {
      "epoch": 0.4587391229182274,
      "grad_norm": 0.764254093170166,
      "learning_rate": 5.90625275331548e-06,
      "loss": 0.7286,
      "step": 15921
    },
    {
      "epoch": 0.4587679363798767,
      "grad_norm": 0.7532302141189575,
      "learning_rate": 5.905793867730344e-06,
      "loss": 0.7313,
      "step": 15922
    },
    {
      "epoch": 0.45879674984152596,
      "grad_norm": 0.750480592250824,
      "learning_rate": 5.905334974256647e-06,
      "loss": 0.7146,
      "step": 15923
    },
    {
      "epoch": 0.45882556330317525,
      "grad_norm": 0.7313959002494812,
      "learning_rate": 5.904876072898389e-06,
      "loss": 0.7169,
      "step": 15924
    },
    {
      "epoch": 0.45885437676482455,
      "grad_norm": 0.7507675290107727,
      "learning_rate": 5.9044171636595645e-06,
      "loss": 0.735,
      "step": 15925
    },
    {
      "epoch": 0.45888319022647384,
      "grad_norm": 0.7929248213768005,
      "learning_rate": 5.903958246544171e-06,
      "loss": 0.7078,
      "step": 15926
    },
    {
      "epoch": 0.45891200368812307,
      "grad_norm": 0.7685974836349487,
      "learning_rate": 5.903499321556204e-06,
      "loss": 0.7334,
      "step": 15927
    },
    {
      "epoch": 0.45894081714977236,
      "grad_norm": 0.7421162128448486,
      "learning_rate": 5.903040388699662e-06,
      "loss": 0.7168,
      "step": 15928
    },
    {
      "epoch": 0.45896963061142165,
      "grad_norm": 0.7180613875389099,
      "learning_rate": 5.902581447978539e-06,
      "loss": 0.6821,
      "step": 15929
    },
    {
      "epoch": 0.45899844407307094,
      "grad_norm": 0.7688829302787781,
      "learning_rate": 5.902122499396836e-06,
      "loss": 0.7219,
      "step": 15930
    },
    {
      "epoch": 0.45902725753472023,
      "grad_norm": 0.7456111311912537,
      "learning_rate": 5.901663542958548e-06,
      "loss": 0.7208,
      "step": 15931
    },
    {
      "epoch": 0.4590560709963695,
      "grad_norm": 0.7749457359313965,
      "learning_rate": 5.901204578667671e-06,
      "loss": 0.7171,
      "step": 15932
    },
    {
      "epoch": 0.45908488445801876,
      "grad_norm": 0.754695475101471,
      "learning_rate": 5.900745606528202e-06,
      "loss": 0.725,
      "step": 15933
    },
    {
      "epoch": 0.45911369791966805,
      "grad_norm": 0.8199664950370789,
      "learning_rate": 5.90028662654414e-06,
      "loss": 0.7436,
      "step": 15934
    },
    {
      "epoch": 0.45914251138131734,
      "grad_norm": 0.7322373986244202,
      "learning_rate": 5.899827638719482e-06,
      "loss": 0.7207,
      "step": 15935
    },
    {
      "epoch": 0.45917132484296663,
      "grad_norm": 0.7631301879882812,
      "learning_rate": 5.899368643058224e-06,
      "loss": 0.7225,
      "step": 15936
    },
    {
      "epoch": 0.4592001383046159,
      "grad_norm": 0.7480021119117737,
      "learning_rate": 5.898909639564366e-06,
      "loss": 0.6845,
      "step": 15937
    },
    {
      "epoch": 0.4592289517662652,
      "grad_norm": 0.7330349683761597,
      "learning_rate": 5.8984506282418995e-06,
      "loss": 0.7131,
      "step": 15938
    },
    {
      "epoch": 0.4592577652279145,
      "grad_norm": 0.8145759701728821,
      "learning_rate": 5.89799160909483e-06,
      "loss": 0.7148,
      "step": 15939
    },
    {
      "epoch": 0.45928657868956374,
      "grad_norm": 0.7634375095367432,
      "learning_rate": 5.89753258212715e-06,
      "loss": 0.6945,
      "step": 15940
    },
    {
      "epoch": 0.45931539215121303,
      "grad_norm": 0.7301532030105591,
      "learning_rate": 5.8970735473428575e-06,
      "loss": 0.7018,
      "step": 15941
    },
    {
      "epoch": 0.4593442056128623,
      "grad_norm": 0.7737174034118652,
      "learning_rate": 5.896614504745952e-06,
      "loss": 0.7238,
      "step": 15942
    },
    {
      "epoch": 0.4593730190745116,
      "grad_norm": 0.7705474495887756,
      "learning_rate": 5.896155454340431e-06,
      "loss": 0.724,
      "step": 15943
    },
    {
      "epoch": 0.4594018325361609,
      "grad_norm": 0.7202634215354919,
      "learning_rate": 5.895696396130291e-06,
      "loss": 0.7054,
      "step": 15944
    },
    {
      "epoch": 0.4594306459978102,
      "grad_norm": 0.7584188580513,
      "learning_rate": 5.895237330119531e-06,
      "loss": 0.7211,
      "step": 15945
    },
    {
      "epoch": 0.4594594594594595,
      "grad_norm": 0.7323319911956787,
      "learning_rate": 5.894778256312149e-06,
      "loss": 0.6983,
      "step": 15946
    },
    {
      "epoch": 0.4594882729211087,
      "grad_norm": 0.7674540877342224,
      "learning_rate": 5.894319174712142e-06,
      "loss": 0.7258,
      "step": 15947
    },
    {
      "epoch": 0.459517086382758,
      "grad_norm": 0.7673949003219604,
      "learning_rate": 5.893860085323509e-06,
      "loss": 0.7188,
      "step": 15948
    },
    {
      "epoch": 0.4595458998444073,
      "grad_norm": 0.7740857601165771,
      "learning_rate": 5.89340098815025e-06,
      "loss": 0.7233,
      "step": 15949
    },
    {
      "epoch": 0.4595747133060566,
      "grad_norm": 0.7365847826004028,
      "learning_rate": 5.89294188319636e-06,
      "loss": 0.7087,
      "step": 15950
    },
    {
      "epoch": 0.4596035267677059,
      "grad_norm": 0.7495239973068237,
      "learning_rate": 5.892482770465839e-06,
      "loss": 0.7298,
      "step": 15951
    },
    {
      "epoch": 0.4596323402293552,
      "grad_norm": 0.7423734068870544,
      "learning_rate": 5.892023649962686e-06,
      "loss": 0.7135,
      "step": 15952
    },
    {
      "epoch": 0.45966115369100446,
      "grad_norm": 0.7897043824195862,
      "learning_rate": 5.891564521690898e-06,
      "loss": 0.7444,
      "step": 15953
    },
    {
      "epoch": 0.4596899671526537,
      "grad_norm": 0.8328381180763245,
      "learning_rate": 5.891105385654474e-06,
      "loss": 0.7217,
      "step": 15954
    },
    {
      "epoch": 0.459718780614303,
      "grad_norm": 0.7654769420623779,
      "learning_rate": 5.890646241857413e-06,
      "loss": 0.7326,
      "step": 15955
    },
    {
      "epoch": 0.4597475940759523,
      "grad_norm": 0.7489615678787231,
      "learning_rate": 5.890187090303714e-06,
      "loss": 0.7118,
      "step": 15956
    },
    {
      "epoch": 0.4597764075376016,
      "grad_norm": 0.7923262119293213,
      "learning_rate": 5.8897279309973745e-06,
      "loss": 0.731,
      "step": 15957
    },
    {
      "epoch": 0.45980522099925086,
      "grad_norm": 0.7410309314727783,
      "learning_rate": 5.889268763942395e-06,
      "loss": 0.7154,
      "step": 15958
    },
    {
      "epoch": 0.45983403446090015,
      "grad_norm": 0.7622255682945251,
      "learning_rate": 5.888809589142772e-06,
      "loss": 0.7235,
      "step": 15959
    },
    {
      "epoch": 0.4598628479225494,
      "grad_norm": 0.7426081895828247,
      "learning_rate": 5.888350406602507e-06,
      "loss": 0.7318,
      "step": 15960
    },
    {
      "epoch": 0.4598916613841987,
      "grad_norm": 0.7888285517692566,
      "learning_rate": 5.887891216325599e-06,
      "loss": 0.7426,
      "step": 15961
    },
    {
      "epoch": 0.45992047484584797,
      "grad_norm": 0.774399995803833,
      "learning_rate": 5.8874320183160446e-06,
      "loss": 0.7364,
      "step": 15962
    },
    {
      "epoch": 0.45994928830749726,
      "grad_norm": 0.740706205368042,
      "learning_rate": 5.886972812577846e-06,
      "loss": 0.6991,
      "step": 15963
    },
    {
      "epoch": 0.45997810176914655,
      "grad_norm": 0.719050943851471,
      "learning_rate": 5.886513599114999e-06,
      "loss": 0.6997,
      "step": 15964
    },
    {
      "epoch": 0.46000691523079584,
      "grad_norm": 0.7498053312301636,
      "learning_rate": 5.886054377931505e-06,
      "loss": 0.7284,
      "step": 15965
    },
    {
      "epoch": 0.46003572869244513,
      "grad_norm": 0.7564039826393127,
      "learning_rate": 5.885595149031365e-06,
      "loss": 0.7397,
      "step": 15966
    },
    {
      "epoch": 0.46006454215409437,
      "grad_norm": 0.798474133014679,
      "learning_rate": 5.8851359124185745e-06,
      "loss": 0.7381,
      "step": 15967
    },
    {
      "epoch": 0.46009335561574366,
      "grad_norm": 0.7513471841812134,
      "learning_rate": 5.8846766680971354e-06,
      "loss": 0.7154,
      "step": 15968
    },
    {
      "epoch": 0.46012216907739295,
      "grad_norm": 0.7311965823173523,
      "learning_rate": 5.8842174160710474e-06,
      "loss": 0.7191,
      "step": 15969
    },
    {
      "epoch": 0.46015098253904224,
      "grad_norm": 0.7437868118286133,
      "learning_rate": 5.88375815634431e-06,
      "loss": 0.7525,
      "step": 15970
    },
    {
      "epoch": 0.46017979600069153,
      "grad_norm": 0.6996088027954102,
      "learning_rate": 5.883298888920921e-06,
      "loss": 0.7039,
      "step": 15971
    },
    {
      "epoch": 0.4602086094623408,
      "grad_norm": 0.7431094646453857,
      "learning_rate": 5.882839613804882e-06,
      "loss": 0.7572,
      "step": 15972
    },
    {
      "epoch": 0.4602374229239901,
      "grad_norm": 0.7564164996147156,
      "learning_rate": 5.8823803310001945e-06,
      "loss": 0.7226,
      "step": 15973
    },
    {
      "epoch": 0.46026623638563935,
      "grad_norm": 0.7328909039497375,
      "learning_rate": 5.881921040510855e-06,
      "loss": 0.7401,
      "step": 15974
    },
    {
      "epoch": 0.46029504984728864,
      "grad_norm": 0.7440444827079773,
      "learning_rate": 5.8814617423408645e-06,
      "loss": 0.7165,
      "step": 15975
    },
    {
      "epoch": 0.46032386330893793,
      "grad_norm": 0.7588438987731934,
      "learning_rate": 5.8810024364942234e-06,
      "loss": 0.7173,
      "step": 15976
    },
    {
      "epoch": 0.4603526767705872,
      "grad_norm": 0.7399049401283264,
      "learning_rate": 5.880543122974932e-06,
      "loss": 0.7603,
      "step": 15977
    },
    {
      "epoch": 0.4603814902322365,
      "grad_norm": 0.730906069278717,
      "learning_rate": 5.880083801786989e-06,
      "loss": 0.7165,
      "step": 15978
    },
    {
      "epoch": 0.4604103036938858,
      "grad_norm": 0.7230467200279236,
      "learning_rate": 5.879624472934398e-06,
      "loss": 0.7045,
      "step": 15979
    },
    {
      "epoch": 0.4604391171555351,
      "grad_norm": 0.7138974666595459,
      "learning_rate": 5.879165136421155e-06,
      "loss": 0.6891,
      "step": 15980
    },
    {
      "epoch": 0.46046793061718433,
      "grad_norm": 0.7106351256370544,
      "learning_rate": 5.878705792251265e-06,
      "loss": 0.7428,
      "step": 15981
    },
    {
      "epoch": 0.4604967440788336,
      "grad_norm": 0.7440205216407776,
      "learning_rate": 5.878246440428725e-06,
      "loss": 0.728,
      "step": 15982
    },
    {
      "epoch": 0.4605255575404829,
      "grad_norm": 0.7520974278450012,
      "learning_rate": 5.877787080957535e-06,
      "loss": 0.7247,
      "step": 15983
    },
    {
      "epoch": 0.4605543710021322,
      "grad_norm": 0.7275938391685486,
      "learning_rate": 5.877327713841697e-06,
      "loss": 0.7188,
      "step": 15984
    },
    {
      "epoch": 0.4605831844637815,
      "grad_norm": 0.7275561690330505,
      "learning_rate": 5.876868339085213e-06,
      "loss": 0.7394,
      "step": 15985
    },
    {
      "epoch": 0.4606119979254308,
      "grad_norm": 0.733361005783081,
      "learning_rate": 5.876408956692084e-06,
      "loss": 0.7075,
      "step": 15986
    },
    {
      "epoch": 0.46064081138708,
      "grad_norm": 0.7409341931343079,
      "learning_rate": 5.875949566666306e-06,
      "loss": 0.7109,
      "step": 15987
    },
    {
      "epoch": 0.4606696248487293,
      "grad_norm": 0.7520536780357361,
      "learning_rate": 5.875490169011884e-06,
      "loss": 0.6928,
      "step": 15988
    },
    {
      "epoch": 0.4606984383103786,
      "grad_norm": 0.7538223266601562,
      "learning_rate": 5.875030763732818e-06,
      "loss": 0.7382,
      "step": 15989
    },
    {
      "epoch": 0.4607272517720279,
      "grad_norm": 0.7587072849273682,
      "learning_rate": 5.874571350833108e-06,
      "loss": 0.7332,
      "step": 15990
    },
    {
      "epoch": 0.4607560652336772,
      "grad_norm": 0.7463625073432922,
      "learning_rate": 5.874111930316756e-06,
      "loss": 0.7014,
      "step": 15991
    },
    {
      "epoch": 0.46078487869532647,
      "grad_norm": 0.7201206088066101,
      "learning_rate": 5.8736525021877635e-06,
      "loss": 0.7241,
      "step": 15992
    },
    {
      "epoch": 0.46081369215697576,
      "grad_norm": 0.7584248185157776,
      "learning_rate": 5.873193066450131e-06,
      "loss": 0.7482,
      "step": 15993
    },
    {
      "epoch": 0.460842505618625,
      "grad_norm": 0.7299190163612366,
      "learning_rate": 5.8727336231078605e-06,
      "loss": 0.7013,
      "step": 15994
    },
    {
      "epoch": 0.4608713190802743,
      "grad_norm": 0.7657397389411926,
      "learning_rate": 5.87227417216495e-06,
      "loss": 0.7308,
      "step": 15995
    },
    {
      "epoch": 0.4609001325419236,
      "grad_norm": 0.7361631393432617,
      "learning_rate": 5.871814713625406e-06,
      "loss": 0.7314,
      "step": 15996
    },
    {
      "epoch": 0.46092894600357287,
      "grad_norm": 0.7404431700706482,
      "learning_rate": 5.871355247493224e-06,
      "loss": 0.7273,
      "step": 15997
    },
    {
      "epoch": 0.46095775946522216,
      "grad_norm": 0.7299777865409851,
      "learning_rate": 5.870895773772412e-06,
      "loss": 0.7536,
      "step": 15998
    },
    {
      "epoch": 0.46098657292687145,
      "grad_norm": 0.7578669786453247,
      "learning_rate": 5.870436292466968e-06,
      "loss": 0.7398,
      "step": 15999
    },
    {
      "epoch": 0.46101538638852074,
      "grad_norm": 0.7530747652053833,
      "learning_rate": 5.869976803580894e-06,
      "loss": 0.7273,
      "step": 16000
    },
    {
      "epoch": 0.46104419985017,
      "grad_norm": 0.7427131533622742,
      "learning_rate": 5.8695173071181896e-06,
      "loss": 0.7131,
      "step": 16001
    },
    {
      "epoch": 0.46107301331181927,
      "grad_norm": 0.7893632054328918,
      "learning_rate": 5.869057803082861e-06,
      "loss": 0.7548,
      "step": 16002
    },
    {
      "epoch": 0.46110182677346856,
      "grad_norm": 0.7056638598442078,
      "learning_rate": 5.868598291478906e-06,
      "loss": 0.7035,
      "step": 16003
    },
    {
      "epoch": 0.46113064023511785,
      "grad_norm": 0.7425496578216553,
      "learning_rate": 5.86813877231033e-06,
      "loss": 0.7287,
      "step": 16004
    },
    {
      "epoch": 0.46115945369676714,
      "grad_norm": 0.7312803864479065,
      "learning_rate": 5.867679245581133e-06,
      "loss": 0.7192,
      "step": 16005
    },
    {
      "epoch": 0.46118826715841643,
      "grad_norm": 0.7254067659378052,
      "learning_rate": 5.867219711295316e-06,
      "loss": 0.7182,
      "step": 16006
    },
    {
      "epoch": 0.46121708062006567,
      "grad_norm": 0.7750325202941895,
      "learning_rate": 5.866760169456881e-06,
      "loss": 0.7424,
      "step": 16007
    },
    {
      "epoch": 0.46124589408171496,
      "grad_norm": 0.7590669393539429,
      "learning_rate": 5.866300620069834e-06,
      "loss": 0.7119,
      "step": 16008
    },
    {
      "epoch": 0.46127470754336425,
      "grad_norm": 0.7336975932121277,
      "learning_rate": 5.865841063138173e-06,
      "loss": 0.7107,
      "step": 16009
    },
    {
      "epoch": 0.46130352100501354,
      "grad_norm": 0.7456032633781433,
      "learning_rate": 5.865381498665903e-06,
      "loss": 0.7362,
      "step": 16010
    },
    {
      "epoch": 0.46133233446666283,
      "grad_norm": 0.7335964441299438,
      "learning_rate": 5.8649219266570245e-06,
      "loss": 0.7096,
      "step": 16011
    },
    {
      "epoch": 0.4613611479283121,
      "grad_norm": 0.752013087272644,
      "learning_rate": 5.8644623471155415e-06,
      "loss": 0.7251,
      "step": 16012
    },
    {
      "epoch": 0.4613899613899614,
      "grad_norm": 0.7411801218986511,
      "learning_rate": 5.864002760045453e-06,
      "loss": 0.7222,
      "step": 16013
    },
    {
      "epoch": 0.46141877485161065,
      "grad_norm": 0.7383880615234375,
      "learning_rate": 5.863543165450767e-06,
      "loss": 0.7108,
      "step": 16014
    },
    {
      "epoch": 0.46144758831325994,
      "grad_norm": 0.7300904393196106,
      "learning_rate": 5.863083563335482e-06,
      "loss": 0.7343,
      "step": 16015
    },
    {
      "epoch": 0.46147640177490923,
      "grad_norm": 0.7164192795753479,
      "learning_rate": 5.862623953703601e-06,
      "loss": 0.6895,
      "step": 16016
    },
    {
      "epoch": 0.4615052152365585,
      "grad_norm": 1.0283410549163818,
      "learning_rate": 5.86216433655913e-06,
      "loss": 0.7308,
      "step": 16017
    },
    {
      "epoch": 0.4615340286982078,
      "grad_norm": 0.7540817260742188,
      "learning_rate": 5.861704711906068e-06,
      "loss": 0.7355,
      "step": 16018
    },
    {
      "epoch": 0.4615628421598571,
      "grad_norm": 0.7657567262649536,
      "learning_rate": 5.8612450797484184e-06,
      "loss": 0.7207,
      "step": 16019
    },
    {
      "epoch": 0.4615916556215064,
      "grad_norm": 0.7403483986854553,
      "learning_rate": 5.8607854400901865e-06,
      "loss": 0.7219,
      "step": 16020
    },
    {
      "epoch": 0.4616204690831556,
      "grad_norm": 0.7459068894386292,
      "learning_rate": 5.860325792935374e-06,
      "loss": 0.7077,
      "step": 16021
    },
    {
      "epoch": 0.4616492825448049,
      "grad_norm": 0.7619481682777405,
      "learning_rate": 5.859866138287984e-06,
      "loss": 0.7261,
      "step": 16022
    },
    {
      "epoch": 0.4616780960064542,
      "grad_norm": 0.7314329147338867,
      "learning_rate": 5.85940647615202e-06,
      "loss": 0.7338,
      "step": 16023
    },
    {
      "epoch": 0.4617069094681035,
      "grad_norm": 0.7379425168037415,
      "learning_rate": 5.858946806531483e-06,
      "loss": 0.7204,
      "step": 16024
    },
    {
      "epoch": 0.4617357229297528,
      "grad_norm": 0.7293897867202759,
      "learning_rate": 5.858487129430378e-06,
      "loss": 0.7013,
      "step": 16025
    },
    {
      "epoch": 0.4617645363914021,
      "grad_norm": 0.7121146321296692,
      "learning_rate": 5.8580274448527094e-06,
      "loss": 0.7131,
      "step": 16026
    },
    {
      "epoch": 0.46179334985305137,
      "grad_norm": 0.7081973552703857,
      "learning_rate": 5.8575677528024785e-06,
      "loss": 0.7131,
      "step": 16027
    },
    {
      "epoch": 0.4618221633147006,
      "grad_norm": 0.7381311655044556,
      "learning_rate": 5.857108053283691e-06,
      "loss": 0.7511,
      "step": 16028
    },
    {
      "epoch": 0.4618509767763499,
      "grad_norm": 0.7421140670776367,
      "learning_rate": 5.856648346300348e-06,
      "loss": 0.7088,
      "step": 16029
    },
    {
      "epoch": 0.4618797902379992,
      "grad_norm": 0.7357193231582642,
      "learning_rate": 5.856188631856455e-06,
      "loss": 0.7176,
      "step": 16030
    },
    {
      "epoch": 0.4619086036996485,
      "grad_norm": 0.7537795305252075,
      "learning_rate": 5.855728909956014e-06,
      "loss": 0.7179,
      "step": 16031
    },
    {
      "epoch": 0.46193741716129777,
      "grad_norm": 0.7228763699531555,
      "learning_rate": 5.855269180603031e-06,
      "loss": 0.7063,
      "step": 16032
    },
    {
      "epoch": 0.46196623062294706,
      "grad_norm": 0.7855859398841858,
      "learning_rate": 5.8548094438015065e-06,
      "loss": 0.7188,
      "step": 16033
    },
    {
      "epoch": 0.4619950440845963,
      "grad_norm": 0.7608886361122131,
      "learning_rate": 5.854349699555449e-06,
      "loss": 0.7396,
      "step": 16034
    },
    {
      "epoch": 0.4620238575462456,
      "grad_norm": 0.7307322025299072,
      "learning_rate": 5.853889947868856e-06,
      "loss": 0.7261,
      "step": 16035
    },
    {
      "epoch": 0.4620526710078949,
      "grad_norm": 0.7593055963516235,
      "learning_rate": 5.8534301887457375e-06,
      "loss": 0.7275,
      "step": 16036
    },
    {
      "epoch": 0.46208148446954417,
      "grad_norm": 0.7120214104652405,
      "learning_rate": 5.852970422190094e-06,
      "loss": 0.7202,
      "step": 16037
    },
    {
      "epoch": 0.46211029793119346,
      "grad_norm": 0.7292835712432861,
      "learning_rate": 5.8525106482059315e-06,
      "loss": 0.7328,
      "step": 16038
    },
    {
      "epoch": 0.46213911139284275,
      "grad_norm": 0.7806122899055481,
      "learning_rate": 5.852050866797254e-06,
      "loss": 0.724,
      "step": 16039
    },
    {
      "epoch": 0.46216792485449204,
      "grad_norm": 0.7986617088317871,
      "learning_rate": 5.851591077968065e-06,
      "loss": 0.7234,
      "step": 16040
    },
    {
      "epoch": 0.4621967383161413,
      "grad_norm": 0.7232659459114075,
      "learning_rate": 5.851131281722368e-06,
      "loss": 0.696,
      "step": 16041
    },
    {
      "epoch": 0.46222555177779057,
      "grad_norm": 0.7361565232276917,
      "learning_rate": 5.850671478064169e-06,
      "loss": 0.7432,
      "step": 16042
    },
    {
      "epoch": 0.46225436523943986,
      "grad_norm": 0.7270938158035278,
      "learning_rate": 5.850211666997472e-06,
      "loss": 0.7382,
      "step": 16043
    },
    {
      "epoch": 0.46228317870108915,
      "grad_norm": 0.7138928771018982,
      "learning_rate": 5.849751848526282e-06,
      "loss": 0.7235,
      "step": 16044
    },
    {
      "epoch": 0.46231199216273844,
      "grad_norm": 0.730358898639679,
      "learning_rate": 5.849292022654602e-06,
      "loss": 0.7322,
      "step": 16045
    },
    {
      "epoch": 0.46234080562438773,
      "grad_norm": 0.7275081872940063,
      "learning_rate": 5.848832189386436e-06,
      "loss": 0.7237,
      "step": 16046
    },
    {
      "epoch": 0.462369619086037,
      "grad_norm": 0.7205197215080261,
      "learning_rate": 5.848372348725793e-06,
      "loss": 0.7116,
      "step": 16047
    },
    {
      "epoch": 0.46239843254768626,
      "grad_norm": 0.7463197112083435,
      "learning_rate": 5.847912500676672e-06,
      "loss": 0.7386,
      "step": 16048
    },
    {
      "epoch": 0.46242724600933555,
      "grad_norm": 0.7314902544021606,
      "learning_rate": 5.847452645243083e-06,
      "loss": 0.718,
      "step": 16049
    },
    {
      "epoch": 0.46245605947098484,
      "grad_norm": 0.7497259974479675,
      "learning_rate": 5.846992782429028e-06,
      "loss": 0.6907,
      "step": 16050
    },
    {
      "epoch": 0.46248487293263413,
      "grad_norm": 0.7295179963111877,
      "learning_rate": 5.8465329122385116e-06,
      "loss": 0.7453,
      "step": 16051
    },
    {
      "epoch": 0.4625136863942834,
      "grad_norm": 0.7425961494445801,
      "learning_rate": 5.846073034675541e-06,
      "loss": 0.7145,
      "step": 16052
    },
    {
      "epoch": 0.4625424998559327,
      "grad_norm": 0.7682022452354431,
      "learning_rate": 5.845613149744121e-06,
      "loss": 0.738,
      "step": 16053
    },
    {
      "epoch": 0.462571313317582,
      "grad_norm": 0.7846354842185974,
      "learning_rate": 5.845153257448253e-06,
      "loss": 0.7392,
      "step": 16054
    },
    {
      "epoch": 0.46260012677923124,
      "grad_norm": 0.7444145083427429,
      "learning_rate": 5.844693357791947e-06,
      "loss": 0.716,
      "step": 16055
    },
    {
      "epoch": 0.4626289402408805,
      "grad_norm": 0.7138367295265198,
      "learning_rate": 5.844233450779203e-06,
      "loss": 0.7174,
      "step": 16056
    },
    {
      "epoch": 0.4626577537025298,
      "grad_norm": 0.7617822289466858,
      "learning_rate": 5.843773536414033e-06,
      "loss": 0.7303,
      "step": 16057
    },
    {
      "epoch": 0.4626865671641791,
      "grad_norm": 0.7308964729309082,
      "learning_rate": 5.843313614700438e-06,
      "loss": 0.6993,
      "step": 16058
    },
    {
      "epoch": 0.4627153806258284,
      "grad_norm": 0.7774739265441895,
      "learning_rate": 5.8428536856424245e-06,
      "loss": 0.7389,
      "step": 16059
    },
    {
      "epoch": 0.4627441940874777,
      "grad_norm": 0.7141070365905762,
      "learning_rate": 5.842393749243997e-06,
      "loss": 0.705,
      "step": 16060
    },
    {
      "epoch": 0.4627730075491269,
      "grad_norm": 0.7429015040397644,
      "learning_rate": 5.841933805509164e-06,
      "loss": 0.7117,
      "step": 16061
    },
    {
      "epoch": 0.4628018210107762,
      "grad_norm": 0.7834364175796509,
      "learning_rate": 5.841473854441927e-06,
      "loss": 0.7534,
      "step": 16062
    },
    {
      "epoch": 0.4628306344724255,
      "grad_norm": 0.744083821773529,
      "learning_rate": 5.8410138960462955e-06,
      "loss": 0.6962,
      "step": 16063
    },
    {
      "epoch": 0.4628594479340748,
      "grad_norm": 0.7352819442749023,
      "learning_rate": 5.840553930326274e-06,
      "loss": 0.7183,
      "step": 16064
    },
    {
      "epoch": 0.4628882613957241,
      "grad_norm": 0.7205616235733032,
      "learning_rate": 5.840093957285866e-06,
      "loss": 0.7273,
      "step": 16065
    },
    {
      "epoch": 0.4629170748573734,
      "grad_norm": 0.701224148273468,
      "learning_rate": 5.83963397692908e-06,
      "loss": 0.7284,
      "step": 16066
    },
    {
      "epoch": 0.46294588831902267,
      "grad_norm": 0.7711036801338196,
      "learning_rate": 5.839173989259923e-06,
      "loss": 0.7434,
      "step": 16067
    },
    {
      "epoch": 0.4629747017806719,
      "grad_norm": 0.7462955713272095,
      "learning_rate": 5.838713994282398e-06,
      "loss": 0.7418,
      "step": 16068
    },
    {
      "epoch": 0.4630035152423212,
      "grad_norm": 0.7203472256660461,
      "learning_rate": 5.838253992000513e-06,
      "loss": 0.7081,
      "step": 16069
    },
    {
      "epoch": 0.4630323287039705,
      "grad_norm": 0.7219346165657043,
      "learning_rate": 5.837793982418274e-06,
      "loss": 0.7213,
      "step": 16070
    },
    {
      "epoch": 0.4630611421656198,
      "grad_norm": 0.7331249117851257,
      "learning_rate": 5.8373339655396865e-06,
      "loss": 0.7248,
      "step": 16071
    },
    {
      "epoch": 0.46308995562726907,
      "grad_norm": 0.7385187149047852,
      "learning_rate": 5.836873941368757e-06,
      "loss": 0.7351,
      "step": 16072
    },
    {
      "epoch": 0.46311876908891836,
      "grad_norm": 0.780951738357544,
      "learning_rate": 5.8364139099094916e-06,
      "loss": 0.7291,
      "step": 16073
    },
    {
      "epoch": 0.46314758255056765,
      "grad_norm": 0.7644454836845398,
      "learning_rate": 5.835953871165898e-06,
      "loss": 0.7687,
      "step": 16074
    },
    {
      "epoch": 0.4631763960122169,
      "grad_norm": 0.7542570233345032,
      "learning_rate": 5.835493825141982e-06,
      "loss": 0.7144,
      "step": 16075
    },
    {
      "epoch": 0.4632052094738662,
      "grad_norm": 0.7617762684822083,
      "learning_rate": 5.835033771841749e-06,
      "loss": 0.7515,
      "step": 16076
    },
    {
      "epoch": 0.46323402293551547,
      "grad_norm": 0.7368104457855225,
      "learning_rate": 5.834573711269207e-06,
      "loss": 0.744,
      "step": 16077
    },
    {
      "epoch": 0.46326283639716476,
      "grad_norm": 0.7388891577720642,
      "learning_rate": 5.834113643428362e-06,
      "loss": 0.724,
      "step": 16078
    },
    {
      "epoch": 0.46329164985881405,
      "grad_norm": 0.7427244186401367,
      "learning_rate": 5.833653568323221e-06,
      "loss": 0.7194,
      "step": 16079
    },
    {
      "epoch": 0.46332046332046334,
      "grad_norm": 0.7471116185188293,
      "learning_rate": 5.833193485957792e-06,
      "loss": 0.7467,
      "step": 16080
    },
    {
      "epoch": 0.46334927678211263,
      "grad_norm": 0.7184661030769348,
      "learning_rate": 5.83273339633608e-06,
      "loss": 0.7187,
      "step": 16081
    },
    {
      "epoch": 0.46337809024376186,
      "grad_norm": 0.7690760493278503,
      "learning_rate": 5.832273299462092e-06,
      "loss": 0.7336,
      "step": 16082
    },
    {
      "epoch": 0.46340690370541116,
      "grad_norm": 0.7267327308654785,
      "learning_rate": 5.831813195339836e-06,
      "loss": 0.7147,
      "step": 16083
    },
    {
      "epoch": 0.46343571716706045,
      "grad_norm": 0.7201931476593018,
      "learning_rate": 5.831353083973318e-06,
      "loss": 0.7214,
      "step": 16084
    },
    {
      "epoch": 0.46346453062870974,
      "grad_norm": 0.7540053725242615,
      "learning_rate": 5.830892965366546e-06,
      "loss": 0.7386,
      "step": 16085
    },
    {
      "epoch": 0.463493344090359,
      "grad_norm": 0.7266811728477478,
      "learning_rate": 5.830432839523528e-06,
      "loss": 0.7232,
      "step": 16086
    },
    {
      "epoch": 0.4635221575520083,
      "grad_norm": 0.7449250221252441,
      "learning_rate": 5.8299727064482684e-06,
      "loss": 0.7502,
      "step": 16087
    },
    {
      "epoch": 0.46355097101365755,
      "grad_norm": 0.7256223559379578,
      "learning_rate": 5.829512566144777e-06,
      "loss": 0.7319,
      "step": 16088
    },
    {
      "epoch": 0.46357978447530684,
      "grad_norm": 0.7382234930992126,
      "learning_rate": 5.829052418617061e-06,
      "loss": 0.7384,
      "step": 16089
    },
    {
      "epoch": 0.46360859793695613,
      "grad_norm": 0.7313773036003113,
      "learning_rate": 5.828592263869125e-06,
      "loss": 0.6864,
      "step": 16090
    },
    {
      "epoch": 0.4636374113986054,
      "grad_norm": 0.7489066123962402,
      "learning_rate": 5.828132101904981e-06,
      "loss": 0.6987,
      "step": 16091
    },
    {
      "epoch": 0.4636662248602547,
      "grad_norm": 0.7510745525360107,
      "learning_rate": 5.8276719327286325e-06,
      "loss": 0.7485,
      "step": 16092
    },
    {
      "epoch": 0.463695038321904,
      "grad_norm": 0.7489298582077026,
      "learning_rate": 5.82721175634409e-06,
      "loss": 0.7273,
      "step": 16093
    },
    {
      "epoch": 0.4637238517835533,
      "grad_norm": 0.7198805212974548,
      "learning_rate": 5.826751572755359e-06,
      "loss": 0.7076,
      "step": 16094
    },
    {
      "epoch": 0.46375266524520253,
      "grad_norm": 0.7108202576637268,
      "learning_rate": 5.826291381966449e-06,
      "loss": 0.7215,
      "step": 16095
    },
    {
      "epoch": 0.4637814787068518,
      "grad_norm": 0.7380794882774353,
      "learning_rate": 5.825831183981368e-06,
      "loss": 0.7119,
      "step": 16096
    },
    {
      "epoch": 0.4638102921685011,
      "grad_norm": 0.7506822347640991,
      "learning_rate": 5.82537097880412e-06,
      "loss": 0.7072,
      "step": 16097
    },
    {
      "epoch": 0.4638391056301504,
      "grad_norm": 0.7804116010665894,
      "learning_rate": 5.824910766438718e-06,
      "loss": 0.7353,
      "step": 16098
    },
    {
      "epoch": 0.4638679190917997,
      "grad_norm": 0.7690550684928894,
      "learning_rate": 5.824450546889168e-06,
      "loss": 0.7072,
      "step": 16099
    },
    {
      "epoch": 0.463896732553449,
      "grad_norm": 0.7802519202232361,
      "learning_rate": 5.823990320159477e-06,
      "loss": 0.7376,
      "step": 16100
    },
    {
      "epoch": 0.4639255460150983,
      "grad_norm": 0.7189123034477234,
      "learning_rate": 5.8235300862536545e-06,
      "loss": 0.6925,
      "step": 16101
    },
    {
      "epoch": 0.4639543594767475,
      "grad_norm": 0.7371626496315002,
      "learning_rate": 5.823069845175708e-06,
      "loss": 0.7254,
      "step": 16102
    },
    {
      "epoch": 0.4639831729383968,
      "grad_norm": 0.7390109896659851,
      "learning_rate": 5.822609596929646e-06,
      "loss": 0.7042,
      "step": 16103
    },
    {
      "epoch": 0.4640119864000461,
      "grad_norm": 0.7521759867668152,
      "learning_rate": 5.822149341519477e-06,
      "loss": 0.7427,
      "step": 16104
    },
    {
      "epoch": 0.4640407998616954,
      "grad_norm": 0.7432475686073303,
      "learning_rate": 5.821689078949209e-06,
      "loss": 0.7173,
      "step": 16105
    },
    {
      "epoch": 0.4640696133233447,
      "grad_norm": 0.7533050179481506,
      "learning_rate": 5.821228809222851e-06,
      "loss": 0.7233,
      "step": 16106
    },
    {
      "epoch": 0.46409842678499397,
      "grad_norm": 0.7641260623931885,
      "learning_rate": 5.8207685323444095e-06,
      "loss": 0.7438,
      "step": 16107
    },
    {
      "epoch": 0.46412724024664326,
      "grad_norm": 0.7693132758140564,
      "learning_rate": 5.820308248317895e-06,
      "loss": 0.7254,
      "step": 16108
    },
    {
      "epoch": 0.4641560537082925,
      "grad_norm": 0.7193330526351929,
      "learning_rate": 5.819847957147317e-06,
      "loss": 0.7368,
      "step": 16109
    },
    {
      "epoch": 0.4641848671699418,
      "grad_norm": 0.7243621945381165,
      "learning_rate": 5.819387658836682e-06,
      "loss": 0.696,
      "step": 16110
    },
    {
      "epoch": 0.4642136806315911,
      "grad_norm": 0.7610070109367371,
      "learning_rate": 5.818927353389999e-06,
      "loss": 0.728,
      "step": 16111
    },
    {
      "epoch": 0.46424249409324037,
      "grad_norm": 0.7211223244667053,
      "learning_rate": 5.818467040811277e-06,
      "loss": 0.7473,
      "step": 16112
    },
    {
      "epoch": 0.46427130755488966,
      "grad_norm": 0.7454432845115662,
      "learning_rate": 5.8180067211045266e-06,
      "loss": 0.7229,
      "step": 16113
    },
    {
      "epoch": 0.46430012101653895,
      "grad_norm": 0.7763615846633911,
      "learning_rate": 5.8175463942737544e-06,
      "loss": 0.7202,
      "step": 16114
    },
    {
      "epoch": 0.4643289344781882,
      "grad_norm": 0.7506204843521118,
      "learning_rate": 5.8170860603229716e-06,
      "loss": 0.7331,
      "step": 16115
    },
    {
      "epoch": 0.4643577479398375,
      "grad_norm": 0.7534321546554565,
      "learning_rate": 5.8166257192561836e-06,
      "loss": 0.736,
      "step": 16116
    },
    {
      "epoch": 0.46438656140148676,
      "grad_norm": 0.7832658290863037,
      "learning_rate": 5.816165371077405e-06,
      "loss": 0.7039,
      "step": 16117
    },
    {
      "epoch": 0.46441537486313605,
      "grad_norm": 0.7143229246139526,
      "learning_rate": 5.81570501579064e-06,
      "loss": 0.714,
      "step": 16118
    },
    {
      "epoch": 0.46444418832478535,
      "grad_norm": 0.6990692615509033,
      "learning_rate": 5.815244653399899e-06,
      "loss": 0.7153,
      "step": 16119
    },
    {
      "epoch": 0.46447300178643464,
      "grad_norm": 0.6983340978622437,
      "learning_rate": 5.814784283909193e-06,
      "loss": 0.7371,
      "step": 16120
    },
    {
      "epoch": 0.4645018152480839,
      "grad_norm": 0.7135276198387146,
      "learning_rate": 5.814323907322531e-06,
      "loss": 0.6993,
      "step": 16121
    },
    {
      "epoch": 0.46453062870973316,
      "grad_norm": 0.7262514233589172,
      "learning_rate": 5.8138635236439215e-06,
      "loss": 0.73,
      "step": 16122
    },
    {
      "epoch": 0.46455944217138245,
      "grad_norm": 0.7479482889175415,
      "learning_rate": 5.8134031328773745e-06,
      "loss": 0.7112,
      "step": 16123
    },
    {
      "epoch": 0.46458825563303174,
      "grad_norm": 0.7312626838684082,
      "learning_rate": 5.812942735026899e-06,
      "loss": 0.6924,
      "step": 16124
    },
    {
      "epoch": 0.46461706909468103,
      "grad_norm": 0.7490389943122864,
      "learning_rate": 5.812482330096505e-06,
      "loss": 0.7337,
      "step": 16125
    },
    {
      "epoch": 0.4646458825563303,
      "grad_norm": 0.7334510684013367,
      "learning_rate": 5.8120219180902e-06,
      "loss": 0.7083,
      "step": 16126
    },
    {
      "epoch": 0.4646746960179796,
      "grad_norm": 0.7068277597427368,
      "learning_rate": 5.811561499011998e-06,
      "loss": 0.7077,
      "step": 16127
    },
    {
      "epoch": 0.4647035094796289,
      "grad_norm": 0.7602952122688293,
      "learning_rate": 5.811101072865908e-06,
      "loss": 0.7046,
      "step": 16128
    },
    {
      "epoch": 0.46473232294127814,
      "grad_norm": 0.7326858043670654,
      "learning_rate": 5.810640639655937e-06,
      "loss": 0.7075,
      "step": 16129
    },
    {
      "epoch": 0.46476113640292743,
      "grad_norm": 0.7647100687026978,
      "learning_rate": 5.810180199386096e-06,
      "loss": 0.7292,
      "step": 16130
    },
    {
      "epoch": 0.4647899498645767,
      "grad_norm": 0.7308558821678162,
      "learning_rate": 5.809719752060395e-06,
      "loss": 0.7134,
      "step": 16131
    },
    {
      "epoch": 0.464818763326226,
      "grad_norm": 0.7229194045066833,
      "learning_rate": 5.8092592976828446e-06,
      "loss": 0.7102,
      "step": 16132
    },
    {
      "epoch": 0.4648475767878753,
      "grad_norm": 0.7187233567237854,
      "learning_rate": 5.8087988362574554e-06,
      "loss": 0.7013,
      "step": 16133
    },
    {
      "epoch": 0.4648763902495246,
      "grad_norm": 0.7691564559936523,
      "learning_rate": 5.808338367788237e-06,
      "loss": 0.6989,
      "step": 16134
    },
    {
      "epoch": 0.4649052037111739,
      "grad_norm": 0.7491254210472107,
      "learning_rate": 5.807877892279197e-06,
      "loss": 0.7382,
      "step": 16135
    },
    {
      "epoch": 0.4649340171728231,
      "grad_norm": 0.7204587459564209,
      "learning_rate": 5.807417409734351e-06,
      "loss": 0.6932,
      "step": 16136
    },
    {
      "epoch": 0.4649628306344724,
      "grad_norm": 0.7649783492088318,
      "learning_rate": 5.806956920157705e-06,
      "loss": 0.7354,
      "step": 16137
    },
    {
      "epoch": 0.4649916440961217,
      "grad_norm": 0.7622607350349426,
      "learning_rate": 5.8064964235532705e-06,
      "loss": 0.7432,
      "step": 16138
    },
    {
      "epoch": 0.465020457557771,
      "grad_norm": 0.7214008569717407,
      "learning_rate": 5.8060359199250594e-06,
      "loss": 0.7372,
      "step": 16139
    },
    {
      "epoch": 0.4650492710194203,
      "grad_norm": 0.7358465790748596,
      "learning_rate": 5.805575409277081e-06,
      "loss": 0.7006,
      "step": 16140
    },
    {
      "epoch": 0.4650780844810696,
      "grad_norm": 0.7281588315963745,
      "learning_rate": 5.805114891613347e-06,
      "loss": 0.7035,
      "step": 16141
    },
    {
      "epoch": 0.4651068979427188,
      "grad_norm": 0.7473697066307068,
      "learning_rate": 5.804654366937867e-06,
      "loss": 0.7039,
      "step": 16142
    },
    {
      "epoch": 0.4651357114043681,
      "grad_norm": 0.7469757199287415,
      "learning_rate": 5.80419383525465e-06,
      "loss": 0.7341,
      "step": 16143
    },
    {
      "epoch": 0.4651645248660174,
      "grad_norm": 0.7235174179077148,
      "learning_rate": 5.80373329656771e-06,
      "loss": 0.7074,
      "step": 16144
    },
    {
      "epoch": 0.4651933383276667,
      "grad_norm": 0.7465478181838989,
      "learning_rate": 5.803272750881055e-06,
      "loss": 0.7497,
      "step": 16145
    },
    {
      "epoch": 0.465222151789316,
      "grad_norm": 0.7404121160507202,
      "learning_rate": 5.8028121981987e-06,
      "loss": 0.7287,
      "step": 16146
    },
    {
      "epoch": 0.46525096525096526,
      "grad_norm": 0.7524109482765198,
      "learning_rate": 5.80235163852465e-06,
      "loss": 0.73,
      "step": 16147
    },
    {
      "epoch": 0.46527977871261456,
      "grad_norm": 0.9486176371574402,
      "learning_rate": 5.801891071862921e-06,
      "loss": 0.7148,
      "step": 16148
    },
    {
      "epoch": 0.4653085921742638,
      "grad_norm": 0.7327105402946472,
      "learning_rate": 5.801430498217524e-06,
      "loss": 0.7285,
      "step": 16149
    },
    {
      "epoch": 0.4653374056359131,
      "grad_norm": 0.7057035565376282,
      "learning_rate": 5.800969917592465e-06,
      "loss": 0.7047,
      "step": 16150
    },
    {
      "epoch": 0.4653662190975624,
      "grad_norm": 0.7301290035247803,
      "learning_rate": 5.800509329991761e-06,
      "loss": 0.7291,
      "step": 16151
    },
    {
      "epoch": 0.46539503255921166,
      "grad_norm": 0.7615072131156921,
      "learning_rate": 5.80004873541942e-06,
      "loss": 0.7356,
      "step": 16152
    },
    {
      "epoch": 0.46542384602086095,
      "grad_norm": 0.7109933495521545,
      "learning_rate": 5.799588133879456e-06,
      "loss": 0.7074,
      "step": 16153
    },
    {
      "epoch": 0.46545265948251024,
      "grad_norm": 0.7512530088424683,
      "learning_rate": 5.799127525375876e-06,
      "loss": 0.7197,
      "step": 16154
    },
    {
      "epoch": 0.46548147294415954,
      "grad_norm": 0.7207416892051697,
      "learning_rate": 5.798666909912696e-06,
      "loss": 0.7125,
      "step": 16155
    },
    {
      "epoch": 0.46551028640580877,
      "grad_norm": 0.742841362953186,
      "learning_rate": 5.798206287493924e-06,
      "loss": 0.7397,
      "step": 16156
    },
    {
      "epoch": 0.46553909986745806,
      "grad_norm": 0.7336885333061218,
      "learning_rate": 5.797745658123574e-06,
      "loss": 0.7552,
      "step": 16157
    },
    {
      "epoch": 0.46556791332910735,
      "grad_norm": 0.7359262108802795,
      "learning_rate": 5.797285021805657e-06,
      "loss": 0.7353,
      "step": 16158
    },
    {
      "epoch": 0.46559672679075664,
      "grad_norm": 0.7193936705589294,
      "learning_rate": 5.796824378544185e-06,
      "loss": 0.7125,
      "step": 16159
    },
    {
      "epoch": 0.46562554025240593,
      "grad_norm": 0.7448914647102356,
      "learning_rate": 5.7963637283431684e-06,
      "loss": 0.7276,
      "step": 16160
    },
    {
      "epoch": 0.4656543537140552,
      "grad_norm": 0.7148468494415283,
      "learning_rate": 5.7959030712066215e-06,
      "loss": 0.7242,
      "step": 16161
    },
    {
      "epoch": 0.4656831671757045,
      "grad_norm": 0.7522292137145996,
      "learning_rate": 5.795442407138551e-06,
      "loss": 0.7225,
      "step": 16162
    },
    {
      "epoch": 0.46571198063735375,
      "grad_norm": 0.7378886342048645,
      "learning_rate": 5.794981736142975e-06,
      "loss": 0.7345,
      "step": 16163
    },
    {
      "epoch": 0.46574079409900304,
      "grad_norm": 0.7464713454246521,
      "learning_rate": 5.794521058223902e-06,
      "loss": 0.7396,
      "step": 16164
    },
    {
      "epoch": 0.46576960756065233,
      "grad_norm": 0.7193312644958496,
      "learning_rate": 5.794060373385345e-06,
      "loss": 0.721,
      "step": 16165
    },
    {
      "epoch": 0.4657984210223016,
      "grad_norm": 0.7491374611854553,
      "learning_rate": 5.793599681631316e-06,
      "loss": 0.711,
      "step": 16166
    },
    {
      "epoch": 0.4658272344839509,
      "grad_norm": 0.7232248783111572,
      "learning_rate": 5.793138982965827e-06,
      "loss": 0.7034,
      "step": 16167
    },
    {
      "epoch": 0.4658560479456002,
      "grad_norm": 0.7407903075218201,
      "learning_rate": 5.792678277392889e-06,
      "loss": 0.7148,
      "step": 16168
    },
    {
      "epoch": 0.46588486140724944,
      "grad_norm": 0.7450896501541138,
      "learning_rate": 5.792217564916517e-06,
      "loss": 0.742,
      "step": 16169
    },
    {
      "epoch": 0.46591367486889873,
      "grad_norm": 0.7877501845359802,
      "learning_rate": 5.7917568455407215e-06,
      "loss": 0.7288,
      "step": 16170
    },
    {
      "epoch": 0.465942488330548,
      "grad_norm": 0.7094285488128662,
      "learning_rate": 5.791296119269516e-06,
      "loss": 0.7302,
      "step": 16171
    },
    {
      "epoch": 0.4659713017921973,
      "grad_norm": 0.7582881450653076,
      "learning_rate": 5.7908353861069124e-06,
      "loss": 0.7233,
      "step": 16172
    },
    {
      "epoch": 0.4660001152538466,
      "grad_norm": 0.7874588370323181,
      "learning_rate": 5.790374646056921e-06,
      "loss": 0.6961,
      "step": 16173
    },
    {
      "epoch": 0.4660289287154959,
      "grad_norm": 0.7302834391593933,
      "learning_rate": 5.789913899123558e-06,
      "loss": 0.736,
      "step": 16174
    },
    {
      "epoch": 0.4660577421771452,
      "grad_norm": 0.7181821465492249,
      "learning_rate": 5.789453145310834e-06,
      "loss": 0.7122,
      "step": 16175
    },
    {
      "epoch": 0.4660865556387944,
      "grad_norm": 0.7502919435501099,
      "learning_rate": 5.7889923846227635e-06,
      "loss": 0.7193,
      "step": 16176
    },
    {
      "epoch": 0.4661153691004437,
      "grad_norm": 0.7643392086029053,
      "learning_rate": 5.788531617063358e-06,
      "loss": 0.72,
      "step": 16177
    },
    {
      "epoch": 0.466144182562093,
      "grad_norm": 0.7566782832145691,
      "learning_rate": 5.7880708426366296e-06,
      "loss": 0.7171,
      "step": 16178
    },
    {
      "epoch": 0.4661729960237423,
      "grad_norm": 0.7323161363601685,
      "learning_rate": 5.787610061346593e-06,
      "loss": 0.6975,
      "step": 16179
    },
    {
      "epoch": 0.4662018094853916,
      "grad_norm": 0.7250576019287109,
      "learning_rate": 5.787149273197258e-06,
      "loss": 0.7187,
      "step": 16180
    },
    {
      "epoch": 0.4662306229470409,
      "grad_norm": 0.7341970801353455,
      "learning_rate": 5.786688478192641e-06,
      "loss": 0.6897,
      "step": 16181
    },
    {
      "epoch": 0.46625943640869016,
      "grad_norm": 0.7587904334068298,
      "learning_rate": 5.786227676336753e-06,
      "loss": 0.7334,
      "step": 16182
    },
    {
      "epoch": 0.4662882498703394,
      "grad_norm": 0.7497917413711548,
      "learning_rate": 5.7857668676336085e-06,
      "loss": 0.7122,
      "step": 16183
    },
    {
      "epoch": 0.4663170633319887,
      "grad_norm": 0.7486211061477661,
      "learning_rate": 5.7853060520872205e-06,
      "loss": 0.7367,
      "step": 16184
    },
    {
      "epoch": 0.466345876793638,
      "grad_norm": 0.7469159364700317,
      "learning_rate": 5.784845229701602e-06,
      "loss": 0.7288,
      "step": 16185
    },
    {
      "epoch": 0.46637469025528727,
      "grad_norm": 0.8052311539649963,
      "learning_rate": 5.784384400480765e-06,
      "loss": 0.7221,
      "step": 16186
    },
    {
      "epoch": 0.46640350371693656,
      "grad_norm": 0.7507591843605042,
      "learning_rate": 5.783923564428724e-06,
      "loss": 0.7037,
      "step": 16187
    },
    {
      "epoch": 0.46643231717858585,
      "grad_norm": 0.7353125810623169,
      "learning_rate": 5.7834627215494935e-06,
      "loss": 0.7027,
      "step": 16188
    },
    {
      "epoch": 0.46646113064023514,
      "grad_norm": 0.7692762017250061,
      "learning_rate": 5.783001871847086e-06,
      "loss": 0.7267,
      "step": 16189
    },
    {
      "epoch": 0.4664899441018844,
      "grad_norm": 0.7235479354858398,
      "learning_rate": 5.782541015325513e-06,
      "loss": 0.6996,
      "step": 16190
    },
    {
      "epoch": 0.46651875756353367,
      "grad_norm": 0.7561032176017761,
      "learning_rate": 5.782080151988791e-06,
      "loss": 0.7313,
      "step": 16191
    },
    {
      "epoch": 0.46654757102518296,
      "grad_norm": 0.7328010201454163,
      "learning_rate": 5.781619281840933e-06,
      "loss": 0.7374,
      "step": 16192
    },
    {
      "epoch": 0.46657638448683225,
      "grad_norm": 0.7492585778236389,
      "learning_rate": 5.7811584048859525e-06,
      "loss": 0.7244,
      "step": 16193
    },
    {
      "epoch": 0.46660519794848154,
      "grad_norm": 0.7817090153694153,
      "learning_rate": 5.780697521127862e-06,
      "loss": 0.7402,
      "step": 16194
    },
    {
      "epoch": 0.46663401141013083,
      "grad_norm": 0.74864661693573,
      "learning_rate": 5.780236630570678e-06,
      "loss": 0.7312,
      "step": 16195
    },
    {
      "epoch": 0.46666282487178007,
      "grad_norm": 0.7399400472640991,
      "learning_rate": 5.779775733218413e-06,
      "loss": 0.6857,
      "step": 16196
    },
    {
      "epoch": 0.46669163833342936,
      "grad_norm": 0.9871280789375305,
      "learning_rate": 5.77931482907508e-06,
      "loss": 0.7158,
      "step": 16197
    },
    {
      "epoch": 0.46672045179507865,
      "grad_norm": 0.7475214600563049,
      "learning_rate": 5.7788539181446925e-06,
      "loss": 0.7072,
      "step": 16198
    },
    {
      "epoch": 0.46674926525672794,
      "grad_norm": 0.762522280216217,
      "learning_rate": 5.778393000431268e-06,
      "loss": 0.7339,
      "step": 16199
    },
    {
      "epoch": 0.46677807871837723,
      "grad_norm": 0.7271504402160645,
      "learning_rate": 5.777932075938818e-06,
      "loss": 0.6956,
      "step": 16200
    },
    {
      "epoch": 0.4668068921800265,
      "grad_norm": 0.7092277407646179,
      "learning_rate": 5.777471144671358e-06,
      "loss": 0.7099,
      "step": 16201
    },
    {
      "epoch": 0.4668357056416758,
      "grad_norm": 0.7485891580581665,
      "learning_rate": 5.7770102066329e-06,
      "loss": 0.7548,
      "step": 16202
    },
    {
      "epoch": 0.46686451910332505,
      "grad_norm": 0.719318687915802,
      "learning_rate": 5.776549261827462e-06,
      "loss": 0.7136,
      "step": 16203
    },
    {
      "epoch": 0.46689333256497434,
      "grad_norm": 0.7695910334587097,
      "learning_rate": 5.776088310259054e-06,
      "loss": 0.7154,
      "step": 16204
    },
    {
      "epoch": 0.46692214602662363,
      "grad_norm": 0.7443058490753174,
      "learning_rate": 5.775627351931693e-06,
      "loss": 0.7252,
      "step": 16205
    },
    {
      "epoch": 0.4669509594882729,
      "grad_norm": 0.7486060857772827,
      "learning_rate": 5.775166386849394e-06,
      "loss": 0.7215,
      "step": 16206
    },
    {
      "epoch": 0.4669797729499222,
      "grad_norm": 0.757416307926178,
      "learning_rate": 5.7747054150161705e-06,
      "loss": 0.7124,
      "step": 16207
    },
    {
      "epoch": 0.4670085864115715,
      "grad_norm": 0.7445022463798523,
      "learning_rate": 5.774244436436037e-06,
      "loss": 0.7193,
      "step": 16208
    },
    {
      "epoch": 0.4670373998732208,
      "grad_norm": 0.7131463885307312,
      "learning_rate": 5.7737834511130085e-06,
      "loss": 0.7006,
      "step": 16209
    },
    {
      "epoch": 0.46706621333487003,
      "grad_norm": 0.7515501976013184,
      "learning_rate": 5.773322459051099e-06,
      "loss": 0.7224,
      "step": 16210
    },
    {
      "epoch": 0.4670950267965193,
      "grad_norm": 0.7537771463394165,
      "learning_rate": 5.772861460254324e-06,
      "loss": 0.7342,
      "step": 16211
    },
    {
      "epoch": 0.4671238402581686,
      "grad_norm": 0.7770379781723022,
      "learning_rate": 5.7724004547267e-06,
      "loss": 0.7593,
      "step": 16212
    },
    {
      "epoch": 0.4671526537198179,
      "grad_norm": 0.745615541934967,
      "learning_rate": 5.771939442472237e-06,
      "loss": 0.7196,
      "step": 16213
    },
    {
      "epoch": 0.4671814671814672,
      "grad_norm": 0.736215353012085,
      "learning_rate": 5.771478423494956e-06,
      "loss": 0.7175,
      "step": 16214
    },
    {
      "epoch": 0.4672102806431165,
      "grad_norm": 0.7285718321800232,
      "learning_rate": 5.771017397798868e-06,
      "loss": 0.7011,
      "step": 16215
    },
    {
      "epoch": 0.4672390941047658,
      "grad_norm": 0.7277564406394958,
      "learning_rate": 5.770556365387988e-06,
      "loss": 0.7082,
      "step": 16216
    },
    {
      "epoch": 0.467267907566415,
      "grad_norm": 0.7473687529563904,
      "learning_rate": 5.770095326266334e-06,
      "loss": 0.7148,
      "step": 16217
    },
    {
      "epoch": 0.4672967210280643,
      "grad_norm": 0.7547780871391296,
      "learning_rate": 5.769634280437919e-06,
      "loss": 0.7271,
      "step": 16218
    },
    {
      "epoch": 0.4673255344897136,
      "grad_norm": 0.7511746287345886,
      "learning_rate": 5.769173227906759e-06,
      "loss": 0.7309,
      "step": 16219
    },
    {
      "epoch": 0.4673543479513629,
      "grad_norm": 0.742500901222229,
      "learning_rate": 5.7687121686768696e-06,
      "loss": 0.7174,
      "step": 16220
    },
    {
      "epoch": 0.46738316141301217,
      "grad_norm": 0.7585791945457458,
      "learning_rate": 5.768251102752266e-06,
      "loss": 0.7344,
      "step": 16221
    },
    {
      "epoch": 0.46741197487466146,
      "grad_norm": 0.7483818531036377,
      "learning_rate": 5.7677900301369615e-06,
      "loss": 0.7439,
      "step": 16222
    },
    {
      "epoch": 0.4674407883363107,
      "grad_norm": 0.7237034440040588,
      "learning_rate": 5.767328950834974e-06,
      "loss": 0.719,
      "step": 16223
    },
    {
      "epoch": 0.46746960179796,
      "grad_norm": 0.7272621989250183,
      "learning_rate": 5.766867864850319e-06,
      "loss": 0.7339,
      "step": 16224
    },
    {
      "epoch": 0.4674984152596093,
      "grad_norm": 0.7185215353965759,
      "learning_rate": 5.766406772187011e-06,
      "loss": 0.7163,
      "step": 16225
    },
    {
      "epoch": 0.46752722872125857,
      "grad_norm": 0.7374772429466248,
      "learning_rate": 5.765945672849066e-06,
      "loss": 0.7167,
      "step": 16226
    },
    {
      "epoch": 0.46755604218290786,
      "grad_norm": 0.7169345617294312,
      "learning_rate": 5.7654845668405e-06,
      "loss": 0.6928,
      "step": 16227
    },
    {
      "epoch": 0.46758485564455715,
      "grad_norm": 0.742332935333252,
      "learning_rate": 5.7650234541653286e-06,
      "loss": 0.7325,
      "step": 16228
    },
    {
      "epoch": 0.46761366910620644,
      "grad_norm": 0.7184644937515259,
      "learning_rate": 5.764562334827568e-06,
      "loss": 0.7259,
      "step": 16229
    },
    {
      "epoch": 0.4676424825678557,
      "grad_norm": 0.7899274826049805,
      "learning_rate": 5.764101208831233e-06,
      "loss": 0.7332,
      "step": 16230
    },
    {
      "epoch": 0.46767129602950497,
      "grad_norm": 0.7391191124916077,
      "learning_rate": 5.763640076180341e-06,
      "loss": 0.7155,
      "step": 16231
    },
    {
      "epoch": 0.46770010949115426,
      "grad_norm": 1.60757315158844,
      "learning_rate": 5.763178936878906e-06,
      "loss": 0.8063,
      "step": 16232
    },
    {
      "epoch": 0.46772892295280355,
      "grad_norm": 0.7248502969741821,
      "learning_rate": 5.762717790930947e-06,
      "loss": 0.7299,
      "step": 16233
    },
    {
      "epoch": 0.46775773641445284,
      "grad_norm": 0.7684924602508545,
      "learning_rate": 5.762256638340478e-06,
      "loss": 0.7241,
      "step": 16234
    },
    {
      "epoch": 0.46778654987610213,
      "grad_norm": 0.7545861601829529,
      "learning_rate": 5.761795479111515e-06,
      "loss": 0.7052,
      "step": 16235
    },
    {
      "epoch": 0.4678153633377514,
      "grad_norm": 0.7511993050575256,
      "learning_rate": 5.761334313248076e-06,
      "loss": 0.762,
      "step": 16236
    },
    {
      "epoch": 0.46784417679940066,
      "grad_norm": 0.738838255405426,
      "learning_rate": 5.760873140754175e-06,
      "loss": 0.7179,
      "step": 16237
    },
    {
      "epoch": 0.46787299026104995,
      "grad_norm": 0.7386723160743713,
      "learning_rate": 5.76041196163383e-06,
      "loss": 0.7399,
      "step": 16238
    },
    {
      "epoch": 0.46790180372269924,
      "grad_norm": 0.762342095375061,
      "learning_rate": 5.759950775891058e-06,
      "loss": 0.7193,
      "step": 16239
    },
    {
      "epoch": 0.46793061718434853,
      "grad_norm": 0.7372989654541016,
      "learning_rate": 5.759489583529872e-06,
      "loss": 0.7582,
      "step": 16240
    },
    {
      "epoch": 0.4679594306459978,
      "grad_norm": 0.7506442070007324,
      "learning_rate": 5.759028384554294e-06,
      "loss": 0.7304,
      "step": 16241
    },
    {
      "epoch": 0.4679882441076471,
      "grad_norm": 0.7330368757247925,
      "learning_rate": 5.758567178968336e-06,
      "loss": 0.7118,
      "step": 16242
    },
    {
      "epoch": 0.4680170575692964,
      "grad_norm": 0.7510859370231628,
      "learning_rate": 5.758105966776017e-06,
      "loss": 0.7399,
      "step": 16243
    },
    {
      "epoch": 0.46804587103094564,
      "grad_norm": 0.7443134784698486,
      "learning_rate": 5.7576447479813516e-06,
      "loss": 0.717,
      "step": 16244
    },
    {
      "epoch": 0.46807468449259493,
      "grad_norm": 0.7221397757530212,
      "learning_rate": 5.757183522588359e-06,
      "loss": 0.7051,
      "step": 16245
    },
    {
      "epoch": 0.4681034979542442,
      "grad_norm": 0.7284433245658875,
      "learning_rate": 5.7567222906010525e-06,
      "loss": 0.7063,
      "step": 16246
    },
    {
      "epoch": 0.4681323114158935,
      "grad_norm": 0.7509209513664246,
      "learning_rate": 5.756261052023453e-06,
      "loss": 0.7314,
      "step": 16247
    },
    {
      "epoch": 0.4681611248775428,
      "grad_norm": 0.7268222570419312,
      "learning_rate": 5.755799806859575e-06,
      "loss": 0.7196,
      "step": 16248
    },
    {
      "epoch": 0.4681899383391921,
      "grad_norm": 0.7675920724868774,
      "learning_rate": 5.755338555113438e-06,
      "loss": 0.7417,
      "step": 16249
    },
    {
      "epoch": 0.4682187518008413,
      "grad_norm": 0.7554382681846619,
      "learning_rate": 5.7548772967890565e-06,
      "loss": 0.7305,
      "step": 16250
    },
    {
      "epoch": 0.4682475652624906,
      "grad_norm": 0.7083373069763184,
      "learning_rate": 5.7544160318904465e-06,
      "loss": 0.7268,
      "step": 16251
    },
    {
      "epoch": 0.4682763787241399,
      "grad_norm": 0.7161439657211304,
      "learning_rate": 5.753954760421628e-06,
      "loss": 0.7143,
      "step": 16252
    },
    {
      "epoch": 0.4683051921857892,
      "grad_norm": 0.7421313524246216,
      "learning_rate": 5.753493482386617e-06,
      "loss": 0.7327,
      "step": 16253
    },
    {
      "epoch": 0.4683340056474385,
      "grad_norm": 0.7503147721290588,
      "learning_rate": 5.753032197789432e-06,
      "loss": 0.7102,
      "step": 16254
    },
    {
      "epoch": 0.4683628191090878,
      "grad_norm": 0.7258586883544922,
      "learning_rate": 5.75257090663409e-06,
      "loss": 0.6998,
      "step": 16255
    },
    {
      "epoch": 0.46839163257073707,
      "grad_norm": 0.7416015863418579,
      "learning_rate": 5.7521096089246056e-06,
      "loss": 0.7295,
      "step": 16256
    },
    {
      "epoch": 0.4684204460323863,
      "grad_norm": 0.7468461394309998,
      "learning_rate": 5.751648304665e-06,
      "loss": 0.7149,
      "step": 16257
    },
    {
      "epoch": 0.4684492594940356,
      "grad_norm": 0.7619380950927734,
      "learning_rate": 5.751186993859287e-06,
      "loss": 0.7137,
      "step": 16258
    },
    {
      "epoch": 0.4684780729556849,
      "grad_norm": 0.7263543605804443,
      "learning_rate": 5.750725676511488e-06,
      "loss": 0.7064,
      "step": 16259
    },
    {
      "epoch": 0.4685068864173342,
      "grad_norm": 0.772106945514679,
      "learning_rate": 5.750264352625619e-06,
      "loss": 0.7326,
      "step": 16260
    },
    {
      "epoch": 0.46853569987898347,
      "grad_norm": 0.7522925138473511,
      "learning_rate": 5.749803022205697e-06,
      "loss": 0.7135,
      "step": 16261
    },
    {
      "epoch": 0.46856451334063276,
      "grad_norm": 0.7482836246490479,
      "learning_rate": 5.74934168525574e-06,
      "loss": 0.7282,
      "step": 16262
    },
    {
      "epoch": 0.46859332680228205,
      "grad_norm": 0.7637324333190918,
      "learning_rate": 5.748880341779766e-06,
      "loss": 0.7313,
      "step": 16263
    },
    {
      "epoch": 0.4686221402639313,
      "grad_norm": 0.7324145436286926,
      "learning_rate": 5.748418991781793e-06,
      "loss": 0.6986,
      "step": 16264
    },
    {
      "epoch": 0.4686509537255806,
      "grad_norm": 0.7544049024581909,
      "learning_rate": 5.747957635265838e-06,
      "loss": 0.7153,
      "step": 16265
    },
    {
      "epoch": 0.46867976718722987,
      "grad_norm": 0.7724384665489197,
      "learning_rate": 5.74749627223592e-06,
      "loss": 0.7214,
      "step": 16266
    },
    {
      "epoch": 0.46870858064887916,
      "grad_norm": 0.7636621594429016,
      "learning_rate": 5.747034902696058e-06,
      "loss": 0.7318,
      "step": 16267
    },
    {
      "epoch": 0.46873739411052845,
      "grad_norm": 0.7756569385528564,
      "learning_rate": 5.746573526650267e-06,
      "loss": 0.7179,
      "step": 16268
    },
    {
      "epoch": 0.46876620757217774,
      "grad_norm": 0.7759689092636108,
      "learning_rate": 5.746112144102568e-06,
      "loss": 0.7149,
      "step": 16269
    },
    {
      "epoch": 0.46879502103382703,
      "grad_norm": 0.7495132088661194,
      "learning_rate": 5.7456507550569775e-06,
      "loss": 0.7169,
      "step": 16270
    },
    {
      "epoch": 0.46882383449547627,
      "grad_norm": 0.7580627202987671,
      "learning_rate": 5.745189359517515e-06,
      "loss": 0.7321,
      "step": 16271
    },
    {
      "epoch": 0.46885264795712556,
      "grad_norm": 0.7480195164680481,
      "learning_rate": 5.744727957488197e-06,
      "loss": 0.7338,
      "step": 16272
    },
    {
      "epoch": 0.46888146141877485,
      "grad_norm": 0.7396832704544067,
      "learning_rate": 5.744266548973043e-06,
      "loss": 0.6984,
      "step": 16273
    },
    {
      "epoch": 0.46891027488042414,
      "grad_norm": 0.7355055809020996,
      "learning_rate": 5.7438051339760715e-06,
      "loss": 0.7414,
      "step": 16274
    },
    {
      "epoch": 0.46893908834207343,
      "grad_norm": 0.7141011357307434,
      "learning_rate": 5.743343712501301e-06,
      "loss": 0.7051,
      "step": 16275
    },
    {
      "epoch": 0.4689679018037227,
      "grad_norm": 0.7151517271995544,
      "learning_rate": 5.74288228455275e-06,
      "loss": 0.7282,
      "step": 16276
    },
    {
      "epoch": 0.46899671526537196,
      "grad_norm": 0.7138208746910095,
      "learning_rate": 5.742420850134437e-06,
      "loss": 0.7284,
      "step": 16277
    },
    {
      "epoch": 0.46902552872702125,
      "grad_norm": 0.7572656273841858,
      "learning_rate": 5.74195940925038e-06,
      "loss": 0.7331,
      "step": 16278
    },
    {
      "epoch": 0.46905434218867054,
      "grad_norm": 0.7605298161506653,
      "learning_rate": 5.7414979619045995e-06,
      "loss": 0.7408,
      "step": 16279
    },
    {
      "epoch": 0.4690831556503198,
      "grad_norm": 0.7361665964126587,
      "learning_rate": 5.7410365081011125e-06,
      "loss": 0.738,
      "step": 16280
    },
    {
      "epoch": 0.4691119691119691,
      "grad_norm": 0.7179305553436279,
      "learning_rate": 5.740575047843938e-06,
      "loss": 0.7191,
      "step": 16281
    },
    {
      "epoch": 0.4691407825736184,
      "grad_norm": 0.8931541442871094,
      "learning_rate": 5.740113581137094e-06,
      "loss": 0.7082,
      "step": 16282
    },
    {
      "epoch": 0.4691695960352677,
      "grad_norm": 0.7431004047393799,
      "learning_rate": 5.739652107984602e-06,
      "loss": 0.6983,
      "step": 16283
    },
    {
      "epoch": 0.46919840949691693,
      "grad_norm": 0.7830125093460083,
      "learning_rate": 5.73919062839048e-06,
      "loss": 0.7385,
      "step": 16284
    },
    {
      "epoch": 0.4692272229585662,
      "grad_norm": 0.75694340467453,
      "learning_rate": 5.738729142358745e-06,
      "loss": 0.7322,
      "step": 16285
    },
    {
      "epoch": 0.4692560364202155,
      "grad_norm": 0.745878279209137,
      "learning_rate": 5.738267649893419e-06,
      "loss": 0.7146,
      "step": 16286
    },
    {
      "epoch": 0.4692848498818648,
      "grad_norm": 0.7417734861373901,
      "learning_rate": 5.73780615099852e-06,
      "loss": 0.7106,
      "step": 16287
    },
    {
      "epoch": 0.4693136633435141,
      "grad_norm": 0.7327975630760193,
      "learning_rate": 5.7373446456780644e-06,
      "loss": 0.725,
      "step": 16288
    },
    {
      "epoch": 0.4693424768051634,
      "grad_norm": 0.7591292858123779,
      "learning_rate": 5.736883133936076e-06,
      "loss": 0.7101,
      "step": 16289
    },
    {
      "epoch": 0.4693712902668127,
      "grad_norm": 0.746109664440155,
      "learning_rate": 5.7364216157765725e-06,
      "loss": 0.7315,
      "step": 16290
    },
    {
      "epoch": 0.4694001037284619,
      "grad_norm": 0.7456250786781311,
      "learning_rate": 5.735960091203571e-06,
      "loss": 0.6971,
      "step": 16291
    },
    {
      "epoch": 0.4694289171901112,
      "grad_norm": 0.7432185411453247,
      "learning_rate": 5.7354985602210945e-06,
      "loss": 0.7258,
      "step": 16292
    },
    {
      "epoch": 0.4694577306517605,
      "grad_norm": 0.7372882962226868,
      "learning_rate": 5.735037022833159e-06,
      "loss": 0.7221,
      "step": 16293
    },
    {
      "epoch": 0.4694865441134098,
      "grad_norm": 0.7470967769622803,
      "learning_rate": 5.734575479043787e-06,
      "loss": 0.7495,
      "step": 16294
    },
    {
      "epoch": 0.4695153575750591,
      "grad_norm": 0.7897334098815918,
      "learning_rate": 5.734113928856998e-06,
      "loss": 0.7564,
      "step": 16295
    },
    {
      "epoch": 0.46954417103670837,
      "grad_norm": 0.7515094876289368,
      "learning_rate": 5.733652372276809e-06,
      "loss": 0.7394,
      "step": 16296
    },
    {
      "epoch": 0.46957298449835766,
      "grad_norm": 0.768149733543396,
      "learning_rate": 5.733190809307242e-06,
      "loss": 0.7171,
      "step": 16297
    },
    {
      "epoch": 0.4696017979600069,
      "grad_norm": 0.7433072328567505,
      "learning_rate": 5.732729239952316e-06,
      "loss": 0.7184,
      "step": 16298
    },
    {
      "epoch": 0.4696306114216562,
      "grad_norm": 0.7148116230964661,
      "learning_rate": 5.732267664216049e-06,
      "loss": 0.6962,
      "step": 16299
    },
    {
      "epoch": 0.4696594248833055,
      "grad_norm": 0.7342842817306519,
      "learning_rate": 5.7318060821024645e-06,
      "loss": 0.7393,
      "step": 16300
    },
    {
      "epoch": 0.46968823834495477,
      "grad_norm": 0.7704523801803589,
      "learning_rate": 5.73134449361558e-06,
      "loss": 0.7349,
      "step": 16301
    },
    {
      "epoch": 0.46971705180660406,
      "grad_norm": 0.8511801362037659,
      "learning_rate": 5.730882898759417e-06,
      "loss": 0.7515,
      "step": 16302
    },
    {
      "epoch": 0.46974586526825335,
      "grad_norm": 0.7400708198547363,
      "learning_rate": 5.730421297537994e-06,
      "loss": 0.735,
      "step": 16303
    },
    {
      "epoch": 0.4697746787299026,
      "grad_norm": 0.7228091359138489,
      "learning_rate": 5.729959689955331e-06,
      "loss": 0.6897,
      "step": 16304
    },
    {
      "epoch": 0.4698034921915519,
      "grad_norm": 0.7287968993186951,
      "learning_rate": 5.729498076015449e-06,
      "loss": 0.7223,
      "step": 16305
    },
    {
      "epoch": 0.46983230565320117,
      "grad_norm": 0.7428905367851257,
      "learning_rate": 5.729036455722369e-06,
      "loss": 0.7104,
      "step": 16306
    },
    {
      "epoch": 0.46986111911485046,
      "grad_norm": 0.7758216857910156,
      "learning_rate": 5.728574829080109e-06,
      "loss": 0.7381,
      "step": 16307
    },
    {
      "epoch": 0.46988993257649975,
      "grad_norm": 0.7776931524276733,
      "learning_rate": 5.728113196092692e-06,
      "loss": 0.7317,
      "step": 16308
    },
    {
      "epoch": 0.46991874603814904,
      "grad_norm": 0.7489044666290283,
      "learning_rate": 5.727651556764136e-06,
      "loss": 0.6955,
      "step": 16309
    },
    {
      "epoch": 0.46994755949979833,
      "grad_norm": 0.731978714466095,
      "learning_rate": 5.727189911098461e-06,
      "loss": 0.7301,
      "step": 16310
    },
    {
      "epoch": 0.46997637296144756,
      "grad_norm": 0.7567004561424255,
      "learning_rate": 5.72672825909969e-06,
      "loss": 0.6901,
      "step": 16311
    },
    {
      "epoch": 0.47000518642309685,
      "grad_norm": 0.7225848436355591,
      "learning_rate": 5.726266600771842e-06,
      "loss": 0.7229,
      "step": 16312
    },
    {
      "epoch": 0.47003399988474615,
      "grad_norm": 0.7582967281341553,
      "learning_rate": 5.725804936118937e-06,
      "loss": 0.7244,
      "step": 16313
    },
    {
      "epoch": 0.47006281334639544,
      "grad_norm": 0.7523930668830872,
      "learning_rate": 5.725343265144999e-06,
      "loss": 0.7366,
      "step": 16314
    },
    {
      "epoch": 0.4700916268080447,
      "grad_norm": 0.7717121839523315,
      "learning_rate": 5.724881587854045e-06,
      "loss": 0.7292,
      "step": 16315
    },
    {
      "epoch": 0.470120440269694,
      "grad_norm": 0.7602924108505249,
      "learning_rate": 5.724419904250097e-06,
      "loss": 0.7076,
      "step": 16316
    },
    {
      "epoch": 0.4701492537313433,
      "grad_norm": 0.7566391229629517,
      "learning_rate": 5.723958214337175e-06,
      "loss": 0.7228,
      "step": 16317
    },
    {
      "epoch": 0.47017806719299254,
      "grad_norm": 0.7385432720184326,
      "learning_rate": 5.723496518119301e-06,
      "loss": 0.7416,
      "step": 16318
    },
    {
      "epoch": 0.47020688065464183,
      "grad_norm": 0.7928712368011475,
      "learning_rate": 5.723034815600497e-06,
      "loss": 0.7073,
      "step": 16319
    },
    {
      "epoch": 0.4702356941162911,
      "grad_norm": 0.758402943611145,
      "learning_rate": 5.722573106784781e-06,
      "loss": 0.7168,
      "step": 16320
    },
    {
      "epoch": 0.4702645075779404,
      "grad_norm": 0.7515597939491272,
      "learning_rate": 5.722111391676177e-06,
      "loss": 0.7122,
      "step": 16321
    },
    {
      "epoch": 0.4702933210395897,
      "grad_norm": 0.7339771389961243,
      "learning_rate": 5.721649670278704e-06,
      "loss": 0.7296,
      "step": 16322
    },
    {
      "epoch": 0.470322134501239,
      "grad_norm": 0.7641220688819885,
      "learning_rate": 5.721187942596382e-06,
      "loss": 0.7333,
      "step": 16323
    },
    {
      "epoch": 0.4703509479628883,
      "grad_norm": 0.7554459571838379,
      "learning_rate": 5.720726208633236e-06,
      "loss": 0.727,
      "step": 16324
    },
    {
      "epoch": 0.4703797614245375,
      "grad_norm": 0.7435435056686401,
      "learning_rate": 5.720264468393285e-06,
      "loss": 0.7305,
      "step": 16325
    },
    {
      "epoch": 0.4704085748861868,
      "grad_norm": 0.7206472754478455,
      "learning_rate": 5.71980272188055e-06,
      "loss": 0.7039,
      "step": 16326
    },
    {
      "epoch": 0.4704373883478361,
      "grad_norm": 0.7263020873069763,
      "learning_rate": 5.719340969099053e-06,
      "loss": 0.7319,
      "step": 16327
    },
    {
      "epoch": 0.4704662018094854,
      "grad_norm": 0.7170253396034241,
      "learning_rate": 5.718879210052816e-06,
      "loss": 0.713,
      "step": 16328
    },
    {
      "epoch": 0.4704950152711347,
      "grad_norm": 0.7427471876144409,
      "learning_rate": 5.718417444745858e-06,
      "loss": 0.7037,
      "step": 16329
    },
    {
      "epoch": 0.470523828732784,
      "grad_norm": 0.7369354963302612,
      "learning_rate": 5.7179556731822025e-06,
      "loss": 0.7271,
      "step": 16330
    },
    {
      "epoch": 0.4705526421944332,
      "grad_norm": 0.7758051753044128,
      "learning_rate": 5.717493895365872e-06,
      "loss": 0.7329,
      "step": 16331
    },
    {
      "epoch": 0.4705814556560825,
      "grad_norm": 0.7960157990455627,
      "learning_rate": 5.717032111300885e-06,
      "loss": 0.7181,
      "step": 16332
    },
    {
      "epoch": 0.4706102691177318,
      "grad_norm": 0.7299719452857971,
      "learning_rate": 5.716570320991267e-06,
      "loss": 0.7053,
      "step": 16333
    },
    {
      "epoch": 0.4706390825793811,
      "grad_norm": 0.7462152242660522,
      "learning_rate": 5.7161085244410376e-06,
      "loss": 0.7434,
      "step": 16334
    },
    {
      "epoch": 0.4706678960410304,
      "grad_norm": 0.7311298251152039,
      "learning_rate": 5.715646721654216e-06,
      "loss": 0.7134,
      "step": 16335
    },
    {
      "epoch": 0.47069670950267967,
      "grad_norm": 0.7650576233863831,
      "learning_rate": 5.71518491263483e-06,
      "loss": 0.7302,
      "step": 16336
    },
    {
      "epoch": 0.47072552296432896,
      "grad_norm": 0.7435930967330933,
      "learning_rate": 5.714723097386898e-06,
      "loss": 0.7221,
      "step": 16337
    },
    {
      "epoch": 0.4707543364259782,
      "grad_norm": 0.7235928773880005,
      "learning_rate": 5.714261275914442e-06,
      "loss": 0.7162,
      "step": 16338
    },
    {
      "epoch": 0.4707831498876275,
      "grad_norm": 0.7419122457504272,
      "learning_rate": 5.713799448221485e-06,
      "loss": 0.7282,
      "step": 16339
    },
    {
      "epoch": 0.4708119633492768,
      "grad_norm": 0.7641946077346802,
      "learning_rate": 5.713337614312046e-06,
      "loss": 0.7515,
      "step": 16340
    },
    {
      "epoch": 0.47084077681092606,
      "grad_norm": 0.8100445866584778,
      "learning_rate": 5.712875774190151e-06,
      "loss": 0.7115,
      "step": 16341
    },
    {
      "epoch": 0.47086959027257536,
      "grad_norm": 0.7094491124153137,
      "learning_rate": 5.712413927859821e-06,
      "loss": 0.7024,
      "step": 16342
    },
    {
      "epoch": 0.47089840373422465,
      "grad_norm": 0.7410751581192017,
      "learning_rate": 5.7119520753250764e-06,
      "loss": 0.742,
      "step": 16343
    },
    {
      "epoch": 0.47092721719587394,
      "grad_norm": 0.7162894010543823,
      "learning_rate": 5.7114902165899436e-06,
      "loss": 0.7338,
      "step": 16344
    },
    {
      "epoch": 0.4709560306575232,
      "grad_norm": 0.7548100352287292,
      "learning_rate": 5.7110283516584394e-06,
      "loss": 0.7424,
      "step": 16345
    },
    {
      "epoch": 0.47098484411917246,
      "grad_norm": 0.7448004484176636,
      "learning_rate": 5.710566480534591e-06,
      "loss": 0.7102,
      "step": 16346
    },
    {
      "epoch": 0.47101365758082175,
      "grad_norm": 0.7156579494476318,
      "learning_rate": 5.710104603222417e-06,
      "loss": 0.7021,
      "step": 16347
    },
    {
      "epoch": 0.47104247104247104,
      "grad_norm": 0.741104245185852,
      "learning_rate": 5.709642719725943e-06,
      "loss": 0.7212,
      "step": 16348
    },
    {
      "epoch": 0.47107128450412034,
      "grad_norm": 0.7639846801757812,
      "learning_rate": 5.70918083004919e-06,
      "loss": 0.7246,
      "step": 16349
    },
    {
      "epoch": 0.4711000979657696,
      "grad_norm": 0.7581343650817871,
      "learning_rate": 5.70871893419618e-06,
      "loss": 0.7201,
      "step": 16350
    },
    {
      "epoch": 0.4711289114274189,
      "grad_norm": 0.7605852484703064,
      "learning_rate": 5.708257032170937e-06,
      "loss": 0.7537,
      "step": 16351
    },
    {
      "epoch": 0.47115772488906815,
      "grad_norm": 0.7376661896705627,
      "learning_rate": 5.7077951239774845e-06,
      "loss": 0.7399,
      "step": 16352
    },
    {
      "epoch": 0.47118653835071744,
      "grad_norm": 0.7625842094421387,
      "learning_rate": 5.707333209619843e-06,
      "loss": 0.712,
      "step": 16353
    },
    {
      "epoch": 0.47121535181236673,
      "grad_norm": 0.7353786826133728,
      "learning_rate": 5.706871289102036e-06,
      "loss": 0.7175,
      "step": 16354
    },
    {
      "epoch": 0.471244165274016,
      "grad_norm": 0.7412909269332886,
      "learning_rate": 5.706409362428088e-06,
      "loss": 0.7147,
      "step": 16355
    },
    {
      "epoch": 0.4712729787356653,
      "grad_norm": 0.7422028183937073,
      "learning_rate": 5.70594742960202e-06,
      "loss": 0.6993,
      "step": 16356
    },
    {
      "epoch": 0.4713017921973146,
      "grad_norm": 0.7729076147079468,
      "learning_rate": 5.705485490627856e-06,
      "loss": 0.7292,
      "step": 16357
    },
    {
      "epoch": 0.47133060565896384,
      "grad_norm": 0.7254570126533508,
      "learning_rate": 5.705023545509618e-06,
      "loss": 0.714,
      "step": 16358
    },
    {
      "epoch": 0.47135941912061313,
      "grad_norm": 0.7735952138900757,
      "learning_rate": 5.704561594251329e-06,
      "loss": 0.7421,
      "step": 16359
    },
    {
      "epoch": 0.4713882325822624,
      "grad_norm": 0.7415897250175476,
      "learning_rate": 5.704099636857014e-06,
      "loss": 0.7233,
      "step": 16360
    },
    {
      "epoch": 0.4714170460439117,
      "grad_norm": 0.7249730825424194,
      "learning_rate": 5.703637673330694e-06,
      "loss": 0.726,
      "step": 16361
    },
    {
      "epoch": 0.471445859505561,
      "grad_norm": 0.7526528835296631,
      "learning_rate": 5.7031757036763934e-06,
      "loss": 0.7204,
      "step": 16362
    },
    {
      "epoch": 0.4714746729672103,
      "grad_norm": 0.7602314949035645,
      "learning_rate": 5.702713727898136e-06,
      "loss": 0.7463,
      "step": 16363
    },
    {
      "epoch": 0.4715034864288596,
      "grad_norm": 0.7160084843635559,
      "learning_rate": 5.702251745999945e-06,
      "loss": 0.7066,
      "step": 16364
    },
    {
      "epoch": 0.4715322998905088,
      "grad_norm": 0.7648904919624329,
      "learning_rate": 5.7017897579858415e-06,
      "loss": 0.7025,
      "step": 16365
    },
    {
      "epoch": 0.4715611133521581,
      "grad_norm": 0.722790539264679,
      "learning_rate": 5.701327763859852e-06,
      "loss": 0.7006,
      "step": 16366
    },
    {
      "epoch": 0.4715899268138074,
      "grad_norm": 0.796536386013031,
      "learning_rate": 5.7008657636259975e-06,
      "loss": 0.7097,
      "step": 16367
    },
    {
      "epoch": 0.4716187402754567,
      "grad_norm": 0.7611997723579407,
      "learning_rate": 5.700403757288305e-06,
      "loss": 0.7082,
      "step": 16368
    },
    {
      "epoch": 0.471647553737106,
      "grad_norm": 0.7246140241622925,
      "learning_rate": 5.699941744850793e-06,
      "loss": 0.717,
      "step": 16369
    },
    {
      "epoch": 0.4716763671987553,
      "grad_norm": 0.7261278033256531,
      "learning_rate": 5.6994797263174906e-06,
      "loss": 0.7236,
      "step": 16370
    },
    {
      "epoch": 0.47170518066040457,
      "grad_norm": 0.7314015626907349,
      "learning_rate": 5.699017701692419e-06,
      "loss": 0.696,
      "step": 16371
    },
    {
      "epoch": 0.4717339941220538,
      "grad_norm": 0.7444297075271606,
      "learning_rate": 5.6985556709796e-06,
      "loss": 0.7225,
      "step": 16372
    },
    {
      "epoch": 0.4717628075837031,
      "grad_norm": 0.7476418614387512,
      "learning_rate": 5.698093634183061e-06,
      "loss": 0.7349,
      "step": 16373
    },
    {
      "epoch": 0.4717916210453524,
      "grad_norm": 0.7331565618515015,
      "learning_rate": 5.6976315913068235e-06,
      "loss": 0.7288,
      "step": 16374
    },
    {
      "epoch": 0.4718204345070017,
      "grad_norm": 0.7577521800994873,
      "learning_rate": 5.697169542354912e-06,
      "loss": 0.7342,
      "step": 16375
    },
    {
      "epoch": 0.47184924796865096,
      "grad_norm": 0.7155104875564575,
      "learning_rate": 5.696707487331352e-06,
      "loss": 0.6901,
      "step": 16376
    },
    {
      "epoch": 0.47187806143030026,
      "grad_norm": 0.7551345825195312,
      "learning_rate": 5.696245426240164e-06,
      "loss": 0.7105,
      "step": 16377
    },
    {
      "epoch": 0.47190687489194955,
      "grad_norm": 0.7489296793937683,
      "learning_rate": 5.695783359085378e-06,
      "loss": 0.7355,
      "step": 16378
    },
    {
      "epoch": 0.4719356883535988,
      "grad_norm": 0.7289673089981079,
      "learning_rate": 5.695321285871012e-06,
      "loss": 0.6822,
      "step": 16379
    },
    {
      "epoch": 0.47196450181524807,
      "grad_norm": 0.7529336214065552,
      "learning_rate": 5.694859206601092e-06,
      "loss": 0.7246,
      "step": 16380
    },
    {
      "epoch": 0.47199331527689736,
      "grad_norm": 0.7088434100151062,
      "learning_rate": 5.694397121279645e-06,
      "loss": 0.7188,
      "step": 16381
    },
    {
      "epoch": 0.47202212873854665,
      "grad_norm": 0.7556015849113464,
      "learning_rate": 5.693935029910692e-06,
      "loss": 0.725,
      "step": 16382
    },
    {
      "epoch": 0.47205094220019594,
      "grad_norm": 0.7257722616195679,
      "learning_rate": 5.693472932498257e-06,
      "loss": 0.7109,
      "step": 16383
    },
    {
      "epoch": 0.47207975566184524,
      "grad_norm": 0.713824450969696,
      "learning_rate": 5.693010829046368e-06,
      "loss": 0.7093,
      "step": 16384
    },
    {
      "epoch": 0.47210856912349447,
      "grad_norm": 0.7283521890640259,
      "learning_rate": 5.692548719559048e-06,
      "loss": 0.718,
      "step": 16385
    },
    {
      "epoch": 0.47213738258514376,
      "grad_norm": 0.7455669641494751,
      "learning_rate": 5.69208660404032e-06,
      "loss": 0.7142,
      "step": 16386
    },
    {
      "epoch": 0.47216619604679305,
      "grad_norm": 0.7460308074951172,
      "learning_rate": 5.69162448249421e-06,
      "loss": 0.7225,
      "step": 16387
    },
    {
      "epoch": 0.47219500950844234,
      "grad_norm": 0.7326490879058838,
      "learning_rate": 5.691162354924741e-06,
      "loss": 0.7416,
      "step": 16388
    },
    {
      "epoch": 0.47222382297009163,
      "grad_norm": 0.7500430941581726,
      "learning_rate": 5.6907002213359395e-06,
      "loss": 0.7309,
      "step": 16389
    },
    {
      "epoch": 0.4722526364317409,
      "grad_norm": 0.7337863445281982,
      "learning_rate": 5.690238081731829e-06,
      "loss": 0.7148,
      "step": 16390
    },
    {
      "epoch": 0.4722814498933902,
      "grad_norm": 0.764339804649353,
      "learning_rate": 5.689775936116435e-06,
      "loss": 0.7325,
      "step": 16391
    },
    {
      "epoch": 0.47231026335503945,
      "grad_norm": 0.7313470244407654,
      "learning_rate": 5.689313784493783e-06,
      "loss": 0.7334,
      "step": 16392
    },
    {
      "epoch": 0.47233907681668874,
      "grad_norm": 0.7308128476142883,
      "learning_rate": 5.6888516268678966e-06,
      "loss": 0.7365,
      "step": 16393
    },
    {
      "epoch": 0.47236789027833803,
      "grad_norm": 0.7385935187339783,
      "learning_rate": 5.6883894632428e-06,
      "loss": 0.7204,
      "step": 16394
    },
    {
      "epoch": 0.4723967037399873,
      "grad_norm": 0.7401861548423767,
      "learning_rate": 5.68792729362252e-06,
      "loss": 0.7513,
      "step": 16395
    },
    {
      "epoch": 0.4724255172016366,
      "grad_norm": 0.7371009588241577,
      "learning_rate": 5.687465118011082e-06,
      "loss": 0.7422,
      "step": 16396
    },
    {
      "epoch": 0.4724543306632859,
      "grad_norm": 0.7399502992630005,
      "learning_rate": 5.687002936412509e-06,
      "loss": 0.7187,
      "step": 16397
    },
    {
      "epoch": 0.4724831441249352,
      "grad_norm": 0.7195896506309509,
      "learning_rate": 5.686540748830828e-06,
      "loss": 0.7384,
      "step": 16398
    },
    {
      "epoch": 0.47251195758658443,
      "grad_norm": 0.7469426989555359,
      "learning_rate": 5.686078555270063e-06,
      "loss": 0.7195,
      "step": 16399
    },
    {
      "epoch": 0.4725407710482337,
      "grad_norm": 0.7501277923583984,
      "learning_rate": 5.685616355734239e-06,
      "loss": 0.7421,
      "step": 16400
    },
    {
      "epoch": 0.472569584509883,
      "grad_norm": 0.7292754650115967,
      "learning_rate": 5.685154150227381e-06,
      "loss": 0.6937,
      "step": 16401
    },
    {
      "epoch": 0.4725983979715323,
      "grad_norm": 0.7591938972473145,
      "learning_rate": 5.684691938753517e-06,
      "loss": 0.7047,
      "step": 16402
    },
    {
      "epoch": 0.4726272114331816,
      "grad_norm": 0.7486067414283752,
      "learning_rate": 5.684229721316671e-06,
      "loss": 0.7267,
      "step": 16403
    },
    {
      "epoch": 0.4726560248948309,
      "grad_norm": 0.7416406869888306,
      "learning_rate": 5.683767497920867e-06,
      "loss": 0.7275,
      "step": 16404
    },
    {
      "epoch": 0.4726848383564802,
      "grad_norm": 0.7292041778564453,
      "learning_rate": 5.683305268570133e-06,
      "loss": 0.7409,
      "step": 16405
    },
    {
      "epoch": 0.4727136518181294,
      "grad_norm": 0.7312073707580566,
      "learning_rate": 5.682843033268491e-06,
      "loss": 0.7172,
      "step": 16406
    },
    {
      "epoch": 0.4727424652797787,
      "grad_norm": 0.7345378994941711,
      "learning_rate": 5.68238079201997e-06,
      "loss": 0.6979,
      "step": 16407
    },
    {
      "epoch": 0.472771278741428,
      "grad_norm": 0.7496793270111084,
      "learning_rate": 5.681918544828594e-06,
      "loss": 0.7518,
      "step": 16408
    },
    {
      "epoch": 0.4728000922030773,
      "grad_norm": 0.7306903600692749,
      "learning_rate": 5.681456291698389e-06,
      "loss": 0.7312,
      "step": 16409
    },
    {
      "epoch": 0.4728289056647266,
      "grad_norm": 0.7557992935180664,
      "learning_rate": 5.6809940326333814e-06,
      "loss": 0.7278,
      "step": 16410
    },
    {
      "epoch": 0.47285771912637586,
      "grad_norm": 0.7312700152397156,
      "learning_rate": 5.680531767637597e-06,
      "loss": 0.7101,
      "step": 16411
    },
    {
      "epoch": 0.4728865325880251,
      "grad_norm": 0.7410195469856262,
      "learning_rate": 5.6800694967150605e-06,
      "loss": 0.6877,
      "step": 16412
    },
    {
      "epoch": 0.4729153460496744,
      "grad_norm": 0.716561496257782,
      "learning_rate": 5.6796072198697985e-06,
      "loss": 0.6973,
      "step": 16413
    },
    {
      "epoch": 0.4729441595113237,
      "grad_norm": 0.7331297397613525,
      "learning_rate": 5.679144937105836e-06,
      "loss": 0.7377,
      "step": 16414
    },
    {
      "epoch": 0.47297297297297297,
      "grad_norm": 0.7427162528038025,
      "learning_rate": 5.678682648427202e-06,
      "loss": 0.7338,
      "step": 16415
    },
    {
      "epoch": 0.47300178643462226,
      "grad_norm": 0.767815113067627,
      "learning_rate": 5.67822035383792e-06,
      "loss": 0.7114,
      "step": 16416
    },
    {
      "epoch": 0.47303059989627155,
      "grad_norm": 0.7169609069824219,
      "learning_rate": 5.677758053342016e-06,
      "loss": 0.718,
      "step": 16417
    },
    {
      "epoch": 0.47305941335792084,
      "grad_norm": 0.71286940574646,
      "learning_rate": 5.677295746943517e-06,
      "loss": 0.7318,
      "step": 16418
    },
    {
      "epoch": 0.4730882268195701,
      "grad_norm": 0.7152239680290222,
      "learning_rate": 5.676833434646449e-06,
      "loss": 0.718,
      "step": 16419
    },
    {
      "epoch": 0.47311704028121937,
      "grad_norm": 0.7228022813796997,
      "learning_rate": 5.67637111645484e-06,
      "loss": 0.7035,
      "step": 16420
    },
    {
      "epoch": 0.47314585374286866,
      "grad_norm": 0.7409286499023438,
      "learning_rate": 5.675908792372713e-06,
      "loss": 0.7196,
      "step": 16421
    },
    {
      "epoch": 0.47317466720451795,
      "grad_norm": 0.7293127179145813,
      "learning_rate": 5.675446462404096e-06,
      "loss": 0.7095,
      "step": 16422
    },
    {
      "epoch": 0.47320348066616724,
      "grad_norm": 0.7551258206367493,
      "learning_rate": 5.674984126553017e-06,
      "loss": 0.7265,
      "step": 16423
    },
    {
      "epoch": 0.47323229412781653,
      "grad_norm": 0.7715526819229126,
      "learning_rate": 5.6745217848235006e-06,
      "loss": 0.7168,
      "step": 16424
    },
    {
      "epoch": 0.4732611075894658,
      "grad_norm": 0.7427007555961609,
      "learning_rate": 5.674059437219571e-06,
      "loss": 0.712,
      "step": 16425
    },
    {
      "epoch": 0.47328992105111506,
      "grad_norm": 0.7390949726104736,
      "learning_rate": 5.67359708374526e-06,
      "loss": 0.7024,
      "step": 16426
    },
    {
      "epoch": 0.47331873451276435,
      "grad_norm": 0.7328099012374878,
      "learning_rate": 5.673134724404592e-06,
      "loss": 0.7326,
      "step": 16427
    },
    {
      "epoch": 0.47334754797441364,
      "grad_norm": 0.7158177495002747,
      "learning_rate": 5.672672359201593e-06,
      "loss": 0.701,
      "step": 16428
    },
    {
      "epoch": 0.47337636143606293,
      "grad_norm": 0.7569079399108887,
      "learning_rate": 5.672209988140289e-06,
      "loss": 0.7269,
      "step": 16429
    },
    {
      "epoch": 0.4734051748977122,
      "grad_norm": 0.7676510214805603,
      "learning_rate": 5.67174761122471e-06,
      "loss": 0.725,
      "step": 16430
    },
    {
      "epoch": 0.4734339883593615,
      "grad_norm": 0.728143036365509,
      "learning_rate": 5.671285228458878e-06,
      "loss": 0.7436,
      "step": 16431
    },
    {
      "epoch": 0.4734628018210108,
      "grad_norm": 0.7639551162719727,
      "learning_rate": 5.670822839846825e-06,
      "loss": 0.739,
      "step": 16432
    },
    {
      "epoch": 0.47349161528266004,
      "grad_norm": 0.7028756737709045,
      "learning_rate": 5.6703604453925756e-06,
      "loss": 0.6881,
      "step": 16433
    },
    {
      "epoch": 0.47352042874430933,
      "grad_norm": 0.7111311554908752,
      "learning_rate": 5.669898045100156e-06,
      "loss": 0.719,
      "step": 16434
    },
    {
      "epoch": 0.4735492422059586,
      "grad_norm": 0.7348019480705261,
      "learning_rate": 5.669435638973595e-06,
      "loss": 0.7008,
      "step": 16435
    },
    {
      "epoch": 0.4735780556676079,
      "grad_norm": 0.7435672283172607,
      "learning_rate": 5.6689732270169185e-06,
      "loss": 0.7337,
      "step": 16436
    },
    {
      "epoch": 0.4736068691292572,
      "grad_norm": 0.7549231052398682,
      "learning_rate": 5.668510809234153e-06,
      "loss": 0.7297,
      "step": 16437
    },
    {
      "epoch": 0.4736356825909065,
      "grad_norm": 0.7375586628913879,
      "learning_rate": 5.668048385629326e-06,
      "loss": 0.7423,
      "step": 16438
    },
    {
      "epoch": 0.47366449605255573,
      "grad_norm": 0.749860405921936,
      "learning_rate": 5.667585956206467e-06,
      "loss": 0.7155,
      "step": 16439
    },
    {
      "epoch": 0.473693309514205,
      "grad_norm": 0.7262625098228455,
      "learning_rate": 5.667123520969602e-06,
      "loss": 0.7066,
      "step": 16440
    },
    {
      "epoch": 0.4737221229758543,
      "grad_norm": 0.7273379564285278,
      "learning_rate": 5.666661079922757e-06,
      "loss": 0.7535,
      "step": 16441
    },
    {
      "epoch": 0.4737509364375036,
      "grad_norm": 0.7423539161682129,
      "learning_rate": 5.666198633069961e-06,
      "loss": 0.6853,
      "step": 16442
    },
    {
      "epoch": 0.4737797498991529,
      "grad_norm": 0.7330166101455688,
      "learning_rate": 5.66573618041524e-06,
      "loss": 0.7035,
      "step": 16443
    },
    {
      "epoch": 0.4738085633608022,
      "grad_norm": 0.7319973707199097,
      "learning_rate": 5.665273721962623e-06,
      "loss": 0.7163,
      "step": 16444
    },
    {
      "epoch": 0.4738373768224515,
      "grad_norm": 0.7522265315055847,
      "learning_rate": 5.664811257716138e-06,
      "loss": 0.7632,
      "step": 16445
    },
    {
      "epoch": 0.4738661902841007,
      "grad_norm": 0.7277442812919617,
      "learning_rate": 5.66434878767981e-06,
      "loss": 0.7301,
      "step": 16446
    },
    {
      "epoch": 0.47389500374575,
      "grad_norm": 0.7395803928375244,
      "learning_rate": 5.663886311857669e-06,
      "loss": 0.7311,
      "step": 16447
    },
    {
      "epoch": 0.4739238172073993,
      "grad_norm": 0.7424360513687134,
      "learning_rate": 5.663423830253741e-06,
      "loss": 0.7068,
      "step": 16448
    },
    {
      "epoch": 0.4739526306690486,
      "grad_norm": 0.7203818559646606,
      "learning_rate": 5.662961342872054e-06,
      "loss": 0.7113,
      "step": 16449
    },
    {
      "epoch": 0.47398144413069787,
      "grad_norm": 0.7128749489784241,
      "learning_rate": 5.662498849716636e-06,
      "loss": 0.6985,
      "step": 16450
    },
    {
      "epoch": 0.47401025759234716,
      "grad_norm": 0.7734473347663879,
      "learning_rate": 5.662036350791518e-06,
      "loss": 0.7517,
      "step": 16451
    },
    {
      "epoch": 0.47403907105399645,
      "grad_norm": 0.7157405614852905,
      "learning_rate": 5.661573846100724e-06,
      "loss": 0.7067,
      "step": 16452
    },
    {
      "epoch": 0.4740678845156457,
      "grad_norm": 0.8227879405021667,
      "learning_rate": 5.6611113356482835e-06,
      "loss": 0.7539,
      "step": 16453
    },
    {
      "epoch": 0.474096697977295,
      "grad_norm": 0.7184422016143799,
      "learning_rate": 5.660648819438223e-06,
      "loss": 0.7113,
      "step": 16454
    },
    {
      "epoch": 0.47412551143894427,
      "grad_norm": 0.749277651309967,
      "learning_rate": 5.6601862974745724e-06,
      "loss": 0.7288,
      "step": 16455
    },
    {
      "epoch": 0.47415432490059356,
      "grad_norm": 0.7425945401191711,
      "learning_rate": 5.6597237697613585e-06,
      "loss": 0.7292,
      "step": 16456
    },
    {
      "epoch": 0.47418313836224285,
      "grad_norm": 0.7660138607025146,
      "learning_rate": 5.659261236302611e-06,
      "loss": 0.7656,
      "step": 16457
    },
    {
      "epoch": 0.47421195182389214,
      "grad_norm": 0.7310730814933777,
      "learning_rate": 5.658798697102357e-06,
      "loss": 0.7442,
      "step": 16458
    },
    {
      "epoch": 0.47424076528554143,
      "grad_norm": 0.7186133861541748,
      "learning_rate": 5.6583361521646255e-06,
      "loss": 0.7223,
      "step": 16459
    },
    {
      "epoch": 0.47426957874719067,
      "grad_norm": 0.7582612037658691,
      "learning_rate": 5.657873601493443e-06,
      "loss": 0.7313,
      "step": 16460
    },
    {
      "epoch": 0.47429839220883996,
      "grad_norm": 0.7797325849533081,
      "learning_rate": 5.657411045092839e-06,
      "loss": 0.7346,
      "step": 16461
    },
    {
      "epoch": 0.47432720567048925,
      "grad_norm": 0.7455533742904663,
      "learning_rate": 5.656948482966842e-06,
      "loss": 0.7326,
      "step": 16462
    },
    {
      "epoch": 0.47435601913213854,
      "grad_norm": 0.711240291595459,
      "learning_rate": 5.6564859151194806e-06,
      "loss": 0.7194,
      "step": 16463
    },
    {
      "epoch": 0.47438483259378783,
      "grad_norm": 0.7451373934745789,
      "learning_rate": 5.656023341554784e-06,
      "loss": 0.7322,
      "step": 16464
    },
    {
      "epoch": 0.4744136460554371,
      "grad_norm": 0.7279585599899292,
      "learning_rate": 5.655560762276779e-06,
      "loss": 0.7131,
      "step": 16465
    },
    {
      "epoch": 0.47444245951708636,
      "grad_norm": 0.731185257434845,
      "learning_rate": 5.655098177289496e-06,
      "loss": 0.6864,
      "step": 16466
    },
    {
      "epoch": 0.47447127297873565,
      "grad_norm": 0.7612107396125793,
      "learning_rate": 5.654635586596962e-06,
      "loss": 0.7343,
      "step": 16467
    },
    {
      "epoch": 0.47450008644038494,
      "grad_norm": 0.7824352383613586,
      "learning_rate": 5.654172990203206e-06,
      "loss": 0.7515,
      "step": 16468
    },
    {
      "epoch": 0.47452889990203423,
      "grad_norm": 0.7367198467254639,
      "learning_rate": 5.653710388112257e-06,
      "loss": 0.7229,
      "step": 16469
    },
    {
      "epoch": 0.4745577133636835,
      "grad_norm": 0.7556087374687195,
      "learning_rate": 5.6532477803281436e-06,
      "loss": 0.7282,
      "step": 16470
    },
    {
      "epoch": 0.4745865268253328,
      "grad_norm": 0.7298765778541565,
      "learning_rate": 5.652785166854897e-06,
      "loss": 0.7212,
      "step": 16471
    },
    {
      "epoch": 0.4746153402869821,
      "grad_norm": 0.7452727556228638,
      "learning_rate": 5.652322547696544e-06,
      "loss": 0.7376,
      "step": 16472
    },
    {
      "epoch": 0.47464415374863134,
      "grad_norm": 0.7516623735427856,
      "learning_rate": 5.651859922857111e-06,
      "loss": 0.7263,
      "step": 16473
    },
    {
      "epoch": 0.4746729672102806,
      "grad_norm": 0.7493687868118286,
      "learning_rate": 5.651397292340632e-06,
      "loss": 0.7398,
      "step": 16474
    },
    {
      "epoch": 0.4747017806719299,
      "grad_norm": 0.7449446320533752,
      "learning_rate": 5.6509346561511334e-06,
      "loss": 0.7229,
      "step": 16475
    },
    {
      "epoch": 0.4747305941335792,
      "grad_norm": 0.7715587019920349,
      "learning_rate": 5.650472014292645e-06,
      "loss": 0.7159,
      "step": 16476
    },
    {
      "epoch": 0.4747594075952285,
      "grad_norm": 0.7392311692237854,
      "learning_rate": 5.650009366769196e-06,
      "loss": 0.706,
      "step": 16477
    },
    {
      "epoch": 0.4747882210568778,
      "grad_norm": 0.7836659550666809,
      "learning_rate": 5.649546713584815e-06,
      "loss": 0.7029,
      "step": 16478
    },
    {
      "epoch": 0.4748170345185271,
      "grad_norm": 0.7924016714096069,
      "learning_rate": 5.649084054743529e-06,
      "loss": 0.7178,
      "step": 16479
    },
    {
      "epoch": 0.4748458479801763,
      "grad_norm": 0.7595053911209106,
      "learning_rate": 5.648621390249371e-06,
      "loss": 0.7139,
      "step": 16480
    },
    {
      "epoch": 0.4748746614418256,
      "grad_norm": 0.7466902136802673,
      "learning_rate": 5.64815872010637e-06,
      "loss": 0.7361,
      "step": 16481
    },
    {
      "epoch": 0.4749034749034749,
      "grad_norm": 0.7545503973960876,
      "learning_rate": 5.6476960443185545e-06,
      "loss": 0.7379,
      "step": 16482
    },
    {
      "epoch": 0.4749322883651242,
      "grad_norm": 0.7991130948066711,
      "learning_rate": 5.647233362889954e-06,
      "loss": 0.734,
      "step": 16483
    },
    {
      "epoch": 0.4749611018267735,
      "grad_norm": 0.7680005431175232,
      "learning_rate": 5.646770675824598e-06,
      "loss": 0.7442,
      "step": 16484
    },
    {
      "epoch": 0.47498991528842277,
      "grad_norm": 0.7543721795082092,
      "learning_rate": 5.646307983126514e-06,
      "loss": 0.7306,
      "step": 16485
    },
    {
      "epoch": 0.47501872875007206,
      "grad_norm": 0.7356469035148621,
      "learning_rate": 5.645845284799736e-06,
      "loss": 0.6932,
      "step": 16486
    },
    {
      "epoch": 0.4750475422117213,
      "grad_norm": 0.7508165240287781,
      "learning_rate": 5.64538258084829e-06,
      "loss": 0.7396,
      "step": 16487
    },
    {
      "epoch": 0.4750763556733706,
      "grad_norm": 0.7462043166160583,
      "learning_rate": 5.644919871276205e-06,
      "loss": 0.7112,
      "step": 16488
    },
    {
      "epoch": 0.4751051691350199,
      "grad_norm": 0.7051938772201538,
      "learning_rate": 5.644457156087515e-06,
      "loss": 0.7302,
      "step": 16489
    },
    {
      "epoch": 0.47513398259666917,
      "grad_norm": 0.7573407888412476,
      "learning_rate": 5.643994435286247e-06,
      "loss": 0.7075,
      "step": 16490
    },
    {
      "epoch": 0.47516279605831846,
      "grad_norm": 0.6956053376197815,
      "learning_rate": 5.643531708876431e-06,
      "loss": 0.7096,
      "step": 16491
    },
    {
      "epoch": 0.47519160951996775,
      "grad_norm": 0.7417377233505249,
      "learning_rate": 5.643068976862097e-06,
      "loss": 0.7094,
      "step": 16492
    },
    {
      "epoch": 0.475220422981617,
      "grad_norm": 0.7377133965492249,
      "learning_rate": 5.6426062392472754e-06,
      "loss": 0.7644,
      "step": 16493
    },
    {
      "epoch": 0.4752492364432663,
      "grad_norm": 0.7136525511741638,
      "learning_rate": 5.642143496035996e-06,
      "loss": 0.7206,
      "step": 16494
    },
    {
      "epoch": 0.47527804990491557,
      "grad_norm": 0.7249821424484253,
      "learning_rate": 5.641680747232289e-06,
      "loss": 0.7086,
      "step": 16495
    },
    {
      "epoch": 0.47530686336656486,
      "grad_norm": 0.7116751670837402,
      "learning_rate": 5.641217992840183e-06,
      "loss": 0.7316,
      "step": 16496
    },
    {
      "epoch": 0.47533567682821415,
      "grad_norm": 0.759253978729248,
      "learning_rate": 5.640755232863709e-06,
      "loss": 0.7165,
      "step": 16497
    },
    {
      "epoch": 0.47536449028986344,
      "grad_norm": 0.7267736196517944,
      "learning_rate": 5.6402924673068994e-06,
      "loss": 0.7101,
      "step": 16498
    },
    {
      "epoch": 0.47539330375151273,
      "grad_norm": 0.7328212857246399,
      "learning_rate": 5.639829696173781e-06,
      "loss": 0.7001,
      "step": 16499
    },
    {
      "epoch": 0.47542211721316197,
      "grad_norm": 0.7352029085159302,
      "learning_rate": 5.639366919468386e-06,
      "loss": 0.7182,
      "step": 16500
    },
    {
      "epoch": 0.47545093067481126,
      "grad_norm": 0.7251425981521606,
      "learning_rate": 5.638904137194745e-06,
      "loss": 0.7279,
      "step": 16501
    },
    {
      "epoch": 0.47547974413646055,
      "grad_norm": 0.7592728734016418,
      "learning_rate": 5.638441349356887e-06,
      "loss": 0.7266,
      "step": 16502
    },
    {
      "epoch": 0.47550855759810984,
      "grad_norm": 0.7368021011352539,
      "learning_rate": 5.637978555958841e-06,
      "loss": 0.7367,
      "step": 16503
    },
    {
      "epoch": 0.47553737105975913,
      "grad_norm": 0.7249210476875305,
      "learning_rate": 5.637515757004641e-06,
      "loss": 0.7148,
      "step": 16504
    },
    {
      "epoch": 0.4755661845214084,
      "grad_norm": 0.7447962164878845,
      "learning_rate": 5.637052952498317e-06,
      "loss": 0.7331,
      "step": 16505
    },
    {
      "epoch": 0.4755949979830577,
      "grad_norm": 0.7341548204421997,
      "learning_rate": 5.636590142443898e-06,
      "loss": 0.7346,
      "step": 16506
    },
    {
      "epoch": 0.47562381144470695,
      "grad_norm": 0.7168158888816833,
      "learning_rate": 5.6361273268454155e-06,
      "loss": 0.7234,
      "step": 16507
    },
    {
      "epoch": 0.47565262490635624,
      "grad_norm": 0.7338623404502869,
      "learning_rate": 5.635664505706899e-06,
      "loss": 0.701,
      "step": 16508
    },
    {
      "epoch": 0.4756814383680055,
      "grad_norm": 0.7103898525238037,
      "learning_rate": 5.63520167903238e-06,
      "loss": 0.6994,
      "step": 16509
    },
    {
      "epoch": 0.4757102518296548,
      "grad_norm": 0.7290337085723877,
      "learning_rate": 5.63473884682589e-06,
      "loss": 0.7208,
      "step": 16510
    },
    {
      "epoch": 0.4757390652913041,
      "grad_norm": 0.7671703696250916,
      "learning_rate": 5.634276009091459e-06,
      "loss": 0.7365,
      "step": 16511
    },
    {
      "epoch": 0.4757678787529534,
      "grad_norm": 0.7289111614227295,
      "learning_rate": 5.633813165833118e-06,
      "loss": 0.7335,
      "step": 16512
    },
    {
      "epoch": 0.4757966922146027,
      "grad_norm": 0.7521519064903259,
      "learning_rate": 5.633350317054899e-06,
      "loss": 0.7405,
      "step": 16513
    },
    {
      "epoch": 0.4758255056762519,
      "grad_norm": 0.748167097568512,
      "learning_rate": 5.632887462760831e-06,
      "loss": 0.7561,
      "step": 16514
    },
    {
      "epoch": 0.4758543191379012,
      "grad_norm": 0.7779188752174377,
      "learning_rate": 5.632424602954945e-06,
      "loss": 0.707,
      "step": 16515
    },
    {
      "epoch": 0.4758831325995505,
      "grad_norm": 0.7312152981758118,
      "learning_rate": 5.6319617376412734e-06,
      "loss": 0.7182,
      "step": 16516
    },
    {
      "epoch": 0.4759119460611998,
      "grad_norm": 0.7262446880340576,
      "learning_rate": 5.631498866823848e-06,
      "loss": 0.7093,
      "step": 16517
    },
    {
      "epoch": 0.4759407595228491,
      "grad_norm": 0.7492262721061707,
      "learning_rate": 5.631035990506698e-06,
      "loss": 0.747,
      "step": 16518
    },
    {
      "epoch": 0.4759695729844984,
      "grad_norm": 0.8484588265419006,
      "learning_rate": 5.630573108693856e-06,
      "loss": 0.7618,
      "step": 16519
    },
    {
      "epoch": 0.4759983864461476,
      "grad_norm": 0.764362633228302,
      "learning_rate": 5.630110221389351e-06,
      "loss": 0.7252,
      "step": 16520
    },
    {
      "epoch": 0.4760271999077969,
      "grad_norm": 0.727074921131134,
      "learning_rate": 5.629647328597215e-06,
      "loss": 0.7308,
      "step": 16521
    },
    {
      "epoch": 0.4760560133694462,
      "grad_norm": 0.7204325795173645,
      "learning_rate": 5.629184430321483e-06,
      "loss": 0.7256,
      "step": 16522
    },
    {
      "epoch": 0.4760848268310955,
      "grad_norm": 0.7420254349708557,
      "learning_rate": 5.628721526566182e-06,
      "loss": 0.7251,
      "step": 16523
    },
    {
      "epoch": 0.4761136402927448,
      "grad_norm": 0.7484375834465027,
      "learning_rate": 5.628258617335346e-06,
      "loss": 0.7277,
      "step": 16524
    },
    {
      "epoch": 0.47614245375439407,
      "grad_norm": 0.7244096398353577,
      "learning_rate": 5.627795702633004e-06,
      "loss": 0.7416,
      "step": 16525
    },
    {
      "epoch": 0.47617126721604336,
      "grad_norm": 0.7145223617553711,
      "learning_rate": 5.62733278246319e-06,
      "loss": 0.7182,
      "step": 16526
    },
    {
      "epoch": 0.4762000806776926,
      "grad_norm": 0.7347941398620605,
      "learning_rate": 5.626869856829934e-06,
      "loss": 0.7075,
      "step": 16527
    },
    {
      "epoch": 0.4762288941393419,
      "grad_norm": 0.7396879196166992,
      "learning_rate": 5.626406925737268e-06,
      "loss": 0.7019,
      "step": 16528
    },
    {
      "epoch": 0.4762577076009912,
      "grad_norm": 0.7248403429985046,
      "learning_rate": 5.625943989189225e-06,
      "loss": 0.7212,
      "step": 16529
    },
    {
      "epoch": 0.47628652106264047,
      "grad_norm": 0.7663093209266663,
      "learning_rate": 5.625481047189835e-06,
      "loss": 0.7306,
      "step": 16530
    },
    {
      "epoch": 0.47631533452428976,
      "grad_norm": 0.7490155100822449,
      "learning_rate": 5.62501809974313e-06,
      "loss": 0.7217,
      "step": 16531
    },
    {
      "epoch": 0.47634414798593905,
      "grad_norm": 0.7473853826522827,
      "learning_rate": 5.624555146853143e-06,
      "loss": 0.7487,
      "step": 16532
    },
    {
      "epoch": 0.47637296144758834,
      "grad_norm": 0.7356196641921997,
      "learning_rate": 5.624092188523903e-06,
      "loss": 0.7277,
      "step": 16533
    },
    {
      "epoch": 0.4764017749092376,
      "grad_norm": 0.7206714749336243,
      "learning_rate": 5.623629224759446e-06,
      "loss": 0.6927,
      "step": 16534
    },
    {
      "epoch": 0.47643058837088686,
      "grad_norm": 0.7368496656417847,
      "learning_rate": 5.623166255563801e-06,
      "loss": 0.7397,
      "step": 16535
    },
    {
      "epoch": 0.47645940183253616,
      "grad_norm": 0.7245197296142578,
      "learning_rate": 5.622703280941001e-06,
      "loss": 0.7471,
      "step": 16536
    },
    {
      "epoch": 0.47648821529418545,
      "grad_norm": 0.7129507064819336,
      "learning_rate": 5.6222403008950786e-06,
      "loss": 0.7351,
      "step": 16537
    },
    {
      "epoch": 0.47651702875583474,
      "grad_norm": 0.73552006483078,
      "learning_rate": 5.621777315430065e-06,
      "loss": 0.7351,
      "step": 16538
    },
    {
      "epoch": 0.47654584221748403,
      "grad_norm": 0.7363921403884888,
      "learning_rate": 5.62131432454999e-06,
      "loss": 0.7131,
      "step": 16539
    },
    {
      "epoch": 0.47657465567913326,
      "grad_norm": 0.7529151439666748,
      "learning_rate": 5.620851328258892e-06,
      "loss": 0.7239,
      "step": 16540
    },
    {
      "epoch": 0.47660346914078255,
      "grad_norm": 0.7405869364738464,
      "learning_rate": 5.6203883265607974e-06,
      "loss": 0.7117,
      "step": 16541
    },
    {
      "epoch": 0.47663228260243184,
      "grad_norm": 0.7434303164482117,
      "learning_rate": 5.619925319459742e-06,
      "loss": 0.7165,
      "step": 16542
    },
    {
      "epoch": 0.47666109606408114,
      "grad_norm": 0.7298169732093811,
      "learning_rate": 5.619462306959756e-06,
      "loss": 0.7043,
      "step": 16543
    },
    {
      "epoch": 0.4766899095257304,
      "grad_norm": 0.7409364581108093,
      "learning_rate": 5.618999289064873e-06,
      "loss": 0.7114,
      "step": 16544
    },
    {
      "epoch": 0.4767187229873797,
      "grad_norm": 0.7461989521980286,
      "learning_rate": 5.618536265779122e-06,
      "loss": 0.7257,
      "step": 16545
    },
    {
      "epoch": 0.476747536449029,
      "grad_norm": 0.7529208064079285,
      "learning_rate": 5.618073237106541e-06,
      "loss": 0.7305,
      "step": 16546
    },
    {
      "epoch": 0.47677634991067824,
      "grad_norm": 0.725124716758728,
      "learning_rate": 5.617610203051159e-06,
      "loss": 0.703,
      "step": 16547
    },
    {
      "epoch": 0.47680516337232753,
      "grad_norm": 0.7394850850105286,
      "learning_rate": 5.617147163617012e-06,
      "loss": 0.7339,
      "step": 16548
    },
    {
      "epoch": 0.4768339768339768,
      "grad_norm": 0.7223778367042542,
      "learning_rate": 5.616684118808128e-06,
      "loss": 0.7377,
      "step": 16549
    },
    {
      "epoch": 0.4768627902956261,
      "grad_norm": 0.7893099188804626,
      "learning_rate": 5.616221068628542e-06,
      "loss": 0.7361,
      "step": 16550
    },
    {
      "epoch": 0.4768916037572754,
      "grad_norm": 0.7347427606582642,
      "learning_rate": 5.6157580130822864e-06,
      "loss": 0.7105,
      "step": 16551
    },
    {
      "epoch": 0.4769204172189247,
      "grad_norm": 0.7364019751548767,
      "learning_rate": 5.615294952173395e-06,
      "loss": 0.6832,
      "step": 16552
    },
    {
      "epoch": 0.476949230680574,
      "grad_norm": 0.7213823795318604,
      "learning_rate": 5.614831885905898e-06,
      "loss": 0.7154,
      "step": 16553
    },
    {
      "epoch": 0.4769780441422232,
      "grad_norm": 0.722463071346283,
      "learning_rate": 5.614368814283831e-06,
      "loss": 0.7039,
      "step": 16554
    },
    {
      "epoch": 0.4770068576038725,
      "grad_norm": 0.7518752217292786,
      "learning_rate": 5.613905737311226e-06,
      "loss": 0.718,
      "step": 16555
    },
    {
      "epoch": 0.4770356710655218,
      "grad_norm": 0.7145728468894958,
      "learning_rate": 5.613442654992116e-06,
      "loss": 0.7177,
      "step": 16556
    },
    {
      "epoch": 0.4770644845271711,
      "grad_norm": 0.7518794536590576,
      "learning_rate": 5.612979567330532e-06,
      "loss": 0.6986,
      "step": 16557
    },
    {
      "epoch": 0.4770932979888204,
      "grad_norm": 0.7539689540863037,
      "learning_rate": 5.612516474330511e-06,
      "loss": 0.6856,
      "step": 16558
    },
    {
      "epoch": 0.4771221114504697,
      "grad_norm": 0.7467823624610901,
      "learning_rate": 5.612053375996082e-06,
      "loss": 0.7342,
      "step": 16559
    },
    {
      "epoch": 0.47715092491211897,
      "grad_norm": 0.7366397976875305,
      "learning_rate": 5.6115902723312806e-06,
      "loss": 0.7113,
      "step": 16560
    },
    {
      "epoch": 0.4771797383737682,
      "grad_norm": 0.7426847815513611,
      "learning_rate": 5.6111271633401395e-06,
      "loss": 0.699,
      "step": 16561
    },
    {
      "epoch": 0.4772085518354175,
      "grad_norm": 0.731723964214325,
      "learning_rate": 5.610664049026691e-06,
      "loss": 0.7116,
      "step": 16562
    },
    {
      "epoch": 0.4772373652970668,
      "grad_norm": 0.7379777431488037,
      "learning_rate": 5.6102009293949685e-06,
      "loss": 0.73,
      "step": 16563
    },
    {
      "epoch": 0.4772661787587161,
      "grad_norm": 0.7257352471351624,
      "learning_rate": 5.609737804449008e-06,
      "loss": 0.7093,
      "step": 16564
    },
    {
      "epoch": 0.47729499222036537,
      "grad_norm": 0.7277331352233887,
      "learning_rate": 5.60927467419284e-06,
      "loss": 0.7141,
      "step": 16565
    },
    {
      "epoch": 0.47732380568201466,
      "grad_norm": 0.724418044090271,
      "learning_rate": 5.608811538630496e-06,
      "loss": 0.7138,
      "step": 16566
    },
    {
      "epoch": 0.4773526191436639,
      "grad_norm": 0.7234103679656982,
      "learning_rate": 5.608348397766016e-06,
      "loss": 0.7169,
      "step": 16567
    },
    {
      "epoch": 0.4773814326053132,
      "grad_norm": 0.7170481085777283,
      "learning_rate": 5.607885251603428e-06,
      "loss": 0.7041,
      "step": 16568
    },
    {
      "epoch": 0.4774102460669625,
      "grad_norm": 0.7151598930358887,
      "learning_rate": 5.607422100146765e-06,
      "loss": 0.7094,
      "step": 16569
    },
    {
      "epoch": 0.47743905952861176,
      "grad_norm": 0.7043372988700867,
      "learning_rate": 5.606958943400066e-06,
      "loss": 0.7254,
      "step": 16570
    },
    {
      "epoch": 0.47746787299026106,
      "grad_norm": 0.7155861854553223,
      "learning_rate": 5.60649578136736e-06,
      "loss": 0.7174,
      "step": 16571
    },
    {
      "epoch": 0.47749668645191035,
      "grad_norm": 0.7264436483383179,
      "learning_rate": 5.606032614052682e-06,
      "loss": 0.72,
      "step": 16572
    },
    {
      "epoch": 0.47752549991355964,
      "grad_norm": 0.7329257130622864,
      "learning_rate": 5.605569441460066e-06,
      "loss": 0.6949,
      "step": 16573
    },
    {
      "epoch": 0.47755431337520887,
      "grad_norm": 0.7349892854690552,
      "learning_rate": 5.605106263593545e-06,
      "loss": 0.7459,
      "step": 16574
    },
    {
      "epoch": 0.47758312683685816,
      "grad_norm": 0.7170813083648682,
      "learning_rate": 5.604643080457154e-06,
      "loss": 0.7122,
      "step": 16575
    },
    {
      "epoch": 0.47761194029850745,
      "grad_norm": 0.7428305745124817,
      "learning_rate": 5.604179892054926e-06,
      "loss": 0.7283,
      "step": 16576
    },
    {
      "epoch": 0.47764075376015674,
      "grad_norm": 0.7260899543762207,
      "learning_rate": 5.603716698390895e-06,
      "loss": 0.7561,
      "step": 16577
    },
    {
      "epoch": 0.47766956722180604,
      "grad_norm": 0.7327352166175842,
      "learning_rate": 5.603253499469095e-06,
      "loss": 0.7379,
      "step": 16578
    },
    {
      "epoch": 0.4776983806834553,
      "grad_norm": 0.7524497509002686,
      "learning_rate": 5.602790295293561e-06,
      "loss": 0.748,
      "step": 16579
    },
    {
      "epoch": 0.4777271941451046,
      "grad_norm": 0.7892935276031494,
      "learning_rate": 5.6023270858683236e-06,
      "loss": 0.6936,
      "step": 16580
    },
    {
      "epoch": 0.47775600760675385,
      "grad_norm": 0.7510440349578857,
      "learning_rate": 5.601863871197422e-06,
      "loss": 0.7151,
      "step": 16581
    },
    {
      "epoch": 0.47778482106840314,
      "grad_norm": 0.7095027565956116,
      "learning_rate": 5.601400651284886e-06,
      "loss": 0.7104,
      "step": 16582
    },
    {
      "epoch": 0.47781363453005243,
      "grad_norm": 0.7653497457504272,
      "learning_rate": 5.600937426134753e-06,
      "loss": 0.733,
      "step": 16583
    },
    {
      "epoch": 0.4778424479917017,
      "grad_norm": 0.7522735595703125,
      "learning_rate": 5.6004741957510545e-06,
      "loss": 0.7082,
      "step": 16584
    },
    {
      "epoch": 0.477871261453351,
      "grad_norm": 0.7708958387374878,
      "learning_rate": 5.600010960137826e-06,
      "loss": 0.7344,
      "step": 16585
    },
    {
      "epoch": 0.4779000749150003,
      "grad_norm": 0.7524880766868591,
      "learning_rate": 5.599547719299102e-06,
      "loss": 0.6918,
      "step": 16586
    },
    {
      "epoch": 0.4779288883766496,
      "grad_norm": 0.7385783195495605,
      "learning_rate": 5.599084473238917e-06,
      "loss": 0.7141,
      "step": 16587
    },
    {
      "epoch": 0.47795770183829883,
      "grad_norm": 0.7133866548538208,
      "learning_rate": 5.598621221961304e-06,
      "loss": 0.7165,
      "step": 16588
    },
    {
      "epoch": 0.4779865152999481,
      "grad_norm": 0.7485241293907166,
      "learning_rate": 5.5981579654703e-06,
      "loss": 0.7241,
      "step": 16589
    },
    {
      "epoch": 0.4780153287615974,
      "grad_norm": 0.7483022809028625,
      "learning_rate": 5.597694703769938e-06,
      "loss": 0.7515,
      "step": 16590
    },
    {
      "epoch": 0.4780441422232467,
      "grad_norm": 0.7001188397407532,
      "learning_rate": 5.597231436864254e-06,
      "loss": 0.6962,
      "step": 16591
    },
    {
      "epoch": 0.478072955684896,
      "grad_norm": 0.7502022981643677,
      "learning_rate": 5.596768164757277e-06,
      "loss": 0.7642,
      "step": 16592
    },
    {
      "epoch": 0.4781017691465453,
      "grad_norm": 0.7514210939407349,
      "learning_rate": 5.596304887453049e-06,
      "loss": 0.7101,
      "step": 16593
    },
    {
      "epoch": 0.4781305826081945,
      "grad_norm": 0.7310863733291626,
      "learning_rate": 5.595841604955601e-06,
      "loss": 0.7284,
      "step": 16594
    },
    {
      "epoch": 0.4781593960698438,
      "grad_norm": 0.7367274165153503,
      "learning_rate": 5.595378317268968e-06,
      "loss": 0.7491,
      "step": 16595
    },
    {
      "epoch": 0.4781882095314931,
      "grad_norm": 0.7307531237602234,
      "learning_rate": 5.594915024397185e-06,
      "loss": 0.7482,
      "step": 16596
    },
    {
      "epoch": 0.4782170229931424,
      "grad_norm": 0.7537792921066284,
      "learning_rate": 5.5944517263442875e-06,
      "loss": 0.7131,
      "step": 16597
    },
    {
      "epoch": 0.4782458364547917,
      "grad_norm": 0.7583703994750977,
      "learning_rate": 5.593988423114308e-06,
      "loss": 0.7199,
      "step": 16598
    },
    {
      "epoch": 0.478274649916441,
      "grad_norm": 0.7496845722198486,
      "learning_rate": 5.593525114711284e-06,
      "loss": 0.7302,
      "step": 16599
    },
    {
      "epoch": 0.47830346337809027,
      "grad_norm": 0.7331599593162537,
      "learning_rate": 5.593061801139251e-06,
      "loss": 0.7024,
      "step": 16600
    },
    {
      "epoch": 0.4783322768397395,
      "grad_norm": 0.7440875768661499,
      "learning_rate": 5.592598482402241e-06,
      "loss": 0.715,
      "step": 16601
    },
    {
      "epoch": 0.4783610903013888,
      "grad_norm": 0.7670915722846985,
      "learning_rate": 5.592135158504291e-06,
      "loss": 0.718,
      "step": 16602
    },
    {
      "epoch": 0.4783899037630381,
      "grad_norm": 0.7481613159179688,
      "learning_rate": 5.591671829449437e-06,
      "loss": 0.7368,
      "step": 16603
    },
    {
      "epoch": 0.4784187172246874,
      "grad_norm": 0.7444038391113281,
      "learning_rate": 5.59120849524171e-06,
      "loss": 0.7042,
      "step": 16604
    },
    {
      "epoch": 0.47844753068633666,
      "grad_norm": 0.7373824715614319,
      "learning_rate": 5.59074515588515e-06,
      "loss": 0.7271,
      "step": 16605
    },
    {
      "epoch": 0.47847634414798595,
      "grad_norm": 0.7648279070854187,
      "learning_rate": 5.59028181138379e-06,
      "loss": 0.7406,
      "step": 16606
    },
    {
      "epoch": 0.47850515760963525,
      "grad_norm": 0.7572075128555298,
      "learning_rate": 5.589818461741664e-06,
      "loss": 0.7334,
      "step": 16607
    },
    {
      "epoch": 0.4785339710712845,
      "grad_norm": 0.7542798519134521,
      "learning_rate": 5.58935510696281e-06,
      "loss": 0.712,
      "step": 16608
    },
    {
      "epoch": 0.47856278453293377,
      "grad_norm": 0.7770881652832031,
      "learning_rate": 5.588891747051263e-06,
      "loss": 0.6991,
      "step": 16609
    },
    {
      "epoch": 0.47859159799458306,
      "grad_norm": 0.7751967310905457,
      "learning_rate": 5.588428382011055e-06,
      "loss": 0.7312,
      "step": 16610
    },
    {
      "epoch": 0.47862041145623235,
      "grad_norm": 0.7363452315330505,
      "learning_rate": 5.587965011846226e-06,
      "loss": 0.7216,
      "step": 16611
    },
    {
      "epoch": 0.47864922491788164,
      "grad_norm": 0.6986455917358398,
      "learning_rate": 5.587501636560809e-06,
      "loss": 0.7093,
      "step": 16612
    },
    {
      "epoch": 0.47867803837953093,
      "grad_norm": 0.7323511838912964,
      "learning_rate": 5.587038256158841e-06,
      "loss": 0.6966,
      "step": 16613
    },
    {
      "epoch": 0.4787068518411802,
      "grad_norm": 0.7672431468963623,
      "learning_rate": 5.5865748706443554e-06,
      "loss": 0.702,
      "step": 16614
    },
    {
      "epoch": 0.47873566530282946,
      "grad_norm": 0.7333734631538391,
      "learning_rate": 5.58611148002139e-06,
      "loss": 0.714,
      "step": 16615
    },
    {
      "epoch": 0.47876447876447875,
      "grad_norm": 0.7236137390136719,
      "learning_rate": 5.585648084293979e-06,
      "loss": 0.7275,
      "step": 16616
    },
    {
      "epoch": 0.47879329222612804,
      "grad_norm": 0.7106077075004578,
      "learning_rate": 5.585184683466159e-06,
      "loss": 0.7361,
      "step": 16617
    },
    {
      "epoch": 0.47882210568777733,
      "grad_norm": 0.7497400641441345,
      "learning_rate": 5.584721277541965e-06,
      "loss": 0.7216,
      "step": 16618
    },
    {
      "epoch": 0.4788509191494266,
      "grad_norm": 0.729786217212677,
      "learning_rate": 5.5842578665254335e-06,
      "loss": 0.7123,
      "step": 16619
    },
    {
      "epoch": 0.4788797326110759,
      "grad_norm": 0.748108983039856,
      "learning_rate": 5.583794450420599e-06,
      "loss": 0.7189,
      "step": 16620
    },
    {
      "epoch": 0.47890854607272515,
      "grad_norm": 0.7344277501106262,
      "learning_rate": 5.5833310292315e-06,
      "loss": 0.7006,
      "step": 16621
    },
    {
      "epoch": 0.47893735953437444,
      "grad_norm": 0.777594804763794,
      "learning_rate": 5.582867602962169e-06,
      "loss": 0.7075,
      "step": 16622
    },
    {
      "epoch": 0.47896617299602373,
      "grad_norm": 0.7598395347595215,
      "learning_rate": 5.582404171616644e-06,
      "loss": 0.7224,
      "step": 16623
    },
    {
      "epoch": 0.478994986457673,
      "grad_norm": 0.7745963335037231,
      "learning_rate": 5.581940735198963e-06,
      "loss": 0.7337,
      "step": 16624
    },
    {
      "epoch": 0.4790237999193223,
      "grad_norm": 0.7637213468551636,
      "learning_rate": 5.581477293713158e-06,
      "loss": 0.7238,
      "step": 16625
    },
    {
      "epoch": 0.4790526133809716,
      "grad_norm": 0.7412738800048828,
      "learning_rate": 5.581013847163267e-06,
      "loss": 0.7263,
      "step": 16626
    },
    {
      "epoch": 0.4790814268426209,
      "grad_norm": 0.7714157104492188,
      "learning_rate": 5.580550395553327e-06,
      "loss": 0.7332,
      "step": 16627
    },
    {
      "epoch": 0.47911024030427013,
      "grad_norm": 0.7595885396003723,
      "learning_rate": 5.580086938887372e-06,
      "loss": 0.7083,
      "step": 16628
    },
    {
      "epoch": 0.4791390537659194,
      "grad_norm": 0.7566149830818176,
      "learning_rate": 5.579623477169442e-06,
      "loss": 0.7207,
      "step": 16629
    },
    {
      "epoch": 0.4791678672275687,
      "grad_norm": 0.7481680512428284,
      "learning_rate": 5.579160010403569e-06,
      "loss": 0.7392,
      "step": 16630
    },
    {
      "epoch": 0.479196680689218,
      "grad_norm": 0.7387374639511108,
      "learning_rate": 5.578696538593793e-06,
      "loss": 0.7134,
      "step": 16631
    },
    {
      "epoch": 0.4792254941508673,
      "grad_norm": 0.7337632775306702,
      "learning_rate": 5.578233061744147e-06,
      "loss": 0.6914,
      "step": 16632
    },
    {
      "epoch": 0.4792543076125166,
      "grad_norm": 0.7490754723548889,
      "learning_rate": 5.57776957985867e-06,
      "loss": 0.7301,
      "step": 16633
    },
    {
      "epoch": 0.4792831210741659,
      "grad_norm": 0.7589907050132751,
      "learning_rate": 5.577306092941397e-06,
      "loss": 0.7172,
      "step": 16634
    },
    {
      "epoch": 0.4793119345358151,
      "grad_norm": 0.7715509533882141,
      "learning_rate": 5.576842600996365e-06,
      "loss": 0.7221,
      "step": 16635
    },
    {
      "epoch": 0.4793407479974644,
      "grad_norm": 0.7644603848457336,
      "learning_rate": 5.576379104027612e-06,
      "loss": 0.7147,
      "step": 16636
    },
    {
      "epoch": 0.4793695614591137,
      "grad_norm": 0.7563074827194214,
      "learning_rate": 5.575915602039172e-06,
      "loss": 0.7377,
      "step": 16637
    },
    {
      "epoch": 0.479398374920763,
      "grad_norm": 0.7529577612876892,
      "learning_rate": 5.575452095035082e-06,
      "loss": 0.7382,
      "step": 16638
    },
    {
      "epoch": 0.4794271883824123,
      "grad_norm": 0.7530434727668762,
      "learning_rate": 5.574988583019381e-06,
      "loss": 0.7302,
      "step": 16639
    },
    {
      "epoch": 0.47945600184406156,
      "grad_norm": 0.7980999946594238,
      "learning_rate": 5.574525065996103e-06,
      "loss": 0.7241,
      "step": 16640
    },
    {
      "epoch": 0.47948481530571085,
      "grad_norm": 0.7607664465904236,
      "learning_rate": 5.574061543969286e-06,
      "loss": 0.7207,
      "step": 16641
    },
    {
      "epoch": 0.4795136287673601,
      "grad_norm": 0.763247549533844,
      "learning_rate": 5.573598016942968e-06,
      "loss": 0.7234,
      "step": 16642
    },
    {
      "epoch": 0.4795424422290094,
      "grad_norm": 0.776050865650177,
      "learning_rate": 5.573134484921184e-06,
      "loss": 0.7226,
      "step": 16643
    },
    {
      "epoch": 0.47957125569065867,
      "grad_norm": 0.7545084953308105,
      "learning_rate": 5.5726709479079725e-06,
      "loss": 0.7073,
      "step": 16644
    },
    {
      "epoch": 0.47960006915230796,
      "grad_norm": 0.733099639415741,
      "learning_rate": 5.572207405907367e-06,
      "loss": 0.7305,
      "step": 16645
    },
    {
      "epoch": 0.47962888261395725,
      "grad_norm": 0.741303026676178,
      "learning_rate": 5.571743858923408e-06,
      "loss": 0.7007,
      "step": 16646
    },
    {
      "epoch": 0.47965769607560654,
      "grad_norm": 0.7260604500770569,
      "learning_rate": 5.571280306960132e-06,
      "loss": 0.6994,
      "step": 16647
    },
    {
      "epoch": 0.4796865095372558,
      "grad_norm": 0.7559642195701599,
      "learning_rate": 5.570816750021576e-06,
      "loss": 0.7134,
      "step": 16648
    },
    {
      "epoch": 0.47971532299890507,
      "grad_norm": 0.7211596369743347,
      "learning_rate": 5.570353188111775e-06,
      "loss": 0.7042,
      "step": 16649
    },
    {
      "epoch": 0.47974413646055436,
      "grad_norm": 0.7304192781448364,
      "learning_rate": 5.569889621234771e-06,
      "loss": 0.7202,
      "step": 16650
    },
    {
      "epoch": 0.47977294992220365,
      "grad_norm": 0.7118525505065918,
      "learning_rate": 5.569426049394597e-06,
      "loss": 0.6909,
      "step": 16651
    },
    {
      "epoch": 0.47980176338385294,
      "grad_norm": 0.7445909380912781,
      "learning_rate": 5.568962472595289e-06,
      "loss": 0.7325,
      "step": 16652
    },
    {
      "epoch": 0.47983057684550223,
      "grad_norm": 0.7416119575500488,
      "learning_rate": 5.568498890840889e-06,
      "loss": 0.7196,
      "step": 16653
    },
    {
      "epoch": 0.4798593903071515,
      "grad_norm": 0.7482998967170715,
      "learning_rate": 5.5680353041354305e-06,
      "loss": 0.7201,
      "step": 16654
    },
    {
      "epoch": 0.47988820376880076,
      "grad_norm": 0.7379928827285767,
      "learning_rate": 5.567571712482953e-06,
      "loss": 0.7105,
      "step": 16655
    },
    {
      "epoch": 0.47991701723045005,
      "grad_norm": 0.7466599345207214,
      "learning_rate": 5.567108115887494e-06,
      "loss": 0.7289,
      "step": 16656
    },
    {
      "epoch": 0.47994583069209934,
      "grad_norm": 0.725746214389801,
      "learning_rate": 5.5666445143530904e-06,
      "loss": 0.7142,
      "step": 16657
    },
    {
      "epoch": 0.47997464415374863,
      "grad_norm": 0.7288442850112915,
      "learning_rate": 5.566180907883777e-06,
      "loss": 0.7208,
      "step": 16658
    },
    {
      "epoch": 0.4800034576153979,
      "grad_norm": 0.7241584062576294,
      "learning_rate": 5.565717296483596e-06,
      "loss": 0.7343,
      "step": 16659
    },
    {
      "epoch": 0.4800322710770472,
      "grad_norm": 0.7261035442352295,
      "learning_rate": 5.565253680156583e-06,
      "loss": 0.7122,
      "step": 16660
    },
    {
      "epoch": 0.4800610845386965,
      "grad_norm": 0.7259412407875061,
      "learning_rate": 5.564790058906775e-06,
      "loss": 0.7068,
      "step": 16661
    },
    {
      "epoch": 0.48008989800034574,
      "grad_norm": 0.7064014077186584,
      "learning_rate": 5.56432643273821e-06,
      "loss": 0.718,
      "step": 16662
    },
    {
      "epoch": 0.48011871146199503,
      "grad_norm": 0.7326989769935608,
      "learning_rate": 5.563862801654925e-06,
      "loss": 0.7432,
      "step": 16663
    },
    {
      "epoch": 0.4801475249236443,
      "grad_norm": 0.7212628722190857,
      "learning_rate": 5.5633991656609595e-06,
      "loss": 0.7104,
      "step": 16664
    },
    {
      "epoch": 0.4801763383852936,
      "grad_norm": 0.7442852258682251,
      "learning_rate": 5.562935524760351e-06,
      "loss": 0.718,
      "step": 16665
    },
    {
      "epoch": 0.4802051518469429,
      "grad_norm": 1.8078420162200928,
      "learning_rate": 5.562471878957135e-06,
      "loss": 0.7424,
      "step": 16666
    },
    {
      "epoch": 0.4802339653085922,
      "grad_norm": 0.7578837871551514,
      "learning_rate": 5.562008228255353e-06,
      "loss": 0.7674,
      "step": 16667
    },
    {
      "epoch": 0.4802627787702415,
      "grad_norm": 0.8270198702812195,
      "learning_rate": 5.561544572659041e-06,
      "loss": 0.7168,
      "step": 16668
    },
    {
      "epoch": 0.4802915922318907,
      "grad_norm": 0.7398512363433838,
      "learning_rate": 5.561080912172237e-06,
      "loss": 0.7077,
      "step": 16669
    },
    {
      "epoch": 0.48032040569354,
      "grad_norm": 0.7506567239761353,
      "learning_rate": 5.560617246798977e-06,
      "loss": 0.7243,
      "step": 16670
    },
    {
      "epoch": 0.4803492191551893,
      "grad_norm": 0.7246556878089905,
      "learning_rate": 5.560153576543304e-06,
      "loss": 0.7229,
      "step": 16671
    },
    {
      "epoch": 0.4803780326168386,
      "grad_norm": 0.7157323956489563,
      "learning_rate": 5.559689901409253e-06,
      "loss": 0.7102,
      "step": 16672
    },
    {
      "epoch": 0.4804068460784879,
      "grad_norm": 0.7742087841033936,
      "learning_rate": 5.559226221400862e-06,
      "loss": 0.7399,
      "step": 16673
    },
    {
      "epoch": 0.48043565954013717,
      "grad_norm": 0.7506381273269653,
      "learning_rate": 5.55876253652217e-06,
      "loss": 0.7128,
      "step": 16674
    },
    {
      "epoch": 0.4804644730017864,
      "grad_norm": 0.7165831327438354,
      "learning_rate": 5.558298846777214e-06,
      "loss": 0.7168,
      "step": 16675
    },
    {
      "epoch": 0.4804932864634357,
      "grad_norm": 0.7465187311172485,
      "learning_rate": 5.5578351521700325e-06,
      "loss": 0.6828,
      "step": 16676
    },
    {
      "epoch": 0.480522099925085,
      "grad_norm": 0.7158063650131226,
      "learning_rate": 5.557371452704665e-06,
      "loss": 0.7097,
      "step": 16677
    },
    {
      "epoch": 0.4805509133867343,
      "grad_norm": 0.7306525707244873,
      "learning_rate": 5.556907748385151e-06,
      "loss": 0.7015,
      "step": 16678
    },
    {
      "epoch": 0.48057972684838357,
      "grad_norm": 0.7581555247306824,
      "learning_rate": 5.556444039215526e-06,
      "loss": 0.721,
      "step": 16679
    },
    {
      "epoch": 0.48060854031003286,
      "grad_norm": 0.7274613380432129,
      "learning_rate": 5.555980325199829e-06,
      "loss": 0.7054,
      "step": 16680
    },
    {
      "epoch": 0.48063735377168215,
      "grad_norm": 0.7675560712814331,
      "learning_rate": 5.5555166063421e-06,
      "loss": 0.737,
      "step": 16681
    },
    {
      "epoch": 0.4806661672333314,
      "grad_norm": 0.760198175907135,
      "learning_rate": 5.5550528826463754e-06,
      "loss": 0.7437,
      "step": 16682
    },
    {
      "epoch": 0.4806949806949807,
      "grad_norm": 0.7182377576828003,
      "learning_rate": 5.554589154116696e-06,
      "loss": 0.7177,
      "step": 16683
    },
    {
      "epoch": 0.48072379415662997,
      "grad_norm": 0.7519062161445618,
      "learning_rate": 5.5541254207571e-06,
      "loss": 0.7158,
      "step": 16684
    },
    {
      "epoch": 0.48075260761827926,
      "grad_norm": 0.7555297017097473,
      "learning_rate": 5.553661682571625e-06,
      "loss": 0.7087,
      "step": 16685
    },
    {
      "epoch": 0.48078142107992855,
      "grad_norm": 0.7217874526977539,
      "learning_rate": 5.55319793956431e-06,
      "loss": 0.7366,
      "step": 16686
    },
    {
      "epoch": 0.48081023454157784,
      "grad_norm": 0.7270641922950745,
      "learning_rate": 5.552734191739194e-06,
      "loss": 0.7146,
      "step": 16687
    },
    {
      "epoch": 0.48083904800322713,
      "grad_norm": 0.7567059993743896,
      "learning_rate": 5.552270439100316e-06,
      "loss": 0.7335,
      "step": 16688
    },
    {
      "epoch": 0.48086786146487637,
      "grad_norm": 0.7324190139770508,
      "learning_rate": 5.551806681651715e-06,
      "loss": 0.6891,
      "step": 16689
    },
    {
      "epoch": 0.48089667492652566,
      "grad_norm": 0.737791121006012,
      "learning_rate": 5.551342919397429e-06,
      "loss": 0.7042,
      "step": 16690
    },
    {
      "epoch": 0.48092548838817495,
      "grad_norm": 0.7626949548721313,
      "learning_rate": 5.550879152341498e-06,
      "loss": 0.7167,
      "step": 16691
    },
    {
      "epoch": 0.48095430184982424,
      "grad_norm": 0.7193794250488281,
      "learning_rate": 5.550415380487958e-06,
      "loss": 0.7271,
      "step": 16692
    },
    {
      "epoch": 0.48098311531147353,
      "grad_norm": 0.7463089227676392,
      "learning_rate": 5.549951603840852e-06,
      "loss": 0.6972,
      "step": 16693
    },
    {
      "epoch": 0.4810119287731228,
      "grad_norm": 0.7326245903968811,
      "learning_rate": 5.549487822404216e-06,
      "loss": 0.7158,
      "step": 16694
    },
    {
      "epoch": 0.4810407422347721,
      "grad_norm": 0.7353671789169312,
      "learning_rate": 5.549024036182091e-06,
      "loss": 0.7313,
      "step": 16695
    },
    {
      "epoch": 0.48106955569642135,
      "grad_norm": 0.7446765303611755,
      "learning_rate": 5.548560245178517e-06,
      "loss": 0.728,
      "step": 16696
    },
    {
      "epoch": 0.48109836915807064,
      "grad_norm": 0.7157720327377319,
      "learning_rate": 5.548096449397531e-06,
      "loss": 0.7249,
      "step": 16697
    },
    {
      "epoch": 0.48112718261971993,
      "grad_norm": 0.71761155128479,
      "learning_rate": 5.547632648843172e-06,
      "loss": 0.696,
      "step": 16698
    },
    {
      "epoch": 0.4811559960813692,
      "grad_norm": 0.7427237033843994,
      "learning_rate": 5.54716884351948e-06,
      "loss": 0.7099,
      "step": 16699
    },
    {
      "epoch": 0.4811848095430185,
      "grad_norm": 0.7394126653671265,
      "learning_rate": 5.546705033430494e-06,
      "loss": 0.7032,
      "step": 16700
    },
    {
      "epoch": 0.4812136230046678,
      "grad_norm": 0.7520869374275208,
      "learning_rate": 5.546241218580255e-06,
      "loss": 0.6996,
      "step": 16701
    },
    {
      "epoch": 0.48124243646631704,
      "grad_norm": 0.7599679827690125,
      "learning_rate": 5.5457773989728e-06,
      "loss": 0.7412,
      "step": 16702
    },
    {
      "epoch": 0.4812712499279663,
      "grad_norm": 0.7520845532417297,
      "learning_rate": 5.54531357461217e-06,
      "loss": 0.7235,
      "step": 16703
    },
    {
      "epoch": 0.4813000633896156,
      "grad_norm": 0.7667508125305176,
      "learning_rate": 5.544849745502403e-06,
      "loss": 0.7289,
      "step": 16704
    },
    {
      "epoch": 0.4813288768512649,
      "grad_norm": 0.7425580620765686,
      "learning_rate": 5.54438591164754e-06,
      "loss": 0.7267,
      "step": 16705
    },
    {
      "epoch": 0.4813576903129142,
      "grad_norm": 0.7138031721115112,
      "learning_rate": 5.543922073051619e-06,
      "loss": 0.7032,
      "step": 16706
    },
    {
      "epoch": 0.4813865037745635,
      "grad_norm": 0.7224751114845276,
      "learning_rate": 5.5434582297186816e-06,
      "loss": 0.7291,
      "step": 16707
    },
    {
      "epoch": 0.4814153172362128,
      "grad_norm": 0.7336099743843079,
      "learning_rate": 5.542994381652767e-06,
      "loss": 0.7068,
      "step": 16708
    },
    {
      "epoch": 0.481444130697862,
      "grad_norm": 0.7735520005226135,
      "learning_rate": 5.542530528857913e-06,
      "loss": 0.7322,
      "step": 16709
    },
    {
      "epoch": 0.4814729441595113,
      "grad_norm": 0.7338825464248657,
      "learning_rate": 5.542066671338161e-06,
      "loss": 0.7337,
      "step": 16710
    },
    {
      "epoch": 0.4815017576211606,
      "grad_norm": 0.735105037689209,
      "learning_rate": 5.541602809097549e-06,
      "loss": 0.7145,
      "step": 16711
    },
    {
      "epoch": 0.4815305710828099,
      "grad_norm": 0.7632430791854858,
      "learning_rate": 5.541138942140118e-06,
      "loss": 0.7425,
      "step": 16712
    },
    {
      "epoch": 0.4815593845444592,
      "grad_norm": 0.7797335982322693,
      "learning_rate": 5.540675070469908e-06,
      "loss": 0.7109,
      "step": 16713
    },
    {
      "epoch": 0.48158819800610847,
      "grad_norm": 0.7736660242080688,
      "learning_rate": 5.54021119409096e-06,
      "loss": 0.7228,
      "step": 16714
    },
    {
      "epoch": 0.48161701146775776,
      "grad_norm": 0.7403555512428284,
      "learning_rate": 5.539747313007312e-06,
      "loss": 0.725,
      "step": 16715
    },
    {
      "epoch": 0.481645824929407,
      "grad_norm": 0.7614259123802185,
      "learning_rate": 5.539283427223005e-06,
      "loss": 0.7069,
      "step": 16716
    },
    {
      "epoch": 0.4816746383910563,
      "grad_norm": 0.7439343929290771,
      "learning_rate": 5.538819536742077e-06,
      "loss": 0.7069,
      "step": 16717
    },
    {
      "epoch": 0.4817034518527056,
      "grad_norm": 0.7487567067146301,
      "learning_rate": 5.538355641568569e-06,
      "loss": 0.7169,
      "step": 16718
    },
    {
      "epoch": 0.48173226531435487,
      "grad_norm": 0.7437763214111328,
      "learning_rate": 5.537891741706522e-06,
      "loss": 0.7254,
      "step": 16719
    },
    {
      "epoch": 0.48176107877600416,
      "grad_norm": 0.7430769205093384,
      "learning_rate": 5.537427837159976e-06,
      "loss": 0.7545,
      "step": 16720
    },
    {
      "epoch": 0.48178989223765345,
      "grad_norm": 0.7402834892272949,
      "learning_rate": 5.536963927932973e-06,
      "loss": 0.7184,
      "step": 16721
    },
    {
      "epoch": 0.48181870569930274,
      "grad_norm": 0.7376667261123657,
      "learning_rate": 5.536500014029548e-06,
      "loss": 0.7166,
      "step": 16722
    },
    {
      "epoch": 0.481847519160952,
      "grad_norm": 0.750857412815094,
      "learning_rate": 5.536036095453744e-06,
      "loss": 0.7177,
      "step": 16723
    },
    {
      "epoch": 0.48187633262260127,
      "grad_norm": 0.7330986857414246,
      "learning_rate": 5.535572172209602e-06,
      "loss": 0.7046,
      "step": 16724
    },
    {
      "epoch": 0.48190514608425056,
      "grad_norm": 0.7800633907318115,
      "learning_rate": 5.535108244301161e-06,
      "loss": 0.7227,
      "step": 16725
    },
    {
      "epoch": 0.48193395954589985,
      "grad_norm": 0.7105046510696411,
      "learning_rate": 5.534644311732465e-06,
      "loss": 0.7081,
      "step": 16726
    },
    {
      "epoch": 0.48196277300754914,
      "grad_norm": 0.7392629981040955,
      "learning_rate": 5.534180374507549e-06,
      "loss": 0.7152,
      "step": 16727
    },
    {
      "epoch": 0.48199158646919843,
      "grad_norm": 0.7155640721321106,
      "learning_rate": 5.533716432630457e-06,
      "loss": 0.7228,
      "step": 16728
    },
    {
      "epoch": 0.48202039993084767,
      "grad_norm": 0.7412738800048828,
      "learning_rate": 5.5332524861052284e-06,
      "loss": 0.7282,
      "step": 16729
    },
    {
      "epoch": 0.48204921339249696,
      "grad_norm": 0.7146337032318115,
      "learning_rate": 5.532788534935903e-06,
      "loss": 0.7211,
      "step": 16730
    },
    {
      "epoch": 0.48207802685414625,
      "grad_norm": 0.7413504123687744,
      "learning_rate": 5.532324579126522e-06,
      "loss": 0.6947,
      "step": 16731
    },
    {
      "epoch": 0.48210684031579554,
      "grad_norm": 0.7137247920036316,
      "learning_rate": 5.5318606186811276e-06,
      "loss": 0.7173,
      "step": 16732
    },
    {
      "epoch": 0.48213565377744483,
      "grad_norm": 0.7294449210166931,
      "learning_rate": 5.531396653603758e-06,
      "loss": 0.7026,
      "step": 16733
    },
    {
      "epoch": 0.4821644672390941,
      "grad_norm": 0.7505067586898804,
      "learning_rate": 5.5309326838984546e-06,
      "loss": 0.7083,
      "step": 16734
    },
    {
      "epoch": 0.4821932807007434,
      "grad_norm": 0.7276819944381714,
      "learning_rate": 5.5304687095692575e-06,
      "loss": 0.7294,
      "step": 16735
    },
    {
      "epoch": 0.48222209416239264,
      "grad_norm": 0.891946017742157,
      "learning_rate": 5.530004730620209e-06,
      "loss": 0.7176,
      "step": 16736
    },
    {
      "epoch": 0.48225090762404194,
      "grad_norm": 0.7658637166023254,
      "learning_rate": 5.529540747055348e-06,
      "loss": 0.7159,
      "step": 16737
    },
    {
      "epoch": 0.4822797210856912,
      "grad_norm": 0.7151915431022644,
      "learning_rate": 5.529076758878719e-06,
      "loss": 0.7225,
      "step": 16738
    },
    {
      "epoch": 0.4823085345473405,
      "grad_norm": 0.7090039849281311,
      "learning_rate": 5.528612766094359e-06,
      "loss": 0.7109,
      "step": 16739
    },
    {
      "epoch": 0.4823373480089898,
      "grad_norm": 0.7412904500961304,
      "learning_rate": 5.52814876870631e-06,
      "loss": 0.7199,
      "step": 16740
    },
    {
      "epoch": 0.4823661614706391,
      "grad_norm": 0.7502892017364502,
      "learning_rate": 5.527684766718613e-06,
      "loss": 0.7242,
      "step": 16741
    },
    {
      "epoch": 0.4823949749322884,
      "grad_norm": 0.7547124624252319,
      "learning_rate": 5.5272207601353086e-06,
      "loss": 0.7235,
      "step": 16742
    },
    {
      "epoch": 0.4824237883939376,
      "grad_norm": 0.7544430494308472,
      "learning_rate": 5.526756748960439e-06,
      "loss": 0.7409,
      "step": 16743
    },
    {
      "epoch": 0.4824526018555869,
      "grad_norm": 0.7307209968566895,
      "learning_rate": 5.5262927331980445e-06,
      "loss": 0.7,
      "step": 16744
    },
    {
      "epoch": 0.4824814153172362,
      "grad_norm": 0.7310968637466431,
      "learning_rate": 5.525828712852167e-06,
      "loss": 0.7067,
      "step": 16745
    },
    {
      "epoch": 0.4825102287788855,
      "grad_norm": 0.7302056550979614,
      "learning_rate": 5.525364687926846e-06,
      "loss": 0.7154,
      "step": 16746
    },
    {
      "epoch": 0.4825390422405348,
      "grad_norm": 0.7315955758094788,
      "learning_rate": 5.524900658426125e-06,
      "loss": 0.7054,
      "step": 16747
    },
    {
      "epoch": 0.4825678557021841,
      "grad_norm": 0.7529934644699097,
      "learning_rate": 5.524436624354043e-06,
      "loss": 0.7525,
      "step": 16748
    },
    {
      "epoch": 0.48259666916383337,
      "grad_norm": 0.7070788145065308,
      "learning_rate": 5.523972585714643e-06,
      "loss": 0.705,
      "step": 16749
    },
    {
      "epoch": 0.4826254826254826,
      "grad_norm": 0.7147388458251953,
      "learning_rate": 5.523508542511965e-06,
      "loss": 0.7073,
      "step": 16750
    },
    {
      "epoch": 0.4826542960871319,
      "grad_norm": 0.7207146286964417,
      "learning_rate": 5.523044494750052e-06,
      "loss": 0.741,
      "step": 16751
    },
    {
      "epoch": 0.4826831095487812,
      "grad_norm": 0.7480258941650391,
      "learning_rate": 5.522580442432943e-06,
      "loss": 0.7479,
      "step": 16752
    },
    {
      "epoch": 0.4827119230104305,
      "grad_norm": 0.7121945023536682,
      "learning_rate": 5.522116385564682e-06,
      "loss": 0.703,
      "step": 16753
    },
    {
      "epoch": 0.48274073647207977,
      "grad_norm": 0.7447675466537476,
      "learning_rate": 5.521652324149307e-06,
      "loss": 0.7361,
      "step": 16754
    },
    {
      "epoch": 0.48276954993372906,
      "grad_norm": 0.7399182319641113,
      "learning_rate": 5.521188258190864e-06,
      "loss": 0.7257,
      "step": 16755
    },
    {
      "epoch": 0.4827983633953783,
      "grad_norm": 0.732476532459259,
      "learning_rate": 5.5207241876933914e-06,
      "loss": 0.7264,
      "step": 16756
    },
    {
      "epoch": 0.4828271768570276,
      "grad_norm": 0.7247036099433899,
      "learning_rate": 5.520260112660933e-06,
      "loss": 0.711,
      "step": 16757
    },
    {
      "epoch": 0.4828559903186769,
      "grad_norm": 0.7406479120254517,
      "learning_rate": 5.519796033097526e-06,
      "loss": 0.7201,
      "step": 16758
    },
    {
      "epoch": 0.48288480378032617,
      "grad_norm": 0.7666088938713074,
      "learning_rate": 5.519331949007218e-06,
      "loss": 0.7341,
      "step": 16759
    },
    {
      "epoch": 0.48291361724197546,
      "grad_norm": 0.7454057335853577,
      "learning_rate": 5.518867860394046e-06,
      "loss": 0.728,
      "step": 16760
    },
    {
      "epoch": 0.48294243070362475,
      "grad_norm": 0.7452501058578491,
      "learning_rate": 5.518403767262055e-06,
      "loss": 0.713,
      "step": 16761
    },
    {
      "epoch": 0.48297124416527404,
      "grad_norm": 0.7247708439826965,
      "learning_rate": 5.517939669615285e-06,
      "loss": 0.6909,
      "step": 16762
    },
    {
      "epoch": 0.4830000576269233,
      "grad_norm": 0.7581302523612976,
      "learning_rate": 5.517475567457777e-06,
      "loss": 0.7228,
      "step": 16763
    },
    {
      "epoch": 0.48302887108857256,
      "grad_norm": 0.8180426955223083,
      "learning_rate": 5.517011460793575e-06,
      "loss": 0.7378,
      "step": 16764
    },
    {
      "epoch": 0.48305768455022186,
      "grad_norm": 0.7149953842163086,
      "learning_rate": 5.516547349626721e-06,
      "loss": 0.7259,
      "step": 16765
    },
    {
      "epoch": 0.48308649801187115,
      "grad_norm": 0.7478095293045044,
      "learning_rate": 5.516083233961254e-06,
      "loss": 0.7266,
      "step": 16766
    },
    {
      "epoch": 0.48311531147352044,
      "grad_norm": 0.7448642253875732,
      "learning_rate": 5.515619113801219e-06,
      "loss": 0.717,
      "step": 16767
    },
    {
      "epoch": 0.4831441249351697,
      "grad_norm": 0.7383393049240112,
      "learning_rate": 5.515154989150657e-06,
      "loss": 0.7181,
      "step": 16768
    },
    {
      "epoch": 0.483172938396819,
      "grad_norm": 0.7446908354759216,
      "learning_rate": 5.51469086001361e-06,
      "loss": 0.7312,
      "step": 16769
    },
    {
      "epoch": 0.48320175185846825,
      "grad_norm": 0.7293996810913086,
      "learning_rate": 5.5142267263941204e-06,
      "loss": 0.711,
      "step": 16770
    },
    {
      "epoch": 0.48323056532011754,
      "grad_norm": 0.7283888459205627,
      "learning_rate": 5.51376258829623e-06,
      "loss": 0.692,
      "step": 16771
    },
    {
      "epoch": 0.48325937878176684,
      "grad_norm": 0.7348889708518982,
      "learning_rate": 5.51329844572398e-06,
      "loss": 0.7219,
      "step": 16772
    },
    {
      "epoch": 0.4832881922434161,
      "grad_norm": 0.7306983470916748,
      "learning_rate": 5.512834298681414e-06,
      "loss": 0.718,
      "step": 16773
    },
    {
      "epoch": 0.4833170057050654,
      "grad_norm": 0.7212983965873718,
      "learning_rate": 5.5123701471725755e-06,
      "loss": 0.7073,
      "step": 16774
    },
    {
      "epoch": 0.4833458191667147,
      "grad_norm": 0.738735020160675,
      "learning_rate": 5.511905991201504e-06,
      "loss": 0.7259,
      "step": 16775
    },
    {
      "epoch": 0.483374632628364,
      "grad_norm": 0.7440963983535767,
      "learning_rate": 5.511441830772243e-06,
      "loss": 0.7212,
      "step": 16776
    },
    {
      "epoch": 0.48340344609001323,
      "grad_norm": 0.7854084968566895,
      "learning_rate": 5.510977665888836e-06,
      "loss": 0.7151,
      "step": 16777
    },
    {
      "epoch": 0.4834322595516625,
      "grad_norm": 0.7519327402114868,
      "learning_rate": 5.510513496555322e-06,
      "loss": 0.7199,
      "step": 16778
    },
    {
      "epoch": 0.4834610730133118,
      "grad_norm": 0.7585299015045166,
      "learning_rate": 5.510049322775746e-06,
      "loss": 0.7199,
      "step": 16779
    },
    {
      "epoch": 0.4834898864749611,
      "grad_norm": 0.7226223945617676,
      "learning_rate": 5.509585144554153e-06,
      "loss": 0.6994,
      "step": 16780
    },
    {
      "epoch": 0.4835186999366104,
      "grad_norm": 0.7410151958465576,
      "learning_rate": 5.509120961894581e-06,
      "loss": 0.7298,
      "step": 16781
    },
    {
      "epoch": 0.4835475133982597,
      "grad_norm": 0.7602745890617371,
      "learning_rate": 5.508656774801073e-06,
      "loss": 0.7258,
      "step": 16782
    },
    {
      "epoch": 0.4835763268599089,
      "grad_norm": 0.7339606285095215,
      "learning_rate": 5.508192583277675e-06,
      "loss": 0.7195,
      "step": 16783
    },
    {
      "epoch": 0.4836051403215582,
      "grad_norm": 2.6407718658447266,
      "learning_rate": 5.507728387328427e-06,
      "loss": 0.7491,
      "step": 16784
    },
    {
      "epoch": 0.4836339537832075,
      "grad_norm": 0.7466810345649719,
      "learning_rate": 5.507264186957372e-06,
      "loss": 0.7565,
      "step": 16785
    },
    {
      "epoch": 0.4836627672448568,
      "grad_norm": 0.751933753490448,
      "learning_rate": 5.506799982168553e-06,
      "loss": 0.7209,
      "step": 16786
    },
    {
      "epoch": 0.4836915807065061,
      "grad_norm": 0.7333232760429382,
      "learning_rate": 5.506335772966014e-06,
      "loss": 0.74,
      "step": 16787
    },
    {
      "epoch": 0.4837203941681554,
      "grad_norm": 0.7123419642448425,
      "learning_rate": 5.505871559353796e-06,
      "loss": 0.6931,
      "step": 16788
    },
    {
      "epoch": 0.48374920762980467,
      "grad_norm": 0.7088101506233215,
      "learning_rate": 5.505407341335942e-06,
      "loss": 0.6979,
      "step": 16789
    },
    {
      "epoch": 0.4837780210914539,
      "grad_norm": 0.7176042199134827,
      "learning_rate": 5.504943118916495e-06,
      "loss": 0.7358,
      "step": 16790
    },
    {
      "epoch": 0.4838068345531032,
      "grad_norm": 0.7100014090538025,
      "learning_rate": 5.504478892099498e-06,
      "loss": 0.7181,
      "step": 16791
    },
    {
      "epoch": 0.4838356480147525,
      "grad_norm": 0.7471333146095276,
      "learning_rate": 5.5040146608889945e-06,
      "loss": 0.7145,
      "step": 16792
    },
    {
      "epoch": 0.4838644614764018,
      "grad_norm": 0.7318114042282104,
      "learning_rate": 5.503550425289028e-06,
      "loss": 0.7097,
      "step": 16793
    },
    {
      "epoch": 0.48389327493805107,
      "grad_norm": 0.7391385436058044,
      "learning_rate": 5.50308618530364e-06,
      "loss": 0.7197,
      "step": 16794
    },
    {
      "epoch": 0.48392208839970036,
      "grad_norm": 0.7519562840461731,
      "learning_rate": 5.502621940936874e-06,
      "loss": 0.749,
      "step": 16795
    },
    {
      "epoch": 0.48395090186134965,
      "grad_norm": 0.748088538646698,
      "learning_rate": 5.5021576921927736e-06,
      "loss": 0.7376,
      "step": 16796
    },
    {
      "epoch": 0.4839797153229989,
      "grad_norm": 0.7309046983718872,
      "learning_rate": 5.5016934390753815e-06,
      "loss": 0.7069,
      "step": 16797
    },
    {
      "epoch": 0.4840085287846482,
      "grad_norm": 0.7508366703987122,
      "learning_rate": 5.501229181588741e-06,
      "loss": 0.7284,
      "step": 16798
    },
    {
      "epoch": 0.48403734224629746,
      "grad_norm": 0.7133965492248535,
      "learning_rate": 5.500764919736896e-06,
      "loss": 0.696,
      "step": 16799
    },
    {
      "epoch": 0.48406615570794675,
      "grad_norm": 0.7042324542999268,
      "learning_rate": 5.500300653523889e-06,
      "loss": 0.7114,
      "step": 16800
    },
    {
      "epoch": 0.48409496916959605,
      "grad_norm": 0.7444375157356262,
      "learning_rate": 5.499836382953763e-06,
      "loss": 0.7142,
      "step": 16801
    },
    {
      "epoch": 0.48412378263124534,
      "grad_norm": 0.7547116875648499,
      "learning_rate": 5.499372108030562e-06,
      "loss": 0.7142,
      "step": 16802
    },
    {
      "epoch": 0.4841525960928946,
      "grad_norm": 0.735902726650238,
      "learning_rate": 5.498907828758328e-06,
      "loss": 0.7133,
      "step": 16803
    },
    {
      "epoch": 0.48418140955454386,
      "grad_norm": 0.7326218485832214,
      "learning_rate": 5.498443545141107e-06,
      "loss": 0.7131,
      "step": 16804
    },
    {
      "epoch": 0.48421022301619315,
      "grad_norm": 0.7436175346374512,
      "learning_rate": 5.497979257182939e-06,
      "loss": 0.7016,
      "step": 16805
    },
    {
      "epoch": 0.48423903647784244,
      "grad_norm": 0.7415751814842224,
      "learning_rate": 5.497514964887873e-06,
      "loss": 0.728,
      "step": 16806
    },
    {
      "epoch": 0.48426784993949173,
      "grad_norm": 0.7286931872367859,
      "learning_rate": 5.497050668259946e-06,
      "loss": 0.7063,
      "step": 16807
    },
    {
      "epoch": 0.484296663401141,
      "grad_norm": 0.7227336764335632,
      "learning_rate": 5.496586367303205e-06,
      "loss": 0.7067,
      "step": 16808
    },
    {
      "epoch": 0.4843254768627903,
      "grad_norm": 0.7431573867797852,
      "learning_rate": 5.496122062021694e-06,
      "loss": 0.7187,
      "step": 16809
    },
    {
      "epoch": 0.48435429032443955,
      "grad_norm": 0.7745595574378967,
      "learning_rate": 5.495657752419456e-06,
      "loss": 0.743,
      "step": 16810
    },
    {
      "epoch": 0.48438310378608884,
      "grad_norm": 0.7946016192436218,
      "learning_rate": 5.495193438500533e-06,
      "loss": 0.7255,
      "step": 16811
    },
    {
      "epoch": 0.48441191724773813,
      "grad_norm": 0.6789360642433167,
      "learning_rate": 5.494729120268973e-06,
      "loss": 0.7173,
      "step": 16812
    },
    {
      "epoch": 0.4844407307093874,
      "grad_norm": 0.717251181602478,
      "learning_rate": 5.494264797728814e-06,
      "loss": 0.7474,
      "step": 16813
    },
    {
      "epoch": 0.4844695441710367,
      "grad_norm": 0.7469701170921326,
      "learning_rate": 5.493800470884103e-06,
      "loss": 0.7014,
      "step": 16814
    },
    {
      "epoch": 0.484498357632686,
      "grad_norm": 0.7513620853424072,
      "learning_rate": 5.493336139738884e-06,
      "loss": 0.7074,
      "step": 16815
    },
    {
      "epoch": 0.4845271710943353,
      "grad_norm": 0.7308911085128784,
      "learning_rate": 5.4928718042972e-06,
      "loss": 0.7226,
      "step": 16816
    },
    {
      "epoch": 0.48455598455598453,
      "grad_norm": 0.7184088230133057,
      "learning_rate": 5.492407464563097e-06,
      "loss": 0.7479,
      "step": 16817
    },
    {
      "epoch": 0.4845847980176338,
      "grad_norm": 0.7283965349197388,
      "learning_rate": 5.491943120540616e-06,
      "loss": 0.7216,
      "step": 16818
    },
    {
      "epoch": 0.4846136114792831,
      "grad_norm": 0.7318004369735718,
      "learning_rate": 5.491478772233803e-06,
      "loss": 0.7268,
      "step": 16819
    },
    {
      "epoch": 0.4846424249409324,
      "grad_norm": 0.739489734172821,
      "learning_rate": 5.4910144196467e-06,
      "loss": 0.7051,
      "step": 16820
    },
    {
      "epoch": 0.4846712384025817,
      "grad_norm": 0.7250241637229919,
      "learning_rate": 5.490550062783352e-06,
      "loss": 0.668,
      "step": 16821
    },
    {
      "epoch": 0.484700051864231,
      "grad_norm": 0.7210108041763306,
      "learning_rate": 5.490085701647805e-06,
      "loss": 0.7278,
      "step": 16822
    },
    {
      "epoch": 0.4847288653258803,
      "grad_norm": 0.709618091583252,
      "learning_rate": 5.489621336244099e-06,
      "loss": 0.7179,
      "step": 16823
    },
    {
      "epoch": 0.4847576787875295,
      "grad_norm": 0.7385750412940979,
      "learning_rate": 5.489156966576282e-06,
      "loss": 0.7223,
      "step": 16824
    },
    {
      "epoch": 0.4847864922491788,
      "grad_norm": 0.7664550542831421,
      "learning_rate": 5.488692592648396e-06,
      "loss": 0.7145,
      "step": 16825
    },
    {
      "epoch": 0.4848153057108281,
      "grad_norm": 0.7305951714515686,
      "learning_rate": 5.488228214464487e-06,
      "loss": 0.6896,
      "step": 16826
    },
    {
      "epoch": 0.4848441191724774,
      "grad_norm": 0.7535414695739746,
      "learning_rate": 5.487763832028599e-06,
      "loss": 0.7267,
      "step": 16827
    },
    {
      "epoch": 0.4848729326341267,
      "grad_norm": 0.7317792773246765,
      "learning_rate": 5.487299445344775e-06,
      "loss": 0.6895,
      "step": 16828
    },
    {
      "epoch": 0.48490174609577597,
      "grad_norm": 0.744870126247406,
      "learning_rate": 5.486835054417059e-06,
      "loss": 0.7394,
      "step": 16829
    },
    {
      "epoch": 0.48493055955742526,
      "grad_norm": 0.7238248586654663,
      "learning_rate": 5.486370659249498e-06,
      "loss": 0.7229,
      "step": 16830
    },
    {
      "epoch": 0.4849593730190745,
      "grad_norm": 0.7120981216430664,
      "learning_rate": 5.485906259846132e-06,
      "loss": 0.719,
      "step": 16831
    },
    {
      "epoch": 0.4849881864807238,
      "grad_norm": 0.71171635389328,
      "learning_rate": 5.485441856211011e-06,
      "loss": 0.6913,
      "step": 16832
    },
    {
      "epoch": 0.4850169999423731,
      "grad_norm": 0.7192398309707642,
      "learning_rate": 5.484977448348176e-06,
      "loss": 0.6885,
      "step": 16833
    },
    {
      "epoch": 0.48504581340402236,
      "grad_norm": 0.729397714138031,
      "learning_rate": 5.484513036261671e-06,
      "loss": 0.7276,
      "step": 16834
    },
    {
      "epoch": 0.48507462686567165,
      "grad_norm": 0.733283281326294,
      "learning_rate": 5.4840486199555444e-06,
      "loss": 0.7357,
      "step": 16835
    },
    {
      "epoch": 0.48510344032732094,
      "grad_norm": 0.7480897307395935,
      "learning_rate": 5.483584199433835e-06,
      "loss": 0.7468,
      "step": 16836
    },
    {
      "epoch": 0.4851322537889702,
      "grad_norm": 0.7127776145935059,
      "learning_rate": 5.483119774700593e-06,
      "loss": 0.6919,
      "step": 16837
    },
    {
      "epoch": 0.48516106725061947,
      "grad_norm": 1.0936225652694702,
      "learning_rate": 5.48265534575986e-06,
      "loss": 0.742,
      "step": 16838
    },
    {
      "epoch": 0.48518988071226876,
      "grad_norm": 0.7307807803153992,
      "learning_rate": 5.482190912615681e-06,
      "loss": 0.7004,
      "step": 16839
    },
    {
      "epoch": 0.48521869417391805,
      "grad_norm": 0.7601779103279114,
      "learning_rate": 5.4817264752721025e-06,
      "loss": 0.7342,
      "step": 16840
    },
    {
      "epoch": 0.48524750763556734,
      "grad_norm": 0.7395563721656799,
      "learning_rate": 5.481262033733166e-06,
      "loss": 0.7077,
      "step": 16841
    },
    {
      "epoch": 0.48527632109721663,
      "grad_norm": 0.776689887046814,
      "learning_rate": 5.480797588002918e-06,
      "loss": 0.7103,
      "step": 16842
    },
    {
      "epoch": 0.4853051345588659,
      "grad_norm": 0.766880452632904,
      "learning_rate": 5.480333138085405e-06,
      "loss": 0.7299,
      "step": 16843
    },
    {
      "epoch": 0.48533394802051516,
      "grad_norm": 0.7343982458114624,
      "learning_rate": 5.4798686839846706e-06,
      "loss": 0.7059,
      "step": 16844
    },
    {
      "epoch": 0.48536276148216445,
      "grad_norm": 0.7342391014099121,
      "learning_rate": 5.4794042257047585e-06,
      "loss": 0.7202,
      "step": 16845
    },
    {
      "epoch": 0.48539157494381374,
      "grad_norm": 0.7587512135505676,
      "learning_rate": 5.478939763249717e-06,
      "loss": 0.7462,
      "step": 16846
    },
    {
      "epoch": 0.48542038840546303,
      "grad_norm": 0.7275137901306152,
      "learning_rate": 5.478475296623588e-06,
      "loss": 0.7071,
      "step": 16847
    },
    {
      "epoch": 0.4854492018671123,
      "grad_norm": 0.7309964299201965,
      "learning_rate": 5.478010825830416e-06,
      "loss": 0.7282,
      "step": 16848
    },
    {
      "epoch": 0.4854780153287616,
      "grad_norm": 0.7522174119949341,
      "learning_rate": 5.4775463508742476e-06,
      "loss": 0.7298,
      "step": 16849
    },
    {
      "epoch": 0.4855068287904109,
      "grad_norm": 0.7462026476860046,
      "learning_rate": 5.47708187175913e-06,
      "loss": 0.742,
      "step": 16850
    },
    {
      "epoch": 0.48553564225206014,
      "grad_norm": 0.6922925114631653,
      "learning_rate": 5.476617388489105e-06,
      "loss": 0.704,
      "step": 16851
    },
    {
      "epoch": 0.48556445571370943,
      "grad_norm": 0.7737824320793152,
      "learning_rate": 5.476152901068219e-06,
      "loss": 0.717,
      "step": 16852
    },
    {
      "epoch": 0.4855932691753587,
      "grad_norm": 0.7207691073417664,
      "learning_rate": 5.475688409500518e-06,
      "loss": 0.7023,
      "step": 16853
    },
    {
      "epoch": 0.485622082637008,
      "grad_norm": 0.7212990522384644,
      "learning_rate": 5.475223913790046e-06,
      "loss": 0.7301,
      "step": 16854
    },
    {
      "epoch": 0.4856508960986573,
      "grad_norm": 0.7412293553352356,
      "learning_rate": 5.4747594139408475e-06,
      "loss": 0.6939,
      "step": 16855
    },
    {
      "epoch": 0.4856797095603066,
      "grad_norm": 0.7416061162948608,
      "learning_rate": 5.47429490995697e-06,
      "loss": 0.7339,
      "step": 16856
    },
    {
      "epoch": 0.4857085230219559,
      "grad_norm": 0.7427890300750732,
      "learning_rate": 5.473830401842458e-06,
      "loss": 0.7311,
      "step": 16857
    },
    {
      "epoch": 0.4857373364836051,
      "grad_norm": 0.7323567271232605,
      "learning_rate": 5.473365889601357e-06,
      "loss": 0.7096,
      "step": 16858
    },
    {
      "epoch": 0.4857661499452544,
      "grad_norm": 0.7735190391540527,
      "learning_rate": 5.472901373237713e-06,
      "loss": 0.747,
      "step": 16859
    },
    {
      "epoch": 0.4857949634069037,
      "grad_norm": 0.713144063949585,
      "learning_rate": 5.472436852755569e-06,
      "loss": 0.7162,
      "step": 16860
    },
    {
      "epoch": 0.485823776868553,
      "grad_norm": 0.7489587068557739,
      "learning_rate": 5.4719723281589735e-06,
      "loss": 0.7144,
      "step": 16861
    },
    {
      "epoch": 0.4858525903302023,
      "grad_norm": 0.7695068717002869,
      "learning_rate": 5.471507799451971e-06,
      "loss": 0.7196,
      "step": 16862
    },
    {
      "epoch": 0.4858814037918516,
      "grad_norm": 0.7554489374160767,
      "learning_rate": 5.471043266638606e-06,
      "loss": 0.6879,
      "step": 16863
    },
    {
      "epoch": 0.4859102172535008,
      "grad_norm": 0.7348440289497375,
      "learning_rate": 5.470578729722925e-06,
      "loss": 0.7096,
      "step": 16864
    },
    {
      "epoch": 0.4859390307151501,
      "grad_norm": 0.73460453748703,
      "learning_rate": 5.470114188708975e-06,
      "loss": 0.7034,
      "step": 16865
    },
    {
      "epoch": 0.4859678441767994,
      "grad_norm": 0.7121290564537048,
      "learning_rate": 5.4696496436007995e-06,
      "loss": 0.6986,
      "step": 16866
    },
    {
      "epoch": 0.4859966576384487,
      "grad_norm": 0.7674098610877991,
      "learning_rate": 5.4691850944024436e-06,
      "loss": 0.7454,
      "step": 16867
    },
    {
      "epoch": 0.48602547110009797,
      "grad_norm": 0.7341053485870361,
      "learning_rate": 5.468720541117957e-06,
      "loss": 0.7141,
      "step": 16868
    },
    {
      "epoch": 0.48605428456174726,
      "grad_norm": 0.7151100039482117,
      "learning_rate": 5.4682559837513825e-06,
      "loss": 0.6873,
      "step": 16869
    },
    {
      "epoch": 0.48608309802339655,
      "grad_norm": 0.7541784644126892,
      "learning_rate": 5.4677914223067665e-06,
      "loss": 0.7243,
      "step": 16870
    },
    {
      "epoch": 0.4861119114850458,
      "grad_norm": 0.7510747313499451,
      "learning_rate": 5.467326856788154e-06,
      "loss": 0.7036,
      "step": 16871
    },
    {
      "epoch": 0.4861407249466951,
      "grad_norm": 0.7603999972343445,
      "learning_rate": 5.4668622871995925e-06,
      "loss": 0.7428,
      "step": 16872
    },
    {
      "epoch": 0.48616953840834437,
      "grad_norm": 0.7427142858505249,
      "learning_rate": 5.466397713545126e-06,
      "loss": 0.7306,
      "step": 16873
    },
    {
      "epoch": 0.48619835186999366,
      "grad_norm": 0.7534748911857605,
      "learning_rate": 5.465933135828802e-06,
      "loss": 0.7243,
      "step": 16874
    },
    {
      "epoch": 0.48622716533164295,
      "grad_norm": 0.7415034174919128,
      "learning_rate": 5.465468554054667e-06,
      "loss": 0.7148,
      "step": 16875
    },
    {
      "epoch": 0.48625597879329224,
      "grad_norm": 0.7574782371520996,
      "learning_rate": 5.465003968226766e-06,
      "loss": 0.7212,
      "step": 16876
    },
    {
      "epoch": 0.48628479225494153,
      "grad_norm": 0.7623410224914551,
      "learning_rate": 5.4645393783491456e-06,
      "loss": 0.7256,
      "step": 16877
    },
    {
      "epoch": 0.48631360571659077,
      "grad_norm": 0.7510899305343628,
      "learning_rate": 5.464074784425852e-06,
      "loss": 0.7463,
      "step": 16878
    },
    {
      "epoch": 0.48634241917824006,
      "grad_norm": 0.775842010974884,
      "learning_rate": 5.463610186460928e-06,
      "loss": 0.7198,
      "step": 16879
    },
    {
      "epoch": 0.48637123263988935,
      "grad_norm": 0.7042752504348755,
      "learning_rate": 5.463145584458425e-06,
      "loss": 0.7342,
      "step": 16880
    },
    {
      "epoch": 0.48640004610153864,
      "grad_norm": 0.7283737063407898,
      "learning_rate": 5.462680978422387e-06,
      "loss": 0.7247,
      "step": 16881
    },
    {
      "epoch": 0.48642885956318793,
      "grad_norm": 0.7603070735931396,
      "learning_rate": 5.4622163683568584e-06,
      "loss": 0.7325,
      "step": 16882
    },
    {
      "epoch": 0.4864576730248372,
      "grad_norm": 0.7593424320220947,
      "learning_rate": 5.46175175426589e-06,
      "loss": 0.7117,
      "step": 16883
    },
    {
      "epoch": 0.4864864864864865,
      "grad_norm": 0.7777803540229797,
      "learning_rate": 5.4612871361535235e-06,
      "loss": 0.7259,
      "step": 16884
    },
    {
      "epoch": 0.48651529994813575,
      "grad_norm": 0.781003475189209,
      "learning_rate": 5.460822514023807e-06,
      "loss": 0.7311,
      "step": 16885
    },
    {
      "epoch": 0.48654411340978504,
      "grad_norm": 0.7718717455863953,
      "learning_rate": 5.4603578878807886e-06,
      "loss": 0.7203,
      "step": 16886
    },
    {
      "epoch": 0.48657292687143433,
      "grad_norm": 0.7663869261741638,
      "learning_rate": 5.459893257728512e-06,
      "loss": 0.7375,
      "step": 16887
    },
    {
      "epoch": 0.4866017403330836,
      "grad_norm": 0.7607789635658264,
      "learning_rate": 5.4594286235710255e-06,
      "loss": 0.7276,
      "step": 16888
    },
    {
      "epoch": 0.4866305537947329,
      "grad_norm": 0.7413244843482971,
      "learning_rate": 5.4589639854123745e-06,
      "loss": 0.7226,
      "step": 16889
    },
    {
      "epoch": 0.4866593672563822,
      "grad_norm": 1.1530557870864868,
      "learning_rate": 5.458499343256607e-06,
      "loss": 0.7052,
      "step": 16890
    },
    {
      "epoch": 0.48668818071803144,
      "grad_norm": 0.7433449029922485,
      "learning_rate": 5.458034697107767e-06,
      "loss": 0.7185,
      "step": 16891
    },
    {
      "epoch": 0.48671699417968073,
      "grad_norm": 0.7241742610931396,
      "learning_rate": 5.457570046969904e-06,
      "loss": 0.7028,
      "step": 16892
    },
    {
      "epoch": 0.48674580764133,
      "grad_norm": 0.7164261937141418,
      "learning_rate": 5.457105392847062e-06,
      "loss": 0.7021,
      "step": 16893
    },
    {
      "epoch": 0.4867746211029793,
      "grad_norm": 0.7320825457572937,
      "learning_rate": 5.45664073474329e-06,
      "loss": 0.7143,
      "step": 16894
    },
    {
      "epoch": 0.4868034345646286,
      "grad_norm": 0.7079203128814697,
      "learning_rate": 5.456176072662633e-06,
      "loss": 0.7205,
      "step": 16895
    },
    {
      "epoch": 0.4868322480262779,
      "grad_norm": 0.7499498128890991,
      "learning_rate": 5.455711406609139e-06,
      "loss": 0.6909,
      "step": 16896
    },
    {
      "epoch": 0.4868610614879272,
      "grad_norm": 0.7420828938484192,
      "learning_rate": 5.455246736586853e-06,
      "loss": 0.7063,
      "step": 16897
    },
    {
      "epoch": 0.4868898749495764,
      "grad_norm": 0.7558904886245728,
      "learning_rate": 5.454782062599825e-06,
      "loss": 0.7268,
      "step": 16898
    },
    {
      "epoch": 0.4869186884112257,
      "grad_norm": 0.7543131709098816,
      "learning_rate": 5.454317384652098e-06,
      "loss": 0.729,
      "step": 16899
    },
    {
      "epoch": 0.486947501872875,
      "grad_norm": 0.7404972910881042,
      "learning_rate": 5.453852702747722e-06,
      "loss": 0.7078,
      "step": 16900
    },
    {
      "epoch": 0.4869763153345243,
      "grad_norm": 0.7304301261901855,
      "learning_rate": 5.4533880168907414e-06,
      "loss": 0.6957,
      "step": 16901
    },
    {
      "epoch": 0.4870051287961736,
      "grad_norm": 0.7169355750083923,
      "learning_rate": 5.452923327085206e-06,
      "loss": 0.7486,
      "step": 16902
    },
    {
      "epoch": 0.48703394225782287,
      "grad_norm": 0.7475118041038513,
      "learning_rate": 5.452458633335159e-06,
      "loss": 0.7139,
      "step": 16903
    },
    {
      "epoch": 0.48706275571947216,
      "grad_norm": 0.7166236639022827,
      "learning_rate": 5.451993935644651e-06,
      "loss": 0.7101,
      "step": 16904
    },
    {
      "epoch": 0.4870915691811214,
      "grad_norm": 0.7456474900245667,
      "learning_rate": 5.451529234017727e-06,
      "loss": 0.728,
      "step": 16905
    },
    {
      "epoch": 0.4871203826427707,
      "grad_norm": 0.7146993279457092,
      "learning_rate": 5.4510645284584365e-06,
      "loss": 0.7078,
      "step": 16906
    },
    {
      "epoch": 0.48714919610442,
      "grad_norm": 0.8115000128746033,
      "learning_rate": 5.450599818970823e-06,
      "loss": 0.7156,
      "step": 16907
    },
    {
      "epoch": 0.48717800956606927,
      "grad_norm": 0.7350112795829773,
      "learning_rate": 5.450135105558937e-06,
      "loss": 0.7106,
      "step": 16908
    },
    {
      "epoch": 0.48720682302771856,
      "grad_norm": 0.7515522241592407,
      "learning_rate": 5.449670388226822e-06,
      "loss": 0.7354,
      "step": 16909
    },
    {
      "epoch": 0.48723563648936785,
      "grad_norm": 0.740871250629425,
      "learning_rate": 5.44920566697853e-06,
      "loss": 0.7229,
      "step": 16910
    },
    {
      "epoch": 0.48726444995101714,
      "grad_norm": 0.691401481628418,
      "learning_rate": 5.4487409418181044e-06,
      "loss": 0.6841,
      "step": 16911
    },
    {
      "epoch": 0.4872932634126664,
      "grad_norm": 0.7278902530670166,
      "learning_rate": 5.448276212749593e-06,
      "loss": 0.7112,
      "step": 16912
    },
    {
      "epoch": 0.48732207687431567,
      "grad_norm": 0.7369861602783203,
      "learning_rate": 5.4478114797770456e-06,
      "loss": 0.7128,
      "step": 16913
    },
    {
      "epoch": 0.48735089033596496,
      "grad_norm": 0.7340229749679565,
      "learning_rate": 5.447346742904508e-06,
      "loss": 0.7234,
      "step": 16914
    },
    {
      "epoch": 0.48737970379761425,
      "grad_norm": 0.7800581455230713,
      "learning_rate": 5.446882002136025e-06,
      "loss": 0.7442,
      "step": 16915
    },
    {
      "epoch": 0.48740851725926354,
      "grad_norm": 0.7350847721099854,
      "learning_rate": 5.446417257475648e-06,
      "loss": 0.7425,
      "step": 16916
    },
    {
      "epoch": 0.48743733072091283,
      "grad_norm": 0.7336879372596741,
      "learning_rate": 5.445952508927423e-06,
      "loss": 0.7246,
      "step": 16917
    },
    {
      "epoch": 0.48746614418256207,
      "grad_norm": 0.7224193811416626,
      "learning_rate": 5.445487756495398e-06,
      "loss": 0.7296,
      "step": 16918
    },
    {
      "epoch": 0.48749495764421136,
      "grad_norm": 0.7093663215637207,
      "learning_rate": 5.445023000183619e-06,
      "loss": 0.7284,
      "step": 16919
    },
    {
      "epoch": 0.48752377110586065,
      "grad_norm": 0.7477984428405762,
      "learning_rate": 5.444558239996134e-06,
      "loss": 0.7171,
      "step": 16920
    },
    {
      "epoch": 0.48755258456750994,
      "grad_norm": 0.7398480772972107,
      "learning_rate": 5.444093475936992e-06,
      "loss": 0.743,
      "step": 16921
    },
    {
      "epoch": 0.48758139802915923,
      "grad_norm": 0.7368444204330444,
      "learning_rate": 5.443628708010239e-06,
      "loss": 0.7093,
      "step": 16922
    },
    {
      "epoch": 0.4876102114908085,
      "grad_norm": 0.7171550393104553,
      "learning_rate": 5.443163936219925e-06,
      "loss": 0.7003,
      "step": 16923
    },
    {
      "epoch": 0.4876390249524578,
      "grad_norm": 0.7310102581977844,
      "learning_rate": 5.4426991605700954e-06,
      "loss": 0.6865,
      "step": 16924
    },
    {
      "epoch": 0.48766783841410705,
      "grad_norm": 0.7280425429344177,
      "learning_rate": 5.442234381064798e-06,
      "loss": 0.7141,
      "step": 16925
    },
    {
      "epoch": 0.48769665187575634,
      "grad_norm": 0.7650890350341797,
      "learning_rate": 5.441769597708082e-06,
      "loss": 0.7435,
      "step": 16926
    },
    {
      "epoch": 0.48772546533740563,
      "grad_norm": 0.7501299381256104,
      "learning_rate": 5.441304810503993e-06,
      "loss": 0.71,
      "step": 16927
    },
    {
      "epoch": 0.4877542787990549,
      "grad_norm": 0.7228778600692749,
      "learning_rate": 5.440840019456582e-06,
      "loss": 0.7291,
      "step": 16928
    },
    {
      "epoch": 0.4877830922607042,
      "grad_norm": 0.7394979000091553,
      "learning_rate": 5.440375224569895e-06,
      "loss": 0.7082,
      "step": 16929
    },
    {
      "epoch": 0.4878119057223535,
      "grad_norm": 0.7555347084999084,
      "learning_rate": 5.439910425847979e-06,
      "loss": 0.7202,
      "step": 16930
    },
    {
      "epoch": 0.4878407191840028,
      "grad_norm": 0.7372070550918579,
      "learning_rate": 5.439445623294884e-06,
      "loss": 0.7276,
      "step": 16931
    },
    {
      "epoch": 0.487869532645652,
      "grad_norm": 0.709436297416687,
      "learning_rate": 5.4389808169146565e-06,
      "loss": 0.7195,
      "step": 16932
    },
    {
      "epoch": 0.4878983461073013,
      "grad_norm": 0.7280899882316589,
      "learning_rate": 5.438516006711344e-06,
      "loss": 0.7209,
      "step": 16933
    },
    {
      "epoch": 0.4879271595689506,
      "grad_norm": 0.7538740038871765,
      "learning_rate": 5.4380511926889966e-06,
      "loss": 0.7144,
      "step": 16934
    },
    {
      "epoch": 0.4879559730305999,
      "grad_norm": 0.7569742202758789,
      "learning_rate": 5.437586374851661e-06,
      "loss": 0.7086,
      "step": 16935
    },
    {
      "epoch": 0.4879847864922492,
      "grad_norm": 0.7239482402801514,
      "learning_rate": 5.437121553203385e-06,
      "loss": 0.706,
      "step": 16936
    },
    {
      "epoch": 0.4880135999538985,
      "grad_norm": 0.7326367497444153,
      "learning_rate": 5.436656727748218e-06,
      "loss": 0.7115,
      "step": 16937
    },
    {
      "epoch": 0.48804241341554777,
      "grad_norm": 0.7804672718048096,
      "learning_rate": 5.436191898490207e-06,
      "loss": 0.7181,
      "step": 16938
    },
    {
      "epoch": 0.488071226877197,
      "grad_norm": 0.761545717716217,
      "learning_rate": 5.435727065433399e-06,
      "loss": 0.728,
      "step": 16939
    },
    {
      "epoch": 0.4881000403388463,
      "grad_norm": 0.7692657113075256,
      "learning_rate": 5.435262228581846e-06,
      "loss": 0.7319,
      "step": 16940
    },
    {
      "epoch": 0.4881288538004956,
      "grad_norm": 0.7234093546867371,
      "learning_rate": 5.434797387939593e-06,
      "loss": 0.7318,
      "step": 16941
    },
    {
      "epoch": 0.4881576672621449,
      "grad_norm": 0.7447591423988342,
      "learning_rate": 5.4343325435106895e-06,
      "loss": 0.7195,
      "step": 16942
    },
    {
      "epoch": 0.48818648072379417,
      "grad_norm": 0.7865151762962341,
      "learning_rate": 5.433867695299184e-06,
      "loss": 0.7149,
      "step": 16943
    },
    {
      "epoch": 0.48821529418544346,
      "grad_norm": 0.7025193572044373,
      "learning_rate": 5.433402843309125e-06,
      "loss": 0.6946,
      "step": 16944
    },
    {
      "epoch": 0.4882441076470927,
      "grad_norm": 0.7340787053108215,
      "learning_rate": 5.43293798754456e-06,
      "loss": 0.7283,
      "step": 16945
    },
    {
      "epoch": 0.488272921108742,
      "grad_norm": 0.7580493092536926,
      "learning_rate": 5.432473128009538e-06,
      "loss": 0.7213,
      "step": 16946
    },
    {
      "epoch": 0.4883017345703913,
      "grad_norm": 0.7305262088775635,
      "learning_rate": 5.432008264708107e-06,
      "loss": 0.7258,
      "step": 16947
    },
    {
      "epoch": 0.48833054803204057,
      "grad_norm": 0.7284015417098999,
      "learning_rate": 5.431543397644317e-06,
      "loss": 0.7297,
      "step": 16948
    },
    {
      "epoch": 0.48835936149368986,
      "grad_norm": 0.719892144203186,
      "learning_rate": 5.431078526822215e-06,
      "loss": 0.7048,
      "step": 16949
    },
    {
      "epoch": 0.48838817495533915,
      "grad_norm": 0.7340116500854492,
      "learning_rate": 5.43061365224585e-06,
      "loss": 0.7285,
      "step": 16950
    },
    {
      "epoch": 0.48841698841698844,
      "grad_norm": 0.7491402626037598,
      "learning_rate": 5.4301487739192685e-06,
      "loss": 0.7092,
      "step": 16951
    },
    {
      "epoch": 0.4884458018786377,
      "grad_norm": 0.7043713331222534,
      "learning_rate": 5.429683891846523e-06,
      "loss": 0.7243,
      "step": 16952
    },
    {
      "epoch": 0.48847461534028697,
      "grad_norm": 0.7347774505615234,
      "learning_rate": 5.4292190060316606e-06,
      "loss": 0.745,
      "step": 16953
    },
    {
      "epoch": 0.48850342880193626,
      "grad_norm": 0.7354212403297424,
      "learning_rate": 5.428754116478729e-06,
      "loss": 0.7177,
      "step": 16954
    },
    {
      "epoch": 0.48853224226358555,
      "grad_norm": 0.7111362814903259,
      "learning_rate": 5.428289223191778e-06,
      "loss": 0.6984,
      "step": 16955
    },
    {
      "epoch": 0.48856105572523484,
      "grad_norm": 0.7619835734367371,
      "learning_rate": 5.427824326174857e-06,
      "loss": 0.729,
      "step": 16956
    },
    {
      "epoch": 0.48858986918688413,
      "grad_norm": 0.740319550037384,
      "learning_rate": 5.427359425432012e-06,
      "loss": 0.7162,
      "step": 16957
    },
    {
      "epoch": 0.4886186826485334,
      "grad_norm": 0.7528384327888489,
      "learning_rate": 5.426894520967294e-06,
      "loss": 0.7393,
      "step": 16958
    },
    {
      "epoch": 0.48864749611018266,
      "grad_norm": 0.7556458711624146,
      "learning_rate": 5.426429612784751e-06,
      "loss": 0.722,
      "step": 16959
    },
    {
      "epoch": 0.48867630957183195,
      "grad_norm": 0.7149136066436768,
      "learning_rate": 5.425964700888433e-06,
      "loss": 0.7241,
      "step": 16960
    },
    {
      "epoch": 0.48870512303348124,
      "grad_norm": 0.7349650859832764,
      "learning_rate": 5.4254997852823884e-06,
      "loss": 0.7049,
      "step": 16961
    },
    {
      "epoch": 0.4887339364951305,
      "grad_norm": 0.8583880066871643,
      "learning_rate": 5.425034865970666e-06,
      "loss": 0.7309,
      "step": 16962
    },
    {
      "epoch": 0.4887627499567798,
      "grad_norm": 0.7493630051612854,
      "learning_rate": 5.424569942957314e-06,
      "loss": 0.7544,
      "step": 16963
    },
    {
      "epoch": 0.4887915634184291,
      "grad_norm": 0.7193743586540222,
      "learning_rate": 5.424105016246384e-06,
      "loss": 0.7231,
      "step": 16964
    },
    {
      "epoch": 0.4888203768800784,
      "grad_norm": 0.7371389269828796,
      "learning_rate": 5.423640085841922e-06,
      "loss": 0.7092,
      "step": 16965
    },
    {
      "epoch": 0.48884919034172764,
      "grad_norm": 0.7635523676872253,
      "learning_rate": 5.423175151747978e-06,
      "loss": 0.7537,
      "step": 16966
    },
    {
      "epoch": 0.4888780038033769,
      "grad_norm": 0.7495578527450562,
      "learning_rate": 5.422710213968603e-06,
      "loss": 0.7361,
      "step": 16967
    },
    {
      "epoch": 0.4889068172650262,
      "grad_norm": 0.7603480815887451,
      "learning_rate": 5.422245272507843e-06,
      "loss": 0.7183,
      "step": 16968
    },
    {
      "epoch": 0.4889356307266755,
      "grad_norm": 0.7616903781890869,
      "learning_rate": 5.421780327369749e-06,
      "loss": 0.7256,
      "step": 16969
    },
    {
      "epoch": 0.4889644441883248,
      "grad_norm": 0.7554761171340942,
      "learning_rate": 5.421315378558371e-06,
      "loss": 0.7448,
      "step": 16970
    },
    {
      "epoch": 0.4889932576499741,
      "grad_norm": 0.7532882690429688,
      "learning_rate": 5.420850426077756e-06,
      "loss": 0.7457,
      "step": 16971
    },
    {
      "epoch": 0.4890220711116233,
      "grad_norm": 0.7265740036964417,
      "learning_rate": 5.420385469931956e-06,
      "loss": 0.6936,
      "step": 16972
    },
    {
      "epoch": 0.4890508845732726,
      "grad_norm": 0.7198699712753296,
      "learning_rate": 5.4199205101250175e-06,
      "loss": 0.7155,
      "step": 16973
    },
    {
      "epoch": 0.4890796980349219,
      "grad_norm": 0.7118257284164429,
      "learning_rate": 5.419455546660992e-06,
      "loss": 0.7046,
      "step": 16974
    },
    {
      "epoch": 0.4891085114965712,
      "grad_norm": 0.7530213594436646,
      "learning_rate": 5.418990579543927e-06,
      "loss": 0.7212,
      "step": 16975
    },
    {
      "epoch": 0.4891373249582205,
      "grad_norm": 0.7577535510063171,
      "learning_rate": 5.418525608777872e-06,
      "loss": 0.7405,
      "step": 16976
    },
    {
      "epoch": 0.4891661384198698,
      "grad_norm": 0.7139708995819092,
      "learning_rate": 5.41806063436688e-06,
      "loss": 0.7103,
      "step": 16977
    },
    {
      "epoch": 0.48919495188151907,
      "grad_norm": 0.7406406402587891,
      "learning_rate": 5.417595656314997e-06,
      "loss": 0.7117,
      "step": 16978
    },
    {
      "epoch": 0.4892237653431683,
      "grad_norm": 0.7486098408699036,
      "learning_rate": 5.417130674626271e-06,
      "loss": 0.7098,
      "step": 16979
    },
    {
      "epoch": 0.4892525788048176,
      "grad_norm": 0.7358320355415344,
      "learning_rate": 5.416665689304756e-06,
      "loss": 0.7101,
      "step": 16980
    },
    {
      "epoch": 0.4892813922664669,
      "grad_norm": 0.7491533160209656,
      "learning_rate": 5.416200700354498e-06,
      "loss": 0.7392,
      "step": 16981
    },
    {
      "epoch": 0.4893102057281162,
      "grad_norm": 0.7091354727745056,
      "learning_rate": 5.4157357077795495e-06,
      "loss": 0.6961,
      "step": 16982
    },
    {
      "epoch": 0.48933901918976547,
      "grad_norm": 0.7480749487876892,
      "learning_rate": 5.415270711583958e-06,
      "loss": 0.7233,
      "step": 16983
    },
    {
      "epoch": 0.48936783265141476,
      "grad_norm": 0.7365549206733704,
      "learning_rate": 5.4148057117717735e-06,
      "loss": 0.7269,
      "step": 16984
    },
    {
      "epoch": 0.48939664611306405,
      "grad_norm": 0.7063134908676147,
      "learning_rate": 5.414340708347045e-06,
      "loss": 0.6825,
      "step": 16985
    },
    {
      "epoch": 0.4894254595747133,
      "grad_norm": 0.7339118719100952,
      "learning_rate": 5.4138757013138255e-06,
      "loss": 0.6983,
      "step": 16986
    },
    {
      "epoch": 0.4894542730363626,
      "grad_norm": 0.7444279789924622,
      "learning_rate": 5.4134106906761594e-06,
      "loss": 0.7451,
      "step": 16987
    },
    {
      "epoch": 0.48948308649801187,
      "grad_norm": 0.7361765503883362,
      "learning_rate": 5.412945676438102e-06,
      "loss": 0.7157,
      "step": 16988
    },
    {
      "epoch": 0.48951189995966116,
      "grad_norm": 0.7385621666908264,
      "learning_rate": 5.412480658603699e-06,
      "loss": 0.7347,
      "step": 16989
    },
    {
      "epoch": 0.48954071342131045,
      "grad_norm": 0.7172850966453552,
      "learning_rate": 5.4120156371770035e-06,
      "loss": 0.7131,
      "step": 16990
    },
    {
      "epoch": 0.48956952688295974,
      "grad_norm": 0.7423264980316162,
      "learning_rate": 5.411550612162063e-06,
      "loss": 0.7322,
      "step": 16991
    },
    {
      "epoch": 0.48959834034460903,
      "grad_norm": 0.7500247359275818,
      "learning_rate": 5.4110855835629276e-06,
      "loss": 0.7264,
      "step": 16992
    },
    {
      "epoch": 0.48962715380625826,
      "grad_norm": 0.7293882966041565,
      "learning_rate": 5.410620551383647e-06,
      "loss": 0.7495,
      "step": 16993
    },
    {
      "epoch": 0.48965596726790755,
      "grad_norm": 0.7228199243545532,
      "learning_rate": 5.410155515628272e-06,
      "loss": 0.7135,
      "step": 16994
    },
    {
      "epoch": 0.48968478072955685,
      "grad_norm": 0.7080143094062805,
      "learning_rate": 5.409690476300854e-06,
      "loss": 0.7084,
      "step": 16995
    },
    {
      "epoch": 0.48971359419120614,
      "grad_norm": 0.7301921844482422,
      "learning_rate": 5.409225433405441e-06,
      "loss": 0.745,
      "step": 16996
    },
    {
      "epoch": 0.4897424076528554,
      "grad_norm": 0.7379408478736877,
      "learning_rate": 5.408760386946082e-06,
      "loss": 0.7281,
      "step": 16997
    },
    {
      "epoch": 0.4897712211145047,
      "grad_norm": 0.7203425168991089,
      "learning_rate": 5.40829533692683e-06,
      "loss": 0.7204,
      "step": 16998
    },
    {
      "epoch": 0.48980003457615395,
      "grad_norm": 0.7083396911621094,
      "learning_rate": 5.407830283351734e-06,
      "loss": 0.7241,
      "step": 16999
    },
    {
      "epoch": 0.48982884803780324,
      "grad_norm": 0.7472553849220276,
      "learning_rate": 5.407365226224842e-06,
      "loss": 0.7524,
      "step": 17000
    },
    {
      "epoch": 0.48985766149945253,
      "grad_norm": 0.7346212267875671,
      "learning_rate": 5.4069001655502075e-06,
      "loss": 0.7017,
      "step": 17001
    },
    {
      "epoch": 0.4898864749611018,
      "grad_norm": 0.7129847407341003,
      "learning_rate": 5.406435101331879e-06,
      "loss": 0.6992,
      "step": 17002
    },
    {
      "epoch": 0.4899152884227511,
      "grad_norm": 0.742302417755127,
      "learning_rate": 5.4059700335739085e-06,
      "loss": 0.7172,
      "step": 17003
    },
    {
      "epoch": 0.4899441018844004,
      "grad_norm": 0.7197604179382324,
      "learning_rate": 5.4055049622803435e-06,
      "loss": 0.72,
      "step": 17004
    },
    {
      "epoch": 0.4899729153460497,
      "grad_norm": 0.7247949838638306,
      "learning_rate": 5.405039887455236e-06,
      "loss": 0.7055,
      "step": 17005
    },
    {
      "epoch": 0.49000172880769893,
      "grad_norm": 0.7320525050163269,
      "learning_rate": 5.404574809102635e-06,
      "loss": 0.7055,
      "step": 17006
    },
    {
      "epoch": 0.4900305422693482,
      "grad_norm": 0.7169275283813477,
      "learning_rate": 5.404109727226593e-06,
      "loss": 0.7068,
      "step": 17007
    },
    {
      "epoch": 0.4900593557309975,
      "grad_norm": 0.6970158815383911,
      "learning_rate": 5.403644641831159e-06,
      "loss": 0.7196,
      "step": 17008
    },
    {
      "epoch": 0.4900881691926468,
      "grad_norm": 0.7307920455932617,
      "learning_rate": 5.403179552920383e-06,
      "loss": 0.7286,
      "step": 17009
    },
    {
      "epoch": 0.4901169826542961,
      "grad_norm": 0.7294957041740417,
      "learning_rate": 5.402714460498318e-06,
      "loss": 0.7034,
      "step": 17010
    },
    {
      "epoch": 0.4901457961159454,
      "grad_norm": 0.7015345692634583,
      "learning_rate": 5.40224936456901e-06,
      "loss": 0.6826,
      "step": 17011
    },
    {
      "epoch": 0.4901746095775947,
      "grad_norm": 0.7652335166931152,
      "learning_rate": 5.4017842651365134e-06,
      "loss": 0.7247,
      "step": 17012
    },
    {
      "epoch": 0.4902034230392439,
      "grad_norm": 0.7332139015197754,
      "learning_rate": 5.401319162204879e-06,
      "loss": 0.7458,
      "step": 17013
    },
    {
      "epoch": 0.4902322365008932,
      "grad_norm": 0.7118077874183655,
      "learning_rate": 5.400854055778154e-06,
      "loss": 0.7293,
      "step": 17014
    },
    {
      "epoch": 0.4902610499625425,
      "grad_norm": 0.7393190264701843,
      "learning_rate": 5.400388945860391e-06,
      "loss": 0.6936,
      "step": 17015
    },
    {
      "epoch": 0.4902898634241918,
      "grad_norm": 0.7025413513183594,
      "learning_rate": 5.399923832455642e-06,
      "loss": 0.6992,
      "step": 17016
    },
    {
      "epoch": 0.4903186768858411,
      "grad_norm": 0.7289223670959473,
      "learning_rate": 5.399458715567953e-06,
      "loss": 0.7205,
      "step": 17017
    },
    {
      "epoch": 0.49034749034749037,
      "grad_norm": 0.7269256114959717,
      "learning_rate": 5.3989935952013804e-06,
      "loss": 0.6978,
      "step": 17018
    },
    {
      "epoch": 0.49037630380913966,
      "grad_norm": 0.7230930328369141,
      "learning_rate": 5.398528471359972e-06,
      "loss": 0.7035,
      "step": 17019
    },
    {
      "epoch": 0.4904051172707889,
      "grad_norm": 0.7296061515808105,
      "learning_rate": 5.398063344047778e-06,
      "loss": 0.7198,
      "step": 17020
    },
    {
      "epoch": 0.4904339307324382,
      "grad_norm": 0.7879016399383545,
      "learning_rate": 5.397598213268852e-06,
      "loss": 0.7278,
      "step": 17021
    },
    {
      "epoch": 0.4904627441940875,
      "grad_norm": 0.7440977096557617,
      "learning_rate": 5.397133079027243e-06,
      "loss": 0.7095,
      "step": 17022
    },
    {
      "epoch": 0.49049155765573677,
      "grad_norm": 0.7335206866264343,
      "learning_rate": 5.396667941327e-06,
      "loss": 0.7191,
      "step": 17023
    },
    {
      "epoch": 0.49052037111738606,
      "grad_norm": 0.7433528304100037,
      "learning_rate": 5.3962028001721765e-06,
      "loss": 0.7301,
      "step": 17024
    },
    {
      "epoch": 0.49054918457903535,
      "grad_norm": 0.7158470153808594,
      "learning_rate": 5.395737655566823e-06,
      "loss": 0.6988,
      "step": 17025
    },
    {
      "epoch": 0.4905779980406846,
      "grad_norm": 0.7417497634887695,
      "learning_rate": 5.39527250751499e-06,
      "loss": 0.7369,
      "step": 17026
    },
    {
      "epoch": 0.4906068115023339,
      "grad_norm": 0.7351358532905579,
      "learning_rate": 5.394807356020729e-06,
      "loss": 0.7395,
      "step": 17027
    },
    {
      "epoch": 0.49063562496398316,
      "grad_norm": 0.7233834266662598,
      "learning_rate": 5.39434220108809e-06,
      "loss": 0.6959,
      "step": 17028
    },
    {
      "epoch": 0.49066443842563245,
      "grad_norm": 0.7216721177101135,
      "learning_rate": 5.393877042721124e-06,
      "loss": 0.7168,
      "step": 17029
    },
    {
      "epoch": 0.49069325188728175,
      "grad_norm": 0.7296134829521179,
      "learning_rate": 5.393411880923884e-06,
      "loss": 0.7302,
      "step": 17030
    },
    {
      "epoch": 0.49072206534893104,
      "grad_norm": 0.7489143013954163,
      "learning_rate": 5.39294671570042e-06,
      "loss": 0.6959,
      "step": 17031
    },
    {
      "epoch": 0.4907508788105803,
      "grad_norm": 0.706453263759613,
      "learning_rate": 5.392481547054782e-06,
      "loss": 0.7133,
      "step": 17032
    },
    {
      "epoch": 0.49077969227222956,
      "grad_norm": 0.7380275130271912,
      "learning_rate": 5.392016374991022e-06,
      "loss": 0.7115,
      "step": 17033
    },
    {
      "epoch": 0.49080850573387885,
      "grad_norm": 0.8099077343940735,
      "learning_rate": 5.391551199513192e-06,
      "loss": 0.7333,
      "step": 17034
    },
    {
      "epoch": 0.49083731919552814,
      "grad_norm": 0.7474621534347534,
      "learning_rate": 5.391086020625341e-06,
      "loss": 0.7523,
      "step": 17035
    },
    {
      "epoch": 0.49086613265717743,
      "grad_norm": 0.7344893217086792,
      "learning_rate": 5.390620838331523e-06,
      "loss": 0.716,
      "step": 17036
    },
    {
      "epoch": 0.4908949461188267,
      "grad_norm": 0.7500784993171692,
      "learning_rate": 5.390155652635788e-06,
      "loss": 0.7244,
      "step": 17037
    },
    {
      "epoch": 0.490923759580476,
      "grad_norm": 0.7142464518547058,
      "learning_rate": 5.389690463542187e-06,
      "loss": 0.702,
      "step": 17038
    },
    {
      "epoch": 0.4909525730421253,
      "grad_norm": 0.7473583817481995,
      "learning_rate": 5.389225271054771e-06,
      "loss": 0.6944,
      "step": 17039
    },
    {
      "epoch": 0.49098138650377454,
      "grad_norm": 0.7345207929611206,
      "learning_rate": 5.388760075177592e-06,
      "loss": 0.7275,
      "step": 17040
    },
    {
      "epoch": 0.49101019996542383,
      "grad_norm": 0.7545759677886963,
      "learning_rate": 5.3882948759147014e-06,
      "loss": 0.7129,
      "step": 17041
    },
    {
      "epoch": 0.4910390134270731,
      "grad_norm": 0.7074885368347168,
      "learning_rate": 5.387829673270152e-06,
      "loss": 0.7017,
      "step": 17042
    },
    {
      "epoch": 0.4910678268887224,
      "grad_norm": 0.7188818454742432,
      "learning_rate": 5.387364467247993e-06,
      "loss": 0.7304,
      "step": 17043
    },
    {
      "epoch": 0.4910966403503717,
      "grad_norm": 0.7237446904182434,
      "learning_rate": 5.386899257852276e-06,
      "loss": 0.7371,
      "step": 17044
    },
    {
      "epoch": 0.491125453812021,
      "grad_norm": 0.734290361404419,
      "learning_rate": 5.386434045087054e-06,
      "loss": 0.7189,
      "step": 17045
    },
    {
      "epoch": 0.4911542672736703,
      "grad_norm": 0.7716913819313049,
      "learning_rate": 5.385968828956378e-06,
      "loss": 0.7261,
      "step": 17046
    },
    {
      "epoch": 0.4911830807353195,
      "grad_norm": 0.7171268463134766,
      "learning_rate": 5.385503609464299e-06,
      "loss": 0.6931,
      "step": 17047
    },
    {
      "epoch": 0.4912118941969688,
      "grad_norm": 0.7099934816360474,
      "learning_rate": 5.385038386614868e-06,
      "loss": 0.6851,
      "step": 17048
    },
    {
      "epoch": 0.4912407076586181,
      "grad_norm": 0.7096533179283142,
      "learning_rate": 5.384573160412138e-06,
      "loss": 0.7197,
      "step": 17049
    },
    {
      "epoch": 0.4912695211202674,
      "grad_norm": 0.7111941576004028,
      "learning_rate": 5.3841079308601615e-06,
      "loss": 0.7364,
      "step": 17050
    },
    {
      "epoch": 0.4912983345819167,
      "grad_norm": 0.7537654638290405,
      "learning_rate": 5.383642697962988e-06,
      "loss": 0.7406,
      "step": 17051
    },
    {
      "epoch": 0.491327148043566,
      "grad_norm": 0.7536649703979492,
      "learning_rate": 5.383177461724671e-06,
      "loss": 0.7279,
      "step": 17052
    },
    {
      "epoch": 0.4913559615052152,
      "grad_norm": 0.7002754807472229,
      "learning_rate": 5.38271222214926e-06,
      "loss": 0.7369,
      "step": 17053
    },
    {
      "epoch": 0.4913847749668645,
      "grad_norm": 0.7173429131507874,
      "learning_rate": 5.382246979240808e-06,
      "loss": 0.7539,
      "step": 17054
    },
    {
      "epoch": 0.4914135884285138,
      "grad_norm": 0.732416570186615,
      "learning_rate": 5.381781733003368e-06,
      "loss": 0.72,
      "step": 17055
    },
    {
      "epoch": 0.4914424018901631,
      "grad_norm": 0.7462132573127747,
      "learning_rate": 5.38131648344099e-06,
      "loss": 0.729,
      "step": 17056
    },
    {
      "epoch": 0.4914712153518124,
      "grad_norm": 0.7439236640930176,
      "learning_rate": 5.380851230557726e-06,
      "loss": 0.7188,
      "step": 17057
    },
    {
      "epoch": 0.49150002881346166,
      "grad_norm": 0.7162737846374512,
      "learning_rate": 5.38038597435763e-06,
      "loss": 0.7048,
      "step": 17058
    },
    {
      "epoch": 0.49152884227511096,
      "grad_norm": 0.7777583003044128,
      "learning_rate": 5.379920714844752e-06,
      "loss": 0.7531,
      "step": 17059
    },
    {
      "epoch": 0.4915576557367602,
      "grad_norm": 0.7758299112319946,
      "learning_rate": 5.379455452023143e-06,
      "loss": 0.7011,
      "step": 17060
    },
    {
      "epoch": 0.4915864691984095,
      "grad_norm": 0.7532773017883301,
      "learning_rate": 5.378990185896857e-06,
      "loss": 0.7003,
      "step": 17061
    },
    {
      "epoch": 0.49161528266005877,
      "grad_norm": 0.7412499189376831,
      "learning_rate": 5.378524916469947e-06,
      "loss": 0.7216,
      "step": 17062
    },
    {
      "epoch": 0.49164409612170806,
      "grad_norm": 0.7417499423027039,
      "learning_rate": 5.378059643746463e-06,
      "loss": 0.7052,
      "step": 17063
    },
    {
      "epoch": 0.49167290958335735,
      "grad_norm": 0.7810169458389282,
      "learning_rate": 5.377594367730456e-06,
      "loss": 0.7311,
      "step": 17064
    },
    {
      "epoch": 0.49170172304500664,
      "grad_norm": 0.7742759585380554,
      "learning_rate": 5.37712908842598e-06,
      "loss": 0.7748,
      "step": 17065
    },
    {
      "epoch": 0.49173053650665594,
      "grad_norm": 0.7794603109359741,
      "learning_rate": 5.376663805837086e-06,
      "loss": 0.7283,
      "step": 17066
    },
    {
      "epoch": 0.49175934996830517,
      "grad_norm": 0.726046085357666,
      "learning_rate": 5.376198519967828e-06,
      "loss": 0.7222,
      "step": 17067
    },
    {
      "epoch": 0.49178816342995446,
      "grad_norm": 0.7172004580497742,
      "learning_rate": 5.375733230822257e-06,
      "loss": 0.7351,
      "step": 17068
    },
    {
      "epoch": 0.49181697689160375,
      "grad_norm": 0.7910158634185791,
      "learning_rate": 5.375267938404425e-06,
      "loss": 0.711,
      "step": 17069
    },
    {
      "epoch": 0.49184579035325304,
      "grad_norm": 0.7655563950538635,
      "learning_rate": 5.374802642718383e-06,
      "loss": 0.7253,
      "step": 17070
    },
    {
      "epoch": 0.49187460381490233,
      "grad_norm": 0.7322496771812439,
      "learning_rate": 5.374337343768186e-06,
      "loss": 0.7182,
      "step": 17071
    },
    {
      "epoch": 0.4919034172765516,
      "grad_norm": 0.751871645450592,
      "learning_rate": 5.373872041557884e-06,
      "loss": 0.731,
      "step": 17072
    },
    {
      "epoch": 0.4919322307382009,
      "grad_norm": 0.7263255715370178,
      "learning_rate": 5.373406736091531e-06,
      "loss": 0.7129,
      "step": 17073
    },
    {
      "epoch": 0.49196104419985015,
      "grad_norm": 0.7453421950340271,
      "learning_rate": 5.3729414273731784e-06,
      "loss": 0.7306,
      "step": 17074
    },
    {
      "epoch": 0.49198985766149944,
      "grad_norm": 0.7787954807281494,
      "learning_rate": 5.3724761154068796e-06,
      "loss": 0.7082,
      "step": 17075
    },
    {
      "epoch": 0.49201867112314873,
      "grad_norm": 0.7223753929138184,
      "learning_rate": 5.372010800196684e-06,
      "loss": 0.7192,
      "step": 17076
    },
    {
      "epoch": 0.492047484584798,
      "grad_norm": 0.7425629496574402,
      "learning_rate": 5.371545481746647e-06,
      "loss": 0.719,
      "step": 17077
    },
    {
      "epoch": 0.4920762980464473,
      "grad_norm": 0.7337983250617981,
      "learning_rate": 5.371080160060822e-06,
      "loss": 0.7256,
      "step": 17078
    },
    {
      "epoch": 0.4921051115080966,
      "grad_norm": 1.789707064628601,
      "learning_rate": 5.370614835143257e-06,
      "loss": 0.7294,
      "step": 17079
    },
    {
      "epoch": 0.49213392496974584,
      "grad_norm": 0.736303448677063,
      "learning_rate": 5.370149506998009e-06,
      "loss": 0.7393,
      "step": 17080
    },
    {
      "epoch": 0.49216273843139513,
      "grad_norm": 0.7481653094291687,
      "learning_rate": 5.36968417562913e-06,
      "loss": 0.7025,
      "step": 17081
    },
    {
      "epoch": 0.4921915518930444,
      "grad_norm": 0.7509998679161072,
      "learning_rate": 5.36921884104067e-06,
      "loss": 0.7148,
      "step": 17082
    },
    {
      "epoch": 0.4922203653546937,
      "grad_norm": 0.6872513294219971,
      "learning_rate": 5.368753503236682e-06,
      "loss": 0.7147,
      "step": 17083
    },
    {
      "epoch": 0.492249178816343,
      "grad_norm": 0.725517988204956,
      "learning_rate": 5.36828816222122e-06,
      "loss": 0.7352,
      "step": 17084
    },
    {
      "epoch": 0.4922779922779923,
      "grad_norm": 0.7203212976455688,
      "learning_rate": 5.367822817998338e-06,
      "loss": 0.7268,
      "step": 17085
    },
    {
      "epoch": 0.4923068057396416,
      "grad_norm": 0.7440850138664246,
      "learning_rate": 5.367357470572086e-06,
      "loss": 0.7125,
      "step": 17086
    },
    {
      "epoch": 0.4923356192012908,
      "grad_norm": 0.7588224411010742,
      "learning_rate": 5.366892119946517e-06,
      "loss": 0.7039,
      "step": 17087
    },
    {
      "epoch": 0.4923644326629401,
      "grad_norm": 0.7030550241470337,
      "learning_rate": 5.366426766125684e-06,
      "loss": 0.7126,
      "step": 17088
    },
    {
      "epoch": 0.4923932461245894,
      "grad_norm": 0.7060596942901611,
      "learning_rate": 5.365961409113641e-06,
      "loss": 0.6982,
      "step": 17089
    },
    {
      "epoch": 0.4924220595862387,
      "grad_norm": 0.7580251693725586,
      "learning_rate": 5.3654960489144404e-06,
      "loss": 0.7288,
      "step": 17090
    },
    {
      "epoch": 0.492450873047888,
      "grad_norm": 0.7381163835525513,
      "learning_rate": 5.365030685532135e-06,
      "loss": 0.722,
      "step": 17091
    },
    {
      "epoch": 0.4924796865095373,
      "grad_norm": 0.7516508102416992,
      "learning_rate": 5.364565318970776e-06,
      "loss": 0.7028,
      "step": 17092
    },
    {
      "epoch": 0.49250849997118656,
      "grad_norm": 0.7371724247932434,
      "learning_rate": 5.364099949234418e-06,
      "loss": 0.7207,
      "step": 17093
    },
    {
      "epoch": 0.4925373134328358,
      "grad_norm": 0.7109203934669495,
      "learning_rate": 5.363634576327112e-06,
      "loss": 0.7093,
      "step": 17094
    },
    {
      "epoch": 0.4925661268944851,
      "grad_norm": 0.7435320615768433,
      "learning_rate": 5.3631692002529145e-06,
      "loss": 0.7131,
      "step": 17095
    },
    {
      "epoch": 0.4925949403561344,
      "grad_norm": 0.7608957290649414,
      "learning_rate": 5.362703821015876e-06,
      "loss": 0.7012,
      "step": 17096
    },
    {
      "epoch": 0.49262375381778367,
      "grad_norm": 0.7654569149017334,
      "learning_rate": 5.362238438620048e-06,
      "loss": 0.7117,
      "step": 17097
    },
    {
      "epoch": 0.49265256727943296,
      "grad_norm": 0.754020631313324,
      "learning_rate": 5.361773053069487e-06,
      "loss": 0.7049,
      "step": 17098
    },
    {
      "epoch": 0.49268138074108225,
      "grad_norm": 0.7242754101753235,
      "learning_rate": 5.361307664368244e-06,
      "loss": 0.745,
      "step": 17099
    },
    {
      "epoch": 0.4927101942027315,
      "grad_norm": 0.7381918430328369,
      "learning_rate": 5.360842272520371e-06,
      "loss": 0.734,
      "step": 17100
    },
    {
      "epoch": 0.4927390076643808,
      "grad_norm": 0.7074543833732605,
      "learning_rate": 5.360376877529924e-06,
      "loss": 0.6863,
      "step": 17101
    },
    {
      "epoch": 0.49276782112603007,
      "grad_norm": 0.7347591519355774,
      "learning_rate": 5.359911479400956e-06,
      "loss": 0.6987,
      "step": 17102
    },
    {
      "epoch": 0.49279663458767936,
      "grad_norm": 0.7613651752471924,
      "learning_rate": 5.359446078137517e-06,
      "loss": 0.7271,
      "step": 17103
    },
    {
      "epoch": 0.49282544804932865,
      "grad_norm": 0.7082949876785278,
      "learning_rate": 5.358980673743662e-06,
      "loss": 0.7166,
      "step": 17104
    },
    {
      "epoch": 0.49285426151097794,
      "grad_norm": 0.7468857765197754,
      "learning_rate": 5.358515266223445e-06,
      "loss": 0.7213,
      "step": 17105
    },
    {
      "epoch": 0.49288307497262723,
      "grad_norm": 0.7571231126785278,
      "learning_rate": 5.358049855580917e-06,
      "loss": 0.7196,
      "step": 17106
    },
    {
      "epoch": 0.49291188843427647,
      "grad_norm": 0.7295190095901489,
      "learning_rate": 5.357584441820133e-06,
      "loss": 0.7215,
      "step": 17107
    },
    {
      "epoch": 0.49294070189592576,
      "grad_norm": 0.7521458268165588,
      "learning_rate": 5.357119024945147e-06,
      "loss": 0.7297,
      "step": 17108
    },
    {
      "epoch": 0.49296951535757505,
      "grad_norm": 0.7416966557502747,
      "learning_rate": 5.356653604960011e-06,
      "loss": 0.7193,
      "step": 17109
    },
    {
      "epoch": 0.49299832881922434,
      "grad_norm": 0.7607327699661255,
      "learning_rate": 5.356188181868779e-06,
      "loss": 0.7261,
      "step": 17110
    },
    {
      "epoch": 0.49302714228087363,
      "grad_norm": 0.7573742270469666,
      "learning_rate": 5.355722755675503e-06,
      "loss": 0.7416,
      "step": 17111
    },
    {
      "epoch": 0.4930559557425229,
      "grad_norm": 0.7325856685638428,
      "learning_rate": 5.355257326384237e-06,
      "loss": 0.7115,
      "step": 17112
    },
    {
      "epoch": 0.4930847692041722,
      "grad_norm": 0.7368029952049255,
      "learning_rate": 5.3547918939990365e-06,
      "loss": 0.7156,
      "step": 17113
    },
    {
      "epoch": 0.49311358266582145,
      "grad_norm": 0.7677544355392456,
      "learning_rate": 5.3543264585239515e-06,
      "loss": 0.7294,
      "step": 17114
    },
    {
      "epoch": 0.49314239612747074,
      "grad_norm": 0.7260977029800415,
      "learning_rate": 5.353861019963038e-06,
      "loss": 0.7059,
      "step": 17115
    },
    {
      "epoch": 0.49317120958912003,
      "grad_norm": 0.7354702949523926,
      "learning_rate": 5.35339557832035e-06,
      "loss": 0.7144,
      "step": 17116
    },
    {
      "epoch": 0.4932000230507693,
      "grad_norm": 0.7620967030525208,
      "learning_rate": 5.352930133599937e-06,
      "loss": 0.7242,
      "step": 17117
    },
    {
      "epoch": 0.4932288365124186,
      "grad_norm": 0.7604235410690308,
      "learning_rate": 5.3524646858058575e-06,
      "loss": 0.7135,
      "step": 17118
    },
    {
      "epoch": 0.4932576499740679,
      "grad_norm": 0.7412566542625427,
      "learning_rate": 5.351999234942161e-06,
      "loss": 0.7217,
      "step": 17119
    },
    {
      "epoch": 0.4932864634357172,
      "grad_norm": 0.7407980561256409,
      "learning_rate": 5.351533781012905e-06,
      "loss": 0.734,
      "step": 17120
    },
    {
      "epoch": 0.49331527689736643,
      "grad_norm": 0.7402373552322388,
      "learning_rate": 5.351068324022141e-06,
      "loss": 0.7369,
      "step": 17121
    },
    {
      "epoch": 0.4933440903590157,
      "grad_norm": 0.7000709772109985,
      "learning_rate": 5.350602863973923e-06,
      "loss": 0.7016,
      "step": 17122
    },
    {
      "epoch": 0.493372903820665,
      "grad_norm": 0.744878351688385,
      "learning_rate": 5.350137400872304e-06,
      "loss": 0.7221,
      "step": 17123
    },
    {
      "epoch": 0.4934017172823143,
      "grad_norm": 0.7175514101982117,
      "learning_rate": 5.3496719347213365e-06,
      "loss": 0.6905,
      "step": 17124
    },
    {
      "epoch": 0.4934305307439636,
      "grad_norm": 0.7380541563034058,
      "learning_rate": 5.349206465525079e-06,
      "loss": 0.7263,
      "step": 17125
    },
    {
      "epoch": 0.4934593442056129,
      "grad_norm": 0.7288358807563782,
      "learning_rate": 5.34874099328758e-06,
      "loss": 0.7324,
      "step": 17126
    },
    {
      "epoch": 0.4934881576672621,
      "grad_norm": 0.7383285760879517,
      "learning_rate": 5.348275518012897e-06,
      "loss": 0.7221,
      "step": 17127
    },
    {
      "epoch": 0.4935169711289114,
      "grad_norm": 0.724989652633667,
      "learning_rate": 5.3478100397050825e-06,
      "loss": 0.7214,
      "step": 17128
    },
    {
      "epoch": 0.4935457845905607,
      "grad_norm": 0.7190002202987671,
      "learning_rate": 5.34734455836819e-06,
      "loss": 0.7203,
      "step": 17129
    },
    {
      "epoch": 0.49357459805221,
      "grad_norm": 0.7178784608840942,
      "learning_rate": 5.346879074006271e-06,
      "loss": 0.6933,
      "step": 17130
    },
    {
      "epoch": 0.4936034115138593,
      "grad_norm": 0.7223835587501526,
      "learning_rate": 5.346413586623386e-06,
      "loss": 0.7238,
      "step": 17131
    },
    {
      "epoch": 0.49363222497550857,
      "grad_norm": 0.7204218506813049,
      "learning_rate": 5.3459480962235836e-06,
      "loss": 0.6994,
      "step": 17132
    },
    {
      "epoch": 0.49366103843715786,
      "grad_norm": 0.7225019335746765,
      "learning_rate": 5.345482602810919e-06,
      "loss": 0.7261,
      "step": 17133
    },
    {
      "epoch": 0.4936898518988071,
      "grad_norm": 0.7748659253120422,
      "learning_rate": 5.345017106389445e-06,
      "loss": 0.7328,
      "step": 17134
    },
    {
      "epoch": 0.4937186653604564,
      "grad_norm": 0.7261534929275513,
      "learning_rate": 5.344551606963218e-06,
      "loss": 0.6984,
      "step": 17135
    },
    {
      "epoch": 0.4937474788221057,
      "grad_norm": 0.7365970015525818,
      "learning_rate": 5.34408610453629e-06,
      "loss": 0.7169,
      "step": 17136
    },
    {
      "epoch": 0.49377629228375497,
      "grad_norm": 0.7400498986244202,
      "learning_rate": 5.343620599112718e-06,
      "loss": 0.7344,
      "step": 17137
    },
    {
      "epoch": 0.49380510574540426,
      "grad_norm": 0.7723128199577332,
      "learning_rate": 5.343155090696551e-06,
      "loss": 0.7184,
      "step": 17138
    },
    {
      "epoch": 0.49383391920705355,
      "grad_norm": 0.771740734577179,
      "learning_rate": 5.342689579291849e-06,
      "loss": 0.7266,
      "step": 17139
    },
    {
      "epoch": 0.49386273266870284,
      "grad_norm": 0.7182226777076721,
      "learning_rate": 5.342224064902662e-06,
      "loss": 0.7397,
      "step": 17140
    },
    {
      "epoch": 0.4938915461303521,
      "grad_norm": 0.7469485402107239,
      "learning_rate": 5.341758547533045e-06,
      "loss": 0.7337,
      "step": 17141
    },
    {
      "epoch": 0.49392035959200137,
      "grad_norm": 0.7991454005241394,
      "learning_rate": 5.341293027187055e-06,
      "loss": 0.7252,
      "step": 17142
    },
    {
      "epoch": 0.49394917305365066,
      "grad_norm": 0.7395808696746826,
      "learning_rate": 5.340827503868742e-06,
      "loss": 0.7093,
      "step": 17143
    },
    {
      "epoch": 0.49397798651529995,
      "grad_norm": 0.7477253079414368,
      "learning_rate": 5.3403619775821625e-06,
      "loss": 0.7296,
      "step": 17144
    },
    {
      "epoch": 0.49400679997694924,
      "grad_norm": 0.7607823014259338,
      "learning_rate": 5.339896448331371e-06,
      "loss": 0.7269,
      "step": 17145
    },
    {
      "epoch": 0.49403561343859853,
      "grad_norm": 0.7302192449569702,
      "learning_rate": 5.33943091612042e-06,
      "loss": 0.6881,
      "step": 17146
    },
    {
      "epoch": 0.4940644269002478,
      "grad_norm": 0.7642880082130432,
      "learning_rate": 5.338965380953367e-06,
      "loss": 0.7062,
      "step": 17147
    },
    {
      "epoch": 0.49409324036189706,
      "grad_norm": 0.7636671662330627,
      "learning_rate": 5.338499842834261e-06,
      "loss": 0.7003,
      "step": 17148
    },
    {
      "epoch": 0.49412205382354635,
      "grad_norm": 0.7251157760620117,
      "learning_rate": 5.338034301767163e-06,
      "loss": 0.6925,
      "step": 17149
    },
    {
      "epoch": 0.49415086728519564,
      "grad_norm": 0.7580877542495728,
      "learning_rate": 5.337568757756123e-06,
      "loss": 0.7322,
      "step": 17150
    },
    {
      "epoch": 0.49417968074684493,
      "grad_norm": 0.8353968262672424,
      "learning_rate": 5.337103210805197e-06,
      "loss": 0.7069,
      "step": 17151
    },
    {
      "epoch": 0.4942084942084942,
      "grad_norm": 0.7361034750938416,
      "learning_rate": 5.336637660918439e-06,
      "loss": 0.7247,
      "step": 17152
    },
    {
      "epoch": 0.4942373076701435,
      "grad_norm": 0.7201599478721619,
      "learning_rate": 5.336172108099903e-06,
      "loss": 0.7117,
      "step": 17153
    },
    {
      "epoch": 0.49426612113179275,
      "grad_norm": 0.7042116522789001,
      "learning_rate": 5.3357065523536435e-06,
      "loss": 0.7188,
      "step": 17154
    },
    {
      "epoch": 0.49429493459344204,
      "grad_norm": 0.7792904376983643,
      "learning_rate": 5.335240993683716e-06,
      "loss": 0.6989,
      "step": 17155
    },
    {
      "epoch": 0.4943237480550913,
      "grad_norm": 0.7740253806114197,
      "learning_rate": 5.334775432094175e-06,
      "loss": 0.723,
      "step": 17156
    },
    {
      "epoch": 0.4943525615167406,
      "grad_norm": 0.7369811534881592,
      "learning_rate": 5.3343098675890735e-06,
      "loss": 0.6942,
      "step": 17157
    },
    {
      "epoch": 0.4943813749783899,
      "grad_norm": 0.7501214742660522,
      "learning_rate": 5.3338443001724685e-06,
      "loss": 0.7214,
      "step": 17158
    },
    {
      "epoch": 0.4944101884400392,
      "grad_norm": 0.7601122260093689,
      "learning_rate": 5.333378729848413e-06,
      "loss": 0.7284,
      "step": 17159
    },
    {
      "epoch": 0.4944390019016885,
      "grad_norm": 0.7736889719963074,
      "learning_rate": 5.332913156620962e-06,
      "loss": 0.7039,
      "step": 17160
    },
    {
      "epoch": 0.4944678153633377,
      "grad_norm": 0.7757100462913513,
      "learning_rate": 5.3324475804941685e-06,
      "loss": 0.7315,
      "step": 17161
    },
    {
      "epoch": 0.494496628824987,
      "grad_norm": 0.7301886677742004,
      "learning_rate": 5.3319820014720915e-06,
      "loss": 0.7124,
      "step": 17162
    },
    {
      "epoch": 0.4945254422866363,
      "grad_norm": 0.7328211665153503,
      "learning_rate": 5.331516419558782e-06,
      "loss": 0.7216,
      "step": 17163
    },
    {
      "epoch": 0.4945542557482856,
      "grad_norm": 0.7129436135292053,
      "learning_rate": 5.331050834758297e-06,
      "loss": 0.7009,
      "step": 17164
    },
    {
      "epoch": 0.4945830692099349,
      "grad_norm": 0.7641342282295227,
      "learning_rate": 5.330585247074689e-06,
      "loss": 0.7139,
      "step": 17165
    },
    {
      "epoch": 0.4946118826715842,
      "grad_norm": 0.7205352783203125,
      "learning_rate": 5.330119656512013e-06,
      "loss": 0.6933,
      "step": 17166
    },
    {
      "epoch": 0.49464069613323347,
      "grad_norm": 0.7437208890914917,
      "learning_rate": 5.329654063074326e-06,
      "loss": 0.6946,
      "step": 17167
    },
    {
      "epoch": 0.4946695095948827,
      "grad_norm": 0.7341585159301758,
      "learning_rate": 5.329188466765682e-06,
      "loss": 0.7262,
      "step": 17168
    },
    {
      "epoch": 0.494698323056532,
      "grad_norm": 0.7604316473007202,
      "learning_rate": 5.328722867590136e-06,
      "loss": 0.694,
      "step": 17169
    },
    {
      "epoch": 0.4947271365181813,
      "grad_norm": 0.8271849155426025,
      "learning_rate": 5.328257265551742e-06,
      "loss": 0.7395,
      "step": 17170
    },
    {
      "epoch": 0.4947559499798306,
      "grad_norm": 0.7281823754310608,
      "learning_rate": 5.327791660654555e-06,
      "loss": 0.72,
      "step": 17171
    },
    {
      "epoch": 0.49478476344147987,
      "grad_norm": 0.7409632205963135,
      "learning_rate": 5.327326052902629e-06,
      "loss": 0.7503,
      "step": 17172
    },
    {
      "epoch": 0.49481357690312916,
      "grad_norm": 0.7358262538909912,
      "learning_rate": 5.326860442300022e-06,
      "loss": 0.7035,
      "step": 17173
    },
    {
      "epoch": 0.49484239036477845,
      "grad_norm": 0.7704005241394043,
      "learning_rate": 5.326394828850786e-06,
      "loss": 0.728,
      "step": 17174
    },
    {
      "epoch": 0.4948712038264277,
      "grad_norm": 0.7486236691474915,
      "learning_rate": 5.325929212558979e-06,
      "loss": 0.7178,
      "step": 17175
    },
    {
      "epoch": 0.494900017288077,
      "grad_norm": 0.734446108341217,
      "learning_rate": 5.3254635934286525e-06,
      "loss": 0.749,
      "step": 17176
    },
    {
      "epoch": 0.49492883074972627,
      "grad_norm": 0.7264869809150696,
      "learning_rate": 5.324997971463864e-06,
      "loss": 0.7274,
      "step": 17177
    },
    {
      "epoch": 0.49495764421137556,
      "grad_norm": 0.7331568002700806,
      "learning_rate": 5.324532346668668e-06,
      "loss": 0.7336,
      "step": 17178
    },
    {
      "epoch": 0.49498645767302485,
      "grad_norm": 0.7428113222122192,
      "learning_rate": 5.324066719047121e-06,
      "loss": 0.7372,
      "step": 17179
    },
    {
      "epoch": 0.49501527113467414,
      "grad_norm": 0.7228140234947205,
      "learning_rate": 5.323601088603275e-06,
      "loss": 0.7404,
      "step": 17180
    },
    {
      "epoch": 0.4950440845963234,
      "grad_norm": 0.7193430066108704,
      "learning_rate": 5.3231354553411886e-06,
      "loss": 0.6963,
      "step": 17181
    },
    {
      "epoch": 0.49507289805797267,
      "grad_norm": 0.7214769124984741,
      "learning_rate": 5.322669819264915e-06,
      "loss": 0.7071,
      "step": 17182
    },
    {
      "epoch": 0.49510171151962196,
      "grad_norm": 0.7361896634101868,
      "learning_rate": 5.322204180378509e-06,
      "loss": 0.7031,
      "step": 17183
    },
    {
      "epoch": 0.49513052498127125,
      "grad_norm": 0.7475284934043884,
      "learning_rate": 5.321738538686027e-06,
      "loss": 0.7307,
      "step": 17184
    },
    {
      "epoch": 0.49515933844292054,
      "grad_norm": 0.7106422185897827,
      "learning_rate": 5.321272894191525e-06,
      "loss": 0.7011,
      "step": 17185
    },
    {
      "epoch": 0.49518815190456983,
      "grad_norm": 0.7344005703926086,
      "learning_rate": 5.320807246899055e-06,
      "loss": 0.7231,
      "step": 17186
    },
    {
      "epoch": 0.4952169653662191,
      "grad_norm": 0.7244041562080383,
      "learning_rate": 5.320341596812677e-06,
      "loss": 0.7122,
      "step": 17187
    },
    {
      "epoch": 0.49524577882786835,
      "grad_norm": 0.7197619080543518,
      "learning_rate": 5.319875943936443e-06,
      "loss": 0.7207,
      "step": 17188
    },
    {
      "epoch": 0.49527459228951765,
      "grad_norm": 0.7112480401992798,
      "learning_rate": 5.31941028827441e-06,
      "loss": 0.7114,
      "step": 17189
    },
    {
      "epoch": 0.49530340575116694,
      "grad_norm": 0.7416876554489136,
      "learning_rate": 5.318944629830631e-06,
      "loss": 0.7181,
      "step": 17190
    },
    {
      "epoch": 0.4953322192128162,
      "grad_norm": 0.6971860527992249,
      "learning_rate": 5.318478968609164e-06,
      "loss": 0.7045,
      "step": 17191
    },
    {
      "epoch": 0.4953610326744655,
      "grad_norm": 0.7264297604560852,
      "learning_rate": 5.318013304614064e-06,
      "loss": 0.7132,
      "step": 17192
    },
    {
      "epoch": 0.4953898461361148,
      "grad_norm": 0.7738998532295227,
      "learning_rate": 5.317547637849386e-06,
      "loss": 0.7354,
      "step": 17193
    },
    {
      "epoch": 0.4954186595977641,
      "grad_norm": 0.7060667276382446,
      "learning_rate": 5.317081968319186e-06,
      "loss": 0.7075,
      "step": 17194
    },
    {
      "epoch": 0.49544747305941333,
      "grad_norm": 0.7130209803581238,
      "learning_rate": 5.316616296027516e-06,
      "loss": 0.7139,
      "step": 17195
    },
    {
      "epoch": 0.4954762865210626,
      "grad_norm": 0.7118685245513916,
      "learning_rate": 5.3161506209784374e-06,
      "loss": 0.7031,
      "step": 17196
    },
    {
      "epoch": 0.4955050999827119,
      "grad_norm": 0.7191215753555298,
      "learning_rate": 5.315684943176001e-06,
      "loss": 0.7111,
      "step": 17197
    },
    {
      "epoch": 0.4955339134443612,
      "grad_norm": 0.7139492034912109,
      "learning_rate": 5.315219262624266e-06,
      "loss": 0.7268,
      "step": 17198
    },
    {
      "epoch": 0.4955627269060105,
      "grad_norm": 0.7004846930503845,
      "learning_rate": 5.314753579327286e-06,
      "loss": 0.7045,
      "step": 17199
    },
    {
      "epoch": 0.4955915403676598,
      "grad_norm": 0.7362034320831299,
      "learning_rate": 5.314287893289116e-06,
      "loss": 0.7229,
      "step": 17200
    },
    {
      "epoch": 0.4956203538293091,
      "grad_norm": 0.723551869392395,
      "learning_rate": 5.313822204513815e-06,
      "loss": 0.7128,
      "step": 17201
    },
    {
      "epoch": 0.4956491672909583,
      "grad_norm": 0.7406463623046875,
      "learning_rate": 5.313356513005433e-06,
      "loss": 0.722,
      "step": 17202
    },
    {
      "epoch": 0.4956779807526076,
      "grad_norm": 0.7327523827552795,
      "learning_rate": 5.312890818768031e-06,
      "loss": 0.7149,
      "step": 17203
    },
    {
      "epoch": 0.4957067942142569,
      "grad_norm": 0.7628555297851562,
      "learning_rate": 5.312425121805663e-06,
      "loss": 0.7514,
      "step": 17204
    },
    {
      "epoch": 0.4957356076759062,
      "grad_norm": 0.7132156491279602,
      "learning_rate": 5.311959422122383e-06,
      "loss": 0.7131,
      "step": 17205
    },
    {
      "epoch": 0.4957644211375555,
      "grad_norm": 0.7190071940422058,
      "learning_rate": 5.31149371972225e-06,
      "loss": 0.7062,
      "step": 17206
    },
    {
      "epoch": 0.49579323459920477,
      "grad_norm": 0.7246820330619812,
      "learning_rate": 5.311028014609319e-06,
      "loss": 0.6873,
      "step": 17207
    },
    {
      "epoch": 0.495822048060854,
      "grad_norm": 0.7315729856491089,
      "learning_rate": 5.310562306787641e-06,
      "loss": 0.7272,
      "step": 17208
    },
    {
      "epoch": 0.4958508615225033,
      "grad_norm": 0.7447746396064758,
      "learning_rate": 5.3100965962612785e-06,
      "loss": 0.7012,
      "step": 17209
    },
    {
      "epoch": 0.4958796749841526,
      "grad_norm": 0.7139427065849304,
      "learning_rate": 5.309630883034285e-06,
      "loss": 0.7094,
      "step": 17210
    },
    {
      "epoch": 0.4959084884458019,
      "grad_norm": 0.7211126685142517,
      "learning_rate": 5.309165167110715e-06,
      "loss": 0.7275,
      "step": 17211
    },
    {
      "epoch": 0.49593730190745117,
      "grad_norm": 0.7155309915542603,
      "learning_rate": 5.308699448494625e-06,
      "loss": 0.7273,
      "step": 17212
    },
    {
      "epoch": 0.49596611536910046,
      "grad_norm": 0.7239570617675781,
      "learning_rate": 5.308233727190073e-06,
      "loss": 0.7142,
      "step": 17213
    },
    {
      "epoch": 0.49599492883074975,
      "grad_norm": 0.7328422665596008,
      "learning_rate": 5.307768003201111e-06,
      "loss": 0.7139,
      "step": 17214
    },
    {
      "epoch": 0.496023742292399,
      "grad_norm": 0.7288126349449158,
      "learning_rate": 5.3073022765318e-06,
      "loss": 0.743,
      "step": 17215
    },
    {
      "epoch": 0.4960525557540483,
      "grad_norm": 0.7167534232139587,
      "learning_rate": 5.306836547186191e-06,
      "loss": 0.7024,
      "step": 17216
    },
    {
      "epoch": 0.49608136921569757,
      "grad_norm": 0.7186635732650757,
      "learning_rate": 5.306370815168344e-06,
      "loss": 0.728,
      "step": 17217
    },
    {
      "epoch": 0.49611018267734686,
      "grad_norm": 0.6992359757423401,
      "learning_rate": 5.305905080482312e-06,
      "loss": 0.6986,
      "step": 17218
    },
    {
      "epoch": 0.49613899613899615,
      "grad_norm": 0.7289332747459412,
      "learning_rate": 5.305439343132153e-06,
      "loss": 0.7115,
      "step": 17219
    },
    {
      "epoch": 0.49616780960064544,
      "grad_norm": 0.7091976404190063,
      "learning_rate": 5.304973603121924e-06,
      "loss": 0.7116,
      "step": 17220
    },
    {
      "epoch": 0.49619662306229473,
      "grad_norm": 0.7121517062187195,
      "learning_rate": 5.304507860455677e-06,
      "loss": 0.7477,
      "step": 17221
    },
    {
      "epoch": 0.49622543652394396,
      "grad_norm": 0.7315922379493713,
      "learning_rate": 5.304042115137474e-06,
      "loss": 0.7228,
      "step": 17222
    },
    {
      "epoch": 0.49625424998559325,
      "grad_norm": 0.7181284427642822,
      "learning_rate": 5.303576367171367e-06,
      "loss": 0.7137,
      "step": 17223
    },
    {
      "epoch": 0.49628306344724255,
      "grad_norm": 0.7295510768890381,
      "learning_rate": 5.303110616561413e-06,
      "loss": 0.7096,
      "step": 17224
    },
    {
      "epoch": 0.49631187690889184,
      "grad_norm": 0.768355131149292,
      "learning_rate": 5.30264486331167e-06,
      "loss": 0.719,
      "step": 17225
    },
    {
      "epoch": 0.4963406903705411,
      "grad_norm": 0.7393004894256592,
      "learning_rate": 5.302179107426191e-06,
      "loss": 0.6941,
      "step": 17226
    },
    {
      "epoch": 0.4963695038321904,
      "grad_norm": 0.7247121334075928,
      "learning_rate": 5.301713348909034e-06,
      "loss": 0.7097,
      "step": 17227
    },
    {
      "epoch": 0.4963983172938397,
      "grad_norm": 0.7051732540130615,
      "learning_rate": 5.301247587764257e-06,
      "loss": 0.7005,
      "step": 17228
    },
    {
      "epoch": 0.49642713075548894,
      "grad_norm": 0.7143356800079346,
      "learning_rate": 5.300781823995913e-06,
      "loss": 0.7279,
      "step": 17229
    },
    {
      "epoch": 0.49645594421713823,
      "grad_norm": 0.7158060073852539,
      "learning_rate": 5.30031605760806e-06,
      "loss": 0.72,
      "step": 17230
    },
    {
      "epoch": 0.4964847576787875,
      "grad_norm": 0.7456088662147522,
      "learning_rate": 5.299850288604755e-06,
      "loss": 0.7468,
      "step": 17231
    },
    {
      "epoch": 0.4965135711404368,
      "grad_norm": 0.7275862693786621,
      "learning_rate": 5.299384516990053e-06,
      "loss": 0.7235,
      "step": 17232
    },
    {
      "epoch": 0.4965423846020861,
      "grad_norm": 0.7278134822845459,
      "learning_rate": 5.298918742768012e-06,
      "loss": 0.7252,
      "step": 17233
    },
    {
      "epoch": 0.4965711980637354,
      "grad_norm": 0.7036857008934021,
      "learning_rate": 5.298452965942687e-06,
      "loss": 0.7214,
      "step": 17234
    },
    {
      "epoch": 0.49660001152538463,
      "grad_norm": 0.6978358626365662,
      "learning_rate": 5.297987186518135e-06,
      "loss": 0.715,
      "step": 17235
    },
    {
      "epoch": 0.4966288249870339,
      "grad_norm": 0.7456051111221313,
      "learning_rate": 5.2975214044984135e-06,
      "loss": 0.7268,
      "step": 17236
    },
    {
      "epoch": 0.4966576384486832,
      "grad_norm": 0.7095485329627991,
      "learning_rate": 5.2970556198875776e-06,
      "loss": 0.7133,
      "step": 17237
    },
    {
      "epoch": 0.4966864519103325,
      "grad_norm": 0.7184340953826904,
      "learning_rate": 5.296589832689683e-06,
      "loss": 0.7201,
      "step": 17238
    },
    {
      "epoch": 0.4967152653719818,
      "grad_norm": 0.7005723118782043,
      "learning_rate": 5.296124042908788e-06,
      "loss": 0.7055,
      "step": 17239
    },
    {
      "epoch": 0.4967440788336311,
      "grad_norm": 0.7119159698486328,
      "learning_rate": 5.295658250548949e-06,
      "loss": 0.7052,
      "step": 17240
    },
    {
      "epoch": 0.4967728922952804,
      "grad_norm": 0.7486398816108704,
      "learning_rate": 5.295192455614223e-06,
      "loss": 0.7499,
      "step": 17241
    },
    {
      "epoch": 0.4968017057569296,
      "grad_norm": 0.7423017621040344,
      "learning_rate": 5.294726658108665e-06,
      "loss": 0.7374,
      "step": 17242
    },
    {
      "epoch": 0.4968305192185789,
      "grad_norm": 0.7202560305595398,
      "learning_rate": 5.294260858036333e-06,
      "loss": 0.696,
      "step": 17243
    },
    {
      "epoch": 0.4968593326802282,
      "grad_norm": 0.7579359412193298,
      "learning_rate": 5.293795055401281e-06,
      "loss": 0.7153,
      "step": 17244
    },
    {
      "epoch": 0.4968881461418775,
      "grad_norm": 0.7270929217338562,
      "learning_rate": 5.2933292502075695e-06,
      "loss": 0.7041,
      "step": 17245
    },
    {
      "epoch": 0.4969169596035268,
      "grad_norm": 0.7174429297447205,
      "learning_rate": 5.292863442459253e-06,
      "loss": 0.7071,
      "step": 17246
    },
    {
      "epoch": 0.49694577306517607,
      "grad_norm": 0.7320489883422852,
      "learning_rate": 5.2923976321603885e-06,
      "loss": 0.723,
      "step": 17247
    },
    {
      "epoch": 0.49697458652682536,
      "grad_norm": 0.7185453176498413,
      "learning_rate": 5.291931819315034e-06,
      "loss": 0.709,
      "step": 17248
    },
    {
      "epoch": 0.4970033999884746,
      "grad_norm": 0.7369484305381775,
      "learning_rate": 5.291466003927245e-06,
      "loss": 0.7146,
      "step": 17249
    },
    {
      "epoch": 0.4970322134501239,
      "grad_norm": 0.7224730253219604,
      "learning_rate": 5.291000186001076e-06,
      "loss": 0.7006,
      "step": 17250
    },
    {
      "epoch": 0.4970610269117732,
      "grad_norm": 0.7794941663742065,
      "learning_rate": 5.29053436554059e-06,
      "loss": 0.7207,
      "step": 17251
    },
    {
      "epoch": 0.49708984037342246,
      "grad_norm": 0.7214218974113464,
      "learning_rate": 5.290068542549838e-06,
      "loss": 0.7162,
      "step": 17252
    },
    {
      "epoch": 0.49711865383507176,
      "grad_norm": 0.7406778335571289,
      "learning_rate": 5.289602717032879e-06,
      "loss": 0.7371,
      "step": 17253
    },
    {
      "epoch": 0.49714746729672105,
      "grad_norm": 0.7243610620498657,
      "learning_rate": 5.2891368889937686e-06,
      "loss": 0.7283,
      "step": 17254
    },
    {
      "epoch": 0.49717628075837034,
      "grad_norm": 0.7589640617370605,
      "learning_rate": 5.288671058436567e-06,
      "loss": 0.7296,
      "step": 17255
    },
    {
      "epoch": 0.49720509422001957,
      "grad_norm": 0.8487892746925354,
      "learning_rate": 5.288205225365328e-06,
      "loss": 0.7375,
      "step": 17256
    },
    {
      "epoch": 0.49723390768166886,
      "grad_norm": 0.7395216226577759,
      "learning_rate": 5.28773938978411e-06,
      "loss": 0.7276,
      "step": 17257
    },
    {
      "epoch": 0.49726272114331815,
      "grad_norm": 0.7275926470756531,
      "learning_rate": 5.28727355169697e-06,
      "loss": 0.6963,
      "step": 17258
    },
    {
      "epoch": 0.49729153460496744,
      "grad_norm": 0.757267415523529,
      "learning_rate": 5.286807711107964e-06,
      "loss": 0.7327,
      "step": 17259
    },
    {
      "epoch": 0.49732034806661674,
      "grad_norm": 0.7627456188201904,
      "learning_rate": 5.28634186802115e-06,
      "loss": 0.7327,
      "step": 17260
    },
    {
      "epoch": 0.497349161528266,
      "grad_norm": 0.71719890832901,
      "learning_rate": 5.2858760224405835e-06,
      "loss": 0.7022,
      "step": 17261
    },
    {
      "epoch": 0.49737797498991526,
      "grad_norm": 0.7407308220863342,
      "learning_rate": 5.285410174370322e-06,
      "loss": 0.7427,
      "step": 17262
    },
    {
      "epoch": 0.49740678845156455,
      "grad_norm": 0.7559714317321777,
      "learning_rate": 5.284944323814425e-06,
      "loss": 0.7313,
      "step": 17263
    },
    {
      "epoch": 0.49743560191321384,
      "grad_norm": 0.7428582906723022,
      "learning_rate": 5.284478470776948e-06,
      "loss": 0.7019,
      "step": 17264
    },
    {
      "epoch": 0.49746441537486313,
      "grad_norm": 0.7171533703804016,
      "learning_rate": 5.284012615261948e-06,
      "loss": 0.7008,
      "step": 17265
    },
    {
      "epoch": 0.4974932288365124,
      "grad_norm": 0.7188900113105774,
      "learning_rate": 5.28354675727348e-06,
      "loss": 0.6824,
      "step": 17266
    },
    {
      "epoch": 0.4975220422981617,
      "grad_norm": 0.7316949367523193,
      "learning_rate": 5.283080896815606e-06,
      "loss": 0.7178,
      "step": 17267
    },
    {
      "epoch": 0.497550855759811,
      "grad_norm": 0.728411078453064,
      "learning_rate": 5.282615033892376e-06,
      "loss": 0.7047,
      "step": 17268
    },
    {
      "epoch": 0.49757966922146024,
      "grad_norm": 0.7289319634437561,
      "learning_rate": 5.282149168507855e-06,
      "loss": 0.7096,
      "step": 17269
    },
    {
      "epoch": 0.49760848268310953,
      "grad_norm": 0.7489740252494812,
      "learning_rate": 5.281683300666096e-06,
      "loss": 0.7167,
      "step": 17270
    },
    {
      "epoch": 0.4976372961447588,
      "grad_norm": 0.7317250370979309,
      "learning_rate": 5.281217430371158e-06,
      "loss": 0.6764,
      "step": 17271
    },
    {
      "epoch": 0.4976661096064081,
      "grad_norm": 0.737343430519104,
      "learning_rate": 5.2807515576270964e-06,
      "loss": 0.7146,
      "step": 17272
    },
    {
      "epoch": 0.4976949230680574,
      "grad_norm": 0.7153491377830505,
      "learning_rate": 5.280285682437969e-06,
      "loss": 0.6859,
      "step": 17273
    },
    {
      "epoch": 0.4977237365297067,
      "grad_norm": 0.7200354933738708,
      "learning_rate": 5.279819804807834e-06,
      "loss": 0.7387,
      "step": 17274
    },
    {
      "epoch": 0.497752549991356,
      "grad_norm": 0.7417486906051636,
      "learning_rate": 5.279353924740748e-06,
      "loss": 0.7375,
      "step": 17275
    },
    {
      "epoch": 0.4977813634530052,
      "grad_norm": 0.7211507558822632,
      "learning_rate": 5.2788880422407674e-06,
      "loss": 0.7081,
      "step": 17276
    },
    {
      "epoch": 0.4978101769146545,
      "grad_norm": 0.7312631011009216,
      "learning_rate": 5.278422157311953e-06,
      "loss": 0.7101,
      "step": 17277
    },
    {
      "epoch": 0.4978389903763038,
      "grad_norm": 0.7641619443893433,
      "learning_rate": 5.277956269958358e-06,
      "loss": 0.7171,
      "step": 17278
    },
    {
      "epoch": 0.4978678038379531,
      "grad_norm": 0.7357386946678162,
      "learning_rate": 5.2774903801840434e-06,
      "loss": 0.7281,
      "step": 17279
    },
    {
      "epoch": 0.4978966172996024,
      "grad_norm": 0.7151428461074829,
      "learning_rate": 5.277024487993063e-06,
      "loss": 0.7204,
      "step": 17280
    },
    {
      "epoch": 0.4979254307612517,
      "grad_norm": 0.7318550944328308,
      "learning_rate": 5.276558593389478e-06,
      "loss": 0.7245,
      "step": 17281
    },
    {
      "epoch": 0.49795424422290097,
      "grad_norm": 0.7283624410629272,
      "learning_rate": 5.276092696377344e-06,
      "loss": 0.7067,
      "step": 17282
    },
    {
      "epoch": 0.4979830576845502,
      "grad_norm": 0.7290146350860596,
      "learning_rate": 5.2756267969607185e-06,
      "loss": 0.7352,
      "step": 17283
    },
    {
      "epoch": 0.4980118711461995,
      "grad_norm": 0.7578892111778259,
      "learning_rate": 5.275160895143659e-06,
      "loss": 0.7283,
      "step": 17284
    },
    {
      "epoch": 0.4980406846078488,
      "grad_norm": 0.7252248525619507,
      "learning_rate": 5.274694990930224e-06,
      "loss": 0.7053,
      "step": 17285
    },
    {
      "epoch": 0.4980694980694981,
      "grad_norm": 0.7482388615608215,
      "learning_rate": 5.274229084324468e-06,
      "loss": 0.7223,
      "step": 17286
    },
    {
      "epoch": 0.49809831153114736,
      "grad_norm": 0.721612274646759,
      "learning_rate": 5.273763175330452e-06,
      "loss": 0.7142,
      "step": 17287
    },
    {
      "epoch": 0.49812712499279665,
      "grad_norm": 0.7355064153671265,
      "learning_rate": 5.273297263952233e-06,
      "loss": 0.7372,
      "step": 17288
    },
    {
      "epoch": 0.4981559384544459,
      "grad_norm": 0.7303947806358337,
      "learning_rate": 5.2728313501938665e-06,
      "loss": 0.7069,
      "step": 17289
    },
    {
      "epoch": 0.4981847519160952,
      "grad_norm": 0.7377131581306458,
      "learning_rate": 5.272365434059413e-06,
      "loss": 0.7313,
      "step": 17290
    },
    {
      "epoch": 0.49821356537774447,
      "grad_norm": 0.7358893156051636,
      "learning_rate": 5.27189951555293e-06,
      "loss": 0.72,
      "step": 17291
    },
    {
      "epoch": 0.49824237883939376,
      "grad_norm": 0.7080510854721069,
      "learning_rate": 5.271433594678471e-06,
      "loss": 0.7181,
      "step": 17292
    },
    {
      "epoch": 0.49827119230104305,
      "grad_norm": 0.7141253352165222,
      "learning_rate": 5.270967671440098e-06,
      "loss": 0.7133,
      "step": 17293
    },
    {
      "epoch": 0.49830000576269234,
      "grad_norm": 0.7882035970687866,
      "learning_rate": 5.270501745841868e-06,
      "loss": 0.7184,
      "step": 17294
    },
    {
      "epoch": 0.49832881922434163,
      "grad_norm": 0.7285835146903992,
      "learning_rate": 5.270035817887836e-06,
      "loss": 0.7091,
      "step": 17295
    },
    {
      "epoch": 0.49835763268599087,
      "grad_norm": 0.7620677947998047,
      "learning_rate": 5.269569887582064e-06,
      "loss": 0.7182,
      "step": 17296
    },
    {
      "epoch": 0.49838644614764016,
      "grad_norm": 0.7297558784484863,
      "learning_rate": 5.269103954928608e-06,
      "loss": 0.71,
      "step": 17297
    },
    {
      "epoch": 0.49841525960928945,
      "grad_norm": 0.7646224498748779,
      "learning_rate": 5.268638019931524e-06,
      "loss": 0.7339,
      "step": 17298
    },
    {
      "epoch": 0.49844407307093874,
      "grad_norm": 0.7436490058898926,
      "learning_rate": 5.268172082594874e-06,
      "loss": 0.7042,
      "step": 17299
    },
    {
      "epoch": 0.49847288653258803,
      "grad_norm": 0.722023069858551,
      "learning_rate": 5.267706142922712e-06,
      "loss": 0.7382,
      "step": 17300
    },
    {
      "epoch": 0.4985016999942373,
      "grad_norm": 0.7408053874969482,
      "learning_rate": 5.267240200919097e-06,
      "loss": 0.708,
      "step": 17301
    },
    {
      "epoch": 0.4985305134558866,
      "grad_norm": 0.7120302319526672,
      "learning_rate": 5.266774256588086e-06,
      "loss": 0.7376,
      "step": 17302
    },
    {
      "epoch": 0.49855932691753585,
      "grad_norm": 0.7144030332565308,
      "learning_rate": 5.26630830993374e-06,
      "loss": 0.7252,
      "step": 17303
    },
    {
      "epoch": 0.49858814037918514,
      "grad_norm": 0.7365641593933105,
      "learning_rate": 5.265842360960112e-06,
      "loss": 0.7303,
      "step": 17304
    },
    {
      "epoch": 0.49861695384083443,
      "grad_norm": 0.72240149974823,
      "learning_rate": 5.265376409671264e-06,
      "loss": 0.7292,
      "step": 17305
    },
    {
      "epoch": 0.4986457673024837,
      "grad_norm": 0.7197166085243225,
      "learning_rate": 5.264910456071254e-06,
      "loss": 0.7121,
      "step": 17306
    },
    {
      "epoch": 0.498674580764133,
      "grad_norm": 0.7359668612480164,
      "learning_rate": 5.264444500164139e-06,
      "loss": 0.7278,
      "step": 17307
    },
    {
      "epoch": 0.4987033942257823,
      "grad_norm": 0.714820921421051,
      "learning_rate": 5.263978541953975e-06,
      "loss": 0.6947,
      "step": 17308
    },
    {
      "epoch": 0.4987322076874316,
      "grad_norm": 0.7267196774482727,
      "learning_rate": 5.263512581444823e-06,
      "loss": 0.7066,
      "step": 17309
    },
    {
      "epoch": 0.49876102114908083,
      "grad_norm": 0.7362599968910217,
      "learning_rate": 5.263046618640738e-06,
      "loss": 0.7084,
      "step": 17310
    },
    {
      "epoch": 0.4987898346107301,
      "grad_norm": 0.7217142581939697,
      "learning_rate": 5.262580653545782e-06,
      "loss": 0.7383,
      "step": 17311
    },
    {
      "epoch": 0.4988186480723794,
      "grad_norm": 0.7218102812767029,
      "learning_rate": 5.262114686164009e-06,
      "loss": 0.7114,
      "step": 17312
    },
    {
      "epoch": 0.4988474615340287,
      "grad_norm": 0.7206340432167053,
      "learning_rate": 5.26164871649948e-06,
      "loss": 0.6945,
      "step": 17313
    },
    {
      "epoch": 0.498876274995678,
      "grad_norm": 0.735933780670166,
      "learning_rate": 5.261182744556252e-06,
      "loss": 0.7459,
      "step": 17314
    },
    {
      "epoch": 0.4989050884573273,
      "grad_norm": 0.7394384741783142,
      "learning_rate": 5.2607167703383845e-06,
      "loss": 0.7228,
      "step": 17315
    },
    {
      "epoch": 0.4989339019189765,
      "grad_norm": 0.7178558707237244,
      "learning_rate": 5.260250793849932e-06,
      "loss": 0.7226,
      "step": 17316
    },
    {
      "epoch": 0.4989627153806258,
      "grad_norm": 0.7570888996124268,
      "learning_rate": 5.259784815094958e-06,
      "loss": 0.687,
      "step": 17317
    },
    {
      "epoch": 0.4989915288422751,
      "grad_norm": 0.7170776128768921,
      "learning_rate": 5.259318834077518e-06,
      "loss": 0.7219,
      "step": 17318
    },
    {
      "epoch": 0.4990203423039244,
      "grad_norm": 0.7000835537910461,
      "learning_rate": 5.258852850801669e-06,
      "loss": 0.7371,
      "step": 17319
    },
    {
      "epoch": 0.4990491557655737,
      "grad_norm": 0.7359552383422852,
      "learning_rate": 5.258386865271471e-06,
      "loss": 0.699,
      "step": 17320
    },
    {
      "epoch": 0.499077969227223,
      "grad_norm": 0.740134596824646,
      "learning_rate": 5.257920877490982e-06,
      "loss": 0.7342,
      "step": 17321
    },
    {
      "epoch": 0.49910678268887226,
      "grad_norm": 0.7659448981285095,
      "learning_rate": 5.257454887464258e-06,
      "loss": 0.7312,
      "step": 17322
    },
    {
      "epoch": 0.4991355961505215,
      "grad_norm": 0.7036277055740356,
      "learning_rate": 5.25698889519536e-06,
      "loss": 0.7226,
      "step": 17323
    },
    {
      "epoch": 0.4991644096121708,
      "grad_norm": 0.7330094575881958,
      "learning_rate": 5.2565229006883465e-06,
      "loss": 0.7386,
      "step": 17324
    },
    {
      "epoch": 0.4991932230738201,
      "grad_norm": 0.7297421097755432,
      "learning_rate": 5.2560569039472755e-06,
      "loss": 0.7121,
      "step": 17325
    },
    {
      "epoch": 0.49922203653546937,
      "grad_norm": 0.7600777745246887,
      "learning_rate": 5.255590904976205e-06,
      "loss": 0.7253,
      "step": 17326
    },
    {
      "epoch": 0.49925084999711866,
      "grad_norm": 0.7458686828613281,
      "learning_rate": 5.255124903779192e-06,
      "loss": 0.726,
      "step": 17327
    },
    {
      "epoch": 0.49927966345876795,
      "grad_norm": 0.7219098806381226,
      "learning_rate": 5.254658900360296e-06,
      "loss": 0.7292,
      "step": 17328
    },
    {
      "epoch": 0.49930847692041724,
      "grad_norm": 0.7228469848632812,
      "learning_rate": 5.254192894723576e-06,
      "loss": 0.7088,
      "step": 17329
    },
    {
      "epoch": 0.4993372903820665,
      "grad_norm": 0.7558174729347229,
      "learning_rate": 5.25372688687309e-06,
      "loss": 0.7257,
      "step": 17330
    },
    {
      "epoch": 0.49936610384371577,
      "grad_norm": 0.7337746024131775,
      "learning_rate": 5.253260876812896e-06,
      "loss": 0.7104,
      "step": 17331
    },
    {
      "epoch": 0.49939491730536506,
      "grad_norm": 0.7303023338317871,
      "learning_rate": 5.252794864547053e-06,
      "loss": 0.7214,
      "step": 17332
    },
    {
      "epoch": 0.49942373076701435,
      "grad_norm": 0.7318271398544312,
      "learning_rate": 5.25232885007962e-06,
      "loss": 0.7435,
      "step": 17333
    },
    {
      "epoch": 0.49945254422866364,
      "grad_norm": 0.7424371838569641,
      "learning_rate": 5.251862833414655e-06,
      "loss": 0.7394,
      "step": 17334
    },
    {
      "epoch": 0.49948135769031293,
      "grad_norm": 0.7303721904754639,
      "learning_rate": 5.251396814556214e-06,
      "loss": 0.7199,
      "step": 17335
    },
    {
      "epoch": 0.4995101711519622,
      "grad_norm": 0.7242385745048523,
      "learning_rate": 5.2509307935083606e-06,
      "loss": 0.7383,
      "step": 17336
    },
    {
      "epoch": 0.49953898461361146,
      "grad_norm": 0.7363427877426147,
      "learning_rate": 5.250464770275151e-06,
      "loss": 0.6789,
      "step": 17337
    },
    {
      "epoch": 0.49956779807526075,
      "grad_norm": 0.7117264866828918,
      "learning_rate": 5.249998744860644e-06,
      "loss": 0.722,
      "step": 17338
    },
    {
      "epoch": 0.49959661153691004,
      "grad_norm": 0.7417152523994446,
      "learning_rate": 5.249532717268897e-06,
      "loss": 0.7098,
      "step": 17339
    },
    {
      "epoch": 0.49962542499855933,
      "grad_norm": 0.7582288384437561,
      "learning_rate": 5.249066687503968e-06,
      "loss": 0.7052,
      "step": 17340
    },
    {
      "epoch": 0.4996542384602086,
      "grad_norm": 0.770820677280426,
      "learning_rate": 5.248600655569919e-06,
      "loss": 0.7221,
      "step": 17341
    },
    {
      "epoch": 0.4996830519218579,
      "grad_norm": 0.7417022585868835,
      "learning_rate": 5.248134621470806e-06,
      "loss": 0.7203,
      "step": 17342
    },
    {
      "epoch": 0.49971186538350715,
      "grad_norm": 0.7295033931732178,
      "learning_rate": 5.247668585210689e-06,
      "loss": 0.7122,
      "step": 17343
    },
    {
      "epoch": 0.49974067884515644,
      "grad_norm": 0.7318732738494873,
      "learning_rate": 5.247202546793626e-06,
      "loss": 0.7222,
      "step": 17344
    },
    {
      "epoch": 0.49976949230680573,
      "grad_norm": 0.7863484025001526,
      "learning_rate": 5.246736506223676e-06,
      "loss": 0.703,
      "step": 17345
    },
    {
      "epoch": 0.499798305768455,
      "grad_norm": 0.7222380042076111,
      "learning_rate": 5.246270463504898e-06,
      "loss": 0.7459,
      "step": 17346
    },
    {
      "epoch": 0.4998271192301043,
      "grad_norm": 0.732517659664154,
      "learning_rate": 5.245804418641349e-06,
      "loss": 0.7036,
      "step": 17347
    },
    {
      "epoch": 0.4998559326917536,
      "grad_norm": 0.7143408060073853,
      "learning_rate": 5.245338371637091e-06,
      "loss": 0.6948,
      "step": 17348
    },
    {
      "epoch": 0.4998847461534029,
      "grad_norm": 0.7027593851089478,
      "learning_rate": 5.244872322496181e-06,
      "loss": 0.6965,
      "step": 17349
    },
    {
      "epoch": 0.4999135596150521,
      "grad_norm": 0.7363053560256958,
      "learning_rate": 5.244406271222677e-06,
      "loss": 0.7522,
      "step": 17350
    },
    {
      "epoch": 0.4999423730767014,
      "grad_norm": 0.7222715020179749,
      "learning_rate": 5.243940217820638e-06,
      "loss": 0.7163,
      "step": 17351
    },
    {
      "epoch": 0.4999711865383507,
      "grad_norm": 0.7528979182243347,
      "learning_rate": 5.2434741622941245e-06,
      "loss": 0.7358,
      "step": 17352
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6960837841033936,
      "learning_rate": 5.243008104647194e-06,
      "loss": 0.7116,
      "step": 17353
    },
    {
      "epoch": 0.5000288134616493,
      "grad_norm": 0.7391934394836426,
      "learning_rate": 5.242542044883906e-06,
      "loss": 0.727,
      "step": 17354
    },
    {
      "epoch": 0.5000576269232986,
      "grad_norm": 0.7290162444114685,
      "learning_rate": 5.242075983008319e-06,
      "loss": 0.7163,
      "step": 17355
    },
    {
      "epoch": 0.5000864403849479,
      "grad_norm": 0.7094374299049377,
      "learning_rate": 5.241609919024493e-06,
      "loss": 0.7184,
      "step": 17356
    },
    {
      "epoch": 0.5001152538465972,
      "grad_norm": 0.7361390590667725,
      "learning_rate": 5.2411438529364855e-06,
      "loss": 0.7286,
      "step": 17357
    },
    {
      "epoch": 0.5001440673082465,
      "grad_norm": 0.7182353734970093,
      "learning_rate": 5.240677784748356e-06,
      "loss": 0.7325,
      "step": 17358
    },
    {
      "epoch": 0.5001728807698957,
      "grad_norm": 0.7148311138153076,
      "learning_rate": 5.240211714464164e-06,
      "loss": 0.6888,
      "step": 17359
    },
    {
      "epoch": 0.5002016942315449,
      "grad_norm": 0.7172682285308838,
      "learning_rate": 5.239745642087969e-06,
      "loss": 0.7234,
      "step": 17360
    },
    {
      "epoch": 0.5002305076931942,
      "grad_norm": 0.7187959551811218,
      "learning_rate": 5.239279567623827e-06,
      "loss": 0.702,
      "step": 17361
    },
    {
      "epoch": 0.5002593211548435,
      "grad_norm": 0.7092073559761047,
      "learning_rate": 5.2388134910758015e-06,
      "loss": 0.718,
      "step": 17362
    },
    {
      "epoch": 0.5002881346164928,
      "grad_norm": 0.7100566625595093,
      "learning_rate": 5.238347412447947e-06,
      "loss": 0.7252,
      "step": 17363
    },
    {
      "epoch": 0.5003169480781421,
      "grad_norm": 0.73355633020401,
      "learning_rate": 5.2378813317443255e-06,
      "loss": 0.7194,
      "step": 17364
    },
    {
      "epoch": 0.5003457615397914,
      "grad_norm": 0.7310348749160767,
      "learning_rate": 5.237415248968997e-06,
      "loss": 0.7262,
      "step": 17365
    },
    {
      "epoch": 0.5003745750014407,
      "grad_norm": 0.7586290836334229,
      "learning_rate": 5.236949164126017e-06,
      "loss": 0.7236,
      "step": 17366
    },
    {
      "epoch": 0.50040338846309,
      "grad_norm": 0.7146082520484924,
      "learning_rate": 5.236483077219448e-06,
      "loss": 0.7299,
      "step": 17367
    },
    {
      "epoch": 0.5004322019247393,
      "grad_norm": 0.7251074314117432,
      "learning_rate": 5.236016988253347e-06,
      "loss": 0.7002,
      "step": 17368
    },
    {
      "epoch": 0.5004610153863885,
      "grad_norm": 0.7438557744026184,
      "learning_rate": 5.235550897231774e-06,
      "loss": 0.7239,
      "step": 17369
    },
    {
      "epoch": 0.5004898288480378,
      "grad_norm": 0.7466745376586914,
      "learning_rate": 5.235084804158788e-06,
      "loss": 0.725,
      "step": 17370
    },
    {
      "epoch": 0.5005186423096871,
      "grad_norm": 0.7104105353355408,
      "learning_rate": 5.234618709038448e-06,
      "loss": 0.7217,
      "step": 17371
    },
    {
      "epoch": 0.5005474557713364,
      "grad_norm": 0.7192004919052124,
      "learning_rate": 5.234152611874814e-06,
      "loss": 0.7149,
      "step": 17372
    },
    {
      "epoch": 0.5005762692329857,
      "grad_norm": 0.7215008735656738,
      "learning_rate": 5.233686512671945e-06,
      "loss": 0.6962,
      "step": 17373
    },
    {
      "epoch": 0.5006050826946349,
      "grad_norm": 0.7471254467964172,
      "learning_rate": 5.2332204114339e-06,
      "loss": 0.7386,
      "step": 17374
    },
    {
      "epoch": 0.5006338961562842,
      "grad_norm": 0.7353280782699585,
      "learning_rate": 5.232754308164738e-06,
      "loss": 0.7306,
      "step": 17375
    },
    {
      "epoch": 0.5006627096179335,
      "grad_norm": 0.7329681515693665,
      "learning_rate": 5.232288202868519e-06,
      "loss": 0.7122,
      "step": 17376
    },
    {
      "epoch": 0.5006915230795828,
      "grad_norm": 0.7324192523956299,
      "learning_rate": 5.231822095549302e-06,
      "loss": 0.7316,
      "step": 17377
    },
    {
      "epoch": 0.500720336541232,
      "grad_norm": 0.7274046540260315,
      "learning_rate": 5.2313559862111465e-06,
      "loss": 0.7171,
      "step": 17378
    },
    {
      "epoch": 0.5007491500028813,
      "grad_norm": 0.7147749662399292,
      "learning_rate": 5.230889874858112e-06,
      "loss": 0.7217,
      "step": 17379
    },
    {
      "epoch": 0.5007779634645306,
      "grad_norm": 0.7265879511833191,
      "learning_rate": 5.2304237614942565e-06,
      "loss": 0.7191,
      "step": 17380
    },
    {
      "epoch": 0.5008067769261799,
      "grad_norm": 0.6884958744049072,
      "learning_rate": 5.22995764612364e-06,
      "loss": 0.705,
      "step": 17381
    },
    {
      "epoch": 0.5008355903878292,
      "grad_norm": 0.7343606948852539,
      "learning_rate": 5.229491528750323e-06,
      "loss": 0.7146,
      "step": 17382
    },
    {
      "epoch": 0.5008644038494785,
      "grad_norm": 0.7393209934234619,
      "learning_rate": 5.229025409378366e-06,
      "loss": 0.7221,
      "step": 17383
    },
    {
      "epoch": 0.5008932173111278,
      "grad_norm": 0.7249369621276855,
      "learning_rate": 5.2285592880118245e-06,
      "loss": 0.7068,
      "step": 17384
    },
    {
      "epoch": 0.5009220307727771,
      "grad_norm": 0.7229852080345154,
      "learning_rate": 5.228093164654761e-06,
      "loss": 0.6841,
      "step": 17385
    },
    {
      "epoch": 0.5009508442344264,
      "grad_norm": 0.719990074634552,
      "learning_rate": 5.227627039311233e-06,
      "loss": 0.6927,
      "step": 17386
    },
    {
      "epoch": 0.5009796576960756,
      "grad_norm": 0.7469003200531006,
      "learning_rate": 5.227160911985301e-06,
      "loss": 0.7248,
      "step": 17387
    },
    {
      "epoch": 0.5010084711577248,
      "grad_norm": 0.7181013226509094,
      "learning_rate": 5.226694782681026e-06,
      "loss": 0.7191,
      "step": 17388
    },
    {
      "epoch": 0.5010372846193741,
      "grad_norm": 0.7122728824615479,
      "learning_rate": 5.226228651402464e-06,
      "loss": 0.6938,
      "step": 17389
    },
    {
      "epoch": 0.5010660980810234,
      "grad_norm": 0.7338123321533203,
      "learning_rate": 5.225762518153679e-06,
      "loss": 0.7127,
      "step": 17390
    },
    {
      "epoch": 0.5010949115426727,
      "grad_norm": 0.7376430034637451,
      "learning_rate": 5.225296382938726e-06,
      "loss": 0.7242,
      "step": 17391
    },
    {
      "epoch": 0.501123725004322,
      "grad_norm": 0.7008709907531738,
      "learning_rate": 5.224830245761667e-06,
      "loss": 0.7184,
      "step": 17392
    },
    {
      "epoch": 0.5011525384659713,
      "grad_norm": 0.7289190292358398,
      "learning_rate": 5.2243641066265625e-06,
      "loss": 0.729,
      "step": 17393
    },
    {
      "epoch": 0.5011813519276206,
      "grad_norm": 0.715440571308136,
      "learning_rate": 5.223897965537469e-06,
      "loss": 0.7169,
      "step": 17394
    },
    {
      "epoch": 0.5012101653892699,
      "grad_norm": 0.7548215389251709,
      "learning_rate": 5.2234318224984495e-06,
      "loss": 0.6896,
      "step": 17395
    },
    {
      "epoch": 0.5012389788509192,
      "grad_norm": 0.7259418368339539,
      "learning_rate": 5.2229656775135626e-06,
      "loss": 0.7297,
      "step": 17396
    },
    {
      "epoch": 0.5012677923125685,
      "grad_norm": 0.7034233808517456,
      "learning_rate": 5.222499530586868e-06,
      "loss": 0.7206,
      "step": 17397
    },
    {
      "epoch": 0.5012966057742178,
      "grad_norm": 0.7522149682044983,
      "learning_rate": 5.222033381722423e-06,
      "loss": 0.7283,
      "step": 17398
    },
    {
      "epoch": 0.501325419235867,
      "grad_norm": 0.7610514163970947,
      "learning_rate": 5.221567230924289e-06,
      "loss": 0.7134,
      "step": 17399
    },
    {
      "epoch": 0.5013542326975163,
      "grad_norm": 0.7084966897964478,
      "learning_rate": 5.2211010781965275e-06,
      "loss": 0.7176,
      "step": 17400
    },
    {
      "epoch": 0.5013830461591655,
      "grad_norm": 0.7416119575500488,
      "learning_rate": 5.2206349235431965e-06,
      "loss": 0.7228,
      "step": 17401
    },
    {
      "epoch": 0.5014118596208148,
      "grad_norm": 0.7576121687889099,
      "learning_rate": 5.220168766968355e-06,
      "loss": 0.7234,
      "step": 17402
    },
    {
      "epoch": 0.5014406730824641,
      "grad_norm": 0.7232834696769714,
      "learning_rate": 5.219702608476065e-06,
      "loss": 0.7149,
      "step": 17403
    },
    {
      "epoch": 0.5014694865441134,
      "grad_norm": 0.7387752532958984,
      "learning_rate": 5.219236448070384e-06,
      "loss": 0.7127,
      "step": 17404
    },
    {
      "epoch": 0.5014983000057627,
      "grad_norm": 0.738699734210968,
      "learning_rate": 5.218770285755371e-06,
      "loss": 0.7094,
      "step": 17405
    },
    {
      "epoch": 0.501527113467412,
      "grad_norm": 0.7260800004005432,
      "learning_rate": 5.218304121535091e-06,
      "loss": 0.6779,
      "step": 17406
    },
    {
      "epoch": 0.5015559269290613,
      "grad_norm": 0.7295577526092529,
      "learning_rate": 5.2178379554135985e-06,
      "loss": 0.7332,
      "step": 17407
    },
    {
      "epoch": 0.5015847403907105,
      "grad_norm": 0.7195022106170654,
      "learning_rate": 5.217371787394955e-06,
      "loss": 0.7023,
      "step": 17408
    },
    {
      "epoch": 0.5016135538523598,
      "grad_norm": 0.725139319896698,
      "learning_rate": 5.216905617483221e-06,
      "loss": 0.7154,
      "step": 17409
    },
    {
      "epoch": 0.5016423673140091,
      "grad_norm": 0.7438543438911438,
      "learning_rate": 5.216439445682455e-06,
      "loss": 0.7025,
      "step": 17410
    },
    {
      "epoch": 0.5016711807756584,
      "grad_norm": 0.7420311570167542,
      "learning_rate": 5.215973271996718e-06,
      "loss": 0.7471,
      "step": 17411
    },
    {
      "epoch": 0.5016999942373077,
      "grad_norm": 0.7431056499481201,
      "learning_rate": 5.215507096430071e-06,
      "loss": 0.7099,
      "step": 17412
    },
    {
      "epoch": 0.501728807698957,
      "grad_norm": 0.7367352843284607,
      "learning_rate": 5.21504091898657e-06,
      "loss": 0.7071,
      "step": 17413
    },
    {
      "epoch": 0.5017576211606062,
      "grad_norm": 0.7224079370498657,
      "learning_rate": 5.214574739670279e-06,
      "loss": 0.7061,
      "step": 17414
    },
    {
      "epoch": 0.5017864346222555,
      "grad_norm": 0.7467042803764343,
      "learning_rate": 5.214108558485258e-06,
      "loss": 0.7256,
      "step": 17415
    },
    {
      "epoch": 0.5018152480839048,
      "grad_norm": 0.728297233581543,
      "learning_rate": 5.2136423754355634e-06,
      "loss": 0.7146,
      "step": 17416
    },
    {
      "epoch": 0.501844061545554,
      "grad_norm": 0.7322311997413635,
      "learning_rate": 5.2131761905252576e-06,
      "loss": 0.7314,
      "step": 17417
    },
    {
      "epoch": 0.5018728750072033,
      "grad_norm": 0.709846556186676,
      "learning_rate": 5.212710003758401e-06,
      "loss": 0.7067,
      "step": 17418
    },
    {
      "epoch": 0.5019016884688526,
      "grad_norm": 0.749021053314209,
      "learning_rate": 5.212243815139053e-06,
      "loss": 0.7236,
      "step": 17419
    },
    {
      "epoch": 0.5019305019305019,
      "grad_norm": 1.7106267213821411,
      "learning_rate": 5.2117776246712736e-06,
      "loss": 0.7552,
      "step": 17420
    },
    {
      "epoch": 0.5019593153921512,
      "grad_norm": 0.7309174537658691,
      "learning_rate": 5.211311432359123e-06,
      "loss": 0.7046,
      "step": 17421
    },
    {
      "epoch": 0.5019881288538005,
      "grad_norm": 0.806059718132019,
      "learning_rate": 5.210845238206661e-06,
      "loss": 0.7314,
      "step": 17422
    },
    {
      "epoch": 0.5020169423154498,
      "grad_norm": 0.7264204621315002,
      "learning_rate": 5.210379042217946e-06,
      "loss": 0.7139,
      "step": 17423
    },
    {
      "epoch": 0.5020457557770991,
      "grad_norm": 0.7771307229995728,
      "learning_rate": 5.209912844397041e-06,
      "loss": 0.7094,
      "step": 17424
    },
    {
      "epoch": 0.5020745692387484,
      "grad_norm": 0.7383360862731934,
      "learning_rate": 5.209446644748005e-06,
      "loss": 0.7251,
      "step": 17425
    },
    {
      "epoch": 0.5021033827003977,
      "grad_norm": 0.7351329922676086,
      "learning_rate": 5.208980443274899e-06,
      "loss": 0.6928,
      "step": 17426
    },
    {
      "epoch": 0.502132196162047,
      "grad_norm": 0.7421258687973022,
      "learning_rate": 5.208514239981781e-06,
      "loss": 0.6935,
      "step": 17427
    },
    {
      "epoch": 0.5021610096236961,
      "grad_norm": 0.7302945852279663,
      "learning_rate": 5.208048034872714e-06,
      "loss": 0.7357,
      "step": 17428
    },
    {
      "epoch": 0.5021898230853454,
      "grad_norm": 0.7474925518035889,
      "learning_rate": 5.207581827951754e-06,
      "loss": 0.75,
      "step": 17429
    },
    {
      "epoch": 0.5022186365469947,
      "grad_norm": 0.7278727293014526,
      "learning_rate": 5.207115619222966e-06,
      "loss": 0.7277,
      "step": 17430
    },
    {
      "epoch": 0.502247450008644,
      "grad_norm": 0.767148494720459,
      "learning_rate": 5.206649408690406e-06,
      "loss": 0.6963,
      "step": 17431
    },
    {
      "epoch": 0.5022762634702933,
      "grad_norm": 0.737274169921875,
      "learning_rate": 5.206183196358136e-06,
      "loss": 0.7052,
      "step": 17432
    },
    {
      "epoch": 0.5023050769319426,
      "grad_norm": 0.7234667539596558,
      "learning_rate": 5.205716982230219e-06,
      "loss": 0.7159,
      "step": 17433
    },
    {
      "epoch": 0.5023338903935919,
      "grad_norm": 0.7261806130409241,
      "learning_rate": 5.205250766310712e-06,
      "loss": 0.7044,
      "step": 17434
    },
    {
      "epoch": 0.5023627038552412,
      "grad_norm": 0.8617154359817505,
      "learning_rate": 5.2047845486036744e-06,
      "loss": 0.7296,
      "step": 17435
    },
    {
      "epoch": 0.5023915173168905,
      "grad_norm": 0.7524994015693665,
      "learning_rate": 5.20431832911317e-06,
      "loss": 0.7194,
      "step": 17436
    },
    {
      "epoch": 0.5024203307785398,
      "grad_norm": 0.7320160269737244,
      "learning_rate": 5.203852107843257e-06,
      "loss": 0.7099,
      "step": 17437
    },
    {
      "epoch": 0.502449144240189,
      "grad_norm": 0.7559020519256592,
      "learning_rate": 5.203385884797995e-06,
      "loss": 0.7054,
      "step": 17438
    },
    {
      "epoch": 0.5024779577018383,
      "grad_norm": 0.7030220627784729,
      "learning_rate": 5.202919659981446e-06,
      "loss": 0.7166,
      "step": 17439
    },
    {
      "epoch": 0.5025067711634876,
      "grad_norm": 0.7349551916122437,
      "learning_rate": 5.20245343339767e-06,
      "loss": 0.7156,
      "step": 17440
    },
    {
      "epoch": 0.5025355846251368,
      "grad_norm": 0.7232559323310852,
      "learning_rate": 5.201987205050725e-06,
      "loss": 0.7355,
      "step": 17441
    },
    {
      "epoch": 0.5025643980867861,
      "grad_norm": 0.7216774821281433,
      "learning_rate": 5.201520974944675e-06,
      "loss": 0.7351,
      "step": 17442
    },
    {
      "epoch": 0.5025932115484354,
      "grad_norm": 0.7380626201629639,
      "learning_rate": 5.2010547430835775e-06,
      "loss": 0.6957,
      "step": 17443
    },
    {
      "epoch": 0.5026220250100847,
      "grad_norm": 0.754226803779602,
      "learning_rate": 5.200588509471496e-06,
      "loss": 0.7147,
      "step": 17444
    },
    {
      "epoch": 0.502650838471734,
      "grad_norm": 0.7327618598937988,
      "learning_rate": 5.200122274112488e-06,
      "loss": 0.7026,
      "step": 17445
    },
    {
      "epoch": 0.5026796519333833,
      "grad_norm": 0.7392842769622803,
      "learning_rate": 5.199656037010615e-06,
      "loss": 0.7041,
      "step": 17446
    },
    {
      "epoch": 0.5027084653950326,
      "grad_norm": 0.6925918459892273,
      "learning_rate": 5.199189798169937e-06,
      "loss": 0.6958,
      "step": 17447
    },
    {
      "epoch": 0.5027372788566818,
      "grad_norm": 0.7521546483039856,
      "learning_rate": 5.198723557594516e-06,
      "loss": 0.709,
      "step": 17448
    },
    {
      "epoch": 0.5027660923183311,
      "grad_norm": 0.7416073083877563,
      "learning_rate": 5.198257315288411e-06,
      "loss": 0.7187,
      "step": 17449
    },
    {
      "epoch": 0.5027949057799804,
      "grad_norm": 0.754818320274353,
      "learning_rate": 5.197791071255684e-06,
      "loss": 0.7016,
      "step": 17450
    },
    {
      "epoch": 0.5028237192416297,
      "grad_norm": 0.7285652756690979,
      "learning_rate": 5.197324825500393e-06,
      "loss": 0.7229,
      "step": 17451
    },
    {
      "epoch": 0.502852532703279,
      "grad_norm": 0.7444002032279968,
      "learning_rate": 5.1968585780266e-06,
      "loss": 0.7418,
      "step": 17452
    },
    {
      "epoch": 0.5028813461649283,
      "grad_norm": 0.7358720302581787,
      "learning_rate": 5.196392328838367e-06,
      "loss": 0.7319,
      "step": 17453
    },
    {
      "epoch": 0.5029101596265776,
      "grad_norm": 0.7394028306007385,
      "learning_rate": 5.19592607793975e-06,
      "loss": 0.7251,
      "step": 17454
    },
    {
      "epoch": 0.5029389730882268,
      "grad_norm": 0.742428183555603,
      "learning_rate": 5.195459825334815e-06,
      "loss": 0.7288,
      "step": 17455
    },
    {
      "epoch": 0.5029677865498761,
      "grad_norm": 0.7212902307510376,
      "learning_rate": 5.194993571027621e-06,
      "loss": 0.7201,
      "step": 17456
    },
    {
      "epoch": 0.5029966000115254,
      "grad_norm": 0.742938756942749,
      "learning_rate": 5.194527315022228e-06,
      "loss": 0.7399,
      "step": 17457
    },
    {
      "epoch": 0.5030254134731746,
      "grad_norm": 0.7751082181930542,
      "learning_rate": 5.1940610573226955e-06,
      "loss": 0.7184,
      "step": 17458
    },
    {
      "epoch": 0.5030542269348239,
      "grad_norm": 0.7468063235282898,
      "learning_rate": 5.193594797933085e-06,
      "loss": 0.7334,
      "step": 17459
    },
    {
      "epoch": 0.5030830403964732,
      "grad_norm": 0.7328944802284241,
      "learning_rate": 5.193128536857458e-06,
      "loss": 0.7342,
      "step": 17460
    },
    {
      "epoch": 0.5031118538581225,
      "grad_norm": 0.7359668016433716,
      "learning_rate": 5.192662274099874e-06,
      "loss": 0.7162,
      "step": 17461
    },
    {
      "epoch": 0.5031406673197718,
      "grad_norm": 0.785891056060791,
      "learning_rate": 5.192196009664395e-06,
      "loss": 0.7278,
      "step": 17462
    },
    {
      "epoch": 0.5031694807814211,
      "grad_norm": 0.743858814239502,
      "learning_rate": 5.191729743555081e-06,
      "loss": 0.7476,
      "step": 17463
    },
    {
      "epoch": 0.5031982942430704,
      "grad_norm": 0.7432137727737427,
      "learning_rate": 5.191263475775992e-06,
      "loss": 0.7089,
      "step": 17464
    },
    {
      "epoch": 0.5032271077047197,
      "grad_norm": 0.7555794715881348,
      "learning_rate": 5.190797206331189e-06,
      "loss": 0.7228,
      "step": 17465
    },
    {
      "epoch": 0.503255921166369,
      "grad_norm": 0.7407946586608887,
      "learning_rate": 5.1903309352247324e-06,
      "loss": 0.7235,
      "step": 17466
    },
    {
      "epoch": 0.5032847346280183,
      "grad_norm": 0.7292322516441345,
      "learning_rate": 5.189864662460685e-06,
      "loss": 0.7261,
      "step": 17467
    },
    {
      "epoch": 0.5033135480896674,
      "grad_norm": 0.7095743417739868,
      "learning_rate": 5.189398388043105e-06,
      "loss": 0.6995,
      "step": 17468
    },
    {
      "epoch": 0.5033423615513167,
      "grad_norm": 1.107356071472168,
      "learning_rate": 5.188932111976056e-06,
      "loss": 0.7268,
      "step": 17469
    },
    {
      "epoch": 0.503371175012966,
      "grad_norm": 0.7289708852767944,
      "learning_rate": 5.188465834263595e-06,
      "loss": 0.7222,
      "step": 17470
    },
    {
      "epoch": 0.5033999884746153,
      "grad_norm": 0.7268871068954468,
      "learning_rate": 5.187999554909787e-06,
      "loss": 0.6987,
      "step": 17471
    },
    {
      "epoch": 0.5034288019362646,
      "grad_norm": 0.7332813739776611,
      "learning_rate": 5.187533273918689e-06,
      "loss": 0.7339,
      "step": 17472
    },
    {
      "epoch": 0.5034576153979139,
      "grad_norm": 0.7233600616455078,
      "learning_rate": 5.187066991294365e-06,
      "loss": 0.7114,
      "step": 17473
    },
    {
      "epoch": 0.5034864288595632,
      "grad_norm": 0.7440423965454102,
      "learning_rate": 5.186600707040874e-06,
      "loss": 0.7088,
      "step": 17474
    },
    {
      "epoch": 0.5035152423212125,
      "grad_norm": 0.7098549604415894,
      "learning_rate": 5.1861344211622776e-06,
      "loss": 0.7365,
      "step": 17475
    },
    {
      "epoch": 0.5035440557828618,
      "grad_norm": 0.7504130601882935,
      "learning_rate": 5.185668133662637e-06,
      "loss": 0.7025,
      "step": 17476
    },
    {
      "epoch": 0.5035728692445111,
      "grad_norm": 0.7669133543968201,
      "learning_rate": 5.185201844546012e-06,
      "loss": 0.7329,
      "step": 17477
    },
    {
      "epoch": 0.5036016827061603,
      "grad_norm": 0.7089622616767883,
      "learning_rate": 5.184735553816464e-06,
      "loss": 0.7239,
      "step": 17478
    },
    {
      "epoch": 0.5036304961678096,
      "grad_norm": 0.7428941130638123,
      "learning_rate": 5.184269261478054e-06,
      "loss": 0.7233,
      "step": 17479
    },
    {
      "epoch": 0.5036593096294589,
      "grad_norm": 0.7461167573928833,
      "learning_rate": 5.183802967534843e-06,
      "loss": 0.74,
      "step": 17480
    },
    {
      "epoch": 0.5036881230911082,
      "grad_norm": 0.696444571018219,
      "learning_rate": 5.183336671990892e-06,
      "loss": 0.7059,
      "step": 17481
    },
    {
      "epoch": 0.5037169365527574,
      "grad_norm": 0.7114288210868835,
      "learning_rate": 5.182870374850262e-06,
      "loss": 0.7049,
      "step": 17482
    },
    {
      "epoch": 0.5037457500144067,
      "grad_norm": 0.7446243166923523,
      "learning_rate": 5.182404076117012e-06,
      "loss": 0.6975,
      "step": 17483
    },
    {
      "epoch": 0.503774563476056,
      "grad_norm": 0.7286477088928223,
      "learning_rate": 5.181937775795207e-06,
      "loss": 0.7034,
      "step": 17484
    },
    {
      "epoch": 0.5038033769377053,
      "grad_norm": 0.7277182340621948,
      "learning_rate": 5.181471473888906e-06,
      "loss": 0.7078,
      "step": 17485
    },
    {
      "epoch": 0.5038321903993546,
      "grad_norm": 0.6982457637786865,
      "learning_rate": 5.1810051704021695e-06,
      "loss": 0.6877,
      "step": 17486
    },
    {
      "epoch": 0.5038610038610039,
      "grad_norm": 0.736659586429596,
      "learning_rate": 5.1805388653390584e-06,
      "loss": 0.7212,
      "step": 17487
    },
    {
      "epoch": 0.5038898173226531,
      "grad_norm": 0.7302623391151428,
      "learning_rate": 5.180072558703635e-06,
      "loss": 0.7096,
      "step": 17488
    },
    {
      "epoch": 0.5039186307843024,
      "grad_norm": 0.7554337978363037,
      "learning_rate": 5.179606250499958e-06,
      "loss": 0.7218,
      "step": 17489
    },
    {
      "epoch": 0.5039474442459517,
      "grad_norm": 0.7040899991989136,
      "learning_rate": 5.179139940732091e-06,
      "loss": 0.6947,
      "step": 17490
    },
    {
      "epoch": 0.503976257707601,
      "grad_norm": 0.7161412239074707,
      "learning_rate": 5.178673629404093e-06,
      "loss": 0.7031,
      "step": 17491
    },
    {
      "epoch": 0.5040050711692503,
      "grad_norm": 0.7102038860321045,
      "learning_rate": 5.178207316520029e-06,
      "loss": 0.7021,
      "step": 17492
    },
    {
      "epoch": 0.5040338846308996,
      "grad_norm": 0.7103283405303955,
      "learning_rate": 5.177741002083956e-06,
      "loss": 0.7103,
      "step": 17493
    },
    {
      "epoch": 0.5040626980925489,
      "grad_norm": 0.728473961353302,
      "learning_rate": 5.177274686099936e-06,
      "loss": 0.7022,
      "step": 17494
    },
    {
      "epoch": 0.5040915115541981,
      "grad_norm": 0.720223069190979,
      "learning_rate": 5.17680836857203e-06,
      "loss": 0.7043,
      "step": 17495
    },
    {
      "epoch": 0.5041203250158474,
      "grad_norm": 0.7136651277542114,
      "learning_rate": 5.176342049504302e-06,
      "loss": 0.701,
      "step": 17496
    },
    {
      "epoch": 0.5041491384774966,
      "grad_norm": 0.7177190184593201,
      "learning_rate": 5.17587572890081e-06,
      "loss": 0.7355,
      "step": 17497
    },
    {
      "epoch": 0.5041779519391459,
      "grad_norm": 0.7211394309997559,
      "learning_rate": 5.175409406765617e-06,
      "loss": 0.7235,
      "step": 17498
    },
    {
      "epoch": 0.5042067654007952,
      "grad_norm": 0.7053492069244385,
      "learning_rate": 5.174943083102782e-06,
      "loss": 0.7049,
      "step": 17499
    },
    {
      "epoch": 0.5042355788624445,
      "grad_norm": 0.7451213598251343,
      "learning_rate": 5.174476757916369e-06,
      "loss": 0.7174,
      "step": 17500
    },
    {
      "epoch": 0.5042643923240938,
      "grad_norm": 0.7064011693000793,
      "learning_rate": 5.174010431210437e-06,
      "loss": 0.721,
      "step": 17501
    },
    {
      "epoch": 0.5042932057857431,
      "grad_norm": 0.7436915040016174,
      "learning_rate": 5.173544102989048e-06,
      "loss": 0.7422,
      "step": 17502
    },
    {
      "epoch": 0.5043220192473924,
      "grad_norm": 0.7192456126213074,
      "learning_rate": 5.173077773256264e-06,
      "loss": 0.7356,
      "step": 17503
    },
    {
      "epoch": 0.5043508327090417,
      "grad_norm": 0.6981735825538635,
      "learning_rate": 5.172611442016145e-06,
      "loss": 0.6943,
      "step": 17504
    },
    {
      "epoch": 0.504379646170691,
      "grad_norm": 0.7025161981582642,
      "learning_rate": 5.172145109272754e-06,
      "loss": 0.7099,
      "step": 17505
    },
    {
      "epoch": 0.5044084596323403,
      "grad_norm": 0.7078136801719666,
      "learning_rate": 5.17167877503015e-06,
      "loss": 0.7235,
      "step": 17506
    },
    {
      "epoch": 0.5044372730939896,
      "grad_norm": 0.7301163077354431,
      "learning_rate": 5.1712124392923946e-06,
      "loss": 0.7239,
      "step": 17507
    },
    {
      "epoch": 0.5044660865556388,
      "grad_norm": 0.7099246978759766,
      "learning_rate": 5.170746102063552e-06,
      "loss": 0.7034,
      "step": 17508
    },
    {
      "epoch": 0.504494900017288,
      "grad_norm": 0.753968358039856,
      "learning_rate": 5.170279763347681e-06,
      "loss": 0.7254,
      "step": 17509
    },
    {
      "epoch": 0.5045237134789373,
      "grad_norm": 0.7250247597694397,
      "learning_rate": 5.169813423148842e-06,
      "loss": 0.7186,
      "step": 17510
    },
    {
      "epoch": 0.5045525269405866,
      "grad_norm": 0.7353544235229492,
      "learning_rate": 5.1693470814711e-06,
      "loss": 0.7511,
      "step": 17511
    },
    {
      "epoch": 0.5045813404022359,
      "grad_norm": 0.7124272584915161,
      "learning_rate": 5.168880738318512e-06,
      "loss": 0.6949,
      "step": 17512
    },
    {
      "epoch": 0.5046101538638852,
      "grad_norm": 0.7254289984703064,
      "learning_rate": 5.1684143936951425e-06,
      "loss": 0.6985,
      "step": 17513
    },
    {
      "epoch": 0.5046389673255345,
      "grad_norm": 0.7200736999511719,
      "learning_rate": 5.167948047605052e-06,
      "loss": 0.6883,
      "step": 17514
    },
    {
      "epoch": 0.5046677807871838,
      "grad_norm": 0.7451792359352112,
      "learning_rate": 5.167481700052304e-06,
      "loss": 0.7184,
      "step": 17515
    },
    {
      "epoch": 0.5046965942488331,
      "grad_norm": 0.7352566719055176,
      "learning_rate": 5.167015351040956e-06,
      "loss": 0.731,
      "step": 17516
    },
    {
      "epoch": 0.5047254077104824,
      "grad_norm": 0.7484868168830872,
      "learning_rate": 5.1665490005750704e-06,
      "loss": 0.6803,
      "step": 17517
    },
    {
      "epoch": 0.5047542211721316,
      "grad_norm": 0.7187138199806213,
      "learning_rate": 5.166082648658711e-06,
      "loss": 0.7208,
      "step": 17518
    },
    {
      "epoch": 0.5047830346337809,
      "grad_norm": 0.7133280634880066,
      "learning_rate": 5.165616295295937e-06,
      "loss": 0.7053,
      "step": 17519
    },
    {
      "epoch": 0.5048118480954302,
      "grad_norm": 0.7296455502510071,
      "learning_rate": 5.16514994049081e-06,
      "loss": 0.7183,
      "step": 17520
    },
    {
      "epoch": 0.5048406615570795,
      "grad_norm": 0.7324258089065552,
      "learning_rate": 5.164683584247394e-06,
      "loss": 0.7406,
      "step": 17521
    },
    {
      "epoch": 0.5048694750187287,
      "grad_norm": 0.7308613657951355,
      "learning_rate": 5.1642172265697475e-06,
      "loss": 0.7081,
      "step": 17522
    },
    {
      "epoch": 0.504898288480378,
      "grad_norm": 0.7122958302497864,
      "learning_rate": 5.163750867461933e-06,
      "loss": 0.7147,
      "step": 17523
    },
    {
      "epoch": 0.5049271019420273,
      "grad_norm": 0.7171646952629089,
      "learning_rate": 5.163284506928012e-06,
      "loss": 0.7226,
      "step": 17524
    },
    {
      "epoch": 0.5049559154036766,
      "grad_norm": 0.7214579582214355,
      "learning_rate": 5.162818144972046e-06,
      "loss": 0.6988,
      "step": 17525
    },
    {
      "epoch": 0.5049847288653259,
      "grad_norm": 0.7191999554634094,
      "learning_rate": 5.162351781598097e-06,
      "loss": 0.7296,
      "step": 17526
    },
    {
      "epoch": 0.5050135423269752,
      "grad_norm": 0.7249466776847839,
      "learning_rate": 5.161885416810227e-06,
      "loss": 0.7144,
      "step": 17527
    },
    {
      "epoch": 0.5050423557886244,
      "grad_norm": 0.749502956867218,
      "learning_rate": 5.161419050612496e-06,
      "loss": 0.7411,
      "step": 17528
    },
    {
      "epoch": 0.5050711692502737,
      "grad_norm": 0.7333077192306519,
      "learning_rate": 5.1609526830089665e-06,
      "loss": 0.6984,
      "step": 17529
    },
    {
      "epoch": 0.505099982711923,
      "grad_norm": 0.7259921431541443,
      "learning_rate": 5.1604863140037e-06,
      "loss": 0.7127,
      "step": 17530
    },
    {
      "epoch": 0.5051287961735723,
      "grad_norm": 0.7110356688499451,
      "learning_rate": 5.160019943600758e-06,
      "loss": 0.71,
      "step": 17531
    },
    {
      "epoch": 0.5051576096352216,
      "grad_norm": 0.7152417898178101,
      "learning_rate": 5.159553571804203e-06,
      "loss": 0.7014,
      "step": 17532
    },
    {
      "epoch": 0.5051864230968709,
      "grad_norm": 0.7369548082351685,
      "learning_rate": 5.159087198618095e-06,
      "loss": 0.7072,
      "step": 17533
    },
    {
      "epoch": 0.5052152365585202,
      "grad_norm": 0.7280025482177734,
      "learning_rate": 5.158620824046497e-06,
      "loss": 0.7092,
      "step": 17534
    },
    {
      "epoch": 0.5052440500201695,
      "grad_norm": 0.7258833050727844,
      "learning_rate": 5.158154448093471e-06,
      "loss": 0.7152,
      "step": 17535
    },
    {
      "epoch": 0.5052728634818187,
      "grad_norm": 0.720580518245697,
      "learning_rate": 5.1576880707630775e-06,
      "loss": 0.7177,
      "step": 17536
    },
    {
      "epoch": 0.505301676943468,
      "grad_norm": 0.7252879738807678,
      "learning_rate": 5.157221692059377e-06,
      "loss": 0.7209,
      "step": 17537
    },
    {
      "epoch": 0.5053304904051172,
      "grad_norm": 0.7355066537857056,
      "learning_rate": 5.156755311986433e-06,
      "loss": 0.7051,
      "step": 17538
    },
    {
      "epoch": 0.5053593038667665,
      "grad_norm": 0.7300795316696167,
      "learning_rate": 5.1562889305483085e-06,
      "loss": 0.7087,
      "step": 17539
    },
    {
      "epoch": 0.5053881173284158,
      "grad_norm": 0.725630521774292,
      "learning_rate": 5.155822547749063e-06,
      "loss": 0.7296,
      "step": 17540
    },
    {
      "epoch": 0.5054169307900651,
      "grad_norm": 0.7365232110023499,
      "learning_rate": 5.155356163592759e-06,
      "loss": 0.7224,
      "step": 17541
    },
    {
      "epoch": 0.5054457442517144,
      "grad_norm": 0.7815935611724854,
      "learning_rate": 5.1548897780834585e-06,
      "loss": 0.6793,
      "step": 17542
    },
    {
      "epoch": 0.5054745577133637,
      "grad_norm": 0.7524962425231934,
      "learning_rate": 5.154423391225221e-06,
      "loss": 0.7316,
      "step": 17543
    },
    {
      "epoch": 0.505503371175013,
      "grad_norm": 0.7461516857147217,
      "learning_rate": 5.153957003022111e-06,
      "loss": 0.7324,
      "step": 17544
    },
    {
      "epoch": 0.5055321846366623,
      "grad_norm": 0.7177671790122986,
      "learning_rate": 5.15349061347819e-06,
      "loss": 0.6886,
      "step": 17545
    },
    {
      "epoch": 0.5055609980983116,
      "grad_norm": 0.7440851330757141,
      "learning_rate": 5.1530242225975195e-06,
      "loss": 0.6942,
      "step": 17546
    },
    {
      "epoch": 0.5055898115599609,
      "grad_norm": 0.766054093837738,
      "learning_rate": 5.152557830384161e-06,
      "loss": 0.7311,
      "step": 17547
    },
    {
      "epoch": 0.5056186250216101,
      "grad_norm": 0.7487585544586182,
      "learning_rate": 5.152091436842175e-06,
      "loss": 0.722,
      "step": 17548
    },
    {
      "epoch": 0.5056474384832593,
      "grad_norm": 0.7155750393867493,
      "learning_rate": 5.151625041975626e-06,
      "loss": 0.7312,
      "step": 17549
    },
    {
      "epoch": 0.5056762519449086,
      "grad_norm": 0.8977711796760559,
      "learning_rate": 5.151158645788573e-06,
      "loss": 0.7029,
      "step": 17550
    },
    {
      "epoch": 0.5057050654065579,
      "grad_norm": 0.7460114359855652,
      "learning_rate": 5.1506922482850794e-06,
      "loss": 0.7092,
      "step": 17551
    },
    {
      "epoch": 0.5057338788682072,
      "grad_norm": 0.7775140404701233,
      "learning_rate": 5.150225849469208e-06,
      "loss": 0.7187,
      "step": 17552
    },
    {
      "epoch": 0.5057626923298565,
      "grad_norm": 0.7144336700439453,
      "learning_rate": 5.1497594493450185e-06,
      "loss": 0.7041,
      "step": 17553
    },
    {
      "epoch": 0.5057915057915058,
      "grad_norm": 0.7223024964332581,
      "learning_rate": 5.149293047916576e-06,
      "loss": 0.7095,
      "step": 17554
    },
    {
      "epoch": 0.5058203192531551,
      "grad_norm": 0.7483353614807129,
      "learning_rate": 5.148826645187938e-06,
      "loss": 0.7207,
      "step": 17555
    },
    {
      "epoch": 0.5058491327148044,
      "grad_norm": 0.7125042080879211,
      "learning_rate": 5.14836024116317e-06,
      "loss": 0.6853,
      "step": 17556
    },
    {
      "epoch": 0.5058779461764537,
      "grad_norm": 0.7326018810272217,
      "learning_rate": 5.147893835846332e-06,
      "loss": 0.7334,
      "step": 17557
    },
    {
      "epoch": 0.5059067596381029,
      "grad_norm": 0.7360860705375671,
      "learning_rate": 5.147427429241487e-06,
      "loss": 0.7166,
      "step": 17558
    },
    {
      "epoch": 0.5059355730997522,
      "grad_norm": 0.7283278703689575,
      "learning_rate": 5.146961021352697e-06,
      "loss": 0.734,
      "step": 17559
    },
    {
      "epoch": 0.5059643865614015,
      "grad_norm": 0.7265938520431519,
      "learning_rate": 5.146494612184022e-06,
      "loss": 0.7302,
      "step": 17560
    },
    {
      "epoch": 0.5059932000230508,
      "grad_norm": 0.7362585067749023,
      "learning_rate": 5.1460282017395255e-06,
      "loss": 0.7069,
      "step": 17561
    },
    {
      "epoch": 0.5060220134847001,
      "grad_norm": 0.739348292350769,
      "learning_rate": 5.1455617900232705e-06,
      "loss": 0.7172,
      "step": 17562
    },
    {
      "epoch": 0.5060508269463493,
      "grad_norm": 0.7220380306243896,
      "learning_rate": 5.1450953770393165e-06,
      "loss": 0.7163,
      "step": 17563
    },
    {
      "epoch": 0.5060796404079986,
      "grad_norm": 0.7339565753936768,
      "learning_rate": 5.144628962791728e-06,
      "loss": 0.6982,
      "step": 17564
    },
    {
      "epoch": 0.5061084538696479,
      "grad_norm": 0.7045885920524597,
      "learning_rate": 5.144162547284565e-06,
      "loss": 0.7024,
      "step": 17565
    },
    {
      "epoch": 0.5061372673312972,
      "grad_norm": 0.7429528832435608,
      "learning_rate": 5.1436961305218914e-06,
      "loss": 0.7354,
      "step": 17566
    },
    {
      "epoch": 0.5061660807929464,
      "grad_norm": 0.7305174469947815,
      "learning_rate": 5.143229712507767e-06,
      "loss": 0.7346,
      "step": 17567
    },
    {
      "epoch": 0.5061948942545957,
      "grad_norm": 0.7334555387496948,
      "learning_rate": 5.142763293246256e-06,
      "loss": 0.7043,
      "step": 17568
    },
    {
      "epoch": 0.506223707716245,
      "grad_norm": 0.6956803798675537,
      "learning_rate": 5.142296872741418e-06,
      "loss": 0.7208,
      "step": 17569
    },
    {
      "epoch": 0.5062525211778943,
      "grad_norm": 0.7235492467880249,
      "learning_rate": 5.141830450997316e-06,
      "loss": 0.7302,
      "step": 17570
    },
    {
      "epoch": 0.5062813346395436,
      "grad_norm": 0.7193825840950012,
      "learning_rate": 5.141364028018015e-06,
      "loss": 0.6966,
      "step": 17571
    },
    {
      "epoch": 0.5063101481011929,
      "grad_norm": 0.7461066246032715,
      "learning_rate": 5.140897603807573e-06,
      "loss": 0.6886,
      "step": 17572
    },
    {
      "epoch": 0.5063389615628422,
      "grad_norm": 0.7485753893852234,
      "learning_rate": 5.140431178370054e-06,
      "loss": 0.7352,
      "step": 17573
    },
    {
      "epoch": 0.5063677750244915,
      "grad_norm": 0.7156184911727905,
      "learning_rate": 5.139964751709519e-06,
      "loss": 0.7384,
      "step": 17574
    },
    {
      "epoch": 0.5063965884861408,
      "grad_norm": 0.7246124148368835,
      "learning_rate": 5.1394983238300335e-06,
      "loss": 0.7122,
      "step": 17575
    },
    {
      "epoch": 0.50642540194779,
      "grad_norm": 0.705228328704834,
      "learning_rate": 5.139031894735656e-06,
      "loss": 0.7126,
      "step": 17576
    },
    {
      "epoch": 0.5064542154094392,
      "grad_norm": 0.7101432085037231,
      "learning_rate": 5.138565464430448e-06,
      "loss": 0.6938,
      "step": 17577
    },
    {
      "epoch": 0.5064830288710885,
      "grad_norm": 0.7201998829841614,
      "learning_rate": 5.138099032918476e-06,
      "loss": 0.6938,
      "step": 17578
    },
    {
      "epoch": 0.5065118423327378,
      "grad_norm": 0.7339751720428467,
      "learning_rate": 5.1376326002037955e-06,
      "loss": 0.72,
      "step": 17579
    },
    {
      "epoch": 0.5065406557943871,
      "grad_norm": 0.722977876663208,
      "learning_rate": 5.1371661662904756e-06,
      "loss": 0.7231,
      "step": 17580
    },
    {
      "epoch": 0.5065694692560364,
      "grad_norm": 0.7330976724624634,
      "learning_rate": 5.136699731182575e-06,
      "loss": 0.7074,
      "step": 17581
    },
    {
      "epoch": 0.5065982827176857,
      "grad_norm": 0.7397657036781311,
      "learning_rate": 5.1362332948841575e-06,
      "loss": 0.7194,
      "step": 17582
    },
    {
      "epoch": 0.506627096179335,
      "grad_norm": 0.7479872107505798,
      "learning_rate": 5.135766857399282e-06,
      "loss": 0.7168,
      "step": 17583
    },
    {
      "epoch": 0.5066559096409843,
      "grad_norm": 0.7342814207077026,
      "learning_rate": 5.135300418732015e-06,
      "loss": 0.7024,
      "step": 17584
    },
    {
      "epoch": 0.5066847231026336,
      "grad_norm": 0.7476248741149902,
      "learning_rate": 5.134833978886415e-06,
      "loss": 0.7416,
      "step": 17585
    },
    {
      "epoch": 0.5067135365642829,
      "grad_norm": 0.716336727142334,
      "learning_rate": 5.134367537866546e-06,
      "loss": 0.7194,
      "step": 17586
    },
    {
      "epoch": 0.5067423500259322,
      "grad_norm": 0.7456820011138916,
      "learning_rate": 5.133901095676471e-06,
      "loss": 0.7148,
      "step": 17587
    },
    {
      "epoch": 0.5067711634875814,
      "grad_norm": 0.7244285345077515,
      "learning_rate": 5.13343465232025e-06,
      "loss": 0.715,
      "step": 17588
    },
    {
      "epoch": 0.5067999769492307,
      "grad_norm": 0.736804723739624,
      "learning_rate": 5.132968207801947e-06,
      "loss": 0.7068,
      "step": 17589
    },
    {
      "epoch": 0.5068287904108799,
      "grad_norm": 0.7399944067001343,
      "learning_rate": 5.132501762125625e-06,
      "loss": 0.7085,
      "step": 17590
    },
    {
      "epoch": 0.5068576038725292,
      "grad_norm": 0.7419739961624146,
      "learning_rate": 5.132035315295343e-06,
      "loss": 0.7278,
      "step": 17591
    },
    {
      "epoch": 0.5068864173341785,
      "grad_norm": 0.7439239025115967,
      "learning_rate": 5.1315688673151665e-06,
      "loss": 0.696,
      "step": 17592
    },
    {
      "epoch": 0.5069152307958278,
      "grad_norm": 0.7200647592544556,
      "learning_rate": 5.131102418189158e-06,
      "loss": 0.723,
      "step": 17593
    },
    {
      "epoch": 0.5069440442574771,
      "grad_norm": 0.7370898723602295,
      "learning_rate": 5.130635967921377e-06,
      "loss": 0.7251,
      "step": 17594
    },
    {
      "epoch": 0.5069728577191264,
      "grad_norm": 0.7232182025909424,
      "learning_rate": 5.1301695165158875e-06,
      "loss": 0.7174,
      "step": 17595
    },
    {
      "epoch": 0.5070016711807757,
      "grad_norm": 0.7282821536064148,
      "learning_rate": 5.129703063976752e-06,
      "loss": 0.7071,
      "step": 17596
    },
    {
      "epoch": 0.507030484642425,
      "grad_norm": 0.7158809304237366,
      "learning_rate": 5.1292366103080316e-06,
      "loss": 0.7122,
      "step": 17597
    },
    {
      "epoch": 0.5070592981040742,
      "grad_norm": 0.72325599193573,
      "learning_rate": 5.12877015551379e-06,
      "loss": 0.7376,
      "step": 17598
    },
    {
      "epoch": 0.5070881115657235,
      "grad_norm": 0.7199602723121643,
      "learning_rate": 5.12830369959809e-06,
      "loss": 0.709,
      "step": 17599
    },
    {
      "epoch": 0.5071169250273728,
      "grad_norm": 0.697425365447998,
      "learning_rate": 5.127837242564993e-06,
      "loss": 0.721,
      "step": 17600
    },
    {
      "epoch": 0.5071457384890221,
      "grad_norm": 0.7390108704566956,
      "learning_rate": 5.12737078441856e-06,
      "loss": 0.7037,
      "step": 17601
    },
    {
      "epoch": 0.5071745519506714,
      "grad_norm": 0.7304027080535889,
      "learning_rate": 5.126904325162856e-06,
      "loss": 0.7045,
      "step": 17602
    },
    {
      "epoch": 0.5072033654123206,
      "grad_norm": 0.7399389743804932,
      "learning_rate": 5.126437864801939e-06,
      "loss": 0.7343,
      "step": 17603
    },
    {
      "epoch": 0.5072321788739699,
      "grad_norm": 0.726009726524353,
      "learning_rate": 5.125971403339879e-06,
      "loss": 0.701,
      "step": 17604
    },
    {
      "epoch": 0.5072609923356192,
      "grad_norm": 0.7442729473114014,
      "learning_rate": 5.125504940780732e-06,
      "loss": 0.7325,
      "step": 17605
    },
    {
      "epoch": 0.5072898057972685,
      "grad_norm": 0.7476503849029541,
      "learning_rate": 5.125038477128563e-06,
      "loss": 0.7382,
      "step": 17606
    },
    {
      "epoch": 0.5073186192589177,
      "grad_norm": 0.7181280255317688,
      "learning_rate": 5.1245720123874334e-06,
      "loss": 0.7172,
      "step": 17607
    },
    {
      "epoch": 0.507347432720567,
      "grad_norm": 0.7133760452270508,
      "learning_rate": 5.124105546561405e-06,
      "loss": 0.7238,
      "step": 17608
    },
    {
      "epoch": 0.5073762461822163,
      "grad_norm": 0.7208607792854309,
      "learning_rate": 5.123639079654543e-06,
      "loss": 0.7056,
      "step": 17609
    },
    {
      "epoch": 0.5074050596438656,
      "grad_norm": 0.7244799733161926,
      "learning_rate": 5.123172611670907e-06,
      "loss": 0.7266,
      "step": 17610
    },
    {
      "epoch": 0.5074338731055149,
      "grad_norm": 0.7127817869186401,
      "learning_rate": 5.122706142614562e-06,
      "loss": 0.7061,
      "step": 17611
    },
    {
      "epoch": 0.5074626865671642,
      "grad_norm": 0.7164139747619629,
      "learning_rate": 5.122239672489569e-06,
      "loss": 0.7213,
      "step": 17612
    },
    {
      "epoch": 0.5074915000288135,
      "grad_norm": 0.7261462807655334,
      "learning_rate": 5.12177320129999e-06,
      "loss": 0.749,
      "step": 17613
    },
    {
      "epoch": 0.5075203134904628,
      "grad_norm": 0.7455873489379883,
      "learning_rate": 5.1213067290498874e-06,
      "loss": 0.7658,
      "step": 17614
    },
    {
      "epoch": 0.5075491269521121,
      "grad_norm": 0.728949785232544,
      "learning_rate": 5.120840255743326e-06,
      "loss": 0.7159,
      "step": 17615
    },
    {
      "epoch": 0.5075779404137614,
      "grad_norm": 0.7196775078773499,
      "learning_rate": 5.1203737813843656e-06,
      "loss": 0.7051,
      "step": 17616
    },
    {
      "epoch": 0.5076067538754105,
      "grad_norm": 0.7121134400367737,
      "learning_rate": 5.119907305977071e-06,
      "loss": 0.7242,
      "step": 17617
    },
    {
      "epoch": 0.5076355673370598,
      "grad_norm": 0.7488341331481934,
      "learning_rate": 5.119440829525504e-06,
      "loss": 0.7199,
      "step": 17618
    },
    {
      "epoch": 0.5076643807987091,
      "grad_norm": 0.7148298025131226,
      "learning_rate": 5.118974352033725e-06,
      "loss": 0.7234,
      "step": 17619
    },
    {
      "epoch": 0.5076931942603584,
      "grad_norm": 0.7470816969871521,
      "learning_rate": 5.118507873505799e-06,
      "loss": 0.6974,
      "step": 17620
    },
    {
      "epoch": 0.5077220077220077,
      "grad_norm": 0.7274456024169922,
      "learning_rate": 5.118041393945788e-06,
      "loss": 0.7209,
      "step": 17621
    },
    {
      "epoch": 0.507750821183657,
      "grad_norm": 0.745172917842865,
      "learning_rate": 5.117574913357754e-06,
      "loss": 0.6954,
      "step": 17622
    },
    {
      "epoch": 0.5077796346453063,
      "grad_norm": 0.7353476881980896,
      "learning_rate": 5.117108431745762e-06,
      "loss": 0.7218,
      "step": 17623
    },
    {
      "epoch": 0.5078084481069556,
      "grad_norm": 0.7550763487815857,
      "learning_rate": 5.116641949113872e-06,
      "loss": 0.726,
      "step": 17624
    },
    {
      "epoch": 0.5078372615686049,
      "grad_norm": 0.7191707491874695,
      "learning_rate": 5.116175465466146e-06,
      "loss": 0.7231,
      "step": 17625
    },
    {
      "epoch": 0.5078660750302542,
      "grad_norm": 0.7430193424224854,
      "learning_rate": 5.115708980806647e-06,
      "loss": 0.7162,
      "step": 17626
    },
    {
      "epoch": 0.5078948884919035,
      "grad_norm": 0.7703564763069153,
      "learning_rate": 5.1152424951394405e-06,
      "loss": 0.724,
      "step": 17627
    },
    {
      "epoch": 0.5079237019535527,
      "grad_norm": 0.710375189781189,
      "learning_rate": 5.114776008468586e-06,
      "loss": 0.7175,
      "step": 17628
    },
    {
      "epoch": 0.507952515415202,
      "grad_norm": 0.746425211429596,
      "learning_rate": 5.114309520798147e-06,
      "loss": 0.7585,
      "step": 17629
    },
    {
      "epoch": 0.5079813288768512,
      "grad_norm": 0.7770578265190125,
      "learning_rate": 5.1138430321321874e-06,
      "loss": 0.7115,
      "step": 17630
    },
    {
      "epoch": 0.5080101423385005,
      "grad_norm": 0.7291212677955627,
      "learning_rate": 5.113376542474769e-06,
      "loss": 0.7214,
      "step": 17631
    },
    {
      "epoch": 0.5080389558001498,
      "grad_norm": 0.7436054944992065,
      "learning_rate": 5.112910051829952e-06,
      "loss": 0.7097,
      "step": 17632
    },
    {
      "epoch": 0.5080677692617991,
      "grad_norm": 0.7172015309333801,
      "learning_rate": 5.112443560201803e-06,
      "loss": 0.719,
      "step": 17633
    },
    {
      "epoch": 0.5080965827234484,
      "grad_norm": 0.7228706479072571,
      "learning_rate": 5.111977067594383e-06,
      "loss": 0.7252,
      "step": 17634
    },
    {
      "epoch": 0.5081253961850977,
      "grad_norm": 0.723945140838623,
      "learning_rate": 5.111510574011754e-06,
      "loss": 0.7097,
      "step": 17635
    },
    {
      "epoch": 0.508154209646747,
      "grad_norm": 0.7583684921264648,
      "learning_rate": 5.111044079457981e-06,
      "loss": 0.7264,
      "step": 17636
    },
    {
      "epoch": 0.5081830231083962,
      "grad_norm": 0.7437198162078857,
      "learning_rate": 5.110577583937123e-06,
      "loss": 0.7202,
      "step": 17637
    },
    {
      "epoch": 0.5082118365700455,
      "grad_norm": 0.7240263223648071,
      "learning_rate": 5.110111087453245e-06,
      "loss": 0.7299,
      "step": 17638
    },
    {
      "epoch": 0.5082406500316948,
      "grad_norm": 0.7140950560569763,
      "learning_rate": 5.10964459001041e-06,
      "loss": 0.6867,
      "step": 17639
    },
    {
      "epoch": 0.5082694634933441,
      "grad_norm": 0.7468494176864624,
      "learning_rate": 5.109178091612681e-06,
      "loss": 0.7154,
      "step": 17640
    },
    {
      "epoch": 0.5082982769549934,
      "grad_norm": 0.7556155323982239,
      "learning_rate": 5.10871159226412e-06,
      "loss": 0.7245,
      "step": 17641
    },
    {
      "epoch": 0.5083270904166427,
      "grad_norm": 0.7418633699417114,
      "learning_rate": 5.108245091968789e-06,
      "loss": 0.7291,
      "step": 17642
    },
    {
      "epoch": 0.508355903878292,
      "grad_norm": 0.7111323475837708,
      "learning_rate": 5.107778590730752e-06,
      "loss": 0.7035,
      "step": 17643
    },
    {
      "epoch": 0.5083847173399412,
      "grad_norm": 0.7320848703384399,
      "learning_rate": 5.10731208855407e-06,
      "loss": 0.6833,
      "step": 17644
    },
    {
      "epoch": 0.5084135308015905,
      "grad_norm": 0.7611026763916016,
      "learning_rate": 5.106845585442808e-06,
      "loss": 0.6982,
      "step": 17645
    },
    {
      "epoch": 0.5084423442632398,
      "grad_norm": 0.7281216382980347,
      "learning_rate": 5.1063790814010275e-06,
      "loss": 0.7013,
      "step": 17646
    },
    {
      "epoch": 0.508471157724889,
      "grad_norm": 0.7049053907394409,
      "learning_rate": 5.105912576432791e-06,
      "loss": 0.6744,
      "step": 17647
    },
    {
      "epoch": 0.5084999711865383,
      "grad_norm": 0.7411231398582458,
      "learning_rate": 5.105446070542162e-06,
      "loss": 0.7093,
      "step": 17648
    },
    {
      "epoch": 0.5085287846481876,
      "grad_norm": 0.7612158060073853,
      "learning_rate": 5.1049795637332035e-06,
      "loss": 0.6989,
      "step": 17649
    },
    {
      "epoch": 0.5085575981098369,
      "grad_norm": 0.7791160941123962,
      "learning_rate": 5.104513056009978e-06,
      "loss": 0.7315,
      "step": 17650
    },
    {
      "epoch": 0.5085864115714862,
      "grad_norm": 0.719368577003479,
      "learning_rate": 5.104046547376548e-06,
      "loss": 0.7027,
      "step": 17651
    },
    {
      "epoch": 0.5086152250331355,
      "grad_norm": 0.723297655582428,
      "learning_rate": 5.103580037836977e-06,
      "loss": 0.703,
      "step": 17652
    },
    {
      "epoch": 0.5086440384947848,
      "grad_norm": 0.7404794692993164,
      "learning_rate": 5.103113527395327e-06,
      "loss": 0.7327,
      "step": 17653
    },
    {
      "epoch": 0.5086728519564341,
      "grad_norm": 0.72983717918396,
      "learning_rate": 5.1026470160556614e-06,
      "loss": 0.7141,
      "step": 17654
    },
    {
      "epoch": 0.5087016654180834,
      "grad_norm": 0.7425085306167603,
      "learning_rate": 5.102180503822043e-06,
      "loss": 0.7154,
      "step": 17655
    },
    {
      "epoch": 0.5087304788797327,
      "grad_norm": 0.7388967871665955,
      "learning_rate": 5.101713990698534e-06,
      "loss": 0.7444,
      "step": 17656
    },
    {
      "epoch": 0.5087592923413818,
      "grad_norm": 0.7598975300788879,
      "learning_rate": 5.101247476689199e-06,
      "loss": 0.7501,
      "step": 17657
    },
    {
      "epoch": 0.5087881058030311,
      "grad_norm": 0.7425800561904907,
      "learning_rate": 5.100780961798098e-06,
      "loss": 0.7261,
      "step": 17658
    },
    {
      "epoch": 0.5088169192646804,
      "grad_norm": 0.7084083557128906,
      "learning_rate": 5.100314446029297e-06,
      "loss": 0.7229,
      "step": 17659
    },
    {
      "epoch": 0.5088457327263297,
      "grad_norm": 0.6944901943206787,
      "learning_rate": 5.099847929386857e-06,
      "loss": 0.715,
      "step": 17660
    },
    {
      "epoch": 0.508874546187979,
      "grad_norm": 0.7348924875259399,
      "learning_rate": 5.099381411874842e-06,
      "loss": 0.7108,
      "step": 17661
    },
    {
      "epoch": 0.5089033596496283,
      "grad_norm": 0.7221212983131409,
      "learning_rate": 5.0989148934973135e-06,
      "loss": 0.7115,
      "step": 17662
    },
    {
      "epoch": 0.5089321731112776,
      "grad_norm": 0.7393284440040588,
      "learning_rate": 5.0984483742583346e-06,
      "loss": 0.713,
      "step": 17663
    },
    {
      "epoch": 0.5089609865729269,
      "grad_norm": 0.720845639705658,
      "learning_rate": 5.097981854161969e-06,
      "loss": 0.7245,
      "step": 17664
    },
    {
      "epoch": 0.5089898000345762,
      "grad_norm": 0.748260498046875,
      "learning_rate": 5.09751533321228e-06,
      "loss": 0.7074,
      "step": 17665
    },
    {
      "epoch": 0.5090186134962255,
      "grad_norm": 0.7264246940612793,
      "learning_rate": 5.097048811413331e-06,
      "loss": 0.7463,
      "step": 17666
    },
    {
      "epoch": 0.5090474269578747,
      "grad_norm": 0.7641575336456299,
      "learning_rate": 5.096582288769182e-06,
      "loss": 0.7088,
      "step": 17667
    },
    {
      "epoch": 0.509076240419524,
      "grad_norm": 0.7264495491981506,
      "learning_rate": 5.096115765283899e-06,
      "loss": 0.723,
      "step": 17668
    },
    {
      "epoch": 0.5091050538811733,
      "grad_norm": 0.748812198638916,
      "learning_rate": 5.095649240961542e-06,
      "loss": 0.7169,
      "step": 17669
    },
    {
      "epoch": 0.5091338673428226,
      "grad_norm": 0.763078510761261,
      "learning_rate": 5.095182715806177e-06,
      "loss": 0.7254,
      "step": 17670
    },
    {
      "epoch": 0.5091626808044718,
      "grad_norm": 0.7568895816802979,
      "learning_rate": 5.094716189821866e-06,
      "loss": 0.7396,
      "step": 17671
    },
    {
      "epoch": 0.5091914942661211,
      "grad_norm": 0.7312013506889343,
      "learning_rate": 5.094249663012671e-06,
      "loss": 0.7025,
      "step": 17672
    },
    {
      "epoch": 0.5092203077277704,
      "grad_norm": 0.7395079731941223,
      "learning_rate": 5.093783135382656e-06,
      "loss": 0.7114,
      "step": 17673
    },
    {
      "epoch": 0.5092491211894197,
      "grad_norm": 0.7246868014335632,
      "learning_rate": 5.093316606935883e-06,
      "loss": 0.7241,
      "step": 17674
    },
    {
      "epoch": 0.509277934651069,
      "grad_norm": 0.7415850758552551,
      "learning_rate": 5.092850077676417e-06,
      "loss": 0.7176,
      "step": 17675
    },
    {
      "epoch": 0.5093067481127183,
      "grad_norm": 0.7354525327682495,
      "learning_rate": 5.092383547608318e-06,
      "loss": 0.7222,
      "step": 17676
    },
    {
      "epoch": 0.5093355615743675,
      "grad_norm": 0.7134260535240173,
      "learning_rate": 5.0919170167356526e-06,
      "loss": 0.7,
      "step": 17677
    },
    {
      "epoch": 0.5093643750360168,
      "grad_norm": 0.7261785864830017,
      "learning_rate": 5.09145048506248e-06,
      "loss": 0.7007,
      "step": 17678
    },
    {
      "epoch": 0.5093931884976661,
      "grad_norm": 0.7110732197761536,
      "learning_rate": 5.090983952592866e-06,
      "loss": 0.7156,
      "step": 17679
    },
    {
      "epoch": 0.5094220019593154,
      "grad_norm": 0.7246667742729187,
      "learning_rate": 5.090517419330873e-06,
      "loss": 0.7262,
      "step": 17680
    },
    {
      "epoch": 0.5094508154209647,
      "grad_norm": 0.7117637395858765,
      "learning_rate": 5.090050885280563e-06,
      "loss": 0.715,
      "step": 17681
    },
    {
      "epoch": 0.509479628882614,
      "grad_norm": 0.7596103549003601,
      "learning_rate": 5.089584350446001e-06,
      "loss": 0.7165,
      "step": 17682
    },
    {
      "epoch": 0.5095084423442633,
      "grad_norm": 0.7379186749458313,
      "learning_rate": 5.089117814831247e-06,
      "loss": 0.7458,
      "step": 17683
    },
    {
      "epoch": 0.5095372558059125,
      "grad_norm": 0.7294559478759766,
      "learning_rate": 5.088651278440367e-06,
      "loss": 0.6911,
      "step": 17684
    },
    {
      "epoch": 0.5095660692675618,
      "grad_norm": 0.7493026852607727,
      "learning_rate": 5.088184741277423e-06,
      "loss": 0.7332,
      "step": 17685
    },
    {
      "epoch": 0.509594882729211,
      "grad_norm": 0.7477341890335083,
      "learning_rate": 5.087718203346478e-06,
      "loss": 0.7315,
      "step": 17686
    },
    {
      "epoch": 0.5096236961908603,
      "grad_norm": 1.303336262702942,
      "learning_rate": 5.087251664651595e-06,
      "loss": 0.74,
      "step": 17687
    },
    {
      "epoch": 0.5096525096525096,
      "grad_norm": 0.7426522970199585,
      "learning_rate": 5.086785125196837e-06,
      "loss": 0.7385,
      "step": 17688
    },
    {
      "epoch": 0.5096813231141589,
      "grad_norm": 0.7389835119247437,
      "learning_rate": 5.086318584986267e-06,
      "loss": 0.7083,
      "step": 17689
    },
    {
      "epoch": 0.5097101365758082,
      "grad_norm": 0.7229263782501221,
      "learning_rate": 5.08585204402395e-06,
      "loss": 0.7104,
      "step": 17690
    },
    {
      "epoch": 0.5097389500374575,
      "grad_norm": 0.7396104335784912,
      "learning_rate": 5.085385502313947e-06,
      "loss": 0.7228,
      "step": 17691
    },
    {
      "epoch": 0.5097677634991068,
      "grad_norm": 0.7700437307357788,
      "learning_rate": 5.0849189598603196e-06,
      "loss": 0.7105,
      "step": 17692
    },
    {
      "epoch": 0.5097965769607561,
      "grad_norm": 0.7269206047058105,
      "learning_rate": 5.0844524166671345e-06,
      "loss": 0.7024,
      "step": 17693
    },
    {
      "epoch": 0.5098253904224054,
      "grad_norm": 0.742378830909729,
      "learning_rate": 5.083985872738454e-06,
      "loss": 0.7107,
      "step": 17694
    },
    {
      "epoch": 0.5098542038840547,
      "grad_norm": 0.7477046251296997,
      "learning_rate": 5.08351932807834e-06,
      "loss": 0.7047,
      "step": 17695
    },
    {
      "epoch": 0.509883017345704,
      "grad_norm": 0.7430925965309143,
      "learning_rate": 5.083052782690856e-06,
      "loss": 0.7174,
      "step": 17696
    },
    {
      "epoch": 0.5099118308073533,
      "grad_norm": 0.7446401715278625,
      "learning_rate": 5.082586236580066e-06,
      "loss": 0.743,
      "step": 17697
    },
    {
      "epoch": 0.5099406442690024,
      "grad_norm": 0.7421453595161438,
      "learning_rate": 5.0821196897500305e-06,
      "loss": 0.7133,
      "step": 17698
    },
    {
      "epoch": 0.5099694577306517,
      "grad_norm": 0.7562634944915771,
      "learning_rate": 5.081653142204816e-06,
      "loss": 0.6949,
      "step": 17699
    },
    {
      "epoch": 0.509998271192301,
      "grad_norm": 0.7446011900901794,
      "learning_rate": 5.081186593948484e-06,
      "loss": 0.7137,
      "step": 17700
    },
    {
      "epoch": 0.5100270846539503,
      "grad_norm": 0.7265300750732422,
      "learning_rate": 5.080720044985098e-06,
      "loss": 0.7128,
      "step": 17701
    },
    {
      "epoch": 0.5100558981155996,
      "grad_norm": 0.7359629273414612,
      "learning_rate": 5.080253495318721e-06,
      "loss": 0.7068,
      "step": 17702
    },
    {
      "epoch": 0.5100847115772489,
      "grad_norm": 0.7239793539047241,
      "learning_rate": 5.079786944953416e-06,
      "loss": 0.7506,
      "step": 17703
    },
    {
      "epoch": 0.5101135250388982,
      "grad_norm": 0.702093780040741,
      "learning_rate": 5.079320393893247e-06,
      "loss": 0.7053,
      "step": 17704
    },
    {
      "epoch": 0.5101423385005475,
      "grad_norm": 0.7303869128227234,
      "learning_rate": 5.078853842142276e-06,
      "loss": 0.7136,
      "step": 17705
    },
    {
      "epoch": 0.5101711519621968,
      "grad_norm": 0.742340624332428,
      "learning_rate": 5.078387289704568e-06,
      "loss": 0.7207,
      "step": 17706
    },
    {
      "epoch": 0.510199965423846,
      "grad_norm": 0.7169813513755798,
      "learning_rate": 5.077920736584183e-06,
      "loss": 0.7248,
      "step": 17707
    },
    {
      "epoch": 0.5102287788854953,
      "grad_norm": 0.7363290786743164,
      "learning_rate": 5.077454182785189e-06,
      "loss": 0.7188,
      "step": 17708
    },
    {
      "epoch": 0.5102575923471446,
      "grad_norm": 0.7308732271194458,
      "learning_rate": 5.076987628311645e-06,
      "loss": 0.7145,
      "step": 17709
    },
    {
      "epoch": 0.5102864058087939,
      "grad_norm": 0.7431098222732544,
      "learning_rate": 5.076521073167615e-06,
      "loss": 0.7009,
      "step": 17710
    },
    {
      "epoch": 0.5103152192704431,
      "grad_norm": 0.7283147573471069,
      "learning_rate": 5.076054517357164e-06,
      "loss": 0.7252,
      "step": 17711
    },
    {
      "epoch": 0.5103440327320924,
      "grad_norm": 0.7479749321937561,
      "learning_rate": 5.075587960884354e-06,
      "loss": 0.7238,
      "step": 17712
    },
    {
      "epoch": 0.5103728461937417,
      "grad_norm": 0.737596333026886,
      "learning_rate": 5.075121403753249e-06,
      "loss": 0.7234,
      "step": 17713
    },
    {
      "epoch": 0.510401659655391,
      "grad_norm": 0.6989437937736511,
      "learning_rate": 5.074654845967912e-06,
      "loss": 0.7041,
      "step": 17714
    },
    {
      "epoch": 0.5104304731170403,
      "grad_norm": 0.7335103154182434,
      "learning_rate": 5.074188287532404e-06,
      "loss": 0.6906,
      "step": 17715
    },
    {
      "epoch": 0.5104592865786896,
      "grad_norm": 0.7117849588394165,
      "learning_rate": 5.07372172845079e-06,
      "loss": 0.6995,
      "step": 17716
    },
    {
      "epoch": 0.5104881000403388,
      "grad_norm": 0.7258259057998657,
      "learning_rate": 5.073255168727136e-06,
      "loss": 0.6988,
      "step": 17717
    },
    {
      "epoch": 0.5105169135019881,
      "grad_norm": 0.7070773839950562,
      "learning_rate": 5.072788608365501e-06,
      "loss": 0.7063,
      "step": 17718
    },
    {
      "epoch": 0.5105457269636374,
      "grad_norm": 0.7308846712112427,
      "learning_rate": 5.072322047369951e-06,
      "loss": 0.7103,
      "step": 17719
    },
    {
      "epoch": 0.5105745404252867,
      "grad_norm": 0.722231388092041,
      "learning_rate": 5.071855485744547e-06,
      "loss": 0.6885,
      "step": 17720
    },
    {
      "epoch": 0.510603353886936,
      "grad_norm": 0.7227950692176819,
      "learning_rate": 5.071388923493354e-06,
      "loss": 0.7062,
      "step": 17721
    },
    {
      "epoch": 0.5106321673485853,
      "grad_norm": 0.711139976978302,
      "learning_rate": 5.070922360620435e-06,
      "loss": 0.7065,
      "step": 17722
    },
    {
      "epoch": 0.5106609808102346,
      "grad_norm": 0.7287235856056213,
      "learning_rate": 5.070455797129853e-06,
      "loss": 0.7177,
      "step": 17723
    },
    {
      "epoch": 0.5106897942718839,
      "grad_norm": 0.7285376787185669,
      "learning_rate": 5.069989233025672e-06,
      "loss": 0.7129,
      "step": 17724
    },
    {
      "epoch": 0.5107186077335331,
      "grad_norm": 0.6863698363304138,
      "learning_rate": 5.069522668311955e-06,
      "loss": 0.7034,
      "step": 17725
    },
    {
      "epoch": 0.5107474211951823,
      "grad_norm": 0.7227075695991516,
      "learning_rate": 5.069056102992762e-06,
      "loss": 0.7203,
      "step": 17726
    },
    {
      "epoch": 0.5107762346568316,
      "grad_norm": 0.715244710445404,
      "learning_rate": 5.068589537072163e-06,
      "loss": 0.7127,
      "step": 17727
    },
    {
      "epoch": 0.5108050481184809,
      "grad_norm": 0.7237220406532288,
      "learning_rate": 5.068122970554215e-06,
      "loss": 0.7114,
      "step": 17728
    },
    {
      "epoch": 0.5108338615801302,
      "grad_norm": 0.7290621995925903,
      "learning_rate": 5.067656403442986e-06,
      "loss": 0.7095,
      "step": 17729
    },
    {
      "epoch": 0.5108626750417795,
      "grad_norm": 0.7133857607841492,
      "learning_rate": 5.067189835742536e-06,
      "loss": 0.7267,
      "step": 17730
    },
    {
      "epoch": 0.5108914885034288,
      "grad_norm": 0.7184416055679321,
      "learning_rate": 5.066723267456931e-06,
      "loss": 0.7052,
      "step": 17731
    },
    {
      "epoch": 0.5109203019650781,
      "grad_norm": 0.728492021560669,
      "learning_rate": 5.066256698590233e-06,
      "loss": 0.7223,
      "step": 17732
    },
    {
      "epoch": 0.5109491154267274,
      "grad_norm": 0.7244483232498169,
      "learning_rate": 5.065790129146505e-06,
      "loss": 0.7111,
      "step": 17733
    },
    {
      "epoch": 0.5109779288883767,
      "grad_norm": 0.7019758820533752,
      "learning_rate": 5.06532355912981e-06,
      "loss": 0.7155,
      "step": 17734
    },
    {
      "epoch": 0.511006742350026,
      "grad_norm": 0.7287944555282593,
      "learning_rate": 5.064856988544213e-06,
      "loss": 0.7262,
      "step": 17735
    },
    {
      "epoch": 0.5110355558116753,
      "grad_norm": 0.6990882158279419,
      "learning_rate": 5.064390417393777e-06,
      "loss": 0.7036,
      "step": 17736
    },
    {
      "epoch": 0.5110643692733245,
      "grad_norm": 0.6945365071296692,
      "learning_rate": 5.063923845682564e-06,
      "loss": 0.6998,
      "step": 17737
    },
    {
      "epoch": 0.5110931827349737,
      "grad_norm": 0.7368948459625244,
      "learning_rate": 5.063457273414638e-06,
      "loss": 0.7073,
      "step": 17738
    },
    {
      "epoch": 0.511121996196623,
      "grad_norm": 0.7094247937202454,
      "learning_rate": 5.062990700594063e-06,
      "loss": 0.6934,
      "step": 17739
    },
    {
      "epoch": 0.5111508096582723,
      "grad_norm": 0.7133501768112183,
      "learning_rate": 5.062524127224902e-06,
      "loss": 0.6945,
      "step": 17740
    },
    {
      "epoch": 0.5111796231199216,
      "grad_norm": 0.7108411192893982,
      "learning_rate": 5.062057553311217e-06,
      "loss": 0.7281,
      "step": 17741
    },
    {
      "epoch": 0.5112084365815709,
      "grad_norm": 0.7319207787513733,
      "learning_rate": 5.0615909788570746e-06,
      "loss": 0.7526,
      "step": 17742
    },
    {
      "epoch": 0.5112372500432202,
      "grad_norm": 0.7211409211158752,
      "learning_rate": 5.061124403866535e-06,
      "loss": 0.7328,
      "step": 17743
    },
    {
      "epoch": 0.5112660635048695,
      "grad_norm": 0.6995422840118408,
      "learning_rate": 5.0606578283436646e-06,
      "loss": 0.7146,
      "step": 17744
    },
    {
      "epoch": 0.5112948769665188,
      "grad_norm": 0.712651789188385,
      "learning_rate": 5.060191252292523e-06,
      "loss": 0.7257,
      "step": 17745
    },
    {
      "epoch": 0.511323690428168,
      "grad_norm": 0.705959141254425,
      "learning_rate": 5.059724675717177e-06,
      "loss": 0.711,
      "step": 17746
    },
    {
      "epoch": 0.5113525038898173,
      "grad_norm": 0.7368792295455933,
      "learning_rate": 5.0592580986216886e-06,
      "loss": 0.7068,
      "step": 17747
    },
    {
      "epoch": 0.5113813173514666,
      "grad_norm": 0.701220691204071,
      "learning_rate": 5.058791521010121e-06,
      "loss": 0.6954,
      "step": 17748
    },
    {
      "epoch": 0.5114101308131159,
      "grad_norm": 0.7223172187805176,
      "learning_rate": 5.058324942886539e-06,
      "loss": 0.7162,
      "step": 17749
    },
    {
      "epoch": 0.5114389442747652,
      "grad_norm": 0.7290837168693542,
      "learning_rate": 5.057858364255005e-06,
      "loss": 0.695,
      "step": 17750
    },
    {
      "epoch": 0.5114677577364145,
      "grad_norm": 0.7259731292724609,
      "learning_rate": 5.057391785119582e-06,
      "loss": 0.7161,
      "step": 17751
    },
    {
      "epoch": 0.5114965711980637,
      "grad_norm": 0.7330029010772705,
      "learning_rate": 5.056925205484333e-06,
      "loss": 0.7268,
      "step": 17752
    },
    {
      "epoch": 0.511525384659713,
      "grad_norm": 0.7040134072303772,
      "learning_rate": 5.056458625353323e-06,
      "loss": 0.6777,
      "step": 17753
    },
    {
      "epoch": 0.5115541981213623,
      "grad_norm": 0.7194922566413879,
      "learning_rate": 5.055992044730616e-06,
      "loss": 0.7119,
      "step": 17754
    },
    {
      "epoch": 0.5115830115830116,
      "grad_norm": 0.7226914763450623,
      "learning_rate": 5.055525463620274e-06,
      "loss": 0.7166,
      "step": 17755
    },
    {
      "epoch": 0.5116118250446609,
      "grad_norm": 0.6969202160835266,
      "learning_rate": 5.05505888202636e-06,
      "loss": 0.6996,
      "step": 17756
    },
    {
      "epoch": 0.5116406385063101,
      "grad_norm": 0.6913061738014221,
      "learning_rate": 5.054592299952938e-06,
      "loss": 0.7119,
      "step": 17757
    },
    {
      "epoch": 0.5116694519679594,
      "grad_norm": 0.721705436706543,
      "learning_rate": 5.054125717404072e-06,
      "loss": 0.7151,
      "step": 17758
    },
    {
      "epoch": 0.5116982654296087,
      "grad_norm": 0.7193055152893066,
      "learning_rate": 5.053659134383825e-06,
      "loss": 0.7152,
      "step": 17759
    },
    {
      "epoch": 0.511727078891258,
      "grad_norm": 0.7305800318717957,
      "learning_rate": 5.05319255089626e-06,
      "loss": 0.7296,
      "step": 17760
    },
    {
      "epoch": 0.5117558923529073,
      "grad_norm": 0.8129780292510986,
      "learning_rate": 5.052725966945442e-06,
      "loss": 0.713,
      "step": 17761
    },
    {
      "epoch": 0.5117847058145566,
      "grad_norm": 0.734470546245575,
      "learning_rate": 5.052259382535433e-06,
      "loss": 0.708,
      "step": 17762
    },
    {
      "epoch": 0.5118135192762059,
      "grad_norm": 0.693881630897522,
      "learning_rate": 5.051792797670298e-06,
      "loss": 0.6982,
      "step": 17763
    },
    {
      "epoch": 0.5118423327378552,
      "grad_norm": 0.7229586839675903,
      "learning_rate": 5.051326212354097e-06,
      "loss": 0.7156,
      "step": 17764
    },
    {
      "epoch": 0.5118711461995044,
      "grad_norm": 0.7276299595832825,
      "learning_rate": 5.0508596265909e-06,
      "loss": 0.727,
      "step": 17765
    },
    {
      "epoch": 0.5118999596611536,
      "grad_norm": 0.7274452447891235,
      "learning_rate": 5.050393040384763e-06,
      "loss": 0.701,
      "step": 17766
    },
    {
      "epoch": 0.5119287731228029,
      "grad_norm": 0.7471339106559753,
      "learning_rate": 5.049926453739754e-06,
      "loss": 0.7166,
      "step": 17767
    },
    {
      "epoch": 0.5119575865844522,
      "grad_norm": 0.7406430244445801,
      "learning_rate": 5.049459866659936e-06,
      "loss": 0.7123,
      "step": 17768
    },
    {
      "epoch": 0.5119864000461015,
      "grad_norm": 0.7337425351142883,
      "learning_rate": 5.048993279149371e-06,
      "loss": 0.7111,
      "step": 17769
    },
    {
      "epoch": 0.5120152135077508,
      "grad_norm": 0.727977991104126,
      "learning_rate": 5.048526691212124e-06,
      "loss": 0.7185,
      "step": 17770
    },
    {
      "epoch": 0.5120440269694001,
      "grad_norm": 0.7032516598701477,
      "learning_rate": 5.048060102852258e-06,
      "loss": 0.6788,
      "step": 17771
    },
    {
      "epoch": 0.5120728404310494,
      "grad_norm": 0.7314479351043701,
      "learning_rate": 5.047593514073837e-06,
      "loss": 0.7211,
      "step": 17772
    },
    {
      "epoch": 0.5121016538926987,
      "grad_norm": 0.7300496697425842,
      "learning_rate": 5.047126924880924e-06,
      "loss": 0.7024,
      "step": 17773
    },
    {
      "epoch": 0.512130467354348,
      "grad_norm": 0.7754871249198914,
      "learning_rate": 5.046660335277583e-06,
      "loss": 0.7061,
      "step": 17774
    },
    {
      "epoch": 0.5121592808159973,
      "grad_norm": 0.7309178709983826,
      "learning_rate": 5.0461937452678764e-06,
      "loss": 0.6796,
      "step": 17775
    },
    {
      "epoch": 0.5121880942776466,
      "grad_norm": 0.7194788455963135,
      "learning_rate": 5.045727154855868e-06,
      "loss": 0.7153,
      "step": 17776
    },
    {
      "epoch": 0.5122169077392958,
      "grad_norm": 0.7741347551345825,
      "learning_rate": 5.045260564045623e-06,
      "loss": 0.7083,
      "step": 17777
    },
    {
      "epoch": 0.5122457212009451,
      "grad_norm": 0.7541149258613586,
      "learning_rate": 5.044793972841203e-06,
      "loss": 0.723,
      "step": 17778
    },
    {
      "epoch": 0.5122745346625943,
      "grad_norm": 0.7524648308753967,
      "learning_rate": 5.044327381246673e-06,
      "loss": 0.7187,
      "step": 17779
    },
    {
      "epoch": 0.5123033481242436,
      "grad_norm": 0.7459410429000854,
      "learning_rate": 5.0438607892660955e-06,
      "loss": 0.7298,
      "step": 17780
    },
    {
      "epoch": 0.5123321615858929,
      "grad_norm": 0.7107844948768616,
      "learning_rate": 5.043394196903534e-06,
      "loss": 0.7002,
      "step": 17781
    },
    {
      "epoch": 0.5123609750475422,
      "grad_norm": 0.7053483724594116,
      "learning_rate": 5.0429276041630515e-06,
      "loss": 0.7161,
      "step": 17782
    },
    {
      "epoch": 0.5123897885091915,
      "grad_norm": 0.7535773515701294,
      "learning_rate": 5.042461011048713e-06,
      "loss": 0.7115,
      "step": 17783
    },
    {
      "epoch": 0.5124186019708408,
      "grad_norm": 0.7364829182624817,
      "learning_rate": 5.041994417564583e-06,
      "loss": 0.742,
      "step": 17784
    },
    {
      "epoch": 0.5124474154324901,
      "grad_norm": 0.7345860004425049,
      "learning_rate": 5.041527823714724e-06,
      "loss": 0.7057,
      "step": 17785
    },
    {
      "epoch": 0.5124762288941394,
      "grad_norm": 0.741270899772644,
      "learning_rate": 5.041061229503196e-06,
      "loss": 0.7056,
      "step": 17786
    },
    {
      "epoch": 0.5125050423557886,
      "grad_norm": 0.7336865663528442,
      "learning_rate": 5.0405946349340676e-06,
      "loss": 0.7275,
      "step": 17787
    },
    {
      "epoch": 0.5125338558174379,
      "grad_norm": 0.7436368465423584,
      "learning_rate": 5.0401280400114e-06,
      "loss": 0.7379,
      "step": 17788
    },
    {
      "epoch": 0.5125626692790872,
      "grad_norm": 0.7349768877029419,
      "learning_rate": 5.039661444739258e-06,
      "loss": 0.7329,
      "step": 17789
    },
    {
      "epoch": 0.5125914827407365,
      "grad_norm": 0.732264518737793,
      "learning_rate": 5.0391948491217045e-06,
      "loss": 0.7164,
      "step": 17790
    },
    {
      "epoch": 0.5126202962023858,
      "grad_norm": 0.7183787226676941,
      "learning_rate": 5.038728253162803e-06,
      "loss": 0.7081,
      "step": 17791
    },
    {
      "epoch": 0.512649109664035,
      "grad_norm": 0.751264750957489,
      "learning_rate": 5.038261656866617e-06,
      "loss": 0.6996,
      "step": 17792
    },
    {
      "epoch": 0.5126779231256843,
      "grad_norm": 0.6936983466148376,
      "learning_rate": 5.0377950602372105e-06,
      "loss": 0.6921,
      "step": 17793
    },
    {
      "epoch": 0.5127067365873336,
      "grad_norm": 0.7147209048271179,
      "learning_rate": 5.0373284632786455e-06,
      "loss": 0.7187,
      "step": 17794
    },
    {
      "epoch": 0.5127355500489829,
      "grad_norm": 0.72137051820755,
      "learning_rate": 5.036861865994988e-06,
      "loss": 0.7181,
      "step": 17795
    },
    {
      "epoch": 0.5127643635106321,
      "grad_norm": 0.7182273268699646,
      "learning_rate": 5.036395268390301e-06,
      "loss": 0.7042,
      "step": 17796
    },
    {
      "epoch": 0.5127931769722814,
      "grad_norm": 0.703520655632019,
      "learning_rate": 5.035928670468646e-06,
      "loss": 0.725,
      "step": 17797
    },
    {
      "epoch": 0.5128219904339307,
      "grad_norm": 0.7033493518829346,
      "learning_rate": 5.035462072234089e-06,
      "loss": 0.6979,
      "step": 17798
    },
    {
      "epoch": 0.51285080389558,
      "grad_norm": 0.7329707741737366,
      "learning_rate": 5.034995473690694e-06,
      "loss": 0.6945,
      "step": 17799
    },
    {
      "epoch": 0.5128796173572293,
      "grad_norm": 0.7129610776901245,
      "learning_rate": 5.034528874842522e-06,
      "loss": 0.7121,
      "step": 17800
    },
    {
      "epoch": 0.5129084308188786,
      "grad_norm": 0.7347707748413086,
      "learning_rate": 5.034062275693637e-06,
      "loss": 0.7083,
      "step": 17801
    },
    {
      "epoch": 0.5129372442805279,
      "grad_norm": 0.7166877388954163,
      "learning_rate": 5.0335956762481065e-06,
      "loss": 0.7417,
      "step": 17802
    },
    {
      "epoch": 0.5129660577421772,
      "grad_norm": 0.7077612280845642,
      "learning_rate": 5.033129076509989e-06,
      "loss": 0.6946,
      "step": 17803
    },
    {
      "epoch": 0.5129948712038265,
      "grad_norm": 0.720490038394928,
      "learning_rate": 5.032662476483351e-06,
      "loss": 0.6947,
      "step": 17804
    },
    {
      "epoch": 0.5130236846654758,
      "grad_norm": 0.7332951426506042,
      "learning_rate": 5.032195876172256e-06,
      "loss": 0.7117,
      "step": 17805
    },
    {
      "epoch": 0.513052498127125,
      "grad_norm": 0.7235827445983887,
      "learning_rate": 5.0317292755807665e-06,
      "loss": 0.7179,
      "step": 17806
    },
    {
      "epoch": 0.5130813115887742,
      "grad_norm": 0.718310534954071,
      "learning_rate": 5.0312626747129454e-06,
      "loss": 0.6955,
      "step": 17807
    },
    {
      "epoch": 0.5131101250504235,
      "grad_norm": 0.7400369644165039,
      "learning_rate": 5.03079607357286e-06,
      "loss": 0.7285,
      "step": 17808
    },
    {
      "epoch": 0.5131389385120728,
      "grad_norm": 0.7701077461242676,
      "learning_rate": 5.03032947216457e-06,
      "loss": 0.7197,
      "step": 17809
    },
    {
      "epoch": 0.5131677519737221,
      "grad_norm": 0.7228721380233765,
      "learning_rate": 5.029862870492142e-06,
      "loss": 0.7459,
      "step": 17810
    },
    {
      "epoch": 0.5131965654353714,
      "grad_norm": 0.7165148854255676,
      "learning_rate": 5.029396268559637e-06,
      "loss": 0.6965,
      "step": 17811
    },
    {
      "epoch": 0.5132253788970207,
      "grad_norm": 0.7429746985435486,
      "learning_rate": 5.028929666371121e-06,
      "loss": 0.6909,
      "step": 17812
    },
    {
      "epoch": 0.51325419235867,
      "grad_norm": 0.795987606048584,
      "learning_rate": 5.028463063930655e-06,
      "loss": 0.7631,
      "step": 17813
    },
    {
      "epoch": 0.5132830058203193,
      "grad_norm": 0.7554196715354919,
      "learning_rate": 5.027996461242306e-06,
      "loss": 0.6877,
      "step": 17814
    },
    {
      "epoch": 0.5133118192819686,
      "grad_norm": 0.7401406168937683,
      "learning_rate": 5.0275298583101354e-06,
      "loss": 0.7079,
      "step": 17815
    },
    {
      "epoch": 0.5133406327436179,
      "grad_norm": 0.7330971956253052,
      "learning_rate": 5.027063255138207e-06,
      "loss": 0.704,
      "step": 17816
    },
    {
      "epoch": 0.5133694462052671,
      "grad_norm": 0.7338109612464905,
      "learning_rate": 5.026596651730584e-06,
      "loss": 0.6878,
      "step": 17817
    },
    {
      "epoch": 0.5133982596669164,
      "grad_norm": 0.7607137560844421,
      "learning_rate": 5.026130048091331e-06,
      "loss": 0.7413,
      "step": 17818
    },
    {
      "epoch": 0.5134270731285656,
      "grad_norm": 0.7239946722984314,
      "learning_rate": 5.025663444224512e-06,
      "loss": 0.7246,
      "step": 17819
    },
    {
      "epoch": 0.5134558865902149,
      "grad_norm": 0.7253960371017456,
      "learning_rate": 5.02519684013419e-06,
      "loss": 0.6812,
      "step": 17820
    },
    {
      "epoch": 0.5134847000518642,
      "grad_norm": 0.7251477837562561,
      "learning_rate": 5.024730235824429e-06,
      "loss": 0.717,
      "step": 17821
    },
    {
      "epoch": 0.5135135135135135,
      "grad_norm": 0.721221387386322,
      "learning_rate": 5.024263631299292e-06,
      "loss": 0.7173,
      "step": 17822
    },
    {
      "epoch": 0.5135423269751628,
      "grad_norm": 0.7483004927635193,
      "learning_rate": 5.023797026562841e-06,
      "loss": 0.6987,
      "step": 17823
    },
    {
      "epoch": 0.5135711404368121,
      "grad_norm": 0.7525590062141418,
      "learning_rate": 5.023330421619145e-06,
      "loss": 0.7384,
      "step": 17824
    },
    {
      "epoch": 0.5135999538984614,
      "grad_norm": 0.7180529236793518,
      "learning_rate": 5.022863816472262e-06,
      "loss": 0.6824,
      "step": 17825
    },
    {
      "epoch": 0.5136287673601106,
      "grad_norm": 0.7349746823310852,
      "learning_rate": 5.0223972111262584e-06,
      "loss": 0.7293,
      "step": 17826
    },
    {
      "epoch": 0.5136575808217599,
      "grad_norm": 0.7396920919418335,
      "learning_rate": 5.021930605585198e-06,
      "loss": 0.6863,
      "step": 17827
    },
    {
      "epoch": 0.5136863942834092,
      "grad_norm": 0.7531265020370483,
      "learning_rate": 5.021463999853145e-06,
      "loss": 0.7322,
      "step": 17828
    },
    {
      "epoch": 0.5137152077450585,
      "grad_norm": 0.7872257828712463,
      "learning_rate": 5.0209973939341605e-06,
      "loss": 0.7223,
      "step": 17829
    },
    {
      "epoch": 0.5137440212067078,
      "grad_norm": 0.7340866327285767,
      "learning_rate": 5.020530787832309e-06,
      "loss": 0.7395,
      "step": 17830
    },
    {
      "epoch": 0.5137728346683571,
      "grad_norm": 0.7200732827186584,
      "learning_rate": 5.020064181551656e-06,
      "loss": 0.7136,
      "step": 17831
    },
    {
      "epoch": 0.5138016481300064,
      "grad_norm": 0.7456483840942383,
      "learning_rate": 5.019597575096265e-06,
      "loss": 0.7162,
      "step": 17832
    },
    {
      "epoch": 0.5138304615916556,
      "grad_norm": 0.7054909467697144,
      "learning_rate": 5.019130968470197e-06,
      "loss": 0.7036,
      "step": 17833
    },
    {
      "epoch": 0.5138592750533049,
      "grad_norm": 0.7254492044448853,
      "learning_rate": 5.0186643616775185e-06,
      "loss": 0.7057,
      "step": 17834
    },
    {
      "epoch": 0.5138880885149542,
      "grad_norm": 0.7164674997329712,
      "learning_rate": 5.018197754722293e-06,
      "loss": 0.6929,
      "step": 17835
    },
    {
      "epoch": 0.5139169019766034,
      "grad_norm": 0.7249496579170227,
      "learning_rate": 5.01773114760858e-06,
      "loss": 0.7067,
      "step": 17836
    },
    {
      "epoch": 0.5139457154382527,
      "grad_norm": 0.7660752534866333,
      "learning_rate": 5.017264540340449e-06,
      "loss": 0.7216,
      "step": 17837
    },
    {
      "epoch": 0.513974528899902,
      "grad_norm": 0.7776814699172974,
      "learning_rate": 5.01679793292196e-06,
      "loss": 0.7227,
      "step": 17838
    },
    {
      "epoch": 0.5140033423615513,
      "grad_norm": 0.741105854511261,
      "learning_rate": 5.016331325357179e-06,
      "loss": 0.7313,
      "step": 17839
    },
    {
      "epoch": 0.5140321558232006,
      "grad_norm": 0.7214758396148682,
      "learning_rate": 5.015864717650168e-06,
      "loss": 0.7332,
      "step": 17840
    },
    {
      "epoch": 0.5140609692848499,
      "grad_norm": 0.7420104146003723,
      "learning_rate": 5.0153981098049906e-06,
      "loss": 0.7392,
      "step": 17841
    },
    {
      "epoch": 0.5140897827464992,
      "grad_norm": 0.730713963508606,
      "learning_rate": 5.0149315018257104e-06,
      "loss": 0.7012,
      "step": 17842
    },
    {
      "epoch": 0.5141185962081485,
      "grad_norm": 0.7446051836013794,
      "learning_rate": 5.014464893716393e-06,
      "loss": 0.7324,
      "step": 17843
    },
    {
      "epoch": 0.5141474096697978,
      "grad_norm": 0.8298376202583313,
      "learning_rate": 5.0139982854811005e-06,
      "loss": 0.737,
      "step": 17844
    },
    {
      "epoch": 0.5141762231314471,
      "grad_norm": 0.7762433290481567,
      "learning_rate": 5.013531677123897e-06,
      "loss": 0.7167,
      "step": 17845
    },
    {
      "epoch": 0.5142050365930962,
      "grad_norm": 0.7282149195671082,
      "learning_rate": 5.013065068648846e-06,
      "loss": 0.7207,
      "step": 17846
    },
    {
      "epoch": 0.5142338500547455,
      "grad_norm": 0.7285406589508057,
      "learning_rate": 5.012598460060012e-06,
      "loss": 0.6915,
      "step": 17847
    },
    {
      "epoch": 0.5142626635163948,
      "grad_norm": 0.7323839664459229,
      "learning_rate": 5.0121318513614555e-06,
      "loss": 0.7367,
      "step": 17848
    },
    {
      "epoch": 0.5142914769780441,
      "grad_norm": 0.7238113880157471,
      "learning_rate": 5.011665242557246e-06,
      "loss": 0.7046,
      "step": 17849
    },
    {
      "epoch": 0.5143202904396934,
      "grad_norm": 0.7234294414520264,
      "learning_rate": 5.011198633651442e-06,
      "loss": 0.6895,
      "step": 17850
    },
    {
      "epoch": 0.5143491039013427,
      "grad_norm": 0.7224965691566467,
      "learning_rate": 5.010732024648111e-06,
      "loss": 0.722,
      "step": 17851
    },
    {
      "epoch": 0.514377917362992,
      "grad_norm": 0.7170881628990173,
      "learning_rate": 5.010265415551313e-06,
      "loss": 0.7385,
      "step": 17852
    },
    {
      "epoch": 0.5144067308246413,
      "grad_norm": 0.7686486840248108,
      "learning_rate": 5.0097988063651135e-06,
      "loss": 0.7259,
      "step": 17853
    },
    {
      "epoch": 0.5144355442862906,
      "grad_norm": 0.728977620601654,
      "learning_rate": 5.009332197093576e-06,
      "loss": 0.712,
      "step": 17854
    },
    {
      "epoch": 0.5144643577479399,
      "grad_norm": 0.7277094721794128,
      "learning_rate": 5.008865587740765e-06,
      "loss": 0.7093,
      "step": 17855
    },
    {
      "epoch": 0.5144931712095892,
      "grad_norm": 0.7475506067276001,
      "learning_rate": 5.008398978310745e-06,
      "loss": 0.7352,
      "step": 17856
    },
    {
      "epoch": 0.5145219846712384,
      "grad_norm": 0.7232934236526489,
      "learning_rate": 5.007932368807576e-06,
      "loss": 0.7219,
      "step": 17857
    },
    {
      "epoch": 0.5145507981328877,
      "grad_norm": 0.7517922520637512,
      "learning_rate": 5.007465759235325e-06,
      "loss": 0.7353,
      "step": 17858
    },
    {
      "epoch": 0.514579611594537,
      "grad_norm": 0.7526842355728149,
      "learning_rate": 5.006999149598055e-06,
      "loss": 0.7266,
      "step": 17859
    },
    {
      "epoch": 0.5146084250561862,
      "grad_norm": 0.7335777282714844,
      "learning_rate": 5.0065325398998275e-06,
      "loss": 0.7098,
      "step": 17860
    },
    {
      "epoch": 0.5146372385178355,
      "grad_norm": 0.7353055477142334,
      "learning_rate": 5.00606593014471e-06,
      "loss": 0.7014,
      "step": 17861
    },
    {
      "epoch": 0.5146660519794848,
      "grad_norm": 0.7223108410835266,
      "learning_rate": 5.005599320336765e-06,
      "loss": 0.7154,
      "step": 17862
    },
    {
      "epoch": 0.5146948654411341,
      "grad_norm": 0.7302128076553345,
      "learning_rate": 5.005132710480054e-06,
      "loss": 0.692,
      "step": 17863
    },
    {
      "epoch": 0.5147236789027834,
      "grad_norm": 0.7446293830871582,
      "learning_rate": 5.004666100578642e-06,
      "loss": 0.7303,
      "step": 17864
    },
    {
      "epoch": 0.5147524923644327,
      "grad_norm": 0.728519856929779,
      "learning_rate": 5.0041994906365935e-06,
      "loss": 0.7317,
      "step": 17865
    },
    {
      "epoch": 0.514781305826082,
      "grad_norm": 0.7189137935638428,
      "learning_rate": 5.003732880657971e-06,
      "loss": 0.6894,
      "step": 17866
    },
    {
      "epoch": 0.5148101192877312,
      "grad_norm": 0.7096177339553833,
      "learning_rate": 5.003266270646839e-06,
      "loss": 0.7135,
      "step": 17867
    },
    {
      "epoch": 0.5148389327493805,
      "grad_norm": 0.706440269947052,
      "learning_rate": 5.002799660607262e-06,
      "loss": 0.7138,
      "step": 17868
    },
    {
      "epoch": 0.5148677462110298,
      "grad_norm": 0.7126495838165283,
      "learning_rate": 5.002333050543303e-06,
      "loss": 0.7044,
      "step": 17869
    },
    {
      "epoch": 0.5148965596726791,
      "grad_norm": 0.6992453336715698,
      "learning_rate": 5.001866440459025e-06,
      "loss": 0.7206,
      "step": 17870
    },
    {
      "epoch": 0.5149253731343284,
      "grad_norm": 0.7356342077255249,
      "learning_rate": 5.001399830358492e-06,
      "loss": 0.7146,
      "step": 17871
    },
    {
      "epoch": 0.5149541865959777,
      "grad_norm": 0.709189772605896,
      "learning_rate": 5.000933220245767e-06,
      "loss": 0.7027,
      "step": 17872
    },
    {
      "epoch": 0.5149830000576269,
      "grad_norm": 0.7325162291526794,
      "learning_rate": 5.000466610124916e-06,
      "loss": 0.7415,
      "step": 17873
    },
    {
      "epoch": 0.5150118135192762,
      "grad_norm": 0.730043888092041,
      "learning_rate": 5e-06,
      "loss": 0.7148,
      "step": 17874
    },
    {
      "epoch": 0.5150406269809255,
      "grad_norm": 0.696411669254303,
      "learning_rate": 4.999533389875086e-06,
      "loss": 0.6967,
      "step": 17875
    },
    {
      "epoch": 0.5150694404425747,
      "grad_norm": 0.7507047057151794,
      "learning_rate": 4.9990667797542346e-06,
      "loss": 0.7356,
      "step": 17876
    },
    {
      "epoch": 0.515098253904224,
      "grad_norm": 0.7409246563911438,
      "learning_rate": 4.9986001696415105e-06,
      "loss": 0.7178,
      "step": 17877
    },
    {
      "epoch": 0.5151270673658733,
      "grad_norm": 0.726615309715271,
      "learning_rate": 4.998133559540978e-06,
      "loss": 0.7371,
      "step": 17878
    },
    {
      "epoch": 0.5151558808275226,
      "grad_norm": 0.7214285731315613,
      "learning_rate": 4.997666949456699e-06,
      "loss": 0.7065,
      "step": 17879
    },
    {
      "epoch": 0.5151846942891719,
      "grad_norm": 0.71120685338974,
      "learning_rate": 4.997200339392738e-06,
      "loss": 0.6964,
      "step": 17880
    },
    {
      "epoch": 0.5152135077508212,
      "grad_norm": 0.7033407092094421,
      "learning_rate": 4.996733729353161e-06,
      "loss": 0.7117,
      "step": 17881
    },
    {
      "epoch": 0.5152423212124705,
      "grad_norm": 0.7346618175506592,
      "learning_rate": 4.996267119342029e-06,
      "loss": 0.6869,
      "step": 17882
    },
    {
      "epoch": 0.5152711346741198,
      "grad_norm": 0.7249078154563904,
      "learning_rate": 4.995800509363407e-06,
      "loss": 0.718,
      "step": 17883
    },
    {
      "epoch": 0.5152999481357691,
      "grad_norm": 0.7227904796600342,
      "learning_rate": 4.99533389942136e-06,
      "loss": 0.7174,
      "step": 17884
    },
    {
      "epoch": 0.5153287615974184,
      "grad_norm": 0.7067409753799438,
      "learning_rate": 4.994867289519947e-06,
      "loss": 0.6844,
      "step": 17885
    },
    {
      "epoch": 0.5153575750590677,
      "grad_norm": 0.7364872694015503,
      "learning_rate": 4.994400679663237e-06,
      "loss": 0.6777,
      "step": 17886
    },
    {
      "epoch": 0.5153863885207168,
      "grad_norm": 0.7259604930877686,
      "learning_rate": 4.993934069855291e-06,
      "loss": 0.7106,
      "step": 17887
    },
    {
      "epoch": 0.5154152019823661,
      "grad_norm": 0.741304337978363,
      "learning_rate": 4.993467460100173e-06,
      "loss": 0.7201,
      "step": 17888
    },
    {
      "epoch": 0.5154440154440154,
      "grad_norm": 0.6924660801887512,
      "learning_rate": 4.993000850401947e-06,
      "loss": 0.701,
      "step": 17889
    },
    {
      "epoch": 0.5154728289056647,
      "grad_norm": 0.7297789454460144,
      "learning_rate": 4.992534240764677e-06,
      "loss": 0.7094,
      "step": 17890
    },
    {
      "epoch": 0.515501642367314,
      "grad_norm": 0.765187680721283,
      "learning_rate": 4.9920676311924264e-06,
      "loss": 0.7241,
      "step": 17891
    },
    {
      "epoch": 0.5155304558289633,
      "grad_norm": 0.71688312292099,
      "learning_rate": 4.991601021689256e-06,
      "loss": 0.7147,
      "step": 17892
    },
    {
      "epoch": 0.5155592692906126,
      "grad_norm": 0.7324289083480835,
      "learning_rate": 4.9911344122592355e-06,
      "loss": 0.723,
      "step": 17893
    },
    {
      "epoch": 0.5155880827522619,
      "grad_norm": 0.7555705308914185,
      "learning_rate": 4.990667802906425e-06,
      "loss": 0.7524,
      "step": 17894
    },
    {
      "epoch": 0.5156168962139112,
      "grad_norm": 0.7486830353736877,
      "learning_rate": 4.990201193634887e-06,
      "loss": 0.7266,
      "step": 17895
    },
    {
      "epoch": 0.5156457096755604,
      "grad_norm": 0.7242740988731384,
      "learning_rate": 4.9897345844486885e-06,
      "loss": 0.7152,
      "step": 17896
    },
    {
      "epoch": 0.5156745231372097,
      "grad_norm": 0.7170915007591248,
      "learning_rate": 4.989267975351892e-06,
      "loss": 0.7149,
      "step": 17897
    },
    {
      "epoch": 0.515703336598859,
      "grad_norm": 0.7248463034629822,
      "learning_rate": 4.988801366348559e-06,
      "loss": 0.7217,
      "step": 17898
    },
    {
      "epoch": 0.5157321500605083,
      "grad_norm": 0.7273274660110474,
      "learning_rate": 4.988334757442756e-06,
      "loss": 0.695,
      "step": 17899
    },
    {
      "epoch": 0.5157609635221575,
      "grad_norm": 0.7254812121391296,
      "learning_rate": 4.987868148638545e-06,
      "loss": 0.7163,
      "step": 17900
    },
    {
      "epoch": 0.5157897769838068,
      "grad_norm": 0.7556173801422119,
      "learning_rate": 4.987401539939991e-06,
      "loss": 0.7094,
      "step": 17901
    },
    {
      "epoch": 0.5158185904454561,
      "grad_norm": 0.7205643653869629,
      "learning_rate": 4.9869349313511555e-06,
      "loss": 0.7138,
      "step": 17902
    },
    {
      "epoch": 0.5158474039071054,
      "grad_norm": 0.7446810007095337,
      "learning_rate": 4.9864683228761045e-06,
      "loss": 0.7052,
      "step": 17903
    },
    {
      "epoch": 0.5158762173687547,
      "grad_norm": 0.7265756130218506,
      "learning_rate": 4.9860017145189e-06,
      "loss": 0.7148,
      "step": 17904
    },
    {
      "epoch": 0.515905030830404,
      "grad_norm": 0.7362923622131348,
      "learning_rate": 4.985535106283608e-06,
      "loss": 0.738,
      "step": 17905
    },
    {
      "epoch": 0.5159338442920532,
      "grad_norm": 0.7092210054397583,
      "learning_rate": 4.98506849817429e-06,
      "loss": 0.7124,
      "step": 17906
    },
    {
      "epoch": 0.5159626577537025,
      "grad_norm": 0.7338472008705139,
      "learning_rate": 4.98460189019501e-06,
      "loss": 0.6872,
      "step": 17907
    },
    {
      "epoch": 0.5159914712153518,
      "grad_norm": 0.7318124175071716,
      "learning_rate": 4.984135282349833e-06,
      "loss": 0.7022,
      "step": 17908
    },
    {
      "epoch": 0.5160202846770011,
      "grad_norm": 0.6982688903808594,
      "learning_rate": 4.9836686746428225e-06,
      "loss": 0.6993,
      "step": 17909
    },
    {
      "epoch": 0.5160490981386504,
      "grad_norm": 0.7153250575065613,
      "learning_rate": 4.983202067078041e-06,
      "loss": 0.7073,
      "step": 17910
    },
    {
      "epoch": 0.5160779116002997,
      "grad_norm": 0.7090291380882263,
      "learning_rate": 4.982735459659553e-06,
      "loss": 0.7,
      "step": 17911
    },
    {
      "epoch": 0.516106725061949,
      "grad_norm": 0.7060283422470093,
      "learning_rate": 4.982268852391421e-06,
      "loss": 0.7086,
      "step": 17912
    },
    {
      "epoch": 0.5161355385235982,
      "grad_norm": 0.7276290655136108,
      "learning_rate": 4.981802245277711e-06,
      "loss": 0.6856,
      "step": 17913
    },
    {
      "epoch": 0.5161643519852475,
      "grad_norm": 0.7371846437454224,
      "learning_rate": 4.981335638322484e-06,
      "loss": 0.7401,
      "step": 17914
    },
    {
      "epoch": 0.5161931654468968,
      "grad_norm": 0.7468152642250061,
      "learning_rate": 4.980869031529806e-06,
      "loss": 0.697,
      "step": 17915
    },
    {
      "epoch": 0.516221978908546,
      "grad_norm": 0.7364240288734436,
      "learning_rate": 4.980402424903736e-06,
      "loss": 0.7046,
      "step": 17916
    },
    {
      "epoch": 0.5162507923701953,
      "grad_norm": 0.7355106472969055,
      "learning_rate": 4.979935818448344e-06,
      "loss": 0.7173,
      "step": 17917
    },
    {
      "epoch": 0.5162796058318446,
      "grad_norm": 0.7246458530426025,
      "learning_rate": 4.979469212167691e-06,
      "loss": 0.7318,
      "step": 17918
    },
    {
      "epoch": 0.5163084192934939,
      "grad_norm": 0.7376742959022522,
      "learning_rate": 4.979002606065841e-06,
      "loss": 0.695,
      "step": 17919
    },
    {
      "epoch": 0.5163372327551432,
      "grad_norm": 0.7332826852798462,
      "learning_rate": 4.978536000146858e-06,
      "loss": 0.7018,
      "step": 17920
    },
    {
      "epoch": 0.5163660462167925,
      "grad_norm": 0.7555997967720032,
      "learning_rate": 4.9780693944148034e-06,
      "loss": 0.7139,
      "step": 17921
    },
    {
      "epoch": 0.5163948596784418,
      "grad_norm": 0.7853091359138489,
      "learning_rate": 4.977602788873743e-06,
      "loss": 0.7405,
      "step": 17922
    },
    {
      "epoch": 0.5164236731400911,
      "grad_norm": 0.7265124320983887,
      "learning_rate": 4.9771361835277395e-06,
      "loss": 0.7224,
      "step": 17923
    },
    {
      "epoch": 0.5164524866017404,
      "grad_norm": 0.7117075324058533,
      "learning_rate": 4.976669578380857e-06,
      "loss": 0.7005,
      "step": 17924
    },
    {
      "epoch": 0.5164813000633897,
      "grad_norm": 0.7272940874099731,
      "learning_rate": 4.97620297343716e-06,
      "loss": 0.731,
      "step": 17925
    },
    {
      "epoch": 0.516510113525039,
      "grad_norm": 0.7582390308380127,
      "learning_rate": 4.975736368700711e-06,
      "loss": 0.7417,
      "step": 17926
    },
    {
      "epoch": 0.5165389269866881,
      "grad_norm": 0.7139357328414917,
      "learning_rate": 4.975269764175574e-06,
      "loss": 0.7233,
      "step": 17927
    },
    {
      "epoch": 0.5165677404483374,
      "grad_norm": 0.7326257228851318,
      "learning_rate": 4.97480315986581e-06,
      "loss": 0.7291,
      "step": 17928
    },
    {
      "epoch": 0.5165965539099867,
      "grad_norm": 0.7589621543884277,
      "learning_rate": 4.974336555775489e-06,
      "loss": 0.7087,
      "step": 17929
    },
    {
      "epoch": 0.516625367371636,
      "grad_norm": 0.7267866730690002,
      "learning_rate": 4.973869951908669e-06,
      "loss": 0.7079,
      "step": 17930
    },
    {
      "epoch": 0.5166541808332853,
      "grad_norm": 0.7426979541778564,
      "learning_rate": 4.973403348269417e-06,
      "loss": 0.7057,
      "step": 17931
    },
    {
      "epoch": 0.5166829942949346,
      "grad_norm": 0.7309293746948242,
      "learning_rate": 4.972936744861794e-06,
      "loss": 0.7176,
      "step": 17932
    },
    {
      "epoch": 0.5167118077565839,
      "grad_norm": 0.7263676524162292,
      "learning_rate": 4.972470141689866e-06,
      "loss": 0.723,
      "step": 17933
    },
    {
      "epoch": 0.5167406212182332,
      "grad_norm": 0.7222920656204224,
      "learning_rate": 4.9720035387576955e-06,
      "loss": 0.7148,
      "step": 17934
    },
    {
      "epoch": 0.5167694346798825,
      "grad_norm": 0.7543646693229675,
      "learning_rate": 4.971536936069346e-06,
      "loss": 0.7162,
      "step": 17935
    },
    {
      "epoch": 0.5167982481415317,
      "grad_norm": 0.756782054901123,
      "learning_rate": 4.971070333628881e-06,
      "loss": 0.7301,
      "step": 17936
    },
    {
      "epoch": 0.516827061603181,
      "grad_norm": 2.0696325302124023,
      "learning_rate": 4.970603731440364e-06,
      "loss": 0.7136,
      "step": 17937
    },
    {
      "epoch": 0.5168558750648303,
      "grad_norm": 0.7284924387931824,
      "learning_rate": 4.97013712950786e-06,
      "loss": 0.7101,
      "step": 17938
    },
    {
      "epoch": 0.5168846885264796,
      "grad_norm": 0.7334122061729431,
      "learning_rate": 4.9696705278354315e-06,
      "loss": 0.696,
      "step": 17939
    },
    {
      "epoch": 0.5169135019881288,
      "grad_norm": 0.7132877707481384,
      "learning_rate": 4.969203926427141e-06,
      "loss": 0.7028,
      "step": 17940
    },
    {
      "epoch": 0.5169423154497781,
      "grad_norm": 0.7104291319847107,
      "learning_rate": 4.9687373252870545e-06,
      "loss": 0.7062,
      "step": 17941
    },
    {
      "epoch": 0.5169711289114274,
      "grad_norm": 0.720157265663147,
      "learning_rate": 4.968270724419235e-06,
      "loss": 0.7092,
      "step": 17942
    },
    {
      "epoch": 0.5169999423730767,
      "grad_norm": 0.7067646384239197,
      "learning_rate": 4.967804123827746e-06,
      "loss": 0.6951,
      "step": 17943
    },
    {
      "epoch": 0.517028755834726,
      "grad_norm": 0.7460079193115234,
      "learning_rate": 4.967337523516651e-06,
      "loss": 0.7332,
      "step": 17944
    },
    {
      "epoch": 0.5170575692963753,
      "grad_norm": 0.7131574749946594,
      "learning_rate": 4.966870923490012e-06,
      "loss": 0.7165,
      "step": 17945
    },
    {
      "epoch": 0.5170863827580245,
      "grad_norm": 0.7117284536361694,
      "learning_rate": 4.966404323751895e-06,
      "loss": 0.6981,
      "step": 17946
    },
    {
      "epoch": 0.5171151962196738,
      "grad_norm": 0.7186157703399658,
      "learning_rate": 4.9659377243063636e-06,
      "loss": 0.7112,
      "step": 17947
    },
    {
      "epoch": 0.5171440096813231,
      "grad_norm": 0.7251515984535217,
      "learning_rate": 4.96547112515748e-06,
      "loss": 0.7082,
      "step": 17948
    },
    {
      "epoch": 0.5171728231429724,
      "grad_norm": 0.7132154107093811,
      "learning_rate": 4.965004526309309e-06,
      "loss": 0.7333,
      "step": 17949
    },
    {
      "epoch": 0.5172016366046217,
      "grad_norm": 0.7091506719589233,
      "learning_rate": 4.964537927765913e-06,
      "loss": 0.7018,
      "step": 17950
    },
    {
      "epoch": 0.517230450066271,
      "grad_norm": 0.7446768283843994,
      "learning_rate": 4.964071329531356e-06,
      "loss": 0.7369,
      "step": 17951
    },
    {
      "epoch": 0.5172592635279203,
      "grad_norm": 0.7140297293663025,
      "learning_rate": 4.9636047316097e-06,
      "loss": 0.7229,
      "step": 17952
    },
    {
      "epoch": 0.5172880769895696,
      "grad_norm": 0.8379387855529785,
      "learning_rate": 4.963138134005013e-06,
      "loss": 0.7209,
      "step": 17953
    },
    {
      "epoch": 0.5173168904512188,
      "grad_norm": 0.9435129165649414,
      "learning_rate": 4.962671536721355e-06,
      "loss": 0.7163,
      "step": 17954
    },
    {
      "epoch": 0.517345703912868,
      "grad_norm": 0.7351441383361816,
      "learning_rate": 4.962204939762791e-06,
      "loss": 0.7063,
      "step": 17955
    },
    {
      "epoch": 0.5173745173745173,
      "grad_norm": 0.7230545878410339,
      "learning_rate": 4.961738343133384e-06,
      "loss": 0.7241,
      "step": 17956
    },
    {
      "epoch": 0.5174033308361666,
      "grad_norm": 0.728309690952301,
      "learning_rate": 4.961271746837198e-06,
      "loss": 0.6897,
      "step": 17957
    },
    {
      "epoch": 0.5174321442978159,
      "grad_norm": 2.1505374908447266,
      "learning_rate": 4.960805150878297e-06,
      "loss": 0.7138,
      "step": 17958
    },
    {
      "epoch": 0.5174609577594652,
      "grad_norm": 0.7278718948364258,
      "learning_rate": 4.960338555260744e-06,
      "loss": 0.7167,
      "step": 17959
    },
    {
      "epoch": 0.5174897712211145,
      "grad_norm": 0.7146894931793213,
      "learning_rate": 4.959871959988601e-06,
      "loss": 0.7003,
      "step": 17960
    },
    {
      "epoch": 0.5175185846827638,
      "grad_norm": 0.7350355982780457,
      "learning_rate": 4.959405365065933e-06,
      "loss": 0.7197,
      "step": 17961
    },
    {
      "epoch": 0.5175473981444131,
      "grad_norm": 0.7428503036499023,
      "learning_rate": 4.958938770496806e-06,
      "loss": 0.7114,
      "step": 17962
    },
    {
      "epoch": 0.5175762116060624,
      "grad_norm": 0.7264119982719421,
      "learning_rate": 4.95847217628528e-06,
      "loss": 0.6864,
      "step": 17963
    },
    {
      "epoch": 0.5176050250677117,
      "grad_norm": 0.7338805198669434,
      "learning_rate": 4.958005582435418e-06,
      "loss": 0.7376,
      "step": 17964
    },
    {
      "epoch": 0.517633838529361,
      "grad_norm": 0.7390952706336975,
      "learning_rate": 4.957538988951287e-06,
      "loss": 0.7119,
      "step": 17965
    },
    {
      "epoch": 0.5176626519910102,
      "grad_norm": 0.7069146633148193,
      "learning_rate": 4.9570723958369485e-06,
      "loss": 0.72,
      "step": 17966
    },
    {
      "epoch": 0.5176914654526594,
      "grad_norm": 0.7090435028076172,
      "learning_rate": 4.956605803096468e-06,
      "loss": 0.7306,
      "step": 17967
    },
    {
      "epoch": 0.5177202789143087,
      "grad_norm": 0.7166078686714172,
      "learning_rate": 4.956139210733906e-06,
      "loss": 0.7119,
      "step": 17968
    },
    {
      "epoch": 0.517749092375958,
      "grad_norm": 0.7236306071281433,
      "learning_rate": 4.9556726187533285e-06,
      "loss": 0.7198,
      "step": 17969
    },
    {
      "epoch": 0.5177779058376073,
      "grad_norm": 0.6868405342102051,
      "learning_rate": 4.955206027158798e-06,
      "loss": 0.6787,
      "step": 17970
    },
    {
      "epoch": 0.5178067192992566,
      "grad_norm": 0.7548145055770874,
      "learning_rate": 4.9547394359543785e-06,
      "loss": 0.7343,
      "step": 17971
    },
    {
      "epoch": 0.5178355327609059,
      "grad_norm": 0.7643333673477173,
      "learning_rate": 4.9542728451441335e-06,
      "loss": 0.7269,
      "step": 17972
    },
    {
      "epoch": 0.5178643462225552,
      "grad_norm": 0.7270323634147644,
      "learning_rate": 4.953806254732126e-06,
      "loss": 0.7214,
      "step": 17973
    },
    {
      "epoch": 0.5178931596842045,
      "grad_norm": 0.7665441036224365,
      "learning_rate": 4.9533396647224195e-06,
      "loss": 0.719,
      "step": 17974
    },
    {
      "epoch": 0.5179219731458538,
      "grad_norm": 0.746671736240387,
      "learning_rate": 4.952873075119079e-06,
      "loss": 0.7131,
      "step": 17975
    },
    {
      "epoch": 0.517950786607503,
      "grad_norm": 0.7411620020866394,
      "learning_rate": 4.9524064859261635e-06,
      "loss": 0.7203,
      "step": 17976
    },
    {
      "epoch": 0.5179796000691523,
      "grad_norm": 0.7535408735275269,
      "learning_rate": 4.951939897147742e-06,
      "loss": 0.7123,
      "step": 17977
    },
    {
      "epoch": 0.5180084135308016,
      "grad_norm": 0.7050634026527405,
      "learning_rate": 4.951473308787876e-06,
      "loss": 0.7143,
      "step": 17978
    },
    {
      "epoch": 0.5180372269924509,
      "grad_norm": 0.7197502255439758,
      "learning_rate": 4.95100672085063e-06,
      "loss": 0.7138,
      "step": 17979
    },
    {
      "epoch": 0.5180660404541002,
      "grad_norm": 0.7239862084388733,
      "learning_rate": 4.950540133340066e-06,
      "loss": 0.7187,
      "step": 17980
    },
    {
      "epoch": 0.5180948539157494,
      "grad_norm": 0.7285915613174438,
      "learning_rate": 4.950073546260248e-06,
      "loss": 0.716,
      "step": 17981
    },
    {
      "epoch": 0.5181236673773987,
      "grad_norm": 0.7347208857536316,
      "learning_rate": 4.949606959615238e-06,
      "loss": 0.713,
      "step": 17982
    },
    {
      "epoch": 0.518152480839048,
      "grad_norm": 0.7199740409851074,
      "learning_rate": 4.949140373409103e-06,
      "loss": 0.7034,
      "step": 17983
    },
    {
      "epoch": 0.5181812943006973,
      "grad_norm": 0.7286975383758545,
      "learning_rate": 4.948673787645904e-06,
      "loss": 0.7113,
      "step": 17984
    },
    {
      "epoch": 0.5182101077623466,
      "grad_norm": 0.7472764253616333,
      "learning_rate": 4.948207202329704e-06,
      "loss": 0.6848,
      "step": 17985
    },
    {
      "epoch": 0.5182389212239958,
      "grad_norm": 0.7156224250793457,
      "learning_rate": 4.947740617464568e-06,
      "loss": 0.7205,
      "step": 17986
    },
    {
      "epoch": 0.5182677346856451,
      "grad_norm": 0.7307929396629333,
      "learning_rate": 4.94727403305456e-06,
      "loss": 0.7217,
      "step": 17987
    },
    {
      "epoch": 0.5182965481472944,
      "grad_norm": 0.7169935703277588,
      "learning_rate": 4.9468074491037395e-06,
      "loss": 0.715,
      "step": 17988
    },
    {
      "epoch": 0.5183253616089437,
      "grad_norm": 0.7408882975578308,
      "learning_rate": 4.946340865616176e-06,
      "loss": 0.709,
      "step": 17989
    },
    {
      "epoch": 0.518354175070593,
      "grad_norm": 0.7113539576530457,
      "learning_rate": 4.945874282595929e-06,
      "loss": 0.6983,
      "step": 17990
    },
    {
      "epoch": 0.5183829885322423,
      "grad_norm": 0.7389634251594543,
      "learning_rate": 4.9454077000470625e-06,
      "loss": 0.7231,
      "step": 17991
    },
    {
      "epoch": 0.5184118019938916,
      "grad_norm": 0.7202456593513489,
      "learning_rate": 4.944941117973641e-06,
      "loss": 0.6926,
      "step": 17992
    },
    {
      "epoch": 0.5184406154555409,
      "grad_norm": 0.7373042702674866,
      "learning_rate": 4.944474536379728e-06,
      "loss": 0.7503,
      "step": 17993
    },
    {
      "epoch": 0.5184694289171901,
      "grad_norm": 0.7358412146568298,
      "learning_rate": 4.944007955269385e-06,
      "loss": 0.7296,
      "step": 17994
    },
    {
      "epoch": 0.5184982423788393,
      "grad_norm": 0.7166950702667236,
      "learning_rate": 4.943541374646678e-06,
      "loss": 0.7187,
      "step": 17995
    },
    {
      "epoch": 0.5185270558404886,
      "grad_norm": 0.7159591317176819,
      "learning_rate": 4.9430747945156686e-06,
      "loss": 0.7176,
      "step": 17996
    },
    {
      "epoch": 0.5185558693021379,
      "grad_norm": 0.729787290096283,
      "learning_rate": 4.942608214880421e-06,
      "loss": 0.7341,
      "step": 17997
    },
    {
      "epoch": 0.5185846827637872,
      "grad_norm": 0.7283382415771484,
      "learning_rate": 4.942141635744998e-06,
      "loss": 0.7186,
      "step": 17998
    },
    {
      "epoch": 0.5186134962254365,
      "grad_norm": 0.7389587163925171,
      "learning_rate": 4.941675057113463e-06,
      "loss": 0.7358,
      "step": 17999
    },
    {
      "epoch": 0.5186423096870858,
      "grad_norm": 0.7229470014572144,
      "learning_rate": 4.941208478989879e-06,
      "loss": 0.7224,
      "step": 18000
    },
    {
      "epoch": 0.5186711231487351,
      "grad_norm": 0.7009609341621399,
      "learning_rate": 4.940741901378312e-06,
      "loss": 0.6953,
      "step": 18001
    },
    {
      "epoch": 0.5186999366103844,
      "grad_norm": 0.7212063670158386,
      "learning_rate": 4.940275324282824e-06,
      "loss": 0.7015,
      "step": 18002
    },
    {
      "epoch": 0.5187287500720337,
      "grad_norm": 0.6960429549217224,
      "learning_rate": 4.939808747707478e-06,
      "loss": 0.6907,
      "step": 18003
    },
    {
      "epoch": 0.518757563533683,
      "grad_norm": 0.7319434285163879,
      "learning_rate": 4.939342171656336e-06,
      "loss": 0.6975,
      "step": 18004
    },
    {
      "epoch": 0.5187863769953323,
      "grad_norm": 0.7497164607048035,
      "learning_rate": 4.9388755961334655e-06,
      "loss": 0.7229,
      "step": 18005
    },
    {
      "epoch": 0.5188151904569815,
      "grad_norm": 0.7270057201385498,
      "learning_rate": 4.938409021142927e-06,
      "loss": 0.6982,
      "step": 18006
    },
    {
      "epoch": 0.5188440039186308,
      "grad_norm": 0.7216282486915588,
      "learning_rate": 4.937942446688784e-06,
      "loss": 0.7242,
      "step": 18007
    },
    {
      "epoch": 0.51887281738028,
      "grad_norm": 0.7290652990341187,
      "learning_rate": 4.937475872775101e-06,
      "loss": 0.7162,
      "step": 18008
    },
    {
      "epoch": 0.5189016308419293,
      "grad_norm": 0.7333051562309265,
      "learning_rate": 4.937009299405939e-06,
      "loss": 0.7487,
      "step": 18009
    },
    {
      "epoch": 0.5189304443035786,
      "grad_norm": 0.7187821269035339,
      "learning_rate": 4.936542726585364e-06,
      "loss": 0.6907,
      "step": 18010
    },
    {
      "epoch": 0.5189592577652279,
      "grad_norm": 0.7263944745063782,
      "learning_rate": 4.936076154317439e-06,
      "loss": 0.7228,
      "step": 18011
    },
    {
      "epoch": 0.5189880712268772,
      "grad_norm": 0.768118679523468,
      "learning_rate": 4.935609582606224e-06,
      "loss": 0.7335,
      "step": 18012
    },
    {
      "epoch": 0.5190168846885265,
      "grad_norm": 0.7362088561058044,
      "learning_rate": 4.935143011455787e-06,
      "loss": 0.7337,
      "step": 18013
    },
    {
      "epoch": 0.5190456981501758,
      "grad_norm": 0.7041334509849548,
      "learning_rate": 4.9346764408701904e-06,
      "loss": 0.7303,
      "step": 18014
    },
    {
      "epoch": 0.519074511611825,
      "grad_norm": 0.6977120637893677,
      "learning_rate": 4.934209870853496e-06,
      "loss": 0.7012,
      "step": 18015
    },
    {
      "epoch": 0.5191033250734743,
      "grad_norm": 0.729032576084137,
      "learning_rate": 4.9337433014097684e-06,
      "loss": 0.7128,
      "step": 18016
    },
    {
      "epoch": 0.5191321385351236,
      "grad_norm": 0.724969208240509,
      "learning_rate": 4.933276732543071e-06,
      "loss": 0.7047,
      "step": 18017
    },
    {
      "epoch": 0.5191609519967729,
      "grad_norm": 0.7214389443397522,
      "learning_rate": 4.932810164257465e-06,
      "loss": 0.7274,
      "step": 18018
    },
    {
      "epoch": 0.5191897654584222,
      "grad_norm": 0.6917661428451538,
      "learning_rate": 4.932343596557016e-06,
      "loss": 0.7129,
      "step": 18019
    },
    {
      "epoch": 0.5192185789200715,
      "grad_norm": 0.7539107799530029,
      "learning_rate": 4.9318770294457856e-06,
      "loss": 0.7105,
      "step": 18020
    },
    {
      "epoch": 0.5192473923817207,
      "grad_norm": 0.7450907826423645,
      "learning_rate": 4.931410462927839e-06,
      "loss": 0.7267,
      "step": 18021
    },
    {
      "epoch": 0.51927620584337,
      "grad_norm": 0.7235465049743652,
      "learning_rate": 4.930943897007239e-06,
      "loss": 0.7077,
      "step": 18022
    },
    {
      "epoch": 0.5193050193050193,
      "grad_norm": 0.7230415940284729,
      "learning_rate": 4.930477331688048e-06,
      "loss": 0.7288,
      "step": 18023
    },
    {
      "epoch": 0.5193338327666686,
      "grad_norm": 0.7142083644866943,
      "learning_rate": 4.93001076697433e-06,
      "loss": 0.6918,
      "step": 18024
    },
    {
      "epoch": 0.5193626462283178,
      "grad_norm": 0.740369975566864,
      "learning_rate": 4.929544202870148e-06,
      "loss": 0.7053,
      "step": 18025
    },
    {
      "epoch": 0.5193914596899671,
      "grad_norm": 0.723281979560852,
      "learning_rate": 4.9290776393795655e-06,
      "loss": 0.7135,
      "step": 18026
    },
    {
      "epoch": 0.5194202731516164,
      "grad_norm": 0.7367302179336548,
      "learning_rate": 4.928611076506647e-06,
      "loss": 0.7083,
      "step": 18027
    },
    {
      "epoch": 0.5194490866132657,
      "grad_norm": 0.7262564301490784,
      "learning_rate": 4.928144514255454e-06,
      "loss": 0.7027,
      "step": 18028
    },
    {
      "epoch": 0.519477900074915,
      "grad_norm": 0.7324110269546509,
      "learning_rate": 4.927677952630051e-06,
      "loss": 0.7272,
      "step": 18029
    },
    {
      "epoch": 0.5195067135365643,
      "grad_norm": 0.7079781889915466,
      "learning_rate": 4.9272113916345e-06,
      "loss": 0.6981,
      "step": 18030
    },
    {
      "epoch": 0.5195355269982136,
      "grad_norm": 0.7394258975982666,
      "learning_rate": 4.926744831272866e-06,
      "loss": 0.7164,
      "step": 18031
    },
    {
      "epoch": 0.5195643404598629,
      "grad_norm": 0.7357417941093445,
      "learning_rate": 4.9262782715492105e-06,
      "loss": 0.7244,
      "step": 18032
    },
    {
      "epoch": 0.5195931539215122,
      "grad_norm": 0.7268593907356262,
      "learning_rate": 4.925811712467598e-06,
      "loss": 0.7239,
      "step": 18033
    },
    {
      "epoch": 0.5196219673831615,
      "grad_norm": 0.6971402764320374,
      "learning_rate": 4.925345154032092e-06,
      "loss": 0.6964,
      "step": 18034
    },
    {
      "epoch": 0.5196507808448106,
      "grad_norm": 0.7069408893585205,
      "learning_rate": 4.924878596246754e-06,
      "loss": 0.7005,
      "step": 18035
    },
    {
      "epoch": 0.5196795943064599,
      "grad_norm": 0.7406966686248779,
      "learning_rate": 4.924412039115646e-06,
      "loss": 0.7329,
      "step": 18036
    },
    {
      "epoch": 0.5197084077681092,
      "grad_norm": 0.6913089156150818,
      "learning_rate": 4.923945482642836e-06,
      "loss": 0.7157,
      "step": 18037
    },
    {
      "epoch": 0.5197372212297585,
      "grad_norm": 0.7196069955825806,
      "learning_rate": 4.923478926832386e-06,
      "loss": 0.6919,
      "step": 18038
    },
    {
      "epoch": 0.5197660346914078,
      "grad_norm": 0.6907258629798889,
      "learning_rate": 4.923012371688357e-06,
      "loss": 0.6931,
      "step": 18039
    },
    {
      "epoch": 0.5197948481530571,
      "grad_norm": 0.7260830402374268,
      "learning_rate": 4.922545817214813e-06,
      "loss": 0.7048,
      "step": 18040
    },
    {
      "epoch": 0.5198236616147064,
      "grad_norm": 0.7452844977378845,
      "learning_rate": 4.9220792634158175e-06,
      "loss": 0.7404,
      "step": 18041
    },
    {
      "epoch": 0.5198524750763557,
      "grad_norm": 0.703240156173706,
      "learning_rate": 4.921612710295433e-06,
      "loss": 0.6953,
      "step": 18042
    },
    {
      "epoch": 0.519881288538005,
      "grad_norm": 0.7197980880737305,
      "learning_rate": 4.921146157857725e-06,
      "loss": 0.7148,
      "step": 18043
    },
    {
      "epoch": 0.5199101019996543,
      "grad_norm": 0.7176377177238464,
      "learning_rate": 4.920679606106755e-06,
      "loss": 0.7061,
      "step": 18044
    },
    {
      "epoch": 0.5199389154613036,
      "grad_norm": 0.742586076259613,
      "learning_rate": 4.9202130550465855e-06,
      "loss": 0.7221,
      "step": 18045
    },
    {
      "epoch": 0.5199677289229528,
      "grad_norm": 0.7285906672477722,
      "learning_rate": 4.919746504681281e-06,
      "loss": 0.7182,
      "step": 18046
    },
    {
      "epoch": 0.5199965423846021,
      "grad_norm": 0.724545419216156,
      "learning_rate": 4.919279955014904e-06,
      "loss": 0.7195,
      "step": 18047
    },
    {
      "epoch": 0.5200253558462513,
      "grad_norm": 0.7136362195014954,
      "learning_rate": 4.9188134060515165e-06,
      "loss": 0.7239,
      "step": 18048
    },
    {
      "epoch": 0.5200541693079006,
      "grad_norm": 0.6940143704414368,
      "learning_rate": 4.918346857795184e-06,
      "loss": 0.7045,
      "step": 18049
    },
    {
      "epoch": 0.5200829827695499,
      "grad_norm": 0.7406476140022278,
      "learning_rate": 4.91788031024997e-06,
      "loss": 0.7244,
      "step": 18050
    },
    {
      "epoch": 0.5201117962311992,
      "grad_norm": 1.3312777280807495,
      "learning_rate": 4.917413763419936e-06,
      "loss": 0.7251,
      "step": 18051
    },
    {
      "epoch": 0.5201406096928485,
      "grad_norm": 0.7096763253211975,
      "learning_rate": 4.916947217309145e-06,
      "loss": 0.6958,
      "step": 18052
    },
    {
      "epoch": 0.5201694231544978,
      "grad_norm": 0.7137464284896851,
      "learning_rate": 4.916480671921662e-06,
      "loss": 0.7017,
      "step": 18053
    },
    {
      "epoch": 0.5201982366161471,
      "grad_norm": 0.7559944987297058,
      "learning_rate": 4.916014127261548e-06,
      "loss": 0.7321,
      "step": 18054
    },
    {
      "epoch": 0.5202270500777963,
      "grad_norm": 0.7762124538421631,
      "learning_rate": 4.915547583332866e-06,
      "loss": 0.7391,
      "step": 18055
    },
    {
      "epoch": 0.5202558635394456,
      "grad_norm": 0.7319069504737854,
      "learning_rate": 4.915081040139681e-06,
      "loss": 0.708,
      "step": 18056
    },
    {
      "epoch": 0.5202846770010949,
      "grad_norm": 0.7106541991233826,
      "learning_rate": 4.914614497686056e-06,
      "loss": 0.6971,
      "step": 18057
    },
    {
      "epoch": 0.5203134904627442,
      "grad_norm": 0.7198246121406555,
      "learning_rate": 4.914147955976052e-06,
      "loss": 0.6951,
      "step": 18058
    },
    {
      "epoch": 0.5203423039243935,
      "grad_norm": 0.7555293440818787,
      "learning_rate": 4.913681415013734e-06,
      "loss": 0.6998,
      "step": 18059
    },
    {
      "epoch": 0.5203711173860428,
      "grad_norm": 0.7192845344543457,
      "learning_rate": 4.913214874803164e-06,
      "loss": 0.7056,
      "step": 18060
    },
    {
      "epoch": 0.5203999308476921,
      "grad_norm": 0.7148814797401428,
      "learning_rate": 4.912748335348406e-06,
      "loss": 0.6986,
      "step": 18061
    },
    {
      "epoch": 0.5204287443093413,
      "grad_norm": 0.7247413396835327,
      "learning_rate": 4.912281796653523e-06,
      "loss": 0.7027,
      "step": 18062
    },
    {
      "epoch": 0.5204575577709906,
      "grad_norm": 0.7045612335205078,
      "learning_rate": 4.911815258722578e-06,
      "loss": 0.7309,
      "step": 18063
    },
    {
      "epoch": 0.5204863712326399,
      "grad_norm": 0.7344616055488586,
      "learning_rate": 4.911348721559634e-06,
      "loss": 0.7326,
      "step": 18064
    },
    {
      "epoch": 0.5205151846942891,
      "grad_norm": 0.7379952073097229,
      "learning_rate": 4.9108821851687535e-06,
      "loss": 0.7002,
      "step": 18065
    },
    {
      "epoch": 0.5205439981559384,
      "grad_norm": 0.7444601655006409,
      "learning_rate": 4.910415649554001e-06,
      "loss": 0.7171,
      "step": 18066
    },
    {
      "epoch": 0.5205728116175877,
      "grad_norm": 0.7976950407028198,
      "learning_rate": 4.909949114719439e-06,
      "loss": 0.6977,
      "step": 18067
    },
    {
      "epoch": 0.520601625079237,
      "grad_norm": 0.7311476469039917,
      "learning_rate": 4.909482580669129e-06,
      "loss": 0.6972,
      "step": 18068
    },
    {
      "epoch": 0.5206304385408863,
      "grad_norm": 0.8570036292076111,
      "learning_rate": 4.909016047407136e-06,
      "loss": 0.7083,
      "step": 18069
    },
    {
      "epoch": 0.5206592520025356,
      "grad_norm": 0.722942054271698,
      "learning_rate": 4.908549514937522e-06,
      "loss": 0.7083,
      "step": 18070
    },
    {
      "epoch": 0.5206880654641849,
      "grad_norm": 0.7523815631866455,
      "learning_rate": 4.908082983264351e-06,
      "loss": 0.7323,
      "step": 18071
    },
    {
      "epoch": 0.5207168789258342,
      "grad_norm": 0.7323500514030457,
      "learning_rate": 4.907616452391682e-06,
      "loss": 0.7251,
      "step": 18072
    },
    {
      "epoch": 0.5207456923874835,
      "grad_norm": 0.7278427481651306,
      "learning_rate": 4.907149922323584e-06,
      "loss": 0.6939,
      "step": 18073
    },
    {
      "epoch": 0.5207745058491328,
      "grad_norm": 0.7317441701889038,
      "learning_rate": 4.906683393064117e-06,
      "loss": 0.7178,
      "step": 18074
    },
    {
      "epoch": 0.5208033193107819,
      "grad_norm": 0.7414091229438782,
      "learning_rate": 4.9062168646173455e-06,
      "loss": 0.7078,
      "step": 18075
    },
    {
      "epoch": 0.5208321327724312,
      "grad_norm": 0.7097771167755127,
      "learning_rate": 4.90575033698733e-06,
      "loss": 0.7043,
      "step": 18076
    },
    {
      "epoch": 0.5208609462340805,
      "grad_norm": 0.7165612578392029,
      "learning_rate": 4.905283810178136e-06,
      "loss": 0.7139,
      "step": 18077
    },
    {
      "epoch": 0.5208897596957298,
      "grad_norm": 0.729905903339386,
      "learning_rate": 4.9048172841938245e-06,
      "loss": 0.7101,
      "step": 18078
    },
    {
      "epoch": 0.5209185731573791,
      "grad_norm": 0.7455853819847107,
      "learning_rate": 4.90435075903846e-06,
      "loss": 0.7106,
      "step": 18079
    },
    {
      "epoch": 0.5209473866190284,
      "grad_norm": 0.7630501985549927,
      "learning_rate": 4.903884234716103e-06,
      "loss": 0.7357,
      "step": 18080
    },
    {
      "epoch": 0.5209762000806777,
      "grad_norm": 0.7167354822158813,
      "learning_rate": 4.90341771123082e-06,
      "loss": 0.6963,
      "step": 18081
    },
    {
      "epoch": 0.521005013542327,
      "grad_norm": 0.7097148299217224,
      "learning_rate": 4.902951188586672e-06,
      "loss": 0.7235,
      "step": 18082
    },
    {
      "epoch": 0.5210338270039763,
      "grad_norm": 0.7492078542709351,
      "learning_rate": 4.902484666787721e-06,
      "loss": 0.7139,
      "step": 18083
    },
    {
      "epoch": 0.5210626404656256,
      "grad_norm": 0.7513207197189331,
      "learning_rate": 4.902018145838031e-06,
      "loss": 0.7303,
      "step": 18084
    },
    {
      "epoch": 0.5210914539272749,
      "grad_norm": 0.7677907943725586,
      "learning_rate": 4.901551625741665e-06,
      "loss": 0.6915,
      "step": 18085
    },
    {
      "epoch": 0.5211202673889241,
      "grad_norm": 0.7303071618080139,
      "learning_rate": 4.901085106502687e-06,
      "loss": 0.6973,
      "step": 18086
    },
    {
      "epoch": 0.5211490808505734,
      "grad_norm": 0.7329155802726746,
      "learning_rate": 4.9006185881251595e-06,
      "loss": 0.7239,
      "step": 18087
    },
    {
      "epoch": 0.5211778943122227,
      "grad_norm": 0.7409541606903076,
      "learning_rate": 4.9001520706131435e-06,
      "loss": 0.7103,
      "step": 18088
    },
    {
      "epoch": 0.5212067077738719,
      "grad_norm": 0.7102764248847961,
      "learning_rate": 4.899685553970704e-06,
      "loss": 0.6844,
      "step": 18089
    },
    {
      "epoch": 0.5212355212355212,
      "grad_norm": 0.7560367584228516,
      "learning_rate": 4.8992190382019024e-06,
      "loss": 0.717,
      "step": 18090
    },
    {
      "epoch": 0.5212643346971705,
      "grad_norm": 0.7229607701301575,
      "learning_rate": 4.898752523310803e-06,
      "loss": 0.7248,
      "step": 18091
    },
    {
      "epoch": 0.5212931481588198,
      "grad_norm": 0.7368506789207458,
      "learning_rate": 4.898286009301467e-06,
      "loss": 0.7001,
      "step": 18092
    },
    {
      "epoch": 0.5213219616204691,
      "grad_norm": 0.7225121259689331,
      "learning_rate": 4.8978194961779594e-06,
      "loss": 0.715,
      "step": 18093
    },
    {
      "epoch": 0.5213507750821184,
      "grad_norm": 0.7210506200790405,
      "learning_rate": 4.897352983944341e-06,
      "loss": 0.6949,
      "step": 18094
    },
    {
      "epoch": 0.5213795885437676,
      "grad_norm": 0.6842076182365417,
      "learning_rate": 4.896886472604675e-06,
      "loss": 0.6789,
      "step": 18095
    },
    {
      "epoch": 0.5214084020054169,
      "grad_norm": 0.7123616933822632,
      "learning_rate": 4.896419962163024e-06,
      "loss": 0.6916,
      "step": 18096
    },
    {
      "epoch": 0.5214372154670662,
      "grad_norm": 0.7733849883079529,
      "learning_rate": 4.895953452623452e-06,
      "loss": 0.7494,
      "step": 18097
    },
    {
      "epoch": 0.5214660289287155,
      "grad_norm": 0.7103502750396729,
      "learning_rate": 4.895486943990023e-06,
      "loss": 0.7147,
      "step": 18098
    },
    {
      "epoch": 0.5214948423903648,
      "grad_norm": 0.7241153120994568,
      "learning_rate": 4.895020436266797e-06,
      "loss": 0.7171,
      "step": 18099
    },
    {
      "epoch": 0.5215236558520141,
      "grad_norm": 0.7257764935493469,
      "learning_rate": 4.894553929457839e-06,
      "loss": 0.6995,
      "step": 18100
    },
    {
      "epoch": 0.5215524693136634,
      "grad_norm": 0.7314204573631287,
      "learning_rate": 4.89408742356721e-06,
      "loss": 0.7189,
      "step": 18101
    },
    {
      "epoch": 0.5215812827753126,
      "grad_norm": 0.7398796677589417,
      "learning_rate": 4.893620918598974e-06,
      "loss": 0.7276,
      "step": 18102
    },
    {
      "epoch": 0.5216100962369619,
      "grad_norm": 0.759350061416626,
      "learning_rate": 4.893154414557194e-06,
      "loss": 0.7343,
      "step": 18103
    },
    {
      "epoch": 0.5216389096986112,
      "grad_norm": 0.7199488282203674,
      "learning_rate": 4.892687911445932e-06,
      "loss": 0.7235,
      "step": 18104
    },
    {
      "epoch": 0.5216677231602604,
      "grad_norm": 0.7144488096237183,
      "learning_rate": 4.89222140926925e-06,
      "loss": 0.6885,
      "step": 18105
    },
    {
      "epoch": 0.5216965366219097,
      "grad_norm": 0.6924922466278076,
      "learning_rate": 4.891754908031213e-06,
      "loss": 0.6983,
      "step": 18106
    },
    {
      "epoch": 0.521725350083559,
      "grad_norm": 0.708167552947998,
      "learning_rate": 4.891288407735883e-06,
      "loss": 0.7278,
      "step": 18107
    },
    {
      "epoch": 0.5217541635452083,
      "grad_norm": 0.7426791787147522,
      "learning_rate": 4.8908219083873195e-06,
      "loss": 0.7321,
      "step": 18108
    },
    {
      "epoch": 0.5217829770068576,
      "grad_norm": 0.7446613907814026,
      "learning_rate": 4.89035540998959e-06,
      "loss": 0.7265,
      "step": 18109
    },
    {
      "epoch": 0.5218117904685069,
      "grad_norm": 0.7548907995223999,
      "learning_rate": 4.8898889125467555e-06,
      "loss": 0.7061,
      "step": 18110
    },
    {
      "epoch": 0.5218406039301562,
      "grad_norm": 0.7490390539169312,
      "learning_rate": 4.889422416062878e-06,
      "loss": 0.7255,
      "step": 18111
    },
    {
      "epoch": 0.5218694173918055,
      "grad_norm": 0.7151121497154236,
      "learning_rate": 4.888955920542021e-06,
      "loss": 0.7006,
      "step": 18112
    },
    {
      "epoch": 0.5218982308534548,
      "grad_norm": 0.7593756914138794,
      "learning_rate": 4.888489425988247e-06,
      "loss": 0.7445,
      "step": 18113
    },
    {
      "epoch": 0.5219270443151041,
      "grad_norm": 0.7332315444946289,
      "learning_rate": 4.888022932405619e-06,
      "loss": 0.7083,
      "step": 18114
    },
    {
      "epoch": 0.5219558577767534,
      "grad_norm": 0.7252590656280518,
      "learning_rate": 4.887556439798199e-06,
      "loss": 0.7128,
      "step": 18115
    },
    {
      "epoch": 0.5219846712384025,
      "grad_norm": 0.7207651734352112,
      "learning_rate": 4.887089948170049e-06,
      "loss": 0.6966,
      "step": 18116
    },
    {
      "epoch": 0.5220134847000518,
      "grad_norm": 0.7183765172958374,
      "learning_rate": 4.886623457525235e-06,
      "loss": 0.7204,
      "step": 18117
    },
    {
      "epoch": 0.5220422981617011,
      "grad_norm": 0.7241623997688293,
      "learning_rate": 4.886156967867814e-06,
      "loss": 0.7128,
      "step": 18118
    },
    {
      "epoch": 0.5220711116233504,
      "grad_norm": 0.7350647449493408,
      "learning_rate": 4.885690479201854e-06,
      "loss": 0.7033,
      "step": 18119
    },
    {
      "epoch": 0.5220999250849997,
      "grad_norm": 0.6952499151229858,
      "learning_rate": 4.885223991531415e-06,
      "loss": 0.7186,
      "step": 18120
    },
    {
      "epoch": 0.522128738546649,
      "grad_norm": 0.7252627611160278,
      "learning_rate": 4.884757504860561e-06,
      "loss": 0.6883,
      "step": 18121
    },
    {
      "epoch": 0.5221575520082983,
      "grad_norm": 0.740608811378479,
      "learning_rate": 4.884291019193353e-06,
      "loss": 0.7078,
      "step": 18122
    },
    {
      "epoch": 0.5221863654699476,
      "grad_norm": 0.7315420508384705,
      "learning_rate": 4.883824534533855e-06,
      "loss": 0.7024,
      "step": 18123
    },
    {
      "epoch": 0.5222151789315969,
      "grad_norm": 0.7212874293327332,
      "learning_rate": 4.88335805088613e-06,
      "loss": 0.7168,
      "step": 18124
    },
    {
      "epoch": 0.5222439923932461,
      "grad_norm": 0.7426908016204834,
      "learning_rate": 4.88289156825424e-06,
      "loss": 0.693,
      "step": 18125
    },
    {
      "epoch": 0.5222728058548954,
      "grad_norm": 0.743731677532196,
      "learning_rate": 4.882425086642246e-06,
      "loss": 0.7245,
      "step": 18126
    },
    {
      "epoch": 0.5223016193165447,
      "grad_norm": 0.7204198241233826,
      "learning_rate": 4.881958606054213e-06,
      "loss": 0.7099,
      "step": 18127
    },
    {
      "epoch": 0.522330432778194,
      "grad_norm": 0.749304473400116,
      "learning_rate": 4.8814921264942024e-06,
      "loss": 0.7365,
      "step": 18128
    },
    {
      "epoch": 0.5223592462398432,
      "grad_norm": 0.7444519400596619,
      "learning_rate": 4.881025647966277e-06,
      "loss": 0.7076,
      "step": 18129
    },
    {
      "epoch": 0.5223880597014925,
      "grad_norm": 0.7563082575798035,
      "learning_rate": 4.8805591704745e-06,
      "loss": 0.7418,
      "step": 18130
    },
    {
      "epoch": 0.5224168731631418,
      "grad_norm": 0.7236095070838928,
      "learning_rate": 4.880092694022932e-06,
      "loss": 0.7426,
      "step": 18131
    },
    {
      "epoch": 0.5224456866247911,
      "grad_norm": 0.7277436852455139,
      "learning_rate": 4.879626218615634e-06,
      "loss": 0.6929,
      "step": 18132
    },
    {
      "epoch": 0.5224745000864404,
      "grad_norm": 0.7276706695556641,
      "learning_rate": 4.879159744256675e-06,
      "loss": 0.7282,
      "step": 18133
    },
    {
      "epoch": 0.5225033135480897,
      "grad_norm": 0.7178232073783875,
      "learning_rate": 4.8786932709501125e-06,
      "loss": 0.7261,
      "step": 18134
    },
    {
      "epoch": 0.522532127009739,
      "grad_norm": 0.7338558435440063,
      "learning_rate": 4.878226798700011e-06,
      "loss": 0.7178,
      "step": 18135
    },
    {
      "epoch": 0.5225609404713882,
      "grad_norm": 0.7225983738899231,
      "learning_rate": 4.877760327510433e-06,
      "loss": 0.7174,
      "step": 18136
    },
    {
      "epoch": 0.5225897539330375,
      "grad_norm": 0.7429869771003723,
      "learning_rate": 4.87729385738544e-06,
      "loss": 0.7149,
      "step": 18137
    },
    {
      "epoch": 0.5226185673946868,
      "grad_norm": 0.7405964732170105,
      "learning_rate": 4.876827388329095e-06,
      "loss": 0.714,
      "step": 18138
    },
    {
      "epoch": 0.5226473808563361,
      "grad_norm": 0.7635725736618042,
      "learning_rate": 4.876360920345458e-06,
      "loss": 0.7014,
      "step": 18139
    },
    {
      "epoch": 0.5226761943179854,
      "grad_norm": 0.7465718388557434,
      "learning_rate": 4.875894453438596e-06,
      "loss": 0.6983,
      "step": 18140
    },
    {
      "epoch": 0.5227050077796347,
      "grad_norm": 0.7490624785423279,
      "learning_rate": 4.875427987612569e-06,
      "loss": 0.7415,
      "step": 18141
    },
    {
      "epoch": 0.522733821241284,
      "grad_norm": 0.7412382364273071,
      "learning_rate": 4.87496152287144e-06,
      "loss": 0.6996,
      "step": 18142
    },
    {
      "epoch": 0.5227626347029332,
      "grad_norm": 0.7493959069252014,
      "learning_rate": 4.874495059219269e-06,
      "loss": 0.6884,
      "step": 18143
    },
    {
      "epoch": 0.5227914481645825,
      "grad_norm": 0.7538542151451111,
      "learning_rate": 4.874028596660122e-06,
      "loss": 0.7242,
      "step": 18144
    },
    {
      "epoch": 0.5228202616262317,
      "grad_norm": 0.7457607984542847,
      "learning_rate": 4.87356213519806e-06,
      "loss": 0.7262,
      "step": 18145
    },
    {
      "epoch": 0.522849075087881,
      "grad_norm": 0.7333888411521912,
      "learning_rate": 4.873095674837146e-06,
      "loss": 0.7296,
      "step": 18146
    },
    {
      "epoch": 0.5228778885495303,
      "grad_norm": 0.7356085777282715,
      "learning_rate": 4.872629215581441e-06,
      "loss": 0.6969,
      "step": 18147
    },
    {
      "epoch": 0.5229067020111796,
      "grad_norm": 0.7526573538780212,
      "learning_rate": 4.872162757435009e-06,
      "loss": 0.7203,
      "step": 18148
    },
    {
      "epoch": 0.5229355154728289,
      "grad_norm": 0.7327576279640198,
      "learning_rate": 4.871696300401912e-06,
      "loss": 0.71,
      "step": 18149
    },
    {
      "epoch": 0.5229643289344782,
      "grad_norm": 0.7136730551719666,
      "learning_rate": 4.871229844486211e-06,
      "loss": 0.703,
      "step": 18150
    },
    {
      "epoch": 0.5229931423961275,
      "grad_norm": 0.753972589969635,
      "learning_rate": 4.87076338969197e-06,
      "loss": 0.7071,
      "step": 18151
    },
    {
      "epoch": 0.5230219558577768,
      "grad_norm": 0.7539016604423523,
      "learning_rate": 4.870296936023251e-06,
      "loss": 0.7064,
      "step": 18152
    },
    {
      "epoch": 0.5230507693194261,
      "grad_norm": 0.7239906787872314,
      "learning_rate": 4.869830483484115e-06,
      "loss": 0.7455,
      "step": 18153
    },
    {
      "epoch": 0.5230795827810754,
      "grad_norm": 0.7425199747085571,
      "learning_rate": 4.869364032078625e-06,
      "loss": 0.711,
      "step": 18154
    },
    {
      "epoch": 0.5231083962427246,
      "grad_norm": 0.728961169719696,
      "learning_rate": 4.868897581810843e-06,
      "loss": 0.7108,
      "step": 18155
    },
    {
      "epoch": 0.5231372097043738,
      "grad_norm": 0.7212636470794678,
      "learning_rate": 4.8684311326848335e-06,
      "loss": 0.7366,
      "step": 18156
    },
    {
      "epoch": 0.5231660231660231,
      "grad_norm": 0.7321879267692566,
      "learning_rate": 4.867964684704658e-06,
      "loss": 0.7081,
      "step": 18157
    },
    {
      "epoch": 0.5231948366276724,
      "grad_norm": 0.7072867751121521,
      "learning_rate": 4.867498237874377e-06,
      "loss": 0.7159,
      "step": 18158
    },
    {
      "epoch": 0.5232236500893217,
      "grad_norm": 0.7354352474212646,
      "learning_rate": 4.867031792198054e-06,
      "loss": 0.7046,
      "step": 18159
    },
    {
      "epoch": 0.523252463550971,
      "grad_norm": 0.7653051018714905,
      "learning_rate": 4.866565347679751e-06,
      "loss": 0.7148,
      "step": 18160
    },
    {
      "epoch": 0.5232812770126203,
      "grad_norm": 0.7516418099403381,
      "learning_rate": 4.86609890432353e-06,
      "loss": 0.7553,
      "step": 18161
    },
    {
      "epoch": 0.5233100904742696,
      "grad_norm": 0.7343370318412781,
      "learning_rate": 4.865632462133456e-06,
      "loss": 0.7216,
      "step": 18162
    },
    {
      "epoch": 0.5233389039359189,
      "grad_norm": 0.7192417979240417,
      "learning_rate": 4.865166021113587e-06,
      "loss": 0.6927,
      "step": 18163
    },
    {
      "epoch": 0.5233677173975682,
      "grad_norm": 0.7462255358695984,
      "learning_rate": 4.864699581267988e-06,
      "loss": 0.6968,
      "step": 18164
    },
    {
      "epoch": 0.5233965308592174,
      "grad_norm": 0.736618161201477,
      "learning_rate": 4.86423314260072e-06,
      "loss": 0.7207,
      "step": 18165
    },
    {
      "epoch": 0.5234253443208667,
      "grad_norm": 0.7164289355278015,
      "learning_rate": 4.863766705115846e-06,
      "loss": 0.719,
      "step": 18166
    },
    {
      "epoch": 0.523454157782516,
      "grad_norm": 0.7176912426948547,
      "learning_rate": 4.863300268817424e-06,
      "loss": 0.7409,
      "step": 18167
    },
    {
      "epoch": 0.5234829712441653,
      "grad_norm": 0.711251974105835,
      "learning_rate": 4.862833833709524e-06,
      "loss": 0.7049,
      "step": 18168
    },
    {
      "epoch": 0.5235117847058146,
      "grad_norm": 0.7219633460044861,
      "learning_rate": 4.862367399796204e-06,
      "loss": 0.7211,
      "step": 18169
    },
    {
      "epoch": 0.5235405981674638,
      "grad_norm": 0.7324638366699219,
      "learning_rate": 4.861900967081527e-06,
      "loss": 0.7295,
      "step": 18170
    },
    {
      "epoch": 0.5235694116291131,
      "grad_norm": 0.7334416508674622,
      "learning_rate": 4.861434535569553e-06,
      "loss": 0.6963,
      "step": 18171
    },
    {
      "epoch": 0.5235982250907624,
      "grad_norm": 0.7014692425727844,
      "learning_rate": 4.860968105264347e-06,
      "loss": 0.7426,
      "step": 18172
    },
    {
      "epoch": 0.5236270385524117,
      "grad_norm": 0.7188059687614441,
      "learning_rate": 4.860501676169968e-06,
      "loss": 0.6981,
      "step": 18173
    },
    {
      "epoch": 0.523655852014061,
      "grad_norm": 0.683620035648346,
      "learning_rate": 4.8600352482904815e-06,
      "loss": 0.6989,
      "step": 18174
    },
    {
      "epoch": 0.5236846654757102,
      "grad_norm": 0.6957666277885437,
      "learning_rate": 4.859568821629948e-06,
      "loss": 0.7331,
      "step": 18175
    },
    {
      "epoch": 0.5237134789373595,
      "grad_norm": 0.7126971483230591,
      "learning_rate": 4.859102396192429e-06,
      "loss": 0.7069,
      "step": 18176
    },
    {
      "epoch": 0.5237422923990088,
      "grad_norm": 0.7354123592376709,
      "learning_rate": 4.858635971981987e-06,
      "loss": 0.7362,
      "step": 18177
    },
    {
      "epoch": 0.5237711058606581,
      "grad_norm": 0.7767757177352905,
      "learning_rate": 4.858169549002685e-06,
      "loss": 0.7294,
      "step": 18178
    },
    {
      "epoch": 0.5237999193223074,
      "grad_norm": 0.7392959594726562,
      "learning_rate": 4.857703127258583e-06,
      "loss": 0.7106,
      "step": 18179
    },
    {
      "epoch": 0.5238287327839567,
      "grad_norm": 0.7365143895149231,
      "learning_rate": 4.8572367067537465e-06,
      "loss": 0.7238,
      "step": 18180
    },
    {
      "epoch": 0.523857546245606,
      "grad_norm": 0.690548837184906,
      "learning_rate": 4.856770287492235e-06,
      "loss": 0.7012,
      "step": 18181
    },
    {
      "epoch": 0.5238863597072553,
      "grad_norm": 0.6898250579833984,
      "learning_rate": 4.85630386947811e-06,
      "loss": 0.7061,
      "step": 18182
    },
    {
      "epoch": 0.5239151731689045,
      "grad_norm": 0.7294155359268188,
      "learning_rate": 4.8558374527154354e-06,
      "loss": 0.7161,
      "step": 18183
    },
    {
      "epoch": 0.5239439866305537,
      "grad_norm": 0.7414243817329407,
      "learning_rate": 4.8553710372082735e-06,
      "loss": 0.727,
      "step": 18184
    },
    {
      "epoch": 0.523972800092203,
      "grad_norm": 0.744581937789917,
      "learning_rate": 4.854904622960684e-06,
      "loss": 0.7411,
      "step": 18185
    },
    {
      "epoch": 0.5240016135538523,
      "grad_norm": 0.7159954905509949,
      "learning_rate": 4.854438209976731e-06,
      "loss": 0.7188,
      "step": 18186
    },
    {
      "epoch": 0.5240304270155016,
      "grad_norm": 0.7173577547073364,
      "learning_rate": 4.853971798260476e-06,
      "loss": 0.7105,
      "step": 18187
    },
    {
      "epoch": 0.5240592404771509,
      "grad_norm": 0.6955465078353882,
      "learning_rate": 4.85350538781598e-06,
      "loss": 0.7272,
      "step": 18188
    },
    {
      "epoch": 0.5240880539388002,
      "grad_norm": 0.7402439117431641,
      "learning_rate": 4.853038978647306e-06,
      "loss": 0.7133,
      "step": 18189
    },
    {
      "epoch": 0.5241168674004495,
      "grad_norm": 0.7148165702819824,
      "learning_rate": 4.852572570758516e-06,
      "loss": 0.7208,
      "step": 18190
    },
    {
      "epoch": 0.5241456808620988,
      "grad_norm": 0.6969513297080994,
      "learning_rate": 4.852106164153668e-06,
      "loss": 0.7167,
      "step": 18191
    },
    {
      "epoch": 0.5241744943237481,
      "grad_norm": 0.7393292188644409,
      "learning_rate": 4.851639758836831e-06,
      "loss": 0.7212,
      "step": 18192
    },
    {
      "epoch": 0.5242033077853974,
      "grad_norm": 0.7304657101631165,
      "learning_rate": 4.851173354812062e-06,
      "loss": 0.7266,
      "step": 18193
    },
    {
      "epoch": 0.5242321212470467,
      "grad_norm": 0.7107395529747009,
      "learning_rate": 4.850706952083426e-06,
      "loss": 0.6949,
      "step": 18194
    },
    {
      "epoch": 0.524260934708696,
      "grad_norm": 0.7140857577323914,
      "learning_rate": 4.850240550654982e-06,
      "loss": 0.7365,
      "step": 18195
    },
    {
      "epoch": 0.5242897481703452,
      "grad_norm": 0.7247562408447266,
      "learning_rate": 4.849774150530794e-06,
      "loss": 0.7281,
      "step": 18196
    },
    {
      "epoch": 0.5243185616319944,
      "grad_norm": 0.7422133684158325,
      "learning_rate": 4.849307751714922e-06,
      "loss": 0.7325,
      "step": 18197
    },
    {
      "epoch": 0.5243473750936437,
      "grad_norm": 0.695290744304657,
      "learning_rate": 4.848841354211429e-06,
      "loss": 0.6948,
      "step": 18198
    },
    {
      "epoch": 0.524376188555293,
      "grad_norm": 0.7218056917190552,
      "learning_rate": 4.848374958024377e-06,
      "loss": 0.7297,
      "step": 18199
    },
    {
      "epoch": 0.5244050020169423,
      "grad_norm": 0.7205285429954529,
      "learning_rate": 4.847908563157827e-06,
      "loss": 0.7383,
      "step": 18200
    },
    {
      "epoch": 0.5244338154785916,
      "grad_norm": 0.7231220006942749,
      "learning_rate": 4.847442169615842e-06,
      "loss": 0.7033,
      "step": 18201
    },
    {
      "epoch": 0.5244626289402409,
      "grad_norm": 0.7009738683700562,
      "learning_rate": 4.846975777402483e-06,
      "loss": 0.7042,
      "step": 18202
    },
    {
      "epoch": 0.5244914424018902,
      "grad_norm": 0.7147380709648132,
      "learning_rate": 4.84650938652181e-06,
      "loss": 0.7132,
      "step": 18203
    },
    {
      "epoch": 0.5245202558635395,
      "grad_norm": 0.7742059230804443,
      "learning_rate": 4.846042996977889e-06,
      "loss": 0.718,
      "step": 18204
    },
    {
      "epoch": 0.5245490693251887,
      "grad_norm": 0.7418673038482666,
      "learning_rate": 4.84557660877478e-06,
      "loss": 0.7325,
      "step": 18205
    },
    {
      "epoch": 0.524577882786838,
      "grad_norm": 0.7184366583824158,
      "learning_rate": 4.845110221916544e-06,
      "loss": 0.718,
      "step": 18206
    },
    {
      "epoch": 0.5246066962484873,
      "grad_norm": 0.7626436352729797,
      "learning_rate": 4.844643836407242e-06,
      "loss": 0.7279,
      "step": 18207
    },
    {
      "epoch": 0.5246355097101366,
      "grad_norm": 0.7334961891174316,
      "learning_rate": 4.844177452250939e-06,
      "loss": 0.7111,
      "step": 18208
    },
    {
      "epoch": 0.5246643231717859,
      "grad_norm": 0.737880289554596,
      "learning_rate": 4.843711069451693e-06,
      "loss": 0.7225,
      "step": 18209
    },
    {
      "epoch": 0.5246931366334351,
      "grad_norm": 0.7311429381370544,
      "learning_rate": 4.843244688013568e-06,
      "loss": 0.7557,
      "step": 18210
    },
    {
      "epoch": 0.5247219500950844,
      "grad_norm": 0.7564249634742737,
      "learning_rate": 4.842778307940625e-06,
      "loss": 0.7254,
      "step": 18211
    },
    {
      "epoch": 0.5247507635567337,
      "grad_norm": 0.7147236466407776,
      "learning_rate": 4.842311929236926e-06,
      "loss": 0.7102,
      "step": 18212
    },
    {
      "epoch": 0.524779577018383,
      "grad_norm": 0.7154345512390137,
      "learning_rate": 4.8418455519065325e-06,
      "loss": 0.702,
      "step": 18213
    },
    {
      "epoch": 0.5248083904800322,
      "grad_norm": 0.7478576302528381,
      "learning_rate": 4.841379175953505e-06,
      "loss": 0.7052,
      "step": 18214
    },
    {
      "epoch": 0.5248372039416815,
      "grad_norm": 0.7276808023452759,
      "learning_rate": 4.840912801381905e-06,
      "loss": 0.7396,
      "step": 18215
    },
    {
      "epoch": 0.5248660174033308,
      "grad_norm": 0.7090876698493958,
      "learning_rate": 4.840446428195798e-06,
      "loss": 0.7176,
      "step": 18216
    },
    {
      "epoch": 0.5248948308649801,
      "grad_norm": 0.7140159010887146,
      "learning_rate": 4.839980056399243e-06,
      "loss": 0.6793,
      "step": 18217
    },
    {
      "epoch": 0.5249236443266294,
      "grad_norm": 0.6956959366798401,
      "learning_rate": 4.839513685996302e-06,
      "loss": 0.7059,
      "step": 18218
    },
    {
      "epoch": 0.5249524577882787,
      "grad_norm": 0.7097735404968262,
      "learning_rate": 4.839047316991035e-06,
      "loss": 0.7117,
      "step": 18219
    },
    {
      "epoch": 0.524981271249928,
      "grad_norm": 0.7330345511436462,
      "learning_rate": 4.838580949387505e-06,
      "loss": 0.7121,
      "step": 18220
    },
    {
      "epoch": 0.5250100847115773,
      "grad_norm": 0.7301748394966125,
      "learning_rate": 4.838114583189775e-06,
      "loss": 0.7209,
      "step": 18221
    },
    {
      "epoch": 0.5250388981732266,
      "grad_norm": 0.7187429070472717,
      "learning_rate": 4.837648218401904e-06,
      "loss": 0.696,
      "step": 18222
    },
    {
      "epoch": 0.5250677116348759,
      "grad_norm": 0.722254753112793,
      "learning_rate": 4.8371818550279554e-06,
      "loss": 0.7073,
      "step": 18223
    },
    {
      "epoch": 0.525096525096525,
      "grad_norm": 0.7229589819908142,
      "learning_rate": 4.83671549307199e-06,
      "loss": 0.7434,
      "step": 18224
    },
    {
      "epoch": 0.5251253385581743,
      "grad_norm": 0.7104611396789551,
      "learning_rate": 4.8362491325380695e-06,
      "loss": 0.6953,
      "step": 18225
    },
    {
      "epoch": 0.5251541520198236,
      "grad_norm": 0.7071658968925476,
      "learning_rate": 4.835782773430255e-06,
      "loss": 0.7046,
      "step": 18226
    },
    {
      "epoch": 0.5251829654814729,
      "grad_norm": 0.7208617925643921,
      "learning_rate": 4.835316415752607e-06,
      "loss": 0.7319,
      "step": 18227
    },
    {
      "epoch": 0.5252117789431222,
      "grad_norm": 0.7370706796646118,
      "learning_rate": 4.83485005950919e-06,
      "loss": 0.7151,
      "step": 18228
    },
    {
      "epoch": 0.5252405924047715,
      "grad_norm": 0.7318857312202454,
      "learning_rate": 4.834383704704064e-06,
      "loss": 0.6925,
      "step": 18229
    },
    {
      "epoch": 0.5252694058664208,
      "grad_norm": 0.7176319360733032,
      "learning_rate": 4.83391735134129e-06,
      "loss": 0.7227,
      "step": 18230
    },
    {
      "epoch": 0.5252982193280701,
      "grad_norm": 0.7063513994216919,
      "learning_rate": 4.83345099942493e-06,
      "loss": 0.6617,
      "step": 18231
    },
    {
      "epoch": 0.5253270327897194,
      "grad_norm": 0.7208842039108276,
      "learning_rate": 4.832984648959046e-06,
      "loss": 0.7263,
      "step": 18232
    },
    {
      "epoch": 0.5253558462513687,
      "grad_norm": 0.7128525376319885,
      "learning_rate": 4.832518299947698e-06,
      "loss": 0.713,
      "step": 18233
    },
    {
      "epoch": 0.525384659713018,
      "grad_norm": 0.7029523849487305,
      "learning_rate": 4.832051952394949e-06,
      "loss": 0.7102,
      "step": 18234
    },
    {
      "epoch": 0.5254134731746672,
      "grad_norm": 0.7247812151908875,
      "learning_rate": 4.831585606304858e-06,
      "loss": 0.7234,
      "step": 18235
    },
    {
      "epoch": 0.5254422866363165,
      "grad_norm": 0.7363582253456116,
      "learning_rate": 4.8311192616814885e-06,
      "loss": 0.7303,
      "step": 18236
    },
    {
      "epoch": 0.5254711000979657,
      "grad_norm": 0.722908616065979,
      "learning_rate": 4.830652918528902e-06,
      "loss": 0.7219,
      "step": 18237
    },
    {
      "epoch": 0.525499913559615,
      "grad_norm": 0.6967519521713257,
      "learning_rate": 4.830186576851159e-06,
      "loss": 0.729,
      "step": 18238
    },
    {
      "epoch": 0.5255287270212643,
      "grad_norm": 0.7309848070144653,
      "learning_rate": 4.82972023665232e-06,
      "loss": 0.7382,
      "step": 18239
    },
    {
      "epoch": 0.5255575404829136,
      "grad_norm": 0.699526846408844,
      "learning_rate": 4.82925389793645e-06,
      "loss": 0.698,
      "step": 18240
    },
    {
      "epoch": 0.5255863539445629,
      "grad_norm": 0.7305209636688232,
      "learning_rate": 4.828787560707605e-06,
      "loss": 0.7186,
      "step": 18241
    },
    {
      "epoch": 0.5256151674062122,
      "grad_norm": 0.7003429532051086,
      "learning_rate": 4.828321224969851e-06,
      "loss": 0.6954,
      "step": 18242
    },
    {
      "epoch": 0.5256439808678615,
      "grad_norm": 0.7195448875427246,
      "learning_rate": 4.827854890727248e-06,
      "loss": 0.6891,
      "step": 18243
    },
    {
      "epoch": 0.5256727943295108,
      "grad_norm": 0.7269355654716492,
      "learning_rate": 4.827388557983856e-06,
      "loss": 0.718,
      "step": 18244
    },
    {
      "epoch": 0.52570160779116,
      "grad_norm": 0.7016419172286987,
      "learning_rate": 4.8269222267437375e-06,
      "loss": 0.6927,
      "step": 18245
    },
    {
      "epoch": 0.5257304212528093,
      "grad_norm": 0.7274549007415771,
      "learning_rate": 4.826455897010954e-06,
      "loss": 0.6969,
      "step": 18246
    },
    {
      "epoch": 0.5257592347144586,
      "grad_norm": 0.7573404908180237,
      "learning_rate": 4.8259895687895655e-06,
      "loss": 0.7397,
      "step": 18247
    },
    {
      "epoch": 0.5257880481761079,
      "grad_norm": 0.7140019536018372,
      "learning_rate": 4.8255232420836335e-06,
      "loss": 0.71,
      "step": 18248
    },
    {
      "epoch": 0.5258168616377572,
      "grad_norm": 0.7173097133636475,
      "learning_rate": 4.82505691689722e-06,
      "loss": 0.7073,
      "step": 18249
    },
    {
      "epoch": 0.5258456750994065,
      "grad_norm": 0.7450119853019714,
      "learning_rate": 4.824590593234386e-06,
      "loss": 0.7019,
      "step": 18250
    },
    {
      "epoch": 0.5258744885610557,
      "grad_norm": 0.7204456925392151,
      "learning_rate": 4.824124271099191e-06,
      "loss": 0.7338,
      "step": 18251
    },
    {
      "epoch": 0.525903302022705,
      "grad_norm": 0.7026956081390381,
      "learning_rate": 4.823657950495699e-06,
      "loss": 0.7127,
      "step": 18252
    },
    {
      "epoch": 0.5259321154843543,
      "grad_norm": 0.7472262978553772,
      "learning_rate": 4.82319163142797e-06,
      "loss": 0.7202,
      "step": 18253
    },
    {
      "epoch": 0.5259609289460035,
      "grad_norm": 0.7638210654258728,
      "learning_rate": 4.822725313900065e-06,
      "loss": 0.7251,
      "step": 18254
    },
    {
      "epoch": 0.5259897424076528,
      "grad_norm": 0.7114267945289612,
      "learning_rate": 4.8222589979160464e-06,
      "loss": 0.7268,
      "step": 18255
    },
    {
      "epoch": 0.5260185558693021,
      "grad_norm": 0.7105697393417358,
      "learning_rate": 4.821792683479974e-06,
      "loss": 0.7065,
      "step": 18256
    },
    {
      "epoch": 0.5260473693309514,
      "grad_norm": 1.5393109321594238,
      "learning_rate": 4.821326370595908e-06,
      "loss": 0.6973,
      "step": 18257
    },
    {
      "epoch": 0.5260761827926007,
      "grad_norm": 0.70965975522995,
      "learning_rate": 4.8208600592679105e-06,
      "loss": 0.7452,
      "step": 18258
    },
    {
      "epoch": 0.52610499625425,
      "grad_norm": 0.7130054831504822,
      "learning_rate": 4.820393749500044e-06,
      "loss": 0.7193,
      "step": 18259
    },
    {
      "epoch": 0.5261338097158993,
      "grad_norm": 0.7219467163085938,
      "learning_rate": 4.8199274412963685e-06,
      "loss": 0.7097,
      "step": 18260
    },
    {
      "epoch": 0.5261626231775486,
      "grad_norm": 0.8962081670761108,
      "learning_rate": 4.819461134660944e-06,
      "loss": 0.7282,
      "step": 18261
    },
    {
      "epoch": 0.5261914366391979,
      "grad_norm": 0.7093482613563538,
      "learning_rate": 4.818994829597833e-06,
      "loss": 0.6973,
      "step": 18262
    },
    {
      "epoch": 0.5262202501008472,
      "grad_norm": 0.7129988670349121,
      "learning_rate": 4.818528526111094e-06,
      "loss": 0.711,
      "step": 18263
    },
    {
      "epoch": 0.5262490635624963,
      "grad_norm": 0.6960892677307129,
      "learning_rate": 4.818062224204793e-06,
      "loss": 0.7201,
      "step": 18264
    },
    {
      "epoch": 0.5262778770241456,
      "grad_norm": 0.7307453751564026,
      "learning_rate": 4.817595923882988e-06,
      "loss": 0.7291,
      "step": 18265
    },
    {
      "epoch": 0.5263066904857949,
      "grad_norm": 0.7078384160995483,
      "learning_rate": 4.8171296251497395e-06,
      "loss": 0.724,
      "step": 18266
    },
    {
      "epoch": 0.5263355039474442,
      "grad_norm": 0.7037187814712524,
      "learning_rate": 4.816663328009109e-06,
      "loss": 0.7076,
      "step": 18267
    },
    {
      "epoch": 0.5263643174090935,
      "grad_norm": 0.7251625061035156,
      "learning_rate": 4.816197032465158e-06,
      "loss": 0.703,
      "step": 18268
    },
    {
      "epoch": 0.5263931308707428,
      "grad_norm": 0.7332080602645874,
      "learning_rate": 4.8157307385219474e-06,
      "loss": 0.717,
      "step": 18269
    },
    {
      "epoch": 0.5264219443323921,
      "grad_norm": 0.7009550333023071,
      "learning_rate": 4.815264446183538e-06,
      "loss": 0.7067,
      "step": 18270
    },
    {
      "epoch": 0.5264507577940414,
      "grad_norm": 0.7243340611457825,
      "learning_rate": 4.81479815545399e-06,
      "loss": 0.723,
      "step": 18271
    },
    {
      "epoch": 0.5264795712556907,
      "grad_norm": 1.7708252668380737,
      "learning_rate": 4.814331866337365e-06,
      "loss": 0.7195,
      "step": 18272
    },
    {
      "epoch": 0.52650838471734,
      "grad_norm": 0.7422600388526917,
      "learning_rate": 4.813865578837724e-06,
      "loss": 0.7142,
      "step": 18273
    },
    {
      "epoch": 0.5265371981789893,
      "grad_norm": 0.74106365442276,
      "learning_rate": 4.813399292959127e-06,
      "loss": 0.736,
      "step": 18274
    },
    {
      "epoch": 0.5265660116406385,
      "grad_norm": 0.7212105989456177,
      "learning_rate": 4.812933008705636e-06,
      "loss": 0.7019,
      "step": 18275
    },
    {
      "epoch": 0.5265948251022878,
      "grad_norm": 0.7260007858276367,
      "learning_rate": 4.812466726081312e-06,
      "loss": 0.7161,
      "step": 18276
    },
    {
      "epoch": 0.5266236385639371,
      "grad_norm": 0.7489156723022461,
      "learning_rate": 4.812000445090215e-06,
      "loss": 0.7365,
      "step": 18277
    },
    {
      "epoch": 0.5266524520255863,
      "grad_norm": 0.7111011743545532,
      "learning_rate": 4.811534165736406e-06,
      "loss": 0.7329,
      "step": 18278
    },
    {
      "epoch": 0.5266812654872356,
      "grad_norm": 0.7548162341117859,
      "learning_rate": 4.811067888023946e-06,
      "loss": 0.7107,
      "step": 18279
    },
    {
      "epoch": 0.5267100789488849,
      "grad_norm": 0.7178398966789246,
      "learning_rate": 4.810601611956896e-06,
      "loss": 0.7394,
      "step": 18280
    },
    {
      "epoch": 0.5267388924105342,
      "grad_norm": 0.7140849828720093,
      "learning_rate": 4.8101353375393165e-06,
      "loss": 0.7234,
      "step": 18281
    },
    {
      "epoch": 0.5267677058721835,
      "grad_norm": 0.7075029611587524,
      "learning_rate": 4.809669064775268e-06,
      "loss": 0.703,
      "step": 18282
    },
    {
      "epoch": 0.5267965193338328,
      "grad_norm": 0.711807370185852,
      "learning_rate": 4.809202793668814e-06,
      "loss": 0.7036,
      "step": 18283
    },
    {
      "epoch": 0.526825332795482,
      "grad_norm": 0.7521527409553528,
      "learning_rate": 4.808736524224011e-06,
      "loss": 0.735,
      "step": 18284
    },
    {
      "epoch": 0.5268541462571313,
      "grad_norm": 0.7298972010612488,
      "learning_rate": 4.808270256444922e-06,
      "loss": 0.7155,
      "step": 18285
    },
    {
      "epoch": 0.5268829597187806,
      "grad_norm": 0.6974954605102539,
      "learning_rate": 4.807803990335607e-06,
      "loss": 0.6965,
      "step": 18286
    },
    {
      "epoch": 0.5269117731804299,
      "grad_norm": 0.742355465888977,
      "learning_rate": 4.807337725900126e-06,
      "loss": 0.7236,
      "step": 18287
    },
    {
      "epoch": 0.5269405866420792,
      "grad_norm": 0.7411687970161438,
      "learning_rate": 4.806871463142542e-06,
      "loss": 0.7134,
      "step": 18288
    },
    {
      "epoch": 0.5269694001037285,
      "grad_norm": 0.702844500541687,
      "learning_rate": 4.806405202066916e-06,
      "loss": 0.7001,
      "step": 18289
    },
    {
      "epoch": 0.5269982135653778,
      "grad_norm": 1.4941987991333008,
      "learning_rate": 4.805938942677305e-06,
      "loss": 0.7259,
      "step": 18290
    },
    {
      "epoch": 0.527027027027027,
      "grad_norm": 0.7310944199562073,
      "learning_rate": 4.805472684977773e-06,
      "loss": 0.7233,
      "step": 18291
    },
    {
      "epoch": 0.5270558404886763,
      "grad_norm": 0.7254542708396912,
      "learning_rate": 4.80500642897238e-06,
      "loss": 0.7094,
      "step": 18292
    },
    {
      "epoch": 0.5270846539503256,
      "grad_norm": 0.7222504615783691,
      "learning_rate": 4.804540174665186e-06,
      "loss": 0.6894,
      "step": 18293
    },
    {
      "epoch": 0.5271134674119748,
      "grad_norm": 0.7430107593536377,
      "learning_rate": 4.804073922060251e-06,
      "loss": 0.7041,
      "step": 18294
    },
    {
      "epoch": 0.5271422808736241,
      "grad_norm": 0.6988725066184998,
      "learning_rate": 4.8036076711616355e-06,
      "loss": 0.7235,
      "step": 18295
    },
    {
      "epoch": 0.5271710943352734,
      "grad_norm": 0.7093551754951477,
      "learning_rate": 4.803141421973401e-06,
      "loss": 0.7036,
      "step": 18296
    },
    {
      "epoch": 0.5271999077969227,
      "grad_norm": 0.7310365438461304,
      "learning_rate": 4.802675174499609e-06,
      "loss": 0.7286,
      "step": 18297
    },
    {
      "epoch": 0.527228721258572,
      "grad_norm": 0.719376266002655,
      "learning_rate": 4.802208928744319e-06,
      "loss": 0.7114,
      "step": 18298
    },
    {
      "epoch": 0.5272575347202213,
      "grad_norm": 0.7132092714309692,
      "learning_rate": 4.80174268471159e-06,
      "loss": 0.7266,
      "step": 18299
    },
    {
      "epoch": 0.5272863481818706,
      "grad_norm": 0.7091084718704224,
      "learning_rate": 4.801276442405484e-06,
      "loss": 0.7275,
      "step": 18300
    },
    {
      "epoch": 0.5273151616435199,
      "grad_norm": 0.7154801487922668,
      "learning_rate": 4.8008102018300625e-06,
      "loss": 0.7043,
      "step": 18301
    },
    {
      "epoch": 0.5273439751051692,
      "grad_norm": 0.7240828275680542,
      "learning_rate": 4.800343962989386e-06,
      "loss": 0.7199,
      "step": 18302
    },
    {
      "epoch": 0.5273727885668185,
      "grad_norm": 0.712517261505127,
      "learning_rate": 4.7998777258875135e-06,
      "loss": 0.7085,
      "step": 18303
    },
    {
      "epoch": 0.5274016020284678,
      "grad_norm": 0.7545091509819031,
      "learning_rate": 4.7994114905285055e-06,
      "loss": 0.722,
      "step": 18304
    },
    {
      "epoch": 0.5274304154901169,
      "grad_norm": 0.7114478349685669,
      "learning_rate": 4.798945256916423e-06,
      "loss": 0.7124,
      "step": 18305
    },
    {
      "epoch": 0.5274592289517662,
      "grad_norm": 0.7212619185447693,
      "learning_rate": 4.798479025055327e-06,
      "loss": 0.7342,
      "step": 18306
    },
    {
      "epoch": 0.5274880424134155,
      "grad_norm": 0.7142083048820496,
      "learning_rate": 4.798012794949277e-06,
      "loss": 0.7097,
      "step": 18307
    },
    {
      "epoch": 0.5275168558750648,
      "grad_norm": 0.7678296566009521,
      "learning_rate": 4.797546566602333e-06,
      "loss": 0.713,
      "step": 18308
    },
    {
      "epoch": 0.5275456693367141,
      "grad_norm": 0.6959379315376282,
      "learning_rate": 4.797080340018556e-06,
      "loss": 0.7115,
      "step": 18309
    },
    {
      "epoch": 0.5275744827983634,
      "grad_norm": 0.7172907590866089,
      "learning_rate": 4.796614115202008e-06,
      "loss": 0.6875,
      "step": 18310
    },
    {
      "epoch": 0.5276032962600127,
      "grad_norm": 0.7182321548461914,
      "learning_rate": 4.796147892156744e-06,
      "loss": 0.6936,
      "step": 18311
    },
    {
      "epoch": 0.527632109721662,
      "grad_norm": 0.7233259081840515,
      "learning_rate": 4.79568167088683e-06,
      "loss": 0.7296,
      "step": 18312
    },
    {
      "epoch": 0.5276609231833113,
      "grad_norm": 0.7002628445625305,
      "learning_rate": 4.7952154513963255e-06,
      "loss": 0.6976,
      "step": 18313
    },
    {
      "epoch": 0.5276897366449605,
      "grad_norm": 0.746087372303009,
      "learning_rate": 4.794749233689289e-06,
      "loss": 0.7175,
      "step": 18314
    },
    {
      "epoch": 0.5277185501066098,
      "grad_norm": 0.7524804472923279,
      "learning_rate": 4.794283017769782e-06,
      "loss": 0.7401,
      "step": 18315
    },
    {
      "epoch": 0.5277473635682591,
      "grad_norm": 0.7164206504821777,
      "learning_rate": 4.7938168036418644e-06,
      "loss": 0.6979,
      "step": 18316
    },
    {
      "epoch": 0.5277761770299084,
      "grad_norm": 0.7245425581932068,
      "learning_rate": 4.7933505913095945e-06,
      "loss": 0.7099,
      "step": 18317
    },
    {
      "epoch": 0.5278049904915576,
      "grad_norm": 0.7332217693328857,
      "learning_rate": 4.792884380777037e-06,
      "loss": 0.7051,
      "step": 18318
    },
    {
      "epoch": 0.5278338039532069,
      "grad_norm": 0.7487154006958008,
      "learning_rate": 4.792418172048247e-06,
      "loss": 0.7305,
      "step": 18319
    },
    {
      "epoch": 0.5278626174148562,
      "grad_norm": 0.6858941316604614,
      "learning_rate": 4.79195196512729e-06,
      "loss": 0.719,
      "step": 18320
    },
    {
      "epoch": 0.5278914308765055,
      "grad_norm": 0.7085921764373779,
      "learning_rate": 4.791485760018221e-06,
      "loss": 0.7209,
      "step": 18321
    },
    {
      "epoch": 0.5279202443381548,
      "grad_norm": 0.7372850179672241,
      "learning_rate": 4.791019556725104e-06,
      "loss": 0.7058,
      "step": 18322
    },
    {
      "epoch": 0.5279490577998041,
      "grad_norm": 0.7367975115776062,
      "learning_rate": 4.790553355251995e-06,
      "loss": 0.696,
      "step": 18323
    },
    {
      "epoch": 0.5279778712614533,
      "grad_norm": 0.7496457695960999,
      "learning_rate": 4.79008715560296e-06,
      "loss": 0.7214,
      "step": 18324
    },
    {
      "epoch": 0.5280066847231026,
      "grad_norm": 0.727094292640686,
      "learning_rate": 4.789620957782055e-06,
      "loss": 0.7215,
      "step": 18325
    },
    {
      "epoch": 0.5280354981847519,
      "grad_norm": 0.7303274273872375,
      "learning_rate": 4.789154761793341e-06,
      "loss": 0.7446,
      "step": 18326
    },
    {
      "epoch": 0.5280643116464012,
      "grad_norm": 0.7306408882141113,
      "learning_rate": 4.788688567640879e-06,
      "loss": 0.7211,
      "step": 18327
    },
    {
      "epoch": 0.5280931251080505,
      "grad_norm": 0.7311638593673706,
      "learning_rate": 4.788222375328728e-06,
      "loss": 0.7294,
      "step": 18328
    },
    {
      "epoch": 0.5281219385696998,
      "grad_norm": 0.7617825269699097,
      "learning_rate": 4.7877561848609484e-06,
      "loss": 0.7369,
      "step": 18329
    },
    {
      "epoch": 0.5281507520313491,
      "grad_norm": 0.7472915053367615,
      "learning_rate": 4.7872899962416005e-06,
      "loss": 0.7145,
      "step": 18330
    },
    {
      "epoch": 0.5281795654929984,
      "grad_norm": 0.7285041809082031,
      "learning_rate": 4.786823809474744e-06,
      "loss": 0.699,
      "step": 18331
    },
    {
      "epoch": 0.5282083789546476,
      "grad_norm": 0.7187921404838562,
      "learning_rate": 4.786357624564438e-06,
      "loss": 0.7117,
      "step": 18332
    },
    {
      "epoch": 0.5282371924162969,
      "grad_norm": 0.7401555180549622,
      "learning_rate": 4.785891441514744e-06,
      "loss": 0.6998,
      "step": 18333
    },
    {
      "epoch": 0.5282660058779461,
      "grad_norm": 0.7414658069610596,
      "learning_rate": 4.785425260329722e-06,
      "loss": 0.6977,
      "step": 18334
    },
    {
      "epoch": 0.5282948193395954,
      "grad_norm": 0.8029593229293823,
      "learning_rate": 4.78495908101343e-06,
      "loss": 0.7424,
      "step": 18335
    },
    {
      "epoch": 0.5283236328012447,
      "grad_norm": 0.7263762950897217,
      "learning_rate": 4.784492903569931e-06,
      "loss": 0.726,
      "step": 18336
    },
    {
      "epoch": 0.528352446262894,
      "grad_norm": 0.7159547805786133,
      "learning_rate": 4.784026728003284e-06,
      "loss": 0.7209,
      "step": 18337
    },
    {
      "epoch": 0.5283812597245433,
      "grad_norm": 0.7551707029342651,
      "learning_rate": 4.783560554317546e-06,
      "loss": 0.7249,
      "step": 18338
    },
    {
      "epoch": 0.5284100731861926,
      "grad_norm": 0.7748745679855347,
      "learning_rate": 4.783094382516781e-06,
      "loss": 0.7316,
      "step": 18339
    },
    {
      "epoch": 0.5284388866478419,
      "grad_norm": 0.7135488986968994,
      "learning_rate": 4.782628212605047e-06,
      "loss": 0.7176,
      "step": 18340
    },
    {
      "epoch": 0.5284677001094912,
      "grad_norm": 0.7361030578613281,
      "learning_rate": 4.782162044586404e-06,
      "loss": 0.7068,
      "step": 18341
    },
    {
      "epoch": 0.5284965135711405,
      "grad_norm": 0.7274897694587708,
      "learning_rate": 4.781695878464912e-06,
      "loss": 0.7176,
      "step": 18342
    },
    {
      "epoch": 0.5285253270327898,
      "grad_norm": 0.7365741729736328,
      "learning_rate": 4.78122971424463e-06,
      "loss": 0.7163,
      "step": 18343
    },
    {
      "epoch": 0.528554140494439,
      "grad_norm": 0.7533118724822998,
      "learning_rate": 4.780763551929618e-06,
      "loss": 0.7334,
      "step": 18344
    },
    {
      "epoch": 0.5285829539560882,
      "grad_norm": 0.761777400970459,
      "learning_rate": 4.7802973915239375e-06,
      "loss": 0.698,
      "step": 18345
    },
    {
      "epoch": 0.5286117674177375,
      "grad_norm": 0.7406346201896667,
      "learning_rate": 4.779831233031647e-06,
      "loss": 0.7017,
      "step": 18346
    },
    {
      "epoch": 0.5286405808793868,
      "grad_norm": 0.7425021529197693,
      "learning_rate": 4.779365076456804e-06,
      "loss": 0.7185,
      "step": 18347
    },
    {
      "epoch": 0.5286693943410361,
      "grad_norm": 0.7178863286972046,
      "learning_rate": 4.778898921803473e-06,
      "loss": 0.7045,
      "step": 18348
    },
    {
      "epoch": 0.5286982078026854,
      "grad_norm": 0.7134808897972107,
      "learning_rate": 4.778432769075711e-06,
      "loss": 0.6773,
      "step": 18349
    },
    {
      "epoch": 0.5287270212643347,
      "grad_norm": 0.7858219742774963,
      "learning_rate": 4.777966618277578e-06,
      "loss": 0.724,
      "step": 18350
    },
    {
      "epoch": 0.528755834725984,
      "grad_norm": 0.7318552732467651,
      "learning_rate": 4.777500469413135e-06,
      "loss": 0.6975,
      "step": 18351
    },
    {
      "epoch": 0.5287846481876333,
      "grad_norm": 0.7357433438301086,
      "learning_rate": 4.777034322486439e-06,
      "loss": 0.7128,
      "step": 18352
    },
    {
      "epoch": 0.5288134616492826,
      "grad_norm": 0.7281168699264526,
      "learning_rate": 4.776568177501551e-06,
      "loss": 0.7216,
      "step": 18353
    },
    {
      "epoch": 0.5288422751109318,
      "grad_norm": 0.7250317931175232,
      "learning_rate": 4.776102034462533e-06,
      "loss": 0.7109,
      "step": 18354
    },
    {
      "epoch": 0.5288710885725811,
      "grad_norm": 0.7394753098487854,
      "learning_rate": 4.775635893373439e-06,
      "loss": 0.7093,
      "step": 18355
    },
    {
      "epoch": 0.5288999020342304,
      "grad_norm": 0.7595610022544861,
      "learning_rate": 4.775169754238334e-06,
      "loss": 0.7201,
      "step": 18356
    },
    {
      "epoch": 0.5289287154958797,
      "grad_norm": 0.778480589389801,
      "learning_rate": 4.774703617061276e-06,
      "loss": 0.7114,
      "step": 18357
    },
    {
      "epoch": 0.528957528957529,
      "grad_norm": 0.6993768215179443,
      "learning_rate": 4.774237481846325e-06,
      "loss": 0.6873,
      "step": 18358
    },
    {
      "epoch": 0.5289863424191782,
      "grad_norm": 0.7252992987632751,
      "learning_rate": 4.773771348597537e-06,
      "loss": 0.7096,
      "step": 18359
    },
    {
      "epoch": 0.5290151558808275,
      "grad_norm": 0.7403355836868286,
      "learning_rate": 4.773305217318975e-06,
      "loss": 0.7169,
      "step": 18360
    },
    {
      "epoch": 0.5290439693424768,
      "grad_norm": 0.7409709692001343,
      "learning_rate": 4.7728390880147e-06,
      "loss": 0.7175,
      "step": 18361
    },
    {
      "epoch": 0.5290727828041261,
      "grad_norm": 0.760477602481842,
      "learning_rate": 4.772372960688768e-06,
      "loss": 0.7255,
      "step": 18362
    },
    {
      "epoch": 0.5291015962657754,
      "grad_norm": 0.7083141803741455,
      "learning_rate": 4.771906835345242e-06,
      "loss": 0.701,
      "step": 18363
    },
    {
      "epoch": 0.5291304097274246,
      "grad_norm": 0.7410694360733032,
      "learning_rate": 4.771440711988177e-06,
      "loss": 0.7074,
      "step": 18364
    },
    {
      "epoch": 0.5291592231890739,
      "grad_norm": 0.7319681644439697,
      "learning_rate": 4.770974590621637e-06,
      "loss": 0.711,
      "step": 18365
    },
    {
      "epoch": 0.5291880366507232,
      "grad_norm": 0.7147519588470459,
      "learning_rate": 4.770508471249678e-06,
      "loss": 0.7137,
      "step": 18366
    },
    {
      "epoch": 0.5292168501123725,
      "grad_norm": 0.7362297773361206,
      "learning_rate": 4.770042353876361e-06,
      "loss": 0.7167,
      "step": 18367
    },
    {
      "epoch": 0.5292456635740218,
      "grad_norm": 0.736458420753479,
      "learning_rate": 4.769576238505745e-06,
      "loss": 0.7088,
      "step": 18368
    },
    {
      "epoch": 0.5292744770356711,
      "grad_norm": 0.7144027948379517,
      "learning_rate": 4.769110125141891e-06,
      "loss": 0.711,
      "step": 18369
    },
    {
      "epoch": 0.5293032904973204,
      "grad_norm": 0.7293190360069275,
      "learning_rate": 4.768644013788856e-06,
      "loss": 0.7188,
      "step": 18370
    },
    {
      "epoch": 0.5293321039589697,
      "grad_norm": 0.7494437098503113,
      "learning_rate": 4.7681779044506984e-06,
      "loss": 0.7009,
      "step": 18371
    },
    {
      "epoch": 0.5293609174206189,
      "grad_norm": 0.7224692702293396,
      "learning_rate": 4.767711797131481e-06,
      "loss": 0.717,
      "step": 18372
    },
    {
      "epoch": 0.5293897308822682,
      "grad_norm": 0.7351012825965881,
      "learning_rate": 4.767245691835263e-06,
      "loss": 0.7194,
      "step": 18373
    },
    {
      "epoch": 0.5294185443439174,
      "grad_norm": 0.7271469235420227,
      "learning_rate": 4.766779588566101e-06,
      "loss": 0.7255,
      "step": 18374
    },
    {
      "epoch": 0.5294473578055667,
      "grad_norm": 0.736799955368042,
      "learning_rate": 4.766313487328057e-06,
      "loss": 0.6908,
      "step": 18375
    },
    {
      "epoch": 0.529476171267216,
      "grad_norm": 0.7558566331863403,
      "learning_rate": 4.765847388125187e-06,
      "loss": 0.7384,
      "step": 18376
    },
    {
      "epoch": 0.5295049847288653,
      "grad_norm": 0.7388220429420471,
      "learning_rate": 4.765381290961553e-06,
      "loss": 0.7107,
      "step": 18377
    },
    {
      "epoch": 0.5295337981905146,
      "grad_norm": 0.7304224967956543,
      "learning_rate": 4.764915195841214e-06,
      "loss": 0.7153,
      "step": 18378
    },
    {
      "epoch": 0.5295626116521639,
      "grad_norm": 0.7387336492538452,
      "learning_rate": 4.7644491027682285e-06,
      "loss": 0.7276,
      "step": 18379
    },
    {
      "epoch": 0.5295914251138132,
      "grad_norm": 0.7762206792831421,
      "learning_rate": 4.763983011746656e-06,
      "loss": 0.7238,
      "step": 18380
    },
    {
      "epoch": 0.5296202385754625,
      "grad_norm": 0.7632978558540344,
      "learning_rate": 4.763516922780555e-06,
      "loss": 0.7513,
      "step": 18381
    },
    {
      "epoch": 0.5296490520371118,
      "grad_norm": 0.7203973531723022,
      "learning_rate": 4.763050835873985e-06,
      "loss": 0.6902,
      "step": 18382
    },
    {
      "epoch": 0.5296778654987611,
      "grad_norm": 0.7464307546615601,
      "learning_rate": 4.762584751031004e-06,
      "loss": 0.7228,
      "step": 18383
    },
    {
      "epoch": 0.5297066789604103,
      "grad_norm": 0.7215530872344971,
      "learning_rate": 4.762118668255674e-06,
      "loss": 0.703,
      "step": 18384
    },
    {
      "epoch": 0.5297354924220596,
      "grad_norm": 0.7280231714248657,
      "learning_rate": 4.7616525875520534e-06,
      "loss": 0.7261,
      "step": 18385
    },
    {
      "epoch": 0.5297643058837088,
      "grad_norm": 0.7382673621177673,
      "learning_rate": 4.7611865089242e-06,
      "loss": 0.7087,
      "step": 18386
    },
    {
      "epoch": 0.5297931193453581,
      "grad_norm": 0.7275159358978271,
      "learning_rate": 4.760720432376174e-06,
      "loss": 0.7145,
      "step": 18387
    },
    {
      "epoch": 0.5298219328070074,
      "grad_norm": 0.7379099130630493,
      "learning_rate": 4.760254357912033e-06,
      "loss": 0.7298,
      "step": 18388
    },
    {
      "epoch": 0.5298507462686567,
      "grad_norm": 0.7460157871246338,
      "learning_rate": 4.759788285535838e-06,
      "loss": 0.727,
      "step": 18389
    },
    {
      "epoch": 0.529879559730306,
      "grad_norm": 0.7510473132133484,
      "learning_rate": 4.759322215251645e-06,
      "loss": 0.7229,
      "step": 18390
    },
    {
      "epoch": 0.5299083731919553,
      "grad_norm": 0.7335126399993896,
      "learning_rate": 4.758856147063516e-06,
      "loss": 0.7069,
      "step": 18391
    },
    {
      "epoch": 0.5299371866536046,
      "grad_norm": 0.7245515584945679,
      "learning_rate": 4.758390080975509e-06,
      "loss": 0.7261,
      "step": 18392
    },
    {
      "epoch": 0.5299660001152539,
      "grad_norm": 0.7366578578948975,
      "learning_rate": 4.757924016991682e-06,
      "loss": 0.714,
      "step": 18393
    },
    {
      "epoch": 0.5299948135769031,
      "grad_norm": 0.6966941952705383,
      "learning_rate": 4.757457955116096e-06,
      "loss": 0.6924,
      "step": 18394
    },
    {
      "epoch": 0.5300236270385524,
      "grad_norm": 0.7253084182739258,
      "learning_rate": 4.7569918953528065e-06,
      "loss": 0.6957,
      "step": 18395
    },
    {
      "epoch": 0.5300524405002017,
      "grad_norm": 0.7119225859642029,
      "learning_rate": 4.756525837705877e-06,
      "loss": 0.7034,
      "step": 18396
    },
    {
      "epoch": 0.530081253961851,
      "grad_norm": 0.7078456282615662,
      "learning_rate": 4.7560597821793636e-06,
      "loss": 0.7193,
      "step": 18397
    },
    {
      "epoch": 0.5301100674235003,
      "grad_norm": 0.7799537777900696,
      "learning_rate": 4.755593728777325e-06,
      "loss": 0.7071,
      "step": 18398
    },
    {
      "epoch": 0.5301388808851495,
      "grad_norm": 0.7412118911743164,
      "learning_rate": 4.755127677503821e-06,
      "loss": 0.7019,
      "step": 18399
    },
    {
      "epoch": 0.5301676943467988,
      "grad_norm": 0.7138177752494812,
      "learning_rate": 4.75466162836291e-06,
      "loss": 0.727,
      "step": 18400
    },
    {
      "epoch": 0.5301965078084481,
      "grad_norm": 0.7232889533042908,
      "learning_rate": 4.754195581358652e-06,
      "loss": 0.7057,
      "step": 18401
    },
    {
      "epoch": 0.5302253212700974,
      "grad_norm": 0.7300010323524475,
      "learning_rate": 4.753729536495104e-06,
      "loss": 0.7116,
      "step": 18402
    },
    {
      "epoch": 0.5302541347317467,
      "grad_norm": 0.7435796856880188,
      "learning_rate": 4.753263493776326e-06,
      "loss": 0.7361,
      "step": 18403
    },
    {
      "epoch": 0.5302829481933959,
      "grad_norm": 0.7325699925422668,
      "learning_rate": 4.752797453206376e-06,
      "loss": 0.7029,
      "step": 18404
    },
    {
      "epoch": 0.5303117616550452,
      "grad_norm": 0.7128349542617798,
      "learning_rate": 4.752331414789314e-06,
      "loss": 0.721,
      "step": 18405
    },
    {
      "epoch": 0.5303405751166945,
      "grad_norm": 0.749136209487915,
      "learning_rate": 4.751865378529194e-06,
      "loss": 0.713,
      "step": 18406
    },
    {
      "epoch": 0.5303693885783438,
      "grad_norm": 0.720302164554596,
      "learning_rate": 4.751399344430082e-06,
      "loss": 0.6962,
      "step": 18407
    },
    {
      "epoch": 0.5303982020399931,
      "grad_norm": 0.7401788830757141,
      "learning_rate": 4.750933312496032e-06,
      "loss": 0.7165,
      "step": 18408
    },
    {
      "epoch": 0.5304270155016424,
      "grad_norm": 0.7147355675697327,
      "learning_rate": 4.750467282731105e-06,
      "loss": 0.7255,
      "step": 18409
    },
    {
      "epoch": 0.5304558289632917,
      "grad_norm": 0.7109745740890503,
      "learning_rate": 4.750001255139358e-06,
      "loss": 0.7064,
      "step": 18410
    },
    {
      "epoch": 0.530484642424941,
      "grad_norm": 0.7454370856285095,
      "learning_rate": 4.749535229724851e-06,
      "loss": 0.7193,
      "step": 18411
    },
    {
      "epoch": 0.5305134558865903,
      "grad_norm": 0.711006760597229,
      "learning_rate": 4.74906920649164e-06,
      "loss": 0.6924,
      "step": 18412
    },
    {
      "epoch": 0.5305422693482394,
      "grad_norm": 0.71184241771698,
      "learning_rate": 4.7486031854437866e-06,
      "loss": 0.7177,
      "step": 18413
    },
    {
      "epoch": 0.5305710828098887,
      "grad_norm": 0.7070825695991516,
      "learning_rate": 4.748137166585347e-06,
      "loss": 0.701,
      "step": 18414
    },
    {
      "epoch": 0.530599896271538,
      "grad_norm": 0.694776713848114,
      "learning_rate": 4.747671149920382e-06,
      "loss": 0.6945,
      "step": 18415
    },
    {
      "epoch": 0.5306287097331873,
      "grad_norm": 0.6942062377929688,
      "learning_rate": 4.747205135452949e-06,
      "loss": 0.6985,
      "step": 18416
    },
    {
      "epoch": 0.5306575231948366,
      "grad_norm": 0.7074558138847351,
      "learning_rate": 4.7467391231871066e-06,
      "loss": 0.7071,
      "step": 18417
    },
    {
      "epoch": 0.5306863366564859,
      "grad_norm": 0.7371032238006592,
      "learning_rate": 4.746273113126912e-06,
      "loss": 0.7245,
      "step": 18418
    },
    {
      "epoch": 0.5307151501181352,
      "grad_norm": 0.7183969020843506,
      "learning_rate": 4.745807105276425e-06,
      "loss": 0.6975,
      "step": 18419
    },
    {
      "epoch": 0.5307439635797845,
      "grad_norm": 0.7303625345230103,
      "learning_rate": 4.745341099639705e-06,
      "loss": 0.7214,
      "step": 18420
    },
    {
      "epoch": 0.5307727770414338,
      "grad_norm": 0.7070521712303162,
      "learning_rate": 4.744875096220809e-06,
      "loss": 0.7015,
      "step": 18421
    },
    {
      "epoch": 0.5308015905030831,
      "grad_norm": 0.7376098036766052,
      "learning_rate": 4.744409095023797e-06,
      "loss": 0.7041,
      "step": 18422
    },
    {
      "epoch": 0.5308304039647324,
      "grad_norm": 0.7330108880996704,
      "learning_rate": 4.743943096052726e-06,
      "loss": 0.713,
      "step": 18423
    },
    {
      "epoch": 0.5308592174263816,
      "grad_norm": 0.713045060634613,
      "learning_rate": 4.743477099311654e-06,
      "loss": 0.6839,
      "step": 18424
    },
    {
      "epoch": 0.5308880308880309,
      "grad_norm": 0.7108554840087891,
      "learning_rate": 4.7430111048046404e-06,
      "loss": 0.6891,
      "step": 18425
    },
    {
      "epoch": 0.5309168443496801,
      "grad_norm": 0.7208558917045593,
      "learning_rate": 4.742545112535743e-06,
      "loss": 0.7072,
      "step": 18426
    },
    {
      "epoch": 0.5309456578113294,
      "grad_norm": 0.7635219097137451,
      "learning_rate": 4.742079122509021e-06,
      "loss": 0.7178,
      "step": 18427
    },
    {
      "epoch": 0.5309744712729787,
      "grad_norm": 0.7169034481048584,
      "learning_rate": 4.741613134728531e-06,
      "loss": 0.6985,
      "step": 18428
    },
    {
      "epoch": 0.531003284734628,
      "grad_norm": 0.7300109267234802,
      "learning_rate": 4.741147149198334e-06,
      "loss": 0.7164,
      "step": 18429
    },
    {
      "epoch": 0.5310320981962773,
      "grad_norm": 0.7191165089607239,
      "learning_rate": 4.740681165922483e-06,
      "loss": 0.7455,
      "step": 18430
    },
    {
      "epoch": 0.5310609116579266,
      "grad_norm": 0.7072800993919373,
      "learning_rate": 4.740215184905042e-06,
      "loss": 0.7039,
      "step": 18431
    },
    {
      "epoch": 0.5310897251195759,
      "grad_norm": 0.7496446967124939,
      "learning_rate": 4.739749206150067e-06,
      "loss": 0.7067,
      "step": 18432
    },
    {
      "epoch": 0.5311185385812252,
      "grad_norm": 0.7450984120368958,
      "learning_rate": 4.739283229661617e-06,
      "loss": 0.7072,
      "step": 18433
    },
    {
      "epoch": 0.5311473520428744,
      "grad_norm": 0.7359236478805542,
      "learning_rate": 4.7388172554437486e-06,
      "loss": 0.729,
      "step": 18434
    },
    {
      "epoch": 0.5311761655045237,
      "grad_norm": 0.7352598309516907,
      "learning_rate": 4.738351283500522e-06,
      "loss": 0.6999,
      "step": 18435
    },
    {
      "epoch": 0.531204978966173,
      "grad_norm": 0.7343339323997498,
      "learning_rate": 4.737885313835992e-06,
      "loss": 0.7301,
      "step": 18436
    },
    {
      "epoch": 0.5312337924278223,
      "grad_norm": 0.7240599989891052,
      "learning_rate": 4.73741934645422e-06,
      "loss": 0.6848,
      "step": 18437
    },
    {
      "epoch": 0.5312626058894716,
      "grad_norm": 0.7343503832817078,
      "learning_rate": 4.736953381359263e-06,
      "loss": 0.7139,
      "step": 18438
    },
    {
      "epoch": 0.5312914193511209,
      "grad_norm": 0.7202447652816772,
      "learning_rate": 4.73648741855518e-06,
      "loss": 0.6921,
      "step": 18439
    },
    {
      "epoch": 0.5313202328127701,
      "grad_norm": 0.7559550404548645,
      "learning_rate": 4.7360214580460275e-06,
      "loss": 0.7079,
      "step": 18440
    },
    {
      "epoch": 0.5313490462744194,
      "grad_norm": 0.7347046732902527,
      "learning_rate": 4.7355554998358645e-06,
      "loss": 0.7167,
      "step": 18441
    },
    {
      "epoch": 0.5313778597360687,
      "grad_norm": 0.7174941301345825,
      "learning_rate": 4.735089543928747e-06,
      "loss": 0.7014,
      "step": 18442
    },
    {
      "epoch": 0.531406673197718,
      "grad_norm": 0.7372229695320129,
      "learning_rate": 4.734623590328736e-06,
      "loss": 0.7361,
      "step": 18443
    },
    {
      "epoch": 0.5314354866593672,
      "grad_norm": 0.7260311841964722,
      "learning_rate": 4.734157639039888e-06,
      "loss": 0.7237,
      "step": 18444
    },
    {
      "epoch": 0.5314643001210165,
      "grad_norm": 0.7234587669372559,
      "learning_rate": 4.733691690066262e-06,
      "loss": 0.716,
      "step": 18445
    },
    {
      "epoch": 0.5314931135826658,
      "grad_norm": 0.7214639782905579,
      "learning_rate": 4.733225743411915e-06,
      "loss": 0.7201,
      "step": 18446
    },
    {
      "epoch": 0.5315219270443151,
      "grad_norm": 0.7487252354621887,
      "learning_rate": 4.7327597990809046e-06,
      "loss": 0.7134,
      "step": 18447
    },
    {
      "epoch": 0.5315507405059644,
      "grad_norm": 0.7235918045043945,
      "learning_rate": 4.73229385707729e-06,
      "loss": 0.7197,
      "step": 18448
    },
    {
      "epoch": 0.5315795539676137,
      "grad_norm": 0.7365262508392334,
      "learning_rate": 4.731827917405128e-06,
      "loss": 0.747,
      "step": 18449
    },
    {
      "epoch": 0.531608367429263,
      "grad_norm": 0.6957788467407227,
      "learning_rate": 4.7313619800684765e-06,
      "loss": 0.6838,
      "step": 18450
    },
    {
      "epoch": 0.5316371808909123,
      "grad_norm": 0.750869870185852,
      "learning_rate": 4.730896045071394e-06,
      "loss": 0.7292,
      "step": 18451
    },
    {
      "epoch": 0.5316659943525616,
      "grad_norm": 0.7528440952301025,
      "learning_rate": 4.730430112417937e-06,
      "loss": 0.7289,
      "step": 18452
    },
    {
      "epoch": 0.5316948078142107,
      "grad_norm": 0.7491815686225891,
      "learning_rate": 4.7299641821121644e-06,
      "loss": 0.6961,
      "step": 18453
    },
    {
      "epoch": 0.53172362127586,
      "grad_norm": 0.7079494595527649,
      "learning_rate": 4.7294982541581334e-06,
      "loss": 0.7072,
      "step": 18454
    },
    {
      "epoch": 0.5317524347375093,
      "grad_norm": 0.6854852437973022,
      "learning_rate": 4.7290323285599035e-06,
      "loss": 0.6886,
      "step": 18455
    },
    {
      "epoch": 0.5317812481991586,
      "grad_norm": 0.7250241637229919,
      "learning_rate": 4.728566405321531e-06,
      "loss": 0.6979,
      "step": 18456
    },
    {
      "epoch": 0.5318100616608079,
      "grad_norm": 0.7505607008934021,
      "learning_rate": 4.728100484447073e-06,
      "loss": 0.7133,
      "step": 18457
    },
    {
      "epoch": 0.5318388751224572,
      "grad_norm": 0.7135950326919556,
      "learning_rate": 4.727634565940588e-06,
      "loss": 0.7056,
      "step": 18458
    },
    {
      "epoch": 0.5318676885841065,
      "grad_norm": 0.7252761125564575,
      "learning_rate": 4.727168649806134e-06,
      "loss": 0.7275,
      "step": 18459
    },
    {
      "epoch": 0.5318965020457558,
      "grad_norm": 0.7313625812530518,
      "learning_rate": 4.726702736047769e-06,
      "loss": 0.728,
      "step": 18460
    },
    {
      "epoch": 0.5319253155074051,
      "grad_norm": 0.7406079173088074,
      "learning_rate": 4.726236824669549e-06,
      "loss": 0.7096,
      "step": 18461
    },
    {
      "epoch": 0.5319541289690544,
      "grad_norm": 0.7356693744659424,
      "learning_rate": 4.7257709156755335e-06,
      "loss": 0.7359,
      "step": 18462
    },
    {
      "epoch": 0.5319829424307037,
      "grad_norm": 0.7315697073936462,
      "learning_rate": 4.72530500906978e-06,
      "loss": 0.7299,
      "step": 18463
    },
    {
      "epoch": 0.5320117558923529,
      "grad_norm": 0.7158712148666382,
      "learning_rate": 4.724839104856343e-06,
      "loss": 0.6997,
      "step": 18464
    },
    {
      "epoch": 0.5320405693540022,
      "grad_norm": 0.7220976948738098,
      "learning_rate": 4.724373203039284e-06,
      "loss": 0.6963,
      "step": 18465
    },
    {
      "epoch": 0.5320693828156515,
      "grad_norm": 0.7505581974983215,
      "learning_rate": 4.723907303622656e-06,
      "loss": 0.7255,
      "step": 18466
    },
    {
      "epoch": 0.5320981962773007,
      "grad_norm": 0.7470589280128479,
      "learning_rate": 4.723441406610522e-06,
      "loss": 0.7362,
      "step": 18467
    },
    {
      "epoch": 0.53212700973895,
      "grad_norm": 0.7024797797203064,
      "learning_rate": 4.722975512006937e-06,
      "loss": 0.6943,
      "step": 18468
    },
    {
      "epoch": 0.5321558232005993,
      "grad_norm": 0.7044962644577026,
      "learning_rate": 4.722509619815957e-06,
      "loss": 0.6942,
      "step": 18469
    },
    {
      "epoch": 0.5321846366622486,
      "grad_norm": 0.7025989890098572,
      "learning_rate": 4.7220437300416426e-06,
      "loss": 0.6893,
      "step": 18470
    },
    {
      "epoch": 0.5322134501238979,
      "grad_norm": 0.7554714679718018,
      "learning_rate": 4.721577842688049e-06,
      "loss": 0.7283,
      "step": 18471
    },
    {
      "epoch": 0.5322422635855472,
      "grad_norm": 0.7064815759658813,
      "learning_rate": 4.721111957759233e-06,
      "loss": 0.6923,
      "step": 18472
    },
    {
      "epoch": 0.5322710770471965,
      "grad_norm": 0.7104281783103943,
      "learning_rate": 4.720646075259253e-06,
      "loss": 0.7154,
      "step": 18473
    },
    {
      "epoch": 0.5322998905088457,
      "grad_norm": 0.7191166877746582,
      "learning_rate": 4.720180195192168e-06,
      "loss": 0.7348,
      "step": 18474
    },
    {
      "epoch": 0.532328703970495,
      "grad_norm": 4.258472919464111,
      "learning_rate": 4.719714317562033e-06,
      "loss": 0.7657,
      "step": 18475
    },
    {
      "epoch": 0.5323575174321443,
      "grad_norm": 0.7301589250564575,
      "learning_rate": 4.719248442372906e-06,
      "loss": 0.7095,
      "step": 18476
    },
    {
      "epoch": 0.5323863308937936,
      "grad_norm": 0.7230568528175354,
      "learning_rate": 4.718782569628845e-06,
      "loss": 0.7024,
      "step": 18477
    },
    {
      "epoch": 0.5324151443554429,
      "grad_norm": 0.6918531060218811,
      "learning_rate": 4.718316699333904e-06,
      "loss": 0.7002,
      "step": 18478
    },
    {
      "epoch": 0.5324439578170922,
      "grad_norm": 0.7107200026512146,
      "learning_rate": 4.717850831492145e-06,
      "loss": 0.6841,
      "step": 18479
    },
    {
      "epoch": 0.5324727712787414,
      "grad_norm": 0.7232306003570557,
      "learning_rate": 4.717384966107623e-06,
      "loss": 0.7138,
      "step": 18480
    },
    {
      "epoch": 0.5325015847403907,
      "grad_norm": 0.7307479977607727,
      "learning_rate": 4.716919103184397e-06,
      "loss": 0.7178,
      "step": 18481
    },
    {
      "epoch": 0.53253039820204,
      "grad_norm": 0.7090417742729187,
      "learning_rate": 4.7164532427265206e-06,
      "loss": 0.7212,
      "step": 18482
    },
    {
      "epoch": 0.5325592116636892,
      "grad_norm": 0.7366560697555542,
      "learning_rate": 4.715987384738055e-06,
      "loss": 0.7101,
      "step": 18483
    },
    {
      "epoch": 0.5325880251253385,
      "grad_norm": 0.7658687233924866,
      "learning_rate": 4.715521529223054e-06,
      "loss": 0.7035,
      "step": 18484
    },
    {
      "epoch": 0.5326168385869878,
      "grad_norm": 0.7307614088058472,
      "learning_rate": 4.715055676185576e-06,
      "loss": 0.7207,
      "step": 18485
    },
    {
      "epoch": 0.5326456520486371,
      "grad_norm": 0.7435385584831238,
      "learning_rate": 4.714589825629679e-06,
      "loss": 0.7196,
      "step": 18486
    },
    {
      "epoch": 0.5326744655102864,
      "grad_norm": 0.7280862927436829,
      "learning_rate": 4.714123977559419e-06,
      "loss": 0.7112,
      "step": 18487
    },
    {
      "epoch": 0.5327032789719357,
      "grad_norm": 0.7357221245765686,
      "learning_rate": 4.7136581319788534e-06,
      "loss": 0.7285,
      "step": 18488
    },
    {
      "epoch": 0.532732092433585,
      "grad_norm": 0.7091528177261353,
      "learning_rate": 4.713192288892039e-06,
      "loss": 0.7128,
      "step": 18489
    },
    {
      "epoch": 0.5327609058952343,
      "grad_norm": 0.7137035727500916,
      "learning_rate": 4.712726448303031e-06,
      "loss": 0.7254,
      "step": 18490
    },
    {
      "epoch": 0.5327897193568836,
      "grad_norm": 0.7122716307640076,
      "learning_rate": 4.712260610215891e-06,
      "loss": 0.7053,
      "step": 18491
    },
    {
      "epoch": 0.5328185328185329,
      "grad_norm": 0.7190842628479004,
      "learning_rate": 4.7117947746346736e-06,
      "loss": 0.7137,
      "step": 18492
    },
    {
      "epoch": 0.5328473462801822,
      "grad_norm": 0.7298935651779175,
      "learning_rate": 4.711328941563434e-06,
      "loss": 0.7141,
      "step": 18493
    },
    {
      "epoch": 0.5328761597418313,
      "grad_norm": 0.7385745048522949,
      "learning_rate": 4.710863111006232e-06,
      "loss": 0.7236,
      "step": 18494
    },
    {
      "epoch": 0.5329049732034806,
      "grad_norm": 0.6811279654502869,
      "learning_rate": 4.710397282967123e-06,
      "loss": 0.6935,
      "step": 18495
    },
    {
      "epoch": 0.5329337866651299,
      "grad_norm": 0.7277519106864929,
      "learning_rate": 4.709931457450165e-06,
      "loss": 0.6991,
      "step": 18496
    },
    {
      "epoch": 0.5329626001267792,
      "grad_norm": 0.7402806878089905,
      "learning_rate": 4.709465634459413e-06,
      "loss": 0.6986,
      "step": 18497
    },
    {
      "epoch": 0.5329914135884285,
      "grad_norm": 0.7169250249862671,
      "learning_rate": 4.708999813998925e-06,
      "loss": 0.6935,
      "step": 18498
    },
    {
      "epoch": 0.5330202270500778,
      "grad_norm": 0.7122630476951599,
      "learning_rate": 4.7085339960727576e-06,
      "loss": 0.6963,
      "step": 18499
    },
    {
      "epoch": 0.5330490405117271,
      "grad_norm": 0.7169521450996399,
      "learning_rate": 4.7080681806849685e-06,
      "loss": 0.7301,
      "step": 18500
    },
    {
      "epoch": 0.5330778539733764,
      "grad_norm": 0.736794650554657,
      "learning_rate": 4.707602367839613e-06,
      "loss": 0.7253,
      "step": 18501
    },
    {
      "epoch": 0.5331066674350257,
      "grad_norm": 0.7174801826477051,
      "learning_rate": 4.707136557540746e-06,
      "loss": 0.7156,
      "step": 18502
    },
    {
      "epoch": 0.533135480896675,
      "grad_norm": 0.7299475073814392,
      "learning_rate": 4.7066707497924305e-06,
      "loss": 0.6843,
      "step": 18503
    },
    {
      "epoch": 0.5331642943583242,
      "grad_norm": 0.7385449409484863,
      "learning_rate": 4.706204944598719e-06,
      "loss": 0.7097,
      "step": 18504
    },
    {
      "epoch": 0.5331931078199735,
      "grad_norm": 0.7141687273979187,
      "learning_rate": 4.705739141963669e-06,
      "loss": 0.7167,
      "step": 18505
    },
    {
      "epoch": 0.5332219212816228,
      "grad_norm": 0.7314359545707703,
      "learning_rate": 4.705273341891336e-06,
      "loss": 0.7116,
      "step": 18506
    },
    {
      "epoch": 0.533250734743272,
      "grad_norm": 0.6971699595451355,
      "learning_rate": 4.704807544385779e-06,
      "loss": 0.6902,
      "step": 18507
    },
    {
      "epoch": 0.5332795482049213,
      "grad_norm": 0.7448853254318237,
      "learning_rate": 4.7043417494510515e-06,
      "loss": 0.7391,
      "step": 18508
    },
    {
      "epoch": 0.5333083616665706,
      "grad_norm": 0.7197855114936829,
      "learning_rate": 4.703875957091213e-06,
      "loss": 0.6936,
      "step": 18509
    },
    {
      "epoch": 0.5333371751282199,
      "grad_norm": 0.7106298804283142,
      "learning_rate": 4.703410167310318e-06,
      "loss": 0.7106,
      "step": 18510
    },
    {
      "epoch": 0.5333659885898692,
      "grad_norm": 0.7106085419654846,
      "learning_rate": 4.702944380112424e-06,
      "loss": 0.6785,
      "step": 18511
    },
    {
      "epoch": 0.5333948020515185,
      "grad_norm": 0.6874844431877136,
      "learning_rate": 4.702478595501588e-06,
      "loss": 0.7087,
      "step": 18512
    },
    {
      "epoch": 0.5334236155131677,
      "grad_norm": 0.7446655035018921,
      "learning_rate": 4.702012813481866e-06,
      "loss": 0.6988,
      "step": 18513
    },
    {
      "epoch": 0.533452428974817,
      "grad_norm": 0.7623725533485413,
      "learning_rate": 4.701547034057314e-06,
      "loss": 0.7137,
      "step": 18514
    },
    {
      "epoch": 0.5334812424364663,
      "grad_norm": 0.7307922840118408,
      "learning_rate": 4.701081257231989e-06,
      "loss": 0.6993,
      "step": 18515
    },
    {
      "epoch": 0.5335100558981156,
      "grad_norm": 0.741798996925354,
      "learning_rate": 4.700615483009947e-06,
      "loss": 0.7146,
      "step": 18516
    },
    {
      "epoch": 0.5335388693597649,
      "grad_norm": 0.7501067519187927,
      "learning_rate": 4.700149711395246e-06,
      "loss": 0.696,
      "step": 18517
    },
    {
      "epoch": 0.5335676828214142,
      "grad_norm": 0.7383867502212524,
      "learning_rate": 4.69968394239194e-06,
      "loss": 0.701,
      "step": 18518
    },
    {
      "epoch": 0.5335964962830635,
      "grad_norm": 0.7413197159767151,
      "learning_rate": 4.699218176004088e-06,
      "loss": 0.718,
      "step": 18519
    },
    {
      "epoch": 0.5336253097447128,
      "grad_norm": 0.7402194738388062,
      "learning_rate": 4.698752412235745e-06,
      "loss": 0.7194,
      "step": 18520
    },
    {
      "epoch": 0.533654123206362,
      "grad_norm": 0.7515285611152649,
      "learning_rate": 4.6982866510909675e-06,
      "loss": 0.6786,
      "step": 18521
    },
    {
      "epoch": 0.5336829366680113,
      "grad_norm": 0.7326042056083679,
      "learning_rate": 4.697820892573811e-06,
      "loss": 0.6914,
      "step": 18522
    },
    {
      "epoch": 0.5337117501296605,
      "grad_norm": 0.721067488193512,
      "learning_rate": 4.697355136688333e-06,
      "loss": 0.7324,
      "step": 18523
    },
    {
      "epoch": 0.5337405635913098,
      "grad_norm": 0.728826105594635,
      "learning_rate": 4.696889383438589e-06,
      "loss": 0.7055,
      "step": 18524
    },
    {
      "epoch": 0.5337693770529591,
      "grad_norm": 0.7251717448234558,
      "learning_rate": 4.696423632828635e-06,
      "loss": 0.7165,
      "step": 18525
    },
    {
      "epoch": 0.5337981905146084,
      "grad_norm": 0.7315953373908997,
      "learning_rate": 4.695957884862527e-06,
      "loss": 0.716,
      "step": 18526
    },
    {
      "epoch": 0.5338270039762577,
      "grad_norm": 0.7572469115257263,
      "learning_rate": 4.695492139544322e-06,
      "loss": 0.737,
      "step": 18527
    },
    {
      "epoch": 0.533855817437907,
      "grad_norm": 0.7449109554290771,
      "learning_rate": 4.695026396878077e-06,
      "loss": 0.694,
      "step": 18528
    },
    {
      "epoch": 0.5338846308995563,
      "grad_norm": 0.7244868874549866,
      "learning_rate": 4.6945606568678474e-06,
      "loss": 0.7228,
      "step": 18529
    },
    {
      "epoch": 0.5339134443612056,
      "grad_norm": 0.7165141701698303,
      "learning_rate": 4.694094919517689e-06,
      "loss": 0.7122,
      "step": 18530
    },
    {
      "epoch": 0.5339422578228549,
      "grad_norm": 0.7618873119354248,
      "learning_rate": 4.693629184831659e-06,
      "loss": 0.7194,
      "step": 18531
    },
    {
      "epoch": 0.5339710712845042,
      "grad_norm": 0.7327884435653687,
      "learning_rate": 4.693163452813811e-06,
      "loss": 0.6986,
      "step": 18532
    },
    {
      "epoch": 0.5339998847461535,
      "grad_norm": 0.7316531538963318,
      "learning_rate": 4.692697723468203e-06,
      "loss": 0.6936,
      "step": 18533
    },
    {
      "epoch": 0.5340286982078026,
      "grad_norm": 0.7282025218009949,
      "learning_rate": 4.69223199679889e-06,
      "loss": 0.7186,
      "step": 18534
    },
    {
      "epoch": 0.5340575116694519,
      "grad_norm": 3.9183990955352783,
      "learning_rate": 4.69176627280993e-06,
      "loss": 0.7074,
      "step": 18535
    },
    {
      "epoch": 0.5340863251311012,
      "grad_norm": 0.7267693877220154,
      "learning_rate": 4.6913005515053766e-06,
      "loss": 0.7222,
      "step": 18536
    },
    {
      "epoch": 0.5341151385927505,
      "grad_norm": 0.7271654009819031,
      "learning_rate": 4.690834832889287e-06,
      "loss": 0.7034,
      "step": 18537
    },
    {
      "epoch": 0.5341439520543998,
      "grad_norm": 0.7524479627609253,
      "learning_rate": 4.690369116965716e-06,
      "loss": 0.71,
      "step": 18538
    },
    {
      "epoch": 0.5341727655160491,
      "grad_norm": 0.772138237953186,
      "learning_rate": 4.6899034037387215e-06,
      "loss": 0.7048,
      "step": 18539
    },
    {
      "epoch": 0.5342015789776984,
      "grad_norm": 0.7267869114875793,
      "learning_rate": 4.6894376932123586e-06,
      "loss": 0.7214,
      "step": 18540
    },
    {
      "epoch": 0.5342303924393477,
      "grad_norm": 0.7175214290618896,
      "learning_rate": 4.688971985390684e-06,
      "loss": 0.7001,
      "step": 18541
    },
    {
      "epoch": 0.534259205900997,
      "grad_norm": 0.7469406723976135,
      "learning_rate": 4.6885062802777506e-06,
      "loss": 0.7351,
      "step": 18542
    },
    {
      "epoch": 0.5342880193626462,
      "grad_norm": 0.7323360443115234,
      "learning_rate": 4.688040577877618e-06,
      "loss": 0.7259,
      "step": 18543
    },
    {
      "epoch": 0.5343168328242955,
      "grad_norm": 0.7507705092430115,
      "learning_rate": 4.6875748781943385e-06,
      "loss": 0.7328,
      "step": 18544
    },
    {
      "epoch": 0.5343456462859448,
      "grad_norm": 0.7387580871582031,
      "learning_rate": 4.68710918123197e-06,
      "loss": 0.7203,
      "step": 18545
    },
    {
      "epoch": 0.5343744597475941,
      "grad_norm": 0.7284646034240723,
      "learning_rate": 4.686643486994568e-06,
      "loss": 0.683,
      "step": 18546
    },
    {
      "epoch": 0.5344032732092434,
      "grad_norm": 0.7306482195854187,
      "learning_rate": 4.686177795486188e-06,
      "loss": 0.7114,
      "step": 18547
    },
    {
      "epoch": 0.5344320866708926,
      "grad_norm": 0.732922375202179,
      "learning_rate": 4.6857121067108855e-06,
      "loss": 0.7243,
      "step": 18548
    },
    {
      "epoch": 0.5344609001325419,
      "grad_norm": 0.7340797185897827,
      "learning_rate": 4.685246420672715e-06,
      "loss": 0.7192,
      "step": 18549
    },
    {
      "epoch": 0.5344897135941912,
      "grad_norm": 0.8451446294784546,
      "learning_rate": 4.684780737375735e-06,
      "loss": 0.7552,
      "step": 18550
    },
    {
      "epoch": 0.5345185270558405,
      "grad_norm": 0.7413761019706726,
      "learning_rate": 4.684315056823999e-06,
      "loss": 0.7123,
      "step": 18551
    },
    {
      "epoch": 0.5345473405174898,
      "grad_norm": 0.7663756012916565,
      "learning_rate": 4.683849379021564e-06,
      "loss": 0.7313,
      "step": 18552
    },
    {
      "epoch": 0.534576153979139,
      "grad_norm": 0.7106480598449707,
      "learning_rate": 4.683383703972485e-06,
      "loss": 0.6857,
      "step": 18553
    },
    {
      "epoch": 0.5346049674407883,
      "grad_norm": 0.7589061260223389,
      "learning_rate": 4.682918031680817e-06,
      "loss": 0.705,
      "step": 18554
    },
    {
      "epoch": 0.5346337809024376,
      "grad_norm": 0.7198286056518555,
      "learning_rate": 4.6824523621506154e-06,
      "loss": 0.6977,
      "step": 18555
    },
    {
      "epoch": 0.5346625943640869,
      "grad_norm": 0.7550093531608582,
      "learning_rate": 4.681986695385937e-06,
      "loss": 0.7274,
      "step": 18556
    },
    {
      "epoch": 0.5346914078257362,
      "grad_norm": 0.72003173828125,
      "learning_rate": 4.681521031390837e-06,
      "loss": 0.7082,
      "step": 18557
    },
    {
      "epoch": 0.5347202212873855,
      "grad_norm": 0.7010286450386047,
      "learning_rate": 4.681055370169371e-06,
      "loss": 0.7089,
      "step": 18558
    },
    {
      "epoch": 0.5347490347490348,
      "grad_norm": 0.7563803791999817,
      "learning_rate": 4.680589711725592e-06,
      "loss": 0.7153,
      "step": 18559
    },
    {
      "epoch": 0.5347778482106841,
      "grad_norm": 0.7260012626647949,
      "learning_rate": 4.680124056063559e-06,
      "loss": 0.7091,
      "step": 18560
    },
    {
      "epoch": 0.5348066616723333,
      "grad_norm": 0.7442737817764282,
      "learning_rate": 4.679658403187325e-06,
      "loss": 0.7162,
      "step": 18561
    },
    {
      "epoch": 0.5348354751339826,
      "grad_norm": 0.7360440492630005,
      "learning_rate": 4.679192753100944e-06,
      "loss": 0.7226,
      "step": 18562
    },
    {
      "epoch": 0.5348642885956318,
      "grad_norm": 0.7294104695320129,
      "learning_rate": 4.678727105808476e-06,
      "loss": 0.7204,
      "step": 18563
    },
    {
      "epoch": 0.5348931020572811,
      "grad_norm": 0.7463657855987549,
      "learning_rate": 4.6782614613139735e-06,
      "loss": 0.6945,
      "step": 18564
    },
    {
      "epoch": 0.5349219155189304,
      "grad_norm": 0.7534452080726624,
      "learning_rate": 4.6777958196214915e-06,
      "loss": 0.735,
      "step": 18565
    },
    {
      "epoch": 0.5349507289805797,
      "grad_norm": 0.7205606698989868,
      "learning_rate": 4.677330180735086e-06,
      "loss": 0.7208,
      "step": 18566
    },
    {
      "epoch": 0.534979542442229,
      "grad_norm": 0.7116948962211609,
      "learning_rate": 4.676864544658813e-06,
      "loss": 0.6888,
      "step": 18567
    },
    {
      "epoch": 0.5350083559038783,
      "grad_norm": 0.7755774855613708,
      "learning_rate": 4.676398911396726e-06,
      "loss": 0.7389,
      "step": 18568
    },
    {
      "epoch": 0.5350371693655276,
      "grad_norm": 0.7060654163360596,
      "learning_rate": 4.675933280952882e-06,
      "loss": 0.6801,
      "step": 18569
    },
    {
      "epoch": 0.5350659828271769,
      "grad_norm": 0.7258759140968323,
      "learning_rate": 4.675467653331334e-06,
      "loss": 0.7104,
      "step": 18570
    },
    {
      "epoch": 0.5350947962888262,
      "grad_norm": 0.7028312087059021,
      "learning_rate": 4.6750020285361366e-06,
      "loss": 0.672,
      "step": 18571
    },
    {
      "epoch": 0.5351236097504755,
      "grad_norm": 0.7435569763183594,
      "learning_rate": 4.674536406571349e-06,
      "loss": 0.7176,
      "step": 18572
    },
    {
      "epoch": 0.5351524232121248,
      "grad_norm": 0.7306133508682251,
      "learning_rate": 4.674070787441024e-06,
      "loss": 0.7139,
      "step": 18573
    },
    {
      "epoch": 0.535181236673774,
      "grad_norm": 0.7487604022026062,
      "learning_rate": 4.673605171149215e-06,
      "loss": 0.7147,
      "step": 18574
    },
    {
      "epoch": 0.5352100501354232,
      "grad_norm": 0.7313891649246216,
      "learning_rate": 4.67313955769998e-06,
      "loss": 0.7146,
      "step": 18575
    },
    {
      "epoch": 0.5352388635970725,
      "grad_norm": 0.756513237953186,
      "learning_rate": 4.672673947097371e-06,
      "loss": 0.7238,
      "step": 18576
    },
    {
      "epoch": 0.5352676770587218,
      "grad_norm": 0.7510395050048828,
      "learning_rate": 4.672208339345447e-06,
      "loss": 0.7099,
      "step": 18577
    },
    {
      "epoch": 0.5352964905203711,
      "grad_norm": 0.7823551893234253,
      "learning_rate": 4.67174273444826e-06,
      "loss": 0.7323,
      "step": 18578
    },
    {
      "epoch": 0.5353253039820204,
      "grad_norm": 0.7281982898712158,
      "learning_rate": 4.671277132409866e-06,
      "loss": 0.7193,
      "step": 18579
    },
    {
      "epoch": 0.5353541174436697,
      "grad_norm": 0.7630137205123901,
      "learning_rate": 4.6708115332343194e-06,
      "loss": 0.7012,
      "step": 18580
    },
    {
      "epoch": 0.535382930905319,
      "grad_norm": 0.7531135082244873,
      "learning_rate": 4.670345936925675e-06,
      "loss": 0.7392,
      "step": 18581
    },
    {
      "epoch": 0.5354117443669683,
      "grad_norm": 0.7337612509727478,
      "learning_rate": 4.669880343487988e-06,
      "loss": 0.6907,
      "step": 18582
    },
    {
      "epoch": 0.5354405578286175,
      "grad_norm": 0.7167478203773499,
      "learning_rate": 4.669414752925314e-06,
      "loss": 0.7028,
      "step": 18583
    },
    {
      "epoch": 0.5354693712902668,
      "grad_norm": 0.7085064649581909,
      "learning_rate": 4.668949165241706e-06,
      "loss": 0.7124,
      "step": 18584
    },
    {
      "epoch": 0.5354981847519161,
      "grad_norm": 0.7232205867767334,
      "learning_rate": 4.668483580441221e-06,
      "loss": 0.6943,
      "step": 18585
    },
    {
      "epoch": 0.5355269982135654,
      "grad_norm": 0.707974910736084,
      "learning_rate": 4.668017998527909e-06,
      "loss": 0.7201,
      "step": 18586
    },
    {
      "epoch": 0.5355558116752147,
      "grad_norm": 0.7301604747772217,
      "learning_rate": 4.667552419505831e-06,
      "loss": 0.7095,
      "step": 18587
    },
    {
      "epoch": 0.5355846251368639,
      "grad_norm": 0.7625668048858643,
      "learning_rate": 4.667086843379039e-06,
      "loss": 0.6991,
      "step": 18588
    },
    {
      "epoch": 0.5356134385985132,
      "grad_norm": 0.7098389863967896,
      "learning_rate": 4.666621270151588e-06,
      "loss": 0.7247,
      "step": 18589
    },
    {
      "epoch": 0.5356422520601625,
      "grad_norm": 0.7224364280700684,
      "learning_rate": 4.666155699827533e-06,
      "loss": 0.7002,
      "step": 18590
    },
    {
      "epoch": 0.5356710655218118,
      "grad_norm": 0.7218223214149475,
      "learning_rate": 4.665690132410928e-06,
      "loss": 0.7082,
      "step": 18591
    },
    {
      "epoch": 0.535699878983461,
      "grad_norm": 0.7074489593505859,
      "learning_rate": 4.665224567905826e-06,
      "loss": 0.7099,
      "step": 18592
    },
    {
      "epoch": 0.5357286924451103,
      "grad_norm": 0.7196916937828064,
      "learning_rate": 4.664759006316285e-06,
      "loss": 0.7367,
      "step": 18593
    },
    {
      "epoch": 0.5357575059067596,
      "grad_norm": 0.684415340423584,
      "learning_rate": 4.664293447646358e-06,
      "loss": 0.6987,
      "step": 18594
    },
    {
      "epoch": 0.5357863193684089,
      "grad_norm": 0.721905529499054,
      "learning_rate": 4.663827891900099e-06,
      "loss": 0.6926,
      "step": 18595
    },
    {
      "epoch": 0.5358151328300582,
      "grad_norm": 0.7407404184341431,
      "learning_rate": 4.663362339081563e-06,
      "loss": 0.7068,
      "step": 18596
    },
    {
      "epoch": 0.5358439462917075,
      "grad_norm": 0.7486644387245178,
      "learning_rate": 4.662896789194805e-06,
      "loss": 0.7146,
      "step": 18597
    },
    {
      "epoch": 0.5358727597533568,
      "grad_norm": 0.7076426148414612,
      "learning_rate": 4.662431242243877e-06,
      "loss": 0.7155,
      "step": 18598
    },
    {
      "epoch": 0.5359015732150061,
      "grad_norm": 0.7219653129577637,
      "learning_rate": 4.661965698232837e-06,
      "loss": 0.7176,
      "step": 18599
    },
    {
      "epoch": 0.5359303866766554,
      "grad_norm": 0.7267800569534302,
      "learning_rate": 4.661500157165738e-06,
      "loss": 0.7215,
      "step": 18600
    },
    {
      "epoch": 0.5359592001383047,
      "grad_norm": 0.7379506230354309,
      "learning_rate": 4.6610346190466345e-06,
      "loss": 0.734,
      "step": 18601
    },
    {
      "epoch": 0.5359880135999538,
      "grad_norm": 0.7265921235084534,
      "learning_rate": 4.660569083879581e-06,
      "loss": 0.7171,
      "step": 18602
    },
    {
      "epoch": 0.5360168270616031,
      "grad_norm": 0.729578971862793,
      "learning_rate": 4.66010355166863e-06,
      "loss": 0.7196,
      "step": 18603
    },
    {
      "epoch": 0.5360456405232524,
      "grad_norm": 0.7484079599380493,
      "learning_rate": 4.659638022417839e-06,
      "loss": 0.7081,
      "step": 18604
    },
    {
      "epoch": 0.5360744539849017,
      "grad_norm": 0.707859992980957,
      "learning_rate": 4.659172496131259e-06,
      "loss": 0.6799,
      "step": 18605
    },
    {
      "epoch": 0.536103267446551,
      "grad_norm": 0.7677052617073059,
      "learning_rate": 4.658706972812948e-06,
      "loss": 0.7241,
      "step": 18606
    },
    {
      "epoch": 0.5361320809082003,
      "grad_norm": 0.7264010310173035,
      "learning_rate": 4.658241452466956e-06,
      "loss": 0.7147,
      "step": 18607
    },
    {
      "epoch": 0.5361608943698496,
      "grad_norm": 0.7304062247276306,
      "learning_rate": 4.65777593509734e-06,
      "loss": 0.6962,
      "step": 18608
    },
    {
      "epoch": 0.5361897078314989,
      "grad_norm": 0.7233988642692566,
      "learning_rate": 4.657310420708153e-06,
      "loss": 0.7242,
      "step": 18609
    },
    {
      "epoch": 0.5362185212931482,
      "grad_norm": 1.5066783428192139,
      "learning_rate": 4.656844909303449e-06,
      "loss": 0.7075,
      "step": 18610
    },
    {
      "epoch": 0.5362473347547975,
      "grad_norm": 0.7599621415138245,
      "learning_rate": 4.656379400887285e-06,
      "loss": 0.7168,
      "step": 18611
    },
    {
      "epoch": 0.5362761482164468,
      "grad_norm": 0.7218727469444275,
      "learning_rate": 4.6559138954637114e-06,
      "loss": 0.6797,
      "step": 18612
    },
    {
      "epoch": 0.536304961678096,
      "grad_norm": 0.7115626931190491,
      "learning_rate": 4.655448393036783e-06,
      "loss": 0.7027,
      "step": 18613
    },
    {
      "epoch": 0.5363337751397453,
      "grad_norm": 0.7304872870445251,
      "learning_rate": 4.654982893610557e-06,
      "loss": 0.6985,
      "step": 18614
    },
    {
      "epoch": 0.5363625886013945,
      "grad_norm": 0.7358362674713135,
      "learning_rate": 4.6545173971890836e-06,
      "loss": 0.7226,
      "step": 18615
    },
    {
      "epoch": 0.5363914020630438,
      "grad_norm": 0.7306164503097534,
      "learning_rate": 4.654051903776419e-06,
      "loss": 0.7174,
      "step": 18616
    },
    {
      "epoch": 0.5364202155246931,
      "grad_norm": 0.7586563229560852,
      "learning_rate": 4.653586413376616e-06,
      "loss": 0.7371,
      "step": 18617
    },
    {
      "epoch": 0.5364490289863424,
      "grad_norm": 0.7615123987197876,
      "learning_rate": 4.65312092599373e-06,
      "loss": 0.7168,
      "step": 18618
    },
    {
      "epoch": 0.5364778424479917,
      "grad_norm": 0.7184858918190002,
      "learning_rate": 4.652655441631813e-06,
      "loss": 0.7165,
      "step": 18619
    },
    {
      "epoch": 0.536506655909641,
      "grad_norm": 0.7381948828697205,
      "learning_rate": 4.652189960294921e-06,
      "loss": 0.7078,
      "step": 18620
    },
    {
      "epoch": 0.5365354693712903,
      "grad_norm": 0.7442135214805603,
      "learning_rate": 4.651724481987105e-06,
      "loss": 0.7119,
      "step": 18621
    },
    {
      "epoch": 0.5365642828329396,
      "grad_norm": 0.7359150052070618,
      "learning_rate": 4.6512590067124195e-06,
      "loss": 0.7089,
      "step": 18622
    },
    {
      "epoch": 0.5365930962945888,
      "grad_norm": 0.7227004170417786,
      "learning_rate": 4.650793534474922e-06,
      "loss": 0.7046,
      "step": 18623
    },
    {
      "epoch": 0.5366219097562381,
      "grad_norm": 0.7591548562049866,
      "learning_rate": 4.6503280652786635e-06,
      "loss": 0.7325,
      "step": 18624
    },
    {
      "epoch": 0.5366507232178874,
      "grad_norm": 0.748024582862854,
      "learning_rate": 4.649862599127698e-06,
      "loss": 0.7158,
      "step": 18625
    },
    {
      "epoch": 0.5366795366795367,
      "grad_norm": 0.7025356888771057,
      "learning_rate": 4.649397136026079e-06,
      "loss": 0.6841,
      "step": 18626
    },
    {
      "epoch": 0.536708350141186,
      "grad_norm": 0.7309522032737732,
      "learning_rate": 4.648931675977861e-06,
      "loss": 0.7184,
      "step": 18627
    },
    {
      "epoch": 0.5367371636028353,
      "grad_norm": 0.7486671805381775,
      "learning_rate": 4.648466218987096e-06,
      "loss": 0.6941,
      "step": 18628
    },
    {
      "epoch": 0.5367659770644845,
      "grad_norm": 0.743501603603363,
      "learning_rate": 4.64800076505784e-06,
      "loss": 0.7179,
      "step": 18629
    },
    {
      "epoch": 0.5367947905261338,
      "grad_norm": 0.7754152417182922,
      "learning_rate": 4.647535314194143e-06,
      "loss": 0.7196,
      "step": 18630
    },
    {
      "epoch": 0.5368236039877831,
      "grad_norm": 0.7504415512084961,
      "learning_rate": 4.647069866400064e-06,
      "loss": 0.7226,
      "step": 18631
    },
    {
      "epoch": 0.5368524174494324,
      "grad_norm": 0.8178760409355164,
      "learning_rate": 4.646604421679653e-06,
      "loss": 0.704,
      "step": 18632
    },
    {
      "epoch": 0.5368812309110816,
      "grad_norm": 0.7942051291465759,
      "learning_rate": 4.6461389800369635e-06,
      "loss": 0.7028,
      "step": 18633
    },
    {
      "epoch": 0.5369100443727309,
      "grad_norm": 0.7530243396759033,
      "learning_rate": 4.645673541476049e-06,
      "loss": 0.7186,
      "step": 18634
    },
    {
      "epoch": 0.5369388578343802,
      "grad_norm": 0.7681589722633362,
      "learning_rate": 4.645208106000964e-06,
      "loss": 0.7322,
      "step": 18635
    },
    {
      "epoch": 0.5369676712960295,
      "grad_norm": 0.7518856525421143,
      "learning_rate": 4.644742673615763e-06,
      "loss": 0.723,
      "step": 18636
    },
    {
      "epoch": 0.5369964847576788,
      "grad_norm": 0.7682174444198608,
      "learning_rate": 4.644277244324498e-06,
      "loss": 0.7017,
      "step": 18637
    },
    {
      "epoch": 0.5370252982193281,
      "grad_norm": 0.7746167778968811,
      "learning_rate": 4.643811818131223e-06,
      "loss": 0.7314,
      "step": 18638
    },
    {
      "epoch": 0.5370541116809774,
      "grad_norm": 0.7110556960105896,
      "learning_rate": 4.64334639503999e-06,
      "loss": 0.7162,
      "step": 18639
    },
    {
      "epoch": 0.5370829251426267,
      "grad_norm": 0.7196870446205139,
      "learning_rate": 4.642880975054854e-06,
      "loss": 0.7197,
      "step": 18640
    },
    {
      "epoch": 0.537111738604276,
      "grad_norm": 0.7186257243156433,
      "learning_rate": 4.642415558179868e-06,
      "loss": 0.6886,
      "step": 18641
    },
    {
      "epoch": 0.5371405520659251,
      "grad_norm": 0.7010065913200378,
      "learning_rate": 4.641950144419085e-06,
      "loss": 0.6977,
      "step": 18642
    },
    {
      "epoch": 0.5371693655275744,
      "grad_norm": 0.7277887463569641,
      "learning_rate": 4.641484733776558e-06,
      "loss": 0.6978,
      "step": 18643
    },
    {
      "epoch": 0.5371981789892237,
      "grad_norm": 0.7403784394264221,
      "learning_rate": 4.641019326256341e-06,
      "loss": 0.7097,
      "step": 18644
    },
    {
      "epoch": 0.537226992450873,
      "grad_norm": 0.7505519390106201,
      "learning_rate": 4.640553921862486e-06,
      "loss": 0.708,
      "step": 18645
    },
    {
      "epoch": 0.5372558059125223,
      "grad_norm": 0.7944642901420593,
      "learning_rate": 4.640088520599045e-06,
      "loss": 0.721,
      "step": 18646
    },
    {
      "epoch": 0.5372846193741716,
      "grad_norm": 0.7196399569511414,
      "learning_rate": 4.6396231224700764e-06,
      "loss": 0.7074,
      "step": 18647
    },
    {
      "epoch": 0.5373134328358209,
      "grad_norm": 0.7110104560852051,
      "learning_rate": 4.639157727479629e-06,
      "loss": 0.711,
      "step": 18648
    },
    {
      "epoch": 0.5373422462974702,
      "grad_norm": 0.713097333908081,
      "learning_rate": 4.638692335631758e-06,
      "loss": 0.6949,
      "step": 18649
    },
    {
      "epoch": 0.5373710597591195,
      "grad_norm": 0.7604856491088867,
      "learning_rate": 4.638226946930515e-06,
      "loss": 0.7464,
      "step": 18650
    },
    {
      "epoch": 0.5373998732207688,
      "grad_norm": 0.7430618405342102,
      "learning_rate": 4.637761561379952e-06,
      "loss": 0.7212,
      "step": 18651
    },
    {
      "epoch": 0.537428686682418,
      "grad_norm": 0.7137237191200256,
      "learning_rate": 4.637296178984125e-06,
      "loss": 0.7091,
      "step": 18652
    },
    {
      "epoch": 0.5374575001440673,
      "grad_norm": 0.7120773792266846,
      "learning_rate": 4.636830799747087e-06,
      "loss": 0.7189,
      "step": 18653
    },
    {
      "epoch": 0.5374863136057166,
      "grad_norm": 0.7501562833786011,
      "learning_rate": 4.636365423672889e-06,
      "loss": 0.7079,
      "step": 18654
    },
    {
      "epoch": 0.5375151270673659,
      "grad_norm": 0.7272403240203857,
      "learning_rate": 4.635900050765584e-06,
      "loss": 0.7015,
      "step": 18655
    },
    {
      "epoch": 0.5375439405290151,
      "grad_norm": 0.7670007348060608,
      "learning_rate": 4.6354346810292265e-06,
      "loss": 0.722,
      "step": 18656
    },
    {
      "epoch": 0.5375727539906644,
      "grad_norm": 0.7476809024810791,
      "learning_rate": 4.634969314467868e-06,
      "loss": 0.7044,
      "step": 18657
    },
    {
      "epoch": 0.5376015674523137,
      "grad_norm": 0.7429390549659729,
      "learning_rate": 4.6345039510855595e-06,
      "loss": 0.7184,
      "step": 18658
    },
    {
      "epoch": 0.537630380913963,
      "grad_norm": 0.7414183616638184,
      "learning_rate": 4.634038590886359e-06,
      "loss": 0.7084,
      "step": 18659
    },
    {
      "epoch": 0.5376591943756123,
      "grad_norm": 0.7667145729064941,
      "learning_rate": 4.6335732338743165e-06,
      "loss": 0.718,
      "step": 18660
    },
    {
      "epoch": 0.5376880078372616,
      "grad_norm": 0.7513254880905151,
      "learning_rate": 4.633107880053484e-06,
      "loss": 0.701,
      "step": 18661
    },
    {
      "epoch": 0.5377168212989109,
      "grad_norm": 0.7213367819786072,
      "learning_rate": 4.632642529427916e-06,
      "loss": 0.704,
      "step": 18662
    },
    {
      "epoch": 0.5377456347605601,
      "grad_norm": 0.7550381422042847,
      "learning_rate": 4.6321771820016635e-06,
      "loss": 0.7343,
      "step": 18663
    },
    {
      "epoch": 0.5377744482222094,
      "grad_norm": 0.7180637121200562,
      "learning_rate": 4.63171183777878e-06,
      "loss": 0.7327,
      "step": 18664
    },
    {
      "epoch": 0.5378032616838587,
      "grad_norm": 0.7437407374382019,
      "learning_rate": 4.63124649676332e-06,
      "loss": 0.7219,
      "step": 18665
    },
    {
      "epoch": 0.537832075145508,
      "grad_norm": 0.7803862690925598,
      "learning_rate": 4.630781158959333e-06,
      "loss": 0.7273,
      "step": 18666
    },
    {
      "epoch": 0.5378608886071573,
      "grad_norm": 0.7400212287902832,
      "learning_rate": 4.630315824370873e-06,
      "loss": 0.7152,
      "step": 18667
    },
    {
      "epoch": 0.5378897020688066,
      "grad_norm": 0.7433269619941711,
      "learning_rate": 4.629850493001992e-06,
      "loss": 0.6925,
      "step": 18668
    },
    {
      "epoch": 0.5379185155304558,
      "grad_norm": 0.6982102394104004,
      "learning_rate": 4.6293851648567425e-06,
      "loss": 0.6989,
      "step": 18669
    },
    {
      "epoch": 0.5379473289921051,
      "grad_norm": 0.9867549538612366,
      "learning_rate": 4.6289198399391795e-06,
      "loss": 0.7122,
      "step": 18670
    },
    {
      "epoch": 0.5379761424537544,
      "grad_norm": 0.704569935798645,
      "learning_rate": 4.628454518253354e-06,
      "loss": 0.7021,
      "step": 18671
    },
    {
      "epoch": 0.5380049559154036,
      "grad_norm": 2.3560049533843994,
      "learning_rate": 4.6279891998033175e-06,
      "loss": 0.6991,
      "step": 18672
    },
    {
      "epoch": 0.5380337693770529,
      "grad_norm": 0.7571591734886169,
      "learning_rate": 4.627523884593122e-06,
      "loss": 0.7034,
      "step": 18673
    },
    {
      "epoch": 0.5380625828387022,
      "grad_norm": 0.776980996131897,
      "learning_rate": 4.627058572626823e-06,
      "loss": 0.7009,
      "step": 18674
    },
    {
      "epoch": 0.5380913963003515,
      "grad_norm": 0.7575401067733765,
      "learning_rate": 4.6265932639084705e-06,
      "loss": 0.7097,
      "step": 18675
    },
    {
      "epoch": 0.5381202097620008,
      "grad_norm": 0.7197191715240479,
      "learning_rate": 4.626127958442118e-06,
      "loss": 0.7293,
      "step": 18676
    },
    {
      "epoch": 0.5381490232236501,
      "grad_norm": 0.7375248074531555,
      "learning_rate": 4.6256626562318165e-06,
      "loss": 0.7022,
      "step": 18677
    },
    {
      "epoch": 0.5381778366852994,
      "grad_norm": 0.75691819190979,
      "learning_rate": 4.625197357281619e-06,
      "loss": 0.6935,
      "step": 18678
    },
    {
      "epoch": 0.5382066501469487,
      "grad_norm": 0.7318138480186462,
      "learning_rate": 4.624732061595578e-06,
      "loss": 0.6914,
      "step": 18679
    },
    {
      "epoch": 0.538235463608598,
      "grad_norm": 0.7190735340118408,
      "learning_rate": 4.624266769177746e-06,
      "loss": 0.7182,
      "step": 18680
    },
    {
      "epoch": 0.5382642770702473,
      "grad_norm": 0.7282930612564087,
      "learning_rate": 4.623801480032172e-06,
      "loss": 0.7181,
      "step": 18681
    },
    {
      "epoch": 0.5382930905318966,
      "grad_norm": 0.7572371959686279,
      "learning_rate": 4.623336194162914e-06,
      "loss": 0.7122,
      "step": 18682
    },
    {
      "epoch": 0.5383219039935457,
      "grad_norm": 0.7494865655899048,
      "learning_rate": 4.622870911574021e-06,
      "loss": 0.7045,
      "step": 18683
    },
    {
      "epoch": 0.538350717455195,
      "grad_norm": 0.7684451937675476,
      "learning_rate": 4.6224056322695455e-06,
      "loss": 0.7365,
      "step": 18684
    },
    {
      "epoch": 0.5383795309168443,
      "grad_norm": 0.7996498346328735,
      "learning_rate": 4.621940356253539e-06,
      "loss": 0.7523,
      "step": 18685
    },
    {
      "epoch": 0.5384083443784936,
      "grad_norm": 0.7865989804267883,
      "learning_rate": 4.621475083530055e-06,
      "loss": 0.7151,
      "step": 18686
    },
    {
      "epoch": 0.5384371578401429,
      "grad_norm": 0.7275128364562988,
      "learning_rate": 4.621009814103144e-06,
      "loss": 0.6917,
      "step": 18687
    },
    {
      "epoch": 0.5384659713017922,
      "grad_norm": 0.7325707674026489,
      "learning_rate": 4.620544547976858e-06,
      "loss": 0.7396,
      "step": 18688
    },
    {
      "epoch": 0.5384947847634415,
      "grad_norm": 0.7593451738357544,
      "learning_rate": 4.6200792851552494e-06,
      "loss": 0.7123,
      "step": 18689
    },
    {
      "epoch": 0.5385235982250908,
      "grad_norm": 0.7086079716682434,
      "learning_rate": 4.619614025642371e-06,
      "loss": 0.7105,
      "step": 18690
    },
    {
      "epoch": 0.5385524116867401,
      "grad_norm": 0.7398648858070374,
      "learning_rate": 4.619148769442275e-06,
      "loss": 0.6979,
      "step": 18691
    },
    {
      "epoch": 0.5385812251483894,
      "grad_norm": 0.7100307941436768,
      "learning_rate": 4.618683516559012e-06,
      "loss": 0.7211,
      "step": 18692
    },
    {
      "epoch": 0.5386100386100386,
      "grad_norm": 0.7282690405845642,
      "learning_rate": 4.6182182669966334e-06,
      "loss": 0.7381,
      "step": 18693
    },
    {
      "epoch": 0.5386388520716879,
      "grad_norm": 0.7119263410568237,
      "learning_rate": 4.617753020759192e-06,
      "loss": 0.6976,
      "step": 18694
    },
    {
      "epoch": 0.5386676655333372,
      "grad_norm": 0.807538628578186,
      "learning_rate": 4.617287777850741e-06,
      "loss": 0.7096,
      "step": 18695
    },
    {
      "epoch": 0.5386964789949864,
      "grad_norm": 0.7419636249542236,
      "learning_rate": 4.61682253827533e-06,
      "loss": 0.7251,
      "step": 18696
    },
    {
      "epoch": 0.5387252924566357,
      "grad_norm": 0.7480245232582092,
      "learning_rate": 4.616357302037013e-06,
      "loss": 0.7173,
      "step": 18697
    },
    {
      "epoch": 0.538754105918285,
      "grad_norm": 0.7196113467216492,
      "learning_rate": 4.615892069139839e-06,
      "loss": 0.7215,
      "step": 18698
    },
    {
      "epoch": 0.5387829193799343,
      "grad_norm": 1.0248918533325195,
      "learning_rate": 4.615426839587862e-06,
      "loss": 0.7124,
      "step": 18699
    },
    {
      "epoch": 0.5388117328415836,
      "grad_norm": 0.7642817497253418,
      "learning_rate": 4.614961613385133e-06,
      "loss": 0.744,
      "step": 18700
    },
    {
      "epoch": 0.5388405463032329,
      "grad_norm": 0.7461541295051575,
      "learning_rate": 4.614496390535703e-06,
      "loss": 0.7161,
      "step": 18701
    },
    {
      "epoch": 0.5388693597648821,
      "grad_norm": 0.7444356679916382,
      "learning_rate": 4.614031171043624e-06,
      "loss": 0.7189,
      "step": 18702
    },
    {
      "epoch": 0.5388981732265314,
      "grad_norm": 0.7345207929611206,
      "learning_rate": 4.613565954912948e-06,
      "loss": 0.7056,
      "step": 18703
    },
    {
      "epoch": 0.5389269866881807,
      "grad_norm": 0.7158993482589722,
      "learning_rate": 4.613100742147726e-06,
      "loss": 0.7012,
      "step": 18704
    },
    {
      "epoch": 0.53895580014983,
      "grad_norm": 0.7457074522972107,
      "learning_rate": 4.612635532752008e-06,
      "loss": 0.7494,
      "step": 18705
    },
    {
      "epoch": 0.5389846136114793,
      "grad_norm": 0.7255617380142212,
      "learning_rate": 4.612170326729849e-06,
      "loss": 0.6936,
      "step": 18706
    },
    {
      "epoch": 0.5390134270731286,
      "grad_norm": 0.7367914319038391,
      "learning_rate": 4.6117051240852985e-06,
      "loss": 0.7119,
      "step": 18707
    },
    {
      "epoch": 0.5390422405347779,
      "grad_norm": 0.7133292555809021,
      "learning_rate": 4.611239924822408e-06,
      "loss": 0.6852,
      "step": 18708
    },
    {
      "epoch": 0.5390710539964272,
      "grad_norm": 0.7279715538024902,
      "learning_rate": 4.61077472894523e-06,
      "loss": 0.7259,
      "step": 18709
    },
    {
      "epoch": 0.5390998674580764,
      "grad_norm": 0.7336022257804871,
      "learning_rate": 4.610309536457816e-06,
      "loss": 0.707,
      "step": 18710
    },
    {
      "epoch": 0.5391286809197257,
      "grad_norm": 0.7082381844520569,
      "learning_rate": 4.6098443473642135e-06,
      "loss": 0.7028,
      "step": 18711
    },
    {
      "epoch": 0.539157494381375,
      "grad_norm": 0.7320845723152161,
      "learning_rate": 4.609379161668478e-06,
      "loss": 0.7046,
      "step": 18712
    },
    {
      "epoch": 0.5391863078430242,
      "grad_norm": 0.7532527446746826,
      "learning_rate": 4.6089139793746605e-06,
      "loss": 0.6959,
      "step": 18713
    },
    {
      "epoch": 0.5392151213046735,
      "grad_norm": 0.7098631858825684,
      "learning_rate": 4.608448800486811e-06,
      "loss": 0.7036,
      "step": 18714
    },
    {
      "epoch": 0.5392439347663228,
      "grad_norm": 0.7239962220191956,
      "learning_rate": 4.60798362500898e-06,
      "loss": 0.7024,
      "step": 18715
    },
    {
      "epoch": 0.5392727482279721,
      "grad_norm": 0.7208250761032104,
      "learning_rate": 4.60751845294522e-06,
      "loss": 0.6912,
      "step": 18716
    },
    {
      "epoch": 0.5393015616896214,
      "grad_norm": 0.7312538623809814,
      "learning_rate": 4.607053284299581e-06,
      "loss": 0.6961,
      "step": 18717
    },
    {
      "epoch": 0.5393303751512707,
      "grad_norm": 0.7223340272903442,
      "learning_rate": 4.606588119076117e-06,
      "loss": 0.7153,
      "step": 18718
    },
    {
      "epoch": 0.53935918861292,
      "grad_norm": 0.7624298930168152,
      "learning_rate": 4.606122957278876e-06,
      "loss": 0.7055,
      "step": 18719
    },
    {
      "epoch": 0.5393880020745693,
      "grad_norm": 0.8075831532478333,
      "learning_rate": 4.605657798911911e-06,
      "loss": 0.709,
      "step": 18720
    },
    {
      "epoch": 0.5394168155362186,
      "grad_norm": 0.7099449634552002,
      "learning_rate": 4.605192643979273e-06,
      "loss": 0.6913,
      "step": 18721
    },
    {
      "epoch": 0.5394456289978679,
      "grad_norm": 0.7342689633369446,
      "learning_rate": 4.604727492485011e-06,
      "loss": 0.7371,
      "step": 18722
    },
    {
      "epoch": 0.539474442459517,
      "grad_norm": 0.7602897882461548,
      "learning_rate": 4.6042623444331785e-06,
      "loss": 0.7269,
      "step": 18723
    },
    {
      "epoch": 0.5395032559211663,
      "grad_norm": 0.7178478240966797,
      "learning_rate": 4.603797199827825e-06,
      "loss": 0.7125,
      "step": 18724
    },
    {
      "epoch": 0.5395320693828156,
      "grad_norm": 0.7193270325660706,
      "learning_rate": 4.603332058673002e-06,
      "loss": 0.7119,
      "step": 18725
    },
    {
      "epoch": 0.5395608828444649,
      "grad_norm": 0.7467480897903442,
      "learning_rate": 4.6028669209727605e-06,
      "loss": 0.7203,
      "step": 18726
    },
    {
      "epoch": 0.5395896963061142,
      "grad_norm": 0.7697263956069946,
      "learning_rate": 4.60240178673115e-06,
      "loss": 0.7308,
      "step": 18727
    },
    {
      "epoch": 0.5396185097677635,
      "grad_norm": 0.716687798500061,
      "learning_rate": 4.601936655952223e-06,
      "loss": 0.7049,
      "step": 18728
    },
    {
      "epoch": 0.5396473232294128,
      "grad_norm": 0.7231616973876953,
      "learning_rate": 4.601471528640029e-06,
      "loss": 0.7161,
      "step": 18729
    },
    {
      "epoch": 0.5396761366910621,
      "grad_norm": 0.7436358332633972,
      "learning_rate": 4.601006404798621e-06,
      "loss": 0.7023,
      "step": 18730
    },
    {
      "epoch": 0.5397049501527114,
      "grad_norm": 0.742897629737854,
      "learning_rate": 4.600541284432048e-06,
      "loss": 0.7108,
      "step": 18731
    },
    {
      "epoch": 0.5397337636143607,
      "grad_norm": 0.7380431294441223,
      "learning_rate": 4.60007616754436e-06,
      "loss": 0.7094,
      "step": 18732
    },
    {
      "epoch": 0.5397625770760099,
      "grad_norm": 0.738090455532074,
      "learning_rate": 4.59961105413961e-06,
      "loss": 0.7135,
      "step": 18733
    },
    {
      "epoch": 0.5397913905376592,
      "grad_norm": 0.8196823000907898,
      "learning_rate": 4.599145944221848e-06,
      "loss": 0.7158,
      "step": 18734
    },
    {
      "epoch": 0.5398202039993085,
      "grad_norm": 0.7430484294891357,
      "learning_rate": 4.598680837795123e-06,
      "loss": 0.6945,
      "step": 18735
    },
    {
      "epoch": 0.5398490174609578,
      "grad_norm": 0.742824912071228,
      "learning_rate": 4.598215734863487e-06,
      "loss": 0.7063,
      "step": 18736
    },
    {
      "epoch": 0.539877830922607,
      "grad_norm": 0.7528170943260193,
      "learning_rate": 4.597750635430991e-06,
      "loss": 0.6861,
      "step": 18737
    },
    {
      "epoch": 0.5399066443842563,
      "grad_norm": 0.7583324313163757,
      "learning_rate": 4.597285539501684e-06,
      "loss": 0.7187,
      "step": 18738
    },
    {
      "epoch": 0.5399354578459056,
      "grad_norm": 0.7592909336090088,
      "learning_rate": 4.5968204470796185e-06,
      "loss": 0.7137,
      "step": 18739
    },
    {
      "epoch": 0.5399642713075549,
      "grad_norm": 0.7711599469184875,
      "learning_rate": 4.596355358168844e-06,
      "loss": 0.6748,
      "step": 18740
    },
    {
      "epoch": 0.5399930847692042,
      "grad_norm": 0.7530016303062439,
      "learning_rate": 4.595890272773407e-06,
      "loss": 0.7113,
      "step": 18741
    },
    {
      "epoch": 0.5400218982308534,
      "grad_norm": 0.7491944432258606,
      "learning_rate": 4.5954251908973655e-06,
      "loss": 0.7193,
      "step": 18742
    },
    {
      "epoch": 0.5400507116925027,
      "grad_norm": 0.7400559186935425,
      "learning_rate": 4.594960112544765e-06,
      "loss": 0.7156,
      "step": 18743
    },
    {
      "epoch": 0.540079525154152,
      "grad_norm": 0.778309166431427,
      "learning_rate": 4.594495037719658e-06,
      "loss": 0.6874,
      "step": 18744
    },
    {
      "epoch": 0.5401083386158013,
      "grad_norm": 0.7371770143508911,
      "learning_rate": 4.594029966426093e-06,
      "loss": 0.6893,
      "step": 18745
    },
    {
      "epoch": 0.5401371520774506,
      "grad_norm": 0.7271918654441833,
      "learning_rate": 4.593564898668122e-06,
      "loss": 0.7041,
      "step": 18746
    },
    {
      "epoch": 0.5401659655390999,
      "grad_norm": 0.7080684304237366,
      "learning_rate": 4.593099834449793e-06,
      "loss": 0.7218,
      "step": 18747
    },
    {
      "epoch": 0.5401947790007492,
      "grad_norm": 0.7379457354545593,
      "learning_rate": 4.592634773775159e-06,
      "loss": 0.718,
      "step": 18748
    },
    {
      "epoch": 0.5402235924623985,
      "grad_norm": 0.7542738914489746,
      "learning_rate": 4.592169716648268e-06,
      "loss": 0.7077,
      "step": 18749
    },
    {
      "epoch": 0.5402524059240477,
      "grad_norm": 0.7206648588180542,
      "learning_rate": 4.591704663073172e-06,
      "loss": 0.6974,
      "step": 18750
    },
    {
      "epoch": 0.540281219385697,
      "grad_norm": 0.7232993245124817,
      "learning_rate": 4.591239613053919e-06,
      "loss": 0.728,
      "step": 18751
    },
    {
      "epoch": 0.5403100328473462,
      "grad_norm": 0.7247612476348877,
      "learning_rate": 4.590774566594562e-06,
      "loss": 0.7113,
      "step": 18752
    },
    {
      "epoch": 0.5403388463089955,
      "grad_norm": 0.7203108072280884,
      "learning_rate": 4.590309523699148e-06,
      "loss": 0.7089,
      "step": 18753
    },
    {
      "epoch": 0.5403676597706448,
      "grad_norm": 0.7454737424850464,
      "learning_rate": 4.589844484371728e-06,
      "loss": 0.7243,
      "step": 18754
    },
    {
      "epoch": 0.5403964732322941,
      "grad_norm": 2.21628999710083,
      "learning_rate": 4.589379448616353e-06,
      "loss": 0.6925,
      "step": 18755
    },
    {
      "epoch": 0.5404252866939434,
      "grad_norm": 0.7547779679298401,
      "learning_rate": 4.588914416437074e-06,
      "loss": 0.7213,
      "step": 18756
    },
    {
      "epoch": 0.5404541001555927,
      "grad_norm": 0.7008587718009949,
      "learning_rate": 4.588449387837939e-06,
      "loss": 0.7289,
      "step": 18757
    },
    {
      "epoch": 0.540482913617242,
      "grad_norm": 0.7164963483810425,
      "learning_rate": 4.587984362822998e-06,
      "loss": 0.7307,
      "step": 18758
    },
    {
      "epoch": 0.5405117270788913,
      "grad_norm": 0.7382416725158691,
      "learning_rate": 4.587519341396302e-06,
      "loss": 0.7383,
      "step": 18759
    },
    {
      "epoch": 0.5405405405405406,
      "grad_norm": 0.7423871755599976,
      "learning_rate": 4.587054323561899e-06,
      "loss": 0.7069,
      "step": 18760
    },
    {
      "epoch": 0.5405693540021899,
      "grad_norm": 0.7667617201805115,
      "learning_rate": 4.586589309323841e-06,
      "loss": 0.7227,
      "step": 18761
    },
    {
      "epoch": 0.5405981674638392,
      "grad_norm": 0.7953338623046875,
      "learning_rate": 4.586124298686177e-06,
      "loss": 0.7156,
      "step": 18762
    },
    {
      "epoch": 0.5406269809254884,
      "grad_norm": 0.7271504402160645,
      "learning_rate": 4.585659291652956e-06,
      "loss": 0.7175,
      "step": 18763
    },
    {
      "epoch": 0.5406557943871376,
      "grad_norm": 0.7345466017723083,
      "learning_rate": 4.585194288228229e-06,
      "loss": 0.7002,
      "step": 18764
    },
    {
      "epoch": 0.5406846078487869,
      "grad_norm": 0.7275170683860779,
      "learning_rate": 4.584729288416043e-06,
      "loss": 0.6957,
      "step": 18765
    },
    {
      "epoch": 0.5407134213104362,
      "grad_norm": 0.7487671971321106,
      "learning_rate": 4.584264292220451e-06,
      "loss": 0.7163,
      "step": 18766
    },
    {
      "epoch": 0.5407422347720855,
      "grad_norm": 0.7284377217292786,
      "learning_rate": 4.583799299645502e-06,
      "loss": 0.7111,
      "step": 18767
    },
    {
      "epoch": 0.5407710482337348,
      "grad_norm": 0.7517327070236206,
      "learning_rate": 4.5833343106952455e-06,
      "loss": 0.7144,
      "step": 18768
    },
    {
      "epoch": 0.5407998616953841,
      "grad_norm": 0.7461457848548889,
      "learning_rate": 4.58286932537373e-06,
      "loss": 0.7079,
      "step": 18769
    },
    {
      "epoch": 0.5408286751570334,
      "grad_norm": 0.7296302914619446,
      "learning_rate": 4.5824043436850054e-06,
      "loss": 0.6956,
      "step": 18770
    },
    {
      "epoch": 0.5408574886186827,
      "grad_norm": 0.7271178960800171,
      "learning_rate": 4.581939365633122e-06,
      "loss": 0.6963,
      "step": 18771
    },
    {
      "epoch": 0.540886302080332,
      "grad_norm": 0.7227829694747925,
      "learning_rate": 4.5814743912221285e-06,
      "loss": 0.7302,
      "step": 18772
    },
    {
      "epoch": 0.5409151155419812,
      "grad_norm": 0.7092658281326294,
      "learning_rate": 4.581009420456075e-06,
      "loss": 0.6957,
      "step": 18773
    },
    {
      "epoch": 0.5409439290036305,
      "grad_norm": 0.7504186034202576,
      "learning_rate": 4.580544453339011e-06,
      "loss": 0.7093,
      "step": 18774
    },
    {
      "epoch": 0.5409727424652798,
      "grad_norm": 0.7006649374961853,
      "learning_rate": 4.580079489874985e-06,
      "loss": 0.7078,
      "step": 18775
    },
    {
      "epoch": 0.5410015559269291,
      "grad_norm": 0.7344843745231628,
      "learning_rate": 4.579614530068047e-06,
      "loss": 0.7407,
      "step": 18776
    },
    {
      "epoch": 0.5410303693885783,
      "grad_norm": 0.7303394675254822,
      "learning_rate": 4.5791495739222445e-06,
      "loss": 0.7257,
      "step": 18777
    },
    {
      "epoch": 0.5410591828502276,
      "grad_norm": 0.7136238217353821,
      "learning_rate": 4.578684621441629e-06,
      "loss": 0.7017,
      "step": 18778
    },
    {
      "epoch": 0.5410879963118769,
      "grad_norm": 0.7147611379623413,
      "learning_rate": 4.578219672630252e-06,
      "loss": 0.697,
      "step": 18779
    },
    {
      "epoch": 0.5411168097735262,
      "grad_norm": 0.7298113703727722,
      "learning_rate": 4.577754727492158e-06,
      "loss": 0.7202,
      "step": 18780
    },
    {
      "epoch": 0.5411456232351755,
      "grad_norm": 0.7104056477546692,
      "learning_rate": 4.577289786031399e-06,
      "loss": 0.7023,
      "step": 18781
    },
    {
      "epoch": 0.5411744366968247,
      "grad_norm": 0.7110889554023743,
      "learning_rate": 4.576824848252023e-06,
      "loss": 0.6969,
      "step": 18782
    },
    {
      "epoch": 0.541203250158474,
      "grad_norm": 0.7229949235916138,
      "learning_rate": 4.576359914158079e-06,
      "loss": 0.7021,
      "step": 18783
    },
    {
      "epoch": 0.5412320636201233,
      "grad_norm": 0.7213910818099976,
      "learning_rate": 4.5758949837536186e-06,
      "loss": 0.7178,
      "step": 18784
    },
    {
      "epoch": 0.5412608770817726,
      "grad_norm": 0.7272641658782959,
      "learning_rate": 4.5754300570426874e-06,
      "loss": 0.7015,
      "step": 18785
    },
    {
      "epoch": 0.5412896905434219,
      "grad_norm": 0.7221591472625732,
      "learning_rate": 4.5749651340293355e-06,
      "loss": 0.7096,
      "step": 18786
    },
    {
      "epoch": 0.5413185040050712,
      "grad_norm": 0.7130221724510193,
      "learning_rate": 4.574500214717613e-06,
      "loss": 0.7225,
      "step": 18787
    },
    {
      "epoch": 0.5413473174667205,
      "grad_norm": 0.7257688641548157,
      "learning_rate": 4.574035299111569e-06,
      "loss": 0.7145,
      "step": 18788
    },
    {
      "epoch": 0.5413761309283698,
      "grad_norm": 2.919856309890747,
      "learning_rate": 4.57357038721525e-06,
      "loss": 0.7065,
      "step": 18789
    },
    {
      "epoch": 0.5414049443900191,
      "grad_norm": 0.7162018418312073,
      "learning_rate": 4.573105479032708e-06,
      "loss": 0.6968,
      "step": 18790
    },
    {
      "epoch": 0.5414337578516683,
      "grad_norm": 0.7052030563354492,
      "learning_rate": 4.572640574567989e-06,
      "loss": 0.6898,
      "step": 18791
    },
    {
      "epoch": 0.5414625713133175,
      "grad_norm": 0.7339329123497009,
      "learning_rate": 4.572175673825145e-06,
      "loss": 0.7321,
      "step": 18792
    },
    {
      "epoch": 0.5414913847749668,
      "grad_norm": 0.726813018321991,
      "learning_rate": 4.571710776808223e-06,
      "loss": 0.7136,
      "step": 18793
    },
    {
      "epoch": 0.5415201982366161,
      "grad_norm": 0.7306279540061951,
      "learning_rate": 4.571245883521272e-06,
      "loss": 0.6932,
      "step": 18794
    },
    {
      "epoch": 0.5415490116982654,
      "grad_norm": 0.718458890914917,
      "learning_rate": 4.570780993968341e-06,
      "loss": 0.6984,
      "step": 18795
    },
    {
      "epoch": 0.5415778251599147,
      "grad_norm": 0.7064930200576782,
      "learning_rate": 4.570316108153479e-06,
      "loss": 0.707,
      "step": 18796
    },
    {
      "epoch": 0.541606638621564,
      "grad_norm": 0.6815741658210754,
      "learning_rate": 4.569851226080732e-06,
      "loss": 0.7132,
      "step": 18797
    },
    {
      "epoch": 0.5416354520832133,
      "grad_norm": 0.709443986415863,
      "learning_rate": 4.569386347754154e-06,
      "loss": 0.7078,
      "step": 18798
    },
    {
      "epoch": 0.5416642655448626,
      "grad_norm": 0.7560135722160339,
      "learning_rate": 4.568921473177788e-06,
      "loss": 0.7105,
      "step": 18799
    },
    {
      "epoch": 0.5416930790065119,
      "grad_norm": 0.7399731874465942,
      "learning_rate": 4.5684566023556855e-06,
      "loss": 0.7253,
      "step": 18800
    },
    {
      "epoch": 0.5417218924681612,
      "grad_norm": 0.7425034642219543,
      "learning_rate": 4.567991735291893e-06,
      "loss": 0.7166,
      "step": 18801
    },
    {
      "epoch": 0.5417507059298104,
      "grad_norm": 0.7543836832046509,
      "learning_rate": 4.567526871990462e-06,
      "loss": 0.7128,
      "step": 18802
    },
    {
      "epoch": 0.5417795193914597,
      "grad_norm": 0.7431557774543762,
      "learning_rate": 4.567062012455441e-06,
      "loss": 0.7221,
      "step": 18803
    },
    {
      "epoch": 0.5418083328531089,
      "grad_norm": 0.7304123044013977,
      "learning_rate": 4.566597156690876e-06,
      "loss": 0.7065,
      "step": 18804
    },
    {
      "epoch": 0.5418371463147582,
      "grad_norm": 0.7195771336555481,
      "learning_rate": 4.566132304700817e-06,
      "loss": 0.6959,
      "step": 18805
    },
    {
      "epoch": 0.5418659597764075,
      "grad_norm": 0.7426347136497498,
      "learning_rate": 4.565667456489311e-06,
      "loss": 0.7379,
      "step": 18806
    },
    {
      "epoch": 0.5418947732380568,
      "grad_norm": 0.7183668613433838,
      "learning_rate": 4.565202612060408e-06,
      "loss": 0.7137,
      "step": 18807
    },
    {
      "epoch": 0.5419235866997061,
      "grad_norm": 0.7127727270126343,
      "learning_rate": 4.564737771418155e-06,
      "loss": 0.7157,
      "step": 18808
    },
    {
      "epoch": 0.5419524001613554,
      "grad_norm": 0.777128279209137,
      "learning_rate": 4.5642729345666015e-06,
      "loss": 0.7205,
      "step": 18809
    },
    {
      "epoch": 0.5419812136230047,
      "grad_norm": 0.69672691822052,
      "learning_rate": 4.563808101509796e-06,
      "loss": 0.6884,
      "step": 18810
    },
    {
      "epoch": 0.542010027084654,
      "grad_norm": 0.7428389191627502,
      "learning_rate": 4.563343272251784e-06,
      "loss": 0.7254,
      "step": 18811
    },
    {
      "epoch": 0.5420388405463032,
      "grad_norm": 0.7362141609191895,
      "learning_rate": 4.562878446796617e-06,
      "loss": 0.7171,
      "step": 18812
    },
    {
      "epoch": 0.5420676540079525,
      "grad_norm": 0.736323893070221,
      "learning_rate": 4.5624136251483395e-06,
      "loss": 0.6933,
      "step": 18813
    },
    {
      "epoch": 0.5420964674696018,
      "grad_norm": 0.7367449998855591,
      "learning_rate": 4.561948807311004e-06,
      "loss": 0.6861,
      "step": 18814
    },
    {
      "epoch": 0.5421252809312511,
      "grad_norm": 0.7173601388931274,
      "learning_rate": 4.561483993288656e-06,
      "loss": 0.6961,
      "step": 18815
    },
    {
      "epoch": 0.5421540943929004,
      "grad_norm": 0.734319806098938,
      "learning_rate": 4.561019183085344e-06,
      "loss": 0.7211,
      "step": 18816
    },
    {
      "epoch": 0.5421829078545497,
      "grad_norm": 0.7433280348777771,
      "learning_rate": 4.560554376705117e-06,
      "loss": 0.7177,
      "step": 18817
    },
    {
      "epoch": 0.5422117213161989,
      "grad_norm": 0.7444132566452026,
      "learning_rate": 4.5600895741520215e-06,
      "loss": 0.7316,
      "step": 18818
    },
    {
      "epoch": 0.5422405347778482,
      "grad_norm": 0.7354813814163208,
      "learning_rate": 4.5596247754301065e-06,
      "loss": 0.7128,
      "step": 18819
    },
    {
      "epoch": 0.5422693482394975,
      "grad_norm": 0.7564262747764587,
      "learning_rate": 4.5591599805434195e-06,
      "loss": 0.7234,
      "step": 18820
    },
    {
      "epoch": 0.5422981617011468,
      "grad_norm": 0.7110199332237244,
      "learning_rate": 4.5586951894960076e-06,
      "loss": 0.7242,
      "step": 18821
    },
    {
      "epoch": 0.542326975162796,
      "grad_norm": 0.7245821356773376,
      "learning_rate": 4.5582304022919195e-06,
      "loss": 0.7135,
      "step": 18822
    },
    {
      "epoch": 0.5423557886244453,
      "grad_norm": 0.710999071598053,
      "learning_rate": 4.557765618935204e-06,
      "loss": 0.7152,
      "step": 18823
    },
    {
      "epoch": 0.5423846020860946,
      "grad_norm": 0.7159926295280457,
      "learning_rate": 4.557300839429906e-06,
      "loss": 0.7065,
      "step": 18824
    },
    {
      "epoch": 0.5424134155477439,
      "grad_norm": 0.7254392504692078,
      "learning_rate": 4.556836063780076e-06,
      "loss": 0.7,
      "step": 18825
    },
    {
      "epoch": 0.5424422290093932,
      "grad_norm": 0.7364533543586731,
      "learning_rate": 4.556371291989761e-06,
      "loss": 0.7105,
      "step": 18826
    },
    {
      "epoch": 0.5424710424710425,
      "grad_norm": 0.7059887647628784,
      "learning_rate": 4.555906524063009e-06,
      "loss": 0.7028,
      "step": 18827
    },
    {
      "epoch": 0.5424998559326918,
      "grad_norm": 0.7211258411407471,
      "learning_rate": 4.555441760003867e-06,
      "loss": 0.7091,
      "step": 18828
    },
    {
      "epoch": 0.5425286693943411,
      "grad_norm": 0.7132704257965088,
      "learning_rate": 4.554976999816382e-06,
      "loss": 0.7081,
      "step": 18829
    },
    {
      "epoch": 0.5425574828559904,
      "grad_norm": 0.7226141095161438,
      "learning_rate": 4.554512243504604e-06,
      "loss": 0.7021,
      "step": 18830
    },
    {
      "epoch": 0.5425862963176395,
      "grad_norm": 0.7191385626792908,
      "learning_rate": 4.5540474910725775e-06,
      "loss": 0.7395,
      "step": 18831
    },
    {
      "epoch": 0.5426151097792888,
      "grad_norm": 0.7178021669387817,
      "learning_rate": 4.5535827425243526e-06,
      "loss": 0.7103,
      "step": 18832
    },
    {
      "epoch": 0.5426439232409381,
      "grad_norm": 0.7357254028320312,
      "learning_rate": 4.553117997863976e-06,
      "loss": 0.7124,
      "step": 18833
    },
    {
      "epoch": 0.5426727367025874,
      "grad_norm": 0.7468987107276917,
      "learning_rate": 4.552653257095495e-06,
      "loss": 0.7377,
      "step": 18834
    },
    {
      "epoch": 0.5427015501642367,
      "grad_norm": 0.7236412763595581,
      "learning_rate": 4.552188520222957e-06,
      "loss": 0.7198,
      "step": 18835
    },
    {
      "epoch": 0.542730363625886,
      "grad_norm": 0.731452465057373,
      "learning_rate": 4.551723787250408e-06,
      "loss": 0.7266,
      "step": 18836
    },
    {
      "epoch": 0.5427591770875353,
      "grad_norm": 0.7133705019950867,
      "learning_rate": 4.551259058181896e-06,
      "loss": 0.7063,
      "step": 18837
    },
    {
      "epoch": 0.5427879905491846,
      "grad_norm": 0.7169626951217651,
      "learning_rate": 4.550794333021471e-06,
      "loss": 0.7064,
      "step": 18838
    },
    {
      "epoch": 0.5428168040108339,
      "grad_norm": 0.7013089060783386,
      "learning_rate": 4.550329611773178e-06,
      "loss": 0.7051,
      "step": 18839
    },
    {
      "epoch": 0.5428456174724832,
      "grad_norm": 0.7329317927360535,
      "learning_rate": 4.549864894441065e-06,
      "loss": 0.7038,
      "step": 18840
    },
    {
      "epoch": 0.5428744309341325,
      "grad_norm": 0.7366257309913635,
      "learning_rate": 4.5494001810291775e-06,
      "loss": 0.725,
      "step": 18841
    },
    {
      "epoch": 0.5429032443957817,
      "grad_norm": 0.7186857461929321,
      "learning_rate": 4.548935471541565e-06,
      "loss": 0.699,
      "step": 18842
    },
    {
      "epoch": 0.542932057857431,
      "grad_norm": 0.7342529892921448,
      "learning_rate": 4.548470765982274e-06,
      "loss": 0.7218,
      "step": 18843
    },
    {
      "epoch": 0.5429608713190803,
      "grad_norm": 0.7124208807945251,
      "learning_rate": 4.54800606435535e-06,
      "loss": 0.7395,
      "step": 18844
    },
    {
      "epoch": 0.5429896847807295,
      "grad_norm": 0.7005149126052856,
      "learning_rate": 4.5475413666648425e-06,
      "loss": 0.6861,
      "step": 18845
    },
    {
      "epoch": 0.5430184982423788,
      "grad_norm": 0.7066167593002319,
      "learning_rate": 4.547076672914796e-06,
      "loss": 0.7115,
      "step": 18846
    },
    {
      "epoch": 0.5430473117040281,
      "grad_norm": 0.7068942785263062,
      "learning_rate": 4.546611983109259e-06,
      "loss": 0.7015,
      "step": 18847
    },
    {
      "epoch": 0.5430761251656774,
      "grad_norm": 0.7360032200813293,
      "learning_rate": 4.546147297252281e-06,
      "loss": 0.7272,
      "step": 18848
    },
    {
      "epoch": 0.5431049386273267,
      "grad_norm": 0.7387974262237549,
      "learning_rate": 4.545682615347903e-06,
      "loss": 0.7194,
      "step": 18849
    },
    {
      "epoch": 0.543133752088976,
      "grad_norm": 0.7174544334411621,
      "learning_rate": 4.545217937400177e-06,
      "loss": 0.7139,
      "step": 18850
    },
    {
      "epoch": 0.5431625655506253,
      "grad_norm": 0.7330710887908936,
      "learning_rate": 4.544753263413147e-06,
      "loss": 0.7042,
      "step": 18851
    },
    {
      "epoch": 0.5431913790122745,
      "grad_norm": 0.7270636558532715,
      "learning_rate": 4.544288593390862e-06,
      "loss": 0.7043,
      "step": 18852
    },
    {
      "epoch": 0.5432201924739238,
      "grad_norm": 0.7220093607902527,
      "learning_rate": 4.543823927337368e-06,
      "loss": 0.7156,
      "step": 18853
    },
    {
      "epoch": 0.5432490059355731,
      "grad_norm": 0.7210941314697266,
      "learning_rate": 4.5433592652567115e-06,
      "loss": 0.7094,
      "step": 18854
    },
    {
      "epoch": 0.5432778193972224,
      "grad_norm": 0.7475382089614868,
      "learning_rate": 4.542894607152939e-06,
      "loss": 0.7138,
      "step": 18855
    },
    {
      "epoch": 0.5433066328588717,
      "grad_norm": 0.7060452103614807,
      "learning_rate": 4.542429953030099e-06,
      "loss": 0.7162,
      "step": 18856
    },
    {
      "epoch": 0.543335446320521,
      "grad_norm": 0.738315224647522,
      "learning_rate": 4.541965302892235e-06,
      "loss": 0.7421,
      "step": 18857
    },
    {
      "epoch": 0.5433642597821702,
      "grad_norm": 0.7238003015518188,
      "learning_rate": 4.541500656743396e-06,
      "loss": 0.7305,
      "step": 18858
    },
    {
      "epoch": 0.5433930732438195,
      "grad_norm": 0.7696155905723572,
      "learning_rate": 4.541036014587627e-06,
      "loss": 0.7126,
      "step": 18859
    },
    {
      "epoch": 0.5434218867054688,
      "grad_norm": 0.7148600816726685,
      "learning_rate": 4.540571376428977e-06,
      "loss": 0.7049,
      "step": 18860
    },
    {
      "epoch": 0.543450700167118,
      "grad_norm": 0.7130028009414673,
      "learning_rate": 4.5401067422714885e-06,
      "loss": 0.7135,
      "step": 18861
    },
    {
      "epoch": 0.5434795136287673,
      "grad_norm": 0.7235983610153198,
      "learning_rate": 4.539642112119212e-06,
      "loss": 0.7154,
      "step": 18862
    },
    {
      "epoch": 0.5435083270904166,
      "grad_norm": 0.6948509812355042,
      "learning_rate": 4.539177485976194e-06,
      "loss": 0.6951,
      "step": 18863
    },
    {
      "epoch": 0.5435371405520659,
      "grad_norm": 0.734696090221405,
      "learning_rate": 4.538712863846478e-06,
      "loss": 0.7198,
      "step": 18864
    },
    {
      "epoch": 0.5435659540137152,
      "grad_norm": 0.7487295866012573,
      "learning_rate": 4.538248245734112e-06,
      "loss": 0.7115,
      "step": 18865
    },
    {
      "epoch": 0.5435947674753645,
      "grad_norm": 0.7187426090240479,
      "learning_rate": 4.537783631643143e-06,
      "loss": 0.7249,
      "step": 18866
    },
    {
      "epoch": 0.5436235809370138,
      "grad_norm": 0.7194507122039795,
      "learning_rate": 4.537319021577615e-06,
      "loss": 0.7175,
      "step": 18867
    },
    {
      "epoch": 0.5436523943986631,
      "grad_norm": 0.7090107798576355,
      "learning_rate": 4.536854415541576e-06,
      "loss": 0.7294,
      "step": 18868
    },
    {
      "epoch": 0.5436812078603124,
      "grad_norm": 0.7286843657493591,
      "learning_rate": 4.5363898135390735e-06,
      "loss": 0.7215,
      "step": 18869
    },
    {
      "epoch": 0.5437100213219617,
      "grad_norm": 0.7310124635696411,
      "learning_rate": 4.535925215574152e-06,
      "loss": 0.7442,
      "step": 18870
    },
    {
      "epoch": 0.543738834783611,
      "grad_norm": 0.7345659732818604,
      "learning_rate": 4.535460621650857e-06,
      "loss": 0.7107,
      "step": 18871
    },
    {
      "epoch": 0.5437676482452601,
      "grad_norm": 0.7222464680671692,
      "learning_rate": 4.534996031773236e-06,
      "loss": 0.7146,
      "step": 18872
    },
    {
      "epoch": 0.5437964617069094,
      "grad_norm": 0.7417537569999695,
      "learning_rate": 4.534531445945334e-06,
      "loss": 0.7177,
      "step": 18873
    },
    {
      "epoch": 0.5438252751685587,
      "grad_norm": 0.7478815913200378,
      "learning_rate": 4.5340668641711985e-06,
      "loss": 0.6861,
      "step": 18874
    },
    {
      "epoch": 0.543854088630208,
      "grad_norm": 0.7616520524024963,
      "learning_rate": 4.533602286454875e-06,
      "loss": 0.7087,
      "step": 18875
    },
    {
      "epoch": 0.5438829020918573,
      "grad_norm": 0.7413826584815979,
      "learning_rate": 4.533137712800409e-06,
      "loss": 0.6784,
      "step": 18876
    },
    {
      "epoch": 0.5439117155535066,
      "grad_norm": 0.7337647080421448,
      "learning_rate": 4.5326731432118475e-06,
      "loss": 0.7044,
      "step": 18877
    },
    {
      "epoch": 0.5439405290151559,
      "grad_norm": 0.7333321571350098,
      "learning_rate": 4.532208577693236e-06,
      "loss": 0.6969,
      "step": 18878
    },
    {
      "epoch": 0.5439693424768052,
      "grad_norm": 0.7509651184082031,
      "learning_rate": 4.531744016248619e-06,
      "loss": 0.7254,
      "step": 18879
    },
    {
      "epoch": 0.5439981559384545,
      "grad_norm": 0.7396859526634216,
      "learning_rate": 4.531279458882045e-06,
      "loss": 0.7228,
      "step": 18880
    },
    {
      "epoch": 0.5440269694001038,
      "grad_norm": 0.7071927785873413,
      "learning_rate": 4.530814905597557e-06,
      "loss": 0.7176,
      "step": 18881
    },
    {
      "epoch": 0.544055782861753,
      "grad_norm": 0.7394638061523438,
      "learning_rate": 4.530350356399203e-06,
      "loss": 0.7047,
      "step": 18882
    },
    {
      "epoch": 0.5440845963234023,
      "grad_norm": 0.7322835326194763,
      "learning_rate": 4.529885811291028e-06,
      "loss": 0.6991,
      "step": 18883
    },
    {
      "epoch": 0.5441134097850516,
      "grad_norm": 0.7440041303634644,
      "learning_rate": 4.529421270277076e-06,
      "loss": 0.7092,
      "step": 18884
    },
    {
      "epoch": 0.5441422232467008,
      "grad_norm": 0.717820405960083,
      "learning_rate": 4.528956733361395e-06,
      "loss": 0.6991,
      "step": 18885
    },
    {
      "epoch": 0.5441710367083501,
      "grad_norm": 0.7162706255912781,
      "learning_rate": 4.52849220054803e-06,
      "loss": 0.7094,
      "step": 18886
    },
    {
      "epoch": 0.5441998501699994,
      "grad_norm": 0.7361534237861633,
      "learning_rate": 4.528027671841027e-06,
      "loss": 0.7187,
      "step": 18887
    },
    {
      "epoch": 0.5442286636316487,
      "grad_norm": 0.7079731822013855,
      "learning_rate": 4.5275631472444324e-06,
      "loss": 0.7099,
      "step": 18888
    },
    {
      "epoch": 0.544257477093298,
      "grad_norm": 0.7386975884437561,
      "learning_rate": 4.527098626762288e-06,
      "loss": 0.7156,
      "step": 18889
    },
    {
      "epoch": 0.5442862905549473,
      "grad_norm": 0.7552595734596252,
      "learning_rate": 4.526634110398644e-06,
      "loss": 0.6978,
      "step": 18890
    },
    {
      "epoch": 0.5443151040165966,
      "grad_norm": 0.7196099162101746,
      "learning_rate": 4.526169598157543e-06,
      "loss": 0.7156,
      "step": 18891
    },
    {
      "epoch": 0.5443439174782458,
      "grad_norm": 0.7345605492591858,
      "learning_rate": 4.525705090043031e-06,
      "loss": 0.7399,
      "step": 18892
    },
    {
      "epoch": 0.5443727309398951,
      "grad_norm": 0.75101238489151,
      "learning_rate": 4.525240586059154e-06,
      "loss": 0.7061,
      "step": 18893
    },
    {
      "epoch": 0.5444015444015444,
      "grad_norm": 0.7470673322677612,
      "learning_rate": 4.524776086209957e-06,
      "loss": 0.6974,
      "step": 18894
    },
    {
      "epoch": 0.5444303578631937,
      "grad_norm": 0.7179431319236755,
      "learning_rate": 4.524311590499485e-06,
      "loss": 0.6909,
      "step": 18895
    },
    {
      "epoch": 0.544459171324843,
      "grad_norm": 0.7454071640968323,
      "learning_rate": 4.523847098931784e-06,
      "loss": 0.712,
      "step": 18896
    },
    {
      "epoch": 0.5444879847864923,
      "grad_norm": 0.7238742113113403,
      "learning_rate": 4.523382611510895e-06,
      "loss": 0.7111,
      "step": 18897
    },
    {
      "epoch": 0.5445167982481416,
      "grad_norm": 0.7567553520202637,
      "learning_rate": 4.522918128240871e-06,
      "loss": 0.7256,
      "step": 18898
    },
    {
      "epoch": 0.5445456117097908,
      "grad_norm": 0.7686258554458618,
      "learning_rate": 4.522453649125752e-06,
      "loss": 0.7133,
      "step": 18899
    },
    {
      "epoch": 0.5445744251714401,
      "grad_norm": 0.7328131198883057,
      "learning_rate": 4.5219891741695844e-06,
      "loss": 0.7152,
      "step": 18900
    },
    {
      "epoch": 0.5446032386330893,
      "grad_norm": 0.7277377843856812,
      "learning_rate": 4.521524703376414e-06,
      "loss": 0.7051,
      "step": 18901
    },
    {
      "epoch": 0.5446320520947386,
      "grad_norm": 0.7326810956001282,
      "learning_rate": 4.521060236750285e-06,
      "loss": 0.6974,
      "step": 18902
    },
    {
      "epoch": 0.5446608655563879,
      "grad_norm": 0.7266961336135864,
      "learning_rate": 4.520595774295242e-06,
      "loss": 0.7014,
      "step": 18903
    },
    {
      "epoch": 0.5446896790180372,
      "grad_norm": 0.7374905943870544,
      "learning_rate": 4.520131316015331e-06,
      "loss": 0.7038,
      "step": 18904
    },
    {
      "epoch": 0.5447184924796865,
      "grad_norm": 0.7292464971542358,
      "learning_rate": 4.519666861914596e-06,
      "loss": 0.7524,
      "step": 18905
    },
    {
      "epoch": 0.5447473059413358,
      "grad_norm": 0.7954539656639099,
      "learning_rate": 4.5192024119970825e-06,
      "loss": 0.7146,
      "step": 18906
    },
    {
      "epoch": 0.5447761194029851,
      "grad_norm": 0.71687251329422,
      "learning_rate": 4.518737966266836e-06,
      "loss": 0.6901,
      "step": 18907
    },
    {
      "epoch": 0.5448049328646344,
      "grad_norm": 0.7381632924079895,
      "learning_rate": 4.518273524727901e-06,
      "loss": 0.7264,
      "step": 18908
    },
    {
      "epoch": 0.5448337463262837,
      "grad_norm": 0.7360635995864868,
      "learning_rate": 4.517809087384321e-06,
      "loss": 0.6919,
      "step": 18909
    },
    {
      "epoch": 0.544862559787933,
      "grad_norm": 0.7536089420318604,
      "learning_rate": 4.517344654240141e-06,
      "loss": 0.7428,
      "step": 18910
    },
    {
      "epoch": 0.5448913732495823,
      "grad_norm": 0.7371735572814941,
      "learning_rate": 4.516880225299408e-06,
      "loss": 0.6881,
      "step": 18911
    },
    {
      "epoch": 0.5449201867112314,
      "grad_norm": 0.7330848574638367,
      "learning_rate": 4.5164158005661655e-06,
      "loss": 0.6858,
      "step": 18912
    },
    {
      "epoch": 0.5449490001728807,
      "grad_norm": 0.7226276993751526,
      "learning_rate": 4.515951380044458e-06,
      "loss": 0.6953,
      "step": 18913
    },
    {
      "epoch": 0.54497781363453,
      "grad_norm": 0.728577733039856,
      "learning_rate": 4.5154869637383294e-06,
      "loss": 0.7294,
      "step": 18914
    },
    {
      "epoch": 0.5450066270961793,
      "grad_norm": 0.6997393369674683,
      "learning_rate": 4.515022551651825e-06,
      "loss": 0.7158,
      "step": 18915
    },
    {
      "epoch": 0.5450354405578286,
      "grad_norm": 0.7268080115318298,
      "learning_rate": 4.514558143788991e-06,
      "loss": 0.7041,
      "step": 18916
    },
    {
      "epoch": 0.5450642540194779,
      "grad_norm": 0.739730954170227,
      "learning_rate": 4.514093740153869e-06,
      "loss": 0.745,
      "step": 18917
    },
    {
      "epoch": 0.5450930674811272,
      "grad_norm": 0.7185631990432739,
      "learning_rate": 4.513629340750505e-06,
      "loss": 0.6785,
      "step": 18918
    },
    {
      "epoch": 0.5451218809427765,
      "grad_norm": 0.7379829287528992,
      "learning_rate": 4.513164945582943e-06,
      "loss": 0.7045,
      "step": 18919
    },
    {
      "epoch": 0.5451506944044258,
      "grad_norm": 0.7801184058189392,
      "learning_rate": 4.512700554655228e-06,
      "loss": 0.7195,
      "step": 18920
    },
    {
      "epoch": 0.545179507866075,
      "grad_norm": 0.7226020097732544,
      "learning_rate": 4.512236167971401e-06,
      "loss": 0.686,
      "step": 18921
    },
    {
      "epoch": 0.5452083213277243,
      "grad_norm": 0.7390033602714539,
      "learning_rate": 4.511771785535513e-06,
      "loss": 0.7124,
      "step": 18922
    },
    {
      "epoch": 0.5452371347893736,
      "grad_norm": 0.7007957696914673,
      "learning_rate": 4.511307407351605e-06,
      "loss": 0.7269,
      "step": 18923
    },
    {
      "epoch": 0.5452659482510229,
      "grad_norm": 0.7384251356124878,
      "learning_rate": 4.5108430334237196e-06,
      "loss": 0.7352,
      "step": 18924
    },
    {
      "epoch": 0.5452947617126722,
      "grad_norm": 0.7460493445396423,
      "learning_rate": 4.510378663755903e-06,
      "loss": 0.7244,
      "step": 18925
    },
    {
      "epoch": 0.5453235751743214,
      "grad_norm": 0.713960587978363,
      "learning_rate": 4.509914298352197e-06,
      "loss": 0.7405,
      "step": 18926
    },
    {
      "epoch": 0.5453523886359707,
      "grad_norm": 0.7042672038078308,
      "learning_rate": 4.50944993721665e-06,
      "loss": 0.6971,
      "step": 18927
    },
    {
      "epoch": 0.54538120209762,
      "grad_norm": 0.7242079377174377,
      "learning_rate": 4.508985580353302e-06,
      "loss": 0.7122,
      "step": 18928
    },
    {
      "epoch": 0.5454100155592693,
      "grad_norm": 0.7414557337760925,
      "learning_rate": 4.5085212277662e-06,
      "loss": 0.715,
      "step": 18929
    },
    {
      "epoch": 0.5454388290209186,
      "grad_norm": 0.7039467096328735,
      "learning_rate": 4.508056879459387e-06,
      "loss": 0.7134,
      "step": 18930
    },
    {
      "epoch": 0.5454676424825678,
      "grad_norm": 0.7566894888877869,
      "learning_rate": 4.507592535436905e-06,
      "loss": 0.7111,
      "step": 18931
    },
    {
      "epoch": 0.5454964559442171,
      "grad_norm": 0.7231560945510864,
      "learning_rate": 4.5071281957028e-06,
      "loss": 0.6858,
      "step": 18932
    },
    {
      "epoch": 0.5455252694058664,
      "grad_norm": 0.7201685905456543,
      "learning_rate": 4.506663860261116e-06,
      "loss": 0.7167,
      "step": 18933
    },
    {
      "epoch": 0.5455540828675157,
      "grad_norm": 0.7359055280685425,
      "learning_rate": 4.5061995291158974e-06,
      "loss": 0.7063,
      "step": 18934
    },
    {
      "epoch": 0.545582896329165,
      "grad_norm": 0.7079033851623535,
      "learning_rate": 4.505735202271187e-06,
      "loss": 0.693,
      "step": 18935
    },
    {
      "epoch": 0.5456117097908143,
      "grad_norm": 0.7330724000930786,
      "learning_rate": 4.50527087973103e-06,
      "loss": 0.6904,
      "step": 18936
    },
    {
      "epoch": 0.5456405232524636,
      "grad_norm": 0.7212926149368286,
      "learning_rate": 4.504806561499468e-06,
      "loss": 0.7196,
      "step": 18937
    },
    {
      "epoch": 0.5456693367141129,
      "grad_norm": 0.7552902698516846,
      "learning_rate": 4.504342247580546e-06,
      "loss": 0.7301,
      "step": 18938
    },
    {
      "epoch": 0.5456981501757621,
      "grad_norm": 0.718562662601471,
      "learning_rate": 4.503877937978308e-06,
      "loss": 0.6926,
      "step": 18939
    },
    {
      "epoch": 0.5457269636374114,
      "grad_norm": 0.7309165596961975,
      "learning_rate": 4.503413632696797e-06,
      "loss": 0.6846,
      "step": 18940
    },
    {
      "epoch": 0.5457557770990606,
      "grad_norm": 0.7489205002784729,
      "learning_rate": 4.502949331740056e-06,
      "loss": 0.7039,
      "step": 18941
    },
    {
      "epoch": 0.5457845905607099,
      "grad_norm": 0.7406396865844727,
      "learning_rate": 4.50248503511213e-06,
      "loss": 0.7198,
      "step": 18942
    },
    {
      "epoch": 0.5458134040223592,
      "grad_norm": 0.7254811525344849,
      "learning_rate": 4.502020742817062e-06,
      "loss": 0.6936,
      "step": 18943
    },
    {
      "epoch": 0.5458422174840085,
      "grad_norm": 0.7290811538696289,
      "learning_rate": 4.501556454858894e-06,
      "loss": 0.7135,
      "step": 18944
    },
    {
      "epoch": 0.5458710309456578,
      "grad_norm": 0.746691882610321,
      "learning_rate": 4.5010921712416725e-06,
      "loss": 0.7053,
      "step": 18945
    },
    {
      "epoch": 0.5458998444073071,
      "grad_norm": 0.723373532295227,
      "learning_rate": 4.5006278919694404e-06,
      "loss": 0.7063,
      "step": 18946
    },
    {
      "epoch": 0.5459286578689564,
      "grad_norm": 0.7100918292999268,
      "learning_rate": 4.500163617046239e-06,
      "loss": 0.6967,
      "step": 18947
    },
    {
      "epoch": 0.5459574713306057,
      "grad_norm": 0.7251238822937012,
      "learning_rate": 4.499699346476112e-06,
      "loss": 0.7194,
      "step": 18948
    },
    {
      "epoch": 0.545986284792255,
      "grad_norm": 0.7432633638381958,
      "learning_rate": 4.499235080263105e-06,
      "loss": 0.7131,
      "step": 18949
    },
    {
      "epoch": 0.5460150982539043,
      "grad_norm": 0.7608281373977661,
      "learning_rate": 4.49877081841126e-06,
      "loss": 0.7255,
      "step": 18950
    },
    {
      "epoch": 0.5460439117155536,
      "grad_norm": 0.7086700797080994,
      "learning_rate": 4.49830656092462e-06,
      "loss": 0.7067,
      "step": 18951
    },
    {
      "epoch": 0.5460727251772028,
      "grad_norm": 0.7284538745880127,
      "learning_rate": 4.497842307807228e-06,
      "loss": 0.7041,
      "step": 18952
    },
    {
      "epoch": 0.546101538638852,
      "grad_norm": 0.7270760536193848,
      "learning_rate": 4.497378059063128e-06,
      "loss": 0.6954,
      "step": 18953
    },
    {
      "epoch": 0.5461303521005013,
      "grad_norm": 0.7167150378227234,
      "learning_rate": 4.496913814696362e-06,
      "loss": 0.7127,
      "step": 18954
    },
    {
      "epoch": 0.5461591655621506,
      "grad_norm": 0.7331246137619019,
      "learning_rate": 4.496449574710974e-06,
      "loss": 0.7108,
      "step": 18955
    },
    {
      "epoch": 0.5461879790237999,
      "grad_norm": 0.7413833141326904,
      "learning_rate": 4.4959853391110055e-06,
      "loss": 0.7274,
      "step": 18956
    },
    {
      "epoch": 0.5462167924854492,
      "grad_norm": 0.7064036726951599,
      "learning_rate": 4.495521107900503e-06,
      "loss": 0.7048,
      "step": 18957
    },
    {
      "epoch": 0.5462456059470985,
      "grad_norm": 0.7036018371582031,
      "learning_rate": 4.495056881083506e-06,
      "loss": 0.7179,
      "step": 18958
    },
    {
      "epoch": 0.5462744194087478,
      "grad_norm": 0.7195911407470703,
      "learning_rate": 4.49459265866406e-06,
      "loss": 0.7209,
      "step": 18959
    },
    {
      "epoch": 0.5463032328703971,
      "grad_norm": 0.7033743262290955,
      "learning_rate": 4.494128440646206e-06,
      "loss": 0.7038,
      "step": 18960
    },
    {
      "epoch": 0.5463320463320464,
      "grad_norm": 0.7320991158485413,
      "learning_rate": 4.493664227033988e-06,
      "loss": 0.7219,
      "step": 18961
    },
    {
      "epoch": 0.5463608597936956,
      "grad_norm": 0.7084595561027527,
      "learning_rate": 4.493200017831448e-06,
      "loss": 0.7285,
      "step": 18962
    },
    {
      "epoch": 0.5463896732553449,
      "grad_norm": 0.7308605313301086,
      "learning_rate": 4.49273581304263e-06,
      "loss": 0.7014,
      "step": 18963
    },
    {
      "epoch": 0.5464184867169942,
      "grad_norm": 0.7279711365699768,
      "learning_rate": 4.4922716126715736e-06,
      "loss": 0.7353,
      "step": 18964
    },
    {
      "epoch": 0.5464473001786435,
      "grad_norm": 0.7467828989028931,
      "learning_rate": 4.4918074167223265e-06,
      "loss": 0.7257,
      "step": 18965
    },
    {
      "epoch": 0.5464761136402927,
      "grad_norm": 0.7083765268325806,
      "learning_rate": 4.491343225198928e-06,
      "loss": 0.6968,
      "step": 18966
    },
    {
      "epoch": 0.546504927101942,
      "grad_norm": 0.7721121907234192,
      "learning_rate": 4.490879038105422e-06,
      "loss": 0.7012,
      "step": 18967
    },
    {
      "epoch": 0.5465337405635913,
      "grad_norm": 0.7467081546783447,
      "learning_rate": 4.490414855445849e-06,
      "loss": 0.7019,
      "step": 18968
    },
    {
      "epoch": 0.5465625540252406,
      "grad_norm": 0.7241957187652588,
      "learning_rate": 4.489950677224253e-06,
      "loss": 0.6971,
      "step": 18969
    },
    {
      "epoch": 0.5465913674868899,
      "grad_norm": 0.7008325457572937,
      "learning_rate": 4.4894865034446786e-06,
      "loss": 0.7054,
      "step": 18970
    },
    {
      "epoch": 0.5466201809485391,
      "grad_norm": 0.7358095645904541,
      "learning_rate": 4.489022334111166e-06,
      "loss": 0.7113,
      "step": 18971
    },
    {
      "epoch": 0.5466489944101884,
      "grad_norm": 0.7397467494010925,
      "learning_rate": 4.488558169227758e-06,
      "loss": 0.7254,
      "step": 18972
    },
    {
      "epoch": 0.5466778078718377,
      "grad_norm": 0.7796778678894043,
      "learning_rate": 4.488094008798498e-06,
      "loss": 0.6947,
      "step": 18973
    },
    {
      "epoch": 0.546706621333487,
      "grad_norm": 0.7296309471130371,
      "learning_rate": 4.487629852827426e-06,
      "loss": 0.7227,
      "step": 18974
    },
    {
      "epoch": 0.5467354347951363,
      "grad_norm": 0.736705482006073,
      "learning_rate": 4.487165701318587e-06,
      "loss": 0.6829,
      "step": 18975
    },
    {
      "epoch": 0.5467642482567856,
      "grad_norm": 0.8563848733901978,
      "learning_rate": 4.486701554276021e-06,
      "loss": 0.697,
      "step": 18976
    },
    {
      "epoch": 0.5467930617184349,
      "grad_norm": 0.7214453220367432,
      "learning_rate": 4.486237411703772e-06,
      "loss": 0.6906,
      "step": 18977
    },
    {
      "epoch": 0.5468218751800842,
      "grad_norm": 0.7264828085899353,
      "learning_rate": 4.485773273605882e-06,
      "loss": 0.7113,
      "step": 18978
    },
    {
      "epoch": 0.5468506886417335,
      "grad_norm": 0.7097421884536743,
      "learning_rate": 4.485309139986392e-06,
      "loss": 0.7069,
      "step": 18979
    },
    {
      "epoch": 0.5468795021033827,
      "grad_norm": 0.7244859933853149,
      "learning_rate": 4.484845010849343e-06,
      "loss": 0.7188,
      "step": 18980
    },
    {
      "epoch": 0.5469083155650319,
      "grad_norm": 0.704491138458252,
      "learning_rate": 4.484380886198782e-06,
      "loss": 0.7104,
      "step": 18981
    },
    {
      "epoch": 0.5469371290266812,
      "grad_norm": 0.7194464802742004,
      "learning_rate": 4.483916766038747e-06,
      "loss": 0.6952,
      "step": 18982
    },
    {
      "epoch": 0.5469659424883305,
      "grad_norm": 0.7324692010879517,
      "learning_rate": 4.483452650373281e-06,
      "loss": 0.706,
      "step": 18983
    },
    {
      "epoch": 0.5469947559499798,
      "grad_norm": 0.7518499493598938,
      "learning_rate": 4.482988539206426e-06,
      "loss": 0.7143,
      "step": 18984
    },
    {
      "epoch": 0.5470235694116291,
      "grad_norm": 0.7238620519638062,
      "learning_rate": 4.482524432542224e-06,
      "loss": 0.6953,
      "step": 18985
    },
    {
      "epoch": 0.5470523828732784,
      "grad_norm": 0.7330762147903442,
      "learning_rate": 4.482060330384716e-06,
      "loss": 0.6988,
      "step": 18986
    },
    {
      "epoch": 0.5470811963349277,
      "grad_norm": 0.7448194026947021,
      "learning_rate": 4.481596232737947e-06,
      "loss": 0.7137,
      "step": 18987
    },
    {
      "epoch": 0.547110009796577,
      "grad_norm": 0.7346213459968567,
      "learning_rate": 4.4811321396059555e-06,
      "loss": 0.7199,
      "step": 18988
    },
    {
      "epoch": 0.5471388232582263,
      "grad_norm": 0.720441997051239,
      "learning_rate": 4.4806680509927845e-06,
      "loss": 0.699,
      "step": 18989
    },
    {
      "epoch": 0.5471676367198756,
      "grad_norm": 0.7525836825370789,
      "learning_rate": 4.480203966902476e-06,
      "loss": 0.7096,
      "step": 18990
    },
    {
      "epoch": 0.5471964501815249,
      "grad_norm": 0.7464028000831604,
      "learning_rate": 4.479739887339071e-06,
      "loss": 0.7311,
      "step": 18991
    },
    {
      "epoch": 0.5472252636431741,
      "grad_norm": 0.7358843684196472,
      "learning_rate": 4.4792758123066085e-06,
      "loss": 0.7176,
      "step": 18992
    },
    {
      "epoch": 0.5472540771048233,
      "grad_norm": 0.7165003418922424,
      "learning_rate": 4.478811741809136e-06,
      "loss": 0.7059,
      "step": 18993
    },
    {
      "epoch": 0.5472828905664726,
      "grad_norm": 0.7314977645874023,
      "learning_rate": 4.478347675850692e-06,
      "loss": 0.7276,
      "step": 18994
    },
    {
      "epoch": 0.5473117040281219,
      "grad_norm": 0.7401649951934814,
      "learning_rate": 4.477883614435319e-06,
      "loss": 0.7048,
      "step": 18995
    },
    {
      "epoch": 0.5473405174897712,
      "grad_norm": 0.7274551391601562,
      "learning_rate": 4.477419557567058e-06,
      "loss": 0.7143,
      "step": 18996
    },
    {
      "epoch": 0.5473693309514205,
      "grad_norm": 0.725342333316803,
      "learning_rate": 4.47695550524995e-06,
      "loss": 0.7195,
      "step": 18997
    },
    {
      "epoch": 0.5473981444130698,
      "grad_norm": 0.7234530448913574,
      "learning_rate": 4.476491457488036e-06,
      "loss": 0.7223,
      "step": 18998
    },
    {
      "epoch": 0.5474269578747191,
      "grad_norm": 0.7158488631248474,
      "learning_rate": 4.476027414285359e-06,
      "loss": 0.6903,
      "step": 18999
    },
    {
      "epoch": 0.5474557713363684,
      "grad_norm": 0.725472629070282,
      "learning_rate": 4.475563375645958e-06,
      "loss": 0.7291,
      "step": 19000
    },
    {
      "epoch": 0.5474845847980176,
      "grad_norm": 0.7209864258766174,
      "learning_rate": 4.475099341573877e-06,
      "loss": 0.6892,
      "step": 19001
    },
    {
      "epoch": 0.5475133982596669,
      "grad_norm": 0.6967813372612,
      "learning_rate": 4.474635312073155e-06,
      "loss": 0.6923,
      "step": 19002
    },
    {
      "epoch": 0.5475422117213162,
      "grad_norm": 0.7331612706184387,
      "learning_rate": 4.474171287147835e-06,
      "loss": 0.7077,
      "step": 19003
    },
    {
      "epoch": 0.5475710251829655,
      "grad_norm": 0.7258742451667786,
      "learning_rate": 4.473707266801956e-06,
      "loss": 0.7004,
      "step": 19004
    },
    {
      "epoch": 0.5475998386446148,
      "grad_norm": 1.3078131675720215,
      "learning_rate": 4.4732432510395625e-06,
      "loss": 0.7427,
      "step": 19005
    },
    {
      "epoch": 0.547628652106264,
      "grad_norm": 0.7652007341384888,
      "learning_rate": 4.472779239864693e-06,
      "loss": 0.7294,
      "step": 19006
    },
    {
      "epoch": 0.5476574655679133,
      "grad_norm": 0.6916742324829102,
      "learning_rate": 4.472315233281389e-06,
      "loss": 0.6949,
      "step": 19007
    },
    {
      "epoch": 0.5476862790295626,
      "grad_norm": 0.7483316659927368,
      "learning_rate": 4.471851231293692e-06,
      "loss": 0.707,
      "step": 19008
    },
    {
      "epoch": 0.5477150924912119,
      "grad_norm": 0.7579226493835449,
      "learning_rate": 4.471387233905643e-06,
      "loss": 0.7105,
      "step": 19009
    },
    {
      "epoch": 0.5477439059528612,
      "grad_norm": 0.7458010315895081,
      "learning_rate": 4.470923241121283e-06,
      "loss": 0.7233,
      "step": 19010
    },
    {
      "epoch": 0.5477727194145104,
      "grad_norm": 0.7279612421989441,
      "learning_rate": 4.4704592529446524e-06,
      "loss": 0.7149,
      "step": 19011
    },
    {
      "epoch": 0.5478015328761597,
      "grad_norm": 0.7252085208892822,
      "learning_rate": 4.469995269379793e-06,
      "loss": 0.7033,
      "step": 19012
    },
    {
      "epoch": 0.547830346337809,
      "grad_norm": 0.697392463684082,
      "learning_rate": 4.469531290430744e-06,
      "loss": 0.6813,
      "step": 19013
    },
    {
      "epoch": 0.5478591597994583,
      "grad_norm": 0.6985365152359009,
      "learning_rate": 4.469067316101549e-06,
      "loss": 0.6832,
      "step": 19014
    },
    {
      "epoch": 0.5478879732611076,
      "grad_norm": 0.7045344710350037,
      "learning_rate": 4.468603346396245e-06,
      "loss": 0.6999,
      "step": 19015
    },
    {
      "epoch": 0.5479167867227569,
      "grad_norm": 0.7023900151252747,
      "learning_rate": 4.468139381318873e-06,
      "loss": 0.6889,
      "step": 19016
    },
    {
      "epoch": 0.5479456001844062,
      "grad_norm": 0.7027722001075745,
      "learning_rate": 4.467675420873478e-06,
      "loss": 0.6914,
      "step": 19017
    },
    {
      "epoch": 0.5479744136460555,
      "grad_norm": 0.7298775315284729,
      "learning_rate": 4.467211465064097e-06,
      "loss": 0.6963,
      "step": 19018
    },
    {
      "epoch": 0.5480032271077048,
      "grad_norm": 0.7392346858978271,
      "learning_rate": 4.466747513894772e-06,
      "loss": 0.7297,
      "step": 19019
    },
    {
      "epoch": 0.548032040569354,
      "grad_norm": 0.7066791653633118,
      "learning_rate": 4.4662835673695445e-06,
      "loss": 0.7247,
      "step": 19020
    },
    {
      "epoch": 0.5480608540310032,
      "grad_norm": 0.732882559299469,
      "learning_rate": 4.465819625492452e-06,
      "loss": 0.7123,
      "step": 19021
    },
    {
      "epoch": 0.5480896674926525,
      "grad_norm": 0.7007479071617126,
      "learning_rate": 4.465355688267537e-06,
      "loss": 0.7334,
      "step": 19022
    },
    {
      "epoch": 0.5481184809543018,
      "grad_norm": 0.7271736860275269,
      "learning_rate": 4.4648917556988395e-06,
      "loss": 0.724,
      "step": 19023
    },
    {
      "epoch": 0.5481472944159511,
      "grad_norm": 0.7268211841583252,
      "learning_rate": 4.464427827790399e-06,
      "loss": 0.7208,
      "step": 19024
    },
    {
      "epoch": 0.5481761078776004,
      "grad_norm": 0.7200124859809875,
      "learning_rate": 4.463963904546258e-06,
      "loss": 0.7228,
      "step": 19025
    },
    {
      "epoch": 0.5482049213392497,
      "grad_norm": 0.7207938432693481,
      "learning_rate": 4.463499985970455e-06,
      "loss": 0.6994,
      "step": 19026
    },
    {
      "epoch": 0.548233734800899,
      "grad_norm": 0.7016176581382751,
      "learning_rate": 4.463036072067031e-06,
      "loss": 0.6932,
      "step": 19027
    },
    {
      "epoch": 0.5482625482625483,
      "grad_norm": 0.7312155365943909,
      "learning_rate": 4.4625721628400245e-06,
      "loss": 0.7091,
      "step": 19028
    },
    {
      "epoch": 0.5482913617241976,
      "grad_norm": 0.7245822548866272,
      "learning_rate": 4.4621082582934776e-06,
      "loss": 0.7223,
      "step": 19029
    },
    {
      "epoch": 0.5483201751858469,
      "grad_norm": 0.7281293869018555,
      "learning_rate": 4.461644358431431e-06,
      "loss": 0.7138,
      "step": 19030
    },
    {
      "epoch": 0.5483489886474961,
      "grad_norm": 0.7360206246376038,
      "learning_rate": 4.461180463257925e-06,
      "loss": 0.7127,
      "step": 19031
    },
    {
      "epoch": 0.5483778021091454,
      "grad_norm": 0.715843141078949,
      "learning_rate": 4.460716572776998e-06,
      "loss": 0.7153,
      "step": 19032
    },
    {
      "epoch": 0.5484066155707946,
      "grad_norm": 0.7268772125244141,
      "learning_rate": 4.4602526869926896e-06,
      "loss": 0.6956,
      "step": 19033
    },
    {
      "epoch": 0.5484354290324439,
      "grad_norm": 0.6769657731056213,
      "learning_rate": 4.459788805909041e-06,
      "loss": 0.6703,
      "step": 19034
    },
    {
      "epoch": 0.5484642424940932,
      "grad_norm": 0.71043860912323,
      "learning_rate": 4.459324929530093e-06,
      "loss": 0.727,
      "step": 19035
    },
    {
      "epoch": 0.5484930559557425,
      "grad_norm": 0.7270376086235046,
      "learning_rate": 4.4588610578598836e-06,
      "loss": 0.7279,
      "step": 19036
    },
    {
      "epoch": 0.5485218694173918,
      "grad_norm": 0.7222435474395752,
      "learning_rate": 4.458397190902453e-06,
      "loss": 0.714,
      "step": 19037
    },
    {
      "epoch": 0.5485506828790411,
      "grad_norm": 0.7536386847496033,
      "learning_rate": 4.457933328661842e-06,
      "loss": 0.7002,
      "step": 19038
    },
    {
      "epoch": 0.5485794963406904,
      "grad_norm": 0.7157284021377563,
      "learning_rate": 4.45746947114209e-06,
      "loss": 0.6967,
      "step": 19039
    },
    {
      "epoch": 0.5486083098023397,
      "grad_norm": 0.7056930661201477,
      "learning_rate": 4.457005618347234e-06,
      "loss": 0.7083,
      "step": 19040
    },
    {
      "epoch": 0.548637123263989,
      "grad_norm": 0.7037466764450073,
      "learning_rate": 4.456541770281318e-06,
      "loss": 0.6988,
      "step": 19041
    },
    {
      "epoch": 0.5486659367256382,
      "grad_norm": 0.7235670685768127,
      "learning_rate": 4.4560779269483815e-06,
      "loss": 0.7058,
      "step": 19042
    },
    {
      "epoch": 0.5486947501872875,
      "grad_norm": 0.6873776912689209,
      "learning_rate": 4.455614088352461e-06,
      "loss": 0.6976,
      "step": 19043
    },
    {
      "epoch": 0.5487235636489368,
      "grad_norm": 0.7351391911506653,
      "learning_rate": 4.455150254497598e-06,
      "loss": 0.6948,
      "step": 19044
    },
    {
      "epoch": 0.5487523771105861,
      "grad_norm": 0.714299201965332,
      "learning_rate": 4.454686425387831e-06,
      "loss": 0.7213,
      "step": 19045
    },
    {
      "epoch": 0.5487811905722354,
      "grad_norm": 0.7395095825195312,
      "learning_rate": 4.454222601027201e-06,
      "loss": 0.7292,
      "step": 19046
    },
    {
      "epoch": 0.5488100040338846,
      "grad_norm": 0.7037145495414734,
      "learning_rate": 4.453758781419747e-06,
      "loss": 0.6922,
      "step": 19047
    },
    {
      "epoch": 0.5488388174955339,
      "grad_norm": 0.7207574844360352,
      "learning_rate": 4.453294966569508e-06,
      "loss": 0.7179,
      "step": 19048
    },
    {
      "epoch": 0.5488676309571832,
      "grad_norm": 0.7294889092445374,
      "learning_rate": 4.4528311564805225e-06,
      "loss": 0.7234,
      "step": 19049
    },
    {
      "epoch": 0.5488964444188325,
      "grad_norm": 0.7154726982116699,
      "learning_rate": 4.4523673511568305e-06,
      "loss": 0.7018,
      "step": 19050
    },
    {
      "epoch": 0.5489252578804817,
      "grad_norm": 0.7204176187515259,
      "learning_rate": 4.451903550602472e-06,
      "loss": 0.6918,
      "step": 19051
    },
    {
      "epoch": 0.548954071342131,
      "grad_norm": 0.7489952445030212,
      "learning_rate": 4.451439754821483e-06,
      "loss": 0.701,
      "step": 19052
    },
    {
      "epoch": 0.5489828848037803,
      "grad_norm": 0.7315383553504944,
      "learning_rate": 4.450975963817908e-06,
      "loss": 0.6882,
      "step": 19053
    },
    {
      "epoch": 0.5490116982654296,
      "grad_norm": 0.7392570972442627,
      "learning_rate": 4.450512177595784e-06,
      "loss": 0.7265,
      "step": 19054
    },
    {
      "epoch": 0.5490405117270789,
      "grad_norm": 0.7158730030059814,
      "learning_rate": 4.4500483961591486e-06,
      "loss": 0.6967,
      "step": 19055
    },
    {
      "epoch": 0.5490693251887282,
      "grad_norm": 0.7256494164466858,
      "learning_rate": 4.449584619512043e-06,
      "loss": 0.7166,
      "step": 19056
    },
    {
      "epoch": 0.5490981386503775,
      "grad_norm": 0.7156983017921448,
      "learning_rate": 4.449120847658505e-06,
      "loss": 0.704,
      "step": 19057
    },
    {
      "epoch": 0.5491269521120268,
      "grad_norm": 0.7789025902748108,
      "learning_rate": 4.448657080602573e-06,
      "loss": 0.7122,
      "step": 19058
    },
    {
      "epoch": 0.5491557655736761,
      "grad_norm": 0.7107499837875366,
      "learning_rate": 4.448193318348287e-06,
      "loss": 0.7254,
      "step": 19059
    },
    {
      "epoch": 0.5491845790353252,
      "grad_norm": 0.7101681232452393,
      "learning_rate": 4.447729560899686e-06,
      "loss": 0.703,
      "step": 19060
    },
    {
      "epoch": 0.5492133924969745,
      "grad_norm": 0.714267373085022,
      "learning_rate": 4.447265808260808e-06,
      "loss": 0.7087,
      "step": 19061
    },
    {
      "epoch": 0.5492422059586238,
      "grad_norm": 0.7173807621002197,
      "learning_rate": 4.446802060435691e-06,
      "loss": 0.7145,
      "step": 19062
    },
    {
      "epoch": 0.5492710194202731,
      "grad_norm": 0.7471798658370972,
      "learning_rate": 4.446338317428377e-06,
      "loss": 0.7176,
      "step": 19063
    },
    {
      "epoch": 0.5492998328819224,
      "grad_norm": 0.7158922553062439,
      "learning_rate": 4.4458745792429005e-06,
      "loss": 0.7208,
      "step": 19064
    },
    {
      "epoch": 0.5493286463435717,
      "grad_norm": 0.7193783521652222,
      "learning_rate": 4.445410845883305e-06,
      "loss": 0.7123,
      "step": 19065
    },
    {
      "epoch": 0.549357459805221,
      "grad_norm": 0.7076957821846008,
      "learning_rate": 4.444947117353625e-06,
      "loss": 0.702,
      "step": 19066
    },
    {
      "epoch": 0.5493862732668703,
      "grad_norm": 0.7274926900863647,
      "learning_rate": 4.444483393657901e-06,
      "loss": 0.6982,
      "step": 19067
    },
    {
      "epoch": 0.5494150867285196,
      "grad_norm": 0.757045328617096,
      "learning_rate": 4.444019674800172e-06,
      "loss": 0.7128,
      "step": 19068
    },
    {
      "epoch": 0.5494439001901689,
      "grad_norm": 0.7143086791038513,
      "learning_rate": 4.443555960784476e-06,
      "loss": 0.7163,
      "step": 19069
    },
    {
      "epoch": 0.5494727136518182,
      "grad_norm": 0.7646188735961914,
      "learning_rate": 4.443092251614851e-06,
      "loss": 0.7274,
      "step": 19070
    },
    {
      "epoch": 0.5495015271134674,
      "grad_norm": 0.7288180589675903,
      "learning_rate": 4.442628547295336e-06,
      "loss": 0.696,
      "step": 19071
    },
    {
      "epoch": 0.5495303405751167,
      "grad_norm": 2.1814863681793213,
      "learning_rate": 4.442164847829969e-06,
      "loss": 0.7053,
      "step": 19072
    },
    {
      "epoch": 0.549559154036766,
      "grad_norm": 0.7360804080963135,
      "learning_rate": 4.441701153222789e-06,
      "loss": 0.7251,
      "step": 19073
    },
    {
      "epoch": 0.5495879674984152,
      "grad_norm": 0.7502268552780151,
      "learning_rate": 4.441237463477833e-06,
      "loss": 0.7241,
      "step": 19074
    },
    {
      "epoch": 0.5496167809600645,
      "grad_norm": 0.7421555519104004,
      "learning_rate": 4.440773778599141e-06,
      "loss": 0.71,
      "step": 19075
    },
    {
      "epoch": 0.5496455944217138,
      "grad_norm": 0.7545925378799438,
      "learning_rate": 4.4403100985907475e-06,
      "loss": 0.7489,
      "step": 19076
    },
    {
      "epoch": 0.5496744078833631,
      "grad_norm": 0.7359157204627991,
      "learning_rate": 4.439846423456696e-06,
      "loss": 0.712,
      "step": 19077
    },
    {
      "epoch": 0.5497032213450124,
      "grad_norm": 0.7493778467178345,
      "learning_rate": 4.439382753201023e-06,
      "loss": 0.7063,
      "step": 19078
    },
    {
      "epoch": 0.5497320348066617,
      "grad_norm": 0.7314649820327759,
      "learning_rate": 4.438919087827765e-06,
      "loss": 0.7058,
      "step": 19079
    },
    {
      "epoch": 0.549760848268311,
      "grad_norm": 0.7316979765892029,
      "learning_rate": 4.4384554273409605e-06,
      "loss": 0.7238,
      "step": 19080
    },
    {
      "epoch": 0.5497896617299602,
      "grad_norm": 0.7146201133728027,
      "learning_rate": 4.437991771744648e-06,
      "loss": 0.7171,
      "step": 19081
    },
    {
      "epoch": 0.5498184751916095,
      "grad_norm": 0.6952105164527893,
      "learning_rate": 4.437528121042867e-06,
      "loss": 0.6788,
      "step": 19082
    },
    {
      "epoch": 0.5498472886532588,
      "grad_norm": 0.7232989072799683,
      "learning_rate": 4.4370644752396504e-06,
      "loss": 0.7139,
      "step": 19083
    },
    {
      "epoch": 0.5498761021149081,
      "grad_norm": 0.7259352803230286,
      "learning_rate": 4.436600834339042e-06,
      "loss": 0.6981,
      "step": 19084
    },
    {
      "epoch": 0.5499049155765574,
      "grad_norm": 0.7345641851425171,
      "learning_rate": 4.436137198345077e-06,
      "loss": 0.7236,
      "step": 19085
    },
    {
      "epoch": 0.5499337290382067,
      "grad_norm": 0.7295746803283691,
      "learning_rate": 4.4356735672617926e-06,
      "loss": 0.7,
      "step": 19086
    },
    {
      "epoch": 0.5499625424998559,
      "grad_norm": 0.7480249404907227,
      "learning_rate": 4.435209941093228e-06,
      "loss": 0.726,
      "step": 19087
    },
    {
      "epoch": 0.5499913559615052,
      "grad_norm": 0.7417561411857605,
      "learning_rate": 4.434746319843418e-06,
      "loss": 0.7007,
      "step": 19088
    },
    {
      "epoch": 0.5500201694231545,
      "grad_norm": 0.73281329870224,
      "learning_rate": 4.434282703516404e-06,
      "loss": 0.7298,
      "step": 19089
    },
    {
      "epoch": 0.5500489828848037,
      "grad_norm": 0.7205604314804077,
      "learning_rate": 4.433819092116223e-06,
      "loss": 0.7166,
      "step": 19090
    },
    {
      "epoch": 0.550077796346453,
      "grad_norm": 0.7459983229637146,
      "learning_rate": 4.433355485646912e-06,
      "loss": 0.6811,
      "step": 19091
    },
    {
      "epoch": 0.5501066098081023,
      "grad_norm": 0.7103729844093323,
      "learning_rate": 4.432891884112507e-06,
      "loss": 0.722,
      "step": 19092
    },
    {
      "epoch": 0.5501354232697516,
      "grad_norm": 0.7483821511268616,
      "learning_rate": 4.4324282875170475e-06,
      "loss": 0.7271,
      "step": 19093
    },
    {
      "epoch": 0.5501642367314009,
      "grad_norm": 0.7301467061042786,
      "learning_rate": 4.43196469586457e-06,
      "loss": 0.6962,
      "step": 19094
    },
    {
      "epoch": 0.5501930501930502,
      "grad_norm": 0.7390128374099731,
      "learning_rate": 4.431501109159113e-06,
      "loss": 0.7098,
      "step": 19095
    },
    {
      "epoch": 0.5502218636546995,
      "grad_norm": 0.7040221095085144,
      "learning_rate": 4.431037527404712e-06,
      "loss": 0.6855,
      "step": 19096
    },
    {
      "epoch": 0.5502506771163488,
      "grad_norm": 0.7520084977149963,
      "learning_rate": 4.430573950605406e-06,
      "loss": 0.7194,
      "step": 19097
    },
    {
      "epoch": 0.5502794905779981,
      "grad_norm": 0.7037469744682312,
      "learning_rate": 4.430110378765232e-06,
      "loss": 0.7178,
      "step": 19098
    },
    {
      "epoch": 0.5503083040396474,
      "grad_norm": 0.7173448801040649,
      "learning_rate": 4.429646811888225e-06,
      "loss": 0.7294,
      "step": 19099
    },
    {
      "epoch": 0.5503371175012967,
      "grad_norm": 0.7241294980049133,
      "learning_rate": 4.429183249978425e-06,
      "loss": 0.7045,
      "step": 19100
    },
    {
      "epoch": 0.5503659309629458,
      "grad_norm": 0.7295152544975281,
      "learning_rate": 4.428719693039869e-06,
      "loss": 0.6955,
      "step": 19101
    },
    {
      "epoch": 0.5503947444245951,
      "grad_norm": 0.73685622215271,
      "learning_rate": 4.428256141076593e-06,
      "loss": 0.7161,
      "step": 19102
    },
    {
      "epoch": 0.5504235578862444,
      "grad_norm": 0.7344449758529663,
      "learning_rate": 4.427792594092635e-06,
      "loss": 0.7263,
      "step": 19103
    },
    {
      "epoch": 0.5504523713478937,
      "grad_norm": 0.7268804311752319,
      "learning_rate": 4.42732905209203e-06,
      "loss": 0.7014,
      "step": 19104
    },
    {
      "epoch": 0.550481184809543,
      "grad_norm": 0.7218884229660034,
      "learning_rate": 4.4268655150788176e-06,
      "loss": 0.6963,
      "step": 19105
    },
    {
      "epoch": 0.5505099982711923,
      "grad_norm": 0.729144275188446,
      "learning_rate": 4.426401983057034e-06,
      "loss": 0.7126,
      "step": 19106
    },
    {
      "epoch": 0.5505388117328416,
      "grad_norm": 0.7306219935417175,
      "learning_rate": 4.425938456030715e-06,
      "loss": 0.7261,
      "step": 19107
    },
    {
      "epoch": 0.5505676251944909,
      "grad_norm": 0.7282046675682068,
      "learning_rate": 4.425474934003899e-06,
      "loss": 0.7081,
      "step": 19108
    },
    {
      "epoch": 0.5505964386561402,
      "grad_norm": 0.9532768130302429,
      "learning_rate": 4.425011416980622e-06,
      "loss": 0.7351,
      "step": 19109
    },
    {
      "epoch": 0.5506252521177895,
      "grad_norm": 0.7016297578811646,
      "learning_rate": 4.42454790496492e-06,
      "loss": 0.7571,
      "step": 19110
    },
    {
      "epoch": 0.5506540655794387,
      "grad_norm": 0.717362642288208,
      "learning_rate": 4.424084397960831e-06,
      "loss": 0.7269,
      "step": 19111
    },
    {
      "epoch": 0.550682879041088,
      "grad_norm": 0.7760412693023682,
      "learning_rate": 4.423620895972389e-06,
      "loss": 0.7223,
      "step": 19112
    },
    {
      "epoch": 0.5507116925027373,
      "grad_norm": 0.7021512389183044,
      "learning_rate": 4.423157399003635e-06,
      "loss": 0.7177,
      "step": 19113
    },
    {
      "epoch": 0.5507405059643865,
      "grad_norm": 0.7266958951950073,
      "learning_rate": 4.422693907058604e-06,
      "loss": 0.728,
      "step": 19114
    },
    {
      "epoch": 0.5507693194260358,
      "grad_norm": 0.7388759851455688,
      "learning_rate": 4.422230420141331e-06,
      "loss": 0.7134,
      "step": 19115
    },
    {
      "epoch": 0.5507981328876851,
      "grad_norm": 0.7295351028442383,
      "learning_rate": 4.421766938255855e-06,
      "loss": 0.7126,
      "step": 19116
    },
    {
      "epoch": 0.5508269463493344,
      "grad_norm": 0.794058620929718,
      "learning_rate": 4.421303461406209e-06,
      "loss": 0.7358,
      "step": 19117
    },
    {
      "epoch": 0.5508557598109837,
      "grad_norm": 0.7055305242538452,
      "learning_rate": 4.4208399895964325e-06,
      "loss": 0.7139,
      "step": 19118
    },
    {
      "epoch": 0.550884573272633,
      "grad_norm": 0.708304762840271,
      "learning_rate": 4.42037652283056e-06,
      "loss": 0.7179,
      "step": 19119
    },
    {
      "epoch": 0.5509133867342823,
      "grad_norm": 0.7062249779701233,
      "learning_rate": 4.419913061112629e-06,
      "loss": 0.7153,
      "step": 19120
    },
    {
      "epoch": 0.5509422001959315,
      "grad_norm": 0.739206075668335,
      "learning_rate": 4.419449604446675e-06,
      "loss": 0.7054,
      "step": 19121
    },
    {
      "epoch": 0.5509710136575808,
      "grad_norm": 0.7084567546844482,
      "learning_rate": 4.418986152836735e-06,
      "loss": 0.6942,
      "step": 19122
    },
    {
      "epoch": 0.5509998271192301,
      "grad_norm": 0.7907010912895203,
      "learning_rate": 4.4185227062868444e-06,
      "loss": 0.7051,
      "step": 19123
    },
    {
      "epoch": 0.5510286405808794,
      "grad_norm": 0.7367691993713379,
      "learning_rate": 4.418059264801039e-06,
      "loss": 0.7355,
      "step": 19124
    },
    {
      "epoch": 0.5510574540425287,
      "grad_norm": 0.7374677062034607,
      "learning_rate": 4.417595828383357e-06,
      "loss": 0.7152,
      "step": 19125
    },
    {
      "epoch": 0.551086267504178,
      "grad_norm": 0.7130693793296814,
      "learning_rate": 4.417132397037832e-06,
      "loss": 0.7075,
      "step": 19126
    },
    {
      "epoch": 0.5511150809658273,
      "grad_norm": 0.7044848799705505,
      "learning_rate": 4.4166689707685014e-06,
      "loss": 0.6937,
      "step": 19127
    },
    {
      "epoch": 0.5511438944274765,
      "grad_norm": 0.718100905418396,
      "learning_rate": 4.416205549579402e-06,
      "loss": 0.714,
      "step": 19128
    },
    {
      "epoch": 0.5511727078891258,
      "grad_norm": 0.7347458004951477,
      "learning_rate": 4.415742133474569e-06,
      "loss": 0.7232,
      "step": 19129
    },
    {
      "epoch": 0.551201521350775,
      "grad_norm": 0.7250934839248657,
      "learning_rate": 4.415278722458037e-06,
      "loss": 0.7195,
      "step": 19130
    },
    {
      "epoch": 0.5512303348124243,
      "grad_norm": 0.7176128029823303,
      "learning_rate": 4.414815316533843e-06,
      "loss": 0.7294,
      "step": 19131
    },
    {
      "epoch": 0.5512591482740736,
      "grad_norm": 0.7347831130027771,
      "learning_rate": 4.414351915706023e-06,
      "loss": 0.6833,
      "step": 19132
    },
    {
      "epoch": 0.5512879617357229,
      "grad_norm": 0.7233069539070129,
      "learning_rate": 4.413888519978612e-06,
      "loss": 0.7225,
      "step": 19133
    },
    {
      "epoch": 0.5513167751973722,
      "grad_norm": 0.7321595549583435,
      "learning_rate": 4.413425129355646e-06,
      "loss": 0.7089,
      "step": 19134
    },
    {
      "epoch": 0.5513455886590215,
      "grad_norm": 0.7402158379554749,
      "learning_rate": 4.412961743841162e-06,
      "loss": 0.7169,
      "step": 19135
    },
    {
      "epoch": 0.5513744021206708,
      "grad_norm": 0.7062199711799622,
      "learning_rate": 4.412498363439191e-06,
      "loss": 0.6865,
      "step": 19136
    },
    {
      "epoch": 0.5514032155823201,
      "grad_norm": 0.7478776574134827,
      "learning_rate": 4.4120349881537735e-06,
      "loss": 0.7276,
      "step": 19137
    },
    {
      "epoch": 0.5514320290439694,
      "grad_norm": 0.741468071937561,
      "learning_rate": 4.411571617988945e-06,
      "loss": 0.7134,
      "step": 19138
    },
    {
      "epoch": 0.5514608425056187,
      "grad_norm": 0.7259402871131897,
      "learning_rate": 4.411108252948739e-06,
      "loss": 0.7277,
      "step": 19139
    },
    {
      "epoch": 0.551489655967268,
      "grad_norm": 0.7298531532287598,
      "learning_rate": 4.410644893037191e-06,
      "loss": 0.7137,
      "step": 19140
    },
    {
      "epoch": 0.5515184694289171,
      "grad_norm": 0.7453038096427917,
      "learning_rate": 4.410181538258338e-06,
      "loss": 0.6898,
      "step": 19141
    },
    {
      "epoch": 0.5515472828905664,
      "grad_norm": 0.7381784915924072,
      "learning_rate": 4.409718188616211e-06,
      "loss": 0.7151,
      "step": 19142
    },
    {
      "epoch": 0.5515760963522157,
      "grad_norm": 0.708397626876831,
      "learning_rate": 4.409254844114851e-06,
      "loss": 0.7141,
      "step": 19143
    },
    {
      "epoch": 0.551604909813865,
      "grad_norm": 0.696118175983429,
      "learning_rate": 4.4087915047582916e-06,
      "loss": 0.713,
      "step": 19144
    },
    {
      "epoch": 0.5516337232755143,
      "grad_norm": 0.740415632724762,
      "learning_rate": 4.408328170550566e-06,
      "loss": 0.7049,
      "step": 19145
    },
    {
      "epoch": 0.5516625367371636,
      "grad_norm": 0.7197487354278564,
      "learning_rate": 4.407864841495711e-06,
      "loss": 0.6998,
      "step": 19146
    },
    {
      "epoch": 0.5516913501988129,
      "grad_norm": 0.7481293678283691,
      "learning_rate": 4.407401517597761e-06,
      "loss": 0.732,
      "step": 19147
    },
    {
      "epoch": 0.5517201636604622,
      "grad_norm": 0.7137627005577087,
      "learning_rate": 4.40693819886075e-06,
      "loss": 0.7182,
      "step": 19148
    },
    {
      "epoch": 0.5517489771221115,
      "grad_norm": 0.700944721698761,
      "learning_rate": 4.406474885288715e-06,
      "loss": 0.6861,
      "step": 19149
    },
    {
      "epoch": 0.5517777905837608,
      "grad_norm": 0.731367290019989,
      "learning_rate": 4.406011576885692e-06,
      "loss": 0.7148,
      "step": 19150
    },
    {
      "epoch": 0.55180660404541,
      "grad_norm": 0.7145614624023438,
      "learning_rate": 4.405548273655713e-06,
      "loss": 0.7231,
      "step": 19151
    },
    {
      "epoch": 0.5518354175070593,
      "grad_norm": 0.716367244720459,
      "learning_rate": 4.405084975602816e-06,
      "loss": 0.7068,
      "step": 19152
    },
    {
      "epoch": 0.5518642309687086,
      "grad_norm": 0.7437546849250793,
      "learning_rate": 4.4046216827310324e-06,
      "loss": 0.7037,
      "step": 19153
    },
    {
      "epoch": 0.5518930444303579,
      "grad_norm": 1.0467537641525269,
      "learning_rate": 4.404158395044401e-06,
      "loss": 0.6826,
      "step": 19154
    },
    {
      "epoch": 0.5519218578920071,
      "grad_norm": 0.7099074721336365,
      "learning_rate": 4.403695112546952e-06,
      "loss": 0.6956,
      "step": 19155
    },
    {
      "epoch": 0.5519506713536564,
      "grad_norm": 0.7140196561813354,
      "learning_rate": 4.403231835242724e-06,
      "loss": 0.7271,
      "step": 19156
    },
    {
      "epoch": 0.5519794848153057,
      "grad_norm": 0.7334246635437012,
      "learning_rate": 4.40276856313575e-06,
      "loss": 0.7019,
      "step": 19157
    },
    {
      "epoch": 0.552008298276955,
      "grad_norm": 0.7395056486129761,
      "learning_rate": 4.402305296230064e-06,
      "loss": 0.7241,
      "step": 19158
    },
    {
      "epoch": 0.5520371117386043,
      "grad_norm": 0.7318294048309326,
      "learning_rate": 4.401842034529701e-06,
      "loss": 0.718,
      "step": 19159
    },
    {
      "epoch": 0.5520659252002535,
      "grad_norm": 0.7223106026649475,
      "learning_rate": 4.401378778038696e-06,
      "loss": 0.6962,
      "step": 19160
    },
    {
      "epoch": 0.5520947386619028,
      "grad_norm": 0.7429628968238831,
      "learning_rate": 4.400915526761084e-06,
      "loss": 0.7256,
      "step": 19161
    },
    {
      "epoch": 0.5521235521235521,
      "grad_norm": 0.7031728029251099,
      "learning_rate": 4.4004522807009e-06,
      "loss": 0.6789,
      "step": 19162
    },
    {
      "epoch": 0.5521523655852014,
      "grad_norm": 0.7269309759140015,
      "learning_rate": 4.399989039862176e-06,
      "loss": 0.7062,
      "step": 19163
    },
    {
      "epoch": 0.5521811790468507,
      "grad_norm": 0.7186570763587952,
      "learning_rate": 4.399525804248947e-06,
      "loss": 0.7093,
      "step": 19164
    },
    {
      "epoch": 0.5522099925085,
      "grad_norm": 0.7386823892593384,
      "learning_rate": 4.399062573865249e-06,
      "loss": 0.7088,
      "step": 19165
    },
    {
      "epoch": 0.5522388059701493,
      "grad_norm": 0.7923043966293335,
      "learning_rate": 4.3985993487151155e-06,
      "loss": 0.7118,
      "step": 19166
    },
    {
      "epoch": 0.5522676194317986,
      "grad_norm": 0.7054582238197327,
      "learning_rate": 4.398136128802581e-06,
      "loss": 0.6985,
      "step": 19167
    },
    {
      "epoch": 0.5522964328934478,
      "grad_norm": 0.702721357345581,
      "learning_rate": 4.397672914131677e-06,
      "loss": 0.7046,
      "step": 19168
    },
    {
      "epoch": 0.552325246355097,
      "grad_norm": 0.9454706907272339,
      "learning_rate": 4.397209704706442e-06,
      "loss": 0.6963,
      "step": 19169
    },
    {
      "epoch": 0.5523540598167463,
      "grad_norm": 0.7001374363899231,
      "learning_rate": 4.396746500530907e-06,
      "loss": 0.6961,
      "step": 19170
    },
    {
      "epoch": 0.5523828732783956,
      "grad_norm": 0.7656009793281555,
      "learning_rate": 4.396283301609107e-06,
      "loss": 0.6816,
      "step": 19171
    },
    {
      "epoch": 0.5524116867400449,
      "grad_norm": 0.7366096377372742,
      "learning_rate": 4.395820107945075e-06,
      "loss": 0.722,
      "step": 19172
    },
    {
      "epoch": 0.5524405002016942,
      "grad_norm": 0.7049761414527893,
      "learning_rate": 4.395356919542847e-06,
      "loss": 0.7123,
      "step": 19173
    },
    {
      "epoch": 0.5524693136633435,
      "grad_norm": 0.7431967258453369,
      "learning_rate": 4.394893736406456e-06,
      "loss": 0.7249,
      "step": 19174
    },
    {
      "epoch": 0.5524981271249928,
      "grad_norm": 0.7225586771965027,
      "learning_rate": 4.394430558539935e-06,
      "loss": 0.7447,
      "step": 19175
    },
    {
      "epoch": 0.5525269405866421,
      "grad_norm": 0.685982346534729,
      "learning_rate": 4.3939673859473195e-06,
      "loss": 0.7127,
      "step": 19176
    },
    {
      "epoch": 0.5525557540482914,
      "grad_norm": 0.7074520587921143,
      "learning_rate": 4.3935042186326415e-06,
      "loss": 0.7089,
      "step": 19177
    },
    {
      "epoch": 0.5525845675099407,
      "grad_norm": 0.7556540966033936,
      "learning_rate": 4.393041056599936e-06,
      "loss": 0.7523,
      "step": 19178
    },
    {
      "epoch": 0.55261338097159,
      "grad_norm": 0.7086122035980225,
      "learning_rate": 4.392577899853236e-06,
      "loss": 0.7023,
      "step": 19179
    },
    {
      "epoch": 0.5526421944332393,
      "grad_norm": 0.7363467216491699,
      "learning_rate": 4.392114748396574e-06,
      "loss": 0.7082,
      "step": 19180
    },
    {
      "epoch": 0.5526710078948885,
      "grad_norm": 0.7285128831863403,
      "learning_rate": 4.391651602233986e-06,
      "loss": 0.6861,
      "step": 19181
    },
    {
      "epoch": 0.5526998213565377,
      "grad_norm": 0.7153435349464417,
      "learning_rate": 4.3911884613695045e-06,
      "loss": 0.6862,
      "step": 19182
    },
    {
      "epoch": 0.552728634818187,
      "grad_norm": 0.6999847292900085,
      "learning_rate": 4.390725325807163e-06,
      "loss": 0.7091,
      "step": 19183
    },
    {
      "epoch": 0.5527574482798363,
      "grad_norm": 0.6978520750999451,
      "learning_rate": 4.3902621955509936e-06,
      "loss": 0.7176,
      "step": 19184
    },
    {
      "epoch": 0.5527862617414856,
      "grad_norm": 0.7489490509033203,
      "learning_rate": 4.3897990706050315e-06,
      "loss": 0.7119,
      "step": 19185
    },
    {
      "epoch": 0.5528150752031349,
      "grad_norm": 0.7035977244377136,
      "learning_rate": 4.38933595097331e-06,
      "loss": 0.6828,
      "step": 19186
    },
    {
      "epoch": 0.5528438886647842,
      "grad_norm": 0.7191630005836487,
      "learning_rate": 4.388872836659862e-06,
      "loss": 0.6984,
      "step": 19187
    },
    {
      "epoch": 0.5528727021264335,
      "grad_norm": 0.900246262550354,
      "learning_rate": 4.388409727668721e-06,
      "loss": 0.7087,
      "step": 19188
    },
    {
      "epoch": 0.5529015155880828,
      "grad_norm": 0.7249576449394226,
      "learning_rate": 4.38794662400392e-06,
      "loss": 0.7017,
      "step": 19189
    },
    {
      "epoch": 0.552930329049732,
      "grad_norm": 0.7269291281700134,
      "learning_rate": 4.387483525669492e-06,
      "loss": 0.7155,
      "step": 19190
    },
    {
      "epoch": 0.5529591425113813,
      "grad_norm": 0.7184542417526245,
      "learning_rate": 4.3870204326694695e-06,
      "loss": 0.7067,
      "step": 19191
    },
    {
      "epoch": 0.5529879559730306,
      "grad_norm": 0.7212217450141907,
      "learning_rate": 4.386557345007887e-06,
      "loss": 0.7029,
      "step": 19192
    },
    {
      "epoch": 0.5530167694346799,
      "grad_norm": 0.7741551995277405,
      "learning_rate": 4.386094262688776e-06,
      "loss": 0.7129,
      "step": 19193
    },
    {
      "epoch": 0.5530455828963292,
      "grad_norm": 0.7476839423179626,
      "learning_rate": 4.385631185716171e-06,
      "loss": 0.7244,
      "step": 19194
    },
    {
      "epoch": 0.5530743963579784,
      "grad_norm": 0.7185420989990234,
      "learning_rate": 4.385168114094102e-06,
      "loss": 0.7135,
      "step": 19195
    },
    {
      "epoch": 0.5531032098196277,
      "grad_norm": 0.7001570463180542,
      "learning_rate": 4.384705047826606e-06,
      "loss": 0.6912,
      "step": 19196
    },
    {
      "epoch": 0.553132023281277,
      "grad_norm": 0.7279003858566284,
      "learning_rate": 4.384241986917714e-06,
      "loss": 0.7109,
      "step": 19197
    },
    {
      "epoch": 0.5531608367429263,
      "grad_norm": 0.7624305486679077,
      "learning_rate": 4.383778931371459e-06,
      "loss": 0.7478,
      "step": 19198
    },
    {
      "epoch": 0.5531896502045756,
      "grad_norm": 0.722989022731781,
      "learning_rate": 4.383315881191874e-06,
      "loss": 0.7227,
      "step": 19199
    },
    {
      "epoch": 0.5532184636662248,
      "grad_norm": 0.7054615616798401,
      "learning_rate": 4.382852836382991e-06,
      "loss": 0.7006,
      "step": 19200
    },
    {
      "epoch": 0.5532472771278741,
      "grad_norm": 0.6999002695083618,
      "learning_rate": 4.382389796948842e-06,
      "loss": 0.7164,
      "step": 19201
    },
    {
      "epoch": 0.5532760905895234,
      "grad_norm": 0.7188425064086914,
      "learning_rate": 4.3819267628934595e-06,
      "loss": 0.7125,
      "step": 19202
    },
    {
      "epoch": 0.5533049040511727,
      "grad_norm": 0.7286154627799988,
      "learning_rate": 4.3814637342208785e-06,
      "loss": 0.6952,
      "step": 19203
    },
    {
      "epoch": 0.553333717512822,
      "grad_norm": 0.7140545845031738,
      "learning_rate": 4.381000710935131e-06,
      "loss": 0.7149,
      "step": 19204
    },
    {
      "epoch": 0.5533625309744713,
      "grad_norm": 0.7058345675468445,
      "learning_rate": 4.380537693040247e-06,
      "loss": 0.6862,
      "step": 19205
    },
    {
      "epoch": 0.5533913444361206,
      "grad_norm": 0.7207743525505066,
      "learning_rate": 4.380074680540261e-06,
      "loss": 0.7059,
      "step": 19206
    },
    {
      "epoch": 0.5534201578977699,
      "grad_norm": 0.7527086734771729,
      "learning_rate": 4.379611673439203e-06,
      "loss": 0.7158,
      "step": 19207
    },
    {
      "epoch": 0.5534489713594192,
      "grad_norm": 0.7518098950386047,
      "learning_rate": 4.379148671741109e-06,
      "loss": 0.7179,
      "step": 19208
    },
    {
      "epoch": 0.5534777848210684,
      "grad_norm": 0.7220008373260498,
      "learning_rate": 4.37868567545001e-06,
      "loss": 0.6907,
      "step": 19209
    },
    {
      "epoch": 0.5535065982827176,
      "grad_norm": 0.7101418375968933,
      "learning_rate": 4.3782226845699375e-06,
      "loss": 0.7068,
      "step": 19210
    },
    {
      "epoch": 0.5535354117443669,
      "grad_norm": 0.7387180328369141,
      "learning_rate": 4.377759699104923e-06,
      "loss": 0.7077,
      "step": 19211
    },
    {
      "epoch": 0.5535642252060162,
      "grad_norm": 0.7201899886131287,
      "learning_rate": 4.377296719059e-06,
      "loss": 0.7167,
      "step": 19212
    },
    {
      "epoch": 0.5535930386676655,
      "grad_norm": 0.7451242208480835,
      "learning_rate": 4.3768337444362006e-06,
      "loss": 0.7009,
      "step": 19213
    },
    {
      "epoch": 0.5536218521293148,
      "grad_norm": 0.743739128112793,
      "learning_rate": 4.376370775240556e-06,
      "loss": 0.7438,
      "step": 19214
    },
    {
      "epoch": 0.5536506655909641,
      "grad_norm": 0.7418831586837769,
      "learning_rate": 4.375907811476098e-06,
      "loss": 0.7344,
      "step": 19215
    },
    {
      "epoch": 0.5536794790526134,
      "grad_norm": 0.7094399929046631,
      "learning_rate": 4.37544485314686e-06,
      "loss": 0.6912,
      "step": 19216
    },
    {
      "epoch": 0.5537082925142627,
      "grad_norm": 0.7057384848594666,
      "learning_rate": 4.3749819002568725e-06,
      "loss": 0.6994,
      "step": 19217
    },
    {
      "epoch": 0.553737105975912,
      "grad_norm": 0.7157174348831177,
      "learning_rate": 4.374518952810167e-06,
      "loss": 0.7142,
      "step": 19218
    },
    {
      "epoch": 0.5537659194375613,
      "grad_norm": 0.7388383746147156,
      "learning_rate": 4.374056010810776e-06,
      "loss": 0.7072,
      "step": 19219
    },
    {
      "epoch": 0.5537947328992106,
      "grad_norm": 0.7348921298980713,
      "learning_rate": 4.373593074262733e-06,
      "loss": 0.7088,
      "step": 19220
    },
    {
      "epoch": 0.5538235463608598,
      "grad_norm": 0.8228987455368042,
      "learning_rate": 4.3731301431700676e-06,
      "loss": 0.7522,
      "step": 19221
    },
    {
      "epoch": 0.553852359822509,
      "grad_norm": 0.7521917223930359,
      "learning_rate": 4.372667217536812e-06,
      "loss": 0.693,
      "step": 19222
    },
    {
      "epoch": 0.5538811732841583,
      "grad_norm": 0.7384029626846313,
      "learning_rate": 4.3722042973669966e-06,
      "loss": 0.6947,
      "step": 19223
    },
    {
      "epoch": 0.5539099867458076,
      "grad_norm": 0.7279737591743469,
      "learning_rate": 4.371741382664656e-06,
      "loss": 0.7014,
      "step": 19224
    },
    {
      "epoch": 0.5539388002074569,
      "grad_norm": 0.7127251625061035,
      "learning_rate": 4.371278473433819e-06,
      "loss": 0.6833,
      "step": 19225
    },
    {
      "epoch": 0.5539676136691062,
      "grad_norm": 0.7337983846664429,
      "learning_rate": 4.370815569678519e-06,
      "loss": 0.695,
      "step": 19226
    },
    {
      "epoch": 0.5539964271307555,
      "grad_norm": 0.7426150441169739,
      "learning_rate": 4.370352671402786e-06,
      "loss": 0.6877,
      "step": 19227
    },
    {
      "epoch": 0.5540252405924048,
      "grad_norm": 0.736095666885376,
      "learning_rate": 4.369889778610652e-06,
      "loss": 0.7155,
      "step": 19228
    },
    {
      "epoch": 0.5540540540540541,
      "grad_norm": 0.7211904525756836,
      "learning_rate": 4.369426891306147e-06,
      "loss": 0.7024,
      "step": 19229
    },
    {
      "epoch": 0.5540828675157033,
      "grad_norm": 0.6891589760780334,
      "learning_rate": 4.3689640094933054e-06,
      "loss": 0.6729,
      "step": 19230
    },
    {
      "epoch": 0.5541116809773526,
      "grad_norm": 0.8005145192146301,
      "learning_rate": 4.368501133176153e-06,
      "loss": 0.6944,
      "step": 19231
    },
    {
      "epoch": 0.5541404944390019,
      "grad_norm": 0.7113571763038635,
      "learning_rate": 4.3680382623587265e-06,
      "loss": 0.6904,
      "step": 19232
    },
    {
      "epoch": 0.5541693079006512,
      "grad_norm": 0.7430193424224854,
      "learning_rate": 4.367575397045055e-06,
      "loss": 0.6807,
      "step": 19233
    },
    {
      "epoch": 0.5541981213623005,
      "grad_norm": 0.7031105756759644,
      "learning_rate": 4.36711253723917e-06,
      "loss": 0.6669,
      "step": 19234
    },
    {
      "epoch": 0.5542269348239498,
      "grad_norm": 0.7184662818908691,
      "learning_rate": 4.366649682945102e-06,
      "loss": 0.6923,
      "step": 19235
    },
    {
      "epoch": 0.554255748285599,
      "grad_norm": 0.7429436445236206,
      "learning_rate": 4.3661868341668825e-06,
      "loss": 0.7008,
      "step": 19236
    },
    {
      "epoch": 0.5542845617472483,
      "grad_norm": 0.7807315587997437,
      "learning_rate": 4.365723990908542e-06,
      "loss": 0.7298,
      "step": 19237
    },
    {
      "epoch": 0.5543133752088976,
      "grad_norm": 0.747021496295929,
      "learning_rate": 4.365261153174111e-06,
      "loss": 0.7078,
      "step": 19238
    },
    {
      "epoch": 0.5543421886705469,
      "grad_norm": 0.7256953716278076,
      "learning_rate": 4.364798320967622e-06,
      "loss": 0.7347,
      "step": 19239
    },
    {
      "epoch": 0.5543710021321961,
      "grad_norm": 0.7601048350334167,
      "learning_rate": 4.364335494293102e-06,
      "loss": 0.6999,
      "step": 19240
    },
    {
      "epoch": 0.5543998155938454,
      "grad_norm": 0.7273496985435486,
      "learning_rate": 4.363872673154587e-06,
      "loss": 0.6999,
      "step": 19241
    },
    {
      "epoch": 0.5544286290554947,
      "grad_norm": 0.7330630421638489,
      "learning_rate": 4.363409857556104e-06,
      "loss": 0.7074,
      "step": 19242
    },
    {
      "epoch": 0.554457442517144,
      "grad_norm": 0.7357615232467651,
      "learning_rate": 4.3629470475016835e-06,
      "loss": 0.7325,
      "step": 19243
    },
    {
      "epoch": 0.5544862559787933,
      "grad_norm": 0.7379831075668335,
      "learning_rate": 4.362484242995358e-06,
      "loss": 0.6793,
      "step": 19244
    },
    {
      "epoch": 0.5545150694404426,
      "grad_norm": 0.7201004028320312,
      "learning_rate": 4.362021444041159e-06,
      "loss": 0.7074,
      "step": 19245
    },
    {
      "epoch": 0.5545438829020919,
      "grad_norm": 0.7723578810691833,
      "learning_rate": 4.361558650643115e-06,
      "loss": 0.7568,
      "step": 19246
    },
    {
      "epoch": 0.5545726963637412,
      "grad_norm": 0.7804636359214783,
      "learning_rate": 4.3610958628052565e-06,
      "loss": 0.7001,
      "step": 19247
    },
    {
      "epoch": 0.5546015098253905,
      "grad_norm": 0.7717404365539551,
      "learning_rate": 4.360633080531616e-06,
      "loss": 0.7069,
      "step": 19248
    },
    {
      "epoch": 0.5546303232870397,
      "grad_norm": 0.7103620171546936,
      "learning_rate": 4.36017030382622e-06,
      "loss": 0.6916,
      "step": 19249
    },
    {
      "epoch": 0.5546591367486889,
      "grad_norm": 0.7243791222572327,
      "learning_rate": 4.359707532693102e-06,
      "loss": 0.7188,
      "step": 19250
    },
    {
      "epoch": 0.5546879502103382,
      "grad_norm": 0.7701201438903809,
      "learning_rate": 4.359244767136292e-06,
      "loss": 0.7001,
      "step": 19251
    },
    {
      "epoch": 0.5547167636719875,
      "grad_norm": 0.7493395805358887,
      "learning_rate": 4.358782007159818e-06,
      "loss": 0.6966,
      "step": 19252
    },
    {
      "epoch": 0.5547455771336368,
      "grad_norm": 0.7238668203353882,
      "learning_rate": 4.358319252767714e-06,
      "loss": 0.7146,
      "step": 19253
    },
    {
      "epoch": 0.5547743905952861,
      "grad_norm": 0.7330614328384399,
      "learning_rate": 4.357856503964006e-06,
      "loss": 0.7194,
      "step": 19254
    },
    {
      "epoch": 0.5548032040569354,
      "grad_norm": 0.7163758277893066,
      "learning_rate": 4.3573937607527245e-06,
      "loss": 0.701,
      "step": 19255
    },
    {
      "epoch": 0.5548320175185847,
      "grad_norm": 0.7528234720230103,
      "learning_rate": 4.356931023137903e-06,
      "loss": 0.7116,
      "step": 19256
    },
    {
      "epoch": 0.554860830980234,
      "grad_norm": 0.7177331447601318,
      "learning_rate": 4.356468291123569e-06,
      "loss": 0.7172,
      "step": 19257
    },
    {
      "epoch": 0.5548896444418833,
      "grad_norm": 0.7271355986595154,
      "learning_rate": 4.356005564713754e-06,
      "loss": 0.6973,
      "step": 19258
    },
    {
      "epoch": 0.5549184579035326,
      "grad_norm": 0.7239652872085571,
      "learning_rate": 4.3555428439124854e-06,
      "loss": 0.7207,
      "step": 19259
    },
    {
      "epoch": 0.5549472713651818,
      "grad_norm": 1.0829410552978516,
      "learning_rate": 4.3550801287237955e-06,
      "loss": 0.7286,
      "step": 19260
    },
    {
      "epoch": 0.5549760848268311,
      "grad_norm": 0.7184580564498901,
      "learning_rate": 4.3546174191517125e-06,
      "loss": 0.7143,
      "step": 19261
    },
    {
      "epoch": 0.5550048982884804,
      "grad_norm": 0.7144178748130798,
      "learning_rate": 4.3541547152002664e-06,
      "loss": 0.6988,
      "step": 19262
    },
    {
      "epoch": 0.5550337117501296,
      "grad_norm": 0.71995609998703,
      "learning_rate": 4.353692016873488e-06,
      "loss": 0.7019,
      "step": 19263
    },
    {
      "epoch": 0.5550625252117789,
      "grad_norm": 0.7377546429634094,
      "learning_rate": 4.353229324175405e-06,
      "loss": 0.7126,
      "step": 19264
    },
    {
      "epoch": 0.5550913386734282,
      "grad_norm": 0.7103573083877563,
      "learning_rate": 4.352766637110048e-06,
      "loss": 0.7005,
      "step": 19265
    },
    {
      "epoch": 0.5551201521350775,
      "grad_norm": 0.7283028960227966,
      "learning_rate": 4.352303955681447e-06,
      "loss": 0.7265,
      "step": 19266
    },
    {
      "epoch": 0.5551489655967268,
      "grad_norm": 0.7218175530433655,
      "learning_rate": 4.351841279893629e-06,
      "loss": 0.7223,
      "step": 19267
    },
    {
      "epoch": 0.5551777790583761,
      "grad_norm": 0.7076535820960999,
      "learning_rate": 4.351378609750628e-06,
      "loss": 0.6823,
      "step": 19268
    },
    {
      "epoch": 0.5552065925200254,
      "grad_norm": 0.7501901388168335,
      "learning_rate": 4.350915945256471e-06,
      "loss": 0.7096,
      "step": 19269
    },
    {
      "epoch": 0.5552354059816746,
      "grad_norm": 0.7514658570289612,
      "learning_rate": 4.350453286415187e-06,
      "loss": 0.7153,
      "step": 19270
    },
    {
      "epoch": 0.5552642194433239,
      "grad_norm": 0.7234139442443848,
      "learning_rate": 4.3499906332308055e-06,
      "loss": 0.7257,
      "step": 19271
    },
    {
      "epoch": 0.5552930329049732,
      "grad_norm": 0.7193683981895447,
      "learning_rate": 4.349527985707356e-06,
      "loss": 0.7143,
      "step": 19272
    },
    {
      "epoch": 0.5553218463666225,
      "grad_norm": 0.7328709363937378,
      "learning_rate": 4.349065343848867e-06,
      "loss": 0.7039,
      "step": 19273
    },
    {
      "epoch": 0.5553506598282718,
      "grad_norm": 0.7204665541648865,
      "learning_rate": 4.348602707659369e-06,
      "loss": 0.6971,
      "step": 19274
    },
    {
      "epoch": 0.5553794732899211,
      "grad_norm": 0.7352365255355835,
      "learning_rate": 4.34814007714289e-06,
      "loss": 0.7169,
      "step": 19275
    },
    {
      "epoch": 0.5554082867515703,
      "grad_norm": 0.7165713906288147,
      "learning_rate": 4.3476774523034596e-06,
      "loss": 0.6964,
      "step": 19276
    },
    {
      "epoch": 0.5554371002132196,
      "grad_norm": 0.7262240648269653,
      "learning_rate": 4.347214833145105e-06,
      "loss": 0.7073,
      "step": 19277
    },
    {
      "epoch": 0.5554659136748689,
      "grad_norm": 0.7229969501495361,
      "learning_rate": 4.346752219671857e-06,
      "loss": 0.6882,
      "step": 19278
    },
    {
      "epoch": 0.5554947271365182,
      "grad_norm": 0.7396512627601624,
      "learning_rate": 4.346289611887744e-06,
      "loss": 0.71,
      "step": 19279
    },
    {
      "epoch": 0.5555235405981674,
      "grad_norm": 0.7045049071311951,
      "learning_rate": 4.3458270097967954e-06,
      "loss": 0.7028,
      "step": 19280
    },
    {
      "epoch": 0.5555523540598167,
      "grad_norm": 0.7305771708488464,
      "learning_rate": 4.345364413403041e-06,
      "loss": 0.7159,
      "step": 19281
    },
    {
      "epoch": 0.555581167521466,
      "grad_norm": 0.7495922446250916,
      "learning_rate": 4.344901822710505e-06,
      "loss": 0.7096,
      "step": 19282
    },
    {
      "epoch": 0.5556099809831153,
      "grad_norm": 0.7169216871261597,
      "learning_rate": 4.344439237723222e-06,
      "loss": 0.7095,
      "step": 19283
    },
    {
      "epoch": 0.5556387944447646,
      "grad_norm": 0.7178419828414917,
      "learning_rate": 4.3439766584452175e-06,
      "loss": 0.7068,
      "step": 19284
    },
    {
      "epoch": 0.5556676079064139,
      "grad_norm": 0.7372326254844666,
      "learning_rate": 4.34351408488052e-06,
      "loss": 0.7137,
      "step": 19285
    },
    {
      "epoch": 0.5556964213680632,
      "grad_norm": 0.7365977764129639,
      "learning_rate": 4.34305151703316e-06,
      "loss": 0.6766,
      "step": 19286
    },
    {
      "epoch": 0.5557252348297125,
      "grad_norm": 0.7227362990379333,
      "learning_rate": 4.342588954907163e-06,
      "loss": 0.7052,
      "step": 19287
    },
    {
      "epoch": 0.5557540482913618,
      "grad_norm": 0.7184492945671082,
      "learning_rate": 4.34212639850656e-06,
      "loss": 0.7194,
      "step": 19288
    },
    {
      "epoch": 0.5557828617530111,
      "grad_norm": 0.7440569400787354,
      "learning_rate": 4.341663847835378e-06,
      "loss": 0.7376,
      "step": 19289
    },
    {
      "epoch": 0.5558116752146602,
      "grad_norm": 0.7157827019691467,
      "learning_rate": 4.341201302897646e-06,
      "loss": 0.6981,
      "step": 19290
    },
    {
      "epoch": 0.5558404886763095,
      "grad_norm": 0.741387128829956,
      "learning_rate": 4.340738763697389e-06,
      "loss": 0.7408,
      "step": 19291
    },
    {
      "epoch": 0.5558693021379588,
      "grad_norm": 0.747930109500885,
      "learning_rate": 4.3402762302386415e-06,
      "loss": 0.7035,
      "step": 19292
    },
    {
      "epoch": 0.5558981155996081,
      "grad_norm": 0.7633156180381775,
      "learning_rate": 4.339813702525428e-06,
      "loss": 0.7317,
      "step": 19293
    },
    {
      "epoch": 0.5559269290612574,
      "grad_norm": 0.7191853523254395,
      "learning_rate": 4.3393511805617775e-06,
      "loss": 0.7258,
      "step": 19294
    },
    {
      "epoch": 0.5559557425229067,
      "grad_norm": 0.7201558947563171,
      "learning_rate": 4.338888664351718e-06,
      "loss": 0.6764,
      "step": 19295
    },
    {
      "epoch": 0.555984555984556,
      "grad_norm": 0.7204212546348572,
      "learning_rate": 4.338426153899277e-06,
      "loss": 0.7183,
      "step": 19296
    },
    {
      "epoch": 0.5560133694462053,
      "grad_norm": 0.7177304029464722,
      "learning_rate": 4.337963649208483e-06,
      "loss": 0.703,
      "step": 19297
    },
    {
      "epoch": 0.5560421829078546,
      "grad_norm": 0.723881721496582,
      "learning_rate": 4.337501150283364e-06,
      "loss": 0.7099,
      "step": 19298
    },
    {
      "epoch": 0.5560709963695039,
      "grad_norm": 0.7655380368232727,
      "learning_rate": 4.337038657127947e-06,
      "loss": 0.725,
      "step": 19299
    },
    {
      "epoch": 0.5560998098311531,
      "grad_norm": 0.7413020730018616,
      "learning_rate": 4.336576169746261e-06,
      "loss": 0.7084,
      "step": 19300
    },
    {
      "epoch": 0.5561286232928024,
      "grad_norm": 0.7290592193603516,
      "learning_rate": 4.336113688142334e-06,
      "loss": 0.7269,
      "step": 19301
    },
    {
      "epoch": 0.5561574367544517,
      "grad_norm": 0.7264972925186157,
      "learning_rate": 4.335651212320192e-06,
      "loss": 0.6994,
      "step": 19302
    },
    {
      "epoch": 0.5561862502161009,
      "grad_norm": 0.7060622572898865,
      "learning_rate": 4.335188742283864e-06,
      "loss": 0.6973,
      "step": 19303
    },
    {
      "epoch": 0.5562150636777502,
      "grad_norm": 0.7336404323577881,
      "learning_rate": 4.334726278037377e-06,
      "loss": 0.6815,
      "step": 19304
    },
    {
      "epoch": 0.5562438771393995,
      "grad_norm": 0.7404970526695251,
      "learning_rate": 4.33426381958476e-06,
      "loss": 0.7202,
      "step": 19305
    },
    {
      "epoch": 0.5562726906010488,
      "grad_norm": 0.7508828639984131,
      "learning_rate": 4.33380136693004e-06,
      "loss": 0.7288,
      "step": 19306
    },
    {
      "epoch": 0.5563015040626981,
      "grad_norm": 0.7352861762046814,
      "learning_rate": 4.333338920077244e-06,
      "loss": 0.7005,
      "step": 19307
    },
    {
      "epoch": 0.5563303175243474,
      "grad_norm": 0.7247849106788635,
      "learning_rate": 4.332876479030399e-06,
      "loss": 0.7238,
      "step": 19308
    },
    {
      "epoch": 0.5563591309859967,
      "grad_norm": 0.731690526008606,
      "learning_rate": 4.332414043793534e-06,
      "loss": 0.7048,
      "step": 19309
    },
    {
      "epoch": 0.5563879444476459,
      "grad_norm": 0.7212468981742859,
      "learning_rate": 4.3319516143706745e-06,
      "loss": 0.6904,
      "step": 19310
    },
    {
      "epoch": 0.5564167579092952,
      "grad_norm": 0.7334845066070557,
      "learning_rate": 4.331489190765849e-06,
      "loss": 0.7089,
      "step": 19311
    },
    {
      "epoch": 0.5564455713709445,
      "grad_norm": 0.7390257120132446,
      "learning_rate": 4.331026772983085e-06,
      "loss": 0.7242,
      "step": 19312
    },
    {
      "epoch": 0.5564743848325938,
      "grad_norm": 0.7366045713424683,
      "learning_rate": 4.330564361026407e-06,
      "loss": 0.6962,
      "step": 19313
    },
    {
      "epoch": 0.5565031982942431,
      "grad_norm": 0.7286285758018494,
      "learning_rate": 4.330101954899847e-06,
      "loss": 0.7036,
      "step": 19314
    },
    {
      "epoch": 0.5565320117558924,
      "grad_norm": 0.713146984577179,
      "learning_rate": 4.329639554607425e-06,
      "loss": 0.6928,
      "step": 19315
    },
    {
      "epoch": 0.5565608252175417,
      "grad_norm": 0.7211351990699768,
      "learning_rate": 4.329177160153175e-06,
      "loss": 0.7084,
      "step": 19316
    },
    {
      "epoch": 0.5565896386791909,
      "grad_norm": 0.7322659492492676,
      "learning_rate": 4.328714771541122e-06,
      "loss": 0.7015,
      "step": 19317
    },
    {
      "epoch": 0.5566184521408402,
      "grad_norm": 0.716546893119812,
      "learning_rate": 4.328252388775292e-06,
      "loss": 0.7058,
      "step": 19318
    },
    {
      "epoch": 0.5566472656024894,
      "grad_norm": 0.7404131293296814,
      "learning_rate": 4.327790011859712e-06,
      "loss": 0.7016,
      "step": 19319
    },
    {
      "epoch": 0.5566760790641387,
      "grad_norm": 0.75339674949646,
      "learning_rate": 4.3273276407984085e-06,
      "loss": 0.7373,
      "step": 19320
    },
    {
      "epoch": 0.556704892525788,
      "grad_norm": 0.7030090093612671,
      "learning_rate": 4.3268652755954095e-06,
      "loss": 0.6732,
      "step": 19321
    },
    {
      "epoch": 0.5567337059874373,
      "grad_norm": 0.7128140926361084,
      "learning_rate": 4.326402916254741e-06,
      "loss": 0.7035,
      "step": 19322
    },
    {
      "epoch": 0.5567625194490866,
      "grad_norm": 0.7356091737747192,
      "learning_rate": 4.32594056278043e-06,
      "loss": 0.7157,
      "step": 19323
    },
    {
      "epoch": 0.5567913329107359,
      "grad_norm": 0.7498476505279541,
      "learning_rate": 4.325478215176503e-06,
      "loss": 0.7196,
      "step": 19324
    },
    {
      "epoch": 0.5568201463723852,
      "grad_norm": 0.7191898226737976,
      "learning_rate": 4.3250158734469865e-06,
      "loss": 0.7047,
      "step": 19325
    },
    {
      "epoch": 0.5568489598340345,
      "grad_norm": 0.7390033006668091,
      "learning_rate": 4.324553537595906e-06,
      "loss": 0.6932,
      "step": 19326
    },
    {
      "epoch": 0.5568777732956838,
      "grad_norm": 0.7405737638473511,
      "learning_rate": 4.324091207627288e-06,
      "loss": 0.7155,
      "step": 19327
    },
    {
      "epoch": 0.5569065867573331,
      "grad_norm": 0.7319471836090088,
      "learning_rate": 4.323628883545161e-06,
      "loss": 0.7002,
      "step": 19328
    },
    {
      "epoch": 0.5569354002189824,
      "grad_norm": 0.7269652485847473,
      "learning_rate": 4.323166565353551e-06,
      "loss": 0.7242,
      "step": 19329
    },
    {
      "epoch": 0.5569642136806315,
      "grad_norm": 0.8139728903770447,
      "learning_rate": 4.322704253056484e-06,
      "loss": 0.7089,
      "step": 19330
    },
    {
      "epoch": 0.5569930271422808,
      "grad_norm": 0.753266453742981,
      "learning_rate": 4.322241946657985e-06,
      "loss": 0.7149,
      "step": 19331
    },
    {
      "epoch": 0.5570218406039301,
      "grad_norm": 0.7387705445289612,
      "learning_rate": 4.321779646162081e-06,
      "loss": 0.7157,
      "step": 19332
    },
    {
      "epoch": 0.5570506540655794,
      "grad_norm": 0.7315277457237244,
      "learning_rate": 4.3213173515728e-06,
      "loss": 0.7273,
      "step": 19333
    },
    {
      "epoch": 0.5570794675272287,
      "grad_norm": 0.7172715067863464,
      "learning_rate": 4.320855062894165e-06,
      "loss": 0.7267,
      "step": 19334
    },
    {
      "epoch": 0.557108280988878,
      "grad_norm": 0.7302702069282532,
      "learning_rate": 4.320392780130204e-06,
      "loss": 0.7088,
      "step": 19335
    },
    {
      "epoch": 0.5571370944505273,
      "grad_norm": 0.7151238322257996,
      "learning_rate": 4.319930503284942e-06,
      "loss": 0.7023,
      "step": 19336
    },
    {
      "epoch": 0.5571659079121766,
      "grad_norm": 0.7552438378334045,
      "learning_rate": 4.319468232362405e-06,
      "loss": 0.7076,
      "step": 19337
    },
    {
      "epoch": 0.5571947213738259,
      "grad_norm": 0.7283350825309753,
      "learning_rate": 4.31900596736662e-06,
      "loss": 0.7283,
      "step": 19338
    },
    {
      "epoch": 0.5572235348354752,
      "grad_norm": 0.7087502479553223,
      "learning_rate": 4.3185437083016115e-06,
      "loss": 0.6989,
      "step": 19339
    },
    {
      "epoch": 0.5572523482971244,
      "grad_norm": 0.7218412756919861,
      "learning_rate": 4.318081455171408e-06,
      "loss": 0.7212,
      "step": 19340
    },
    {
      "epoch": 0.5572811617587737,
      "grad_norm": 0.720757007598877,
      "learning_rate": 4.317619207980031e-06,
      "loss": 0.6999,
      "step": 19341
    },
    {
      "epoch": 0.557309975220423,
      "grad_norm": 0.7323803901672363,
      "learning_rate": 4.317156966731509e-06,
      "loss": 0.7168,
      "step": 19342
    },
    {
      "epoch": 0.5573387886820723,
      "grad_norm": 0.733945369720459,
      "learning_rate": 4.316694731429869e-06,
      "loss": 0.7338,
      "step": 19343
    },
    {
      "epoch": 0.5573676021437215,
      "grad_norm": 0.7250244617462158,
      "learning_rate": 4.316232502079134e-06,
      "loss": 0.6961,
      "step": 19344
    },
    {
      "epoch": 0.5573964156053708,
      "grad_norm": 0.7249734997749329,
      "learning_rate": 4.3157702786833305e-06,
      "loss": 0.694,
      "step": 19345
    },
    {
      "epoch": 0.5574252290670201,
      "grad_norm": 0.7501100897789001,
      "learning_rate": 4.3153080612464835e-06,
      "loss": 0.7013,
      "step": 19346
    },
    {
      "epoch": 0.5574540425286694,
      "grad_norm": 0.7566485404968262,
      "learning_rate": 4.31484584977262e-06,
      "loss": 0.7345,
      "step": 19347
    },
    {
      "epoch": 0.5574828559903187,
      "grad_norm": 0.7101729512214661,
      "learning_rate": 4.3143836442657635e-06,
      "loss": 0.6991,
      "step": 19348
    },
    {
      "epoch": 0.557511669451968,
      "grad_norm": 0.7348969578742981,
      "learning_rate": 4.31392144472994e-06,
      "loss": 0.7036,
      "step": 19349
    },
    {
      "epoch": 0.5575404829136172,
      "grad_norm": 0.8058189153671265,
      "learning_rate": 4.3134592511691745e-06,
      "loss": 0.7193,
      "step": 19350
    },
    {
      "epoch": 0.5575692963752665,
      "grad_norm": 0.7419092655181885,
      "learning_rate": 4.312997063587491e-06,
      "loss": 0.7191,
      "step": 19351
    },
    {
      "epoch": 0.5575981098369158,
      "grad_norm": 0.7752796411514282,
      "learning_rate": 4.312534881988919e-06,
      "loss": 0.704,
      "step": 19352
    },
    {
      "epoch": 0.5576269232985651,
      "grad_norm": 0.7464997172355652,
      "learning_rate": 4.312072706377481e-06,
      "loss": 0.7385,
      "step": 19353
    },
    {
      "epoch": 0.5576557367602144,
      "grad_norm": 0.739115834236145,
      "learning_rate": 4.3116105367572005e-06,
      "loss": 0.7139,
      "step": 19354
    },
    {
      "epoch": 0.5576845502218637,
      "grad_norm": 0.7263401746749878,
      "learning_rate": 4.311148373132105e-06,
      "loss": 0.7015,
      "step": 19355
    },
    {
      "epoch": 0.557713363683513,
      "grad_norm": 0.7242717742919922,
      "learning_rate": 4.310686215506219e-06,
      "loss": 0.676,
      "step": 19356
    },
    {
      "epoch": 0.5577421771451622,
      "grad_norm": 0.7806122899055481,
      "learning_rate": 4.310224063883566e-06,
      "loss": 0.745,
      "step": 19357
    },
    {
      "epoch": 0.5577709906068115,
      "grad_norm": 0.7152660489082336,
      "learning_rate": 4.309761918268171e-06,
      "loss": 0.7209,
      "step": 19358
    },
    {
      "epoch": 0.5577998040684607,
      "grad_norm": 0.7147924304008484,
      "learning_rate": 4.309299778664062e-06,
      "loss": 0.7005,
      "step": 19359
    },
    {
      "epoch": 0.55782861753011,
      "grad_norm": 0.7719119191169739,
      "learning_rate": 4.308837645075261e-06,
      "loss": 0.7177,
      "step": 19360
    },
    {
      "epoch": 0.5578574309917593,
      "grad_norm": 0.7158815860748291,
      "learning_rate": 4.308375517505793e-06,
      "loss": 0.7126,
      "step": 19361
    },
    {
      "epoch": 0.5578862444534086,
      "grad_norm": 0.7595251798629761,
      "learning_rate": 4.307913395959683e-06,
      "loss": 0.7152,
      "step": 19362
    },
    {
      "epoch": 0.5579150579150579,
      "grad_norm": 0.7683264017105103,
      "learning_rate": 4.307451280440953e-06,
      "loss": 0.7109,
      "step": 19363
    },
    {
      "epoch": 0.5579438713767072,
      "grad_norm": 0.7662054896354675,
      "learning_rate": 4.306989170953632e-06,
      "loss": 0.6656,
      "step": 19364
    },
    {
      "epoch": 0.5579726848383565,
      "grad_norm": 0.7319592237472534,
      "learning_rate": 4.306527067501743e-06,
      "loss": 0.6886,
      "step": 19365
    },
    {
      "epoch": 0.5580014983000058,
      "grad_norm": 0.7477492690086365,
      "learning_rate": 4.306064970089309e-06,
      "loss": 0.7151,
      "step": 19366
    },
    {
      "epoch": 0.5580303117616551,
      "grad_norm": 0.7623788714408875,
      "learning_rate": 4.305602878720357e-06,
      "loss": 0.7103,
      "step": 19367
    },
    {
      "epoch": 0.5580591252233044,
      "grad_norm": 0.7117445468902588,
      "learning_rate": 4.305140793398909e-06,
      "loss": 0.6924,
      "step": 19368
    },
    {
      "epoch": 0.5580879386849537,
      "grad_norm": 0.7309309244155884,
      "learning_rate": 4.3046787141289905e-06,
      "loss": 0.7092,
      "step": 19369
    },
    {
      "epoch": 0.558116752146603,
      "grad_norm": 0.7307368516921997,
      "learning_rate": 4.304216640914625e-06,
      "loss": 0.6845,
      "step": 19370
    },
    {
      "epoch": 0.5581455656082521,
      "grad_norm": 0.7046940326690674,
      "learning_rate": 4.3037545737598365e-06,
      "loss": 0.6953,
      "step": 19371
    },
    {
      "epoch": 0.5581743790699014,
      "grad_norm": 0.7174378037452698,
      "learning_rate": 4.30329251266865e-06,
      "loss": 0.7013,
      "step": 19372
    },
    {
      "epoch": 0.5582031925315507,
      "grad_norm": 0.7341066002845764,
      "learning_rate": 4.30283045764509e-06,
      "loss": 0.6918,
      "step": 19373
    },
    {
      "epoch": 0.5582320059932,
      "grad_norm": 0.7583805918693542,
      "learning_rate": 4.302368408693179e-06,
      "loss": 0.7341,
      "step": 19374
    },
    {
      "epoch": 0.5582608194548493,
      "grad_norm": 0.7303003072738647,
      "learning_rate": 4.30190636581694e-06,
      "loss": 0.7032,
      "step": 19375
    },
    {
      "epoch": 0.5582896329164986,
      "grad_norm": 0.721443772315979,
      "learning_rate": 4.3014443290204e-06,
      "loss": 0.7148,
      "step": 19376
    },
    {
      "epoch": 0.5583184463781479,
      "grad_norm": 0.7001248002052307,
      "learning_rate": 4.300982298307583e-06,
      "loss": 0.7075,
      "step": 19377
    },
    {
      "epoch": 0.5583472598397972,
      "grad_norm": 0.7281387448310852,
      "learning_rate": 4.300520273682511e-06,
      "loss": 0.7121,
      "step": 19378
    },
    {
      "epoch": 0.5583760733014465,
      "grad_norm": 0.7272725701332092,
      "learning_rate": 4.300058255149208e-06,
      "loss": 0.7297,
      "step": 19379
    },
    {
      "epoch": 0.5584048867630957,
      "grad_norm": 0.6929779052734375,
      "learning_rate": 4.299596242711697e-06,
      "loss": 0.6958,
      "step": 19380
    },
    {
      "epoch": 0.558433700224745,
      "grad_norm": 0.7121738791465759,
      "learning_rate": 4.299134236374003e-06,
      "loss": 0.699,
      "step": 19381
    },
    {
      "epoch": 0.5584625136863943,
      "grad_norm": 0.6817530393600464,
      "learning_rate": 4.29867223614015e-06,
      "loss": 0.6852,
      "step": 19382
    },
    {
      "epoch": 0.5584913271480436,
      "grad_norm": 0.7287575006484985,
      "learning_rate": 4.29821024201416e-06,
      "loss": 0.7135,
      "step": 19383
    },
    {
      "epoch": 0.5585201406096928,
      "grad_norm": 0.7333751320838928,
      "learning_rate": 4.297748254000058e-06,
      "loss": 0.7215,
      "step": 19384
    },
    {
      "epoch": 0.5585489540713421,
      "grad_norm": 0.7022879719734192,
      "learning_rate": 4.297286272101866e-06,
      "loss": 0.6824,
      "step": 19385
    },
    {
      "epoch": 0.5585777675329914,
      "grad_norm": 0.7243294715881348,
      "learning_rate": 4.296824296323608e-06,
      "loss": 0.713,
      "step": 19386
    },
    {
      "epoch": 0.5586065809946407,
      "grad_norm": 0.7138406038284302,
      "learning_rate": 4.296362326669307e-06,
      "loss": 0.7174,
      "step": 19387
    },
    {
      "epoch": 0.55863539445629,
      "grad_norm": 0.7302077412605286,
      "learning_rate": 4.295900363142987e-06,
      "loss": 0.7111,
      "step": 19388
    },
    {
      "epoch": 0.5586642079179392,
      "grad_norm": 0.8219637870788574,
      "learning_rate": 4.295438405748672e-06,
      "loss": 0.6986,
      "step": 19389
    },
    {
      "epoch": 0.5586930213795885,
      "grad_norm": 0.7240742444992065,
      "learning_rate": 4.294976454490384e-06,
      "loss": 0.7221,
      "step": 19390
    },
    {
      "epoch": 0.5587218348412378,
      "grad_norm": 0.7303766012191772,
      "learning_rate": 4.294514509372147e-06,
      "loss": 0.7104,
      "step": 19391
    },
    {
      "epoch": 0.5587506483028871,
      "grad_norm": 0.7185215353965759,
      "learning_rate": 4.294052570397982e-06,
      "loss": 0.6786,
      "step": 19392
    },
    {
      "epoch": 0.5587794617645364,
      "grad_norm": 0.7698697447776794,
      "learning_rate": 4.293590637571914e-06,
      "loss": 0.7417,
      "step": 19393
    },
    {
      "epoch": 0.5588082752261857,
      "grad_norm": 0.7012985944747925,
      "learning_rate": 4.293128710897965e-06,
      "loss": 0.6995,
      "step": 19394
    },
    {
      "epoch": 0.558837088687835,
      "grad_norm": 0.7070648670196533,
      "learning_rate": 4.292666790380159e-06,
      "loss": 0.702,
      "step": 19395
    },
    {
      "epoch": 0.5588659021494843,
      "grad_norm": 0.7286941409111023,
      "learning_rate": 4.292204876022516e-06,
      "loss": 0.7345,
      "step": 19396
    },
    {
      "epoch": 0.5588947156111336,
      "grad_norm": 0.7322090864181519,
      "learning_rate": 4.291742967829064e-06,
      "loss": 0.7382,
      "step": 19397
    },
    {
      "epoch": 0.5589235290727828,
      "grad_norm": 0.7267071008682251,
      "learning_rate": 4.291281065803821e-06,
      "loss": 0.7087,
      "step": 19398
    },
    {
      "epoch": 0.558952342534432,
      "grad_norm": 0.7814803123474121,
      "learning_rate": 4.2908191699508115e-06,
      "loss": 0.6738,
      "step": 19399
    },
    {
      "epoch": 0.5589811559960813,
      "grad_norm": 0.7807788848876953,
      "learning_rate": 4.290357280274058e-06,
      "loss": 0.6852,
      "step": 19400
    },
    {
      "epoch": 0.5590099694577306,
      "grad_norm": 0.7127170562744141,
      "learning_rate": 4.289895396777584e-06,
      "loss": 0.723,
      "step": 19401
    },
    {
      "epoch": 0.5590387829193799,
      "grad_norm": 0.730177640914917,
      "learning_rate": 4.28943351946541e-06,
      "loss": 0.6804,
      "step": 19402
    },
    {
      "epoch": 0.5590675963810292,
      "grad_norm": 0.7290357351303101,
      "learning_rate": 4.288971648341561e-06,
      "loss": 0.6809,
      "step": 19403
    },
    {
      "epoch": 0.5590964098426785,
      "grad_norm": 0.7438573241233826,
      "learning_rate": 4.288509783410059e-06,
      "loss": 0.7145,
      "step": 19404
    },
    {
      "epoch": 0.5591252233043278,
      "grad_norm": 0.7061891555786133,
      "learning_rate": 4.288047924674924e-06,
      "loss": 0.714,
      "step": 19405
    },
    {
      "epoch": 0.5591540367659771,
      "grad_norm": 0.7053154110908508,
      "learning_rate": 4.287586072140181e-06,
      "loss": 0.6905,
      "step": 19406
    },
    {
      "epoch": 0.5591828502276264,
      "grad_norm": 0.7394749522209167,
      "learning_rate": 4.287124225809851e-06,
      "loss": 0.7389,
      "step": 19407
    },
    {
      "epoch": 0.5592116636892757,
      "grad_norm": 0.6890641450881958,
      "learning_rate": 4.286662385687955e-06,
      "loss": 0.696,
      "step": 19408
    },
    {
      "epoch": 0.559240477150925,
      "grad_norm": 0.7191857099533081,
      "learning_rate": 4.286200551778518e-06,
      "loss": 0.7282,
      "step": 19409
    },
    {
      "epoch": 0.5592692906125742,
      "grad_norm": 0.7197828888893127,
      "learning_rate": 4.28573872408556e-06,
      "loss": 0.7053,
      "step": 19410
    },
    {
      "epoch": 0.5592981040742234,
      "grad_norm": 0.7203465700149536,
      "learning_rate": 4.2852769026131025e-06,
      "loss": 0.7103,
      "step": 19411
    },
    {
      "epoch": 0.5593269175358727,
      "grad_norm": 0.7454715371131897,
      "learning_rate": 4.28481508736517e-06,
      "loss": 0.7268,
      "step": 19412
    },
    {
      "epoch": 0.559355730997522,
      "grad_norm": 0.7499179244041443,
      "learning_rate": 4.284353278345783e-06,
      "loss": 0.715,
      "step": 19413
    },
    {
      "epoch": 0.5593845444591713,
      "grad_norm": 0.7373188734054565,
      "learning_rate": 4.283891475558964e-06,
      "loss": 0.7001,
      "step": 19414
    },
    {
      "epoch": 0.5594133579208206,
      "grad_norm": 0.7288045287132263,
      "learning_rate": 4.283429679008735e-06,
      "loss": 0.727,
      "step": 19415
    },
    {
      "epoch": 0.5594421713824699,
      "grad_norm": 0.7621766328811646,
      "learning_rate": 4.282967888699116e-06,
      "loss": 0.7179,
      "step": 19416
    },
    {
      "epoch": 0.5594709848441192,
      "grad_norm": 0.7641093134880066,
      "learning_rate": 4.28250610463413e-06,
      "loss": 0.7046,
      "step": 19417
    },
    {
      "epoch": 0.5594997983057685,
      "grad_norm": 0.7032044529914856,
      "learning_rate": 4.282044326817798e-06,
      "loss": 0.7144,
      "step": 19418
    },
    {
      "epoch": 0.5595286117674177,
      "grad_norm": 0.7288829684257507,
      "learning_rate": 4.281582555254143e-06,
      "loss": 0.7145,
      "step": 19419
    },
    {
      "epoch": 0.559557425229067,
      "grad_norm": 0.7316784262657166,
      "learning_rate": 4.2811207899471864e-06,
      "loss": 0.7225,
      "step": 19420
    },
    {
      "epoch": 0.5595862386907163,
      "grad_norm": 0.7367585301399231,
      "learning_rate": 4.2806590309009485e-06,
      "loss": 0.7393,
      "step": 19421
    },
    {
      "epoch": 0.5596150521523656,
      "grad_norm": 0.7346960306167603,
      "learning_rate": 4.280197278119453e-06,
      "loss": 0.7103,
      "step": 19422
    },
    {
      "epoch": 0.5596438656140149,
      "grad_norm": 0.7454673647880554,
      "learning_rate": 4.279735531606715e-06,
      "loss": 0.7068,
      "step": 19423
    },
    {
      "epoch": 0.5596726790756642,
      "grad_norm": 0.7330553531646729,
      "learning_rate": 4.279273791366765e-06,
      "loss": 0.7012,
      "step": 19424
    },
    {
      "epoch": 0.5597014925373134,
      "grad_norm": 0.7251576781272888,
      "learning_rate": 4.2788120574036176e-06,
      "loss": 0.692,
      "step": 19425
    },
    {
      "epoch": 0.5597303059989627,
      "grad_norm": 0.7492062449455261,
      "learning_rate": 4.278350329721298e-06,
      "loss": 0.7057,
      "step": 19426
    },
    {
      "epoch": 0.559759119460612,
      "grad_norm": 0.7589002847671509,
      "learning_rate": 4.277888608323824e-06,
      "loss": 0.7279,
      "step": 19427
    },
    {
      "epoch": 0.5597879329222613,
      "grad_norm": 0.708895742893219,
      "learning_rate": 4.27742689321522e-06,
      "loss": 0.6954,
      "step": 19428
    },
    {
      "epoch": 0.5598167463839105,
      "grad_norm": 0.6830632090568542,
      "learning_rate": 4.276965184399504e-06,
      "loss": 0.7162,
      "step": 19429
    },
    {
      "epoch": 0.5598455598455598,
      "grad_norm": 0.7322263121604919,
      "learning_rate": 4.2765034818807e-06,
      "loss": 0.6921,
      "step": 19430
    },
    {
      "epoch": 0.5598743733072091,
      "grad_norm": 0.7388405203819275,
      "learning_rate": 4.276041785662826e-06,
      "loss": 0.7174,
      "step": 19431
    },
    {
      "epoch": 0.5599031867688584,
      "grad_norm": 0.7051711678504944,
      "learning_rate": 4.275580095749905e-06,
      "loss": 0.7029,
      "step": 19432
    },
    {
      "epoch": 0.5599320002305077,
      "grad_norm": 0.7127894163131714,
      "learning_rate": 4.2751184121459576e-06,
      "loss": 0.7152,
      "step": 19433
    },
    {
      "epoch": 0.559960813692157,
      "grad_norm": 0.7210041284561157,
      "learning_rate": 4.2746567348550026e-06,
      "loss": 0.676,
      "step": 19434
    },
    {
      "epoch": 0.5599896271538063,
      "grad_norm": 0.7016038298606873,
      "learning_rate": 4.2741950638810626e-06,
      "loss": 0.6894,
      "step": 19435
    },
    {
      "epoch": 0.5600184406154556,
      "grad_norm": 0.708867609500885,
      "learning_rate": 4.273733399228159e-06,
      "loss": 0.698,
      "step": 19436
    },
    {
      "epoch": 0.5600472540771049,
      "grad_norm": 0.7217006087303162,
      "learning_rate": 4.273271740900312e-06,
      "loss": 0.7338,
      "step": 19437
    },
    {
      "epoch": 0.560076067538754,
      "grad_norm": 0.7313050031661987,
      "learning_rate": 4.2728100889015405e-06,
      "loss": 0.7239,
      "step": 19438
    },
    {
      "epoch": 0.5601048810004033,
      "grad_norm": 0.7010241746902466,
      "learning_rate": 4.272348443235867e-06,
      "loss": 0.7079,
      "step": 19439
    },
    {
      "epoch": 0.5601336944620526,
      "grad_norm": 0.706616997718811,
      "learning_rate": 4.2718868039073106e-06,
      "loss": 0.7115,
      "step": 19440
    },
    {
      "epoch": 0.5601625079237019,
      "grad_norm": 0.7138108015060425,
      "learning_rate": 4.271425170919892e-06,
      "loss": 0.6976,
      "step": 19441
    },
    {
      "epoch": 0.5601913213853512,
      "grad_norm": 0.7149459719657898,
      "learning_rate": 4.270963544277633e-06,
      "loss": 0.7218,
      "step": 19442
    },
    {
      "epoch": 0.5602201348470005,
      "grad_norm": 0.7103540301322937,
      "learning_rate": 4.2705019239845525e-06,
      "loss": 0.7254,
      "step": 19443
    },
    {
      "epoch": 0.5602489483086498,
      "grad_norm": 0.7337926030158997,
      "learning_rate": 4.2700403100446705e-06,
      "loss": 0.7253,
      "step": 19444
    },
    {
      "epoch": 0.5602777617702991,
      "grad_norm": 0.7336564660072327,
      "learning_rate": 4.269578702462008e-06,
      "loss": 0.7025,
      "step": 19445
    },
    {
      "epoch": 0.5603065752319484,
      "grad_norm": 0.7169981598854065,
      "learning_rate": 4.269117101240585e-06,
      "loss": 0.7157,
      "step": 19446
    },
    {
      "epoch": 0.5603353886935977,
      "grad_norm": 0.7232709527015686,
      "learning_rate": 4.26865550638442e-06,
      "loss": 0.7197,
      "step": 19447
    },
    {
      "epoch": 0.560364202155247,
      "grad_norm": 0.7059689164161682,
      "learning_rate": 4.2681939178975355e-06,
      "loss": 0.6968,
      "step": 19448
    },
    {
      "epoch": 0.5603930156168963,
      "grad_norm": 0.7533220052719116,
      "learning_rate": 4.267732335783951e-06,
      "loss": 0.7123,
      "step": 19449
    },
    {
      "epoch": 0.5604218290785455,
      "grad_norm": 0.7315029501914978,
      "learning_rate": 4.267270760047685e-06,
      "loss": 0.708,
      "step": 19450
    },
    {
      "epoch": 0.5604506425401948,
      "grad_norm": 0.717534065246582,
      "learning_rate": 4.2668091906927595e-06,
      "loss": 0.7251,
      "step": 19451
    },
    {
      "epoch": 0.560479456001844,
      "grad_norm": 0.7296955585479736,
      "learning_rate": 4.266347627723192e-06,
      "loss": 0.7203,
      "step": 19452
    },
    {
      "epoch": 0.5605082694634933,
      "grad_norm": 0.7183418273925781,
      "learning_rate": 4.265886071143004e-06,
      "loss": 0.7155,
      "step": 19453
    },
    {
      "epoch": 0.5605370829251426,
      "grad_norm": 0.7477745413780212,
      "learning_rate": 4.265424520956214e-06,
      "loss": 0.7093,
      "step": 19454
    },
    {
      "epoch": 0.5605658963867919,
      "grad_norm": 0.7352813482284546,
      "learning_rate": 4.2649629771668414e-06,
      "loss": 0.7015,
      "step": 19455
    },
    {
      "epoch": 0.5605947098484412,
      "grad_norm": 0.7335013747215271,
      "learning_rate": 4.264501439778907e-06,
      "loss": 0.7056,
      "step": 19456
    },
    {
      "epoch": 0.5606235233100905,
      "grad_norm": 0.7252349853515625,
      "learning_rate": 4.26403990879643e-06,
      "loss": 0.7028,
      "step": 19457
    },
    {
      "epoch": 0.5606523367717398,
      "grad_norm": 0.7283176183700562,
      "learning_rate": 4.263578384223429e-06,
      "loss": 0.6971,
      "step": 19458
    },
    {
      "epoch": 0.560681150233389,
      "grad_norm": 0.7324978113174438,
      "learning_rate": 4.2631168660639255e-06,
      "loss": 0.754,
      "step": 19459
    },
    {
      "epoch": 0.5607099636950383,
      "grad_norm": 0.7591308951377869,
      "learning_rate": 4.2626553543219355e-06,
      "loss": 0.7206,
      "step": 19460
    },
    {
      "epoch": 0.5607387771566876,
      "grad_norm": 0.7147789597511292,
      "learning_rate": 4.262193849001482e-06,
      "loss": 0.7161,
      "step": 19461
    },
    {
      "epoch": 0.5607675906183369,
      "grad_norm": 0.7030715346336365,
      "learning_rate": 4.261732350106582e-06,
      "loss": 0.7462,
      "step": 19462
    },
    {
      "epoch": 0.5607964040799862,
      "grad_norm": 0.7107481360435486,
      "learning_rate": 4.261270857641256e-06,
      "loss": 0.705,
      "step": 19463
    },
    {
      "epoch": 0.5608252175416355,
      "grad_norm": 0.7849970459938049,
      "learning_rate": 4.260809371609522e-06,
      "loss": 0.7325,
      "step": 19464
    },
    {
      "epoch": 0.5608540310032847,
      "grad_norm": 0.7172250151634216,
      "learning_rate": 4.260347892015399e-06,
      "loss": 0.6939,
      "step": 19465
    },
    {
      "epoch": 0.560882844464934,
      "grad_norm": 0.7247611880302429,
      "learning_rate": 4.2598864188629065e-06,
      "loss": 0.6973,
      "step": 19466
    },
    {
      "epoch": 0.5609116579265833,
      "grad_norm": 0.7376636862754822,
      "learning_rate": 4.259424952156064e-06,
      "loss": 0.6829,
      "step": 19467
    },
    {
      "epoch": 0.5609404713882326,
      "grad_norm": 0.7011313438415527,
      "learning_rate": 4.25896349189889e-06,
      "loss": 0.7109,
      "step": 19468
    },
    {
      "epoch": 0.5609692848498818,
      "grad_norm": 0.7297033071517944,
      "learning_rate": 4.258502038095403e-06,
      "loss": 0.7312,
      "step": 19469
    },
    {
      "epoch": 0.5609980983115311,
      "grad_norm": 0.7258613109588623,
      "learning_rate": 4.25804059074962e-06,
      "loss": 0.7032,
      "step": 19470
    },
    {
      "epoch": 0.5610269117731804,
      "grad_norm": 0.7329587340354919,
      "learning_rate": 4.257579149865564e-06,
      "loss": 0.7125,
      "step": 19471
    },
    {
      "epoch": 0.5610557252348297,
      "grad_norm": 0.7292168736457825,
      "learning_rate": 4.257117715447251e-06,
      "loss": 0.7175,
      "step": 19472
    },
    {
      "epoch": 0.561084538696479,
      "grad_norm": 0.7132347822189331,
      "learning_rate": 4.2566562874987e-06,
      "loss": 0.6978,
      "step": 19473
    },
    {
      "epoch": 0.5611133521581283,
      "grad_norm": 0.7442913055419922,
      "learning_rate": 4.256194866023929e-06,
      "loss": 0.6986,
      "step": 19474
    },
    {
      "epoch": 0.5611421656197776,
      "grad_norm": 0.7229923009872437,
      "learning_rate": 4.255733451026959e-06,
      "loss": 0.7083,
      "step": 19475
    },
    {
      "epoch": 0.5611709790814269,
      "grad_norm": 0.7545821666717529,
      "learning_rate": 4.255272042511806e-06,
      "loss": 0.7224,
      "step": 19476
    },
    {
      "epoch": 0.5611997925430762,
      "grad_norm": 0.7120653390884399,
      "learning_rate": 4.254810640482488e-06,
      "loss": 0.7133,
      "step": 19477
    },
    {
      "epoch": 0.5612286060047255,
      "grad_norm": 0.733839750289917,
      "learning_rate": 4.254349244943024e-06,
      "loss": 0.7261,
      "step": 19478
    },
    {
      "epoch": 0.5612574194663746,
      "grad_norm": 0.7339690923690796,
      "learning_rate": 4.253887855897434e-06,
      "loss": 0.7072,
      "step": 19479
    },
    {
      "epoch": 0.5612862329280239,
      "grad_norm": 0.7218970060348511,
      "learning_rate": 4.253426473349735e-06,
      "loss": 0.6981,
      "step": 19480
    },
    {
      "epoch": 0.5613150463896732,
      "grad_norm": 0.7275791764259338,
      "learning_rate": 4.252965097303945e-06,
      "loss": 0.7051,
      "step": 19481
    },
    {
      "epoch": 0.5613438598513225,
      "grad_norm": 0.7246162295341492,
      "learning_rate": 4.2525037277640805e-06,
      "loss": 0.7336,
      "step": 19482
    },
    {
      "epoch": 0.5613726733129718,
      "grad_norm": 0.7557042241096497,
      "learning_rate": 4.2520423647341625e-06,
      "loss": 0.6918,
      "step": 19483
    },
    {
      "epoch": 0.5614014867746211,
      "grad_norm": 0.7012669444084167,
      "learning_rate": 4.251581008218208e-06,
      "loss": 0.7247,
      "step": 19484
    },
    {
      "epoch": 0.5614303002362704,
      "grad_norm": 0.7017831802368164,
      "learning_rate": 4.251119658220235e-06,
      "loss": 0.6815,
      "step": 19485
    },
    {
      "epoch": 0.5614591136979197,
      "grad_norm": 0.716042160987854,
      "learning_rate": 4.250658314744262e-06,
      "loss": 0.7082,
      "step": 19486
    },
    {
      "epoch": 0.561487927159569,
      "grad_norm": 0.7428545355796814,
      "learning_rate": 4.250196977794305e-06,
      "loss": 0.7136,
      "step": 19487
    },
    {
      "epoch": 0.5615167406212183,
      "grad_norm": 0.7230521440505981,
      "learning_rate": 4.249735647374383e-06,
      "loss": 0.7111,
      "step": 19488
    },
    {
      "epoch": 0.5615455540828675,
      "grad_norm": 0.7240169644355774,
      "learning_rate": 4.249274323488514e-06,
      "loss": 0.7247,
      "step": 19489
    },
    {
      "epoch": 0.5615743675445168,
      "grad_norm": 0.7440519332885742,
      "learning_rate": 4.248813006140714e-06,
      "loss": 0.7123,
      "step": 19490
    },
    {
      "epoch": 0.5616031810061661,
      "grad_norm": 0.7114600539207458,
      "learning_rate": 4.248351695335003e-06,
      "loss": 0.7106,
      "step": 19491
    },
    {
      "epoch": 0.5616319944678153,
      "grad_norm": 0.7125750780105591,
      "learning_rate": 4.247890391075397e-06,
      "loss": 0.7033,
      "step": 19492
    },
    {
      "epoch": 0.5616608079294646,
      "grad_norm": 0.7363449335098267,
      "learning_rate": 4.247429093365912e-06,
      "loss": 0.7183,
      "step": 19493
    },
    {
      "epoch": 0.5616896213911139,
      "grad_norm": 0.7066996097564697,
      "learning_rate": 4.246967802210569e-06,
      "loss": 0.7029,
      "step": 19494
    },
    {
      "epoch": 0.5617184348527632,
      "grad_norm": 0.7299464344978333,
      "learning_rate": 4.246506517613384e-06,
      "loss": 0.6991,
      "step": 19495
    },
    {
      "epoch": 0.5617472483144125,
      "grad_norm": 0.7241610288619995,
      "learning_rate": 4.246045239578373e-06,
      "loss": 0.7323,
      "step": 19496
    },
    {
      "epoch": 0.5617760617760618,
      "grad_norm": 0.7305166721343994,
      "learning_rate": 4.245583968109555e-06,
      "loss": 0.7152,
      "step": 19497
    },
    {
      "epoch": 0.561804875237711,
      "grad_norm": 0.7249348163604736,
      "learning_rate": 4.245122703210945e-06,
      "loss": 0.693,
      "step": 19498
    },
    {
      "epoch": 0.5618336886993603,
      "grad_norm": 0.7102232575416565,
      "learning_rate": 4.244661444886564e-06,
      "loss": 0.698,
      "step": 19499
    },
    {
      "epoch": 0.5618625021610096,
      "grad_norm": 0.7213401198387146,
      "learning_rate": 4.2442001931404255e-06,
      "loss": 0.6933,
      "step": 19500
    },
    {
      "epoch": 0.5618913156226589,
      "grad_norm": 0.7228655815124512,
      "learning_rate": 4.243738947976548e-06,
      "loss": 0.6957,
      "step": 19501
    },
    {
      "epoch": 0.5619201290843082,
      "grad_norm": 0.7317777276039124,
      "learning_rate": 4.243277709398949e-06,
      "loss": 0.6913,
      "step": 19502
    },
    {
      "epoch": 0.5619489425459575,
      "grad_norm": 0.7338554263114929,
      "learning_rate": 4.242816477411644e-06,
      "loss": 0.6901,
      "step": 19503
    },
    {
      "epoch": 0.5619777560076068,
      "grad_norm": 0.7255223393440247,
      "learning_rate": 4.242355252018651e-06,
      "loss": 0.6909,
      "step": 19504
    },
    {
      "epoch": 0.5620065694692561,
      "grad_norm": 0.7332341074943542,
      "learning_rate": 4.241894033223987e-06,
      "loss": 0.6979,
      "step": 19505
    },
    {
      "epoch": 0.5620353829309053,
      "grad_norm": 0.6995463967323303,
      "learning_rate": 4.241432821031665e-06,
      "loss": 0.7048,
      "step": 19506
    },
    {
      "epoch": 0.5620641963925546,
      "grad_norm": 0.7466979026794434,
      "learning_rate": 4.240971615445707e-06,
      "loss": 0.7272,
      "step": 19507
    },
    {
      "epoch": 0.5620930098542039,
      "grad_norm": 0.7258826494216919,
      "learning_rate": 4.240510416470127e-06,
      "loss": 0.6984,
      "step": 19508
    },
    {
      "epoch": 0.5621218233158531,
      "grad_norm": 0.7340562343597412,
      "learning_rate": 4.240049224108943e-06,
      "loss": 0.699,
      "step": 19509
    },
    {
      "epoch": 0.5621506367775024,
      "grad_norm": 0.7111634612083435,
      "learning_rate": 4.239588038366171e-06,
      "loss": 0.7038,
      "step": 19510
    },
    {
      "epoch": 0.5621794502391517,
      "grad_norm": 0.706802487373352,
      "learning_rate": 4.239126859245826e-06,
      "loss": 0.6955,
      "step": 19511
    },
    {
      "epoch": 0.562208263700801,
      "grad_norm": 0.72086101770401,
      "learning_rate": 4.238665686751926e-06,
      "loss": 0.7163,
      "step": 19512
    },
    {
      "epoch": 0.5622370771624503,
      "grad_norm": 0.7601830959320068,
      "learning_rate": 4.2382045208884865e-06,
      "loss": 0.7075,
      "step": 19513
    },
    {
      "epoch": 0.5622658906240996,
      "grad_norm": 0.7186294198036194,
      "learning_rate": 4.237743361659525e-06,
      "loss": 0.7179,
      "step": 19514
    },
    {
      "epoch": 0.5622947040857489,
      "grad_norm": 0.7399386167526245,
      "learning_rate": 4.237282209069054e-06,
      "loss": 0.7095,
      "step": 19515
    },
    {
      "epoch": 0.5623235175473982,
      "grad_norm": 0.7136863470077515,
      "learning_rate": 4.236821063121095e-06,
      "loss": 0.6954,
      "step": 19516
    },
    {
      "epoch": 0.5623523310090475,
      "grad_norm": 0.7356709837913513,
      "learning_rate": 4.236359923819661e-06,
      "loss": 0.7326,
      "step": 19517
    },
    {
      "epoch": 0.5623811444706968,
      "grad_norm": 0.7131011486053467,
      "learning_rate": 4.235898791168768e-06,
      "loss": 0.7191,
      "step": 19518
    },
    {
      "epoch": 0.5624099579323459,
      "grad_norm": 0.7122602462768555,
      "learning_rate": 4.235437665172434e-06,
      "loss": 0.7236,
      "step": 19519
    },
    {
      "epoch": 0.5624387713939952,
      "grad_norm": 0.7546557188034058,
      "learning_rate": 4.234976545834672e-06,
      "loss": 0.7195,
      "step": 19520
    },
    {
      "epoch": 0.5624675848556445,
      "grad_norm": 0.7153882384300232,
      "learning_rate": 4.234515433159501e-06,
      "loss": 0.6918,
      "step": 19521
    },
    {
      "epoch": 0.5624963983172938,
      "grad_norm": 0.7218692302703857,
      "learning_rate": 4.234054327150935e-06,
      "loss": 0.6823,
      "step": 19522
    },
    {
      "epoch": 0.5625252117789431,
      "grad_norm": 0.7902252674102783,
      "learning_rate": 4.23359322781299e-06,
      "loss": 0.735,
      "step": 19523
    },
    {
      "epoch": 0.5625540252405924,
      "grad_norm": 0.713310956954956,
      "learning_rate": 4.233132135149682e-06,
      "loss": 0.6688,
      "step": 19524
    },
    {
      "epoch": 0.5625828387022417,
      "grad_norm": 0.7718071937561035,
      "learning_rate": 4.232671049165027e-06,
      "loss": 0.7061,
      "step": 19525
    },
    {
      "epoch": 0.562611652163891,
      "grad_norm": 0.7432312369346619,
      "learning_rate": 4.23220996986304e-06,
      "loss": 0.7007,
      "step": 19526
    },
    {
      "epoch": 0.5626404656255403,
      "grad_norm": 0.7192946672439575,
      "learning_rate": 4.231748897247737e-06,
      "loss": 0.7121,
      "step": 19527
    },
    {
      "epoch": 0.5626692790871896,
      "grad_norm": 0.7253904342651367,
      "learning_rate": 4.231287831323133e-06,
      "loss": 0.7106,
      "step": 19528
    },
    {
      "epoch": 0.5626980925488388,
      "grad_norm": 0.739554762840271,
      "learning_rate": 4.230826772093243e-06,
      "loss": 0.696,
      "step": 19529
    },
    {
      "epoch": 0.5627269060104881,
      "grad_norm": 0.7301630973815918,
      "learning_rate": 4.230365719562081e-06,
      "loss": 0.6868,
      "step": 19530
    },
    {
      "epoch": 0.5627557194721374,
      "grad_norm": 0.7406406402587891,
      "learning_rate": 4.2299046737336655e-06,
      "loss": 0.7113,
      "step": 19531
    },
    {
      "epoch": 0.5627845329337867,
      "grad_norm": 0.7250484824180603,
      "learning_rate": 4.229443634612012e-06,
      "loss": 0.7163,
      "step": 19532
    },
    {
      "epoch": 0.5628133463954359,
      "grad_norm": 0.6947176456451416,
      "learning_rate": 4.228982602201134e-06,
      "loss": 0.6983,
      "step": 19533
    },
    {
      "epoch": 0.5628421598570852,
      "grad_norm": 0.7232486605644226,
      "learning_rate": 4.228521576505046e-06,
      "loss": 0.7046,
      "step": 19534
    },
    {
      "epoch": 0.5628709733187345,
      "grad_norm": 0.7077807784080505,
      "learning_rate": 4.2280605575277635e-06,
      "loss": 0.6778,
      "step": 19535
    },
    {
      "epoch": 0.5628997867803838,
      "grad_norm": 0.7356722950935364,
      "learning_rate": 4.227599545273302e-06,
      "loss": 0.7275,
      "step": 19536
    },
    {
      "epoch": 0.5629286002420331,
      "grad_norm": 0.7615329027175903,
      "learning_rate": 4.227138539745677e-06,
      "loss": 0.7393,
      "step": 19537
    },
    {
      "epoch": 0.5629574137036824,
      "grad_norm": 0.7224112749099731,
      "learning_rate": 4.226677540948902e-06,
      "loss": 0.7214,
      "step": 19538
    },
    {
      "epoch": 0.5629862271653316,
      "grad_norm": 0.7238897085189819,
      "learning_rate": 4.226216548886994e-06,
      "loss": 0.6925,
      "step": 19539
    },
    {
      "epoch": 0.5630150406269809,
      "grad_norm": 0.7146919965744019,
      "learning_rate": 4.225755563563966e-06,
      "loss": 0.7091,
      "step": 19540
    },
    {
      "epoch": 0.5630438540886302,
      "grad_norm": 0.7647107839584351,
      "learning_rate": 4.225294584983832e-06,
      "loss": 0.7219,
      "step": 19541
    },
    {
      "epoch": 0.5630726675502795,
      "grad_norm": 0.7331885695457458,
      "learning_rate": 4.224833613150606e-06,
      "loss": 0.7132,
      "step": 19542
    },
    {
      "epoch": 0.5631014810119288,
      "grad_norm": 0.7140501141548157,
      "learning_rate": 4.224372648068307e-06,
      "loss": 0.7182,
      "step": 19543
    },
    {
      "epoch": 0.5631302944735781,
      "grad_norm": 0.7453955411911011,
      "learning_rate": 4.223911689740947e-06,
      "loss": 0.6988,
      "step": 19544
    },
    {
      "epoch": 0.5631591079352274,
      "grad_norm": 0.7505368590354919,
      "learning_rate": 4.22345073817254e-06,
      "loss": 0.7144,
      "step": 19545
    },
    {
      "epoch": 0.5631879213968766,
      "grad_norm": 0.7272033095359802,
      "learning_rate": 4.2229897933671005e-06,
      "loss": 0.6964,
      "step": 19546
    },
    {
      "epoch": 0.5632167348585259,
      "grad_norm": 0.7183753252029419,
      "learning_rate": 4.222528855328644e-06,
      "loss": 0.71,
      "step": 19547
    },
    {
      "epoch": 0.5632455483201751,
      "grad_norm": 0.731913685798645,
      "learning_rate": 4.222067924061184e-06,
      "loss": 0.6998,
      "step": 19548
    },
    {
      "epoch": 0.5632743617818244,
      "grad_norm": 0.712299108505249,
      "learning_rate": 4.2216069995687335e-06,
      "loss": 0.7005,
      "step": 19549
    },
    {
      "epoch": 0.5633031752434737,
      "grad_norm": 0.726637065410614,
      "learning_rate": 4.221146081855308e-06,
      "loss": 0.7049,
      "step": 19550
    },
    {
      "epoch": 0.563331988705123,
      "grad_norm": 0.7285045981407166,
      "learning_rate": 4.220685170924923e-06,
      "loss": 0.7214,
      "step": 19551
    },
    {
      "epoch": 0.5633608021667723,
      "grad_norm": 0.719440758228302,
      "learning_rate": 4.22022426678159e-06,
      "loss": 0.7159,
      "step": 19552
    },
    {
      "epoch": 0.5633896156284216,
      "grad_norm": 0.7543953061103821,
      "learning_rate": 4.219763369429324e-06,
      "loss": 0.7032,
      "step": 19553
    },
    {
      "epoch": 0.5634184290900709,
      "grad_norm": 0.7461857199668884,
      "learning_rate": 4.2193024788721385e-06,
      "loss": 0.7413,
      "step": 19554
    },
    {
      "epoch": 0.5634472425517202,
      "grad_norm": 0.7586621642112732,
      "learning_rate": 4.218841595114049e-06,
      "loss": 0.7191,
      "step": 19555
    },
    {
      "epoch": 0.5634760560133695,
      "grad_norm": 0.7531154751777649,
      "learning_rate": 4.218380718159068e-06,
      "loss": 0.7186,
      "step": 19556
    },
    {
      "epoch": 0.5635048694750188,
      "grad_norm": 0.7369619011878967,
      "learning_rate": 4.21791984801121e-06,
      "loss": 0.7252,
      "step": 19557
    },
    {
      "epoch": 0.5635336829366681,
      "grad_norm": 0.7304530739784241,
      "learning_rate": 4.217458984674487e-06,
      "loss": 0.7216,
      "step": 19558
    },
    {
      "epoch": 0.5635624963983173,
      "grad_norm": 0.7145882248878479,
      "learning_rate": 4.216998128152916e-06,
      "loss": 0.6838,
      "step": 19559
    },
    {
      "epoch": 0.5635913098599665,
      "grad_norm": 0.7240351438522339,
      "learning_rate": 4.216537278450507e-06,
      "loss": 0.6809,
      "step": 19560
    },
    {
      "epoch": 0.5636201233216158,
      "grad_norm": 0.7231440544128418,
      "learning_rate": 4.216076435571277e-06,
      "loss": 0.715,
      "step": 19561
    },
    {
      "epoch": 0.5636489367832651,
      "grad_norm": 0.7421388626098633,
      "learning_rate": 4.2156155995192366e-06,
      "loss": 0.7136,
      "step": 19562
    },
    {
      "epoch": 0.5636777502449144,
      "grad_norm": 0.7486557960510254,
      "learning_rate": 4.2151547702984e-06,
      "loss": 0.7051,
      "step": 19563
    },
    {
      "epoch": 0.5637065637065637,
      "grad_norm": 0.7262318730354309,
      "learning_rate": 4.214693947912781e-06,
      "loss": 0.7025,
      "step": 19564
    },
    {
      "epoch": 0.563735377168213,
      "grad_norm": 0.7164513468742371,
      "learning_rate": 4.214233132366393e-06,
      "loss": 0.7053,
      "step": 19565
    },
    {
      "epoch": 0.5637641906298623,
      "grad_norm": 0.7152954936027527,
      "learning_rate": 4.213772323663247e-06,
      "loss": 0.7059,
      "step": 19566
    },
    {
      "epoch": 0.5637930040915116,
      "grad_norm": 0.7224941849708557,
      "learning_rate": 4.213311521807359e-06,
      "loss": 0.7182,
      "step": 19567
    },
    {
      "epoch": 0.5638218175531609,
      "grad_norm": 0.737392246723175,
      "learning_rate": 4.2128507268027426e-06,
      "loss": 0.6889,
      "step": 19568
    },
    {
      "epoch": 0.5638506310148101,
      "grad_norm": 0.7555577158927917,
      "learning_rate": 4.21238993865341e-06,
      "loss": 0.7313,
      "step": 19569
    },
    {
      "epoch": 0.5638794444764594,
      "grad_norm": 0.7333041429519653,
      "learning_rate": 4.211929157363372e-06,
      "loss": 0.7226,
      "step": 19570
    },
    {
      "epoch": 0.5639082579381087,
      "grad_norm": 0.7374141216278076,
      "learning_rate": 4.211468382936644e-06,
      "loss": 0.7149,
      "step": 19571
    },
    {
      "epoch": 0.563937071399758,
      "grad_norm": 0.7411205172538757,
      "learning_rate": 4.211007615377238e-06,
      "loss": 0.7231,
      "step": 19572
    },
    {
      "epoch": 0.5639658848614072,
      "grad_norm": 0.747115969657898,
      "learning_rate": 4.210546854689167e-06,
      "loss": 0.7139,
      "step": 19573
    },
    {
      "epoch": 0.5639946983230565,
      "grad_norm": 0.7555550932884216,
      "learning_rate": 4.2100861008764425e-06,
      "loss": 0.7106,
      "step": 19574
    },
    {
      "epoch": 0.5640235117847058,
      "grad_norm": 0.715432345867157,
      "learning_rate": 4.20962535394308e-06,
      "loss": 0.7201,
      "step": 19575
    },
    {
      "epoch": 0.5640523252463551,
      "grad_norm": 0.7214940786361694,
      "learning_rate": 4.209164613893091e-06,
      "loss": 0.6855,
      "step": 19576
    },
    {
      "epoch": 0.5640811387080044,
      "grad_norm": 0.748920202255249,
      "learning_rate": 4.208703880730486e-06,
      "loss": 0.7116,
      "step": 19577
    },
    {
      "epoch": 0.5641099521696537,
      "grad_norm": 0.7079659700393677,
      "learning_rate": 4.208243154459279e-06,
      "loss": 0.699,
      "step": 19578
    },
    {
      "epoch": 0.5641387656313029,
      "grad_norm": 0.716941237449646,
      "learning_rate": 4.207782435083483e-06,
      "loss": 0.7035,
      "step": 19579
    },
    {
      "epoch": 0.5641675790929522,
      "grad_norm": 0.7200295925140381,
      "learning_rate": 4.207321722607111e-06,
      "loss": 0.7179,
      "step": 19580
    },
    {
      "epoch": 0.5641963925546015,
      "grad_norm": 0.7015926837921143,
      "learning_rate": 4.206861017034174e-06,
      "loss": 0.6948,
      "step": 19581
    },
    {
      "epoch": 0.5642252060162508,
      "grad_norm": 0.6943456530570984,
      "learning_rate": 4.206400318368686e-06,
      "loss": 0.6881,
      "step": 19582
    },
    {
      "epoch": 0.5642540194779001,
      "grad_norm": 0.7617952227592468,
      "learning_rate": 4.2059396266146564e-06,
      "loss": 0.7265,
      "step": 19583
    },
    {
      "epoch": 0.5642828329395494,
      "grad_norm": 0.7307103872299194,
      "learning_rate": 4.2054789417761e-06,
      "loss": 0.7343,
      "step": 19584
    },
    {
      "epoch": 0.5643116464011987,
      "grad_norm": 0.709835946559906,
      "learning_rate": 4.205018263857027e-06,
      "loss": 0.7118,
      "step": 19585
    },
    {
      "epoch": 0.564340459862848,
      "grad_norm": 0.7112618684768677,
      "learning_rate": 4.20455759286145e-06,
      "loss": 0.7012,
      "step": 19586
    },
    {
      "epoch": 0.5643692733244972,
      "grad_norm": 0.7383577227592468,
      "learning_rate": 4.204096928793382e-06,
      "loss": 0.705,
      "step": 19587
    },
    {
      "epoch": 0.5643980867861464,
      "grad_norm": 0.7422275543212891,
      "learning_rate": 4.203636271656833e-06,
      "loss": 0.6947,
      "step": 19588
    },
    {
      "epoch": 0.5644269002477957,
      "grad_norm": 0.7192673087120056,
      "learning_rate": 4.2031756214558175e-06,
      "loss": 0.7181,
      "step": 19589
    },
    {
      "epoch": 0.564455713709445,
      "grad_norm": 0.7246209383010864,
      "learning_rate": 4.202714978194344e-06,
      "loss": 0.7449,
      "step": 19590
    },
    {
      "epoch": 0.5644845271710943,
      "grad_norm": 0.7288272380828857,
      "learning_rate": 4.2022543418764265e-06,
      "loss": 0.7115,
      "step": 19591
    },
    {
      "epoch": 0.5645133406327436,
      "grad_norm": 0.705743670463562,
      "learning_rate": 4.201793712506076e-06,
      "loss": 0.7125,
      "step": 19592
    },
    {
      "epoch": 0.5645421540943929,
      "grad_norm": 0.7174003720283508,
      "learning_rate": 4.201333090087306e-06,
      "loss": 0.7084,
      "step": 19593
    },
    {
      "epoch": 0.5645709675560422,
      "grad_norm": 0.7229511737823486,
      "learning_rate": 4.200872474624125e-06,
      "loss": 0.7224,
      "step": 19594
    },
    {
      "epoch": 0.5645997810176915,
      "grad_norm": 0.705208420753479,
      "learning_rate": 4.2004118661205464e-06,
      "loss": 0.689,
      "step": 19595
    },
    {
      "epoch": 0.5646285944793408,
      "grad_norm": 0.7282698154449463,
      "learning_rate": 4.199951264580581e-06,
      "loss": 0.6971,
      "step": 19596
    },
    {
      "epoch": 0.5646574079409901,
      "grad_norm": 0.738632321357727,
      "learning_rate": 4.19949067000824e-06,
      "loss": 0.7419,
      "step": 19597
    },
    {
      "epoch": 0.5646862214026394,
      "grad_norm": 0.7246930003166199,
      "learning_rate": 4.199030082407536e-06,
      "loss": 0.7172,
      "step": 19598
    },
    {
      "epoch": 0.5647150348642886,
      "grad_norm": 0.749721109867096,
      "learning_rate": 4.198569501782479e-06,
      "loss": 0.7379,
      "step": 19599
    },
    {
      "epoch": 0.5647438483259378,
      "grad_norm": 0.695932924747467,
      "learning_rate": 4.19810892813708e-06,
      "loss": 0.692,
      "step": 19600
    },
    {
      "epoch": 0.5647726617875871,
      "grad_norm": 0.723176896572113,
      "learning_rate": 4.197648361475351e-06,
      "loss": 0.7216,
      "step": 19601
    },
    {
      "epoch": 0.5648014752492364,
      "grad_norm": 0.7576623558998108,
      "learning_rate": 4.197187801801301e-06,
      "loss": 0.7123,
      "step": 19602
    },
    {
      "epoch": 0.5648302887108857,
      "grad_norm": 0.7368777394294739,
      "learning_rate": 4.196727249118945e-06,
      "loss": 0.6932,
      "step": 19603
    },
    {
      "epoch": 0.564859102172535,
      "grad_norm": 0.773543119430542,
      "learning_rate": 4.196266703432291e-06,
      "loss": 0.7009,
      "step": 19604
    },
    {
      "epoch": 0.5648879156341843,
      "grad_norm": 0.7050905227661133,
      "learning_rate": 4.1958061647453505e-06,
      "loss": 0.6907,
      "step": 19605
    },
    {
      "epoch": 0.5649167290958336,
      "grad_norm": 0.7007887363433838,
      "learning_rate": 4.195345633062135e-06,
      "loss": 0.6894,
      "step": 19606
    },
    {
      "epoch": 0.5649455425574829,
      "grad_norm": 0.7163359522819519,
      "learning_rate": 4.1948851083866545e-06,
      "loss": 0.6895,
      "step": 19607
    },
    {
      "epoch": 0.5649743560191322,
      "grad_norm": 0.7163243889808655,
      "learning_rate": 4.194424590722919e-06,
      "loss": 0.7058,
      "step": 19608
    },
    {
      "epoch": 0.5650031694807814,
      "grad_norm": 0.7411450147628784,
      "learning_rate": 4.193964080074941e-06,
      "loss": 0.7088,
      "step": 19609
    },
    {
      "epoch": 0.5650319829424307,
      "grad_norm": 0.7602841854095459,
      "learning_rate": 4.193503576446731e-06,
      "loss": 0.7378,
      "step": 19610
    },
    {
      "epoch": 0.56506079640408,
      "grad_norm": 0.7586194276809692,
      "learning_rate": 4.193043079842297e-06,
      "loss": 0.7316,
      "step": 19611
    },
    {
      "epoch": 0.5650896098657293,
      "grad_norm": 0.696464478969574,
      "learning_rate": 4.192582590265651e-06,
      "loss": 0.7112,
      "step": 19612
    },
    {
      "epoch": 0.5651184233273786,
      "grad_norm": 0.746454119682312,
      "learning_rate": 4.192122107720804e-06,
      "loss": 0.7253,
      "step": 19613
    },
    {
      "epoch": 0.5651472367890278,
      "grad_norm": 0.7672178745269775,
      "learning_rate": 4.191661632211766e-06,
      "loss": 0.7325,
      "step": 19614
    },
    {
      "epoch": 0.5651760502506771,
      "grad_norm": 0.7096680998802185,
      "learning_rate": 4.191201163742546e-06,
      "loss": 0.7081,
      "step": 19615
    },
    {
      "epoch": 0.5652048637123264,
      "grad_norm": 0.7280484437942505,
      "learning_rate": 4.190740702317157e-06,
      "loss": 0.694,
      "step": 19616
    },
    {
      "epoch": 0.5652336771739757,
      "grad_norm": 0.7430151104927063,
      "learning_rate": 4.1902802479396056e-06,
      "loss": 0.7089,
      "step": 19617
    },
    {
      "epoch": 0.565262490635625,
      "grad_norm": 0.7173912525177002,
      "learning_rate": 4.189819800613906e-06,
      "loss": 0.6996,
      "step": 19618
    },
    {
      "epoch": 0.5652913040972742,
      "grad_norm": 0.7621369361877441,
      "learning_rate": 4.189359360344065e-06,
      "loss": 0.7083,
      "step": 19619
    },
    {
      "epoch": 0.5653201175589235,
      "grad_norm": 0.7138421535491943,
      "learning_rate": 4.1888989271340945e-06,
      "loss": 0.7248,
      "step": 19620
    },
    {
      "epoch": 0.5653489310205728,
      "grad_norm": 0.6982003450393677,
      "learning_rate": 4.188438500988003e-06,
      "loss": 0.699,
      "step": 19621
    },
    {
      "epoch": 0.5653777444822221,
      "grad_norm": 0.7300769090652466,
      "learning_rate": 4.187978081909801e-06,
      "loss": 0.6999,
      "step": 19622
    },
    {
      "epoch": 0.5654065579438714,
      "grad_norm": 0.7200302481651306,
      "learning_rate": 4.187517669903498e-06,
      "loss": 0.7152,
      "step": 19623
    },
    {
      "epoch": 0.5654353714055207,
      "grad_norm": 0.7205941677093506,
      "learning_rate": 4.187057264973104e-06,
      "loss": 0.7299,
      "step": 19624
    },
    {
      "epoch": 0.56546418486717,
      "grad_norm": 0.7260777950286865,
      "learning_rate": 4.186596867122628e-06,
      "loss": 0.6933,
      "step": 19625
    },
    {
      "epoch": 0.5654929983288193,
      "grad_norm": 0.7154983878135681,
      "learning_rate": 4.1861364763560785e-06,
      "loss": 0.7037,
      "step": 19626
    },
    {
      "epoch": 0.5655218117904685,
      "grad_norm": 0.7795827388763428,
      "learning_rate": 4.185676092677469e-06,
      "loss": 0.6999,
      "step": 19627
    },
    {
      "epoch": 0.5655506252521177,
      "grad_norm": 0.7066949009895325,
      "learning_rate": 4.185215716090807e-06,
      "loss": 0.7211,
      "step": 19628
    },
    {
      "epoch": 0.565579438713767,
      "grad_norm": 0.6932422518730164,
      "learning_rate": 4.184755346600101e-06,
      "loss": 0.681,
      "step": 19629
    },
    {
      "epoch": 0.5656082521754163,
      "grad_norm": 0.694231390953064,
      "learning_rate": 4.1842949842093615e-06,
      "loss": 0.6789,
      "step": 19630
    },
    {
      "epoch": 0.5656370656370656,
      "grad_norm": 0.7331354022026062,
      "learning_rate": 4.183834628922597e-06,
      "loss": 0.7011,
      "step": 19631
    },
    {
      "epoch": 0.5656658790987149,
      "grad_norm": 0.7402907609939575,
      "learning_rate": 4.183374280743817e-06,
      "loss": 0.727,
      "step": 19632
    },
    {
      "epoch": 0.5656946925603642,
      "grad_norm": 0.7120527625083923,
      "learning_rate": 4.18291393967703e-06,
      "loss": 0.6941,
      "step": 19633
    },
    {
      "epoch": 0.5657235060220135,
      "grad_norm": 0.7385700345039368,
      "learning_rate": 4.182453605726246e-06,
      "loss": 0.7101,
      "step": 19634
    },
    {
      "epoch": 0.5657523194836628,
      "grad_norm": 0.7244153022766113,
      "learning_rate": 4.181993278895475e-06,
      "loss": 0.7299,
      "step": 19635
    },
    {
      "epoch": 0.5657811329453121,
      "grad_norm": 0.7612526416778564,
      "learning_rate": 4.181532959188724e-06,
      "loss": 0.6904,
      "step": 19636
    },
    {
      "epoch": 0.5658099464069614,
      "grad_norm": 0.7631896734237671,
      "learning_rate": 4.181072646610004e-06,
      "loss": 0.7417,
      "step": 19637
    },
    {
      "epoch": 0.5658387598686107,
      "grad_norm": 0.761986255645752,
      "learning_rate": 4.180612341163319e-06,
      "loss": 0.7029,
      "step": 19638
    },
    {
      "epoch": 0.5658675733302599,
      "grad_norm": 0.739495038986206,
      "learning_rate": 4.180152042852684e-06,
      "loss": 0.7219,
      "step": 19639
    },
    {
      "epoch": 0.5658963867919092,
      "grad_norm": 0.7294878363609314,
      "learning_rate": 4.179691751682106e-06,
      "loss": 0.7316,
      "step": 19640
    },
    {
      "epoch": 0.5659252002535584,
      "grad_norm": 0.7286807298660278,
      "learning_rate": 4.179231467655591e-06,
      "loss": 0.6958,
      "step": 19641
    },
    {
      "epoch": 0.5659540137152077,
      "grad_norm": 0.7340981960296631,
      "learning_rate": 4.1787711907771515e-06,
      "loss": 0.721,
      "step": 19642
    },
    {
      "epoch": 0.565982827176857,
      "grad_norm": 0.728818416595459,
      "learning_rate": 4.178310921050793e-06,
      "loss": 0.7261,
      "step": 19643
    },
    {
      "epoch": 0.5660116406385063,
      "grad_norm": 0.7058225274085999,
      "learning_rate": 4.177850658480524e-06,
      "loss": 0.7002,
      "step": 19644
    },
    {
      "epoch": 0.5660404541001556,
      "grad_norm": 0.7249895334243774,
      "learning_rate": 4.1773904030703554e-06,
      "loss": 0.7248,
      "step": 19645
    },
    {
      "epoch": 0.5660692675618049,
      "grad_norm": 0.718288004398346,
      "learning_rate": 4.176930154824293e-06,
      "loss": 0.6983,
      "step": 19646
    },
    {
      "epoch": 0.5660980810234542,
      "grad_norm": 0.741279125213623,
      "learning_rate": 4.176469913746348e-06,
      "loss": 0.7095,
      "step": 19647
    },
    {
      "epoch": 0.5661268944851034,
      "grad_norm": 0.7467444539070129,
      "learning_rate": 4.176009679840526e-06,
      "loss": 0.7063,
      "step": 19648
    },
    {
      "epoch": 0.5661557079467527,
      "grad_norm": 0.7180765271186829,
      "learning_rate": 4.175549453110835e-06,
      "loss": 0.7255,
      "step": 19649
    },
    {
      "epoch": 0.566184521408402,
      "grad_norm": 0.740763783454895,
      "learning_rate": 4.1750892335612824e-06,
      "loss": 0.7082,
      "step": 19650
    },
    {
      "epoch": 0.5662133348700513,
      "grad_norm": 0.8670125007629395,
      "learning_rate": 4.17462902119588e-06,
      "loss": 0.7211,
      "step": 19651
    },
    {
      "epoch": 0.5662421483317006,
      "grad_norm": 0.7159919142723083,
      "learning_rate": 4.174168816018634e-06,
      "loss": 0.7331,
      "step": 19652
    },
    {
      "epoch": 0.5662709617933499,
      "grad_norm": 0.7257015705108643,
      "learning_rate": 4.173708618033552e-06,
      "loss": 0.6859,
      "step": 19653
    },
    {
      "epoch": 0.5662997752549991,
      "grad_norm": 0.7380920052528381,
      "learning_rate": 4.173248427244643e-06,
      "loss": 0.7229,
      "step": 19654
    },
    {
      "epoch": 0.5663285887166484,
      "grad_norm": 0.8145814538002014,
      "learning_rate": 4.172788243655912e-06,
      "loss": 0.7125,
      "step": 19655
    },
    {
      "epoch": 0.5663574021782977,
      "grad_norm": 0.7569436430931091,
      "learning_rate": 4.172328067271368e-06,
      "loss": 0.7255,
      "step": 19656
    },
    {
      "epoch": 0.566386215639947,
      "grad_norm": 0.7064402103424072,
      "learning_rate": 4.171867898095021e-06,
      "loss": 0.6789,
      "step": 19657
    },
    {
      "epoch": 0.5664150291015962,
      "grad_norm": 0.7234563827514648,
      "learning_rate": 4.1714077361308764e-06,
      "loss": 0.7212,
      "step": 19658
    },
    {
      "epoch": 0.5664438425632455,
      "grad_norm": 0.7424296140670776,
      "learning_rate": 4.170947581382942e-06,
      "loss": 0.7245,
      "step": 19659
    },
    {
      "epoch": 0.5664726560248948,
      "grad_norm": 0.7183544635772705,
      "learning_rate": 4.170487433855225e-06,
      "loss": 0.6929,
      "step": 19660
    },
    {
      "epoch": 0.5665014694865441,
      "grad_norm": 0.7269642353057861,
      "learning_rate": 4.170027293551734e-06,
      "loss": 0.693,
      "step": 19661
    },
    {
      "epoch": 0.5665302829481934,
      "grad_norm": 0.6969347596168518,
      "learning_rate": 4.169567160476473e-06,
      "loss": 0.6831,
      "step": 19662
    },
    {
      "epoch": 0.5665590964098427,
      "grad_norm": 0.7235945463180542,
      "learning_rate": 4.169107034633455e-06,
      "loss": 0.714,
      "step": 19663
    },
    {
      "epoch": 0.566587909871492,
      "grad_norm": 0.6806153059005737,
      "learning_rate": 4.168646916026682e-06,
      "loss": 0.6858,
      "step": 19664
    },
    {
      "epoch": 0.5666167233331413,
      "grad_norm": 0.7504202723503113,
      "learning_rate": 4.168186804660165e-06,
      "loss": 0.701,
      "step": 19665
    },
    {
      "epoch": 0.5666455367947906,
      "grad_norm": 0.7215688228607178,
      "learning_rate": 4.167726700537909e-06,
      "loss": 0.7196,
      "step": 19666
    },
    {
      "epoch": 0.5666743502564399,
      "grad_norm": 0.7322152853012085,
      "learning_rate": 4.167266603663922e-06,
      "loss": 0.7056,
      "step": 19667
    },
    {
      "epoch": 0.566703163718089,
      "grad_norm": 0.7214760184288025,
      "learning_rate": 4.1668065140422095e-06,
      "loss": 0.722,
      "step": 19668
    },
    {
      "epoch": 0.5667319771797383,
      "grad_norm": 0.725416362285614,
      "learning_rate": 4.16634643167678e-06,
      "loss": 0.7077,
      "step": 19669
    },
    {
      "epoch": 0.5667607906413876,
      "grad_norm": 0.7035599946975708,
      "learning_rate": 4.16588635657164e-06,
      "loss": 0.6958,
      "step": 19670
    },
    {
      "epoch": 0.5667896041030369,
      "grad_norm": 0.7138715386390686,
      "learning_rate": 4.165426288730794e-06,
      "loss": 0.7312,
      "step": 19671
    },
    {
      "epoch": 0.5668184175646862,
      "grad_norm": 0.7372941970825195,
      "learning_rate": 4.164966228158253e-06,
      "loss": 0.7065,
      "step": 19672
    },
    {
      "epoch": 0.5668472310263355,
      "grad_norm": 0.7074500918388367,
      "learning_rate": 4.16450617485802e-06,
      "loss": 0.6916,
      "step": 19673
    },
    {
      "epoch": 0.5668760444879848,
      "grad_norm": 0.7272542119026184,
      "learning_rate": 4.164046128834103e-06,
      "loss": 0.6975,
      "step": 19674
    },
    {
      "epoch": 0.5669048579496341,
      "grad_norm": 0.706821620464325,
      "learning_rate": 4.163586090090509e-06,
      "loss": 0.7118,
      "step": 19675
    },
    {
      "epoch": 0.5669336714112834,
      "grad_norm": 0.707883894443512,
      "learning_rate": 4.163126058631244e-06,
      "loss": 0.6964,
      "step": 19676
    },
    {
      "epoch": 0.5669624848729327,
      "grad_norm": 0.7271609306335449,
      "learning_rate": 4.162666034460314e-06,
      "loss": 0.7351,
      "step": 19677
    },
    {
      "epoch": 0.566991298334582,
      "grad_norm": 0.7229691743850708,
      "learning_rate": 4.162206017581727e-06,
      "loss": 0.6962,
      "step": 19678
    },
    {
      "epoch": 0.5670201117962312,
      "grad_norm": 0.7369662523269653,
      "learning_rate": 4.161746007999488e-06,
      "loss": 0.7278,
      "step": 19679
    },
    {
      "epoch": 0.5670489252578805,
      "grad_norm": 0.6963277459144592,
      "learning_rate": 4.161286005717603e-06,
      "loss": 0.6983,
      "step": 19680
    },
    {
      "epoch": 0.5670777387195297,
      "grad_norm": 0.7143937945365906,
      "learning_rate": 4.160826010740079e-06,
      "loss": 0.7216,
      "step": 19681
    },
    {
      "epoch": 0.567106552181179,
      "grad_norm": 0.732300877571106,
      "learning_rate": 4.1603660230709205e-06,
      "loss": 0.6977,
      "step": 19682
    },
    {
      "epoch": 0.5671353656428283,
      "grad_norm": 0.7163258194923401,
      "learning_rate": 4.1599060427141355e-06,
      "loss": 0.6871,
      "step": 19683
    },
    {
      "epoch": 0.5671641791044776,
      "grad_norm": 0.7107528448104858,
      "learning_rate": 4.15944606967373e-06,
      "loss": 0.7027,
      "step": 19684
    },
    {
      "epoch": 0.5671929925661269,
      "grad_norm": 0.7344526052474976,
      "learning_rate": 4.158986103953707e-06,
      "loss": 0.708,
      "step": 19685
    },
    {
      "epoch": 0.5672218060277762,
      "grad_norm": 0.7101230025291443,
      "learning_rate": 4.158526145558073e-06,
      "loss": 0.6837,
      "step": 19686
    },
    {
      "epoch": 0.5672506194894255,
      "grad_norm": 0.7348304986953735,
      "learning_rate": 4.158066194490837e-06,
      "loss": 0.6957,
      "step": 19687
    },
    {
      "epoch": 0.5672794329510747,
      "grad_norm": 0.71766197681427,
      "learning_rate": 4.157606250756003e-06,
      "loss": 0.7159,
      "step": 19688
    },
    {
      "epoch": 0.567308246412724,
      "grad_norm": 0.7562296986579895,
      "learning_rate": 4.157146314357577e-06,
      "loss": 0.7215,
      "step": 19689
    },
    {
      "epoch": 0.5673370598743733,
      "grad_norm": 0.6893007755279541,
      "learning_rate": 4.156686385299564e-06,
      "loss": 0.6917,
      "step": 19690
    },
    {
      "epoch": 0.5673658733360226,
      "grad_norm": 0.7330052852630615,
      "learning_rate": 4.156226463585969e-06,
      "loss": 0.6705,
      "step": 19691
    },
    {
      "epoch": 0.5673946867976719,
      "grad_norm": 0.7322446703910828,
      "learning_rate": 4.155766549220798e-06,
      "loss": 0.6819,
      "step": 19692
    },
    {
      "epoch": 0.5674235002593212,
      "grad_norm": 0.731306791305542,
      "learning_rate": 4.155306642208056e-06,
      "loss": 0.7186,
      "step": 19693
    },
    {
      "epoch": 0.5674523137209705,
      "grad_norm": 0.7723024487495422,
      "learning_rate": 4.154846742551749e-06,
      "loss": 0.7258,
      "step": 19694
    },
    {
      "epoch": 0.5674811271826197,
      "grad_norm": 0.764153242111206,
      "learning_rate": 4.154386850255883e-06,
      "loss": 0.6984,
      "step": 19695
    },
    {
      "epoch": 0.567509940644269,
      "grad_norm": 0.7525814771652222,
      "learning_rate": 4.1539269653244615e-06,
      "loss": 0.7088,
      "step": 19696
    },
    {
      "epoch": 0.5675387541059183,
      "grad_norm": 0.6964426636695862,
      "learning_rate": 4.15346708776149e-06,
      "loss": 0.6951,
      "step": 19697
    },
    {
      "epoch": 0.5675675675675675,
      "grad_norm": 0.726674497127533,
      "learning_rate": 4.153007217570973e-06,
      "loss": 0.7196,
      "step": 19698
    },
    {
      "epoch": 0.5675963810292168,
      "grad_norm": 0.7238730788230896,
      "learning_rate": 4.152547354756918e-06,
      "loss": 0.7265,
      "step": 19699
    },
    {
      "epoch": 0.5676251944908661,
      "grad_norm": 0.744804322719574,
      "learning_rate": 4.152087499323329e-06,
      "loss": 0.72,
      "step": 19700
    },
    {
      "epoch": 0.5676540079525154,
      "grad_norm": 0.7477046847343445,
      "learning_rate": 4.151627651274209e-06,
      "loss": 0.6997,
      "step": 19701
    },
    {
      "epoch": 0.5676828214141647,
      "grad_norm": 0.7714676856994629,
      "learning_rate": 4.1511678106135645e-06,
      "loss": 0.6838,
      "step": 19702
    },
    {
      "epoch": 0.567711634875814,
      "grad_norm": 0.7290273308753967,
      "learning_rate": 4.1507079773454e-06,
      "loss": 0.7051,
      "step": 19703
    },
    {
      "epoch": 0.5677404483374633,
      "grad_norm": 0.709766685962677,
      "learning_rate": 4.1502481514737205e-06,
      "loss": 0.7252,
      "step": 19704
    },
    {
      "epoch": 0.5677692617991126,
      "grad_norm": 0.7302556037902832,
      "learning_rate": 4.14978833300253e-06,
      "loss": 0.7158,
      "step": 19705
    },
    {
      "epoch": 0.5677980752607619,
      "grad_norm": 0.7668378353118896,
      "learning_rate": 4.149328521935832e-06,
      "loss": 0.7068,
      "step": 19706
    },
    {
      "epoch": 0.5678268887224112,
      "grad_norm": 0.7670040726661682,
      "learning_rate": 4.148868718277634e-06,
      "loss": 0.728,
      "step": 19707
    },
    {
      "epoch": 0.5678557021840603,
      "grad_norm": 0.7483181953430176,
      "learning_rate": 4.148408922031938e-06,
      "loss": 0.7022,
      "step": 19708
    },
    {
      "epoch": 0.5678845156457096,
      "grad_norm": 0.7214446663856506,
      "learning_rate": 4.1479491332027475e-06,
      "loss": 0.685,
      "step": 19709
    },
    {
      "epoch": 0.5679133291073589,
      "grad_norm": 0.7780531048774719,
      "learning_rate": 4.1474893517940685e-06,
      "loss": 0.7098,
      "step": 19710
    },
    {
      "epoch": 0.5679421425690082,
      "grad_norm": 0.7201117873191833,
      "learning_rate": 4.147029577809906e-06,
      "loss": 0.7185,
      "step": 19711
    },
    {
      "epoch": 0.5679709560306575,
      "grad_norm": 0.7106620073318481,
      "learning_rate": 4.146569811254264e-06,
      "loss": 0.6951,
      "step": 19712
    },
    {
      "epoch": 0.5679997694923068,
      "grad_norm": 0.7257257103919983,
      "learning_rate": 4.146110052131145e-06,
      "loss": 0.7152,
      "step": 19713
    },
    {
      "epoch": 0.5680285829539561,
      "grad_norm": 0.7639932036399841,
      "learning_rate": 4.145650300444553e-06,
      "loss": 0.7269,
      "step": 19714
    },
    {
      "epoch": 0.5680573964156054,
      "grad_norm": 0.7316114902496338,
      "learning_rate": 4.145190556198494e-06,
      "loss": 0.7019,
      "step": 19715
    },
    {
      "epoch": 0.5680862098772547,
      "grad_norm": 0.7019981741905212,
      "learning_rate": 4.144730819396971e-06,
      "loss": 0.6941,
      "step": 19716
    },
    {
      "epoch": 0.568115023338904,
      "grad_norm": 0.7278223633766174,
      "learning_rate": 4.1442710900439875e-06,
      "loss": 0.6964,
      "step": 19717
    },
    {
      "epoch": 0.5681438368005532,
      "grad_norm": 0.7336973547935486,
      "learning_rate": 4.143811368143547e-06,
      "loss": 0.7015,
      "step": 19718
    },
    {
      "epoch": 0.5681726502622025,
      "grad_norm": 0.7029855251312256,
      "learning_rate": 4.143351653699654e-06,
      "loss": 0.7188,
      "step": 19719
    },
    {
      "epoch": 0.5682014637238518,
      "grad_norm": 0.7311148047447205,
      "learning_rate": 4.142891946716311e-06,
      "loss": 0.6799,
      "step": 19720
    },
    {
      "epoch": 0.5682302771855011,
      "grad_norm": 0.7266220450401306,
      "learning_rate": 4.1424322471975214e-06,
      "loss": 0.6985,
      "step": 19721
    },
    {
      "epoch": 0.5682590906471503,
      "grad_norm": 0.7469485402107239,
      "learning_rate": 4.141972555147291e-06,
      "loss": 0.7376,
      "step": 19722
    },
    {
      "epoch": 0.5682879041087996,
      "grad_norm": 0.7713415622711182,
      "learning_rate": 4.141512870569622e-06,
      "loss": 0.6964,
      "step": 19723
    },
    {
      "epoch": 0.5683167175704489,
      "grad_norm": 0.74601811170578,
      "learning_rate": 4.1410531934685185e-06,
      "loss": 0.692,
      "step": 19724
    },
    {
      "epoch": 0.5683455310320982,
      "grad_norm": 0.7604082822799683,
      "learning_rate": 4.140593523847983e-06,
      "loss": 0.7098,
      "step": 19725
    },
    {
      "epoch": 0.5683743444937475,
      "grad_norm": 0.7748610377311707,
      "learning_rate": 4.140133861712017e-06,
      "loss": 0.7087,
      "step": 19726
    },
    {
      "epoch": 0.5684031579553968,
      "grad_norm": 0.7519271373748779,
      "learning_rate": 4.139674207064627e-06,
      "loss": 0.7012,
      "step": 19727
    },
    {
      "epoch": 0.568431971417046,
      "grad_norm": 0.7373956441879272,
      "learning_rate": 4.139214559909814e-06,
      "loss": 0.7204,
      "step": 19728
    },
    {
      "epoch": 0.5684607848786953,
      "grad_norm": 0.7473019957542419,
      "learning_rate": 4.138754920251583e-06,
      "loss": 0.7094,
      "step": 19729
    },
    {
      "epoch": 0.5684895983403446,
      "grad_norm": 0.6836168169975281,
      "learning_rate": 4.138295288093935e-06,
      "loss": 0.7015,
      "step": 19730
    },
    {
      "epoch": 0.5685184118019939,
      "grad_norm": 0.7056286334991455,
      "learning_rate": 4.137835663440872e-06,
      "loss": 0.6919,
      "step": 19731
    },
    {
      "epoch": 0.5685472252636432,
      "grad_norm": 0.7930786609649658,
      "learning_rate": 4.1373760462964e-06,
      "loss": 0.7,
      "step": 19732
    },
    {
      "epoch": 0.5685760387252925,
      "grad_norm": 0.7147344350814819,
      "learning_rate": 4.1369164366645194e-06,
      "loss": 0.6893,
      "step": 19733
    },
    {
      "epoch": 0.5686048521869418,
      "grad_norm": 0.7058773040771484,
      "learning_rate": 4.136456834549235e-06,
      "loss": 0.6859,
      "step": 19734
    },
    {
      "epoch": 0.568633665648591,
      "grad_norm": 0.7037660479545593,
      "learning_rate": 4.1359972399545465e-06,
      "loss": 0.7214,
      "step": 19735
    },
    {
      "epoch": 0.5686624791102403,
      "grad_norm": 0.7472472190856934,
      "learning_rate": 4.13553765288446e-06,
      "loss": 0.7263,
      "step": 19736
    },
    {
      "epoch": 0.5686912925718896,
      "grad_norm": 0.7272248268127441,
      "learning_rate": 4.135078073342977e-06,
      "loss": 0.7233,
      "step": 19737
    },
    {
      "epoch": 0.5687201060335388,
      "grad_norm": 0.7132159471511841,
      "learning_rate": 4.134618501334098e-06,
      "loss": 0.7152,
      "step": 19738
    },
    {
      "epoch": 0.5687489194951881,
      "grad_norm": 0.7332382202148438,
      "learning_rate": 4.134158936861828e-06,
      "loss": 0.714,
      "step": 19739
    },
    {
      "epoch": 0.5687777329568374,
      "grad_norm": 0.7252174615859985,
      "learning_rate": 4.133699379930168e-06,
      "loss": 0.7083,
      "step": 19740
    },
    {
      "epoch": 0.5688065464184867,
      "grad_norm": 0.7501510977745056,
      "learning_rate": 4.13323983054312e-06,
      "loss": 0.7159,
      "step": 19741
    },
    {
      "epoch": 0.568835359880136,
      "grad_norm": 0.7288205623626709,
      "learning_rate": 4.132780288704687e-06,
      "loss": 0.6897,
      "step": 19742
    },
    {
      "epoch": 0.5688641733417853,
      "grad_norm": 0.6994662880897522,
      "learning_rate": 4.13232075441887e-06,
      "loss": 0.7054,
      "step": 19743
    },
    {
      "epoch": 0.5688929868034346,
      "grad_norm": 0.7186816334724426,
      "learning_rate": 4.131861227689673e-06,
      "loss": 0.7179,
      "step": 19744
    },
    {
      "epoch": 0.5689218002650839,
      "grad_norm": 0.7209891080856323,
      "learning_rate": 4.131401708521094e-06,
      "loss": 0.727,
      "step": 19745
    },
    {
      "epoch": 0.5689506137267332,
      "grad_norm": 0.7149856090545654,
      "learning_rate": 4.13094219691714e-06,
      "loss": 0.7072,
      "step": 19746
    },
    {
      "epoch": 0.5689794271883825,
      "grad_norm": 0.7439841628074646,
      "learning_rate": 4.130482692881811e-06,
      "loss": 0.7261,
      "step": 19747
    },
    {
      "epoch": 0.5690082406500317,
      "grad_norm": 0.6938281059265137,
      "learning_rate": 4.130023196419109e-06,
      "loss": 0.6957,
      "step": 19748
    },
    {
      "epoch": 0.5690370541116809,
      "grad_norm": 0.7473028302192688,
      "learning_rate": 4.1295637075330344e-06,
      "loss": 0.7151,
      "step": 19749
    },
    {
      "epoch": 0.5690658675733302,
      "grad_norm": 0.7040585279464722,
      "learning_rate": 4.12910422622759e-06,
      "loss": 0.7227,
      "step": 19750
    },
    {
      "epoch": 0.5690946810349795,
      "grad_norm": 0.7607155442237854,
      "learning_rate": 4.128644752506777e-06,
      "loss": 0.7089,
      "step": 19751
    },
    {
      "epoch": 0.5691234944966288,
      "grad_norm": 0.7576556205749512,
      "learning_rate": 4.128185286374596e-06,
      "loss": 0.7226,
      "step": 19752
    },
    {
      "epoch": 0.5691523079582781,
      "grad_norm": 0.717419445514679,
      "learning_rate": 4.127725827835052e-06,
      "loss": 0.72,
      "step": 19753
    },
    {
      "epoch": 0.5691811214199274,
      "grad_norm": 0.6867669224739075,
      "learning_rate": 4.127266376892143e-06,
      "loss": 0.6892,
      "step": 19754
    },
    {
      "epoch": 0.5692099348815767,
      "grad_norm": 0.7087495923042297,
      "learning_rate": 4.1268069335498714e-06,
      "loss": 0.7012,
      "step": 19755
    },
    {
      "epoch": 0.569238748343226,
      "grad_norm": 0.7409068942070007,
      "learning_rate": 4.126347497812239e-06,
      "loss": 0.7072,
      "step": 19756
    },
    {
      "epoch": 0.5692675618048753,
      "grad_norm": 0.7351295948028564,
      "learning_rate": 4.1258880696832434e-06,
      "loss": 0.6959,
      "step": 19757
    },
    {
      "epoch": 0.5692963752665245,
      "grad_norm": 0.7799451351165771,
      "learning_rate": 4.125428649166893e-06,
      "loss": 0.722,
      "step": 19758
    },
    {
      "epoch": 0.5693251887281738,
      "grad_norm": 0.729622483253479,
      "learning_rate": 4.124969236267183e-06,
      "loss": 0.6994,
      "step": 19759
    },
    {
      "epoch": 0.5693540021898231,
      "grad_norm": 0.7032943367958069,
      "learning_rate": 4.124509830988117e-06,
      "loss": 0.7199,
      "step": 19760
    },
    {
      "epoch": 0.5693828156514724,
      "grad_norm": 0.7174384593963623,
      "learning_rate": 4.124050433333695e-06,
      "loss": 0.699,
      "step": 19761
    },
    {
      "epoch": 0.5694116291131216,
      "grad_norm": 0.7266953587532043,
      "learning_rate": 4.123591043307918e-06,
      "loss": 0.6963,
      "step": 19762
    },
    {
      "epoch": 0.5694404425747709,
      "grad_norm": 0.7274816036224365,
      "learning_rate": 4.1231316609147876e-06,
      "loss": 0.7267,
      "step": 19763
    },
    {
      "epoch": 0.5694692560364202,
      "grad_norm": 0.7364214658737183,
      "learning_rate": 4.122672286158304e-06,
      "loss": 0.7274,
      "step": 19764
    },
    {
      "epoch": 0.5694980694980695,
      "grad_norm": 0.7203687429428101,
      "learning_rate": 4.122212919042467e-06,
      "loss": 0.7184,
      "step": 19765
    },
    {
      "epoch": 0.5695268829597188,
      "grad_norm": 0.745512843132019,
      "learning_rate": 4.121753559571278e-06,
      "loss": 0.7053,
      "step": 19766
    },
    {
      "epoch": 0.569555696421368,
      "grad_norm": 0.7414875626564026,
      "learning_rate": 4.121294207748738e-06,
      "loss": 0.705,
      "step": 19767
    },
    {
      "epoch": 0.5695845098830173,
      "grad_norm": 0.7194366455078125,
      "learning_rate": 4.120834863578846e-06,
      "loss": 0.6977,
      "step": 19768
    },
    {
      "epoch": 0.5696133233446666,
      "grad_norm": 0.7190135717391968,
      "learning_rate": 4.120375527065603e-06,
      "loss": 0.6922,
      "step": 19769
    },
    {
      "epoch": 0.5696421368063159,
      "grad_norm": 0.7397460341453552,
      "learning_rate": 4.119916198213011e-06,
      "loss": 0.7196,
      "step": 19770
    },
    {
      "epoch": 0.5696709502679652,
      "grad_norm": 0.7407922148704529,
      "learning_rate": 4.11945687702507e-06,
      "loss": 0.7093,
      "step": 19771
    },
    {
      "epoch": 0.5696997637296145,
      "grad_norm": 0.7069786190986633,
      "learning_rate": 4.118997563505778e-06,
      "loss": 0.7218,
      "step": 19772
    },
    {
      "epoch": 0.5697285771912638,
      "grad_norm": 0.7243744134902954,
      "learning_rate": 4.118538257659136e-06,
      "loss": 0.7122,
      "step": 19773
    },
    {
      "epoch": 0.5697573906529131,
      "grad_norm": 0.7161064743995667,
      "learning_rate": 4.118078959489147e-06,
      "loss": 0.7392,
      "step": 19774
    },
    {
      "epoch": 0.5697862041145624,
      "grad_norm": 0.7027847170829773,
      "learning_rate": 4.117619668999807e-06,
      "loss": 0.6986,
      "step": 19775
    },
    {
      "epoch": 0.5698150175762116,
      "grad_norm": 0.733587920665741,
      "learning_rate": 4.1171603861951185e-06,
      "loss": 0.7107,
      "step": 19776
    },
    {
      "epoch": 0.5698438310378608,
      "grad_norm": 0.7024879455566406,
      "learning_rate": 4.11670111107908e-06,
      "loss": 0.6941,
      "step": 19777
    },
    {
      "epoch": 0.5698726444995101,
      "grad_norm": 0.6980885863304138,
      "learning_rate": 4.116241843655692e-06,
      "loss": 0.6865,
      "step": 19778
    },
    {
      "epoch": 0.5699014579611594,
      "grad_norm": 0.7465974688529968,
      "learning_rate": 4.115782583928954e-06,
      "loss": 0.7169,
      "step": 19779
    },
    {
      "epoch": 0.5699302714228087,
      "grad_norm": 0.7336255311965942,
      "learning_rate": 4.115323331902866e-06,
      "loss": 0.7088,
      "step": 19780
    },
    {
      "epoch": 0.569959084884458,
      "grad_norm": 0.7135959267616272,
      "learning_rate": 4.114864087581426e-06,
      "loss": 0.7096,
      "step": 19781
    },
    {
      "epoch": 0.5699878983461073,
      "grad_norm": 0.730607271194458,
      "learning_rate": 4.114404850968636e-06,
      "loss": 0.6954,
      "step": 19782
    },
    {
      "epoch": 0.5700167118077566,
      "grad_norm": 0.7812539935112,
      "learning_rate": 4.113945622068495e-06,
      "loss": 0.7173,
      "step": 19783
    },
    {
      "epoch": 0.5700455252694059,
      "grad_norm": 0.7106948494911194,
      "learning_rate": 4.1134864008850015e-06,
      "loss": 0.6878,
      "step": 19784
    },
    {
      "epoch": 0.5700743387310552,
      "grad_norm": 0.6935390830039978,
      "learning_rate": 4.113027187422156e-06,
      "loss": 0.6957,
      "step": 19785
    },
    {
      "epoch": 0.5701031521927045,
      "grad_norm": 0.7688036561012268,
      "learning_rate": 4.112567981683956e-06,
      "loss": 0.7016,
      "step": 19786
    },
    {
      "epoch": 0.5701319656543538,
      "grad_norm": 0.718509316444397,
      "learning_rate": 4.112108783674403e-06,
      "loss": 0.7132,
      "step": 19787
    },
    {
      "epoch": 0.570160779116003,
      "grad_norm": 0.7153339982032776,
      "learning_rate": 4.111649593397494e-06,
      "loss": 0.6903,
      "step": 19788
    },
    {
      "epoch": 0.5701895925776522,
      "grad_norm": 0.7007201313972473,
      "learning_rate": 4.11119041085723e-06,
      "loss": 0.6994,
      "step": 19789
    },
    {
      "epoch": 0.5702184060393015,
      "grad_norm": 0.7394458055496216,
      "learning_rate": 4.110731236057607e-06,
      "loss": 0.6851,
      "step": 19790
    },
    {
      "epoch": 0.5702472195009508,
      "grad_norm": 0.7085838317871094,
      "learning_rate": 4.110272069002627e-06,
      "loss": 0.6986,
      "step": 19791
    },
    {
      "epoch": 0.5702760329626001,
      "grad_norm": 0.7338452935218811,
      "learning_rate": 4.109812909696288e-06,
      "loss": 0.7147,
      "step": 19792
    },
    {
      "epoch": 0.5703048464242494,
      "grad_norm": 0.7722080945968628,
      "learning_rate": 4.109353758142588e-06,
      "loss": 0.7106,
      "step": 19793
    },
    {
      "epoch": 0.5703336598858987,
      "grad_norm": 0.7001037001609802,
      "learning_rate": 4.1088946143455275e-06,
      "loss": 0.686,
      "step": 19794
    },
    {
      "epoch": 0.570362473347548,
      "grad_norm": 0.7150998115539551,
      "learning_rate": 4.108435478309103e-06,
      "loss": 0.6984,
      "step": 19795
    },
    {
      "epoch": 0.5703912868091973,
      "grad_norm": 0.7543278336524963,
      "learning_rate": 4.1079763500373155e-06,
      "loss": 0.7134,
      "step": 19796
    },
    {
      "epoch": 0.5704201002708466,
      "grad_norm": 0.7457995414733887,
      "learning_rate": 4.107517229534162e-06,
      "loss": 0.7207,
      "step": 19797
    },
    {
      "epoch": 0.5704489137324958,
      "grad_norm": 0.8621892929077148,
      "learning_rate": 4.107058116803641e-06,
      "loss": 0.6956,
      "step": 19798
    },
    {
      "epoch": 0.5704777271941451,
      "grad_norm": 0.7132675051689148,
      "learning_rate": 4.106599011849752e-06,
      "loss": 0.7157,
      "step": 19799
    },
    {
      "epoch": 0.5705065406557944,
      "grad_norm": 0.718707263469696,
      "learning_rate": 4.1061399146764914e-06,
      "loss": 0.7116,
      "step": 19800
    },
    {
      "epoch": 0.5705353541174437,
      "grad_norm": 0.7437512278556824,
      "learning_rate": 4.10568082528786e-06,
      "loss": 0.6988,
      "step": 19801
    },
    {
      "epoch": 0.570564167579093,
      "grad_norm": 0.7602890133857727,
      "learning_rate": 4.105221743687853e-06,
      "loss": 0.697,
      "step": 19802
    },
    {
      "epoch": 0.5705929810407422,
      "grad_norm": 0.7109087705612183,
      "learning_rate": 4.104762669880471e-06,
      "loss": 0.698,
      "step": 19803
    },
    {
      "epoch": 0.5706217945023915,
      "grad_norm": 0.7484856843948364,
      "learning_rate": 4.104303603869712e-06,
      "loss": 0.7117,
      "step": 19804
    },
    {
      "epoch": 0.5706506079640408,
      "grad_norm": 0.7289592623710632,
      "learning_rate": 4.10384454565957e-06,
      "loss": 0.7084,
      "step": 19805
    },
    {
      "epoch": 0.5706794214256901,
      "grad_norm": 0.6951165199279785,
      "learning_rate": 4.103385495254049e-06,
      "loss": 0.7011,
      "step": 19806
    },
    {
      "epoch": 0.5707082348873393,
      "grad_norm": 0.6982230544090271,
      "learning_rate": 4.102926452657143e-06,
      "loss": 0.7177,
      "step": 19807
    },
    {
      "epoch": 0.5707370483489886,
      "grad_norm": 0.716666579246521,
      "learning_rate": 4.102467417872852e-06,
      "loss": 0.7174,
      "step": 19808
    },
    {
      "epoch": 0.5707658618106379,
      "grad_norm": 0.7129825949668884,
      "learning_rate": 4.1020083909051725e-06,
      "loss": 0.695,
      "step": 19809
    },
    {
      "epoch": 0.5707946752722872,
      "grad_norm": 0.733206570148468,
      "learning_rate": 4.101549371758101e-06,
      "loss": 0.6897,
      "step": 19810
    },
    {
      "epoch": 0.5708234887339365,
      "grad_norm": 0.7486475706100464,
      "learning_rate": 4.101090360435637e-06,
      "loss": 0.7173,
      "step": 19811
    },
    {
      "epoch": 0.5708523021955858,
      "grad_norm": 0.7328701615333557,
      "learning_rate": 4.100631356941777e-06,
      "loss": 0.6979,
      "step": 19812
    },
    {
      "epoch": 0.5708811156572351,
      "grad_norm": 0.7228407263755798,
      "learning_rate": 4.10017236128052e-06,
      "loss": 0.7041,
      "step": 19813
    },
    {
      "epoch": 0.5709099291188844,
      "grad_norm": 0.7348405718803406,
      "learning_rate": 4.099713373455861e-06,
      "loss": 0.7058,
      "step": 19814
    },
    {
      "epoch": 0.5709387425805337,
      "grad_norm": 0.7194302678108215,
      "learning_rate": 4.0992543934718e-06,
      "loss": 0.687,
      "step": 19815
    },
    {
      "epoch": 0.5709675560421829,
      "grad_norm": 0.7415726780891418,
      "learning_rate": 4.098795421332332e-06,
      "loss": 0.7094,
      "step": 19816
    },
    {
      "epoch": 0.5709963695038321,
      "grad_norm": 0.7080027461051941,
      "learning_rate": 4.098336457041453e-06,
      "loss": 0.6986,
      "step": 19817
    },
    {
      "epoch": 0.5710251829654814,
      "grad_norm": 0.7632450461387634,
      "learning_rate": 4.097877500603164e-06,
      "loss": 0.7397,
      "step": 19818
    },
    {
      "epoch": 0.5710539964271307,
      "grad_norm": 0.7572528719902039,
      "learning_rate": 4.097418552021461e-06,
      "loss": 0.7319,
      "step": 19819
    },
    {
      "epoch": 0.57108280988878,
      "grad_norm": 0.7379891276359558,
      "learning_rate": 4.096959611300339e-06,
      "loss": 0.7256,
      "step": 19820
    },
    {
      "epoch": 0.5711116233504293,
      "grad_norm": 0.7363461852073669,
      "learning_rate": 4.096500678443797e-06,
      "loss": 0.6914,
      "step": 19821
    },
    {
      "epoch": 0.5711404368120786,
      "grad_norm": 0.7111622095108032,
      "learning_rate": 4.09604175345583e-06,
      "loss": 0.7215,
      "step": 19822
    },
    {
      "epoch": 0.5711692502737279,
      "grad_norm": 0.7298962473869324,
      "learning_rate": 4.095582836340436e-06,
      "loss": 0.7052,
      "step": 19823
    },
    {
      "epoch": 0.5711980637353772,
      "grad_norm": 0.7507344484329224,
      "learning_rate": 4.095123927101613e-06,
      "loss": 0.7209,
      "step": 19824
    },
    {
      "epoch": 0.5712268771970265,
      "grad_norm": 0.7404847145080566,
      "learning_rate": 4.094665025743354e-06,
      "loss": 0.7124,
      "step": 19825
    },
    {
      "epoch": 0.5712556906586758,
      "grad_norm": 0.738768458366394,
      "learning_rate": 4.094206132269658e-06,
      "loss": 0.7092,
      "step": 19826
    },
    {
      "epoch": 0.571284504120325,
      "grad_norm": 0.7142630815505981,
      "learning_rate": 4.093747246684522e-06,
      "loss": 0.6933,
      "step": 19827
    },
    {
      "epoch": 0.5713133175819743,
      "grad_norm": 0.7206262946128845,
      "learning_rate": 4.09328836899194e-06,
      "loss": 0.6885,
      "step": 19828
    },
    {
      "epoch": 0.5713421310436236,
      "grad_norm": 0.7387617826461792,
      "learning_rate": 4.09282949919591e-06,
      "loss": 0.7042,
      "step": 19829
    },
    {
      "epoch": 0.5713709445052728,
      "grad_norm": 0.7260662913322449,
      "learning_rate": 4.092370637300429e-06,
      "loss": 0.7098,
      "step": 19830
    },
    {
      "epoch": 0.5713997579669221,
      "grad_norm": 0.7465123534202576,
      "learning_rate": 4.091911783309492e-06,
      "loss": 0.704,
      "step": 19831
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.7482101321220398,
      "learning_rate": 4.091452937227097e-06,
      "loss": 0.7128,
      "step": 19832
    },
    {
      "epoch": 0.5714573848902207,
      "grad_norm": 0.7239298820495605,
      "learning_rate": 4.090994099057237e-06,
      "loss": 0.7268,
      "step": 19833
    },
    {
      "epoch": 0.57148619835187,
      "grad_norm": 0.7369198799133301,
      "learning_rate": 4.09053526880391e-06,
      "loss": 0.7134,
      "step": 19834
    },
    {
      "epoch": 0.5715150118135193,
      "grad_norm": 0.7261044383049011,
      "learning_rate": 4.090076446471113e-06,
      "loss": 0.7084,
      "step": 19835
    },
    {
      "epoch": 0.5715438252751686,
      "grad_norm": 0.7171579003334045,
      "learning_rate": 4.089617632062839e-06,
      "loss": 0.7018,
      "step": 19836
    },
    {
      "epoch": 0.5715726387368179,
      "grad_norm": 0.7106176614761353,
      "learning_rate": 4.089158825583087e-06,
      "loss": 0.7204,
      "step": 19837
    },
    {
      "epoch": 0.5716014521984671,
      "grad_norm": 0.7230300903320312,
      "learning_rate": 4.08870002703585e-06,
      "loss": 0.7152,
      "step": 19838
    },
    {
      "epoch": 0.5716302656601164,
      "grad_norm": 0.754098653793335,
      "learning_rate": 4.088241236425126e-06,
      "loss": 0.6812,
      "step": 19839
    },
    {
      "epoch": 0.5716590791217657,
      "grad_norm": 0.726962685585022,
      "learning_rate": 4.087782453754909e-06,
      "loss": 0.7019,
      "step": 19840
    },
    {
      "epoch": 0.571687892583415,
      "grad_norm": 0.7302975654602051,
      "learning_rate": 4.087323679029191e-06,
      "loss": 0.7166,
      "step": 19841
    },
    {
      "epoch": 0.5717167060450643,
      "grad_norm": 0.7497353553771973,
      "learning_rate": 4.086864912251976e-06,
      "loss": 0.6778,
      "step": 19842
    },
    {
      "epoch": 0.5717455195067135,
      "grad_norm": 0.7451965808868408,
      "learning_rate": 4.086406153427253e-06,
      "loss": 0.7012,
      "step": 19843
    },
    {
      "epoch": 0.5717743329683628,
      "grad_norm": 0.7271080017089844,
      "learning_rate": 4.0859474025590205e-06,
      "loss": 0.7205,
      "step": 19844
    },
    {
      "epoch": 0.5718031464300121,
      "grad_norm": 0.717031717300415,
      "learning_rate": 4.0854886596512715e-06,
      "loss": 0.7112,
      "step": 19845
    },
    {
      "epoch": 0.5718319598916614,
      "grad_norm": 0.6882858872413635,
      "learning_rate": 4.085029924708003e-06,
      "loss": 0.7009,
      "step": 19846
    },
    {
      "epoch": 0.5718607733533106,
      "grad_norm": 0.7426251173019409,
      "learning_rate": 4.084571197733208e-06,
      "loss": 0.7024,
      "step": 19847
    },
    {
      "epoch": 0.5718895868149599,
      "grad_norm": 0.7341274619102478,
      "learning_rate": 4.084112478730884e-06,
      "loss": 0.7092,
      "step": 19848
    },
    {
      "epoch": 0.5719184002766092,
      "grad_norm": 0.7190229892730713,
      "learning_rate": 4.083653767705023e-06,
      "loss": 0.7208,
      "step": 19849
    },
    {
      "epoch": 0.5719472137382585,
      "grad_norm": 0.7165867686271667,
      "learning_rate": 4.083195064659623e-06,
      "loss": 0.7203,
      "step": 19850
    },
    {
      "epoch": 0.5719760271999078,
      "grad_norm": 0.7260928750038147,
      "learning_rate": 4.082736369598677e-06,
      "loss": 0.7402,
      "step": 19851
    },
    {
      "epoch": 0.5720048406615571,
      "grad_norm": 0.7299246191978455,
      "learning_rate": 4.08227768252618e-06,
      "loss": 0.7147,
      "step": 19852
    },
    {
      "epoch": 0.5720336541232064,
      "grad_norm": 0.7051205039024353,
      "learning_rate": 4.081819003446127e-06,
      "loss": 0.6803,
      "step": 19853
    },
    {
      "epoch": 0.5720624675848557,
      "grad_norm": 0.7162731289863586,
      "learning_rate": 4.081360332362512e-06,
      "loss": 0.7123,
      "step": 19854
    },
    {
      "epoch": 0.572091281046505,
      "grad_norm": 0.779942512512207,
      "learning_rate": 4.080901669279332e-06,
      "loss": 0.7083,
      "step": 19855
    },
    {
      "epoch": 0.5721200945081543,
      "grad_norm": 0.7244445085525513,
      "learning_rate": 4.080443014200579e-06,
      "loss": 0.6992,
      "step": 19856
    },
    {
      "epoch": 0.5721489079698034,
      "grad_norm": 0.7367802858352661,
      "learning_rate": 4.079984367130249e-06,
      "loss": 0.7048,
      "step": 19857
    },
    {
      "epoch": 0.5721777214314527,
      "grad_norm": 0.7348257899284363,
      "learning_rate": 4.079525728072335e-06,
      "loss": 0.7058,
      "step": 19858
    },
    {
      "epoch": 0.572206534893102,
      "grad_norm": 0.749697208404541,
      "learning_rate": 4.079067097030831e-06,
      "loss": 0.6899,
      "step": 19859
    },
    {
      "epoch": 0.5722353483547513,
      "grad_norm": 0.7390693426132202,
      "learning_rate": 4.078608474009732e-06,
      "loss": 0.7014,
      "step": 19860
    },
    {
      "epoch": 0.5722641618164006,
      "grad_norm": 0.729519784450531,
      "learning_rate": 4.078149859013033e-06,
      "loss": 0.7269,
      "step": 19861
    },
    {
      "epoch": 0.5722929752780499,
      "grad_norm": 0.7316480875015259,
      "learning_rate": 4.077691252044727e-06,
      "loss": 0.6934,
      "step": 19862
    },
    {
      "epoch": 0.5723217887396992,
      "grad_norm": 0.7192084193229675,
      "learning_rate": 4.077232653108808e-06,
      "loss": 0.679,
      "step": 19863
    },
    {
      "epoch": 0.5723506022013485,
      "grad_norm": 0.754294216632843,
      "learning_rate": 4.07677406220927e-06,
      "loss": 0.758,
      "step": 19864
    },
    {
      "epoch": 0.5723794156629978,
      "grad_norm": 0.7368844151496887,
      "learning_rate": 4.076315479350106e-06,
      "loss": 0.7166,
      "step": 19865
    },
    {
      "epoch": 0.5724082291246471,
      "grad_norm": 0.7437299489974976,
      "learning_rate": 4.075856904535311e-06,
      "loss": 0.6963,
      "step": 19866
    },
    {
      "epoch": 0.5724370425862964,
      "grad_norm": 0.6965279579162598,
      "learning_rate": 4.075398337768881e-06,
      "loss": 0.7179,
      "step": 19867
    },
    {
      "epoch": 0.5724658560479456,
      "grad_norm": 0.6837489008903503,
      "learning_rate": 4.074939779054805e-06,
      "loss": 0.6984,
      "step": 19868
    },
    {
      "epoch": 0.5724946695095949,
      "grad_norm": 0.7285782694816589,
      "learning_rate": 4.07448122839708e-06,
      "loss": 0.7038,
      "step": 19869
    },
    {
      "epoch": 0.5725234829712441,
      "grad_norm": 0.7475273609161377,
      "learning_rate": 4.074022685799698e-06,
      "loss": 0.7114,
      "step": 19870
    },
    {
      "epoch": 0.5725522964328934,
      "grad_norm": 0.7423056960105896,
      "learning_rate": 4.073564151266652e-06,
      "loss": 0.7188,
      "step": 19871
    },
    {
      "epoch": 0.5725811098945427,
      "grad_norm": 0.7542703747749329,
      "learning_rate": 4.073105624801937e-06,
      "loss": 0.7212,
      "step": 19872
    },
    {
      "epoch": 0.572609923356192,
      "grad_norm": 0.7560063004493713,
      "learning_rate": 4.072647106409545e-06,
      "loss": 0.6844,
      "step": 19873
    },
    {
      "epoch": 0.5726387368178413,
      "grad_norm": 0.7675074934959412,
      "learning_rate": 4.07218859609347e-06,
      "loss": 0.7094,
      "step": 19874
    },
    {
      "epoch": 0.5726675502794906,
      "grad_norm": 0.7502689957618713,
      "learning_rate": 4.071730093857706e-06,
      "loss": 0.73,
      "step": 19875
    },
    {
      "epoch": 0.5726963637411399,
      "grad_norm": 0.726288914680481,
      "learning_rate": 4.071271599706244e-06,
      "loss": 0.7267,
      "step": 19876
    },
    {
      "epoch": 0.5727251772027891,
      "grad_norm": 0.7186914086341858,
      "learning_rate": 4.070813113643075e-06,
      "loss": 0.6772,
      "step": 19877
    },
    {
      "epoch": 0.5727539906644384,
      "grad_norm": 0.7143376469612122,
      "learning_rate": 4.070354635672198e-06,
      "loss": 0.6908,
      "step": 19878
    },
    {
      "epoch": 0.5727828041260877,
      "grad_norm": 0.7189128398895264,
      "learning_rate": 4.069896165797603e-06,
      "loss": 0.7024,
      "step": 19879
    },
    {
      "epoch": 0.572811617587737,
      "grad_norm": 0.6867086887359619,
      "learning_rate": 4.0694377040232815e-06,
      "loss": 0.6739,
      "step": 19880
    },
    {
      "epoch": 0.5728404310493863,
      "grad_norm": 0.7278002500534058,
      "learning_rate": 4.068979250353228e-06,
      "loss": 0.7155,
      "step": 19881
    },
    {
      "epoch": 0.5728692445110356,
      "grad_norm": 0.7297666668891907,
      "learning_rate": 4.068520804791435e-06,
      "loss": 0.6852,
      "step": 19882
    },
    {
      "epoch": 0.5728980579726849,
      "grad_norm": 0.7259580492973328,
      "learning_rate": 4.068062367341894e-06,
      "loss": 0.7211,
      "step": 19883
    },
    {
      "epoch": 0.5729268714343341,
      "grad_norm": 0.7112907767295837,
      "learning_rate": 4.067603938008599e-06,
      "loss": 0.7027,
      "step": 19884
    },
    {
      "epoch": 0.5729556848959834,
      "grad_norm": 0.7525763511657715,
      "learning_rate": 4.067145516795542e-06,
      "loss": 0.7372,
      "step": 19885
    },
    {
      "epoch": 0.5729844983576327,
      "grad_norm": 0.7134621143341064,
      "learning_rate": 4.066687103706713e-06,
      "loss": 0.6941,
      "step": 19886
    },
    {
      "epoch": 0.573013311819282,
      "grad_norm": 0.738407552242279,
      "learning_rate": 4.066228698746107e-06,
      "loss": 0.7117,
      "step": 19887
    },
    {
      "epoch": 0.5730421252809312,
      "grad_norm": 0.7418609261512756,
      "learning_rate": 4.065770301917716e-06,
      "loss": 0.7267,
      "step": 19888
    },
    {
      "epoch": 0.5730709387425805,
      "grad_norm": 0.7052547335624695,
      "learning_rate": 4.065311913225531e-06,
      "loss": 0.7017,
      "step": 19889
    },
    {
      "epoch": 0.5730997522042298,
      "grad_norm": 0.7266028523445129,
      "learning_rate": 4.0648535326735464e-06,
      "loss": 0.727,
      "step": 19890
    },
    {
      "epoch": 0.5731285656658791,
      "grad_norm": 0.7177461981773376,
      "learning_rate": 4.064395160265751e-06,
      "loss": 0.6953,
      "step": 19891
    },
    {
      "epoch": 0.5731573791275284,
      "grad_norm": 0.7103464603424072,
      "learning_rate": 4.0639367960061395e-06,
      "loss": 0.6963,
      "step": 19892
    },
    {
      "epoch": 0.5731861925891777,
      "grad_norm": 0.7050599455833435,
      "learning_rate": 4.063478439898702e-06,
      "loss": 0.6875,
      "step": 19893
    },
    {
      "epoch": 0.573215006050827,
      "grad_norm": 0.7083255052566528,
      "learning_rate": 4.063020091947433e-06,
      "loss": 0.6842,
      "step": 19894
    },
    {
      "epoch": 0.5732438195124763,
      "grad_norm": 0.725174605846405,
      "learning_rate": 4.062561752156321e-06,
      "loss": 0.6961,
      "step": 19895
    },
    {
      "epoch": 0.5732726329741256,
      "grad_norm": 0.7378025650978088,
      "learning_rate": 4.06210342052936e-06,
      "loss": 0.7324,
      "step": 19896
    },
    {
      "epoch": 0.5733014464357747,
      "grad_norm": 0.7178457975387573,
      "learning_rate": 4.06164509707054e-06,
      "loss": 0.693,
      "step": 19897
    },
    {
      "epoch": 0.573330259897424,
      "grad_norm": 0.7317941784858704,
      "learning_rate": 4.061186781783853e-06,
      "loss": 0.7349,
      "step": 19898
    },
    {
      "epoch": 0.5733590733590733,
      "grad_norm": 0.7132099270820618,
      "learning_rate": 4.060728474673291e-06,
      "loss": 0.7122,
      "step": 19899
    },
    {
      "epoch": 0.5733878868207226,
      "grad_norm": 0.7452116012573242,
      "learning_rate": 4.060270175742845e-06,
      "loss": 0.7079,
      "step": 19900
    },
    {
      "epoch": 0.5734167002823719,
      "grad_norm": 0.7229160666465759,
      "learning_rate": 4.059811884996505e-06,
      "loss": 0.7156,
      "step": 19901
    },
    {
      "epoch": 0.5734455137440212,
      "grad_norm": 0.7357956171035767,
      "learning_rate": 4.059353602438265e-06,
      "loss": 0.7198,
      "step": 19902
    },
    {
      "epoch": 0.5734743272056705,
      "grad_norm": 0.7179601192474365,
      "learning_rate": 4.058895328072114e-06,
      "loss": 0.7137,
      "step": 19903
    },
    {
      "epoch": 0.5735031406673198,
      "grad_norm": 0.7263159155845642,
      "learning_rate": 4.058437061902045e-06,
      "loss": 0.7274,
      "step": 19904
    },
    {
      "epoch": 0.5735319541289691,
      "grad_norm": 0.7031205892562866,
      "learning_rate": 4.057978803932048e-06,
      "loss": 0.6942,
      "step": 19905
    },
    {
      "epoch": 0.5735607675906184,
      "grad_norm": 0.7128348350524902,
      "learning_rate": 4.057520554166114e-06,
      "loss": 0.6934,
      "step": 19906
    },
    {
      "epoch": 0.5735895810522676,
      "grad_norm": 0.7432795166969299,
      "learning_rate": 4.057062312608233e-06,
      "loss": 0.7279,
      "step": 19907
    },
    {
      "epoch": 0.5736183945139169,
      "grad_norm": 0.7466549873352051,
      "learning_rate": 4.056604079262396e-06,
      "loss": 0.713,
      "step": 19908
    },
    {
      "epoch": 0.5736472079755662,
      "grad_norm": 0.727174699306488,
      "learning_rate": 4.056145854132596e-06,
      "loss": 0.7106,
      "step": 19909
    },
    {
      "epoch": 0.5736760214372155,
      "grad_norm": 0.7246691584587097,
      "learning_rate": 4.055687637222822e-06,
      "loss": 0.6885,
      "step": 19910
    },
    {
      "epoch": 0.5737048348988647,
      "grad_norm": 0.7429329752922058,
      "learning_rate": 4.055229428537064e-06,
      "loss": 0.7253,
      "step": 19911
    },
    {
      "epoch": 0.573733648360514,
      "grad_norm": 0.7247593402862549,
      "learning_rate": 4.0547712280793136e-06,
      "loss": 0.7011,
      "step": 19912
    },
    {
      "epoch": 0.5737624618221633,
      "grad_norm": 0.7204939723014832,
      "learning_rate": 4.05431303585356e-06,
      "loss": 0.6909,
      "step": 19913
    },
    {
      "epoch": 0.5737912752838126,
      "grad_norm": 0.7157400250434875,
      "learning_rate": 4.053854851863795e-06,
      "loss": 0.6943,
      "step": 19914
    },
    {
      "epoch": 0.5738200887454619,
      "grad_norm": 0.7163251042366028,
      "learning_rate": 4.053396676114008e-06,
      "loss": 0.7042,
      "step": 19915
    },
    {
      "epoch": 0.5738489022071112,
      "grad_norm": 0.7036064863204956,
      "learning_rate": 4.052938508608192e-06,
      "loss": 0.6935,
      "step": 19916
    },
    {
      "epoch": 0.5738777156687604,
      "grad_norm": 0.7234310507774353,
      "learning_rate": 4.052480349350333e-06,
      "loss": 0.6975,
      "step": 19917
    },
    {
      "epoch": 0.5739065291304097,
      "grad_norm": 0.7471958994865417,
      "learning_rate": 4.052022198344424e-06,
      "loss": 0.7185,
      "step": 19918
    },
    {
      "epoch": 0.573935342592059,
      "grad_norm": 0.7188088297843933,
      "learning_rate": 4.051564055594454e-06,
      "loss": 0.7013,
      "step": 19919
    },
    {
      "epoch": 0.5739641560537083,
      "grad_norm": 0.7050384283065796,
      "learning_rate": 4.051105921104412e-06,
      "loss": 0.7049,
      "step": 19920
    },
    {
      "epoch": 0.5739929695153576,
      "grad_norm": 0.7121617794036865,
      "learning_rate": 4.0506477948782895e-06,
      "loss": 0.7161,
      "step": 19921
    },
    {
      "epoch": 0.5740217829770069,
      "grad_norm": 0.6934986114501953,
      "learning_rate": 4.050189676920075e-06,
      "loss": 0.7088,
      "step": 19922
    },
    {
      "epoch": 0.5740505964386562,
      "grad_norm": 0.7375710606575012,
      "learning_rate": 4.04973156723376e-06,
      "loss": 0.702,
      "step": 19923
    },
    {
      "epoch": 0.5740794099003054,
      "grad_norm": 0.7113412022590637,
      "learning_rate": 4.049273465823333e-06,
      "loss": 0.7218,
      "step": 19924
    },
    {
      "epoch": 0.5741082233619547,
      "grad_norm": 0.7016412615776062,
      "learning_rate": 4.048815372692781e-06,
      "loss": 0.7077,
      "step": 19925
    },
    {
      "epoch": 0.574137036823604,
      "grad_norm": 0.7043140530586243,
      "learning_rate": 4.0483572878460985e-06,
      "loss": 0.6858,
      "step": 19926
    },
    {
      "epoch": 0.5741658502852532,
      "grad_norm": 0.7187795639038086,
      "learning_rate": 4.047899211287273e-06,
      "loss": 0.6941,
      "step": 19927
    },
    {
      "epoch": 0.5741946637469025,
      "grad_norm": 0.713779628276825,
      "learning_rate": 4.0474411430202935e-06,
      "loss": 0.7143,
      "step": 19928
    },
    {
      "epoch": 0.5742234772085518,
      "grad_norm": 0.7271739840507507,
      "learning_rate": 4.04698308304915e-06,
      "loss": 0.7156,
      "step": 19929
    },
    {
      "epoch": 0.5742522906702011,
      "grad_norm": 0.7102159857749939,
      "learning_rate": 4.046525031377829e-06,
      "loss": 0.7161,
      "step": 19930
    },
    {
      "epoch": 0.5742811041318504,
      "grad_norm": 0.7245779633522034,
      "learning_rate": 4.046066988010324e-06,
      "loss": 0.694,
      "step": 19931
    },
    {
      "epoch": 0.5743099175934997,
      "grad_norm": 0.7436326146125793,
      "learning_rate": 4.045608952950622e-06,
      "loss": 0.7086,
      "step": 19932
    },
    {
      "epoch": 0.574338731055149,
      "grad_norm": 0.7401350736618042,
      "learning_rate": 4.045150926202711e-06,
      "loss": 0.7079,
      "step": 19933
    },
    {
      "epoch": 0.5743675445167983,
      "grad_norm": 0.7159131765365601,
      "learning_rate": 4.044692907770581e-06,
      "loss": 0.7044,
      "step": 19934
    },
    {
      "epoch": 0.5743963579784476,
      "grad_norm": 0.7156891822814941,
      "learning_rate": 4.044234897658222e-06,
      "loss": 0.7042,
      "step": 19935
    },
    {
      "epoch": 0.5744251714400969,
      "grad_norm": 0.7001582980155945,
      "learning_rate": 4.043776895869621e-06,
      "loss": 0.6871,
      "step": 19936
    },
    {
      "epoch": 0.5744539849017462,
      "grad_norm": 0.702747642993927,
      "learning_rate": 4.043318902408765e-06,
      "loss": 0.7159,
      "step": 19937
    },
    {
      "epoch": 0.5744827983633953,
      "grad_norm": 0.7050946354866028,
      "learning_rate": 4.0428609172796465e-06,
      "loss": 0.7074,
      "step": 19938
    },
    {
      "epoch": 0.5745116118250446,
      "grad_norm": 0.7089359164237976,
      "learning_rate": 4.042402940486253e-06,
      "loss": 0.708,
      "step": 19939
    },
    {
      "epoch": 0.5745404252866939,
      "grad_norm": 0.7194564938545227,
      "learning_rate": 4.041944972032573e-06,
      "loss": 0.7223,
      "step": 19940
    },
    {
      "epoch": 0.5745692387483432,
      "grad_norm": 0.7133142948150635,
      "learning_rate": 4.041487011922595e-06,
      "loss": 0.7104,
      "step": 19941
    },
    {
      "epoch": 0.5745980522099925,
      "grad_norm": 0.7066864371299744,
      "learning_rate": 4.041029060160306e-06,
      "loss": 0.7152,
      "step": 19942
    },
    {
      "epoch": 0.5746268656716418,
      "grad_norm": 0.7174503207206726,
      "learning_rate": 4.040571116749697e-06,
      "loss": 0.697,
      "step": 19943
    },
    {
      "epoch": 0.5746556791332911,
      "grad_norm": 0.7290204167366028,
      "learning_rate": 4.0401131816947524e-06,
      "loss": 0.7066,
      "step": 19944
    },
    {
      "epoch": 0.5746844925949404,
      "grad_norm": 0.7167713642120361,
      "learning_rate": 4.0396552549994635e-06,
      "loss": 0.7011,
      "step": 19945
    },
    {
      "epoch": 0.5747133060565897,
      "grad_norm": 0.7445876002311707,
      "learning_rate": 4.039197336667816e-06,
      "loss": 0.713,
      "step": 19946
    },
    {
      "epoch": 0.574742119518239,
      "grad_norm": 0.7473498582839966,
      "learning_rate": 4.0387394267038e-06,
      "loss": 0.7051,
      "step": 19947
    },
    {
      "epoch": 0.5747709329798882,
      "grad_norm": 0.7220942974090576,
      "learning_rate": 4.038281525111404e-06,
      "loss": 0.7097,
      "step": 19948
    },
    {
      "epoch": 0.5747997464415375,
      "grad_norm": 0.7297239303588867,
      "learning_rate": 4.037823631894612e-06,
      "loss": 0.7089,
      "step": 19949
    },
    {
      "epoch": 0.5748285599031868,
      "grad_norm": 0.7492796182632446,
      "learning_rate": 4.037365747057416e-06,
      "loss": 0.6996,
      "step": 19950
    },
    {
      "epoch": 0.574857373364836,
      "grad_norm": 0.7685693502426147,
      "learning_rate": 4.036907870603801e-06,
      "loss": 0.7269,
      "step": 19951
    },
    {
      "epoch": 0.5748861868264853,
      "grad_norm": 0.7567213773727417,
      "learning_rate": 4.036450002537757e-06,
      "loss": 0.7122,
      "step": 19952
    },
    {
      "epoch": 0.5749150002881346,
      "grad_norm": 0.6890741586685181,
      "learning_rate": 4.03599214286327e-06,
      "loss": 0.7118,
      "step": 19953
    },
    {
      "epoch": 0.5749438137497839,
      "grad_norm": 0.7273120284080505,
      "learning_rate": 4.035534291584328e-06,
      "loss": 0.7005,
      "step": 19954
    },
    {
      "epoch": 0.5749726272114332,
      "grad_norm": 0.742914080619812,
      "learning_rate": 4.0350764487049184e-06,
      "loss": 0.6985,
      "step": 19955
    },
    {
      "epoch": 0.5750014406730825,
      "grad_norm": 0.7048923969268799,
      "learning_rate": 4.034618614229028e-06,
      "loss": 0.6959,
      "step": 19956
    },
    {
      "epoch": 0.5750302541347317,
      "grad_norm": 0.720090925693512,
      "learning_rate": 4.034160788160644e-06,
      "loss": 0.7101,
      "step": 19957
    },
    {
      "epoch": 0.575059067596381,
      "grad_norm": 0.7391262054443359,
      "learning_rate": 4.033702970503755e-06,
      "loss": 0.7292,
      "step": 19958
    },
    {
      "epoch": 0.5750878810580303,
      "grad_norm": 0.7308517694473267,
      "learning_rate": 4.033245161262346e-06,
      "loss": 0.7056,
      "step": 19959
    },
    {
      "epoch": 0.5751166945196796,
      "grad_norm": 0.801243782043457,
      "learning_rate": 4.032787360440407e-06,
      "loss": 0.6729,
      "step": 19960
    },
    {
      "epoch": 0.5751455079813289,
      "grad_norm": 0.6965730786323547,
      "learning_rate": 4.03232956804192e-06,
      "loss": 0.6865,
      "step": 19961
    },
    {
      "epoch": 0.5751743214429782,
      "grad_norm": 0.7356316447257996,
      "learning_rate": 4.031871784070877e-06,
      "loss": 0.7016,
      "step": 19962
    },
    {
      "epoch": 0.5752031349046275,
      "grad_norm": 0.7615039348602295,
      "learning_rate": 4.031414008531263e-06,
      "loss": 0.711,
      "step": 19963
    },
    {
      "epoch": 0.5752319483662768,
      "grad_norm": 0.6966525316238403,
      "learning_rate": 4.0309562414270655e-06,
      "loss": 0.6902,
      "step": 19964
    },
    {
      "epoch": 0.575260761827926,
      "grad_norm": 0.705275297164917,
      "learning_rate": 4.0304984827622695e-06,
      "loss": 0.6847,
      "step": 19965
    },
    {
      "epoch": 0.5752895752895753,
      "grad_norm": 0.7142578363418579,
      "learning_rate": 4.030040732540863e-06,
      "loss": 0.6973,
      "step": 19966
    },
    {
      "epoch": 0.5753183887512245,
      "grad_norm": 0.6923538446426392,
      "learning_rate": 4.029582990766832e-06,
      "loss": 0.7018,
      "step": 19967
    },
    {
      "epoch": 0.5753472022128738,
      "grad_norm": 0.7332477569580078,
      "learning_rate": 4.029125257444163e-06,
      "loss": 0.7172,
      "step": 19968
    },
    {
      "epoch": 0.5753760156745231,
      "grad_norm": 0.7339456677436829,
      "learning_rate": 4.028667532576842e-06,
      "loss": 0.7077,
      "step": 19969
    },
    {
      "epoch": 0.5754048291361724,
      "grad_norm": 0.7124704718589783,
      "learning_rate": 4.028209816168857e-06,
      "loss": 0.7205,
      "step": 19970
    },
    {
      "epoch": 0.5754336425978217,
      "grad_norm": 0.720319926738739,
      "learning_rate": 4.027752108224192e-06,
      "loss": 0.7128,
      "step": 19971
    },
    {
      "epoch": 0.575462456059471,
      "grad_norm": 0.7108628153800964,
      "learning_rate": 4.027294408746835e-06,
      "loss": 0.7009,
      "step": 19972
    },
    {
      "epoch": 0.5754912695211203,
      "grad_norm": 0.7094858288764954,
      "learning_rate": 4.0268367177407695e-06,
      "loss": 0.7241,
      "step": 19973
    },
    {
      "epoch": 0.5755200829827696,
      "grad_norm": 0.6984139680862427,
      "learning_rate": 4.026379035209985e-06,
      "loss": 0.7051,
      "step": 19974
    },
    {
      "epoch": 0.5755488964444189,
      "grad_norm": 0.7368590831756592,
      "learning_rate": 4.025921361158466e-06,
      "loss": 0.7132,
      "step": 19975
    },
    {
      "epoch": 0.5755777099060682,
      "grad_norm": 0.7555180191993713,
      "learning_rate": 4.025463695590198e-06,
      "loss": 0.6876,
      "step": 19976
    },
    {
      "epoch": 0.5756065233677174,
      "grad_norm": 0.724200427532196,
      "learning_rate": 4.025006038509167e-06,
      "loss": 0.7032,
      "step": 19977
    },
    {
      "epoch": 0.5756353368293666,
      "grad_norm": 0.7234055995941162,
      "learning_rate": 4.02454838991936e-06,
      "loss": 0.723,
      "step": 19978
    },
    {
      "epoch": 0.5756641502910159,
      "grad_norm": 0.7324970960617065,
      "learning_rate": 4.02409074982476e-06,
      "loss": 0.7245,
      "step": 19979
    },
    {
      "epoch": 0.5756929637526652,
      "grad_norm": 0.7119717597961426,
      "learning_rate": 4.0236331182293545e-06,
      "loss": 0.6879,
      "step": 19980
    },
    {
      "epoch": 0.5757217772143145,
      "grad_norm": 0.7440829873085022,
      "learning_rate": 4.023175495137128e-06,
      "loss": 0.6942,
      "step": 19981
    },
    {
      "epoch": 0.5757505906759638,
      "grad_norm": 0.7272929549217224,
      "learning_rate": 4.022717880552067e-06,
      "loss": 0.6974,
      "step": 19982
    },
    {
      "epoch": 0.5757794041376131,
      "grad_norm": 0.7465358972549438,
      "learning_rate": 4.022260274478156e-06,
      "loss": 0.7084,
      "step": 19983
    },
    {
      "epoch": 0.5758082175992624,
      "grad_norm": 0.7394012212753296,
      "learning_rate": 4.021802676919379e-06,
      "loss": 0.7099,
      "step": 19984
    },
    {
      "epoch": 0.5758370310609117,
      "grad_norm": 0.7042896151542664,
      "learning_rate": 4.021345087879725e-06,
      "loss": 0.6799,
      "step": 19985
    },
    {
      "epoch": 0.575865844522561,
      "grad_norm": 0.7196751236915588,
      "learning_rate": 4.020887507363177e-06,
      "loss": 0.7059,
      "step": 19986
    },
    {
      "epoch": 0.5758946579842102,
      "grad_norm": 0.7390108704566956,
      "learning_rate": 4.020429935373719e-06,
      "loss": 0.7041,
      "step": 19987
    },
    {
      "epoch": 0.5759234714458595,
      "grad_norm": 0.9136196970939636,
      "learning_rate": 4.019972371915338e-06,
      "loss": 0.7094,
      "step": 19988
    },
    {
      "epoch": 0.5759522849075088,
      "grad_norm": 0.7433750629425049,
      "learning_rate": 4.019514816992017e-06,
      "loss": 0.728,
      "step": 19989
    },
    {
      "epoch": 0.5759810983691581,
      "grad_norm": 0.7252927422523499,
      "learning_rate": 4.019057270607742e-06,
      "loss": 0.7178,
      "step": 19990
    },
    {
      "epoch": 0.5760099118308074,
      "grad_norm": 0.7585101127624512,
      "learning_rate": 4.018599732766498e-06,
      "loss": 0.723,
      "step": 19991
    },
    {
      "epoch": 0.5760387252924566,
      "grad_norm": 0.7322094440460205,
      "learning_rate": 4.0181422034722704e-06,
      "loss": 0.7111,
      "step": 19992
    },
    {
      "epoch": 0.5760675387541059,
      "grad_norm": 0.713672935962677,
      "learning_rate": 4.017684682729042e-06,
      "loss": 0.7067,
      "step": 19993
    },
    {
      "epoch": 0.5760963522157552,
      "grad_norm": 0.7180473208427429,
      "learning_rate": 4.017227170540797e-06,
      "loss": 0.6998,
      "step": 19994
    },
    {
      "epoch": 0.5761251656774045,
      "grad_norm": 0.7312642931938171,
      "learning_rate": 4.016769666911522e-06,
      "loss": 0.7295,
      "step": 19995
    },
    {
      "epoch": 0.5761539791390538,
      "grad_norm": 0.7326171398162842,
      "learning_rate": 4.0163121718451985e-06,
      "loss": 0.7032,
      "step": 19996
    },
    {
      "epoch": 0.576182792600703,
      "grad_norm": 0.7004364132881165,
      "learning_rate": 4.015854685345814e-06,
      "loss": 0.6958,
      "step": 19997
    },
    {
      "epoch": 0.5762116060623523,
      "grad_norm": 0.7221063375473022,
      "learning_rate": 4.015397207417351e-06,
      "loss": 0.6861,
      "step": 19998
    },
    {
      "epoch": 0.5762404195240016,
      "grad_norm": 0.7147477865219116,
      "learning_rate": 4.014939738063795e-06,
      "loss": 0.7199,
      "step": 19999
    },
    {
      "epoch": 0.5762692329856509,
      "grad_norm": 0.7391548156738281,
      "learning_rate": 4.014482277289129e-06,
      "loss": 0.7204,
      "step": 20000
    },
    {
      "epoch": 0.5762980464473002,
      "grad_norm": 0.7230455875396729,
      "learning_rate": 4.014024825097337e-06,
      "loss": 0.6988,
      "step": 20001
    },
    {
      "epoch": 0.5763268599089495,
      "grad_norm": 0.7222679853439331,
      "learning_rate": 4.013567381492403e-06,
      "loss": 0.7098,
      "step": 20002
    },
    {
      "epoch": 0.5763556733705988,
      "grad_norm": 0.7267243266105652,
      "learning_rate": 4.013109946478312e-06,
      "loss": 0.7035,
      "step": 20003
    },
    {
      "epoch": 0.5763844868322481,
      "grad_norm": 0.7241532206535339,
      "learning_rate": 4.012652520059046e-06,
      "loss": 0.7241,
      "step": 20004
    },
    {
      "epoch": 0.5764133002938973,
      "grad_norm": 0.7301020622253418,
      "learning_rate": 4.01219510223859e-06,
      "loss": 0.6868,
      "step": 20005
    },
    {
      "epoch": 0.5764421137555465,
      "grad_norm": 0.7156946063041687,
      "learning_rate": 4.011737693020926e-06,
      "loss": 0.7009,
      "step": 20006
    },
    {
      "epoch": 0.5764709272171958,
      "grad_norm": 0.736758828163147,
      "learning_rate": 4.0112802924100394e-06,
      "loss": 0.7069,
      "step": 20007
    },
    {
      "epoch": 0.5764997406788451,
      "grad_norm": 0.7089994549751282,
      "learning_rate": 4.0108229004099125e-06,
      "loss": 0.7111,
      "step": 20008
    },
    {
      "epoch": 0.5765285541404944,
      "grad_norm": 0.7338113188743591,
      "learning_rate": 4.01036551702453e-06,
      "loss": 0.7048,
      "step": 20009
    },
    {
      "epoch": 0.5765573676021437,
      "grad_norm": 0.7318087220191956,
      "learning_rate": 4.009908142257875e-06,
      "loss": 0.7166,
      "step": 20010
    },
    {
      "epoch": 0.576586181063793,
      "grad_norm": 0.7098971009254456,
      "learning_rate": 4.00945077611393e-06,
      "loss": 0.6936,
      "step": 20011
    },
    {
      "epoch": 0.5766149945254423,
      "grad_norm": 0.7138672471046448,
      "learning_rate": 4.0089934185966785e-06,
      "loss": 0.7092,
      "step": 20012
    },
    {
      "epoch": 0.5766438079870916,
      "grad_norm": 0.7432908415794373,
      "learning_rate": 4.008536069710103e-06,
      "loss": 0.7292,
      "step": 20013
    },
    {
      "epoch": 0.5766726214487409,
      "grad_norm": 0.7193833589553833,
      "learning_rate": 4.008078729458189e-06,
      "loss": 0.7144,
      "step": 20014
    },
    {
      "epoch": 0.5767014349103902,
      "grad_norm": 0.7396256327629089,
      "learning_rate": 4.007621397844916e-06,
      "loss": 0.6932,
      "step": 20015
    },
    {
      "epoch": 0.5767302483720395,
      "grad_norm": 0.7384440302848816,
      "learning_rate": 4.00716407487427e-06,
      "loss": 0.7035,
      "step": 20016
    },
    {
      "epoch": 0.5767590618336887,
      "grad_norm": 0.7245886325836182,
      "learning_rate": 4.006706760550232e-06,
      "loss": 0.7105,
      "step": 20017
    },
    {
      "epoch": 0.576787875295338,
      "grad_norm": 0.7303420901298523,
      "learning_rate": 4.006249454876785e-06,
      "loss": 0.6921,
      "step": 20018
    },
    {
      "epoch": 0.5768166887569872,
      "grad_norm": 0.7444576621055603,
      "learning_rate": 4.0057921578579115e-06,
      "loss": 0.7301,
      "step": 20019
    },
    {
      "epoch": 0.5768455022186365,
      "grad_norm": 0.7282289266586304,
      "learning_rate": 4.0053348694975935e-06,
      "loss": 0.7225,
      "step": 20020
    },
    {
      "epoch": 0.5768743156802858,
      "grad_norm": 0.6926485300064087,
      "learning_rate": 4.004877589799816e-06,
      "loss": 0.7002,
      "step": 20021
    },
    {
      "epoch": 0.5769031291419351,
      "grad_norm": 0.7298030853271484,
      "learning_rate": 4.004420318768559e-06,
      "loss": 0.7022,
      "step": 20022
    },
    {
      "epoch": 0.5769319426035844,
      "grad_norm": 0.7010557055473328,
      "learning_rate": 4.003963056407807e-06,
      "loss": 0.6922,
      "step": 20023
    },
    {
      "epoch": 0.5769607560652337,
      "grad_norm": 0.7204756736755371,
      "learning_rate": 4.003505802721541e-06,
      "loss": 0.6948,
      "step": 20024
    },
    {
      "epoch": 0.576989569526883,
      "grad_norm": 0.7066394090652466,
      "learning_rate": 4.003048557713743e-06,
      "loss": 0.7086,
      "step": 20025
    },
    {
      "epoch": 0.5770183829885323,
      "grad_norm": 0.7346580028533936,
      "learning_rate": 4.002591321388396e-06,
      "loss": 0.7277,
      "step": 20026
    },
    {
      "epoch": 0.5770471964501815,
      "grad_norm": 0.7214959263801575,
      "learning_rate": 4.002134093749479e-06,
      "loss": 0.6903,
      "step": 20027
    },
    {
      "epoch": 0.5770760099118308,
      "grad_norm": 0.7212896943092346,
      "learning_rate": 4.00167687480098e-06,
      "loss": 0.7187,
      "step": 20028
    },
    {
      "epoch": 0.5771048233734801,
      "grad_norm": 0.7000017762184143,
      "learning_rate": 4.001219664546875e-06,
      "loss": 0.7193,
      "step": 20029
    },
    {
      "epoch": 0.5771336368351294,
      "grad_norm": 0.7115835547447205,
      "learning_rate": 4.00076246299115e-06,
      "loss": 0.7109,
      "step": 20030
    },
    {
      "epoch": 0.5771624502967787,
      "grad_norm": 0.7309755086898804,
      "learning_rate": 4.000305270137785e-06,
      "loss": 0.7246,
      "step": 20031
    },
    {
      "epoch": 0.5771912637584279,
      "grad_norm": 0.7198075652122498,
      "learning_rate": 3.999848085990758e-06,
      "loss": 0.7253,
      "step": 20032
    },
    {
      "epoch": 0.5772200772200772,
      "grad_norm": 0.7088006138801575,
      "learning_rate": 3.999390910554058e-06,
      "loss": 0.6798,
      "step": 20033
    },
    {
      "epoch": 0.5772488906817265,
      "grad_norm": 0.72420334815979,
      "learning_rate": 3.998933743831663e-06,
      "loss": 0.7221,
      "step": 20034
    },
    {
      "epoch": 0.5772777041433758,
      "grad_norm": 0.7016867995262146,
      "learning_rate": 3.998476585827554e-06,
      "loss": 0.693,
      "step": 20035
    },
    {
      "epoch": 0.577306517605025,
      "grad_norm": 0.7366052865982056,
      "learning_rate": 3.998019436545712e-06,
      "loss": 0.7147,
      "step": 20036
    },
    {
      "epoch": 0.5773353310666743,
      "grad_norm": 0.7107483148574829,
      "learning_rate": 3.99756229599012e-06,
      "loss": 0.7031,
      "step": 20037
    },
    {
      "epoch": 0.5773641445283236,
      "grad_norm": 0.7207172513008118,
      "learning_rate": 3.997105164164759e-06,
      "loss": 0.7133,
      "step": 20038
    },
    {
      "epoch": 0.5773929579899729,
      "grad_norm": 0.7180500626564026,
      "learning_rate": 3.996648041073608e-06,
      "loss": 0.7097,
      "step": 20039
    },
    {
      "epoch": 0.5774217714516222,
      "grad_norm": 0.7205531597137451,
      "learning_rate": 3.99619092672065e-06,
      "loss": 0.7242,
      "step": 20040
    },
    {
      "epoch": 0.5774505849132715,
      "grad_norm": 0.7572861909866333,
      "learning_rate": 3.995733821109866e-06,
      "loss": 0.713,
      "step": 20041
    },
    {
      "epoch": 0.5774793983749208,
      "grad_norm": 0.7251279354095459,
      "learning_rate": 3.995276724245237e-06,
      "loss": 0.6855,
      "step": 20042
    },
    {
      "epoch": 0.5775082118365701,
      "grad_norm": 0.7239702343940735,
      "learning_rate": 3.994819636130742e-06,
      "loss": 0.7176,
      "step": 20043
    },
    {
      "epoch": 0.5775370252982194,
      "grad_norm": 0.6846866607666016,
      "learning_rate": 3.994362556770362e-06,
      "loss": 0.7092,
      "step": 20044
    },
    {
      "epoch": 0.5775658387598687,
      "grad_norm": 0.7348563075065613,
      "learning_rate": 3.993905486168081e-06,
      "loss": 0.6992,
      "step": 20045
    },
    {
      "epoch": 0.5775946522215178,
      "grad_norm": 0.7178941965103149,
      "learning_rate": 3.993448424327877e-06,
      "loss": 0.7118,
      "step": 20046
    },
    {
      "epoch": 0.5776234656831671,
      "grad_norm": 0.7371703386306763,
      "learning_rate": 3.992991371253731e-06,
      "loss": 0.7143,
      "step": 20047
    },
    {
      "epoch": 0.5776522791448164,
      "grad_norm": 0.7377166152000427,
      "learning_rate": 3.992534326949624e-06,
      "loss": 0.6893,
      "step": 20048
    },
    {
      "epoch": 0.5776810926064657,
      "grad_norm": 0.7086170315742493,
      "learning_rate": 3.992077291419535e-06,
      "loss": 0.7074,
      "step": 20049
    },
    {
      "epoch": 0.577709906068115,
      "grad_norm": 0.7082664370536804,
      "learning_rate": 3.991620264667446e-06,
      "loss": 0.7057,
      "step": 20050
    },
    {
      "epoch": 0.5777387195297643,
      "grad_norm": 0.7462476491928101,
      "learning_rate": 3.991163246697336e-06,
      "loss": 0.7068,
      "step": 20051
    },
    {
      "epoch": 0.5777675329914136,
      "grad_norm": 0.770427405834198,
      "learning_rate": 3.990706237513186e-06,
      "loss": 0.6893,
      "step": 20052
    },
    {
      "epoch": 0.5777963464530629,
      "grad_norm": 0.7322568893432617,
      "learning_rate": 3.990249237118976e-06,
      "loss": 0.6978,
      "step": 20053
    },
    {
      "epoch": 0.5778251599147122,
      "grad_norm": 0.7321311831474304,
      "learning_rate": 3.989792245518686e-06,
      "loss": 0.7387,
      "step": 20054
    },
    {
      "epoch": 0.5778539733763615,
      "grad_norm": 0.7139188647270203,
      "learning_rate": 3.989335262716295e-06,
      "loss": 0.7158,
      "step": 20055
    },
    {
      "epoch": 0.5778827868380108,
      "grad_norm": 0.7108044624328613,
      "learning_rate": 3.988878288715782e-06,
      "loss": 0.7156,
      "step": 20056
    },
    {
      "epoch": 0.57791160029966,
      "grad_norm": 0.7408888936042786,
      "learning_rate": 3.9884213235211296e-06,
      "loss": 0.6873,
      "step": 20057
    },
    {
      "epoch": 0.5779404137613093,
      "grad_norm": 0.7409921884536743,
      "learning_rate": 3.987964367136317e-06,
      "loss": 0.6929,
      "step": 20058
    },
    {
      "epoch": 0.5779692272229585,
      "grad_norm": 0.7256609201431274,
      "learning_rate": 3.987507419565323e-06,
      "loss": 0.7181,
      "step": 20059
    },
    {
      "epoch": 0.5779980406846078,
      "grad_norm": 0.72445148229599,
      "learning_rate": 3.987050480812127e-06,
      "loss": 0.6956,
      "step": 20060
    },
    {
      "epoch": 0.5780268541462571,
      "grad_norm": 0.7367987036705017,
      "learning_rate": 3.986593550880709e-06,
      "loss": 0.7252,
      "step": 20061
    },
    {
      "epoch": 0.5780556676079064,
      "grad_norm": 0.7270298600196838,
      "learning_rate": 3.986136629775048e-06,
      "loss": 0.7112,
      "step": 20062
    },
    {
      "epoch": 0.5780844810695557,
      "grad_norm": 0.7057446241378784,
      "learning_rate": 3.985679717499123e-06,
      "loss": 0.7086,
      "step": 20063
    },
    {
      "epoch": 0.578113294531205,
      "grad_norm": 0.7177121639251709,
      "learning_rate": 3.985222814056915e-06,
      "loss": 0.7154,
      "step": 20064
    },
    {
      "epoch": 0.5781421079928543,
      "grad_norm": 0.7333115339279175,
      "learning_rate": 3.9847659194524e-06,
      "loss": 0.7065,
      "step": 20065
    },
    {
      "epoch": 0.5781709214545036,
      "grad_norm": 0.7105408310890198,
      "learning_rate": 3.98430903368956e-06,
      "loss": 0.7024,
      "step": 20066
    },
    {
      "epoch": 0.5781997349161528,
      "grad_norm": 0.7309815883636475,
      "learning_rate": 3.9838521567723735e-06,
      "loss": 0.7048,
      "step": 20067
    },
    {
      "epoch": 0.5782285483778021,
      "grad_norm": 0.7125903367996216,
      "learning_rate": 3.9833952887048175e-06,
      "loss": 0.7042,
      "step": 20068
    },
    {
      "epoch": 0.5782573618394514,
      "grad_norm": 0.7312868237495422,
      "learning_rate": 3.982938429490873e-06,
      "loss": 0.7098,
      "step": 20069
    },
    {
      "epoch": 0.5782861753011007,
      "grad_norm": 0.7290487885475159,
      "learning_rate": 3.982481579134518e-06,
      "loss": 0.6919,
      "step": 20070
    },
    {
      "epoch": 0.57831498876275,
      "grad_norm": 0.7107909917831421,
      "learning_rate": 3.982024737639731e-06,
      "loss": 0.7305,
      "step": 20071
    },
    {
      "epoch": 0.5783438022243992,
      "grad_norm": 0.7320802807807922,
      "learning_rate": 3.981567905010492e-06,
      "loss": 0.7272,
      "step": 20072
    },
    {
      "epoch": 0.5783726156860485,
      "grad_norm": 0.7489588856697083,
      "learning_rate": 3.981111081250778e-06,
      "loss": 0.6924,
      "step": 20073
    },
    {
      "epoch": 0.5784014291476978,
      "grad_norm": 0.7351298928260803,
      "learning_rate": 3.980654266364569e-06,
      "loss": 0.7198,
      "step": 20074
    },
    {
      "epoch": 0.5784302426093471,
      "grad_norm": 0.7087358832359314,
      "learning_rate": 3.98019746035584e-06,
      "loss": 0.7041,
      "step": 20075
    },
    {
      "epoch": 0.5784590560709963,
      "grad_norm": 0.7357577085494995,
      "learning_rate": 3.979740663228574e-06,
      "loss": 0.7205,
      "step": 20076
    },
    {
      "epoch": 0.5784878695326456,
      "grad_norm": 0.7155153751373291,
      "learning_rate": 3.979283874986747e-06,
      "loss": 0.6964,
      "step": 20077
    },
    {
      "epoch": 0.5785166829942949,
      "grad_norm": 0.7074885964393616,
      "learning_rate": 3.978827095634336e-06,
      "loss": 0.6918,
      "step": 20078
    },
    {
      "epoch": 0.5785454964559442,
      "grad_norm": 0.7251901030540466,
      "learning_rate": 3.97837032517532e-06,
      "loss": 0.7091,
      "step": 20079
    },
    {
      "epoch": 0.5785743099175935,
      "grad_norm": 0.7200910449028015,
      "learning_rate": 3.977913563613676e-06,
      "loss": 0.7044,
      "step": 20080
    },
    {
      "epoch": 0.5786031233792428,
      "grad_norm": 0.7188688516616821,
      "learning_rate": 3.977456810953384e-06,
      "loss": 0.6918,
      "step": 20081
    },
    {
      "epoch": 0.5786319368408921,
      "grad_norm": 0.7151405811309814,
      "learning_rate": 3.977000067198422e-06,
      "loss": 0.6889,
      "step": 20082
    },
    {
      "epoch": 0.5786607503025414,
      "grad_norm": 0.7575345635414124,
      "learning_rate": 3.976543332352767e-06,
      "loss": 0.7129,
      "step": 20083
    },
    {
      "epoch": 0.5786895637641907,
      "grad_norm": 0.7200107574462891,
      "learning_rate": 3.976086606420396e-06,
      "loss": 0.7139,
      "step": 20084
    },
    {
      "epoch": 0.57871837722584,
      "grad_norm": 0.7216672301292419,
      "learning_rate": 3.975629889405287e-06,
      "loss": 0.7081,
      "step": 20085
    },
    {
      "epoch": 0.5787471906874891,
      "grad_norm": 0.7613641619682312,
      "learning_rate": 3.975173181311417e-06,
      "loss": 0.6884,
      "step": 20086
    },
    {
      "epoch": 0.5787760041491384,
      "grad_norm": 0.7140392661094666,
      "learning_rate": 3.9747164821427645e-06,
      "loss": 0.6887,
      "step": 20087
    },
    {
      "epoch": 0.5788048176107877,
      "grad_norm": 0.7284187078475952,
      "learning_rate": 3.974259791903308e-06,
      "loss": 0.7117,
      "step": 20088
    },
    {
      "epoch": 0.578833631072437,
      "grad_norm": 0.7141442894935608,
      "learning_rate": 3.973803110597022e-06,
      "loss": 0.7069,
      "step": 20089
    },
    {
      "epoch": 0.5788624445340863,
      "grad_norm": 0.7069556713104248,
      "learning_rate": 3.9733464382278856e-06,
      "loss": 0.6979,
      "step": 20090
    },
    {
      "epoch": 0.5788912579957356,
      "grad_norm": 0.7054656147956848,
      "learning_rate": 3.972889774799875e-06,
      "loss": 0.7003,
      "step": 20091
    },
    {
      "epoch": 0.5789200714573849,
      "grad_norm": 0.719034731388092,
      "learning_rate": 3.972433120316966e-06,
      "loss": 0.6986,
      "step": 20092
    },
    {
      "epoch": 0.5789488849190342,
      "grad_norm": 0.7339128851890564,
      "learning_rate": 3.971976474783139e-06,
      "loss": 0.7036,
      "step": 20093
    },
    {
      "epoch": 0.5789776983806835,
      "grad_norm": 0.7048594951629639,
      "learning_rate": 3.97151983820237e-06,
      "loss": 0.6837,
      "step": 20094
    },
    {
      "epoch": 0.5790065118423328,
      "grad_norm": 0.7232036590576172,
      "learning_rate": 3.9710632105786355e-06,
      "loss": 0.6932,
      "step": 20095
    },
    {
      "epoch": 0.579035325303982,
      "grad_norm": 0.7250344157218933,
      "learning_rate": 3.970606591915911e-06,
      "loss": 0.7151,
      "step": 20096
    },
    {
      "epoch": 0.5790641387656313,
      "grad_norm": 0.7213476896286011,
      "learning_rate": 3.970149982218173e-06,
      "loss": 0.7102,
      "step": 20097
    },
    {
      "epoch": 0.5790929522272806,
      "grad_norm": 0.7042356133460999,
      "learning_rate": 3.969693381489401e-06,
      "loss": 0.688,
      "step": 20098
    },
    {
      "epoch": 0.5791217656889298,
      "grad_norm": 0.7191218137741089,
      "learning_rate": 3.969236789733569e-06,
      "loss": 0.7121,
      "step": 20099
    },
    {
      "epoch": 0.5791505791505791,
      "grad_norm": 0.7478200197219849,
      "learning_rate": 3.968780206954654e-06,
      "loss": 0.7219,
      "step": 20100
    },
    {
      "epoch": 0.5791793926122284,
      "grad_norm": 0.7154759168624878,
      "learning_rate": 3.968323633156634e-06,
      "loss": 0.7125,
      "step": 20101
    },
    {
      "epoch": 0.5792082060738777,
      "grad_norm": 0.7546247839927673,
      "learning_rate": 3.967867068343481e-06,
      "loss": 0.7316,
      "step": 20102
    },
    {
      "epoch": 0.579237019535527,
      "grad_norm": 0.7260411977767944,
      "learning_rate": 3.967410512519176e-06,
      "loss": 0.6789,
      "step": 20103
    },
    {
      "epoch": 0.5792658329971763,
      "grad_norm": 0.7109570503234863,
      "learning_rate": 3.966953965687692e-06,
      "loss": 0.6899,
      "step": 20104
    },
    {
      "epoch": 0.5792946464588256,
      "grad_norm": 0.6988444328308105,
      "learning_rate": 3.966497427853006e-06,
      "loss": 0.6874,
      "step": 20105
    },
    {
      "epoch": 0.5793234599204748,
      "grad_norm": 0.7254341840744019,
      "learning_rate": 3.966040899019096e-06,
      "loss": 0.6929,
      "step": 20106
    },
    {
      "epoch": 0.5793522733821241,
      "grad_norm": 0.7245362997055054,
      "learning_rate": 3.965584379189936e-06,
      "loss": 0.7066,
      "step": 20107
    },
    {
      "epoch": 0.5793810868437734,
      "grad_norm": 0.7121627926826477,
      "learning_rate": 3.965127868369501e-06,
      "loss": 0.6934,
      "step": 20108
    },
    {
      "epoch": 0.5794099003054227,
      "grad_norm": 0.7258414030075073,
      "learning_rate": 3.964671366561768e-06,
      "loss": 0.7049,
      "step": 20109
    },
    {
      "epoch": 0.579438713767072,
      "grad_norm": 0.7291973233222961,
      "learning_rate": 3.964214873770713e-06,
      "loss": 0.7186,
      "step": 20110
    },
    {
      "epoch": 0.5794675272287213,
      "grad_norm": 0.7128391861915588,
      "learning_rate": 3.963758390000311e-06,
      "loss": 0.7158,
      "step": 20111
    },
    {
      "epoch": 0.5794963406903706,
      "grad_norm": 0.7167853713035583,
      "learning_rate": 3.963301915254538e-06,
      "loss": 0.6948,
      "step": 20112
    },
    {
      "epoch": 0.5795251541520198,
      "grad_norm": 0.7178062796592712,
      "learning_rate": 3.962845449537368e-06,
      "loss": 0.7067,
      "step": 20113
    },
    {
      "epoch": 0.5795539676136691,
      "grad_norm": 0.73222815990448,
      "learning_rate": 3.962388992852778e-06,
      "loss": 0.7022,
      "step": 20114
    },
    {
      "epoch": 0.5795827810753184,
      "grad_norm": 0.7268752455711365,
      "learning_rate": 3.961932545204743e-06,
      "loss": 0.6929,
      "step": 20115
    },
    {
      "epoch": 0.5796115945369676,
      "grad_norm": 0.7266660928726196,
      "learning_rate": 3.961476106597235e-06,
      "loss": 0.6947,
      "step": 20116
    },
    {
      "epoch": 0.5796404079986169,
      "grad_norm": 0.7207236289978027,
      "learning_rate": 3.9610196770342345e-06,
      "loss": 0.7157,
      "step": 20117
    },
    {
      "epoch": 0.5796692214602662,
      "grad_norm": 0.7726648449897766,
      "learning_rate": 3.9605632565197135e-06,
      "loss": 0.6936,
      "step": 20118
    },
    {
      "epoch": 0.5796980349219155,
      "grad_norm": 0.7408387660980225,
      "learning_rate": 3.960106845057649e-06,
      "loss": 0.7318,
      "step": 20119
    },
    {
      "epoch": 0.5797268483835648,
      "grad_norm": 0.7307618856430054,
      "learning_rate": 3.959650442652013e-06,
      "loss": 0.7031,
      "step": 20120
    },
    {
      "epoch": 0.5797556618452141,
      "grad_norm": 0.7143833637237549,
      "learning_rate": 3.959194049306782e-06,
      "loss": 0.6956,
      "step": 20121
    },
    {
      "epoch": 0.5797844753068634,
      "grad_norm": 0.7083417177200317,
      "learning_rate": 3.95873766502593e-06,
      "loss": 0.7181,
      "step": 20122
    },
    {
      "epoch": 0.5798132887685127,
      "grad_norm": 0.7050224542617798,
      "learning_rate": 3.958281289813433e-06,
      "loss": 0.6996,
      "step": 20123
    },
    {
      "epoch": 0.579842102230162,
      "grad_norm": 0.7278926372528076,
      "learning_rate": 3.957824923673263e-06,
      "loss": 0.7228,
      "step": 20124
    },
    {
      "epoch": 0.5798709156918113,
      "grad_norm": 0.7128801345825195,
      "learning_rate": 3.957368566609396e-06,
      "loss": 0.7057,
      "step": 20125
    },
    {
      "epoch": 0.5798997291534604,
      "grad_norm": 0.7212271094322205,
      "learning_rate": 3.956912218625808e-06,
      "loss": 0.7152,
      "step": 20126
    },
    {
      "epoch": 0.5799285426151097,
      "grad_norm": 0.735095739364624,
      "learning_rate": 3.956455879726472e-06,
      "loss": 0.6944,
      "step": 20127
    },
    {
      "epoch": 0.579957356076759,
      "grad_norm": 0.7029622197151184,
      "learning_rate": 3.95599954991536e-06,
      "loss": 0.6747,
      "step": 20128
    },
    {
      "epoch": 0.5799861695384083,
      "grad_norm": 0.7293274998664856,
      "learning_rate": 3.9555432291964495e-06,
      "loss": 0.7074,
      "step": 20129
    },
    {
      "epoch": 0.5800149830000576,
      "grad_norm": 0.7323285341262817,
      "learning_rate": 3.9550869175737135e-06,
      "loss": 0.7102,
      "step": 20130
    },
    {
      "epoch": 0.5800437964617069,
      "grad_norm": 0.7211315631866455,
      "learning_rate": 3.954630615051127e-06,
      "loss": 0.6878,
      "step": 20131
    },
    {
      "epoch": 0.5800726099233562,
      "grad_norm": 0.7232693433761597,
      "learning_rate": 3.954174321632661e-06,
      "loss": 0.7176,
      "step": 20132
    },
    {
      "epoch": 0.5801014233850055,
      "grad_norm": 0.734108567237854,
      "learning_rate": 3.953718037322293e-06,
      "loss": 0.6988,
      "step": 20133
    },
    {
      "epoch": 0.5801302368466548,
      "grad_norm": 0.7321568727493286,
      "learning_rate": 3.9532617621239945e-06,
      "loss": 0.7153,
      "step": 20134
    },
    {
      "epoch": 0.5801590503083041,
      "grad_norm": 0.7439985871315002,
      "learning_rate": 3.952805496041739e-06,
      "loss": 0.7272,
      "step": 20135
    },
    {
      "epoch": 0.5801878637699533,
      "grad_norm": 0.7088432908058167,
      "learning_rate": 3.9523492390795015e-06,
      "loss": 0.704,
      "step": 20136
    },
    {
      "epoch": 0.5802166772316026,
      "grad_norm": 0.6793501973152161,
      "learning_rate": 3.951892991241255e-06,
      "loss": 0.6741,
      "step": 20137
    },
    {
      "epoch": 0.5802454906932519,
      "grad_norm": 0.7264344096183777,
      "learning_rate": 3.951436752530973e-06,
      "loss": 0.7097,
      "step": 20138
    },
    {
      "epoch": 0.5802743041549012,
      "grad_norm": 0.7457954287528992,
      "learning_rate": 3.950980522952628e-06,
      "loss": 0.7195,
      "step": 20139
    },
    {
      "epoch": 0.5803031176165504,
      "grad_norm": 0.7127888798713684,
      "learning_rate": 3.950524302510192e-06,
      "loss": 0.6983,
      "step": 20140
    },
    {
      "epoch": 0.5803319310781997,
      "grad_norm": 0.7122254967689514,
      "learning_rate": 3.9500680912076426e-06,
      "loss": 0.7251,
      "step": 20141
    },
    {
      "epoch": 0.580360744539849,
      "grad_norm": 0.7121716141700745,
      "learning_rate": 3.9496118890489505e-06,
      "loss": 0.7057,
      "step": 20142
    },
    {
      "epoch": 0.5803895580014983,
      "grad_norm": 0.7492533922195435,
      "learning_rate": 3.949155696038088e-06,
      "loss": 0.731,
      "step": 20143
    },
    {
      "epoch": 0.5804183714631476,
      "grad_norm": 0.713716447353363,
      "learning_rate": 3.9486995121790305e-06,
      "loss": 0.7097,
      "step": 20144
    },
    {
      "epoch": 0.5804471849247969,
      "grad_norm": 0.7239062190055847,
      "learning_rate": 3.9482433374757485e-06,
      "loss": 0.6929,
      "step": 20145
    },
    {
      "epoch": 0.5804759983864461,
      "grad_norm": 0.7412858605384827,
      "learning_rate": 3.947787171932215e-06,
      "loss": 0.7117,
      "step": 20146
    },
    {
      "epoch": 0.5805048118480954,
      "grad_norm": 0.7190899848937988,
      "learning_rate": 3.947331015552403e-06,
      "loss": 0.719,
      "step": 20147
    },
    {
      "epoch": 0.5805336253097447,
      "grad_norm": 0.7029191851615906,
      "learning_rate": 3.946874868340287e-06,
      "loss": 0.7118,
      "step": 20148
    },
    {
      "epoch": 0.580562438771394,
      "grad_norm": 0.6875964403152466,
      "learning_rate": 3.946418730299839e-06,
      "loss": 0.6794,
      "step": 20149
    },
    {
      "epoch": 0.5805912522330433,
      "grad_norm": 0.7203632593154907,
      "learning_rate": 3.945962601435029e-06,
      "loss": 0.6911,
      "step": 20150
    },
    {
      "epoch": 0.5806200656946926,
      "grad_norm": 0.704930305480957,
      "learning_rate": 3.9455064817498315e-06,
      "loss": 0.6856,
      "step": 20151
    },
    {
      "epoch": 0.5806488791563419,
      "grad_norm": 0.7392358183860779,
      "learning_rate": 3.9450503712482164e-06,
      "loss": 0.7252,
      "step": 20152
    },
    {
      "epoch": 0.5806776926179911,
      "grad_norm": 0.7255198359489441,
      "learning_rate": 3.94459426993416e-06,
      "loss": 0.6927,
      "step": 20153
    },
    {
      "epoch": 0.5807065060796404,
      "grad_norm": 0.7436606287956238,
      "learning_rate": 3.944138177811633e-06,
      "loss": 0.7116,
      "step": 20154
    },
    {
      "epoch": 0.5807353195412897,
      "grad_norm": 0.7117560505867004,
      "learning_rate": 3.943682094884607e-06,
      "loss": 0.7051,
      "step": 20155
    },
    {
      "epoch": 0.5807641330029389,
      "grad_norm": 0.709290087223053,
      "learning_rate": 3.9432260211570535e-06,
      "loss": 0.7191,
      "step": 20156
    },
    {
      "epoch": 0.5807929464645882,
      "grad_norm": 0.7292981147766113,
      "learning_rate": 3.942769956632945e-06,
      "loss": 0.6753,
      "step": 20157
    },
    {
      "epoch": 0.5808217599262375,
      "grad_norm": 0.6926228404045105,
      "learning_rate": 3.942313901316254e-06,
      "loss": 0.7208,
      "step": 20158
    },
    {
      "epoch": 0.5808505733878868,
      "grad_norm": 0.7226572036743164,
      "learning_rate": 3.941857855210951e-06,
      "loss": 0.6997,
      "step": 20159
    },
    {
      "epoch": 0.5808793868495361,
      "grad_norm": 0.7195788025856018,
      "learning_rate": 3.941401818321008e-06,
      "loss": 0.6961,
      "step": 20160
    },
    {
      "epoch": 0.5809082003111854,
      "grad_norm": 0.7109841704368591,
      "learning_rate": 3.940945790650397e-06,
      "loss": 0.6785,
      "step": 20161
    },
    {
      "epoch": 0.5809370137728347,
      "grad_norm": 0.7710865139961243,
      "learning_rate": 3.940489772203088e-06,
      "loss": 0.7177,
      "step": 20162
    },
    {
      "epoch": 0.580965827234484,
      "grad_norm": 0.731947660446167,
      "learning_rate": 3.940033762983057e-06,
      "loss": 0.6973,
      "step": 20163
    },
    {
      "epoch": 0.5809946406961333,
      "grad_norm": 0.7134810090065002,
      "learning_rate": 3.93957776299427e-06,
      "loss": 0.7,
      "step": 20164
    },
    {
      "epoch": 0.5810234541577826,
      "grad_norm": 0.7411278486251831,
      "learning_rate": 3.939121772240701e-06,
      "loss": 0.7152,
      "step": 20165
    },
    {
      "epoch": 0.5810522676194319,
      "grad_norm": 0.734957218170166,
      "learning_rate": 3.938665790726323e-06,
      "loss": 0.7319,
      "step": 20166
    },
    {
      "epoch": 0.581081081081081,
      "grad_norm": 0.7189474105834961,
      "learning_rate": 3.938209818455102e-06,
      "loss": 0.7266,
      "step": 20167
    },
    {
      "epoch": 0.5811098945427303,
      "grad_norm": 0.7149302363395691,
      "learning_rate": 3.937753855431014e-06,
      "loss": 0.6744,
      "step": 20168
    },
    {
      "epoch": 0.5811387080043796,
      "grad_norm": 0.7256224751472473,
      "learning_rate": 3.937297901658028e-06,
      "loss": 0.7066,
      "step": 20169
    },
    {
      "epoch": 0.5811675214660289,
      "grad_norm": 0.7361341714859009,
      "learning_rate": 3.936841957140115e-06,
      "loss": 0.7025,
      "step": 20170
    },
    {
      "epoch": 0.5811963349276782,
      "grad_norm": 0.7137372493743896,
      "learning_rate": 3.936386021881246e-06,
      "loss": 0.7025,
      "step": 20171
    },
    {
      "epoch": 0.5812251483893275,
      "grad_norm": 0.7545584440231323,
      "learning_rate": 3.935930095885391e-06,
      "loss": 0.7085,
      "step": 20172
    },
    {
      "epoch": 0.5812539618509768,
      "grad_norm": 0.7425175905227661,
      "learning_rate": 3.935474179156521e-06,
      "loss": 0.7381,
      "step": 20173
    },
    {
      "epoch": 0.5812827753126261,
      "grad_norm": 0.7224605679512024,
      "learning_rate": 3.935018271698607e-06,
      "loss": 0.7139,
      "step": 20174
    },
    {
      "epoch": 0.5813115887742754,
      "grad_norm": 0.7445029616355896,
      "learning_rate": 3.934562373515619e-06,
      "loss": 0.6879,
      "step": 20175
    },
    {
      "epoch": 0.5813404022359246,
      "grad_norm": 0.7452827095985413,
      "learning_rate": 3.934106484611526e-06,
      "loss": 0.7007,
      "step": 20176
    },
    {
      "epoch": 0.5813692156975739,
      "grad_norm": 0.7093279957771301,
      "learning_rate": 3.933650604990302e-06,
      "loss": 0.7077,
      "step": 20177
    },
    {
      "epoch": 0.5813980291592232,
      "grad_norm": 0.7329966425895691,
      "learning_rate": 3.933194734655916e-06,
      "loss": 0.7175,
      "step": 20178
    },
    {
      "epoch": 0.5814268426208725,
      "grad_norm": 0.7495335936546326,
      "learning_rate": 3.9327388736123366e-06,
      "loss": 0.7162,
      "step": 20179
    },
    {
      "epoch": 0.5814556560825217,
      "grad_norm": 0.7756819725036621,
      "learning_rate": 3.932283021863535e-06,
      "loss": 0.7276,
      "step": 20180
    },
    {
      "epoch": 0.581484469544171,
      "grad_norm": 0.738808274269104,
      "learning_rate": 3.931827179413481e-06,
      "loss": 0.6995,
      "step": 20181
    },
    {
      "epoch": 0.5815132830058203,
      "grad_norm": 0.697687566280365,
      "learning_rate": 3.931371346266144e-06,
      "loss": 0.6785,
      "step": 20182
    },
    {
      "epoch": 0.5815420964674696,
      "grad_norm": 0.73344486951828,
      "learning_rate": 3.930915522425494e-06,
      "loss": 0.7141,
      "step": 20183
    },
    {
      "epoch": 0.5815709099291189,
      "grad_norm": 0.7249916195869446,
      "learning_rate": 3.9304597078955005e-06,
      "loss": 0.7232,
      "step": 20184
    },
    {
      "epoch": 0.5815997233907682,
      "grad_norm": 0.6819983720779419,
      "learning_rate": 3.930003902680136e-06,
      "loss": 0.6736,
      "step": 20185
    },
    {
      "epoch": 0.5816285368524174,
      "grad_norm": 0.7104019522666931,
      "learning_rate": 3.929548106783367e-06,
      "loss": 0.6797,
      "step": 20186
    },
    {
      "epoch": 0.5816573503140667,
      "grad_norm": 0.726004421710968,
      "learning_rate": 3.929092320209163e-06,
      "loss": 0.7091,
      "step": 20187
    },
    {
      "epoch": 0.581686163775716,
      "grad_norm": 0.7342800498008728,
      "learning_rate": 3.928636542961495e-06,
      "loss": 0.693,
      "step": 20188
    },
    {
      "epoch": 0.5817149772373653,
      "grad_norm": 0.711305558681488,
      "learning_rate": 3.9281807750443305e-06,
      "loss": 0.7098,
      "step": 20189
    },
    {
      "epoch": 0.5817437906990146,
      "grad_norm": 0.7119581699371338,
      "learning_rate": 3.927725016461642e-06,
      "loss": 0.6969,
      "step": 20190
    },
    {
      "epoch": 0.5817726041606639,
      "grad_norm": 0.7585002779960632,
      "learning_rate": 3.927269267217396e-06,
      "loss": 0.7307,
      "step": 20191
    },
    {
      "epoch": 0.5818014176223132,
      "grad_norm": 0.7280385494232178,
      "learning_rate": 3.926813527315563e-06,
      "loss": 0.6932,
      "step": 20192
    },
    {
      "epoch": 0.5818302310839625,
      "grad_norm": 0.7309803366661072,
      "learning_rate": 3.9263577967601104e-06,
      "loss": 0.7042,
      "step": 20193
    },
    {
      "epoch": 0.5818590445456117,
      "grad_norm": 0.7448665499687195,
      "learning_rate": 3.925902075555009e-06,
      "loss": 0.714,
      "step": 20194
    },
    {
      "epoch": 0.581887858007261,
      "grad_norm": 0.7208306193351746,
      "learning_rate": 3.925446363704227e-06,
      "loss": 0.6977,
      "step": 20195
    },
    {
      "epoch": 0.5819166714689102,
      "grad_norm": 0.7156243920326233,
      "learning_rate": 3.924990661211732e-06,
      "loss": 0.7042,
      "step": 20196
    },
    {
      "epoch": 0.5819454849305595,
      "grad_norm": 0.7366439700126648,
      "learning_rate": 3.924534968081495e-06,
      "loss": 0.7398,
      "step": 20197
    },
    {
      "epoch": 0.5819742983922088,
      "grad_norm": 0.7394742965698242,
      "learning_rate": 3.924079284317482e-06,
      "loss": 0.7387,
      "step": 20198
    },
    {
      "epoch": 0.5820031118538581,
      "grad_norm": 0.7460702657699585,
      "learning_rate": 3.923623609923664e-06,
      "loss": 0.7192,
      "step": 20199
    },
    {
      "epoch": 0.5820319253155074,
      "grad_norm": 0.7443220615386963,
      "learning_rate": 3.923167944904006e-06,
      "loss": 0.7369,
      "step": 20200
    },
    {
      "epoch": 0.5820607387771567,
      "grad_norm": 0.7344780564308167,
      "learning_rate": 3.92271228926248e-06,
      "loss": 0.7263,
      "step": 20201
    },
    {
      "epoch": 0.582089552238806,
      "grad_norm": 0.7297662496566772,
      "learning_rate": 3.922256643003054e-06,
      "loss": 0.6892,
      "step": 20202
    },
    {
      "epoch": 0.5821183657004553,
      "grad_norm": 0.7594982385635376,
      "learning_rate": 3.921801006129695e-06,
      "loss": 0.7305,
      "step": 20203
    },
    {
      "epoch": 0.5821471791621046,
      "grad_norm": 0.7162021994590759,
      "learning_rate": 3.921345378646372e-06,
      "loss": 0.7069,
      "step": 20204
    },
    {
      "epoch": 0.5821759926237539,
      "grad_norm": 0.7360182404518127,
      "learning_rate": 3.9208897605570505e-06,
      "loss": 0.7086,
      "step": 20205
    },
    {
      "epoch": 0.5822048060854031,
      "grad_norm": 0.7076606750488281,
      "learning_rate": 3.920434151865702e-06,
      "loss": 0.7085,
      "step": 20206
    },
    {
      "epoch": 0.5822336195470523,
      "grad_norm": 0.7023105621337891,
      "learning_rate": 3.919978552576292e-06,
      "loss": 0.683,
      "step": 20207
    },
    {
      "epoch": 0.5822624330087016,
      "grad_norm": 0.7118605971336365,
      "learning_rate": 3.91952296269279e-06,
      "loss": 0.6938,
      "step": 20208
    },
    {
      "epoch": 0.5822912464703509,
      "grad_norm": 0.7117988467216492,
      "learning_rate": 3.919067382219164e-06,
      "loss": 0.6842,
      "step": 20209
    },
    {
      "epoch": 0.5823200599320002,
      "grad_norm": 0.7130577564239502,
      "learning_rate": 3.918611811159379e-06,
      "loss": 0.6937,
      "step": 20210
    },
    {
      "epoch": 0.5823488733936495,
      "grad_norm": 0.7346508502960205,
      "learning_rate": 3.918156249517405e-06,
      "loss": 0.7108,
      "step": 20211
    },
    {
      "epoch": 0.5823776868552988,
      "grad_norm": 0.7058601975440979,
      "learning_rate": 3.917700697297205e-06,
      "loss": 0.6947,
      "step": 20212
    },
    {
      "epoch": 0.5824065003169481,
      "grad_norm": 0.7065528035163879,
      "learning_rate": 3.917245154502754e-06,
      "loss": 0.6989,
      "step": 20213
    },
    {
      "epoch": 0.5824353137785974,
      "grad_norm": 0.7147529125213623,
      "learning_rate": 3.9167896211380146e-06,
      "loss": 0.6772,
      "step": 20214
    },
    {
      "epoch": 0.5824641272402467,
      "grad_norm": 0.7321820259094238,
      "learning_rate": 3.916334097206954e-06,
      "loss": 0.7225,
      "step": 20215
    },
    {
      "epoch": 0.582492940701896,
      "grad_norm": 0.7219451069831848,
      "learning_rate": 3.915878582713541e-06,
      "loss": 0.7039,
      "step": 20216
    },
    {
      "epoch": 0.5825217541635452,
      "grad_norm": 0.7285202741622925,
      "learning_rate": 3.915423077661742e-06,
      "loss": 0.7039,
      "step": 20217
    },
    {
      "epoch": 0.5825505676251945,
      "grad_norm": 0.7006787061691284,
      "learning_rate": 3.914967582055524e-06,
      "loss": 0.7057,
      "step": 20218
    },
    {
      "epoch": 0.5825793810868438,
      "grad_norm": 0.7375966906547546,
      "learning_rate": 3.9145120958988535e-06,
      "loss": 0.6974,
      "step": 20219
    },
    {
      "epoch": 0.5826081945484931,
      "grad_norm": 0.7226738929748535,
      "learning_rate": 3.914056619195697e-06,
      "loss": 0.6842,
      "step": 20220
    },
    {
      "epoch": 0.5826370080101423,
      "grad_norm": 0.7329463958740234,
      "learning_rate": 3.9136011519500226e-06,
      "loss": 0.729,
      "step": 20221
    },
    {
      "epoch": 0.5826658214717916,
      "grad_norm": 0.7007284760475159,
      "learning_rate": 3.913145694165794e-06,
      "loss": 0.6949,
      "step": 20222
    },
    {
      "epoch": 0.5826946349334409,
      "grad_norm": 0.7036109566688538,
      "learning_rate": 3.912690245846982e-06,
      "loss": 0.6874,
      "step": 20223
    },
    {
      "epoch": 0.5827234483950902,
      "grad_norm": 0.7006140351295471,
      "learning_rate": 3.912234806997549e-06,
      "loss": 0.6757,
      "step": 20224
    },
    {
      "epoch": 0.5827522618567395,
      "grad_norm": 0.7065570950508118,
      "learning_rate": 3.911779377621466e-06,
      "loss": 0.7,
      "step": 20225
    },
    {
      "epoch": 0.5827810753183887,
      "grad_norm": 0.717424213886261,
      "learning_rate": 3.911323957722695e-06,
      "loss": 0.7006,
      "step": 20226
    },
    {
      "epoch": 0.582809888780038,
      "grad_norm": 0.7235425114631653,
      "learning_rate": 3.910868547305204e-06,
      "loss": 0.7113,
      "step": 20227
    },
    {
      "epoch": 0.5828387022416873,
      "grad_norm": 0.7276061773300171,
      "learning_rate": 3.91041314637296e-06,
      "loss": 0.7025,
      "step": 20228
    },
    {
      "epoch": 0.5828675157033366,
      "grad_norm": 0.7164051532745361,
      "learning_rate": 3.909957754929929e-06,
      "loss": 0.7198,
      "step": 20229
    },
    {
      "epoch": 0.5828963291649859,
      "grad_norm": 0.7357739210128784,
      "learning_rate": 3.9095023729800766e-06,
      "loss": 0.697,
      "step": 20230
    },
    {
      "epoch": 0.5829251426266352,
      "grad_norm": 0.720158576965332,
      "learning_rate": 3.909047000527367e-06,
      "loss": 0.7157,
      "step": 20231
    },
    {
      "epoch": 0.5829539560882845,
      "grad_norm": 0.714015543460846,
      "learning_rate": 3.908591637575768e-06,
      "loss": 0.6786,
      "step": 20232
    },
    {
      "epoch": 0.5829827695499338,
      "grad_norm": 0.7499525547027588,
      "learning_rate": 3.908136284129245e-06,
      "loss": 0.7205,
      "step": 20233
    },
    {
      "epoch": 0.583011583011583,
      "grad_norm": 0.7242482304573059,
      "learning_rate": 3.907680940191764e-06,
      "loss": 0.679,
      "step": 20234
    },
    {
      "epoch": 0.5830403964732322,
      "grad_norm": 0.7326055765151978,
      "learning_rate": 3.907225605767289e-06,
      "loss": 0.7036,
      "step": 20235
    },
    {
      "epoch": 0.5830692099348815,
      "grad_norm": 0.7233127951622009,
      "learning_rate": 3.906770280859785e-06,
      "loss": 0.6976,
      "step": 20236
    },
    {
      "epoch": 0.5830980233965308,
      "grad_norm": 0.7304344177246094,
      "learning_rate": 3.906314965473221e-06,
      "loss": 0.6998,
      "step": 20237
    },
    {
      "epoch": 0.5831268368581801,
      "grad_norm": 0.7244289517402649,
      "learning_rate": 3.905859659611561e-06,
      "loss": 0.7119,
      "step": 20238
    },
    {
      "epoch": 0.5831556503198294,
      "grad_norm": 0.7408344745635986,
      "learning_rate": 3.905404363278769e-06,
      "loss": 0.7047,
      "step": 20239
    },
    {
      "epoch": 0.5831844637814787,
      "grad_norm": 0.7458330988883972,
      "learning_rate": 3.904949076478811e-06,
      "loss": 0.6812,
      "step": 20240
    },
    {
      "epoch": 0.583213277243128,
      "grad_norm": 0.7251101732254028,
      "learning_rate": 3.904493799215652e-06,
      "loss": 0.7092,
      "step": 20241
    },
    {
      "epoch": 0.5832420907047773,
      "grad_norm": 0.6955528855323792,
      "learning_rate": 3.904038531493257e-06,
      "loss": 0.6846,
      "step": 20242
    },
    {
      "epoch": 0.5832709041664266,
      "grad_norm": 0.7100752592086792,
      "learning_rate": 3.903583273315589e-06,
      "loss": 0.6974,
      "step": 20243
    },
    {
      "epoch": 0.5832997176280759,
      "grad_norm": 0.7270651459693909,
      "learning_rate": 3.903128024686617e-06,
      "loss": 0.6941,
      "step": 20244
    },
    {
      "epoch": 0.5833285310897252,
      "grad_norm": 0.7279752492904663,
      "learning_rate": 3.902672785610302e-06,
      "loss": 0.7115,
      "step": 20245
    },
    {
      "epoch": 0.5833573445513744,
      "grad_norm": 0.7121752500534058,
      "learning_rate": 3.902217556090611e-06,
      "loss": 0.6935,
      "step": 20246
    },
    {
      "epoch": 0.5833861580130237,
      "grad_norm": 0.7166033387184143,
      "learning_rate": 3.901762336131506e-06,
      "loss": 0.7108,
      "step": 20247
    },
    {
      "epoch": 0.5834149714746729,
      "grad_norm": 0.7639037370681763,
      "learning_rate": 3.901307125736955e-06,
      "loss": 0.7017,
      "step": 20248
    },
    {
      "epoch": 0.5834437849363222,
      "grad_norm": 0.7592332363128662,
      "learning_rate": 3.90085192491092e-06,
      "loss": 0.7152,
      "step": 20249
    },
    {
      "epoch": 0.5834725983979715,
      "grad_norm": 0.7284128665924072,
      "learning_rate": 3.900396733657366e-06,
      "loss": 0.673,
      "step": 20250
    },
    {
      "epoch": 0.5835014118596208,
      "grad_norm": 1.951973795890808,
      "learning_rate": 3.899941551980258e-06,
      "loss": 0.6967,
      "step": 20251
    },
    {
      "epoch": 0.5835302253212701,
      "grad_norm": 0.7116521000862122,
      "learning_rate": 3.899486379883559e-06,
      "loss": 0.6941,
      "step": 20252
    },
    {
      "epoch": 0.5835590387829194,
      "grad_norm": 0.7279707193374634,
      "learning_rate": 3.899031217371234e-06,
      "loss": 0.7042,
      "step": 20253
    },
    {
      "epoch": 0.5835878522445687,
      "grad_norm": 0.7261725664138794,
      "learning_rate": 3.898576064447245e-06,
      "loss": 0.6971,
      "step": 20254
    },
    {
      "epoch": 0.583616665706218,
      "grad_norm": 0.7641341090202332,
      "learning_rate": 3.898120921115559e-06,
      "loss": 0.714,
      "step": 20255
    },
    {
      "epoch": 0.5836454791678672,
      "grad_norm": 0.7206140756607056,
      "learning_rate": 3.897665787380138e-06,
      "loss": 0.6969,
      "step": 20256
    },
    {
      "epoch": 0.5836742926295165,
      "grad_norm": 1.3717228174209595,
      "learning_rate": 3.897210663244946e-06,
      "loss": 0.7422,
      "step": 20257
    },
    {
      "epoch": 0.5837031060911658,
      "grad_norm": 0.7223513126373291,
      "learning_rate": 3.896755548713946e-06,
      "loss": 0.6952,
      "step": 20258
    },
    {
      "epoch": 0.5837319195528151,
      "grad_norm": 0.7099565863609314,
      "learning_rate": 3.896300443791101e-06,
      "loss": 0.7113,
      "step": 20259
    },
    {
      "epoch": 0.5837607330144644,
      "grad_norm": 0.7148914337158203,
      "learning_rate": 3.8958453484803784e-06,
      "loss": 0.722,
      "step": 20260
    },
    {
      "epoch": 0.5837895464761136,
      "grad_norm": 0.7385704517364502,
      "learning_rate": 3.895390262785738e-06,
      "loss": 0.706,
      "step": 20261
    },
    {
      "epoch": 0.5838183599377629,
      "grad_norm": 0.727340042591095,
      "learning_rate": 3.894935186711145e-06,
      "loss": 0.6995,
      "step": 20262
    },
    {
      "epoch": 0.5838471733994122,
      "grad_norm": 0.7284776568412781,
      "learning_rate": 3.894480120260561e-06,
      "loss": 0.6628,
      "step": 20263
    },
    {
      "epoch": 0.5838759868610615,
      "grad_norm": 0.7363590002059937,
      "learning_rate": 3.89402506343795e-06,
      "loss": 0.7093,
      "step": 20264
    },
    {
      "epoch": 0.5839048003227107,
      "grad_norm": 0.7199879288673401,
      "learning_rate": 3.893570016247276e-06,
      "loss": 0.7043,
      "step": 20265
    },
    {
      "epoch": 0.58393361378436,
      "grad_norm": 0.7509673237800598,
      "learning_rate": 3.8931149786925e-06,
      "loss": 0.726,
      "step": 20266
    },
    {
      "epoch": 0.5839624272460093,
      "grad_norm": 0.730711042881012,
      "learning_rate": 3.892659950777588e-06,
      "loss": 0.7133,
      "step": 20267
    },
    {
      "epoch": 0.5839912407076586,
      "grad_norm": 0.7193305492401123,
      "learning_rate": 3.892204932506499e-06,
      "loss": 0.7253,
      "step": 20268
    },
    {
      "epoch": 0.5840200541693079,
      "grad_norm": 0.7226778268814087,
      "learning_rate": 3.891749923883199e-06,
      "loss": 0.717,
      "step": 20269
    },
    {
      "epoch": 0.5840488676309572,
      "grad_norm": 0.7047404646873474,
      "learning_rate": 3.891294924911649e-06,
      "loss": 0.6749,
      "step": 20270
    },
    {
      "epoch": 0.5840776810926065,
      "grad_norm": 0.7257170081138611,
      "learning_rate": 3.8908399355958095e-06,
      "loss": 0.7091,
      "step": 20271
    },
    {
      "epoch": 0.5841064945542558,
      "grad_norm": 0.7389448881149292,
      "learning_rate": 3.890384955939647e-06,
      "loss": 0.7105,
      "step": 20272
    },
    {
      "epoch": 0.5841353080159051,
      "grad_norm": 0.7171602845191956,
      "learning_rate": 3.889929985947123e-06,
      "loss": 0.7147,
      "step": 20273
    },
    {
      "epoch": 0.5841641214775544,
      "grad_norm": 0.7090232372283936,
      "learning_rate": 3.889475025622199e-06,
      "loss": 0.718,
      "step": 20274
    },
    {
      "epoch": 0.5841929349392035,
      "grad_norm": 0.717026948928833,
      "learning_rate": 3.889020074968838e-06,
      "loss": 0.7283,
      "step": 20275
    },
    {
      "epoch": 0.5842217484008528,
      "grad_norm": 0.7432416677474976,
      "learning_rate": 3.8885651339910006e-06,
      "loss": 0.712,
      "step": 20276
    },
    {
      "epoch": 0.5842505618625021,
      "grad_norm": 0.7192625403404236,
      "learning_rate": 3.888110202692651e-06,
      "loss": 0.7181,
      "step": 20277
    },
    {
      "epoch": 0.5842793753241514,
      "grad_norm": 0.6914308071136475,
      "learning_rate": 3.887655281077749e-06,
      "loss": 0.6989,
      "step": 20278
    },
    {
      "epoch": 0.5843081887858007,
      "grad_norm": 0.7052670121192932,
      "learning_rate": 3.887200369150258e-06,
      "loss": 0.6948,
      "step": 20279
    },
    {
      "epoch": 0.58433700224745,
      "grad_norm": 0.7153542041778564,
      "learning_rate": 3.886745466914139e-06,
      "loss": 0.7069,
      "step": 20280
    },
    {
      "epoch": 0.5843658157090993,
      "grad_norm": 0.7522360682487488,
      "learning_rate": 3.886290574373354e-06,
      "loss": 0.7034,
      "step": 20281
    },
    {
      "epoch": 0.5843946291707486,
      "grad_norm": 0.754163920879364,
      "learning_rate": 3.885835691531865e-06,
      "loss": 0.7095,
      "step": 20282
    },
    {
      "epoch": 0.5844234426323979,
      "grad_norm": 0.7356115579605103,
      "learning_rate": 3.885380818393633e-06,
      "loss": 0.7014,
      "step": 20283
    },
    {
      "epoch": 0.5844522560940472,
      "grad_norm": 0.7062368392944336,
      "learning_rate": 3.88492595496262e-06,
      "loss": 0.7065,
      "step": 20284
    },
    {
      "epoch": 0.5844810695556965,
      "grad_norm": 0.70784592628479,
      "learning_rate": 3.8844711012427875e-06,
      "loss": 0.7168,
      "step": 20285
    },
    {
      "epoch": 0.5845098830173457,
      "grad_norm": 0.7362136840820312,
      "learning_rate": 3.8840162572380955e-06,
      "loss": 0.6778,
      "step": 20286
    },
    {
      "epoch": 0.584538696478995,
      "grad_norm": 0.7573645710945129,
      "learning_rate": 3.8835614229525085e-06,
      "loss": 0.7039,
      "step": 20287
    },
    {
      "epoch": 0.5845675099406442,
      "grad_norm": 0.7429183125495911,
      "learning_rate": 3.883106598389984e-06,
      "loss": 0.7002,
      "step": 20288
    },
    {
      "epoch": 0.5845963234022935,
      "grad_norm": 0.716566801071167,
      "learning_rate": 3.882651783554485e-06,
      "loss": 0.6848,
      "step": 20289
    },
    {
      "epoch": 0.5846251368639428,
      "grad_norm": 0.7340354323387146,
      "learning_rate": 3.882196978449972e-06,
      "loss": 0.7122,
      "step": 20290
    },
    {
      "epoch": 0.5846539503255921,
      "grad_norm": 0.7163376212120056,
      "learning_rate": 3.881742183080407e-06,
      "loss": 0.6896,
      "step": 20291
    },
    {
      "epoch": 0.5846827637872414,
      "grad_norm": 0.7390971779823303,
      "learning_rate": 3.881287397449749e-06,
      "loss": 0.7076,
      "step": 20292
    },
    {
      "epoch": 0.5847115772488907,
      "grad_norm": 0.7231804132461548,
      "learning_rate": 3.880832621561959e-06,
      "loss": 0.7047,
      "step": 20293
    },
    {
      "epoch": 0.58474039071054,
      "grad_norm": 0.7289844155311584,
      "learning_rate": 3.880377855420998e-06,
      "loss": 0.7306,
      "step": 20294
    },
    {
      "epoch": 0.5847692041721892,
      "grad_norm": 0.730832040309906,
      "learning_rate": 3.879923099030825e-06,
      "loss": 0.7296,
      "step": 20295
    },
    {
      "epoch": 0.5847980176338385,
      "grad_norm": 0.7287191152572632,
      "learning_rate": 3.8794683523954046e-06,
      "loss": 0.7092,
      "step": 20296
    },
    {
      "epoch": 0.5848268310954878,
      "grad_norm": 0.7280545830726624,
      "learning_rate": 3.879013615518694e-06,
      "loss": 0.699,
      "step": 20297
    },
    {
      "epoch": 0.5848556445571371,
      "grad_norm": 0.7154130339622498,
      "learning_rate": 3.878558888404655e-06,
      "loss": 0.7145,
      "step": 20298
    },
    {
      "epoch": 0.5848844580187864,
      "grad_norm": 0.7447324395179749,
      "learning_rate": 3.878104171057246e-06,
      "loss": 0.7026,
      "step": 20299
    },
    {
      "epoch": 0.5849132714804357,
      "grad_norm": 0.7216975092887878,
      "learning_rate": 3.87764946348043e-06,
      "loss": 0.6942,
      "step": 20300
    },
    {
      "epoch": 0.584942084942085,
      "grad_norm": 0.7071232199668884,
      "learning_rate": 3.877194765678164e-06,
      "loss": 0.7038,
      "step": 20301
    },
    {
      "epoch": 0.5849708984037342,
      "grad_norm": 0.7488963603973389,
      "learning_rate": 3.876740077654408e-06,
      "loss": 0.6874,
      "step": 20302
    },
    {
      "epoch": 0.5849997118653835,
      "grad_norm": 0.6963945627212524,
      "learning_rate": 3.876285399413124e-06,
      "loss": 0.6978,
      "step": 20303
    },
    {
      "epoch": 0.5850285253270328,
      "grad_norm": 0.7252909541130066,
      "learning_rate": 3.875830730958271e-06,
      "loss": 0.7036,
      "step": 20304
    },
    {
      "epoch": 0.585057338788682,
      "grad_norm": 0.7019846439361572,
      "learning_rate": 3.875376072293809e-06,
      "loss": 0.7052,
      "step": 20305
    },
    {
      "epoch": 0.5850861522503313,
      "grad_norm": 0.710834801197052,
      "learning_rate": 3.874921423423697e-06,
      "loss": 0.6984,
      "step": 20306
    },
    {
      "epoch": 0.5851149657119806,
      "grad_norm": 0.7308875322341919,
      "learning_rate": 3.874466784351893e-06,
      "loss": 0.7065,
      "step": 20307
    },
    {
      "epoch": 0.5851437791736299,
      "grad_norm": 0.6799665093421936,
      "learning_rate": 3.874012155082359e-06,
      "loss": 0.6864,
      "step": 20308
    },
    {
      "epoch": 0.5851725926352792,
      "grad_norm": 0.6944016814231873,
      "learning_rate": 3.873557535619054e-06,
      "loss": 0.7082,
      "step": 20309
    },
    {
      "epoch": 0.5852014060969285,
      "grad_norm": 0.7139101624488831,
      "learning_rate": 3.873102925965938e-06,
      "loss": 0.7206,
      "step": 20310
    },
    {
      "epoch": 0.5852302195585778,
      "grad_norm": 0.7221068143844604,
      "learning_rate": 3.872648326126969e-06,
      "loss": 0.7019,
      "step": 20311
    },
    {
      "epoch": 0.5852590330202271,
      "grad_norm": 0.6992765069007874,
      "learning_rate": 3.872193736106105e-06,
      "loss": 0.6933,
      "step": 20312
    },
    {
      "epoch": 0.5852878464818764,
      "grad_norm": 0.7149155139923096,
      "learning_rate": 3.871739155907307e-06,
      "loss": 0.6822,
      "step": 20313
    },
    {
      "epoch": 0.5853166599435257,
      "grad_norm": 0.7377601861953735,
      "learning_rate": 3.871284585534533e-06,
      "loss": 0.7051,
      "step": 20314
    },
    {
      "epoch": 0.5853454734051748,
      "grad_norm": 0.7099842429161072,
      "learning_rate": 3.870830024991742e-06,
      "loss": 0.6795,
      "step": 20315
    },
    {
      "epoch": 0.5853742868668241,
      "grad_norm": 0.717763364315033,
      "learning_rate": 3.870375474282891e-06,
      "loss": 0.6891,
      "step": 20316
    },
    {
      "epoch": 0.5854031003284734,
      "grad_norm": 0.746256411075592,
      "learning_rate": 3.869920933411942e-06,
      "loss": 0.7265,
      "step": 20317
    },
    {
      "epoch": 0.5854319137901227,
      "grad_norm": 0.71756911277771,
      "learning_rate": 3.869466402382852e-06,
      "loss": 0.6963,
      "step": 20318
    },
    {
      "epoch": 0.585460727251772,
      "grad_norm": 0.6952123641967773,
      "learning_rate": 3.869011881199578e-06,
      "loss": 0.6859,
      "step": 20319
    },
    {
      "epoch": 0.5854895407134213,
      "grad_norm": 0.7191793322563171,
      "learning_rate": 3.8685573698660805e-06,
      "loss": 0.7326,
      "step": 20320
    },
    {
      "epoch": 0.5855183541750706,
      "grad_norm": 0.7706355452537537,
      "learning_rate": 3.868102868386318e-06,
      "loss": 0.7087,
      "step": 20321
    },
    {
      "epoch": 0.5855471676367199,
      "grad_norm": 0.7434828877449036,
      "learning_rate": 3.867648376764248e-06,
      "loss": 0.7326,
      "step": 20322
    },
    {
      "epoch": 0.5855759810983692,
      "grad_norm": 0.7459347248077393,
      "learning_rate": 3.867193895003829e-06,
      "loss": 0.6773,
      "step": 20323
    },
    {
      "epoch": 0.5856047945600185,
      "grad_norm": 0.7004815340042114,
      "learning_rate": 3.8667394231090175e-06,
      "loss": 0.6964,
      "step": 20324
    },
    {
      "epoch": 0.5856336080216678,
      "grad_norm": 0.729836106300354,
      "learning_rate": 3.866284961083774e-06,
      "loss": 0.7096,
      "step": 20325
    },
    {
      "epoch": 0.585662421483317,
      "grad_norm": 0.7528555989265442,
      "learning_rate": 3.865830508932055e-06,
      "loss": 0.7087,
      "step": 20326
    },
    {
      "epoch": 0.5856912349449663,
      "grad_norm": 0.722513735294342,
      "learning_rate": 3.865376066657819e-06,
      "loss": 0.7069,
      "step": 20327
    },
    {
      "epoch": 0.5857200484066156,
      "grad_norm": 0.7351158857345581,
      "learning_rate": 3.864921634265023e-06,
      "loss": 0.7338,
      "step": 20328
    },
    {
      "epoch": 0.5857488618682648,
      "grad_norm": 0.7040007710456848,
      "learning_rate": 3.864467211757624e-06,
      "loss": 0.7244,
      "step": 20329
    },
    {
      "epoch": 0.5857776753299141,
      "grad_norm": 0.7272464632987976,
      "learning_rate": 3.864012799139582e-06,
      "loss": 0.7124,
      "step": 20330
    },
    {
      "epoch": 0.5858064887915634,
      "grad_norm": 0.7345746159553528,
      "learning_rate": 3.86355839641485e-06,
      "loss": 0.714,
      "step": 20331
    },
    {
      "epoch": 0.5858353022532127,
      "grad_norm": 0.7398650050163269,
      "learning_rate": 3.86310400358739e-06,
      "loss": 0.6954,
      "step": 20332
    },
    {
      "epoch": 0.585864115714862,
      "grad_norm": 0.7336695790290833,
      "learning_rate": 3.8626496206611584e-06,
      "loss": 0.7092,
      "step": 20333
    },
    {
      "epoch": 0.5858929291765113,
      "grad_norm": 0.7140284180641174,
      "learning_rate": 3.862195247640111e-06,
      "loss": 0.7098,
      "step": 20334
    },
    {
      "epoch": 0.5859217426381605,
      "grad_norm": 0.7077670693397522,
      "learning_rate": 3.861740884528206e-06,
      "loss": 0.6799,
      "step": 20335
    },
    {
      "epoch": 0.5859505560998098,
      "grad_norm": 0.7505239248275757,
      "learning_rate": 3.8612865313294e-06,
      "loss": 0.7253,
      "step": 20336
    },
    {
      "epoch": 0.5859793695614591,
      "grad_norm": 0.7634007930755615,
      "learning_rate": 3.8608321880476505e-06,
      "loss": 0.7018,
      "step": 20337
    },
    {
      "epoch": 0.5860081830231084,
      "grad_norm": 0.739158570766449,
      "learning_rate": 3.860377854686913e-06,
      "loss": 0.7097,
      "step": 20338
    },
    {
      "epoch": 0.5860369964847577,
      "grad_norm": 0.7130547165870667,
      "learning_rate": 3.859923531251146e-06,
      "loss": 0.7012,
      "step": 20339
    },
    {
      "epoch": 0.586065809946407,
      "grad_norm": 0.7193528413772583,
      "learning_rate": 3.859469217744304e-06,
      "loss": 0.7294,
      "step": 20340
    },
    {
      "epoch": 0.5860946234080563,
      "grad_norm": 0.718906819820404,
      "learning_rate": 3.859014914170346e-06,
      "loss": 0.7118,
      "step": 20341
    },
    {
      "epoch": 0.5861234368697055,
      "grad_norm": 0.7598457932472229,
      "learning_rate": 3.858560620533227e-06,
      "loss": 0.7125,
      "step": 20342
    },
    {
      "epoch": 0.5861522503313548,
      "grad_norm": 0.7499111294746399,
      "learning_rate": 3.858106336836904e-06,
      "loss": 0.6984,
      "step": 20343
    },
    {
      "epoch": 0.586181063793004,
      "grad_norm": 0.7395504117012024,
      "learning_rate": 3.857652063085333e-06,
      "loss": 0.7171,
      "step": 20344
    },
    {
      "epoch": 0.5862098772546533,
      "grad_norm": 0.7094885110855103,
      "learning_rate": 3.857197799282471e-06,
      "loss": 0.6905,
      "step": 20345
    },
    {
      "epoch": 0.5862386907163026,
      "grad_norm": 0.6845084428787231,
      "learning_rate": 3.8567435454322745e-06,
      "loss": 0.6817,
      "step": 20346
    },
    {
      "epoch": 0.5862675041779519,
      "grad_norm": 0.7276617288589478,
      "learning_rate": 3.8562893015386984e-06,
      "loss": 0.7063,
      "step": 20347
    },
    {
      "epoch": 0.5862963176396012,
      "grad_norm": 0.7413918375968933,
      "learning_rate": 3.8558350676056995e-06,
      "loss": 0.7135,
      "step": 20348
    },
    {
      "epoch": 0.5863251311012505,
      "grad_norm": 0.7410320043563843,
      "learning_rate": 3.855380843637233e-06,
      "loss": 0.6935,
      "step": 20349
    },
    {
      "epoch": 0.5863539445628998,
      "grad_norm": 0.7230333089828491,
      "learning_rate": 3.854926629637256e-06,
      "loss": 0.7002,
      "step": 20350
    },
    {
      "epoch": 0.5863827580245491,
      "grad_norm": 0.7297428846359253,
      "learning_rate": 3.854472425609723e-06,
      "loss": 0.7252,
      "step": 20351
    },
    {
      "epoch": 0.5864115714861984,
      "grad_norm": 0.7332834005355835,
      "learning_rate": 3.854018231558589e-06,
      "loss": 0.7077,
      "step": 20352
    },
    {
      "epoch": 0.5864403849478477,
      "grad_norm": 0.7246006727218628,
      "learning_rate": 3.853564047487813e-06,
      "loss": 0.6948,
      "step": 20353
    },
    {
      "epoch": 0.586469198409497,
      "grad_norm": 0.7145023345947266,
      "learning_rate": 3.853109873401346e-06,
      "loss": 0.7017,
      "step": 20354
    },
    {
      "epoch": 0.5864980118711463,
      "grad_norm": 0.6898782849311829,
      "learning_rate": 3.8526557093031435e-06,
      "loss": 0.6767,
      "step": 20355
    },
    {
      "epoch": 0.5865268253327954,
      "grad_norm": 0.7075121998786926,
      "learning_rate": 3.8522015551971655e-06,
      "loss": 0.7128,
      "step": 20356
    },
    {
      "epoch": 0.5865556387944447,
      "grad_norm": 0.713577926158905,
      "learning_rate": 3.8517474110873656e-06,
      "loss": 0.7119,
      "step": 20357
    },
    {
      "epoch": 0.586584452256094,
      "grad_norm": 0.7083842158317566,
      "learning_rate": 3.851293276977697e-06,
      "loss": 0.7118,
      "step": 20358
    },
    {
      "epoch": 0.5866132657177433,
      "grad_norm": 0.7265333533287048,
      "learning_rate": 3.8508391528721155e-06,
      "loss": 0.6937,
      "step": 20359
    },
    {
      "epoch": 0.5866420791793926,
      "grad_norm": 0.7220755219459534,
      "learning_rate": 3.850385038774576e-06,
      "loss": 0.6818,
      "step": 20360
    },
    {
      "epoch": 0.5866708926410419,
      "grad_norm": 0.7273165583610535,
      "learning_rate": 3.849930934689034e-06,
      "loss": 0.7051,
      "step": 20361
    },
    {
      "epoch": 0.5866997061026912,
      "grad_norm": 0.7072898149490356,
      "learning_rate": 3.849476840619443e-06,
      "loss": 0.7003,
      "step": 20362
    },
    {
      "epoch": 0.5867285195643405,
      "grad_norm": 0.7578589916229248,
      "learning_rate": 3.84902275656976e-06,
      "loss": 0.728,
      "step": 20363
    },
    {
      "epoch": 0.5867573330259898,
      "grad_norm": 0.7018632292747498,
      "learning_rate": 3.848568682543937e-06,
      "loss": 0.6867,
      "step": 20364
    },
    {
      "epoch": 0.586786146487639,
      "grad_norm": 0.716083288192749,
      "learning_rate": 3.8481146185459305e-06,
      "loss": 0.7119,
      "step": 20365
    },
    {
      "epoch": 0.5868149599492883,
      "grad_norm": 0.7024412155151367,
      "learning_rate": 3.847660564579695e-06,
      "loss": 0.7031,
      "step": 20366
    },
    {
      "epoch": 0.5868437734109376,
      "grad_norm": 0.7358956933021545,
      "learning_rate": 3.847206520649181e-06,
      "loss": 0.7217,
      "step": 20367
    },
    {
      "epoch": 0.5868725868725869,
      "grad_norm": 0.7229484915733337,
      "learning_rate": 3.846752486758348e-06,
      "loss": 0.674,
      "step": 20368
    },
    {
      "epoch": 0.5869014003342361,
      "grad_norm": 0.7046389579772949,
      "learning_rate": 3.846298462911148e-06,
      "loss": 0.7096,
      "step": 20369
    },
    {
      "epoch": 0.5869302137958854,
      "grad_norm": 0.7083885669708252,
      "learning_rate": 3.845844449111535e-06,
      "loss": 0.6831,
      "step": 20370
    },
    {
      "epoch": 0.5869590272575347,
      "grad_norm": 0.7000373005867004,
      "learning_rate": 3.845390445363464e-06,
      "loss": 0.6885,
      "step": 20371
    },
    {
      "epoch": 0.586987840719184,
      "grad_norm": 0.7302008867263794,
      "learning_rate": 3.844936451670887e-06,
      "loss": 0.7439,
      "step": 20372
    },
    {
      "epoch": 0.5870166541808333,
      "grad_norm": 0.725999653339386,
      "learning_rate": 3.84448246803776e-06,
      "loss": 0.7153,
      "step": 20373
    },
    {
      "epoch": 0.5870454676424826,
      "grad_norm": 0.7271222472190857,
      "learning_rate": 3.844028494468034e-06,
      "loss": 0.7129,
      "step": 20374
    },
    {
      "epoch": 0.5870742811041318,
      "grad_norm": 0.7219181060791016,
      "learning_rate": 3.8435745309656655e-06,
      "loss": 0.7361,
      "step": 20375
    },
    {
      "epoch": 0.5871030945657811,
      "grad_norm": 0.6984508037567139,
      "learning_rate": 3.843120577534607e-06,
      "loss": 0.6917,
      "step": 20376
    },
    {
      "epoch": 0.5871319080274304,
      "grad_norm": 0.7580145001411438,
      "learning_rate": 3.842666634178811e-06,
      "loss": 0.6912,
      "step": 20377
    },
    {
      "epoch": 0.5871607214890797,
      "grad_norm": 0.7389573454856873,
      "learning_rate": 3.842212700902231e-06,
      "loss": 0.7257,
      "step": 20378
    },
    {
      "epoch": 0.587189534950729,
      "grad_norm": 0.7519391775131226,
      "learning_rate": 3.841758777708821e-06,
      "loss": 0.7126,
      "step": 20379
    },
    {
      "epoch": 0.5872183484123783,
      "grad_norm": 0.7049451470375061,
      "learning_rate": 3.841304864602535e-06,
      "loss": 0.6869,
      "step": 20380
    },
    {
      "epoch": 0.5872471618740276,
      "grad_norm": 0.7416338920593262,
      "learning_rate": 3.840850961587326e-06,
      "loss": 0.709,
      "step": 20381
    },
    {
      "epoch": 0.5872759753356769,
      "grad_norm": 0.9078823924064636,
      "learning_rate": 3.840397068667146e-06,
      "loss": 0.7251,
      "step": 20382
    },
    {
      "epoch": 0.5873047887973261,
      "grad_norm": 0.7354185581207275,
      "learning_rate": 3.839943185845947e-06,
      "loss": 0.684,
      "step": 20383
    },
    {
      "epoch": 0.5873336022589754,
      "grad_norm": 0.78809654712677,
      "learning_rate": 3.839489313127684e-06,
      "loss": 0.6935,
      "step": 20384
    },
    {
      "epoch": 0.5873624157206246,
      "grad_norm": 0.7298718690872192,
      "learning_rate": 3.83903545051631e-06,
      "loss": 0.7023,
      "step": 20385
    },
    {
      "epoch": 0.5873912291822739,
      "grad_norm": 0.7405486106872559,
      "learning_rate": 3.838581598015776e-06,
      "loss": 0.7034,
      "step": 20386
    },
    {
      "epoch": 0.5874200426439232,
      "grad_norm": 0.7097988128662109,
      "learning_rate": 3.8381277556300355e-06,
      "loss": 0.6952,
      "step": 20387
    },
    {
      "epoch": 0.5874488561055725,
      "grad_norm": 0.7362288236618042,
      "learning_rate": 3.837673923363041e-06,
      "loss": 0.7155,
      "step": 20388
    },
    {
      "epoch": 0.5874776695672218,
      "grad_norm": 0.7327970266342163,
      "learning_rate": 3.837220101218745e-06,
      "loss": 0.6806,
      "step": 20389
    },
    {
      "epoch": 0.5875064830288711,
      "grad_norm": 0.7520233392715454,
      "learning_rate": 3.8367662892010994e-06,
      "loss": 0.7052,
      "step": 20390
    },
    {
      "epoch": 0.5875352964905204,
      "grad_norm": 0.7359675168991089,
      "learning_rate": 3.836312487314054e-06,
      "loss": 0.7035,
      "step": 20391
    },
    {
      "epoch": 0.5875641099521697,
      "grad_norm": 0.7104225754737854,
      "learning_rate": 3.835858695561566e-06,
      "loss": 0.7186,
      "step": 20392
    },
    {
      "epoch": 0.587592923413819,
      "grad_norm": 0.7206280827522278,
      "learning_rate": 3.835404913947584e-06,
      "loss": 0.7162,
      "step": 20393
    },
    {
      "epoch": 0.5876217368754683,
      "grad_norm": 0.7418248653411865,
      "learning_rate": 3.834951142476063e-06,
      "loss": 0.7096,
      "step": 20394
    },
    {
      "epoch": 0.5876505503371175,
      "grad_norm": 0.763189971446991,
      "learning_rate": 3.8344973811509514e-06,
      "loss": 0.6794,
      "step": 20395
    },
    {
      "epoch": 0.5876793637987667,
      "grad_norm": 0.7284237146377563,
      "learning_rate": 3.834043629976204e-06,
      "loss": 0.7252,
      "step": 20396
    },
    {
      "epoch": 0.587708177260416,
      "grad_norm": 0.7386654615402222,
      "learning_rate": 3.83358988895577e-06,
      "loss": 0.6915,
      "step": 20397
    },
    {
      "epoch": 0.5877369907220653,
      "grad_norm": 0.7334575057029724,
      "learning_rate": 3.833136158093601e-06,
      "loss": 0.6865,
      "step": 20398
    },
    {
      "epoch": 0.5877658041837146,
      "grad_norm": 0.721352219581604,
      "learning_rate": 3.83268243739365e-06,
      "loss": 0.7252,
      "step": 20399
    },
    {
      "epoch": 0.5877946176453639,
      "grad_norm": 0.7661341428756714,
      "learning_rate": 3.832228726859869e-06,
      "loss": 0.7186,
      "step": 20400
    },
    {
      "epoch": 0.5878234311070132,
      "grad_norm": 0.7067715525627136,
      "learning_rate": 3.831775026496208e-06,
      "loss": 0.6918,
      "step": 20401
    },
    {
      "epoch": 0.5878522445686625,
      "grad_norm": 0.7169959545135498,
      "learning_rate": 3.8313213363066195e-06,
      "loss": 0.7057,
      "step": 20402
    },
    {
      "epoch": 0.5878810580303118,
      "grad_norm": 0.6988946795463562,
      "learning_rate": 3.8308676562950525e-06,
      "loss": 0.6919,
      "step": 20403
    },
    {
      "epoch": 0.5879098714919611,
      "grad_norm": 0.7332013249397278,
      "learning_rate": 3.83041398646546e-06,
      "loss": 0.7089,
      "step": 20404
    },
    {
      "epoch": 0.5879386849536103,
      "grad_norm": 0.7119176387786865,
      "learning_rate": 3.829960326821793e-06,
      "loss": 0.6779,
      "step": 20405
    },
    {
      "epoch": 0.5879674984152596,
      "grad_norm": 0.7232277989387512,
      "learning_rate": 3.829506677368002e-06,
      "loss": 0.7015,
      "step": 20406
    },
    {
      "epoch": 0.5879963118769089,
      "grad_norm": 0.716716468334198,
      "learning_rate": 3.829053038108037e-06,
      "loss": 0.7289,
      "step": 20407
    },
    {
      "epoch": 0.5880251253385582,
      "grad_norm": 0.7172746658325195,
      "learning_rate": 3.828599409045851e-06,
      "loss": 0.7084,
      "step": 20408
    },
    {
      "epoch": 0.5880539388002075,
      "grad_norm": 0.7241240739822388,
      "learning_rate": 3.828145790185392e-06,
      "loss": 0.6814,
      "step": 20409
    },
    {
      "epoch": 0.5880827522618567,
      "grad_norm": 0.7125242352485657,
      "learning_rate": 3.827692181530612e-06,
      "loss": 0.7086,
      "step": 20410
    },
    {
      "epoch": 0.588111565723506,
      "grad_norm": 0.7117651700973511,
      "learning_rate": 3.827238583085462e-06,
      "loss": 0.7139,
      "step": 20411
    },
    {
      "epoch": 0.5881403791851553,
      "grad_norm": 0.7274720072746277,
      "learning_rate": 3.8267849948538904e-06,
      "loss": 0.7259,
      "step": 20412
    },
    {
      "epoch": 0.5881691926468046,
      "grad_norm": 0.7024668455123901,
      "learning_rate": 3.82633141683985e-06,
      "loss": 0.7209,
      "step": 20413
    },
    {
      "epoch": 0.5881980061084539,
      "grad_norm": 0.7116650938987732,
      "learning_rate": 3.825877849047289e-06,
      "loss": 0.7139,
      "step": 20414
    },
    {
      "epoch": 0.5882268195701031,
      "grad_norm": 0.7138357758522034,
      "learning_rate": 3.825424291480155e-06,
      "loss": 0.6781,
      "step": 20415
    },
    {
      "epoch": 0.5882556330317524,
      "grad_norm": 0.7363631129264832,
      "learning_rate": 3.824970744142405e-06,
      "loss": 0.7176,
      "step": 20416
    },
    {
      "epoch": 0.5882844464934017,
      "grad_norm": 0.6944588422775269,
      "learning_rate": 3.824517207037985e-06,
      "loss": 0.7011,
      "step": 20417
    },
    {
      "epoch": 0.588313259955051,
      "grad_norm": 0.7138090133666992,
      "learning_rate": 3.824063680170845e-06,
      "loss": 0.7013,
      "step": 20418
    },
    {
      "epoch": 0.5883420734167003,
      "grad_norm": 0.711592972278595,
      "learning_rate": 3.823610163544934e-06,
      "loss": 0.6917,
      "step": 20419
    },
    {
      "epoch": 0.5883708868783496,
      "grad_norm": 0.7079024314880371,
      "learning_rate": 3.823156657164204e-06,
      "loss": 0.7123,
      "step": 20420
    },
    {
      "epoch": 0.5883997003399989,
      "grad_norm": 0.7410203814506531,
      "learning_rate": 3.822703161032601e-06,
      "loss": 0.7493,
      "step": 20421
    },
    {
      "epoch": 0.5884285138016482,
      "grad_norm": 0.7112249135971069,
      "learning_rate": 3.822249675154078e-06,
      "loss": 0.7069,
      "step": 20422
    },
    {
      "epoch": 0.5884573272632974,
      "grad_norm": 0.7283188104629517,
      "learning_rate": 3.821796199532583e-06,
      "loss": 0.7145,
      "step": 20423
    },
    {
      "epoch": 0.5884861407249466,
      "grad_norm": 0.7489566206932068,
      "learning_rate": 3.821342734172065e-06,
      "loss": 0.7196,
      "step": 20424
    },
    {
      "epoch": 0.5885149541865959,
      "grad_norm": 0.7327931523323059,
      "learning_rate": 3.820889279076474e-06,
      "loss": 0.7023,
      "step": 20425
    },
    {
      "epoch": 0.5885437676482452,
      "grad_norm": 0.7552239894866943,
      "learning_rate": 3.820435834249758e-06,
      "loss": 0.7104,
      "step": 20426
    },
    {
      "epoch": 0.5885725811098945,
      "grad_norm": 0.725715696811676,
      "learning_rate": 3.819982399695865e-06,
      "loss": 0.7393,
      "step": 20427
    },
    {
      "epoch": 0.5886013945715438,
      "grad_norm": 0.7146695852279663,
      "learning_rate": 3.819528975418748e-06,
      "loss": 0.7007,
      "step": 20428
    },
    {
      "epoch": 0.5886302080331931,
      "grad_norm": 0.7357956767082214,
      "learning_rate": 3.819075561422354e-06,
      "loss": 0.6965,
      "step": 20429
    },
    {
      "epoch": 0.5886590214948424,
      "grad_norm": 0.7262626886367798,
      "learning_rate": 3.818622157710631e-06,
      "loss": 0.7131,
      "step": 20430
    },
    {
      "epoch": 0.5886878349564917,
      "grad_norm": 0.7103663682937622,
      "learning_rate": 3.818168764287528e-06,
      "loss": 0.7132,
      "step": 20431
    },
    {
      "epoch": 0.588716648418141,
      "grad_norm": 0.7143649458885193,
      "learning_rate": 3.817715381156993e-06,
      "loss": 0.7105,
      "step": 20432
    },
    {
      "epoch": 0.5887454618797903,
      "grad_norm": 0.7711280584335327,
      "learning_rate": 3.817262008322976e-06,
      "loss": 0.7288,
      "step": 20433
    },
    {
      "epoch": 0.5887742753414396,
      "grad_norm": 0.7334055304527283,
      "learning_rate": 3.816808645789425e-06,
      "loss": 0.6912,
      "step": 20434
    },
    {
      "epoch": 0.5888030888030888,
      "grad_norm": 0.7334337830543518,
      "learning_rate": 3.816355293560287e-06,
      "loss": 0.7096,
      "step": 20435
    },
    {
      "epoch": 0.5888319022647381,
      "grad_norm": 0.7501797080039978,
      "learning_rate": 3.815901951639512e-06,
      "loss": 0.7293,
      "step": 20436
    },
    {
      "epoch": 0.5888607157263873,
      "grad_norm": 0.6854690909385681,
      "learning_rate": 3.815448620031045e-06,
      "loss": 0.6994,
      "step": 20437
    },
    {
      "epoch": 0.5888895291880366,
      "grad_norm": 0.7115635275840759,
      "learning_rate": 3.814995298738839e-06,
      "loss": 0.7209,
      "step": 20438
    },
    {
      "epoch": 0.5889183426496859,
      "grad_norm": 0.7088375091552734,
      "learning_rate": 3.814541987766837e-06,
      "loss": 0.6991,
      "step": 20439
    },
    {
      "epoch": 0.5889471561113352,
      "grad_norm": 0.7099027037620544,
      "learning_rate": 3.8140886871189914e-06,
      "loss": 0.7105,
      "step": 20440
    },
    {
      "epoch": 0.5889759695729845,
      "grad_norm": 0.7332227230072021,
      "learning_rate": 3.8136353967992474e-06,
      "loss": 0.6941,
      "step": 20441
    },
    {
      "epoch": 0.5890047830346338,
      "grad_norm": 0.7275317907333374,
      "learning_rate": 3.8131821168115517e-06,
      "loss": 0.7274,
      "step": 20442
    },
    {
      "epoch": 0.5890335964962831,
      "grad_norm": 0.7007375955581665,
      "learning_rate": 3.812728847159855e-06,
      "loss": 0.6875,
      "step": 20443
    },
    {
      "epoch": 0.5890624099579324,
      "grad_norm": 0.7178142666816711,
      "learning_rate": 3.812275587848103e-06,
      "loss": 0.7151,
      "step": 20444
    },
    {
      "epoch": 0.5890912234195816,
      "grad_norm": 0.7177863121032715,
      "learning_rate": 3.8118223388802434e-06,
      "loss": 0.6949,
      "step": 20445
    },
    {
      "epoch": 0.5891200368812309,
      "grad_norm": 0.7101871967315674,
      "learning_rate": 3.811369100260223e-06,
      "loss": 0.6979,
      "step": 20446
    },
    {
      "epoch": 0.5891488503428802,
      "grad_norm": 0.723145604133606,
      "learning_rate": 3.8109158719919902e-06,
      "loss": 0.7158,
      "step": 20447
    },
    {
      "epoch": 0.5891776638045295,
      "grad_norm": 0.7046802043914795,
      "learning_rate": 3.8104626540794915e-06,
      "loss": 0.7065,
      "step": 20448
    },
    {
      "epoch": 0.5892064772661788,
      "grad_norm": 0.7223619222640991,
      "learning_rate": 3.8100094465266733e-06,
      "loss": 0.6996,
      "step": 20449
    },
    {
      "epoch": 0.589235290727828,
      "grad_norm": 0.7283222675323486,
      "learning_rate": 3.8095562493374845e-06,
      "loss": 0.6954,
      "step": 20450
    },
    {
      "epoch": 0.5892641041894773,
      "grad_norm": 0.7157007455825806,
      "learning_rate": 3.8091030625158676e-06,
      "loss": 0.6928,
      "step": 20451
    },
    {
      "epoch": 0.5892929176511266,
      "grad_norm": 0.7104197144508362,
      "learning_rate": 3.808649886065775e-06,
      "loss": 0.7088,
      "step": 20452
    },
    {
      "epoch": 0.5893217311127759,
      "grad_norm": 0.7280068397521973,
      "learning_rate": 3.808196719991151e-06,
      "loss": 0.7004,
      "step": 20453
    },
    {
      "epoch": 0.5893505445744252,
      "grad_norm": 0.7214069962501526,
      "learning_rate": 3.807743564295942e-06,
      "loss": 0.7199,
      "step": 20454
    },
    {
      "epoch": 0.5893793580360744,
      "grad_norm": 0.717689037322998,
      "learning_rate": 3.807290418984095e-06,
      "loss": 0.6744,
      "step": 20455
    },
    {
      "epoch": 0.5894081714977237,
      "grad_norm": 0.7317858934402466,
      "learning_rate": 3.8068372840595556e-06,
      "loss": 0.7077,
      "step": 20456
    },
    {
      "epoch": 0.589436984959373,
      "grad_norm": 0.7702366709709167,
      "learning_rate": 3.8063841595262716e-06,
      "loss": 0.7146,
      "step": 20457
    },
    {
      "epoch": 0.5894657984210223,
      "grad_norm": 0.7376474142074585,
      "learning_rate": 3.805931045388188e-06,
      "loss": 0.7048,
      "step": 20458
    },
    {
      "epoch": 0.5894946118826716,
      "grad_norm": 0.7289241552352905,
      "learning_rate": 3.8054779416492505e-06,
      "loss": 0.694,
      "step": 20459
    },
    {
      "epoch": 0.5895234253443209,
      "grad_norm": 0.746371328830719,
      "learning_rate": 3.8050248483134068e-06,
      "loss": 0.7122,
      "step": 20460
    },
    {
      "epoch": 0.5895522388059702,
      "grad_norm": 0.7057744860649109,
      "learning_rate": 3.804571765384602e-06,
      "loss": 0.7009,
      "step": 20461
    },
    {
      "epoch": 0.5895810522676195,
      "grad_norm": 0.7136983871459961,
      "learning_rate": 3.804118692866782e-06,
      "loss": 0.7034,
      "step": 20462
    },
    {
      "epoch": 0.5896098657292688,
      "grad_norm": 0.7693399786949158,
      "learning_rate": 3.8036656307638926e-06,
      "loss": 0.716,
      "step": 20463
    },
    {
      "epoch": 0.589638679190918,
      "grad_norm": 0.7153896689414978,
      "learning_rate": 3.803212579079879e-06,
      "loss": 0.6942,
      "step": 20464
    },
    {
      "epoch": 0.5896674926525672,
      "grad_norm": 0.7067050933837891,
      "learning_rate": 3.8027595378186886e-06,
      "loss": 0.7099,
      "step": 20465
    },
    {
      "epoch": 0.5896963061142165,
      "grad_norm": 0.7352688908576965,
      "learning_rate": 3.802306506984266e-06,
      "loss": 0.6999,
      "step": 20466
    },
    {
      "epoch": 0.5897251195758658,
      "grad_norm": 0.7016528844833374,
      "learning_rate": 3.8018534865805564e-06,
      "loss": 0.6993,
      "step": 20467
    },
    {
      "epoch": 0.5897539330375151,
      "grad_norm": 0.7278615832328796,
      "learning_rate": 3.8014004766115046e-06,
      "loss": 0.7115,
      "step": 20468
    },
    {
      "epoch": 0.5897827464991644,
      "grad_norm": 0.736565887928009,
      "learning_rate": 3.8009474770810573e-06,
      "loss": 0.7149,
      "step": 20469
    },
    {
      "epoch": 0.5898115599608137,
      "grad_norm": 0.7139707803726196,
      "learning_rate": 3.8004944879931587e-06,
      "loss": 0.697,
      "step": 20470
    },
    {
      "epoch": 0.589840373422463,
      "grad_norm": 0.7016146183013916,
      "learning_rate": 3.800041509351754e-06,
      "loss": 0.6909,
      "step": 20471
    },
    {
      "epoch": 0.5898691868841123,
      "grad_norm": 0.7215824127197266,
      "learning_rate": 3.799588541160788e-06,
      "loss": 0.7141,
      "step": 20472
    },
    {
      "epoch": 0.5898980003457616,
      "grad_norm": 0.7265306115150452,
      "learning_rate": 3.799135583424206e-06,
      "loss": 0.7295,
      "step": 20473
    },
    {
      "epoch": 0.5899268138074109,
      "grad_norm": 0.7042701840400696,
      "learning_rate": 3.798682636145953e-06,
      "loss": 0.7109,
      "step": 20474
    },
    {
      "epoch": 0.5899556272690601,
      "grad_norm": 0.7080634832382202,
      "learning_rate": 3.7982296993299707e-06,
      "loss": 0.7027,
      "step": 20475
    },
    {
      "epoch": 0.5899844407307094,
      "grad_norm": 0.7318976521492004,
      "learning_rate": 3.7977767729802084e-06,
      "loss": 0.7007,
      "step": 20476
    },
    {
      "epoch": 0.5900132541923586,
      "grad_norm": 0.7196399569511414,
      "learning_rate": 3.797323857100609e-06,
      "loss": 0.7073,
      "step": 20477
    },
    {
      "epoch": 0.5900420676540079,
      "grad_norm": 0.7415076494216919,
      "learning_rate": 3.7968709516951165e-06,
      "loss": 0.7078,
      "step": 20478
    },
    {
      "epoch": 0.5900708811156572,
      "grad_norm": 0.7127624750137329,
      "learning_rate": 3.7964180567676757e-06,
      "loss": 0.7108,
      "step": 20479
    },
    {
      "epoch": 0.5900996945773065,
      "grad_norm": 0.7008267045021057,
      "learning_rate": 3.795965172322229e-06,
      "loss": 0.7279,
      "step": 20480
    },
    {
      "epoch": 0.5901285080389558,
      "grad_norm": 0.7042570114135742,
      "learning_rate": 3.795512298362723e-06,
      "loss": 0.6974,
      "step": 20481
    },
    {
      "epoch": 0.5901573215006051,
      "grad_norm": 0.7264920473098755,
      "learning_rate": 3.795059434893101e-06,
      "loss": 0.7226,
      "step": 20482
    },
    {
      "epoch": 0.5901861349622544,
      "grad_norm": 0.709931492805481,
      "learning_rate": 3.7946065819173067e-06,
      "loss": 0.691,
      "step": 20483
    },
    {
      "epoch": 0.5902149484239037,
      "grad_norm": 0.711749792098999,
      "learning_rate": 3.794153739439284e-06,
      "loss": 0.7082,
      "step": 20484
    },
    {
      "epoch": 0.5902437618855529,
      "grad_norm": 0.7132747173309326,
      "learning_rate": 3.793700907462977e-06,
      "loss": 0.7188,
      "step": 20485
    },
    {
      "epoch": 0.5902725753472022,
      "grad_norm": 0.7120835781097412,
      "learning_rate": 3.79324808599233e-06,
      "loss": 0.7107,
      "step": 20486
    },
    {
      "epoch": 0.5903013888088515,
      "grad_norm": 0.736298143863678,
      "learning_rate": 3.7927952750312824e-06,
      "loss": 0.7006,
      "step": 20487
    },
    {
      "epoch": 0.5903302022705008,
      "grad_norm": 0.7318924069404602,
      "learning_rate": 3.7923424745837846e-06,
      "loss": 0.6997,
      "step": 20488
    },
    {
      "epoch": 0.5903590157321501,
      "grad_norm": 0.7099792957305908,
      "learning_rate": 3.7918896846537755e-06,
      "loss": 0.7123,
      "step": 20489
    },
    {
      "epoch": 0.5903878291937994,
      "grad_norm": 0.7108732461929321,
      "learning_rate": 3.7914369052452005e-06,
      "loss": 0.7149,
      "step": 20490
    },
    {
      "epoch": 0.5904166426554486,
      "grad_norm": 0.716224730014801,
      "learning_rate": 3.7909841363620014e-06,
      "loss": 0.712,
      "step": 20491
    },
    {
      "epoch": 0.5904454561170979,
      "grad_norm": 0.7340056300163269,
      "learning_rate": 3.7905313780081223e-06,
      "loss": 0.7024,
      "step": 20492
    },
    {
      "epoch": 0.5904742695787472,
      "grad_norm": 0.7265262007713318,
      "learning_rate": 3.790078630187505e-06,
      "loss": 0.7143,
      "step": 20493
    },
    {
      "epoch": 0.5905030830403964,
      "grad_norm": 0.7238891124725342,
      "learning_rate": 3.789625892904094e-06,
      "loss": 0.7136,
      "step": 20494
    },
    {
      "epoch": 0.5905318965020457,
      "grad_norm": 0.7237240672111511,
      "learning_rate": 3.7891731661618315e-06,
      "loss": 0.7047,
      "step": 20495
    },
    {
      "epoch": 0.590560709963695,
      "grad_norm": 0.7014868855476379,
      "learning_rate": 3.78872044996466e-06,
      "loss": 0.692,
      "step": 20496
    },
    {
      "epoch": 0.5905895234253443,
      "grad_norm": 0.7082638740539551,
      "learning_rate": 3.7882677443165215e-06,
      "loss": 0.688,
      "step": 20497
    },
    {
      "epoch": 0.5906183368869936,
      "grad_norm": 0.7153737545013428,
      "learning_rate": 3.7878150492213617e-06,
      "loss": 0.6896,
      "step": 20498
    },
    {
      "epoch": 0.5906471503486429,
      "grad_norm": 0.7289956212043762,
      "learning_rate": 3.787362364683119e-06,
      "loss": 0.6828,
      "step": 20499
    },
    {
      "epoch": 0.5906759638102922,
      "grad_norm": 0.6945709586143494,
      "learning_rate": 3.7869096907057397e-06,
      "loss": 0.6918,
      "step": 20500
    },
    {
      "epoch": 0.5907047772719415,
      "grad_norm": 0.6998553276062012,
      "learning_rate": 3.7864570272931633e-06,
      "loss": 0.6865,
      "step": 20501
    },
    {
      "epoch": 0.5907335907335908,
      "grad_norm": 0.7193357944488525,
      "learning_rate": 3.786004374449333e-06,
      "loss": 0.7004,
      "step": 20502
    },
    {
      "epoch": 0.5907624041952401,
      "grad_norm": 0.7129896283149719,
      "learning_rate": 3.785551732178192e-06,
      "loss": 0.7251,
      "step": 20503
    },
    {
      "epoch": 0.5907912176568892,
      "grad_norm": 0.7478763461112976,
      "learning_rate": 3.7850991004836813e-06,
      "loss": 0.7211,
      "step": 20504
    },
    {
      "epoch": 0.5908200311185385,
      "grad_norm": 0.7434826493263245,
      "learning_rate": 3.7846464793697425e-06,
      "loss": 0.7069,
      "step": 20505
    },
    {
      "epoch": 0.5908488445801878,
      "grad_norm": 0.7282230854034424,
      "learning_rate": 3.7841938688403184e-06,
      "loss": 0.714,
      "step": 20506
    },
    {
      "epoch": 0.5908776580418371,
      "grad_norm": 0.7061296105384827,
      "learning_rate": 3.7837412688993504e-06,
      "loss": 0.703,
      "step": 20507
    },
    {
      "epoch": 0.5909064715034864,
      "grad_norm": 0.7307397127151489,
      "learning_rate": 3.7832886795507793e-06,
      "loss": 0.6957,
      "step": 20508
    },
    {
      "epoch": 0.5909352849651357,
      "grad_norm": 0.740705132484436,
      "learning_rate": 3.782836100798549e-06,
      "loss": 0.7037,
      "step": 20509
    },
    {
      "epoch": 0.590964098426785,
      "grad_norm": 0.7537950873374939,
      "learning_rate": 3.782383532646596e-06,
      "loss": 0.7104,
      "step": 20510
    },
    {
      "epoch": 0.5909929118884343,
      "grad_norm": 0.7224553227424622,
      "learning_rate": 3.7819309750988686e-06,
      "loss": 0.6953,
      "step": 20511
    },
    {
      "epoch": 0.5910217253500836,
      "grad_norm": 0.7532817721366882,
      "learning_rate": 3.7814784281593043e-06,
      "loss": 0.7319,
      "step": 20512
    },
    {
      "epoch": 0.5910505388117329,
      "grad_norm": 0.7280975580215454,
      "learning_rate": 3.7810258918318443e-06,
      "loss": 0.735,
      "step": 20513
    },
    {
      "epoch": 0.5910793522733822,
      "grad_norm": 0.7191411256790161,
      "learning_rate": 3.7805733661204307e-06,
      "loss": 0.6896,
      "step": 20514
    },
    {
      "epoch": 0.5911081657350314,
      "grad_norm": 0.7345561385154724,
      "learning_rate": 3.7801208510290045e-06,
      "loss": 0.7041,
      "step": 20515
    },
    {
      "epoch": 0.5911369791966807,
      "grad_norm": 0.7261468172073364,
      "learning_rate": 3.779668346561506e-06,
      "loss": 0.7039,
      "step": 20516
    },
    {
      "epoch": 0.59116579265833,
      "grad_norm": 0.7137490510940552,
      "learning_rate": 3.779215852721876e-06,
      "loss": 0.7244,
      "step": 20517
    },
    {
      "epoch": 0.5911946061199792,
      "grad_norm": 0.6938609480857849,
      "learning_rate": 3.7787633695140547e-06,
      "loss": 0.6994,
      "step": 20518
    },
    {
      "epoch": 0.5912234195816285,
      "grad_norm": 0.7395081520080566,
      "learning_rate": 3.7783108969419845e-06,
      "loss": 0.6988,
      "step": 20519
    },
    {
      "epoch": 0.5912522330432778,
      "grad_norm": 0.728611409664154,
      "learning_rate": 3.7778584350096053e-06,
      "loss": 0.7153,
      "step": 20520
    },
    {
      "epoch": 0.5912810465049271,
      "grad_norm": 0.7145503163337708,
      "learning_rate": 3.7774059837208575e-06,
      "loss": 0.6792,
      "step": 20521
    },
    {
      "epoch": 0.5913098599665764,
      "grad_norm": 0.7203702330589294,
      "learning_rate": 3.77695354307968e-06,
      "loss": 0.7077,
      "step": 20522
    },
    {
      "epoch": 0.5913386734282257,
      "grad_norm": 0.7125963568687439,
      "learning_rate": 3.7765011130900153e-06,
      "loss": 0.7062,
      "step": 20523
    },
    {
      "epoch": 0.591367486889875,
      "grad_norm": 0.7181194424629211,
      "learning_rate": 3.776048693755803e-06,
      "loss": 0.6956,
      "step": 20524
    },
    {
      "epoch": 0.5913963003515242,
      "grad_norm": 0.71488356590271,
      "learning_rate": 3.7755962850809843e-06,
      "loss": 0.6933,
      "step": 20525
    },
    {
      "epoch": 0.5914251138131735,
      "grad_norm": 0.7414246201515198,
      "learning_rate": 3.7751438870694967e-06,
      "loss": 0.699,
      "step": 20526
    },
    {
      "epoch": 0.5914539272748228,
      "grad_norm": 0.758746862411499,
      "learning_rate": 3.7746914997252828e-06,
      "loss": 0.7111,
      "step": 20527
    },
    {
      "epoch": 0.5914827407364721,
      "grad_norm": 0.7617975473403931,
      "learning_rate": 3.77423912305228e-06,
      "loss": 0.716,
      "step": 20528
    },
    {
      "epoch": 0.5915115541981214,
      "grad_norm": 0.7145469188690186,
      "learning_rate": 3.7737867570544294e-06,
      "loss": 0.701,
      "step": 20529
    },
    {
      "epoch": 0.5915403676597707,
      "grad_norm": 0.7502279877662659,
      "learning_rate": 3.7733344017356702e-06,
      "loss": 0.6838,
      "step": 20530
    },
    {
      "epoch": 0.5915691811214199,
      "grad_norm": 0.74349045753479,
      "learning_rate": 3.7728820570999426e-06,
      "loss": 0.7018,
      "step": 20531
    },
    {
      "epoch": 0.5915979945830692,
      "grad_norm": 0.7406591773033142,
      "learning_rate": 3.7724297231511852e-06,
      "loss": 0.701,
      "step": 20532
    },
    {
      "epoch": 0.5916268080447185,
      "grad_norm": 0.7117815613746643,
      "learning_rate": 3.771977399893338e-06,
      "loss": 0.7148,
      "step": 20533
    },
    {
      "epoch": 0.5916556215063677,
      "grad_norm": 0.7046873569488525,
      "learning_rate": 3.7715250873303376e-06,
      "loss": 0.6757,
      "step": 20534
    },
    {
      "epoch": 0.591684434968017,
      "grad_norm": 0.7580666542053223,
      "learning_rate": 3.771072785466128e-06,
      "loss": 0.7058,
      "step": 20535
    },
    {
      "epoch": 0.5917132484296663,
      "grad_norm": 0.7349903583526611,
      "learning_rate": 3.7706204943046464e-06,
      "loss": 0.6915,
      "step": 20536
    },
    {
      "epoch": 0.5917420618913156,
      "grad_norm": 0.7280613780021667,
      "learning_rate": 3.770168213849831e-06,
      "loss": 0.7144,
      "step": 20537
    },
    {
      "epoch": 0.5917708753529649,
      "grad_norm": 0.71302330493927,
      "learning_rate": 3.7697159441056207e-06,
      "loss": 0.6812,
      "step": 20538
    },
    {
      "epoch": 0.5917996888146142,
      "grad_norm": 0.7387728095054626,
      "learning_rate": 3.7692636850759555e-06,
      "loss": 0.719,
      "step": 20539
    },
    {
      "epoch": 0.5918285022762635,
      "grad_norm": 0.7108661532402039,
      "learning_rate": 3.7688114367647723e-06,
      "loss": 0.7255,
      "step": 20540
    },
    {
      "epoch": 0.5918573157379128,
      "grad_norm": 0.7662805318832397,
      "learning_rate": 3.768359199176011e-06,
      "loss": 0.7362,
      "step": 20541
    },
    {
      "epoch": 0.5918861291995621,
      "grad_norm": 0.7222380042076111,
      "learning_rate": 3.7679069723136104e-06,
      "loss": 0.714,
      "step": 20542
    },
    {
      "epoch": 0.5919149426612114,
      "grad_norm": 0.7348629832267761,
      "learning_rate": 3.767454756181508e-06,
      "loss": 0.7196,
      "step": 20543
    },
    {
      "epoch": 0.5919437561228607,
      "grad_norm": 0.7112271785736084,
      "learning_rate": 3.767002550783643e-06,
      "loss": 0.6808,
      "step": 20544
    },
    {
      "epoch": 0.5919725695845098,
      "grad_norm": 0.7305935621261597,
      "learning_rate": 3.766550356123953e-06,
      "loss": 0.7035,
      "step": 20545
    },
    {
      "epoch": 0.5920013830461591,
      "grad_norm": 0.7274163961410522,
      "learning_rate": 3.7660981722063745e-06,
      "loss": 0.7,
      "step": 20546
    },
    {
      "epoch": 0.5920301965078084,
      "grad_norm": 0.6977097988128662,
      "learning_rate": 3.765645999034849e-06,
      "loss": 0.6897,
      "step": 20547
    },
    {
      "epoch": 0.5920590099694577,
      "grad_norm": 0.7455203533172607,
      "learning_rate": 3.7651938366133133e-06,
      "loss": 0.7039,
      "step": 20548
    },
    {
      "epoch": 0.592087823431107,
      "grad_norm": 0.7322452068328857,
      "learning_rate": 3.7647416849457052e-06,
      "loss": 0.7151,
      "step": 20549
    },
    {
      "epoch": 0.5921166368927563,
      "grad_norm": 0.7229534983634949,
      "learning_rate": 3.7642895440359624e-06,
      "loss": 0.695,
      "step": 20550
    },
    {
      "epoch": 0.5921454503544056,
      "grad_norm": 0.7759037017822266,
      "learning_rate": 3.7638374138880217e-06,
      "loss": 0.7273,
      "step": 20551
    },
    {
      "epoch": 0.5921742638160549,
      "grad_norm": 0.7069367170333862,
      "learning_rate": 3.763385294505822e-06,
      "loss": 0.6765,
      "step": 20552
    },
    {
      "epoch": 0.5922030772777042,
      "grad_norm": 0.7249887585639954,
      "learning_rate": 3.7629331858933e-06,
      "loss": 0.7106,
      "step": 20553
    },
    {
      "epoch": 0.5922318907393535,
      "grad_norm": 0.7438222169876099,
      "learning_rate": 3.7624810880543935e-06,
      "loss": 0.7001,
      "step": 20554
    },
    {
      "epoch": 0.5922607042010027,
      "grad_norm": 0.7315263152122498,
      "learning_rate": 3.7620290009930393e-06,
      "loss": 0.6851,
      "step": 20555
    },
    {
      "epoch": 0.592289517662652,
      "grad_norm": 0.7166036367416382,
      "learning_rate": 3.761576924713174e-06,
      "loss": 0.7044,
      "step": 20556
    },
    {
      "epoch": 0.5923183311243013,
      "grad_norm": 0.7185754776000977,
      "learning_rate": 3.7611248592187373e-06,
      "loss": 0.7066,
      "step": 20557
    },
    {
      "epoch": 0.5923471445859505,
      "grad_norm": 0.7296732068061829,
      "learning_rate": 3.760672804513663e-06,
      "loss": 0.7024,
      "step": 20558
    },
    {
      "epoch": 0.5923759580475998,
      "grad_norm": 0.7104141712188721,
      "learning_rate": 3.760220760601891e-06,
      "loss": 0.7,
      "step": 20559
    },
    {
      "epoch": 0.5924047715092491,
      "grad_norm": 0.7077128291130066,
      "learning_rate": 3.7597687274873566e-06,
      "loss": 0.7019,
      "step": 20560
    },
    {
      "epoch": 0.5924335849708984,
      "grad_norm": 0.699042797088623,
      "learning_rate": 3.759316705173996e-06,
      "loss": 0.7011,
      "step": 20561
    },
    {
      "epoch": 0.5924623984325477,
      "grad_norm": 0.7302191257476807,
      "learning_rate": 3.758864693665748e-06,
      "loss": 0.7265,
      "step": 20562
    },
    {
      "epoch": 0.592491211894197,
      "grad_norm": 0.7158313393592834,
      "learning_rate": 3.758412692966547e-06,
      "loss": 0.7152,
      "step": 20563
    },
    {
      "epoch": 0.5925200253558462,
      "grad_norm": 0.7261031270027161,
      "learning_rate": 3.757960703080331e-06,
      "loss": 0.7068,
      "step": 20564
    },
    {
      "epoch": 0.5925488388174955,
      "grad_norm": 0.7490659952163696,
      "learning_rate": 3.7575087240110355e-06,
      "loss": 0.7113,
      "step": 20565
    },
    {
      "epoch": 0.5925776522791448,
      "grad_norm": 0.7199265360832214,
      "learning_rate": 3.757056755762597e-06,
      "loss": 0.7013,
      "step": 20566
    },
    {
      "epoch": 0.5926064657407941,
      "grad_norm": 0.6998358964920044,
      "learning_rate": 3.756604798338952e-06,
      "loss": 0.6909,
      "step": 20567
    },
    {
      "epoch": 0.5926352792024434,
      "grad_norm": 0.719555139541626,
      "learning_rate": 3.7561528517440353e-06,
      "loss": 0.7161,
      "step": 20568
    },
    {
      "epoch": 0.5926640926640927,
      "grad_norm": 0.7075812220573425,
      "learning_rate": 3.755700915981785e-06,
      "loss": 0.6949,
      "step": 20569
    },
    {
      "epoch": 0.592692906125742,
      "grad_norm": 0.7207998633384705,
      "learning_rate": 3.755248991056133e-06,
      "loss": 0.7072,
      "step": 20570
    },
    {
      "epoch": 0.5927217195873913,
      "grad_norm": 0.7531525492668152,
      "learning_rate": 3.754797076971021e-06,
      "loss": 0.6969,
      "step": 20571
    },
    {
      "epoch": 0.5927505330490405,
      "grad_norm": 0.7210949659347534,
      "learning_rate": 3.7543451737303805e-06,
      "loss": 0.729,
      "step": 20572
    },
    {
      "epoch": 0.5927793465106898,
      "grad_norm": 0.7108808755874634,
      "learning_rate": 3.753893281338149e-06,
      "loss": 0.697,
      "step": 20573
    },
    {
      "epoch": 0.592808159972339,
      "grad_norm": 0.7199864983558655,
      "learning_rate": 3.753441399798261e-06,
      "loss": 0.6873,
      "step": 20574
    },
    {
      "epoch": 0.5928369734339883,
      "grad_norm": 0.7019858956336975,
      "learning_rate": 3.752989529114653e-06,
      "loss": 0.71,
      "step": 20575
    },
    {
      "epoch": 0.5928657868956376,
      "grad_norm": 0.7399867177009583,
      "learning_rate": 3.752537669291259e-06,
      "loss": 0.7109,
      "step": 20576
    },
    {
      "epoch": 0.5928946003572869,
      "grad_norm": 0.7182762026786804,
      "learning_rate": 3.7520858203320143e-06,
      "loss": 0.7149,
      "step": 20577
    },
    {
      "epoch": 0.5929234138189362,
      "grad_norm": 0.7028853893280029,
      "learning_rate": 3.7516339822408555e-06,
      "loss": 0.713,
      "step": 20578
    },
    {
      "epoch": 0.5929522272805855,
      "grad_norm": 0.7202377319335938,
      "learning_rate": 3.7511821550217166e-06,
      "loss": 0.7231,
      "step": 20579
    },
    {
      "epoch": 0.5929810407422348,
      "grad_norm": 0.7246623635292053,
      "learning_rate": 3.750730338678533e-06,
      "loss": 0.6913,
      "step": 20580
    },
    {
      "epoch": 0.5930098542038841,
      "grad_norm": 0.767572283744812,
      "learning_rate": 3.750278533215239e-06,
      "loss": 0.7237,
      "step": 20581
    },
    {
      "epoch": 0.5930386676655334,
      "grad_norm": 0.7107143402099609,
      "learning_rate": 3.749826738635768e-06,
      "loss": 0.7004,
      "step": 20582
    },
    {
      "epoch": 0.5930674811271827,
      "grad_norm": 0.7021060585975647,
      "learning_rate": 3.7493749549440585e-06,
      "loss": 0.6941,
      "step": 20583
    },
    {
      "epoch": 0.593096294588832,
      "grad_norm": 0.7179740071296692,
      "learning_rate": 3.748923182144043e-06,
      "loss": 0.7065,
      "step": 20584
    },
    {
      "epoch": 0.5931251080504811,
      "grad_norm": 0.7475925087928772,
      "learning_rate": 3.7484714202396555e-06,
      "loss": 0.7129,
      "step": 20585
    },
    {
      "epoch": 0.5931539215121304,
      "grad_norm": 0.7361772656440735,
      "learning_rate": 3.7480196692348315e-06,
      "loss": 0.7225,
      "step": 20586
    },
    {
      "epoch": 0.5931827349737797,
      "grad_norm": 0.7249301075935364,
      "learning_rate": 3.747567929133504e-06,
      "loss": 0.7147,
      "step": 20587
    },
    {
      "epoch": 0.593211548435429,
      "grad_norm": 0.7369803786277771,
      "learning_rate": 3.7471161999396082e-06,
      "loss": 0.7166,
      "step": 20588
    },
    {
      "epoch": 0.5932403618970783,
      "grad_norm": 0.7446512579917908,
      "learning_rate": 3.746664481657078e-06,
      "loss": 0.7137,
      "step": 20589
    },
    {
      "epoch": 0.5932691753587276,
      "grad_norm": 0.7136803865432739,
      "learning_rate": 3.746212774289847e-06,
      "loss": 0.701,
      "step": 20590
    },
    {
      "epoch": 0.5932979888203769,
      "grad_norm": 0.7232342958450317,
      "learning_rate": 3.7457610778418495e-06,
      "loss": 0.7079,
      "step": 20591
    },
    {
      "epoch": 0.5933268022820262,
      "grad_norm": 0.6964899301528931,
      "learning_rate": 3.7453093923170196e-06,
      "loss": 0.7006,
      "step": 20592
    },
    {
      "epoch": 0.5933556157436755,
      "grad_norm": 0.7628815770149231,
      "learning_rate": 3.7448577177192904e-06,
      "loss": 0.7113,
      "step": 20593
    },
    {
      "epoch": 0.5933844292053247,
      "grad_norm": 0.7809932827949524,
      "learning_rate": 3.744406054052594e-06,
      "loss": 0.6678,
      "step": 20594
    },
    {
      "epoch": 0.593413242666974,
      "grad_norm": 0.7548699975013733,
      "learning_rate": 3.7439544013208672e-06,
      "loss": 0.7318,
      "step": 20595
    },
    {
      "epoch": 0.5934420561286233,
      "grad_norm": 0.7098520994186401,
      "learning_rate": 3.7435027595280422e-06,
      "loss": 0.6858,
      "step": 20596
    },
    {
      "epoch": 0.5934708695902726,
      "grad_norm": 0.6786742806434631,
      "learning_rate": 3.7430511286780524e-06,
      "loss": 0.6926,
      "step": 20597
    },
    {
      "epoch": 0.5934996830519219,
      "grad_norm": 0.7250788807868958,
      "learning_rate": 3.74259950877483e-06,
      "loss": 0.7159,
      "step": 20598
    },
    {
      "epoch": 0.5935284965135711,
      "grad_norm": 0.7119684815406799,
      "learning_rate": 3.742147899822309e-06,
      "loss": 0.7052,
      "step": 20599
    },
    {
      "epoch": 0.5935573099752204,
      "grad_norm": 0.7237600684165955,
      "learning_rate": 3.741696301824423e-06,
      "loss": 0.7136,
      "step": 20600
    },
    {
      "epoch": 0.5935861234368697,
      "grad_norm": 0.7100088000297546,
      "learning_rate": 3.741244714785104e-06,
      "loss": 0.6954,
      "step": 20601
    },
    {
      "epoch": 0.593614936898519,
      "grad_norm": 0.7040026783943176,
      "learning_rate": 3.7407931387082853e-06,
      "loss": 0.6738,
      "step": 20602
    },
    {
      "epoch": 0.5936437503601683,
      "grad_norm": 0.7021121382713318,
      "learning_rate": 3.7403415735978997e-06,
      "loss": 0.6959,
      "step": 20603
    },
    {
      "epoch": 0.5936725638218175,
      "grad_norm": 0.730302631855011,
      "learning_rate": 3.73989001945788e-06,
      "loss": 0.6972,
      "step": 20604
    },
    {
      "epoch": 0.5937013772834668,
      "grad_norm": 0.7168562412261963,
      "learning_rate": 3.7394384762921586e-06,
      "loss": 0.7076,
      "step": 20605
    },
    {
      "epoch": 0.5937301907451161,
      "grad_norm": 0.7114185094833374,
      "learning_rate": 3.738986944104666e-06,
      "loss": 0.708,
      "step": 20606
    },
    {
      "epoch": 0.5937590042067654,
      "grad_norm": 0.7278897762298584,
      "learning_rate": 3.7385354228993387e-06,
      "loss": 0.6726,
      "step": 20607
    },
    {
      "epoch": 0.5937878176684147,
      "grad_norm": 0.7387956976890564,
      "learning_rate": 3.738083912680106e-06,
      "loss": 0.6886,
      "step": 20608
    },
    {
      "epoch": 0.593816631130064,
      "grad_norm": 0.7433728575706482,
      "learning_rate": 3.7376324134509025e-06,
      "loss": 0.7266,
      "step": 20609
    },
    {
      "epoch": 0.5938454445917133,
      "grad_norm": 0.6978248357772827,
      "learning_rate": 3.7371809252156577e-06,
      "loss": 0.7016,
      "step": 20610
    },
    {
      "epoch": 0.5938742580533626,
      "grad_norm": 0.6972334980964661,
      "learning_rate": 3.7367294479783043e-06,
      "loss": 0.6908,
      "step": 20611
    },
    {
      "epoch": 0.5939030715150118,
      "grad_norm": 0.71198970079422,
      "learning_rate": 3.7362779817427752e-06,
      "loss": 0.7274,
      "step": 20612
    },
    {
      "epoch": 0.593931884976661,
      "grad_norm": 0.724443793296814,
      "learning_rate": 3.735826526513002e-06,
      "loss": 0.705,
      "step": 20613
    },
    {
      "epoch": 0.5939606984383103,
      "grad_norm": 0.723358690738678,
      "learning_rate": 3.7353750822929157e-06,
      "loss": 0.7244,
      "step": 20614
    },
    {
      "epoch": 0.5939895118999596,
      "grad_norm": 0.6939184069633484,
      "learning_rate": 3.734923649086447e-06,
      "loss": 0.6735,
      "step": 20615
    },
    {
      "epoch": 0.5940183253616089,
      "grad_norm": 0.7224708199501038,
      "learning_rate": 3.73447222689753e-06,
      "loss": 0.7135,
      "step": 20616
    },
    {
      "epoch": 0.5940471388232582,
      "grad_norm": 0.7212996482849121,
      "learning_rate": 3.7340208157300945e-06,
      "loss": 0.697,
      "step": 20617
    },
    {
      "epoch": 0.5940759522849075,
      "grad_norm": 0.7218008637428284,
      "learning_rate": 3.7335694155880715e-06,
      "loss": 0.6874,
      "step": 20618
    },
    {
      "epoch": 0.5941047657465568,
      "grad_norm": 0.7222065329551697,
      "learning_rate": 3.7331180264753937e-06,
      "loss": 0.7198,
      "step": 20619
    },
    {
      "epoch": 0.5941335792082061,
      "grad_norm": 0.722005307674408,
      "learning_rate": 3.7326666483959904e-06,
      "loss": 0.7213,
      "step": 20620
    },
    {
      "epoch": 0.5941623926698554,
      "grad_norm": 0.7324408292770386,
      "learning_rate": 3.732215281353795e-06,
      "loss": 0.7234,
      "step": 20621
    },
    {
      "epoch": 0.5941912061315047,
      "grad_norm": 0.7305825352668762,
      "learning_rate": 3.731763925352737e-06,
      "loss": 0.7111,
      "step": 20622
    },
    {
      "epoch": 0.594220019593154,
      "grad_norm": 0.7389093637466431,
      "learning_rate": 3.7313125803967478e-06,
      "loss": 0.7298,
      "step": 20623
    },
    {
      "epoch": 0.5942488330548032,
      "grad_norm": 0.7269149422645569,
      "learning_rate": 3.7308612464897574e-06,
      "loss": 0.7144,
      "step": 20624
    },
    {
      "epoch": 0.5942776465164525,
      "grad_norm": 0.7173669338226318,
      "learning_rate": 3.7304099236356974e-06,
      "loss": 0.7062,
      "step": 20625
    },
    {
      "epoch": 0.5943064599781017,
      "grad_norm": 0.7138620018959045,
      "learning_rate": 3.7299586118384967e-06,
      "loss": 0.7214,
      "step": 20626
    },
    {
      "epoch": 0.594335273439751,
      "grad_norm": 0.7135277986526489,
      "learning_rate": 3.7295073111020878e-06,
      "loss": 0.691,
      "step": 20627
    },
    {
      "epoch": 0.5943640869014003,
      "grad_norm": 0.7215166091918945,
      "learning_rate": 3.7290560214304008e-06,
      "loss": 0.7027,
      "step": 20628
    },
    {
      "epoch": 0.5943929003630496,
      "grad_norm": 0.7366349101066589,
      "learning_rate": 3.7286047428273642e-06,
      "loss": 0.7156,
      "step": 20629
    },
    {
      "epoch": 0.5944217138246989,
      "grad_norm": 0.7256220579147339,
      "learning_rate": 3.728153475296907e-06,
      "loss": 0.7011,
      "step": 20630
    },
    {
      "epoch": 0.5944505272863482,
      "grad_norm": 0.7329807281494141,
      "learning_rate": 3.7277022188429648e-06,
      "loss": 0.6827,
      "step": 20631
    },
    {
      "epoch": 0.5944793407479975,
      "grad_norm": 0.7358022928237915,
      "learning_rate": 3.727250973469464e-06,
      "loss": 0.7461,
      "step": 20632
    },
    {
      "epoch": 0.5945081542096468,
      "grad_norm": 0.7164691686630249,
      "learning_rate": 3.7267997391803355e-06,
      "loss": 0.7219,
      "step": 20633
    },
    {
      "epoch": 0.594536967671296,
      "grad_norm": 0.729300320148468,
      "learning_rate": 3.7263485159795078e-06,
      "loss": 0.7095,
      "step": 20634
    },
    {
      "epoch": 0.5945657811329453,
      "grad_norm": 0.7117292881011963,
      "learning_rate": 3.725897303870912e-06,
      "loss": 0.7153,
      "step": 20635
    },
    {
      "epoch": 0.5945945945945946,
      "grad_norm": 0.6975612044334412,
      "learning_rate": 3.7254461028584765e-06,
      "loss": 0.7021,
      "step": 20636
    },
    {
      "epoch": 0.5946234080562439,
      "grad_norm": 0.703719973564148,
      "learning_rate": 3.7249949129461304e-06,
      "loss": 0.711,
      "step": 20637
    },
    {
      "epoch": 0.5946522215178932,
      "grad_norm": 0.6870402097702026,
      "learning_rate": 3.7245437341378048e-06,
      "loss": 0.689,
      "step": 20638
    },
    {
      "epoch": 0.5946810349795424,
      "grad_norm": 0.7255417108535767,
      "learning_rate": 3.724092566437429e-06,
      "loss": 0.6977,
      "step": 20639
    },
    {
      "epoch": 0.5947098484411917,
      "grad_norm": 0.7020689249038696,
      "learning_rate": 3.723641409848931e-06,
      "loss": 0.7027,
      "step": 20640
    },
    {
      "epoch": 0.594738661902841,
      "grad_norm": 0.7072681188583374,
      "learning_rate": 3.7231902643762406e-06,
      "loss": 0.6835,
      "step": 20641
    },
    {
      "epoch": 0.5947674753644903,
      "grad_norm": 0.7140173316001892,
      "learning_rate": 3.7227391300232847e-06,
      "loss": 0.7062,
      "step": 20642
    },
    {
      "epoch": 0.5947962888261396,
      "grad_norm": 0.7339028120040894,
      "learning_rate": 3.722288006793996e-06,
      "loss": 0.7134,
      "step": 20643
    },
    {
      "epoch": 0.5948251022877888,
      "grad_norm": 0.7499225735664368,
      "learning_rate": 3.7218368946923015e-06,
      "loss": 0.6917,
      "step": 20644
    },
    {
      "epoch": 0.5948539157494381,
      "grad_norm": 0.7119706273078918,
      "learning_rate": 3.72138579372213e-06,
      "loss": 0.7135,
      "step": 20645
    },
    {
      "epoch": 0.5948827292110874,
      "grad_norm": 0.7096022963523865,
      "learning_rate": 3.7209347038874098e-06,
      "loss": 0.6776,
      "step": 20646
    },
    {
      "epoch": 0.5949115426727367,
      "grad_norm": 0.7148486971855164,
      "learning_rate": 3.7204836251920706e-06,
      "loss": 0.7136,
      "step": 20647
    },
    {
      "epoch": 0.594940356134386,
      "grad_norm": 0.6925159096717834,
      "learning_rate": 3.72003255764004e-06,
      "loss": 0.6954,
      "step": 20648
    },
    {
      "epoch": 0.5949691695960353,
      "grad_norm": 0.7402757406234741,
      "learning_rate": 3.7195815012352452e-06,
      "loss": 0.6872,
      "step": 20649
    },
    {
      "epoch": 0.5949979830576846,
      "grad_norm": 0.7003016471862793,
      "learning_rate": 3.7191304559816165e-06,
      "loss": 0.6875,
      "step": 20650
    },
    {
      "epoch": 0.5950267965193339,
      "grad_norm": 0.6919886469841003,
      "learning_rate": 3.718679421883081e-06,
      "loss": 0.687,
      "step": 20651
    },
    {
      "epoch": 0.5950556099809832,
      "grad_norm": 0.7090640664100647,
      "learning_rate": 3.7182283989435665e-06,
      "loss": 0.7178,
      "step": 20652
    },
    {
      "epoch": 0.5950844234426323,
      "grad_norm": 0.7272199988365173,
      "learning_rate": 3.717777387167e-06,
      "loss": 0.7119,
      "step": 20653
    },
    {
      "epoch": 0.5951132369042816,
      "grad_norm": 0.7309809923171997,
      "learning_rate": 3.717326386557312e-06,
      "loss": 0.7044,
      "step": 20654
    },
    {
      "epoch": 0.5951420503659309,
      "grad_norm": 0.7152271866798401,
      "learning_rate": 3.716875397118429e-06,
      "loss": 0.7081,
      "step": 20655
    },
    {
      "epoch": 0.5951708638275802,
      "grad_norm": 0.6853678822517395,
      "learning_rate": 3.7164244188542784e-06,
      "loss": 0.6869,
      "step": 20656
    },
    {
      "epoch": 0.5951996772892295,
      "grad_norm": 0.7152343988418579,
      "learning_rate": 3.7159734517687883e-06,
      "loss": 0.713,
      "step": 20657
    },
    {
      "epoch": 0.5952284907508788,
      "grad_norm": 0.6924015879631042,
      "learning_rate": 3.715522495865885e-06,
      "loss": 0.7018,
      "step": 20658
    },
    {
      "epoch": 0.5952573042125281,
      "grad_norm": 0.7546364665031433,
      "learning_rate": 3.7150715511494984e-06,
      "loss": 0.7519,
      "step": 20659
    },
    {
      "epoch": 0.5952861176741774,
      "grad_norm": 0.7463259696960449,
      "learning_rate": 3.714620617623553e-06,
      "loss": 0.7061,
      "step": 20660
    },
    {
      "epoch": 0.5953149311358267,
      "grad_norm": 0.7224077582359314,
      "learning_rate": 3.714169695291978e-06,
      "loss": 0.7227,
      "step": 20661
    },
    {
      "epoch": 0.595343744597476,
      "grad_norm": 0.7353902459144592,
      "learning_rate": 3.7137187841586993e-06,
      "loss": 0.6959,
      "step": 20662
    },
    {
      "epoch": 0.5953725580591253,
      "grad_norm": 0.7220001816749573,
      "learning_rate": 3.7132678842276435e-06,
      "loss": 0.723,
      "step": 20663
    },
    {
      "epoch": 0.5954013715207745,
      "grad_norm": 0.7374183535575867,
      "learning_rate": 3.7128169955027392e-06,
      "loss": 0.6967,
      "step": 20664
    },
    {
      "epoch": 0.5954301849824238,
      "grad_norm": 0.6737292408943176,
      "learning_rate": 3.7123661179879115e-06,
      "loss": 0.6656,
      "step": 20665
    },
    {
      "epoch": 0.595458998444073,
      "grad_norm": 0.6896215677261353,
      "learning_rate": 3.711915251687086e-06,
      "loss": 0.6851,
      "step": 20666
    },
    {
      "epoch": 0.5954878119057223,
      "grad_norm": 0.7514870762825012,
      "learning_rate": 3.711464396604193e-06,
      "loss": 0.7071,
      "step": 20667
    },
    {
      "epoch": 0.5955166253673716,
      "grad_norm": 0.7578734755516052,
      "learning_rate": 3.7110135527431572e-06,
      "loss": 0.7119,
      "step": 20668
    },
    {
      "epoch": 0.5955454388290209,
      "grad_norm": 0.7165732979774475,
      "learning_rate": 3.7105627201079048e-06,
      "loss": 0.708,
      "step": 20669
    },
    {
      "epoch": 0.5955742522906702,
      "grad_norm": 0.7349852919578552,
      "learning_rate": 3.7101118987023623e-06,
      "loss": 0.7159,
      "step": 20670
    },
    {
      "epoch": 0.5956030657523195,
      "grad_norm": 0.7445263862609863,
      "learning_rate": 3.7096610885304554e-06,
      "loss": 0.7283,
      "step": 20671
    },
    {
      "epoch": 0.5956318792139688,
      "grad_norm": 0.7460106611251831,
      "learning_rate": 3.7092102895961104e-06,
      "loss": 0.7022,
      "step": 20672
    },
    {
      "epoch": 0.595660692675618,
      "grad_norm": 0.7437129616737366,
      "learning_rate": 3.7087595019032542e-06,
      "loss": 0.725,
      "step": 20673
    },
    {
      "epoch": 0.5956895061372673,
      "grad_norm": 0.7191125750541687,
      "learning_rate": 3.7083087254558116e-06,
      "loss": 0.7107,
      "step": 20674
    },
    {
      "epoch": 0.5957183195989166,
      "grad_norm": 0.7298349142074585,
      "learning_rate": 3.707857960257707e-06,
      "loss": 0.7045,
      "step": 20675
    },
    {
      "epoch": 0.5957471330605659,
      "grad_norm": 0.8408474922180176,
      "learning_rate": 3.7074072063128698e-06,
      "loss": 0.7108,
      "step": 20676
    },
    {
      "epoch": 0.5957759465222152,
      "grad_norm": 0.7160432934761047,
      "learning_rate": 3.706956463625223e-06,
      "loss": 0.7335,
      "step": 20677
    },
    {
      "epoch": 0.5958047599838645,
      "grad_norm": 0.7238327264785767,
      "learning_rate": 3.7065057321986924e-06,
      "loss": 0.7329,
      "step": 20678
    },
    {
      "epoch": 0.5958335734455138,
      "grad_norm": 0.7172996401786804,
      "learning_rate": 3.7060550120372053e-06,
      "loss": 0.6828,
      "step": 20679
    },
    {
      "epoch": 0.595862386907163,
      "grad_norm": 0.7412083745002747,
      "learning_rate": 3.7056043031446836e-06,
      "loss": 0.6973,
      "step": 20680
    },
    {
      "epoch": 0.5958912003688123,
      "grad_norm": 0.7246427536010742,
      "learning_rate": 3.7051536055250557e-06,
      "loss": 0.721,
      "step": 20681
    },
    {
      "epoch": 0.5959200138304616,
      "grad_norm": 0.6976116299629211,
      "learning_rate": 3.704702919182246e-06,
      "loss": 0.6864,
      "step": 20682
    },
    {
      "epoch": 0.5959488272921108,
      "grad_norm": 0.718110978603363,
      "learning_rate": 3.704252244120179e-06,
      "loss": 0.6991,
      "step": 20683
    },
    {
      "epoch": 0.5959776407537601,
      "grad_norm": 0.7079578638076782,
      "learning_rate": 3.7038015803427796e-06,
      "loss": 0.7163,
      "step": 20684
    },
    {
      "epoch": 0.5960064542154094,
      "grad_norm": 0.7338765859603882,
      "learning_rate": 3.7033509278539726e-06,
      "loss": 0.7182,
      "step": 20685
    },
    {
      "epoch": 0.5960352676770587,
      "grad_norm": 0.7072324752807617,
      "learning_rate": 3.7029002866576823e-06,
      "loss": 0.7004,
      "step": 20686
    },
    {
      "epoch": 0.596064081138708,
      "grad_norm": 0.7201652526855469,
      "learning_rate": 3.702449656757835e-06,
      "loss": 0.7036,
      "step": 20687
    },
    {
      "epoch": 0.5960928946003573,
      "grad_norm": 0.7031732201576233,
      "learning_rate": 3.7019990381583532e-06,
      "loss": 0.6905,
      "step": 20688
    },
    {
      "epoch": 0.5961217080620066,
      "grad_norm": 0.7500439286231995,
      "learning_rate": 3.7015484308631623e-06,
      "loss": 0.6833,
      "step": 20689
    },
    {
      "epoch": 0.5961505215236559,
      "grad_norm": 0.7099084854125977,
      "learning_rate": 3.701097834876185e-06,
      "loss": 0.6952,
      "step": 20690
    },
    {
      "epoch": 0.5961793349853052,
      "grad_norm": 0.719283401966095,
      "learning_rate": 3.700647250201349e-06,
      "loss": 0.701,
      "step": 20691
    },
    {
      "epoch": 0.5962081484469545,
      "grad_norm": 0.7215315699577332,
      "learning_rate": 3.7001966768425767e-06,
      "loss": 0.7094,
      "step": 20692
    },
    {
      "epoch": 0.5962369619086036,
      "grad_norm": 0.7072629332542419,
      "learning_rate": 3.6997461148037916e-06,
      "loss": 0.696,
      "step": 20693
    },
    {
      "epoch": 0.5962657753702529,
      "grad_norm": 0.6986560225486755,
      "learning_rate": 3.6992955640889184e-06,
      "loss": 0.709,
      "step": 20694
    },
    {
      "epoch": 0.5962945888319022,
      "grad_norm": 0.7127872705459595,
      "learning_rate": 3.69884502470188e-06,
      "loss": 0.717,
      "step": 20695
    },
    {
      "epoch": 0.5963234022935515,
      "grad_norm": 0.7397634983062744,
      "learning_rate": 3.698394496646601e-06,
      "loss": 0.7071,
      "step": 20696
    },
    {
      "epoch": 0.5963522157552008,
      "grad_norm": 0.7714131474494934,
      "learning_rate": 3.697943979927005e-06,
      "loss": 0.7107,
      "step": 20697
    },
    {
      "epoch": 0.5963810292168501,
      "grad_norm": 0.723635733127594,
      "learning_rate": 3.697493474547016e-06,
      "loss": 0.6986,
      "step": 20698
    },
    {
      "epoch": 0.5964098426784994,
      "grad_norm": 0.7383568286895752,
      "learning_rate": 3.697042980510556e-06,
      "loss": 0.7078,
      "step": 20699
    },
    {
      "epoch": 0.5964386561401487,
      "grad_norm": 0.7758994698524475,
      "learning_rate": 3.6965924978215496e-06,
      "loss": 0.7069,
      "step": 20700
    },
    {
      "epoch": 0.596467469601798,
      "grad_norm": 0.7233620285987854,
      "learning_rate": 3.6961420264839187e-06,
      "loss": 0.7177,
      "step": 20701
    },
    {
      "epoch": 0.5964962830634473,
      "grad_norm": 0.7098838686943054,
      "learning_rate": 3.6956915665015864e-06,
      "loss": 0.7133,
      "step": 20702
    },
    {
      "epoch": 0.5965250965250966,
      "grad_norm": 0.7261576056480408,
      "learning_rate": 3.695241117878479e-06,
      "loss": 0.7174,
      "step": 20703
    },
    {
      "epoch": 0.5965539099867458,
      "grad_norm": 0.7339076399803162,
      "learning_rate": 3.6947906806185164e-06,
      "loss": 0.7162,
      "step": 20704
    },
    {
      "epoch": 0.5965827234483951,
      "grad_norm": 0.7293874025344849,
      "learning_rate": 3.6943402547256223e-06,
      "loss": 0.6883,
      "step": 20705
    },
    {
      "epoch": 0.5966115369100444,
      "grad_norm": 0.7557618021965027,
      "learning_rate": 3.6938898402037194e-06,
      "loss": 0.7242,
      "step": 20706
    },
    {
      "epoch": 0.5966403503716936,
      "grad_norm": 0.7163515686988831,
      "learning_rate": 3.6934394370567308e-06,
      "loss": 0.7311,
      "step": 20707
    },
    {
      "epoch": 0.5966691638333429,
      "grad_norm": 0.723486065864563,
      "learning_rate": 3.692989045288578e-06,
      "loss": 0.6948,
      "step": 20708
    },
    {
      "epoch": 0.5966979772949922,
      "grad_norm": 0.7167050838470459,
      "learning_rate": 3.6925386649031847e-06,
      "loss": 0.7204,
      "step": 20709
    },
    {
      "epoch": 0.5967267907566415,
      "grad_norm": 0.7053078413009644,
      "learning_rate": 3.6920882959044726e-06,
      "loss": 0.7132,
      "step": 20710
    },
    {
      "epoch": 0.5967556042182908,
      "grad_norm": 0.7210693955421448,
      "learning_rate": 3.6916379382963634e-06,
      "loss": 0.7236,
      "step": 20711
    },
    {
      "epoch": 0.5967844176799401,
      "grad_norm": 0.7350145578384399,
      "learning_rate": 3.69118759208278e-06,
      "loss": 0.692,
      "step": 20712
    },
    {
      "epoch": 0.5968132311415894,
      "grad_norm": 0.7171065807342529,
      "learning_rate": 3.690737257267644e-06,
      "loss": 0.6954,
      "step": 20713
    },
    {
      "epoch": 0.5968420446032386,
      "grad_norm": 0.698544442653656,
      "learning_rate": 3.6902869338548773e-06,
      "loss": 0.7092,
      "step": 20714
    },
    {
      "epoch": 0.5968708580648879,
      "grad_norm": 0.7385860681533813,
      "learning_rate": 3.6898366218484026e-06,
      "loss": 0.7143,
      "step": 20715
    },
    {
      "epoch": 0.5968996715265372,
      "grad_norm": 0.7334088683128357,
      "learning_rate": 3.689386321252142e-06,
      "loss": 0.6858,
      "step": 20716
    },
    {
      "epoch": 0.5969284849881865,
      "grad_norm": 0.7237518429756165,
      "learning_rate": 3.688936032070015e-06,
      "loss": 0.7246,
      "step": 20717
    },
    {
      "epoch": 0.5969572984498358,
      "grad_norm": 0.7102307081222534,
      "learning_rate": 3.688485754305946e-06,
      "loss": 0.6956,
      "step": 20718
    },
    {
      "epoch": 0.5969861119114851,
      "grad_norm": 0.7002034187316895,
      "learning_rate": 3.6880354879638546e-06,
      "loss": 0.6891,
      "step": 20719
    },
    {
      "epoch": 0.5970149253731343,
      "grad_norm": 0.7046633362770081,
      "learning_rate": 3.6875852330476636e-06,
      "loss": 0.7058,
      "step": 20720
    },
    {
      "epoch": 0.5970437388347836,
      "grad_norm": 0.723811149597168,
      "learning_rate": 3.6871349895612923e-06,
      "loss": 0.7096,
      "step": 20721
    },
    {
      "epoch": 0.5970725522964329,
      "grad_norm": 0.7362416386604309,
      "learning_rate": 3.6866847575086626e-06,
      "loss": 0.731,
      "step": 20722
    },
    {
      "epoch": 0.5971013657580821,
      "grad_norm": 0.7283684611320496,
      "learning_rate": 3.6862345368936967e-06,
      "loss": 0.6986,
      "step": 20723
    },
    {
      "epoch": 0.5971301792197314,
      "grad_norm": 0.7322108745574951,
      "learning_rate": 3.685784327720314e-06,
      "loss": 0.7183,
      "step": 20724
    },
    {
      "epoch": 0.5971589926813807,
      "grad_norm": 0.7045606970787048,
      "learning_rate": 3.6853341299924373e-06,
      "loss": 0.7022,
      "step": 20725
    },
    {
      "epoch": 0.59718780614303,
      "grad_norm": 0.7119572162628174,
      "learning_rate": 3.6848839437139827e-06,
      "loss": 0.7041,
      "step": 20726
    },
    {
      "epoch": 0.5972166196046793,
      "grad_norm": 0.7246286273002625,
      "learning_rate": 3.6844337688888763e-06,
      "loss": 0.7008,
      "step": 20727
    },
    {
      "epoch": 0.5972454330663286,
      "grad_norm": 0.7083394527435303,
      "learning_rate": 3.6839836055210374e-06,
      "loss": 0.6861,
      "step": 20728
    },
    {
      "epoch": 0.5972742465279779,
      "grad_norm": 0.7113374471664429,
      "learning_rate": 3.683533453614386e-06,
      "loss": 0.6928,
      "step": 20729
    },
    {
      "epoch": 0.5973030599896272,
      "grad_norm": 0.7244629263877869,
      "learning_rate": 3.6830833131728416e-06,
      "loss": 0.7293,
      "step": 20730
    },
    {
      "epoch": 0.5973318734512765,
      "grad_norm": 0.7147168517112732,
      "learning_rate": 3.682633184200326e-06,
      "loss": 0.7151,
      "step": 20731
    },
    {
      "epoch": 0.5973606869129258,
      "grad_norm": 0.7024534940719604,
      "learning_rate": 3.6821830667007574e-06,
      "loss": 0.7177,
      "step": 20732
    },
    {
      "epoch": 0.597389500374575,
      "grad_norm": 0.7005594968795776,
      "learning_rate": 3.681732960678057e-06,
      "loss": 0.6683,
      "step": 20733
    },
    {
      "epoch": 0.5974183138362242,
      "grad_norm": 0.7077967524528503,
      "learning_rate": 3.6812828661361446e-06,
      "loss": 0.7099,
      "step": 20734
    },
    {
      "epoch": 0.5974471272978735,
      "grad_norm": 0.70991450548172,
      "learning_rate": 3.6808327830789405e-06,
      "loss": 0.7087,
      "step": 20735
    },
    {
      "epoch": 0.5974759407595228,
      "grad_norm": 0.7190231680870056,
      "learning_rate": 3.6803827115103645e-06,
      "loss": 0.6929,
      "step": 20736
    },
    {
      "epoch": 0.5975047542211721,
      "grad_norm": 0.7143057584762573,
      "learning_rate": 3.6799326514343357e-06,
      "loss": 0.6926,
      "step": 20737
    },
    {
      "epoch": 0.5975335676828214,
      "grad_norm": 0.7360687851905823,
      "learning_rate": 3.679482602854773e-06,
      "loss": 0.7347,
      "step": 20738
    },
    {
      "epoch": 0.5975623811444707,
      "grad_norm": 0.6975166201591492,
      "learning_rate": 3.6790325657755973e-06,
      "loss": 0.708,
      "step": 20739
    },
    {
      "epoch": 0.59759119460612,
      "grad_norm": 0.7204293012619019,
      "learning_rate": 3.678582540200728e-06,
      "loss": 0.6855,
      "step": 20740
    },
    {
      "epoch": 0.5976200080677693,
      "grad_norm": 0.7319963574409485,
      "learning_rate": 3.678132526134084e-06,
      "loss": 0.7187,
      "step": 20741
    },
    {
      "epoch": 0.5976488215294186,
      "grad_norm": 0.7219365239143372,
      "learning_rate": 3.6776825235795838e-06,
      "loss": 0.7,
      "step": 20742
    },
    {
      "epoch": 0.5976776349910679,
      "grad_norm": 0.773269534111023,
      "learning_rate": 3.6772325325411473e-06,
      "loss": 0.7113,
      "step": 20743
    },
    {
      "epoch": 0.5977064484527171,
      "grad_norm": 0.7418818473815918,
      "learning_rate": 3.676782553022693e-06,
      "loss": 0.7104,
      "step": 20744
    },
    {
      "epoch": 0.5977352619143664,
      "grad_norm": 0.7214593887329102,
      "learning_rate": 3.6763325850281408e-06,
      "loss": 0.7014,
      "step": 20745
    },
    {
      "epoch": 0.5977640753760157,
      "grad_norm": 0.7155462503433228,
      "learning_rate": 3.675882628561408e-06,
      "loss": 0.6893,
      "step": 20746
    },
    {
      "epoch": 0.5977928888376649,
      "grad_norm": 0.702256977558136,
      "learning_rate": 3.675432683626413e-06,
      "loss": 0.7058,
      "step": 20747
    },
    {
      "epoch": 0.5978217022993142,
      "grad_norm": 0.7258961200714111,
      "learning_rate": 3.6749827502270763e-06,
      "loss": 0.7102,
      "step": 20748
    },
    {
      "epoch": 0.5978505157609635,
      "grad_norm": 0.7818047404289246,
      "learning_rate": 3.674532828367315e-06,
      "loss": 0.7035,
      "step": 20749
    },
    {
      "epoch": 0.5978793292226128,
      "grad_norm": 0.7503392696380615,
      "learning_rate": 3.674082918051046e-06,
      "loss": 0.723,
      "step": 20750
    },
    {
      "epoch": 0.5979081426842621,
      "grad_norm": 0.7097684144973755,
      "learning_rate": 3.673633019282191e-06,
      "loss": 0.6954,
      "step": 20751
    },
    {
      "epoch": 0.5979369561459114,
      "grad_norm": 0.6962099075317383,
      "learning_rate": 3.673183132064667e-06,
      "loss": 0.6973,
      "step": 20752
    },
    {
      "epoch": 0.5979657696075606,
      "grad_norm": 0.7649878263473511,
      "learning_rate": 3.672733256402391e-06,
      "loss": 0.703,
      "step": 20753
    },
    {
      "epoch": 0.5979945830692099,
      "grad_norm": 0.7267587184906006,
      "learning_rate": 3.6722833922992824e-06,
      "loss": 0.6861,
      "step": 20754
    },
    {
      "epoch": 0.5980233965308592,
      "grad_norm": 0.7389863729476929,
      "learning_rate": 3.6718335397592566e-06,
      "loss": 0.6889,
      "step": 20755
    },
    {
      "epoch": 0.5980522099925085,
      "grad_norm": 0.7008274793624878,
      "learning_rate": 3.671383698786234e-06,
      "loss": 0.6822,
      "step": 20756
    },
    {
      "epoch": 0.5980810234541578,
      "grad_norm": 0.7514971494674683,
      "learning_rate": 3.6709338693841314e-06,
      "loss": 0.7229,
      "step": 20757
    },
    {
      "epoch": 0.5981098369158071,
      "grad_norm": 0.7343454360961914,
      "learning_rate": 3.6704840515568663e-06,
      "loss": 0.6818,
      "step": 20758
    },
    {
      "epoch": 0.5981386503774564,
      "grad_norm": 0.7371471524238586,
      "learning_rate": 3.670034245308356e-06,
      "loss": 0.6899,
      "step": 20759
    },
    {
      "epoch": 0.5981674638391057,
      "grad_norm": 0.7759815454483032,
      "learning_rate": 3.669584450642518e-06,
      "loss": 0.7115,
      "step": 20760
    },
    {
      "epoch": 0.5981962773007549,
      "grad_norm": 0.7156925201416016,
      "learning_rate": 3.6691346675632687e-06,
      "loss": 0.6873,
      "step": 20761
    },
    {
      "epoch": 0.5982250907624042,
      "grad_norm": 0.7264539003372192,
      "learning_rate": 3.668684896074525e-06,
      "loss": 0.7062,
      "step": 20762
    },
    {
      "epoch": 0.5982539042240534,
      "grad_norm": 0.7048860788345337,
      "learning_rate": 3.668235136180207e-06,
      "loss": 0.7299,
      "step": 20763
    },
    {
      "epoch": 0.5982827176857027,
      "grad_norm": 0.7601941823959351,
      "learning_rate": 3.6677853878842294e-06,
      "loss": 0.7116,
      "step": 20764
    },
    {
      "epoch": 0.598311531147352,
      "grad_norm": 0.7210016846656799,
      "learning_rate": 3.667335651190509e-06,
      "loss": 0.6997,
      "step": 20765
    },
    {
      "epoch": 0.5983403446090013,
      "grad_norm": 0.7556668519973755,
      "learning_rate": 3.666885926102963e-06,
      "loss": 0.7006,
      "step": 20766
    },
    {
      "epoch": 0.5983691580706506,
      "grad_norm": 0.7370720505714417,
      "learning_rate": 3.6664362126255087e-06,
      "loss": 0.7151,
      "step": 20767
    },
    {
      "epoch": 0.5983979715322999,
      "grad_norm": 0.7048178911209106,
      "learning_rate": 3.6659865107620617e-06,
      "loss": 0.67,
      "step": 20768
    },
    {
      "epoch": 0.5984267849939492,
      "grad_norm": 0.7540643811225891,
      "learning_rate": 3.6655368205165385e-06,
      "loss": 0.7194,
      "step": 20769
    },
    {
      "epoch": 0.5984555984555985,
      "grad_norm": 0.7488542199134827,
      "learning_rate": 3.665087141892856e-06,
      "loss": 0.7204,
      "step": 20770
    },
    {
      "epoch": 0.5984844119172478,
      "grad_norm": 0.730512797832489,
      "learning_rate": 3.664637474894929e-06,
      "loss": 0.7021,
      "step": 20771
    },
    {
      "epoch": 0.5985132253788971,
      "grad_norm": 0.7266911864280701,
      "learning_rate": 3.6641878195266745e-06,
      "loss": 0.7096,
      "step": 20772
    },
    {
      "epoch": 0.5985420388405464,
      "grad_norm": 0.72114497423172,
      "learning_rate": 3.6637381757920087e-06,
      "loss": 0.7012,
      "step": 20773
    },
    {
      "epoch": 0.5985708523021955,
      "grad_norm": 0.7219867706298828,
      "learning_rate": 3.6632885436948484e-06,
      "loss": 0.7151,
      "step": 20774
    },
    {
      "epoch": 0.5985996657638448,
      "grad_norm": 0.7437293529510498,
      "learning_rate": 3.6628389232391087e-06,
      "loss": 0.7427,
      "step": 20775
    },
    {
      "epoch": 0.5986284792254941,
      "grad_norm": 0.7021557688713074,
      "learning_rate": 3.6623893144287053e-06,
      "loss": 0.6768,
      "step": 20776
    },
    {
      "epoch": 0.5986572926871434,
      "grad_norm": 0.7470837235450745,
      "learning_rate": 3.661939717267553e-06,
      "loss": 0.7074,
      "step": 20777
    },
    {
      "epoch": 0.5986861061487927,
      "grad_norm": 0.706393301486969,
      "learning_rate": 3.6614901317595695e-06,
      "loss": 0.7044,
      "step": 20778
    },
    {
      "epoch": 0.598714919610442,
      "grad_norm": 0.7243131995201111,
      "learning_rate": 3.6610405579086684e-06,
      "loss": 0.7101,
      "step": 20779
    },
    {
      "epoch": 0.5987437330720913,
      "grad_norm": 0.7199962139129639,
      "learning_rate": 3.6605909957187657e-06,
      "loss": 0.6777,
      "step": 20780
    },
    {
      "epoch": 0.5987725465337406,
      "grad_norm": 0.7378746271133423,
      "learning_rate": 3.660141445193777e-06,
      "loss": 0.7098,
      "step": 20781
    },
    {
      "epoch": 0.5988013599953899,
      "grad_norm": 0.7115344405174255,
      "learning_rate": 3.6596919063376163e-06,
      "loss": 0.697,
      "step": 20782
    },
    {
      "epoch": 0.5988301734570391,
      "grad_norm": 0.7222005128860474,
      "learning_rate": 3.659242379154199e-06,
      "loss": 0.7032,
      "step": 20783
    },
    {
      "epoch": 0.5988589869186884,
      "grad_norm": 0.7050880789756775,
      "learning_rate": 3.658792863647441e-06,
      "loss": 0.7118,
      "step": 20784
    },
    {
      "epoch": 0.5988878003803377,
      "grad_norm": 0.7174785733222961,
      "learning_rate": 3.6583433598212544e-06,
      "loss": 0.6903,
      "step": 20785
    },
    {
      "epoch": 0.598916613841987,
      "grad_norm": 0.7221607565879822,
      "learning_rate": 3.6578938676795576e-06,
      "loss": 0.6899,
      "step": 20786
    },
    {
      "epoch": 0.5989454273036363,
      "grad_norm": 0.7221559882164001,
      "learning_rate": 3.6574443872262638e-06,
      "loss": 0.6922,
      "step": 20787
    },
    {
      "epoch": 0.5989742407652855,
      "grad_norm": 0.7189710736274719,
      "learning_rate": 3.6569949184652874e-06,
      "loss": 0.6672,
      "step": 20788
    },
    {
      "epoch": 0.5990030542269348,
      "grad_norm": 0.7367322444915771,
      "learning_rate": 3.6565454614005425e-06,
      "loss": 0.7111,
      "step": 20789
    },
    {
      "epoch": 0.5990318676885841,
      "grad_norm": 0.7450379133224487,
      "learning_rate": 3.656096016035944e-06,
      "loss": 0.6971,
      "step": 20790
    },
    {
      "epoch": 0.5990606811502334,
      "grad_norm": 0.7260771989822388,
      "learning_rate": 3.6556465823754063e-06,
      "loss": 0.6951,
      "step": 20791
    },
    {
      "epoch": 0.5990894946118827,
      "grad_norm": 0.7429428100585938,
      "learning_rate": 3.655197160422842e-06,
      "loss": 0.7085,
      "step": 20792
    },
    {
      "epoch": 0.599118308073532,
      "grad_norm": 0.7429869174957275,
      "learning_rate": 3.6547477501821647e-06,
      "loss": 0.6952,
      "step": 20793
    },
    {
      "epoch": 0.5991471215351812,
      "grad_norm": 0.7217182517051697,
      "learning_rate": 3.6542983516572916e-06,
      "loss": 0.7088,
      "step": 20794
    },
    {
      "epoch": 0.5991759349968305,
      "grad_norm": 0.7298756241798401,
      "learning_rate": 3.653848964852135e-06,
      "loss": 0.7024,
      "step": 20795
    },
    {
      "epoch": 0.5992047484584798,
      "grad_norm": 0.7442299723625183,
      "learning_rate": 3.6533995897706074e-06,
      "loss": 0.7088,
      "step": 20796
    },
    {
      "epoch": 0.5992335619201291,
      "grad_norm": 0.7099366188049316,
      "learning_rate": 3.652950226416623e-06,
      "loss": 0.6937,
      "step": 20797
    },
    {
      "epoch": 0.5992623753817784,
      "grad_norm": 0.7163823246955872,
      "learning_rate": 3.652500874794095e-06,
      "loss": 0.6963,
      "step": 20798
    },
    {
      "epoch": 0.5992911888434277,
      "grad_norm": 0.6979102492332458,
      "learning_rate": 3.6520515349069384e-06,
      "loss": 0.6901,
      "step": 20799
    },
    {
      "epoch": 0.599320002305077,
      "grad_norm": 0.7067794799804688,
      "learning_rate": 3.651602206759066e-06,
      "loss": 0.694,
      "step": 20800
    },
    {
      "epoch": 0.5993488157667262,
      "grad_norm": 0.7120316624641418,
      "learning_rate": 3.6511528903543904e-06,
      "loss": 0.7125,
      "step": 20801
    },
    {
      "epoch": 0.5993776292283755,
      "grad_norm": 0.7332307696342468,
      "learning_rate": 3.650703585696824e-06,
      "loss": 0.7213,
      "step": 20802
    },
    {
      "epoch": 0.5994064426900247,
      "grad_norm": 0.7672549486160278,
      "learning_rate": 3.650254292790282e-06,
      "loss": 0.7113,
      "step": 20803
    },
    {
      "epoch": 0.599435256151674,
      "grad_norm": 0.6983882188796997,
      "learning_rate": 3.6498050116386745e-06,
      "loss": 0.6821,
      "step": 20804
    },
    {
      "epoch": 0.5994640696133233,
      "grad_norm": 0.723326563835144,
      "learning_rate": 3.649355742245916e-06,
      "loss": 0.7027,
      "step": 20805
    },
    {
      "epoch": 0.5994928830749726,
      "grad_norm": 0.7175735235214233,
      "learning_rate": 3.6489064846159193e-06,
      "loss": 0.7139,
      "step": 20806
    },
    {
      "epoch": 0.5995216965366219,
      "grad_norm": 0.7277495265007019,
      "learning_rate": 3.6484572387525964e-06,
      "loss": 0.6898,
      "step": 20807
    },
    {
      "epoch": 0.5995505099982712,
      "grad_norm": 0.7215704321861267,
      "learning_rate": 3.6480080046598593e-06,
      "loss": 0.701,
      "step": 20808
    },
    {
      "epoch": 0.5995793234599205,
      "grad_norm": 0.7287200689315796,
      "learning_rate": 3.647558782341619e-06,
      "loss": 0.7225,
      "step": 20809
    },
    {
      "epoch": 0.5996081369215698,
      "grad_norm": 0.6987705826759338,
      "learning_rate": 3.6471095718017924e-06,
      "loss": 0.7124,
      "step": 20810
    },
    {
      "epoch": 0.5996369503832191,
      "grad_norm": 0.7170248031616211,
      "learning_rate": 3.6466603730442883e-06,
      "loss": 0.7059,
      "step": 20811
    },
    {
      "epoch": 0.5996657638448684,
      "grad_norm": 0.7351465225219727,
      "learning_rate": 3.6462111860730202e-06,
      "loss": 0.7068,
      "step": 20812
    },
    {
      "epoch": 0.5996945773065177,
      "grad_norm": 0.7031297087669373,
      "learning_rate": 3.645762010891899e-06,
      "loss": 0.7009,
      "step": 20813
    },
    {
      "epoch": 0.5997233907681669,
      "grad_norm": 0.7437852025032043,
      "learning_rate": 3.6453128475048373e-06,
      "loss": 0.6943,
      "step": 20814
    },
    {
      "epoch": 0.5997522042298161,
      "grad_norm": 0.731213390827179,
      "learning_rate": 3.6448636959157453e-06,
      "loss": 0.7146,
      "step": 20815
    },
    {
      "epoch": 0.5997810176914654,
      "grad_norm": 0.7088233232498169,
      "learning_rate": 3.6444145561285373e-06,
      "loss": 0.7049,
      "step": 20816
    },
    {
      "epoch": 0.5998098311531147,
      "grad_norm": 0.7000434398651123,
      "learning_rate": 3.6439654281471226e-06,
      "loss": 0.6923,
      "step": 20817
    },
    {
      "epoch": 0.599838644614764,
      "grad_norm": 0.7216241359710693,
      "learning_rate": 3.6435163119754134e-06,
      "loss": 0.7188,
      "step": 20818
    },
    {
      "epoch": 0.5998674580764133,
      "grad_norm": 0.7290047407150269,
      "learning_rate": 3.643067207617322e-06,
      "loss": 0.6989,
      "step": 20819
    },
    {
      "epoch": 0.5998962715380626,
      "grad_norm": 0.7181235551834106,
      "learning_rate": 3.642618115076758e-06,
      "loss": 0.7147,
      "step": 20820
    },
    {
      "epoch": 0.5999250849997119,
      "grad_norm": 0.7089766263961792,
      "learning_rate": 3.6421690343576323e-06,
      "loss": 0.6844,
      "step": 20821
    },
    {
      "epoch": 0.5999538984613612,
      "grad_norm": 0.683009147644043,
      "learning_rate": 3.641719965463858e-06,
      "loss": 0.6662,
      "step": 20822
    },
    {
      "epoch": 0.5999827119230104,
      "grad_norm": 0.7204024195671082,
      "learning_rate": 3.6412709083993452e-06,
      "loss": 0.7083,
      "step": 20823
    },
    {
      "epoch": 0.6000115253846597,
      "grad_norm": 0.7168985605239868,
      "learning_rate": 3.640821863168005e-06,
      "loss": 0.6852,
      "step": 20824
    },
    {
      "epoch": 0.600040338846309,
      "grad_norm": 0.6960815787315369,
      "learning_rate": 3.640372829773747e-06,
      "loss": 0.695,
      "step": 20825
    },
    {
      "epoch": 0.6000691523079583,
      "grad_norm": 0.7186582684516907,
      "learning_rate": 3.6399238082204836e-06,
      "loss": 0.6751,
      "step": 20826
    },
    {
      "epoch": 0.6000979657696076,
      "grad_norm": 0.7134377360343933,
      "learning_rate": 3.6394747985121236e-06,
      "loss": 0.709,
      "step": 20827
    },
    {
      "epoch": 0.6001267792312568,
      "grad_norm": 0.7072638273239136,
      "learning_rate": 3.639025800652578e-06,
      "loss": 0.7202,
      "step": 20828
    },
    {
      "epoch": 0.6001555926929061,
      "grad_norm": 0.7078589797019958,
      "learning_rate": 3.6385768146457573e-06,
      "loss": 0.6877,
      "step": 20829
    },
    {
      "epoch": 0.6001844061545554,
      "grad_norm": 0.7060692310333252,
      "learning_rate": 3.638127840495571e-06,
      "loss": 0.695,
      "step": 20830
    },
    {
      "epoch": 0.6002132196162047,
      "grad_norm": 0.7348540425300598,
      "learning_rate": 3.6376788782059304e-06,
      "loss": 0.6827,
      "step": 20831
    },
    {
      "epoch": 0.600242033077854,
      "grad_norm": 0.7340205311775208,
      "learning_rate": 3.6372299277807445e-06,
      "loss": 0.717,
      "step": 20832
    },
    {
      "epoch": 0.6002708465395032,
      "grad_norm": 0.7367086410522461,
      "learning_rate": 3.636780989223923e-06,
      "loss": 0.7002,
      "step": 20833
    },
    {
      "epoch": 0.6002996600011525,
      "grad_norm": 0.7349328994750977,
      "learning_rate": 3.6363320625393777e-06,
      "loss": 0.7025,
      "step": 20834
    },
    {
      "epoch": 0.6003284734628018,
      "grad_norm": 0.7220914959907532,
      "learning_rate": 3.6358831477310163e-06,
      "loss": 0.7015,
      "step": 20835
    },
    {
      "epoch": 0.6003572869244511,
      "grad_norm": 0.7080073356628418,
      "learning_rate": 3.635434244802749e-06,
      "loss": 0.6994,
      "step": 20836
    },
    {
      "epoch": 0.6003861003861004,
      "grad_norm": 0.7203418612480164,
      "learning_rate": 3.6349853537584854e-06,
      "loss": 0.6943,
      "step": 20837
    },
    {
      "epoch": 0.6004149138477497,
      "grad_norm": 0.7645055055618286,
      "learning_rate": 3.6345364746021363e-06,
      "loss": 0.7015,
      "step": 20838
    },
    {
      "epoch": 0.600443727309399,
      "grad_norm": 0.7362925410270691,
      "learning_rate": 3.6340876073376084e-06,
      "loss": 0.7199,
      "step": 20839
    },
    {
      "epoch": 0.6004725407710483,
      "grad_norm": 0.7213548421859741,
      "learning_rate": 3.633638751968812e-06,
      "loss": 0.7001,
      "step": 20840
    },
    {
      "epoch": 0.6005013542326976,
      "grad_norm": 0.7512950301170349,
      "learning_rate": 3.6331899084996563e-06,
      "loss": 0.7151,
      "step": 20841
    },
    {
      "epoch": 0.6005301676943468,
      "grad_norm": 0.7115629315376282,
      "learning_rate": 3.6327410769340502e-06,
      "loss": 0.7007,
      "step": 20842
    },
    {
      "epoch": 0.600558981155996,
      "grad_norm": 0.725292980670929,
      "learning_rate": 3.6322922572759033e-06,
      "loss": 0.7155,
      "step": 20843
    },
    {
      "epoch": 0.6005877946176453,
      "grad_norm": 0.7098820805549622,
      "learning_rate": 3.631843449529123e-06,
      "loss": 0.6821,
      "step": 20844
    },
    {
      "epoch": 0.6006166080792946,
      "grad_norm": 0.6983896493911743,
      "learning_rate": 3.6313946536976164e-06,
      "loss": 0.6869,
      "step": 20845
    },
    {
      "epoch": 0.6006454215409439,
      "grad_norm": 0.7460038065910339,
      "learning_rate": 3.630945869785296e-06,
      "loss": 0.6966,
      "step": 20846
    },
    {
      "epoch": 0.6006742350025932,
      "grad_norm": 0.725420355796814,
      "learning_rate": 3.6304970977960697e-06,
      "loss": 0.6975,
      "step": 20847
    },
    {
      "epoch": 0.6007030484642425,
      "grad_norm": 0.7370278239250183,
      "learning_rate": 3.6300483377338434e-06,
      "loss": 0.7024,
      "step": 20848
    },
    {
      "epoch": 0.6007318619258918,
      "grad_norm": 0.7295473217964172,
      "learning_rate": 3.629599589602527e-06,
      "loss": 0.7152,
      "step": 20849
    },
    {
      "epoch": 0.6007606753875411,
      "grad_norm": 0.7112430334091187,
      "learning_rate": 3.6291508534060293e-06,
      "loss": 0.7122,
      "step": 20850
    },
    {
      "epoch": 0.6007894888491904,
      "grad_norm": 0.7255517244338989,
      "learning_rate": 3.628702129148256e-06,
      "loss": 0.7105,
      "step": 20851
    },
    {
      "epoch": 0.6008183023108397,
      "grad_norm": 0.6997869610786438,
      "learning_rate": 3.6282534168331175e-06,
      "loss": 0.6979,
      "step": 20852
    },
    {
      "epoch": 0.600847115772489,
      "grad_norm": 0.7195890545845032,
      "learning_rate": 3.627804716464518e-06,
      "loss": 0.6764,
      "step": 20853
    },
    {
      "epoch": 0.6008759292341382,
      "grad_norm": 0.7329708933830261,
      "learning_rate": 3.62735602804637e-06,
      "loss": 0.6921,
      "step": 20854
    },
    {
      "epoch": 0.6009047426957874,
      "grad_norm": 0.7402015328407288,
      "learning_rate": 3.6269073515825783e-06,
      "loss": 0.6958,
      "step": 20855
    },
    {
      "epoch": 0.6009335561574367,
      "grad_norm": 0.7530601024627686,
      "learning_rate": 3.6264586870770513e-06,
      "loss": 0.7108,
      "step": 20856
    },
    {
      "epoch": 0.600962369619086,
      "grad_norm": 0.7104648947715759,
      "learning_rate": 3.6260100345336946e-06,
      "loss": 0.7018,
      "step": 20857
    },
    {
      "epoch": 0.6009911830807353,
      "grad_norm": 0.704299807548523,
      "learning_rate": 3.625561393956417e-06,
      "loss": 0.7151,
      "step": 20858
    },
    {
      "epoch": 0.6010199965423846,
      "grad_norm": 0.6955706477165222,
      "learning_rate": 3.6251127653491274e-06,
      "loss": 0.7002,
      "step": 20859
    },
    {
      "epoch": 0.6010488100040339,
      "grad_norm": 0.7569897770881653,
      "learning_rate": 3.6246641487157307e-06,
      "loss": 0.7364,
      "step": 20860
    },
    {
      "epoch": 0.6010776234656832,
      "grad_norm": 0.7231436371803284,
      "learning_rate": 3.624215544060135e-06,
      "loss": 0.7076,
      "step": 20861
    },
    {
      "epoch": 0.6011064369273325,
      "grad_norm": 0.766176700592041,
      "learning_rate": 3.623766951386246e-06,
      "loss": 0.7093,
      "step": 20862
    },
    {
      "epoch": 0.6011352503889817,
      "grad_norm": 0.7219798564910889,
      "learning_rate": 3.6233183706979713e-06,
      "loss": 0.68,
      "step": 20863
    },
    {
      "epoch": 0.601164063850631,
      "grad_norm": 0.7121400237083435,
      "learning_rate": 3.622869801999218e-06,
      "loss": 0.7256,
      "step": 20864
    },
    {
      "epoch": 0.6011928773122803,
      "grad_norm": 0.7170897126197815,
      "learning_rate": 3.6224212452938913e-06,
      "loss": 0.7179,
      "step": 20865
    },
    {
      "epoch": 0.6012216907739296,
      "grad_norm": 0.7251383662223816,
      "learning_rate": 3.6219727005858985e-06,
      "loss": 0.6899,
      "step": 20866
    },
    {
      "epoch": 0.6012505042355789,
      "grad_norm": 0.7429195046424866,
      "learning_rate": 3.621524167879147e-06,
      "loss": 0.7142,
      "step": 20867
    },
    {
      "epoch": 0.6012793176972282,
      "grad_norm": 0.7402200698852539,
      "learning_rate": 3.621075647177541e-06,
      "loss": 0.693,
      "step": 20868
    },
    {
      "epoch": 0.6013081311588774,
      "grad_norm": 0.7312841415405273,
      "learning_rate": 3.6206271384849857e-06,
      "loss": 0.6954,
      "step": 20869
    },
    {
      "epoch": 0.6013369446205267,
      "grad_norm": 0.7578033804893494,
      "learning_rate": 3.6201786418053915e-06,
      "loss": 0.6999,
      "step": 20870
    },
    {
      "epoch": 0.601365758082176,
      "grad_norm": 0.7025348544120789,
      "learning_rate": 3.619730157142662e-06,
      "loss": 0.6865,
      "step": 20871
    },
    {
      "epoch": 0.6013945715438253,
      "grad_norm": 0.7226232886314392,
      "learning_rate": 3.6192816845007026e-06,
      "loss": 0.699,
      "step": 20872
    },
    {
      "epoch": 0.6014233850054745,
      "grad_norm": 0.7096837162971497,
      "learning_rate": 3.61883322388342e-06,
      "loss": 0.7009,
      "step": 20873
    },
    {
      "epoch": 0.6014521984671238,
      "grad_norm": 0.7249834537506104,
      "learning_rate": 3.6183847752947182e-06,
      "loss": 0.6932,
      "step": 20874
    },
    {
      "epoch": 0.6014810119287731,
      "grad_norm": 0.7489407658576965,
      "learning_rate": 3.6179363387385047e-06,
      "loss": 0.7155,
      "step": 20875
    },
    {
      "epoch": 0.6015098253904224,
      "grad_norm": 0.7653695940971375,
      "learning_rate": 3.617487914218685e-06,
      "loss": 0.7001,
      "step": 20876
    },
    {
      "epoch": 0.6015386388520717,
      "grad_norm": 0.7141615152359009,
      "learning_rate": 3.6170395017391623e-06,
      "loss": 0.6867,
      "step": 20877
    },
    {
      "epoch": 0.601567452313721,
      "grad_norm": 0.6983102560043335,
      "learning_rate": 3.616591101303844e-06,
      "loss": 0.6856,
      "step": 20878
    },
    {
      "epoch": 0.6015962657753703,
      "grad_norm": 0.7276378273963928,
      "learning_rate": 3.6161427129166336e-06,
      "loss": 0.7058,
      "step": 20879
    },
    {
      "epoch": 0.6016250792370196,
      "grad_norm": 0.7268280982971191,
      "learning_rate": 3.615694336581438e-06,
      "loss": 0.6917,
      "step": 20880
    },
    {
      "epoch": 0.6016538926986689,
      "grad_norm": 0.7353107333183289,
      "learning_rate": 3.615245972302158e-06,
      "loss": 0.7004,
      "step": 20881
    },
    {
      "epoch": 0.601682706160318,
      "grad_norm": 0.7181265354156494,
      "learning_rate": 3.614797620082703e-06,
      "loss": 0.7061,
      "step": 20882
    },
    {
      "epoch": 0.6017115196219673,
      "grad_norm": 0.7336665391921997,
      "learning_rate": 3.6143492799269763e-06,
      "loss": 0.7008,
      "step": 20883
    },
    {
      "epoch": 0.6017403330836166,
      "grad_norm": 0.7179293036460876,
      "learning_rate": 3.6139009518388828e-06,
      "loss": 0.6863,
      "step": 20884
    },
    {
      "epoch": 0.6017691465452659,
      "grad_norm": 0.7523141503334045,
      "learning_rate": 3.6134526358223267e-06,
      "loss": 0.7134,
      "step": 20885
    },
    {
      "epoch": 0.6017979600069152,
      "grad_norm": 0.7190448641777039,
      "learning_rate": 3.6130043318812113e-06,
      "loss": 0.7064,
      "step": 20886
    },
    {
      "epoch": 0.6018267734685645,
      "grad_norm": 0.6959463357925415,
      "learning_rate": 3.6125560400194417e-06,
      "loss": 0.7183,
      "step": 20887
    },
    {
      "epoch": 0.6018555869302138,
      "grad_norm": 0.6997365951538086,
      "learning_rate": 3.6121077602409223e-06,
      "loss": 0.7045,
      "step": 20888
    },
    {
      "epoch": 0.6018844003918631,
      "grad_norm": 0.7208220958709717,
      "learning_rate": 3.6116594925495563e-06,
      "loss": 0.7047,
      "step": 20889
    },
    {
      "epoch": 0.6019132138535124,
      "grad_norm": 0.7256035804748535,
      "learning_rate": 3.611211236949248e-06,
      "loss": 0.7131,
      "step": 20890
    },
    {
      "epoch": 0.6019420273151617,
      "grad_norm": 0.7044106125831604,
      "learning_rate": 3.6107629934439026e-06,
      "loss": 0.6709,
      "step": 20891
    },
    {
      "epoch": 0.601970840776811,
      "grad_norm": 0.6767367124557495,
      "learning_rate": 3.6103147620374223e-06,
      "loss": 0.6954,
      "step": 20892
    },
    {
      "epoch": 0.6019996542384602,
      "grad_norm": 0.7220925092697144,
      "learning_rate": 3.609866542733711e-06,
      "loss": 0.6717,
      "step": 20893
    },
    {
      "epoch": 0.6020284677001095,
      "grad_norm": 0.9727131128311157,
      "learning_rate": 3.609418335536673e-06,
      "loss": 0.7088,
      "step": 20894
    },
    {
      "epoch": 0.6020572811617588,
      "grad_norm": 0.7229704856872559,
      "learning_rate": 3.6089701404502096e-06,
      "loss": 0.6935,
      "step": 20895
    },
    {
      "epoch": 0.602086094623408,
      "grad_norm": 0.7236185073852539,
      "learning_rate": 3.6085219574782278e-06,
      "loss": 0.7152,
      "step": 20896
    },
    {
      "epoch": 0.6021149080850573,
      "grad_norm": 0.7327321767807007,
      "learning_rate": 3.6080737866246286e-06,
      "loss": 0.7011,
      "step": 20897
    },
    {
      "epoch": 0.6021437215467066,
      "grad_norm": 0.7369648218154907,
      "learning_rate": 3.6076256278933147e-06,
      "loss": 0.7125,
      "step": 20898
    },
    {
      "epoch": 0.6021725350083559,
      "grad_norm": 0.7489016056060791,
      "learning_rate": 3.6071774812881897e-06,
      "loss": 0.7043,
      "step": 20899
    },
    {
      "epoch": 0.6022013484700052,
      "grad_norm": 0.7198607921600342,
      "learning_rate": 3.606729346813157e-06,
      "loss": 0.6819,
      "step": 20900
    },
    {
      "epoch": 0.6022301619316545,
      "grad_norm": 0.7217922210693359,
      "learning_rate": 3.606281224472119e-06,
      "loss": 0.6961,
      "step": 20901
    },
    {
      "epoch": 0.6022589753933038,
      "grad_norm": 0.7030352354049683,
      "learning_rate": 3.6058331142689786e-06,
      "loss": 0.6797,
      "step": 20902
    },
    {
      "epoch": 0.602287788854953,
      "grad_norm": 0.7229472398757935,
      "learning_rate": 3.6053850162076375e-06,
      "loss": 0.7343,
      "step": 20903
    },
    {
      "epoch": 0.6023166023166023,
      "grad_norm": 0.7544835209846497,
      "learning_rate": 3.604936930291999e-06,
      "loss": 0.7248,
      "step": 20904
    },
    {
      "epoch": 0.6023454157782516,
      "grad_norm": 0.7255149483680725,
      "learning_rate": 3.6044888565259628e-06,
      "loss": 0.7142,
      "step": 20905
    },
    {
      "epoch": 0.6023742292399009,
      "grad_norm": 0.6892030835151672,
      "learning_rate": 3.6040407949134366e-06,
      "loss": 0.6969,
      "step": 20906
    },
    {
      "epoch": 0.6024030427015502,
      "grad_norm": 0.74974125623703,
      "learning_rate": 3.6035927454583193e-06,
      "loss": 0.701,
      "step": 20907
    },
    {
      "epoch": 0.6024318561631995,
      "grad_norm": 0.7374292016029358,
      "learning_rate": 3.603144708164513e-06,
      "loss": 0.71,
      "step": 20908
    },
    {
      "epoch": 0.6024606696248487,
      "grad_norm": 0.7427417039871216,
      "learning_rate": 3.60269668303592e-06,
      "loss": 0.7126,
      "step": 20909
    },
    {
      "epoch": 0.602489483086498,
      "grad_norm": 0.7104411721229553,
      "learning_rate": 3.6022486700764423e-06,
      "loss": 0.6941,
      "step": 20910
    },
    {
      "epoch": 0.6025182965481473,
      "grad_norm": 0.7105824947357178,
      "learning_rate": 3.601800669289981e-06,
      "loss": 0.7124,
      "step": 20911
    },
    {
      "epoch": 0.6025471100097965,
      "grad_norm": 0.7230333089828491,
      "learning_rate": 3.601352680680438e-06,
      "loss": 0.7255,
      "step": 20912
    },
    {
      "epoch": 0.6025759234714458,
      "grad_norm": 0.6922183632850647,
      "learning_rate": 3.600904704251715e-06,
      "loss": 0.7002,
      "step": 20913
    },
    {
      "epoch": 0.6026047369330951,
      "grad_norm": 0.715657114982605,
      "learning_rate": 3.600456740007714e-06,
      "loss": 0.7069,
      "step": 20914
    },
    {
      "epoch": 0.6026335503947444,
      "grad_norm": 0.728306233882904,
      "learning_rate": 3.600008787952336e-06,
      "loss": 0.717,
      "step": 20915
    },
    {
      "epoch": 0.6026623638563937,
      "grad_norm": 0.7132197022438049,
      "learning_rate": 3.5995608480894815e-06,
      "loss": 0.7082,
      "step": 20916
    },
    {
      "epoch": 0.602691177318043,
      "grad_norm": 0.701961874961853,
      "learning_rate": 3.59911292042305e-06,
      "loss": 0.6955,
      "step": 20917
    },
    {
      "epoch": 0.6027199907796923,
      "grad_norm": 0.7049961090087891,
      "learning_rate": 3.598665004956947e-06,
      "loss": 0.7141,
      "step": 20918
    },
    {
      "epoch": 0.6027488042413416,
      "grad_norm": 0.721976637840271,
      "learning_rate": 3.5982171016950712e-06,
      "loss": 0.7202,
      "step": 20919
    },
    {
      "epoch": 0.6027776177029909,
      "grad_norm": 0.740233302116394,
      "learning_rate": 3.597769210641323e-06,
      "loss": 0.7169,
      "step": 20920
    },
    {
      "epoch": 0.6028064311646402,
      "grad_norm": 0.7237915396690369,
      "learning_rate": 3.597321331799603e-06,
      "loss": 0.7093,
      "step": 20921
    },
    {
      "epoch": 0.6028352446262895,
      "grad_norm": 0.7145600914955139,
      "learning_rate": 3.5968734651738124e-06,
      "loss": 0.6851,
      "step": 20922
    },
    {
      "epoch": 0.6028640580879386,
      "grad_norm": 0.7102901339530945,
      "learning_rate": 3.596425610767851e-06,
      "loss": 0.7293,
      "step": 20923
    },
    {
      "epoch": 0.6028928715495879,
      "grad_norm": 0.7158185839653015,
      "learning_rate": 3.5959777685856197e-06,
      "loss": 0.6878,
      "step": 20924
    },
    {
      "epoch": 0.6029216850112372,
      "grad_norm": 0.7116808295249939,
      "learning_rate": 3.5955299386310185e-06,
      "loss": 0.6945,
      "step": 20925
    },
    {
      "epoch": 0.6029504984728865,
      "grad_norm": 0.7192535996437073,
      "learning_rate": 3.5950821209079465e-06,
      "loss": 0.7204,
      "step": 20926
    },
    {
      "epoch": 0.6029793119345358,
      "grad_norm": 0.7083394527435303,
      "learning_rate": 3.5946343154203066e-06,
      "loss": 0.7007,
      "step": 20927
    },
    {
      "epoch": 0.6030081253961851,
      "grad_norm": 0.7039783596992493,
      "learning_rate": 3.594186522171995e-06,
      "loss": 0.7084,
      "step": 20928
    },
    {
      "epoch": 0.6030369388578344,
      "grad_norm": 0.7265200614929199,
      "learning_rate": 3.593738741166913e-06,
      "loss": 0.7036,
      "step": 20929
    },
    {
      "epoch": 0.6030657523194837,
      "grad_norm": 0.7139396071434021,
      "learning_rate": 3.5932909724089626e-06,
      "loss": 0.7104,
      "step": 20930
    },
    {
      "epoch": 0.603094565781133,
      "grad_norm": 0.7114752531051636,
      "learning_rate": 3.592843215902042e-06,
      "loss": 0.7097,
      "step": 20931
    },
    {
      "epoch": 0.6031233792427823,
      "grad_norm": 0.6998922824859619,
      "learning_rate": 3.59239547165005e-06,
      "loss": 0.6759,
      "step": 20932
    },
    {
      "epoch": 0.6031521927044315,
      "grad_norm": 0.7364124059677124,
      "learning_rate": 3.5919477396568846e-06,
      "loss": 0.69,
      "step": 20933
    },
    {
      "epoch": 0.6031810061660808,
      "grad_norm": 0.691426694393158,
      "learning_rate": 3.5915000199264482e-06,
      "loss": 0.6778,
      "step": 20934
    },
    {
      "epoch": 0.6032098196277301,
      "grad_norm": 0.7056014537811279,
      "learning_rate": 3.5910523124626397e-06,
      "loss": 0.6855,
      "step": 20935
    },
    {
      "epoch": 0.6032386330893793,
      "grad_norm": 0.7288603782653809,
      "learning_rate": 3.5906046172693553e-06,
      "loss": 0.7145,
      "step": 20936
    },
    {
      "epoch": 0.6032674465510286,
      "grad_norm": 0.7148092985153198,
      "learning_rate": 3.5901569343504974e-06,
      "loss": 0.7181,
      "step": 20937
    },
    {
      "epoch": 0.6032962600126779,
      "grad_norm": 0.7076324224472046,
      "learning_rate": 3.589709263709963e-06,
      "loss": 0.6725,
      "step": 20938
    },
    {
      "epoch": 0.6033250734743272,
      "grad_norm": 0.7173283100128174,
      "learning_rate": 3.5892616053516504e-06,
      "loss": 0.6958,
      "step": 20939
    },
    {
      "epoch": 0.6033538869359765,
      "grad_norm": 0.7018452286720276,
      "learning_rate": 3.5888139592794592e-06,
      "loss": 0.6997,
      "step": 20940
    },
    {
      "epoch": 0.6033827003976258,
      "grad_norm": 0.6984603404998779,
      "learning_rate": 3.588366325497286e-06,
      "loss": 0.6976,
      "step": 20941
    },
    {
      "epoch": 0.603411513859275,
      "grad_norm": 0.7054502964019775,
      "learning_rate": 3.587918704009033e-06,
      "loss": 0.6902,
      "step": 20942
    },
    {
      "epoch": 0.6034403273209243,
      "grad_norm": 0.7010514736175537,
      "learning_rate": 3.5874710948185963e-06,
      "loss": 0.6986,
      "step": 20943
    },
    {
      "epoch": 0.6034691407825736,
      "grad_norm": 0.6939646601676941,
      "learning_rate": 3.5870234979298746e-06,
      "loss": 0.6929,
      "step": 20944
    },
    {
      "epoch": 0.6034979542442229,
      "grad_norm": 0.7039128541946411,
      "learning_rate": 3.586575913346766e-06,
      "loss": 0.7058,
      "step": 20945
    },
    {
      "epoch": 0.6035267677058722,
      "grad_norm": 0.6963468790054321,
      "learning_rate": 3.586128341073167e-06,
      "loss": 0.7121,
      "step": 20946
    },
    {
      "epoch": 0.6035555811675215,
      "grad_norm": 0.718734860420227,
      "learning_rate": 3.5856807811129784e-06,
      "loss": 0.6992,
      "step": 20947
    },
    {
      "epoch": 0.6035843946291708,
      "grad_norm": 0.7151004076004028,
      "learning_rate": 3.585233233470096e-06,
      "loss": 0.6842,
      "step": 20948
    },
    {
      "epoch": 0.6036132080908201,
      "grad_norm": 0.7073171734809875,
      "learning_rate": 3.5847856981484175e-06,
      "loss": 0.7089,
      "step": 20949
    },
    {
      "epoch": 0.6036420215524693,
      "grad_norm": 0.7198441028594971,
      "learning_rate": 3.5843381751518398e-06,
      "loss": 0.7,
      "step": 20950
    },
    {
      "epoch": 0.6036708350141186,
      "grad_norm": 0.7330031991004944,
      "learning_rate": 3.583890664484263e-06,
      "loss": 0.7229,
      "step": 20951
    },
    {
      "epoch": 0.6036996484757678,
      "grad_norm": 0.7099316716194153,
      "learning_rate": 3.583443166149582e-06,
      "loss": 0.691,
      "step": 20952
    },
    {
      "epoch": 0.6037284619374171,
      "grad_norm": 0.744031548500061,
      "learning_rate": 3.5829956801516954e-06,
      "loss": 0.6909,
      "step": 20953
    },
    {
      "epoch": 0.6037572753990664,
      "grad_norm": 1.4306280612945557,
      "learning_rate": 3.5825482064944993e-06,
      "loss": 0.7067,
      "step": 20954
    },
    {
      "epoch": 0.6037860888607157,
      "grad_norm": 0.7443450093269348,
      "learning_rate": 3.582100745181891e-06,
      "loss": 0.6918,
      "step": 20955
    },
    {
      "epoch": 0.603814902322365,
      "grad_norm": 0.7274857759475708,
      "learning_rate": 3.5816532962177696e-06,
      "loss": 0.7115,
      "step": 20956
    },
    {
      "epoch": 0.6038437157840143,
      "grad_norm": 0.6837559938430786,
      "learning_rate": 3.5812058596060295e-06,
      "loss": 0.6916,
      "step": 20957
    },
    {
      "epoch": 0.6038725292456636,
      "grad_norm": 0.7085933089256287,
      "learning_rate": 3.580758435350568e-06,
      "loss": 0.6811,
      "step": 20958
    },
    {
      "epoch": 0.6039013427073129,
      "grad_norm": 0.7152843475341797,
      "learning_rate": 3.5803110234552816e-06,
      "loss": 0.6936,
      "step": 20959
    },
    {
      "epoch": 0.6039301561689622,
      "grad_norm": 0.7127935886383057,
      "learning_rate": 3.579863623924068e-06,
      "loss": 0.699,
      "step": 20960
    },
    {
      "epoch": 0.6039589696306115,
      "grad_norm": 0.7194913029670715,
      "learning_rate": 3.579416236760822e-06,
      "loss": 0.6923,
      "step": 20961
    },
    {
      "epoch": 0.6039877830922608,
      "grad_norm": 0.7296553254127502,
      "learning_rate": 3.5789688619694406e-06,
      "loss": 0.7139,
      "step": 20962
    },
    {
      "epoch": 0.6040165965539099,
      "grad_norm": 0.7060565948486328,
      "learning_rate": 3.5785214995538197e-06,
      "loss": 0.6833,
      "step": 20963
    },
    {
      "epoch": 0.6040454100155592,
      "grad_norm": 0.7197301387786865,
      "learning_rate": 3.5780741495178554e-06,
      "loss": 0.7079,
      "step": 20964
    },
    {
      "epoch": 0.6040742234772085,
      "grad_norm": 0.7094221711158752,
      "learning_rate": 3.577626811865442e-06,
      "loss": 0.6657,
      "step": 20965
    },
    {
      "epoch": 0.6041030369388578,
      "grad_norm": 0.7303697466850281,
      "learning_rate": 3.5771794866004796e-06,
      "loss": 0.717,
      "step": 20966
    },
    {
      "epoch": 0.6041318504005071,
      "grad_norm": 0.7025587558746338,
      "learning_rate": 3.576732173726861e-06,
      "loss": 0.6994,
      "step": 20967
    },
    {
      "epoch": 0.6041606638621564,
      "grad_norm": 0.7347086071968079,
      "learning_rate": 3.5762848732484835e-06,
      "loss": 0.7294,
      "step": 20968
    },
    {
      "epoch": 0.6041894773238057,
      "grad_norm": 0.7169905304908752,
      "learning_rate": 3.5758375851692406e-06,
      "loss": 0.7082,
      "step": 20969
    },
    {
      "epoch": 0.604218290785455,
      "grad_norm": 0.7073343992233276,
      "learning_rate": 3.5753903094930287e-06,
      "loss": 0.6854,
      "step": 20970
    },
    {
      "epoch": 0.6042471042471043,
      "grad_norm": 0.7379081845283508,
      "learning_rate": 3.5749430462237424e-06,
      "loss": 0.7179,
      "step": 20971
    },
    {
      "epoch": 0.6042759177087536,
      "grad_norm": 0.6981736421585083,
      "learning_rate": 3.5744957953652786e-06,
      "loss": 0.6958,
      "step": 20972
    },
    {
      "epoch": 0.6043047311704028,
      "grad_norm": 0.7314359545707703,
      "learning_rate": 3.5740485569215316e-06,
      "loss": 0.6953,
      "step": 20973
    },
    {
      "epoch": 0.6043335446320521,
      "grad_norm": 0.7440698146820068,
      "learning_rate": 3.573601330896396e-06,
      "loss": 0.75,
      "step": 20974
    },
    {
      "epoch": 0.6043623580937014,
      "grad_norm": 0.7179579138755798,
      "learning_rate": 3.5731541172937665e-06,
      "loss": 0.7132,
      "step": 20975
    },
    {
      "epoch": 0.6043911715553507,
      "grad_norm": 0.7359161376953125,
      "learning_rate": 3.572706916117539e-06,
      "loss": 0.71,
      "step": 20976
    },
    {
      "epoch": 0.6044199850169999,
      "grad_norm": 0.6973876953125,
      "learning_rate": 3.5722597273716054e-06,
      "loss": 0.7215,
      "step": 20977
    },
    {
      "epoch": 0.6044487984786492,
      "grad_norm": 0.7142995595932007,
      "learning_rate": 3.5718125510598646e-06,
      "loss": 0.6861,
      "step": 20978
    },
    {
      "epoch": 0.6044776119402985,
      "grad_norm": 0.7051156759262085,
      "learning_rate": 3.5713653871862085e-06,
      "loss": 0.7065,
      "step": 20979
    },
    {
      "epoch": 0.6045064254019478,
      "grad_norm": 0.7031550407409668,
      "learning_rate": 3.570918235754531e-06,
      "loss": 0.7309,
      "step": 20980
    },
    {
      "epoch": 0.6045352388635971,
      "grad_norm": 0.7103055715560913,
      "learning_rate": 3.5704710967687283e-06,
      "loss": 0.7183,
      "step": 20981
    },
    {
      "epoch": 0.6045640523252463,
      "grad_norm": 0.7019908428192139,
      "learning_rate": 3.5700239702326927e-06,
      "loss": 0.7042,
      "step": 20982
    },
    {
      "epoch": 0.6045928657868956,
      "grad_norm": 0.7049282193183899,
      "learning_rate": 3.569576856150319e-06,
      "loss": 0.6945,
      "step": 20983
    },
    {
      "epoch": 0.6046216792485449,
      "grad_norm": 0.6976235508918762,
      "learning_rate": 3.569129754525501e-06,
      "loss": 0.6901,
      "step": 20984
    },
    {
      "epoch": 0.6046504927101942,
      "grad_norm": 0.7420141100883484,
      "learning_rate": 3.568682665362133e-06,
      "loss": 0.6987,
      "step": 20985
    },
    {
      "epoch": 0.6046793061718435,
      "grad_norm": 0.7249619960784912,
      "learning_rate": 3.568235588664107e-06,
      "loss": 0.6796,
      "step": 20986
    },
    {
      "epoch": 0.6047081196334928,
      "grad_norm": 0.689254641532898,
      "learning_rate": 3.567788524435319e-06,
      "loss": 0.6756,
      "step": 20987
    },
    {
      "epoch": 0.6047369330951421,
      "grad_norm": 0.7086905837059021,
      "learning_rate": 3.5673414726796595e-06,
      "loss": 0.689,
      "step": 20988
    },
    {
      "epoch": 0.6047657465567914,
      "grad_norm": 0.7570555210113525,
      "learning_rate": 3.566894433401024e-06,
      "loss": 0.6945,
      "step": 20989
    },
    {
      "epoch": 0.6047945600184406,
      "grad_norm": 0.710046648979187,
      "learning_rate": 3.566447406603306e-06,
      "loss": 0.6968,
      "step": 20990
    },
    {
      "epoch": 0.6048233734800899,
      "grad_norm": 0.7074782252311707,
      "learning_rate": 3.5660003922903984e-06,
      "loss": 0.6942,
      "step": 20991
    },
    {
      "epoch": 0.6048521869417391,
      "grad_norm": 0.6998465061187744,
      "learning_rate": 3.565553390466194e-06,
      "loss": 0.7131,
      "step": 20992
    },
    {
      "epoch": 0.6048810004033884,
      "grad_norm": 0.748894453048706,
      "learning_rate": 3.565106401134585e-06,
      "loss": 0.7072,
      "step": 20993
    },
    {
      "epoch": 0.6049098138650377,
      "grad_norm": 0.7093406319618225,
      "learning_rate": 3.564659424299465e-06,
      "loss": 0.6912,
      "step": 20994
    },
    {
      "epoch": 0.604938627326687,
      "grad_norm": 0.7161188721656799,
      "learning_rate": 3.564212459964727e-06,
      "loss": 0.7144,
      "step": 20995
    },
    {
      "epoch": 0.6049674407883363,
      "grad_norm": 0.7560216188430786,
      "learning_rate": 3.563765508134263e-06,
      "loss": 0.7153,
      "step": 20996
    },
    {
      "epoch": 0.6049962542499856,
      "grad_norm": 0.7249768376350403,
      "learning_rate": 3.5633185688119665e-06,
      "loss": 0.7066,
      "step": 20997
    },
    {
      "epoch": 0.6050250677116349,
      "grad_norm": 0.7006835341453552,
      "learning_rate": 3.5628716420017284e-06,
      "loss": 0.6902,
      "step": 20998
    },
    {
      "epoch": 0.6050538811732842,
      "grad_norm": 0.7346388697624207,
      "learning_rate": 3.562424727707442e-06,
      "loss": 0.7255,
      "step": 20999
    },
    {
      "epoch": 0.6050826946349335,
      "grad_norm": 0.7089472413063049,
      "learning_rate": 3.5619778259329986e-06,
      "loss": 0.6951,
      "step": 21000
    },
    {
      "epoch": 0.6051115080965828,
      "grad_norm": 0.7130894064903259,
      "learning_rate": 3.5615309366822894e-06,
      "loss": 0.7115,
      "step": 21001
    },
    {
      "epoch": 0.605140321558232,
      "grad_norm": 0.7261847257614136,
      "learning_rate": 3.5610840599592093e-06,
      "loss": 0.7093,
      "step": 21002
    },
    {
      "epoch": 0.6051691350198813,
      "grad_norm": 0.7258558869361877,
      "learning_rate": 3.560637195767649e-06,
      "loss": 0.7111,
      "step": 21003
    },
    {
      "epoch": 0.6051979484815305,
      "grad_norm": 0.7050837874412537,
      "learning_rate": 3.5601903441115e-06,
      "loss": 0.6855,
      "step": 21004
    },
    {
      "epoch": 0.6052267619431798,
      "grad_norm": 0.7281344532966614,
      "learning_rate": 3.5597435049946537e-06,
      "loss": 0.6941,
      "step": 21005
    },
    {
      "epoch": 0.6052555754048291,
      "grad_norm": 0.7147849798202515,
      "learning_rate": 3.5592966784210016e-06,
      "loss": 0.7121,
      "step": 21006
    },
    {
      "epoch": 0.6052843888664784,
      "grad_norm": 0.7694924473762512,
      "learning_rate": 3.558849864394436e-06,
      "loss": 0.7177,
      "step": 21007
    },
    {
      "epoch": 0.6053132023281277,
      "grad_norm": 0.7429482340812683,
      "learning_rate": 3.558403062918846e-06,
      "loss": 0.7104,
      "step": 21008
    },
    {
      "epoch": 0.605342015789777,
      "grad_norm": 0.7231078743934631,
      "learning_rate": 3.5579562739981243e-06,
      "loss": 0.6696,
      "step": 21009
    },
    {
      "epoch": 0.6053708292514263,
      "grad_norm": 0.7227749824523926,
      "learning_rate": 3.557509497636163e-06,
      "loss": 0.7118,
      "step": 21010
    },
    {
      "epoch": 0.6053996427130756,
      "grad_norm": 0.7502860426902771,
      "learning_rate": 3.557062733836851e-06,
      "loss": 0.7293,
      "step": 21011
    },
    {
      "epoch": 0.6054284561747248,
      "grad_norm": 0.7266488671302795,
      "learning_rate": 3.556615982604081e-06,
      "loss": 0.7231,
      "step": 21012
    },
    {
      "epoch": 0.6054572696363741,
      "grad_norm": 0.7329099774360657,
      "learning_rate": 3.556169243941742e-06,
      "loss": 0.7191,
      "step": 21013
    },
    {
      "epoch": 0.6054860830980234,
      "grad_norm": 0.7156665325164795,
      "learning_rate": 3.5557225178537257e-06,
      "loss": 0.711,
      "step": 21014
    },
    {
      "epoch": 0.6055148965596727,
      "grad_norm": 0.7421817183494568,
      "learning_rate": 3.5552758043439233e-06,
      "loss": 0.7052,
      "step": 21015
    },
    {
      "epoch": 0.605543710021322,
      "grad_norm": 0.7394518256187439,
      "learning_rate": 3.554829103416224e-06,
      "loss": 0.7122,
      "step": 21016
    },
    {
      "epoch": 0.6055725234829712,
      "grad_norm": 0.7002280950546265,
      "learning_rate": 3.5543824150745195e-06,
      "loss": 0.6997,
      "step": 21017
    },
    {
      "epoch": 0.6056013369446205,
      "grad_norm": 0.7141042351722717,
      "learning_rate": 3.553935739322698e-06,
      "loss": 0.6825,
      "step": 21018
    },
    {
      "epoch": 0.6056301504062698,
      "grad_norm": 0.7069087028503418,
      "learning_rate": 3.553489076164652e-06,
      "loss": 0.7015,
      "step": 21019
    },
    {
      "epoch": 0.6056589638679191,
      "grad_norm": 0.7617918252944946,
      "learning_rate": 3.5530424256042697e-06,
      "loss": 0.7089,
      "step": 21020
    },
    {
      "epoch": 0.6056877773295684,
      "grad_norm": 0.7487912178039551,
      "learning_rate": 3.552595787645441e-06,
      "loss": 0.7074,
      "step": 21021
    },
    {
      "epoch": 0.6057165907912176,
      "grad_norm": 0.7093132138252258,
      "learning_rate": 3.5521491622920558e-06,
      "loss": 0.7051,
      "step": 21022
    },
    {
      "epoch": 0.6057454042528669,
      "grad_norm": 0.7447383999824524,
      "learning_rate": 3.5517025495480045e-06,
      "loss": 0.7088,
      "step": 21023
    },
    {
      "epoch": 0.6057742177145162,
      "grad_norm": 0.7249227166175842,
      "learning_rate": 3.551255949417176e-06,
      "loss": 0.6938,
      "step": 21024
    },
    {
      "epoch": 0.6058030311761655,
      "grad_norm": 0.7065134048461914,
      "learning_rate": 3.5508093619034584e-06,
      "loss": 0.6885,
      "step": 21025
    },
    {
      "epoch": 0.6058318446378148,
      "grad_norm": 0.7045168876647949,
      "learning_rate": 3.5503627870107443e-06,
      "loss": 0.6835,
      "step": 21026
    },
    {
      "epoch": 0.6058606580994641,
      "grad_norm": 0.7083144187927246,
      "learning_rate": 3.5499162247429212e-06,
      "loss": 0.6966,
      "step": 21027
    },
    {
      "epoch": 0.6058894715611134,
      "grad_norm": 0.7212281227111816,
      "learning_rate": 3.5494696751038787e-06,
      "loss": 0.7048,
      "step": 21028
    },
    {
      "epoch": 0.6059182850227627,
      "grad_norm": 0.7439959645271301,
      "learning_rate": 3.549023138097505e-06,
      "loss": 0.7168,
      "step": 21029
    },
    {
      "epoch": 0.605947098484412,
      "grad_norm": 0.7213823199272156,
      "learning_rate": 3.5485766137276894e-06,
      "loss": 0.7085,
      "step": 21030
    },
    {
      "epoch": 0.6059759119460612,
      "grad_norm": 0.7273546457290649,
      "learning_rate": 3.5481301019983196e-06,
      "loss": 0.6895,
      "step": 21031
    },
    {
      "epoch": 0.6060047254077104,
      "grad_norm": 0.7172096967697144,
      "learning_rate": 3.547683602913286e-06,
      "loss": 0.7153,
      "step": 21032
    },
    {
      "epoch": 0.6060335388693597,
      "grad_norm": 0.718426525592804,
      "learning_rate": 3.5472371164764764e-06,
      "loss": 0.6609,
      "step": 21033
    },
    {
      "epoch": 0.606062352331009,
      "grad_norm": 0.739433228969574,
      "learning_rate": 3.5467906426917797e-06,
      "loss": 0.7157,
      "step": 21034
    },
    {
      "epoch": 0.6060911657926583,
      "grad_norm": 0.7224498391151428,
      "learning_rate": 3.546344181563084e-06,
      "loss": 0.6813,
      "step": 21035
    },
    {
      "epoch": 0.6061199792543076,
      "grad_norm": 0.770973265171051,
      "learning_rate": 3.5458977330942747e-06,
      "loss": 0.6907,
      "step": 21036
    },
    {
      "epoch": 0.6061487927159569,
      "grad_norm": 0.7038608193397522,
      "learning_rate": 3.545451297289245e-06,
      "loss": 0.6908,
      "step": 21037
    },
    {
      "epoch": 0.6061776061776062,
      "grad_norm": 0.7167696356773376,
      "learning_rate": 3.54500487415188e-06,
      "loss": 0.7,
      "step": 21038
    },
    {
      "epoch": 0.6062064196392555,
      "grad_norm": 0.7252507209777832,
      "learning_rate": 3.5445584636860684e-06,
      "loss": 0.7265,
      "step": 21039
    },
    {
      "epoch": 0.6062352331009048,
      "grad_norm": 0.7295836210250854,
      "learning_rate": 3.544112065895698e-06,
      "loss": 0.7051,
      "step": 21040
    },
    {
      "epoch": 0.6062640465625541,
      "grad_norm": 0.7267245054244995,
      "learning_rate": 3.543665680784656e-06,
      "loss": 0.6838,
      "step": 21041
    },
    {
      "epoch": 0.6062928600242034,
      "grad_norm": 0.7170661687850952,
      "learning_rate": 3.54321930835683e-06,
      "loss": 0.6978,
      "step": 21042
    },
    {
      "epoch": 0.6063216734858526,
      "grad_norm": 0.7554608583450317,
      "learning_rate": 3.542772948616107e-06,
      "loss": 0.6833,
      "step": 21043
    },
    {
      "epoch": 0.6063504869475018,
      "grad_norm": 0.7161653637886047,
      "learning_rate": 3.542326601566376e-06,
      "loss": 0.6798,
      "step": 21044
    },
    {
      "epoch": 0.6063793004091511,
      "grad_norm": 0.7201554775238037,
      "learning_rate": 3.5418802672115228e-06,
      "loss": 0.72,
      "step": 21045
    },
    {
      "epoch": 0.6064081138708004,
      "grad_norm": 0.7382150888442993,
      "learning_rate": 3.5414339455554345e-06,
      "loss": 0.7228,
      "step": 21046
    },
    {
      "epoch": 0.6064369273324497,
      "grad_norm": 0.7288964986801147,
      "learning_rate": 3.540987636601998e-06,
      "loss": 0.7162,
      "step": 21047
    },
    {
      "epoch": 0.606465740794099,
      "grad_norm": 0.7065891027450562,
      "learning_rate": 3.5405413403551e-06,
      "loss": 0.7067,
      "step": 21048
    },
    {
      "epoch": 0.6064945542557483,
      "grad_norm": 0.732499361038208,
      "learning_rate": 3.5400950568186295e-06,
      "loss": 0.7133,
      "step": 21049
    },
    {
      "epoch": 0.6065233677173976,
      "grad_norm": 0.7111031413078308,
      "learning_rate": 3.5396487859964714e-06,
      "loss": 0.6907,
      "step": 21050
    },
    {
      "epoch": 0.6065521811790469,
      "grad_norm": 0.7170252799987793,
      "learning_rate": 3.539202527892513e-06,
      "loss": 0.6972,
      "step": 21051
    },
    {
      "epoch": 0.6065809946406961,
      "grad_norm": 0.7304619550704956,
      "learning_rate": 3.5387562825106392e-06,
      "loss": 0.6985,
      "step": 21052
    },
    {
      "epoch": 0.6066098081023454,
      "grad_norm": 0.7338573336601257,
      "learning_rate": 3.5383100498547384e-06,
      "loss": 0.7214,
      "step": 21053
    },
    {
      "epoch": 0.6066386215639947,
      "grad_norm": 0.6932106614112854,
      "learning_rate": 3.5378638299286954e-06,
      "loss": 0.6961,
      "step": 21054
    },
    {
      "epoch": 0.606667435025644,
      "grad_norm": 0.6971368193626404,
      "learning_rate": 3.5374176227363977e-06,
      "loss": 0.6954,
      "step": 21055
    },
    {
      "epoch": 0.6066962484872933,
      "grad_norm": 0.7131045460700989,
      "learning_rate": 3.5369714282817302e-06,
      "loss": 0.7117,
      "step": 21056
    },
    {
      "epoch": 0.6067250619489426,
      "grad_norm": 0.7244419455528259,
      "learning_rate": 3.5365252465685783e-06,
      "loss": 0.6954,
      "step": 21057
    },
    {
      "epoch": 0.6067538754105918,
      "grad_norm": 0.7244676947593689,
      "learning_rate": 3.536079077600829e-06,
      "loss": 0.7118,
      "step": 21058
    },
    {
      "epoch": 0.6067826888722411,
      "grad_norm": 0.7234742045402527,
      "learning_rate": 3.535632921382368e-06,
      "loss": 0.7176,
      "step": 21059
    },
    {
      "epoch": 0.6068115023338904,
      "grad_norm": 0.6950367093086243,
      "learning_rate": 3.5351867779170786e-06,
      "loss": 0.6931,
      "step": 21060
    },
    {
      "epoch": 0.6068403157955397,
      "grad_norm": 0.7606542110443115,
      "learning_rate": 3.5347406472088495e-06,
      "loss": 0.7151,
      "step": 21061
    },
    {
      "epoch": 0.6068691292571889,
      "grad_norm": 0.7417978644371033,
      "learning_rate": 3.5342945292615648e-06,
      "loss": 0.6736,
      "step": 21062
    },
    {
      "epoch": 0.6068979427188382,
      "grad_norm": 0.7018498182296753,
      "learning_rate": 3.5338484240791093e-06,
      "loss": 0.6864,
      "step": 21063
    },
    {
      "epoch": 0.6069267561804875,
      "grad_norm": 0.7204033136367798,
      "learning_rate": 3.5334023316653685e-06,
      "loss": 0.7035,
      "step": 21064
    },
    {
      "epoch": 0.6069555696421368,
      "grad_norm": 0.7549755573272705,
      "learning_rate": 3.5329562520242277e-06,
      "loss": 0.7071,
      "step": 21065
    },
    {
      "epoch": 0.6069843831037861,
      "grad_norm": 0.7303200364112854,
      "learning_rate": 3.5325101851595717e-06,
      "loss": 0.6751,
      "step": 21066
    },
    {
      "epoch": 0.6070131965654354,
      "grad_norm": 0.7107045650482178,
      "learning_rate": 3.5320641310752848e-06,
      "loss": 0.6937,
      "step": 21067
    },
    {
      "epoch": 0.6070420100270847,
      "grad_norm": 0.6964973211288452,
      "learning_rate": 3.5316180897752504e-06,
      "loss": 0.687,
      "step": 21068
    },
    {
      "epoch": 0.607070823488734,
      "grad_norm": 0.7073804140090942,
      "learning_rate": 3.531172061263356e-06,
      "loss": 0.7072,
      "step": 21069
    },
    {
      "epoch": 0.6070996369503833,
      "grad_norm": 0.7410145998001099,
      "learning_rate": 3.5307260455434846e-06,
      "loss": 0.7061,
      "step": 21070
    },
    {
      "epoch": 0.6071284504120324,
      "grad_norm": 0.7564051151275635,
      "learning_rate": 3.5302800426195205e-06,
      "loss": 0.73,
      "step": 21071
    },
    {
      "epoch": 0.6071572638736817,
      "grad_norm": 0.7132195234298706,
      "learning_rate": 3.529834052495347e-06,
      "loss": 0.6983,
      "step": 21072
    },
    {
      "epoch": 0.607186077335331,
      "grad_norm": 0.7021605968475342,
      "learning_rate": 3.5293880751748493e-06,
      "loss": 0.716,
      "step": 21073
    },
    {
      "epoch": 0.6072148907969803,
      "grad_norm": 0.6975491046905518,
      "learning_rate": 3.5289421106619125e-06,
      "loss": 0.6877,
      "step": 21074
    },
    {
      "epoch": 0.6072437042586296,
      "grad_norm": 0.7138308882713318,
      "learning_rate": 3.5284961589604193e-06,
      "loss": 0.7109,
      "step": 21075
    },
    {
      "epoch": 0.6072725177202789,
      "grad_norm": 0.7447543144226074,
      "learning_rate": 3.528050220074254e-06,
      "loss": 0.7334,
      "step": 21076
    },
    {
      "epoch": 0.6073013311819282,
      "grad_norm": 0.7142292261123657,
      "learning_rate": 3.5276042940072998e-06,
      "loss": 0.6726,
      "step": 21077
    },
    {
      "epoch": 0.6073301446435775,
      "grad_norm": 0.6972784996032715,
      "learning_rate": 3.5271583807634398e-06,
      "loss": 0.6783,
      "step": 21078
    },
    {
      "epoch": 0.6073589581052268,
      "grad_norm": 0.7388511896133423,
      "learning_rate": 3.526712480346558e-06,
      "loss": 0.7064,
      "step": 21079
    },
    {
      "epoch": 0.6073877715668761,
      "grad_norm": 0.7633275985717773,
      "learning_rate": 3.526266592760538e-06,
      "loss": 0.6969,
      "step": 21080
    },
    {
      "epoch": 0.6074165850285254,
      "grad_norm": 0.7252844572067261,
      "learning_rate": 3.525820718009263e-06,
      "loss": 0.7032,
      "step": 21081
    },
    {
      "epoch": 0.6074453984901746,
      "grad_norm": 0.7187150716781616,
      "learning_rate": 3.525374856096616e-06,
      "loss": 0.7121,
      "step": 21082
    },
    {
      "epoch": 0.6074742119518239,
      "grad_norm": 0.701909601688385,
      "learning_rate": 3.5249290070264797e-06,
      "loss": 0.6886,
      "step": 21083
    },
    {
      "epoch": 0.6075030254134732,
      "grad_norm": 0.7249831557273865,
      "learning_rate": 3.524483170802735e-06,
      "loss": 0.7079,
      "step": 21084
    },
    {
      "epoch": 0.6075318388751224,
      "grad_norm": 0.7520536780357361,
      "learning_rate": 3.5240373474292688e-06,
      "loss": 0.696,
      "step": 21085
    },
    {
      "epoch": 0.6075606523367717,
      "grad_norm": 0.7423651814460754,
      "learning_rate": 3.5235915369099622e-06,
      "loss": 0.6815,
      "step": 21086
    },
    {
      "epoch": 0.607589465798421,
      "grad_norm": 0.7841752171516418,
      "learning_rate": 3.523145739248697e-06,
      "loss": 0.7307,
      "step": 21087
    },
    {
      "epoch": 0.6076182792600703,
      "grad_norm": 0.7285637855529785,
      "learning_rate": 3.5226999544493567e-06,
      "loss": 0.7243,
      "step": 21088
    },
    {
      "epoch": 0.6076470927217196,
      "grad_norm": 0.719388484954834,
      "learning_rate": 3.522254182515823e-06,
      "loss": 0.6908,
      "step": 21089
    },
    {
      "epoch": 0.6076759061833689,
      "grad_norm": 0.7302815318107605,
      "learning_rate": 3.521808423451977e-06,
      "loss": 0.6976,
      "step": 21090
    },
    {
      "epoch": 0.6077047196450182,
      "grad_norm": 0.7366974353790283,
      "learning_rate": 3.5213626772617026e-06,
      "loss": 0.723,
      "step": 21091
    },
    {
      "epoch": 0.6077335331066674,
      "grad_norm": 0.6991862654685974,
      "learning_rate": 3.520916943948882e-06,
      "loss": 0.7143,
      "step": 21092
    },
    {
      "epoch": 0.6077623465683167,
      "grad_norm": 0.7150223255157471,
      "learning_rate": 3.5204712235173956e-06,
      "loss": 0.7033,
      "step": 21093
    },
    {
      "epoch": 0.607791160029966,
      "grad_norm": 0.7312870621681213,
      "learning_rate": 3.520025515971126e-06,
      "loss": 0.7349,
      "step": 21094
    },
    {
      "epoch": 0.6078199734916153,
      "grad_norm": 0.6951905488967896,
      "learning_rate": 3.5195798213139554e-06,
      "loss": 0.7023,
      "step": 21095
    },
    {
      "epoch": 0.6078487869532646,
      "grad_norm": 0.7329648733139038,
      "learning_rate": 3.5191341395497614e-06,
      "loss": 0.7133,
      "step": 21096
    },
    {
      "epoch": 0.6078776004149139,
      "grad_norm": 0.7470287680625916,
      "learning_rate": 3.518688470682432e-06,
      "loss": 0.7163,
      "step": 21097
    },
    {
      "epoch": 0.6079064138765631,
      "grad_norm": 0.7619885206222534,
      "learning_rate": 3.518242814715844e-06,
      "loss": 0.7293,
      "step": 21098
    },
    {
      "epoch": 0.6079352273382124,
      "grad_norm": 0.703127384185791,
      "learning_rate": 3.5177971716538807e-06,
      "loss": 0.6933,
      "step": 21099
    },
    {
      "epoch": 0.6079640407998617,
      "grad_norm": 0.7102954387664795,
      "learning_rate": 3.5173515415004223e-06,
      "loss": 0.695,
      "step": 21100
    },
    {
      "epoch": 0.607992854261511,
      "grad_norm": 0.7009575366973877,
      "learning_rate": 3.5169059242593497e-06,
      "loss": 0.706,
      "step": 21101
    },
    {
      "epoch": 0.6080216677231602,
      "grad_norm": 0.7458292841911316,
      "learning_rate": 3.5164603199345436e-06,
      "loss": 0.6986,
      "step": 21102
    },
    {
      "epoch": 0.6080504811848095,
      "grad_norm": 0.7184535264968872,
      "learning_rate": 3.5160147285298864e-06,
      "loss": 0.7013,
      "step": 21103
    },
    {
      "epoch": 0.6080792946464588,
      "grad_norm": 0.7115961313247681,
      "learning_rate": 3.515569150049256e-06,
      "loss": 0.7161,
      "step": 21104
    },
    {
      "epoch": 0.6081081081081081,
      "grad_norm": 0.7128688097000122,
      "learning_rate": 3.5151235844965354e-06,
      "loss": 0.7019,
      "step": 21105
    },
    {
      "epoch": 0.6081369215697574,
      "grad_norm": 0.7152462601661682,
      "learning_rate": 3.5146780318756025e-06,
      "loss": 0.6927,
      "step": 21106
    },
    {
      "epoch": 0.6081657350314067,
      "grad_norm": 0.7476747035980225,
      "learning_rate": 3.5142324921903405e-06,
      "loss": 0.7437,
      "step": 21107
    },
    {
      "epoch": 0.608194548493056,
      "grad_norm": 0.6959807872772217,
      "learning_rate": 3.5137869654446264e-06,
      "loss": 0.6988,
      "step": 21108
    },
    {
      "epoch": 0.6082233619547053,
      "grad_norm": 0.7211486101150513,
      "learning_rate": 3.513341451642345e-06,
      "loss": 0.6914,
      "step": 21109
    },
    {
      "epoch": 0.6082521754163546,
      "grad_norm": 0.7224115133285522,
      "learning_rate": 3.5128959507873715e-06,
      "loss": 0.7089,
      "step": 21110
    },
    {
      "epoch": 0.6082809888780039,
      "grad_norm": 0.7713883519172668,
      "learning_rate": 3.5124504628835874e-06,
      "loss": 0.7098,
      "step": 21111
    },
    {
      "epoch": 0.608309802339653,
      "grad_norm": 0.758582353591919,
      "learning_rate": 3.512004987934874e-06,
      "loss": 0.7314,
      "step": 21112
    },
    {
      "epoch": 0.6083386158013023,
      "grad_norm": 0.7353163957595825,
      "learning_rate": 3.5115595259451096e-06,
      "loss": 0.7067,
      "step": 21113
    },
    {
      "epoch": 0.6083674292629516,
      "grad_norm": 0.691745400428772,
      "learning_rate": 3.511114076918174e-06,
      "loss": 0.6951,
      "step": 21114
    },
    {
      "epoch": 0.6083962427246009,
      "grad_norm": 0.7367416620254517,
      "learning_rate": 3.510668640857946e-06,
      "loss": 0.6936,
      "step": 21115
    },
    {
      "epoch": 0.6084250561862502,
      "grad_norm": 0.7389132380485535,
      "learning_rate": 3.510223217768306e-06,
      "loss": 0.6987,
      "step": 21116
    },
    {
      "epoch": 0.6084538696478995,
      "grad_norm": 0.7065582275390625,
      "learning_rate": 3.5097778076531318e-06,
      "loss": 0.7022,
      "step": 21117
    },
    {
      "epoch": 0.6084826831095488,
      "grad_norm": 0.741288959980011,
      "learning_rate": 3.509332410516304e-06,
      "loss": 0.6786,
      "step": 21118
    },
    {
      "epoch": 0.6085114965711981,
      "grad_norm": 0.7180743217468262,
      "learning_rate": 3.5088870263617e-06,
      "loss": 0.7016,
      "step": 21119
    },
    {
      "epoch": 0.6085403100328474,
      "grad_norm": 0.7214019298553467,
      "learning_rate": 3.508441655193198e-06,
      "loss": 0.7177,
      "step": 21120
    },
    {
      "epoch": 0.6085691234944967,
      "grad_norm": 0.7177156209945679,
      "learning_rate": 3.5079962970146796e-06,
      "loss": 0.715,
      "step": 21121
    },
    {
      "epoch": 0.608597936956146,
      "grad_norm": 0.7564079165458679,
      "learning_rate": 3.5075509518300226e-06,
      "loss": 0.7062,
      "step": 21122
    },
    {
      "epoch": 0.6086267504177952,
      "grad_norm": 0.7465296983718872,
      "learning_rate": 3.5071056196431047e-06,
      "loss": 0.6809,
      "step": 21123
    },
    {
      "epoch": 0.6086555638794445,
      "grad_norm": 0.717591404914856,
      "learning_rate": 3.506660300457805e-06,
      "loss": 0.7293,
      "step": 21124
    },
    {
      "epoch": 0.6086843773410937,
      "grad_norm": 0.7339202165603638,
      "learning_rate": 3.506214994278e-06,
      "loss": 0.6865,
      "step": 21125
    },
    {
      "epoch": 0.608713190802743,
      "grad_norm": 0.6923955082893372,
      "learning_rate": 3.5057697011075703e-06,
      "loss": 0.6959,
      "step": 21126
    },
    {
      "epoch": 0.6087420042643923,
      "grad_norm": 0.7150111794471741,
      "learning_rate": 3.5053244209503923e-06,
      "loss": 0.6928,
      "step": 21127
    },
    {
      "epoch": 0.6087708177260416,
      "grad_norm": 0.7196464538574219,
      "learning_rate": 3.5048791538103435e-06,
      "loss": 0.7061,
      "step": 21128
    },
    {
      "epoch": 0.6087996311876909,
      "grad_norm": 0.6974805593490601,
      "learning_rate": 3.504433899691304e-06,
      "loss": 0.683,
      "step": 21129
    },
    {
      "epoch": 0.6088284446493402,
      "grad_norm": 0.7613328099250793,
      "learning_rate": 3.50398865859715e-06,
      "loss": 0.7167,
      "step": 21130
    },
    {
      "epoch": 0.6088572581109895,
      "grad_norm": 0.721049427986145,
      "learning_rate": 3.5035434305317594e-06,
      "loss": 0.7149,
      "step": 21131
    },
    {
      "epoch": 0.6088860715726387,
      "grad_norm": 0.7070043683052063,
      "learning_rate": 3.5030982154990082e-06,
      "loss": 0.6842,
      "step": 21132
    },
    {
      "epoch": 0.608914885034288,
      "grad_norm": 0.7067862153053284,
      "learning_rate": 3.502653013502776e-06,
      "loss": 0.6891,
      "step": 21133
    },
    {
      "epoch": 0.6089436984959373,
      "grad_norm": 0.730598509311676,
      "learning_rate": 3.5022078245469395e-06,
      "loss": 0.7204,
      "step": 21134
    },
    {
      "epoch": 0.6089725119575866,
      "grad_norm": 0.7092750072479248,
      "learning_rate": 3.501762648635376e-06,
      "loss": 0.6817,
      "step": 21135
    },
    {
      "epoch": 0.6090013254192359,
      "grad_norm": 1.2167608737945557,
      "learning_rate": 3.501317485771962e-06,
      "loss": 0.7204,
      "step": 21136
    },
    {
      "epoch": 0.6090301388808852,
      "grad_norm": 0.7364619970321655,
      "learning_rate": 3.5008723359605747e-06,
      "loss": 0.7236,
      "step": 21137
    },
    {
      "epoch": 0.6090589523425345,
      "grad_norm": 0.7241188287734985,
      "learning_rate": 3.500427199205091e-06,
      "loss": 0.7028,
      "step": 21138
    },
    {
      "epoch": 0.6090877658041837,
      "grad_norm": 0.7204611301422119,
      "learning_rate": 3.499982075509387e-06,
      "loss": 0.7032,
      "step": 21139
    },
    {
      "epoch": 0.609116579265833,
      "grad_norm": 0.7399390935897827,
      "learning_rate": 3.4995369648773393e-06,
      "loss": 0.7244,
      "step": 21140
    },
    {
      "epoch": 0.6091453927274822,
      "grad_norm": 0.7404844760894775,
      "learning_rate": 3.499091867312825e-06,
      "loss": 0.7361,
      "step": 21141
    },
    {
      "epoch": 0.6091742061891315,
      "grad_norm": 0.7286943793296814,
      "learning_rate": 3.4986467828197203e-06,
      "loss": 0.6997,
      "step": 21142
    },
    {
      "epoch": 0.6092030196507808,
      "grad_norm": 0.7166956663131714,
      "learning_rate": 3.4982017114019017e-06,
      "loss": 0.6812,
      "step": 21143
    },
    {
      "epoch": 0.6092318331124301,
      "grad_norm": 0.7221027612686157,
      "learning_rate": 3.4977566530632424e-06,
      "loss": 0.7108,
      "step": 21144
    },
    {
      "epoch": 0.6092606465740794,
      "grad_norm": 0.7049935460090637,
      "learning_rate": 3.497311607807623e-06,
      "loss": 0.6738,
      "step": 21145
    },
    {
      "epoch": 0.6092894600357287,
      "grad_norm": 0.7381775975227356,
      "learning_rate": 3.4968665756389177e-06,
      "loss": 0.7013,
      "step": 21146
    },
    {
      "epoch": 0.609318273497378,
      "grad_norm": 0.7147257328033447,
      "learning_rate": 3.4964215565610017e-06,
      "loss": 0.7093,
      "step": 21147
    },
    {
      "epoch": 0.6093470869590273,
      "grad_norm": 0.7117981910705566,
      "learning_rate": 3.495976550577751e-06,
      "loss": 0.7142,
      "step": 21148
    },
    {
      "epoch": 0.6093759004206766,
      "grad_norm": 0.7257875204086304,
      "learning_rate": 3.4955315576930392e-06,
      "loss": 0.7049,
      "step": 21149
    },
    {
      "epoch": 0.6094047138823259,
      "grad_norm": 0.7935450077056885,
      "learning_rate": 3.4950865779107456e-06,
      "loss": 0.6945,
      "step": 21150
    },
    {
      "epoch": 0.6094335273439752,
      "grad_norm": 0.7299418449401855,
      "learning_rate": 3.4946416112347436e-06,
      "loss": 0.7097,
      "step": 21151
    },
    {
      "epoch": 0.6094623408056243,
      "grad_norm": 0.7163031697273254,
      "learning_rate": 3.494196657668908e-06,
      "loss": 0.7314,
      "step": 21152
    },
    {
      "epoch": 0.6094911542672736,
      "grad_norm": 0.7287850379943848,
      "learning_rate": 3.4937517172171144e-06,
      "loss": 0.7099,
      "step": 21153
    },
    {
      "epoch": 0.6095199677289229,
      "grad_norm": 0.705326497554779,
      "learning_rate": 3.4933067898832373e-06,
      "loss": 0.7035,
      "step": 21154
    },
    {
      "epoch": 0.6095487811905722,
      "grad_norm": 0.7112399935722351,
      "learning_rate": 3.492861875671153e-06,
      "loss": 0.7192,
      "step": 21155
    },
    {
      "epoch": 0.6095775946522215,
      "grad_norm": 0.7249777913093567,
      "learning_rate": 3.492416974584732e-06,
      "loss": 0.6695,
      "step": 21156
    },
    {
      "epoch": 0.6096064081138708,
      "grad_norm": 0.7198106050491333,
      "learning_rate": 3.4919720866278546e-06,
      "loss": 0.6987,
      "step": 21157
    },
    {
      "epoch": 0.6096352215755201,
      "grad_norm": 0.7217079997062683,
      "learning_rate": 3.491527211804393e-06,
      "loss": 0.7074,
      "step": 21158
    },
    {
      "epoch": 0.6096640350371694,
      "grad_norm": 0.7138576507568359,
      "learning_rate": 3.4910823501182213e-06,
      "loss": 0.7179,
      "step": 21159
    },
    {
      "epoch": 0.6096928484988187,
      "grad_norm": 0.7340227365493774,
      "learning_rate": 3.4906375015732135e-06,
      "loss": 0.7182,
      "step": 21160
    },
    {
      "epoch": 0.609721661960468,
      "grad_norm": 0.7237694263458252,
      "learning_rate": 3.490192666173245e-06,
      "loss": 0.7106,
      "step": 21161
    },
    {
      "epoch": 0.6097504754221172,
      "grad_norm": 0.709377110004425,
      "learning_rate": 3.4897478439221887e-06,
      "loss": 0.7073,
      "step": 21162
    },
    {
      "epoch": 0.6097792888837665,
      "grad_norm": 0.7364779710769653,
      "learning_rate": 3.4893030348239187e-06,
      "loss": 0.7066,
      "step": 21163
    },
    {
      "epoch": 0.6098081023454158,
      "grad_norm": 0.7149826288223267,
      "learning_rate": 3.48885823888231e-06,
      "loss": 0.7063,
      "step": 21164
    },
    {
      "epoch": 0.609836915807065,
      "grad_norm": 0.7014926075935364,
      "learning_rate": 3.488413456101235e-06,
      "loss": 0.6953,
      "step": 21165
    },
    {
      "epoch": 0.6098657292687143,
      "grad_norm": 0.7031119465827942,
      "learning_rate": 3.4879686864845664e-06,
      "loss": 0.6919,
      "step": 21166
    },
    {
      "epoch": 0.6098945427303636,
      "grad_norm": 0.7313164472579956,
      "learning_rate": 3.487523930036181e-06,
      "loss": 0.7159,
      "step": 21167
    },
    {
      "epoch": 0.6099233561920129,
      "grad_norm": 0.7216303944587708,
      "learning_rate": 3.487079186759948e-06,
      "loss": 0.6918,
      "step": 21168
    },
    {
      "epoch": 0.6099521696536622,
      "grad_norm": 0.7223238945007324,
      "learning_rate": 3.4866344566597445e-06,
      "loss": 0.7066,
      "step": 21169
    },
    {
      "epoch": 0.6099809831153115,
      "grad_norm": 0.7014347314834595,
      "learning_rate": 3.486189739739442e-06,
      "loss": 0.6979,
      "step": 21170
    },
    {
      "epoch": 0.6100097965769607,
      "grad_norm": 0.738032341003418,
      "learning_rate": 3.4857450360029128e-06,
      "loss": 0.69,
      "step": 21171
    },
    {
      "epoch": 0.61003861003861,
      "grad_norm": 0.7170391082763672,
      "learning_rate": 3.4853003454540315e-06,
      "loss": 0.7235,
      "step": 21172
    },
    {
      "epoch": 0.6100674235002593,
      "grad_norm": 0.7143610119819641,
      "learning_rate": 3.4848556680966706e-06,
      "loss": 0.736,
      "step": 21173
    },
    {
      "epoch": 0.6100962369619086,
      "grad_norm": 0.7186572551727295,
      "learning_rate": 3.484411003934701e-06,
      "loss": 0.7079,
      "step": 21174
    },
    {
      "epoch": 0.6101250504235579,
      "grad_norm": 0.7097194194793701,
      "learning_rate": 3.4839663529719973e-06,
      "loss": 0.705,
      "step": 21175
    },
    {
      "epoch": 0.6101538638852072,
      "grad_norm": 0.7299306392669678,
      "learning_rate": 3.483521715212431e-06,
      "loss": 0.6868,
      "step": 21176
    },
    {
      "epoch": 0.6101826773468565,
      "grad_norm": 0.7373113036155701,
      "learning_rate": 3.4830770906598754e-06,
      "loss": 0.6785,
      "step": 21177
    },
    {
      "epoch": 0.6102114908085058,
      "grad_norm": 0.7309085130691528,
      "learning_rate": 3.4826324793182007e-06,
      "loss": 0.6916,
      "step": 21178
    },
    {
      "epoch": 0.610240304270155,
      "grad_norm": 0.7481828927993774,
      "learning_rate": 3.482187881191281e-06,
      "loss": 0.7037,
      "step": 21179
    },
    {
      "epoch": 0.6102691177318043,
      "grad_norm": 0.6964400410652161,
      "learning_rate": 3.4817432962829856e-06,
      "loss": 0.6947,
      "step": 21180
    },
    {
      "epoch": 0.6102979311934535,
      "grad_norm": 0.744434118270874,
      "learning_rate": 3.481298724597191e-06,
      "loss": 0.6929,
      "step": 21181
    },
    {
      "epoch": 0.6103267446551028,
      "grad_norm": 0.7063440680503845,
      "learning_rate": 3.4808541661377652e-06,
      "loss": 0.6983,
      "step": 21182
    },
    {
      "epoch": 0.6103555581167521,
      "grad_norm": 0.7177086472511292,
      "learning_rate": 3.4804096209085812e-06,
      "loss": 0.682,
      "step": 21183
    },
    {
      "epoch": 0.6103843715784014,
      "grad_norm": 0.6915759444236755,
      "learning_rate": 3.479965088913511e-06,
      "loss": 0.6962,
      "step": 21184
    },
    {
      "epoch": 0.6104131850400507,
      "grad_norm": 0.7303822040557861,
      "learning_rate": 3.479520570156425e-06,
      "loss": 0.7012,
      "step": 21185
    },
    {
      "epoch": 0.6104419985017,
      "grad_norm": 0.7153183817863464,
      "learning_rate": 3.4790760646411953e-06,
      "loss": 0.685,
      "step": 21186
    },
    {
      "epoch": 0.6104708119633493,
      "grad_norm": 0.7065637707710266,
      "learning_rate": 3.4786315723716914e-06,
      "loss": 0.7158,
      "step": 21187
    },
    {
      "epoch": 0.6104996254249986,
      "grad_norm": 0.7129140496253967,
      "learning_rate": 3.478187093351787e-06,
      "loss": 0.6781,
      "step": 21188
    },
    {
      "epoch": 0.6105284388866479,
      "grad_norm": 0.7263572812080383,
      "learning_rate": 3.477742627585351e-06,
      "loss": 0.6953,
      "step": 21189
    },
    {
      "epoch": 0.6105572523482972,
      "grad_norm": 0.7096492648124695,
      "learning_rate": 3.477298175076256e-06,
      "loss": 0.7129,
      "step": 21190
    },
    {
      "epoch": 0.6105860658099465,
      "grad_norm": 0.7580726146697998,
      "learning_rate": 3.4768537358283704e-06,
      "loss": 0.7281,
      "step": 21191
    },
    {
      "epoch": 0.6106148792715956,
      "grad_norm": 0.7283644080162048,
      "learning_rate": 3.4764093098455655e-06,
      "loss": 0.6881,
      "step": 21192
    },
    {
      "epoch": 0.6106436927332449,
      "grad_norm": 0.7475310564041138,
      "learning_rate": 3.475964897131714e-06,
      "loss": 0.7211,
      "step": 21193
    },
    {
      "epoch": 0.6106725061948942,
      "grad_norm": 0.725187361240387,
      "learning_rate": 3.4755204976906844e-06,
      "loss": 0.6975,
      "step": 21194
    },
    {
      "epoch": 0.6107013196565435,
      "grad_norm": 0.7300293445587158,
      "learning_rate": 3.475076111526348e-06,
      "loss": 0.7113,
      "step": 21195
    },
    {
      "epoch": 0.6107301331181928,
      "grad_norm": 0.7078375816345215,
      "learning_rate": 3.474631738642574e-06,
      "loss": 0.6838,
      "step": 21196
    },
    {
      "epoch": 0.6107589465798421,
      "grad_norm": 0.7248975038528442,
      "learning_rate": 3.4741873790432323e-06,
      "loss": 0.6768,
      "step": 21197
    },
    {
      "epoch": 0.6107877600414914,
      "grad_norm": 0.7041966915130615,
      "learning_rate": 3.473743032732194e-06,
      "loss": 0.6914,
      "step": 21198
    },
    {
      "epoch": 0.6108165735031407,
      "grad_norm": 0.7539559602737427,
      "learning_rate": 3.473298699713328e-06,
      "loss": 0.7106,
      "step": 21199
    },
    {
      "epoch": 0.61084538696479,
      "grad_norm": 0.7117295265197754,
      "learning_rate": 3.472854379990504e-06,
      "loss": 0.6907,
      "step": 21200
    },
    {
      "epoch": 0.6108742004264393,
      "grad_norm": 0.7204748392105103,
      "learning_rate": 3.4724100735675913e-06,
      "loss": 0.6885,
      "step": 21201
    },
    {
      "epoch": 0.6109030138880885,
      "grad_norm": 0.7533184885978699,
      "learning_rate": 3.4719657804484607e-06,
      "loss": 0.7089,
      "step": 21202
    },
    {
      "epoch": 0.6109318273497378,
      "grad_norm": 0.7427067756652832,
      "learning_rate": 3.4715215006369796e-06,
      "loss": 0.6938,
      "step": 21203
    },
    {
      "epoch": 0.6109606408113871,
      "grad_norm": 0.7673174738883972,
      "learning_rate": 3.471077234137018e-06,
      "loss": 0.701,
      "step": 21204
    },
    {
      "epoch": 0.6109894542730364,
      "grad_norm": 0.733544647693634,
      "learning_rate": 3.4706329809524463e-06,
      "loss": 0.7009,
      "step": 21205
    },
    {
      "epoch": 0.6110182677346856,
      "grad_norm": 0.7212480306625366,
      "learning_rate": 3.4701887410871325e-06,
      "loss": 0.672,
      "step": 21206
    },
    {
      "epoch": 0.6110470811963349,
      "grad_norm": 0.7158746719360352,
      "learning_rate": 3.469744514544946e-06,
      "loss": 0.7225,
      "step": 21207
    },
    {
      "epoch": 0.6110758946579842,
      "grad_norm": 0.7174445390701294,
      "learning_rate": 3.4693003013297534e-06,
      "loss": 0.696,
      "step": 21208
    },
    {
      "epoch": 0.6111047081196335,
      "grad_norm": 0.7168608903884888,
      "learning_rate": 3.4688561014454265e-06,
      "loss": 0.7067,
      "step": 21209
    },
    {
      "epoch": 0.6111335215812828,
      "grad_norm": 0.6941128373146057,
      "learning_rate": 3.4684119148958317e-06,
      "loss": 0.6958,
      "step": 21210
    },
    {
      "epoch": 0.611162335042932,
      "grad_norm": 0.7296271324157715,
      "learning_rate": 3.467967741684839e-06,
      "loss": 0.6951,
      "step": 21211
    },
    {
      "epoch": 0.6111911485045813,
      "grad_norm": 0.7110561728477478,
      "learning_rate": 3.4675235818163154e-06,
      "loss": 0.7176,
      "step": 21212
    },
    {
      "epoch": 0.6112199619662306,
      "grad_norm": 0.7150813937187195,
      "learning_rate": 3.4670794352941295e-06,
      "loss": 0.7056,
      "step": 21213
    },
    {
      "epoch": 0.6112487754278799,
      "grad_norm": 0.7430864572525024,
      "learning_rate": 3.4666353021221493e-06,
      "loss": 0.7195,
      "step": 21214
    },
    {
      "epoch": 0.6112775888895292,
      "grad_norm": 0.7159491777420044,
      "learning_rate": 3.4661911823042427e-06,
      "loss": 0.7155,
      "step": 21215
    },
    {
      "epoch": 0.6113064023511785,
      "grad_norm": 0.7307013869285583,
      "learning_rate": 3.4657470758442757e-06,
      "loss": 0.6943,
      "step": 21216
    },
    {
      "epoch": 0.6113352158128278,
      "grad_norm": 0.7144821286201477,
      "learning_rate": 3.4653029827461205e-06,
      "loss": 0.7014,
      "step": 21217
    },
    {
      "epoch": 0.6113640292744771,
      "grad_norm": 0.7236940860748291,
      "learning_rate": 3.4648589030136416e-06,
      "loss": 0.7141,
      "step": 21218
    },
    {
      "epoch": 0.6113928427361263,
      "grad_norm": 0.7673560380935669,
      "learning_rate": 3.464414836650707e-06,
      "loss": 0.6909,
      "step": 21219
    },
    {
      "epoch": 0.6114216561977756,
      "grad_norm": 0.7194344401359558,
      "learning_rate": 3.463970783661185e-06,
      "loss": 0.7157,
      "step": 21220
    },
    {
      "epoch": 0.6114504696594248,
      "grad_norm": 0.7537233829498291,
      "learning_rate": 3.463526744048941e-06,
      "loss": 0.7073,
      "step": 21221
    },
    {
      "epoch": 0.6114792831210741,
      "grad_norm": 0.7136300802230835,
      "learning_rate": 3.463082717817844e-06,
      "loss": 0.6557,
      "step": 21222
    },
    {
      "epoch": 0.6115080965827234,
      "grad_norm": 0.7370343208312988,
      "learning_rate": 3.4626387049717593e-06,
      "loss": 0.7217,
      "step": 21223
    },
    {
      "epoch": 0.6115369100443727,
      "grad_norm": 0.7453932166099548,
      "learning_rate": 3.4621947055145554e-06,
      "loss": 0.6941,
      "step": 21224
    },
    {
      "epoch": 0.611565723506022,
      "grad_norm": 0.745080828666687,
      "learning_rate": 3.4617507194500965e-06,
      "loss": 0.7172,
      "step": 21225
    },
    {
      "epoch": 0.6115945369676713,
      "grad_norm": 0.7435272336006165,
      "learning_rate": 3.461306746782253e-06,
      "loss": 0.7017,
      "step": 21226
    },
    {
      "epoch": 0.6116233504293206,
      "grad_norm": 0.7210168838500977,
      "learning_rate": 3.4608627875148893e-06,
      "loss": 0.6796,
      "step": 21227
    },
    {
      "epoch": 0.6116521638909699,
      "grad_norm": 0.7539207339286804,
      "learning_rate": 3.4604188416518715e-06,
      "loss": 0.7186,
      "step": 21228
    },
    {
      "epoch": 0.6116809773526192,
      "grad_norm": 0.740469753742218,
      "learning_rate": 3.4599749091970667e-06,
      "loss": 0.6903,
      "step": 21229
    },
    {
      "epoch": 0.6117097908142685,
      "grad_norm": 0.7206722497940063,
      "learning_rate": 3.4595309901543405e-06,
      "loss": 0.7401,
      "step": 21230
    },
    {
      "epoch": 0.6117386042759178,
      "grad_norm": 0.7544403672218323,
      "learning_rate": 3.459087084527561e-06,
      "loss": 0.7095,
      "step": 21231
    },
    {
      "epoch": 0.611767417737567,
      "grad_norm": 0.7195907235145569,
      "learning_rate": 3.458643192320592e-06,
      "loss": 0.69,
      "step": 21232
    },
    {
      "epoch": 0.6117962311992162,
      "grad_norm": 0.7363622188568115,
      "learning_rate": 3.4581993135373002e-06,
      "loss": 0.6794,
      "step": 21233
    },
    {
      "epoch": 0.6118250446608655,
      "grad_norm": 0.827239990234375,
      "learning_rate": 3.4577554481815513e-06,
      "loss": 0.6902,
      "step": 21234
    },
    {
      "epoch": 0.6118538581225148,
      "grad_norm": 0.7375057935714722,
      "learning_rate": 3.45731159625721e-06,
      "loss": 0.7038,
      "step": 21235
    },
    {
      "epoch": 0.6118826715841641,
      "grad_norm": 0.7568856477737427,
      "learning_rate": 3.456867757768144e-06,
      "loss": 0.7053,
      "step": 21236
    },
    {
      "epoch": 0.6119114850458134,
      "grad_norm": 0.7212337851524353,
      "learning_rate": 3.456423932718216e-06,
      "loss": 0.6904,
      "step": 21237
    },
    {
      "epoch": 0.6119402985074627,
      "grad_norm": 0.7057485580444336,
      "learning_rate": 3.455980121111293e-06,
      "loss": 0.6754,
      "step": 21238
    },
    {
      "epoch": 0.611969111969112,
      "grad_norm": 0.7151290774345398,
      "learning_rate": 3.4555363229512396e-06,
      "loss": 0.7123,
      "step": 21239
    },
    {
      "epoch": 0.6119979254307613,
      "grad_norm": 0.7237271070480347,
      "learning_rate": 3.4550925382419185e-06,
      "loss": 0.6857,
      "step": 21240
    },
    {
      "epoch": 0.6120267388924105,
      "grad_norm": 0.7136909365653992,
      "learning_rate": 3.4546487669872e-06,
      "loss": 0.6991,
      "step": 21241
    },
    {
      "epoch": 0.6120555523540598,
      "grad_norm": 0.708245575428009,
      "learning_rate": 3.454205009190945e-06,
      "loss": 0.6931,
      "step": 21242
    },
    {
      "epoch": 0.6120843658157091,
      "grad_norm": 0.7173810601234436,
      "learning_rate": 3.4537612648570194e-06,
      "loss": 0.7107,
      "step": 21243
    },
    {
      "epoch": 0.6121131792773584,
      "grad_norm": 0.724145233631134,
      "learning_rate": 3.4533175339892883e-06,
      "loss": 0.7238,
      "step": 21244
    },
    {
      "epoch": 0.6121419927390077,
      "grad_norm": 0.7115667462348938,
      "learning_rate": 3.4528738165916147e-06,
      "loss": 0.6924,
      "step": 21245
    },
    {
      "epoch": 0.6121708062006569,
      "grad_norm": 0.7157032489776611,
      "learning_rate": 3.4524301126678628e-06,
      "loss": 0.7128,
      "step": 21246
    },
    {
      "epoch": 0.6121996196623062,
      "grad_norm": 0.7214358448982239,
      "learning_rate": 3.4519864222218983e-06,
      "loss": 0.6788,
      "step": 21247
    },
    {
      "epoch": 0.6122284331239555,
      "grad_norm": 0.7079905867576599,
      "learning_rate": 3.4515427452575847e-06,
      "loss": 0.6935,
      "step": 21248
    },
    {
      "epoch": 0.6122572465856048,
      "grad_norm": 0.7342614531517029,
      "learning_rate": 3.451099081778786e-06,
      "loss": 0.7081,
      "step": 21249
    },
    {
      "epoch": 0.612286060047254,
      "grad_norm": 0.7267321944236755,
      "learning_rate": 3.4506554317893657e-06,
      "loss": 0.6982,
      "step": 21250
    },
    {
      "epoch": 0.6123148735089033,
      "grad_norm": 0.7048943042755127,
      "learning_rate": 3.4502117952931883e-06,
      "loss": 0.6958,
      "step": 21251
    },
    {
      "epoch": 0.6123436869705526,
      "grad_norm": 0.7153779864311218,
      "learning_rate": 3.4497681722941146e-06,
      "loss": 0.7055,
      "step": 21252
    },
    {
      "epoch": 0.6123725004322019,
      "grad_norm": 0.709383487701416,
      "learning_rate": 3.4493245627960125e-06,
      "loss": 0.7071,
      "step": 21253
    },
    {
      "epoch": 0.6124013138938512,
      "grad_norm": 0.7422047257423401,
      "learning_rate": 3.4488809668027436e-06,
      "loss": 0.719,
      "step": 21254
    },
    {
      "epoch": 0.6124301273555005,
      "grad_norm": 0.7132028937339783,
      "learning_rate": 3.4484373843181705e-06,
      "loss": 0.7045,
      "step": 21255
    },
    {
      "epoch": 0.6124589408171498,
      "grad_norm": 0.7238273620605469,
      "learning_rate": 3.4479938153461578e-06,
      "loss": 0.703,
      "step": 21256
    },
    {
      "epoch": 0.6124877542787991,
      "grad_norm": 0.7373000979423523,
      "learning_rate": 3.4475502598905665e-06,
      "loss": 0.7029,
      "step": 21257
    },
    {
      "epoch": 0.6125165677404484,
      "grad_norm": 0.7191250324249268,
      "learning_rate": 3.447106717955261e-06,
      "loss": 0.6793,
      "step": 21258
    },
    {
      "epoch": 0.6125453812020977,
      "grad_norm": 0.6988528966903687,
      "learning_rate": 3.446663189544103e-06,
      "loss": 0.6905,
      "step": 21259
    },
    {
      "epoch": 0.6125741946637469,
      "grad_norm": 0.706325352191925,
      "learning_rate": 3.4462196746609565e-06,
      "loss": 0.6946,
      "step": 21260
    },
    {
      "epoch": 0.6126030081253961,
      "grad_norm": 0.7563307881355286,
      "learning_rate": 3.4457761733096833e-06,
      "loss": 0.7295,
      "step": 21261
    },
    {
      "epoch": 0.6126318215870454,
      "grad_norm": 0.7472084164619446,
      "learning_rate": 3.445332685494146e-06,
      "loss": 0.7185,
      "step": 21262
    },
    {
      "epoch": 0.6126606350486947,
      "grad_norm": 0.7441907525062561,
      "learning_rate": 3.444889211218206e-06,
      "loss": 0.692,
      "step": 21263
    },
    {
      "epoch": 0.612689448510344,
      "grad_norm": 0.7407787442207336,
      "learning_rate": 3.444445750485726e-06,
      "loss": 0.707,
      "step": 21264
    },
    {
      "epoch": 0.6127182619719933,
      "grad_norm": 0.7428044080734253,
      "learning_rate": 3.44400230330057e-06,
      "loss": 0.6997,
      "step": 21265
    },
    {
      "epoch": 0.6127470754336426,
      "grad_norm": 0.7435526847839355,
      "learning_rate": 3.4435588696665977e-06,
      "loss": 0.6927,
      "step": 21266
    },
    {
      "epoch": 0.6127758888952919,
      "grad_norm": 0.7434531450271606,
      "learning_rate": 3.443115449587673e-06,
      "loss": 0.691,
      "step": 21267
    },
    {
      "epoch": 0.6128047023569412,
      "grad_norm": 0.7025397419929504,
      "learning_rate": 3.4426720430676546e-06,
      "loss": 0.7129,
      "step": 21268
    },
    {
      "epoch": 0.6128335158185905,
      "grad_norm": 0.7362412810325623,
      "learning_rate": 3.4422286501104076e-06,
      "loss": 0.6788,
      "step": 21269
    },
    {
      "epoch": 0.6128623292802398,
      "grad_norm": 0.7288925051689148,
      "learning_rate": 3.4417852707197908e-06,
      "loss": 0.7066,
      "step": 21270
    },
    {
      "epoch": 0.612891142741889,
      "grad_norm": 0.736190140247345,
      "learning_rate": 3.4413419048996675e-06,
      "loss": 0.6944,
      "step": 21271
    },
    {
      "epoch": 0.6129199562035383,
      "grad_norm": 0.718207597732544,
      "learning_rate": 3.4408985526538974e-06,
      "loss": 0.6845,
      "step": 21272
    },
    {
      "epoch": 0.6129487696651875,
      "grad_norm": 0.7218483090400696,
      "learning_rate": 3.4404552139863436e-06,
      "loss": 0.6936,
      "step": 21273
    },
    {
      "epoch": 0.6129775831268368,
      "grad_norm": 0.7283541560173035,
      "learning_rate": 3.440011888900865e-06,
      "loss": 0.6784,
      "step": 21274
    },
    {
      "epoch": 0.6130063965884861,
      "grad_norm": 0.7379668951034546,
      "learning_rate": 3.4395685774013233e-06,
      "loss": 0.7329,
      "step": 21275
    },
    {
      "epoch": 0.6130352100501354,
      "grad_norm": 0.7164595127105713,
      "learning_rate": 3.4391252794915774e-06,
      "loss": 0.6857,
      "step": 21276
    },
    {
      "epoch": 0.6130640235117847,
      "grad_norm": 0.7213370203971863,
      "learning_rate": 3.4386819951754923e-06,
      "loss": 0.676,
      "step": 21277
    },
    {
      "epoch": 0.613092836973434,
      "grad_norm": 0.7625737190246582,
      "learning_rate": 3.4382387244569267e-06,
      "loss": 0.6739,
      "step": 21278
    },
    {
      "epoch": 0.6131216504350833,
      "grad_norm": 0.7593829035758972,
      "learning_rate": 3.43779546733974e-06,
      "loss": 0.7087,
      "step": 21279
    },
    {
      "epoch": 0.6131504638967326,
      "grad_norm": 0.7158359885215759,
      "learning_rate": 3.4373522238277934e-06,
      "loss": 0.7062,
      "step": 21280
    },
    {
      "epoch": 0.6131792773583818,
      "grad_norm": 0.7419588565826416,
      "learning_rate": 3.436908993924947e-06,
      "loss": 0.7033,
      "step": 21281
    },
    {
      "epoch": 0.6132080908200311,
      "grad_norm": 0.7500599026679993,
      "learning_rate": 3.4364657776350607e-06,
      "loss": 0.7184,
      "step": 21282
    },
    {
      "epoch": 0.6132369042816804,
      "grad_norm": 0.7224976420402527,
      "learning_rate": 3.4360225749619945e-06,
      "loss": 0.7049,
      "step": 21283
    },
    {
      "epoch": 0.6132657177433297,
      "grad_norm": 0.7319057583808899,
      "learning_rate": 3.4355793859096065e-06,
      "loss": 0.698,
      "step": 21284
    },
    {
      "epoch": 0.613294531204979,
      "grad_norm": 0.7227859497070312,
      "learning_rate": 3.4351362104817604e-06,
      "loss": 0.7055,
      "step": 21285
    },
    {
      "epoch": 0.6133233446666283,
      "grad_norm": 0.721458911895752,
      "learning_rate": 3.4346930486823127e-06,
      "loss": 0.6966,
      "step": 21286
    },
    {
      "epoch": 0.6133521581282775,
      "grad_norm": 0.7209321856498718,
      "learning_rate": 3.434249900515124e-06,
      "loss": 0.6903,
      "step": 21287
    },
    {
      "epoch": 0.6133809715899268,
      "grad_norm": 0.7375530004501343,
      "learning_rate": 3.4338067659840527e-06,
      "loss": 0.718,
      "step": 21288
    },
    {
      "epoch": 0.6134097850515761,
      "grad_norm": 0.7105850577354431,
      "learning_rate": 3.433363645092959e-06,
      "loss": 0.7047,
      "step": 21289
    },
    {
      "epoch": 0.6134385985132254,
      "grad_norm": 0.721458375453949,
      "learning_rate": 3.432920537845703e-06,
      "loss": 0.6874,
      "step": 21290
    },
    {
      "epoch": 0.6134674119748746,
      "grad_norm": 0.7309148907661438,
      "learning_rate": 3.4324774442461424e-06,
      "loss": 0.6971,
      "step": 21291
    },
    {
      "epoch": 0.6134962254365239,
      "grad_norm": 0.720429003238678,
      "learning_rate": 3.4320343642981358e-06,
      "loss": 0.6978,
      "step": 21292
    },
    {
      "epoch": 0.6135250388981732,
      "grad_norm": 0.7285169959068298,
      "learning_rate": 3.4315912980055433e-06,
      "loss": 0.6991,
      "step": 21293
    },
    {
      "epoch": 0.6135538523598225,
      "grad_norm": 0.728563129901886,
      "learning_rate": 3.4311482453722223e-06,
      "loss": 0.7169,
      "step": 21294
    },
    {
      "epoch": 0.6135826658214718,
      "grad_norm": 0.6991394758224487,
      "learning_rate": 3.430705206402032e-06,
      "loss": 0.7069,
      "step": 21295
    },
    {
      "epoch": 0.6136114792831211,
      "grad_norm": 0.7303568720817566,
      "learning_rate": 3.43026218109883e-06,
      "loss": 0.6949,
      "step": 21296
    },
    {
      "epoch": 0.6136402927447704,
      "grad_norm": 0.7179474830627441,
      "learning_rate": 3.429819169466476e-06,
      "loss": 0.7078,
      "step": 21297
    },
    {
      "epoch": 0.6136691062064197,
      "grad_norm": 0.7027584314346313,
      "learning_rate": 3.429376171508827e-06,
      "loss": 0.6854,
      "step": 21298
    },
    {
      "epoch": 0.613697919668069,
      "grad_norm": 0.7095968723297119,
      "learning_rate": 3.42893318722974e-06,
      "loss": 0.6933,
      "step": 21299
    },
    {
      "epoch": 0.6137267331297181,
      "grad_norm": 0.7256805896759033,
      "learning_rate": 3.4284902166330758e-06,
      "loss": 0.7178,
      "step": 21300
    },
    {
      "epoch": 0.6137555465913674,
      "grad_norm": 0.7571820616722107,
      "learning_rate": 3.428047259722691e-06,
      "loss": 0.6836,
      "step": 21301
    },
    {
      "epoch": 0.6137843600530167,
      "grad_norm": 0.7047550678253174,
      "learning_rate": 3.4276043165024443e-06,
      "loss": 0.7075,
      "step": 21302
    },
    {
      "epoch": 0.613813173514666,
      "grad_norm": 0.7342797517776489,
      "learning_rate": 3.427161386976191e-06,
      "loss": 0.6956,
      "step": 21303
    },
    {
      "epoch": 0.6138419869763153,
      "grad_norm": 0.7157461047172546,
      "learning_rate": 3.42671847114779e-06,
      "loss": 0.6873,
      "step": 21304
    },
    {
      "epoch": 0.6138708004379646,
      "grad_norm": 0.7267674803733826,
      "learning_rate": 3.4262755690210982e-06,
      "loss": 0.6944,
      "step": 21305
    },
    {
      "epoch": 0.6138996138996139,
      "grad_norm": 0.7438297271728516,
      "learning_rate": 3.425832680599972e-06,
      "loss": 0.703,
      "step": 21306
    },
    {
      "epoch": 0.6139284273612632,
      "grad_norm": 0.7113437056541443,
      "learning_rate": 3.425389805888271e-06,
      "loss": 0.7162,
      "step": 21307
    },
    {
      "epoch": 0.6139572408229125,
      "grad_norm": 0.7086157202720642,
      "learning_rate": 3.424946944889851e-06,
      "loss": 0.6963,
      "step": 21308
    },
    {
      "epoch": 0.6139860542845618,
      "grad_norm": 0.7190143465995789,
      "learning_rate": 3.424504097608568e-06,
      "loss": 0.6902,
      "step": 21309
    },
    {
      "epoch": 0.6140148677462111,
      "grad_norm": 0.7124858498573303,
      "learning_rate": 3.4240612640482797e-06,
      "loss": 0.6987,
      "step": 21310
    },
    {
      "epoch": 0.6140436812078603,
      "grad_norm": 0.7549629807472229,
      "learning_rate": 3.4236184442128404e-06,
      "loss": 0.7143,
      "step": 21311
    },
    {
      "epoch": 0.6140724946695096,
      "grad_norm": 0.7201518416404724,
      "learning_rate": 3.4231756381061104e-06,
      "loss": 0.7114,
      "step": 21312
    },
    {
      "epoch": 0.6141013081311589,
      "grad_norm": 0.7252528667449951,
      "learning_rate": 3.4227328457319448e-06,
      "loss": 0.7086,
      "step": 21313
    },
    {
      "epoch": 0.6141301215928081,
      "grad_norm": 0.7204814553260803,
      "learning_rate": 3.4222900670941993e-06,
      "loss": 0.6761,
      "step": 21314
    },
    {
      "epoch": 0.6141589350544574,
      "grad_norm": 0.7081509232521057,
      "learning_rate": 3.42184730219673e-06,
      "loss": 0.7271,
      "step": 21315
    },
    {
      "epoch": 0.6141877485161067,
      "grad_norm": 0.7265672087669373,
      "learning_rate": 3.421404551043393e-06,
      "loss": 0.7082,
      "step": 21316
    },
    {
      "epoch": 0.614216561977756,
      "grad_norm": 0.6992964148521423,
      "learning_rate": 3.4209618136380447e-06,
      "loss": 0.7153,
      "step": 21317
    },
    {
      "epoch": 0.6142453754394053,
      "grad_norm": 0.7312230467796326,
      "learning_rate": 3.4205190899845405e-06,
      "loss": 0.7049,
      "step": 21318
    },
    {
      "epoch": 0.6142741889010546,
      "grad_norm": 0.7034890651702881,
      "learning_rate": 3.4200763800867353e-06,
      "loss": 0.7137,
      "step": 21319
    },
    {
      "epoch": 0.6143030023627039,
      "grad_norm": 0.6932654976844788,
      "learning_rate": 3.419633683948487e-06,
      "loss": 0.7078,
      "step": 21320
    },
    {
      "epoch": 0.6143318158243531,
      "grad_norm": 0.7249900698661804,
      "learning_rate": 3.4191910015736485e-06,
      "loss": 0.6846,
      "step": 21321
    },
    {
      "epoch": 0.6143606292860024,
      "grad_norm": 0.7472226619720459,
      "learning_rate": 3.4187483329660753e-06,
      "loss": 0.6994,
      "step": 21322
    },
    {
      "epoch": 0.6143894427476517,
      "grad_norm": 0.7138211131095886,
      "learning_rate": 3.418305678129623e-06,
      "loss": 0.6998,
      "step": 21323
    },
    {
      "epoch": 0.614418256209301,
      "grad_norm": 0.7307422757148743,
      "learning_rate": 3.417863037068149e-06,
      "loss": 0.6898,
      "step": 21324
    },
    {
      "epoch": 0.6144470696709503,
      "grad_norm": 0.7288249731063843,
      "learning_rate": 3.4174204097855057e-06,
      "loss": 0.7369,
      "step": 21325
    },
    {
      "epoch": 0.6144758831325996,
      "grad_norm": 0.7014259099960327,
      "learning_rate": 3.4169777962855488e-06,
      "loss": 0.7007,
      "step": 21326
    },
    {
      "epoch": 0.6145046965942488,
      "grad_norm": 0.7055713534355164,
      "learning_rate": 3.4165351965721322e-06,
      "loss": 0.7133,
      "step": 21327
    },
    {
      "epoch": 0.6145335100558981,
      "grad_norm": 0.7287423610687256,
      "learning_rate": 3.416092610649112e-06,
      "loss": 0.711,
      "step": 21328
    },
    {
      "epoch": 0.6145623235175474,
      "grad_norm": 0.7244777679443359,
      "learning_rate": 3.4156500385203417e-06,
      "loss": 0.6796,
      "step": 21329
    },
    {
      "epoch": 0.6145911369791967,
      "grad_norm": 0.7164641618728638,
      "learning_rate": 3.415207480189676e-06,
      "loss": 0.6804,
      "step": 21330
    },
    {
      "epoch": 0.6146199504408459,
      "grad_norm": 0.7332765460014343,
      "learning_rate": 3.4147649356609693e-06,
      "loss": 0.6993,
      "step": 21331
    },
    {
      "epoch": 0.6146487639024952,
      "grad_norm": 0.713084876537323,
      "learning_rate": 3.414322404938074e-06,
      "loss": 0.7092,
      "step": 21332
    },
    {
      "epoch": 0.6146775773641445,
      "grad_norm": 0.7303382754325867,
      "learning_rate": 3.413879888024847e-06,
      "loss": 0.7084,
      "step": 21333
    },
    {
      "epoch": 0.6147063908257938,
      "grad_norm": 0.7226505875587463,
      "learning_rate": 3.4134373849251403e-06,
      "loss": 0.7098,
      "step": 21334
    },
    {
      "epoch": 0.6147352042874431,
      "grad_norm": 0.7498582601547241,
      "learning_rate": 3.412994895642805e-06,
      "loss": 0.7244,
      "step": 21335
    },
    {
      "epoch": 0.6147640177490924,
      "grad_norm": 0.724466860294342,
      "learning_rate": 3.4125524201817008e-06,
      "loss": 0.6778,
      "step": 21336
    },
    {
      "epoch": 0.6147928312107417,
      "grad_norm": 0.7366090416908264,
      "learning_rate": 3.4121099585456775e-06,
      "loss": 0.6872,
      "step": 21337
    },
    {
      "epoch": 0.614821644672391,
      "grad_norm": 0.746033251285553,
      "learning_rate": 3.411667510738589e-06,
      "loss": 0.6903,
      "step": 21338
    },
    {
      "epoch": 0.6148504581340403,
      "grad_norm": 0.7236374020576477,
      "learning_rate": 3.4112250767642896e-06,
      "loss": 0.6743,
      "step": 21339
    },
    {
      "epoch": 0.6148792715956896,
      "grad_norm": 0.7409535050392151,
      "learning_rate": 3.410782656626631e-06,
      "loss": 0.6841,
      "step": 21340
    },
    {
      "epoch": 0.6149080850573387,
      "grad_norm": 0.7107043266296387,
      "learning_rate": 3.410340250329467e-06,
      "loss": 0.7113,
      "step": 21341
    },
    {
      "epoch": 0.614936898518988,
      "grad_norm": 0.7408311367034912,
      "learning_rate": 3.40989785787665e-06,
      "loss": 0.6876,
      "step": 21342
    },
    {
      "epoch": 0.6149657119806373,
      "grad_norm": 0.7193507552146912,
      "learning_rate": 3.4094554792720336e-06,
      "loss": 0.7317,
      "step": 21343
    },
    {
      "epoch": 0.6149945254422866,
      "grad_norm": 0.7184339165687561,
      "learning_rate": 3.409013114519468e-06,
      "loss": 0.7239,
      "step": 21344
    },
    {
      "epoch": 0.6150233389039359,
      "grad_norm": 0.7222334146499634,
      "learning_rate": 3.4085707636228104e-06,
      "loss": 0.7137,
      "step": 21345
    },
    {
      "epoch": 0.6150521523655852,
      "grad_norm": 0.7104118466377258,
      "learning_rate": 3.4081284265859092e-06,
      "loss": 0.6863,
      "step": 21346
    },
    {
      "epoch": 0.6150809658272345,
      "grad_norm": 0.7364882826805115,
      "learning_rate": 3.407686103412618e-06,
      "loss": 0.7122,
      "step": 21347
    },
    {
      "epoch": 0.6151097792888838,
      "grad_norm": 0.7199593782424927,
      "learning_rate": 3.407243794106789e-06,
      "loss": 0.6966,
      "step": 21348
    },
    {
      "epoch": 0.6151385927505331,
      "grad_norm": 0.7422816753387451,
      "learning_rate": 3.406801498672275e-06,
      "loss": 0.6922,
      "step": 21349
    },
    {
      "epoch": 0.6151674062121824,
      "grad_norm": 0.743977963924408,
      "learning_rate": 3.406359217112928e-06,
      "loss": 0.6919,
      "step": 21350
    },
    {
      "epoch": 0.6151962196738316,
      "grad_norm": 0.7226698398590088,
      "learning_rate": 3.405916949432598e-06,
      "loss": 0.6932,
      "step": 21351
    },
    {
      "epoch": 0.6152250331354809,
      "grad_norm": 0.702631950378418,
      "learning_rate": 3.4054746956351392e-06,
      "loss": 0.7114,
      "step": 21352
    },
    {
      "epoch": 0.6152538465971302,
      "grad_norm": 0.7297876477241516,
      "learning_rate": 3.405032455724401e-06,
      "loss": 0.7292,
      "step": 21353
    },
    {
      "epoch": 0.6152826600587794,
      "grad_norm": 0.7307542562484741,
      "learning_rate": 3.4045902297042355e-06,
      "loss": 0.7156,
      "step": 21354
    },
    {
      "epoch": 0.6153114735204287,
      "grad_norm": 0.7391929626464844,
      "learning_rate": 3.404148017578495e-06,
      "loss": 0.7149,
      "step": 21355
    },
    {
      "epoch": 0.615340286982078,
      "grad_norm": 0.7436595559120178,
      "learning_rate": 3.4037058193510297e-06,
      "loss": 0.7029,
      "step": 21356
    },
    {
      "epoch": 0.6153691004437273,
      "grad_norm": 0.7408331036567688,
      "learning_rate": 3.403263635025691e-06,
      "loss": 0.7087,
      "step": 21357
    },
    {
      "epoch": 0.6153979139053766,
      "grad_norm": 0.7249069213867188,
      "learning_rate": 3.40282146460633e-06,
      "loss": 0.7019,
      "step": 21358
    },
    {
      "epoch": 0.6154267273670259,
      "grad_norm": 0.7169705033302307,
      "learning_rate": 3.4023793080967953e-06,
      "loss": 0.7048,
      "step": 21359
    },
    {
      "epoch": 0.6154555408286752,
      "grad_norm": 0.7061245441436768,
      "learning_rate": 3.4019371655009412e-06,
      "loss": 0.6921,
      "step": 21360
    },
    {
      "epoch": 0.6154843542903244,
      "grad_norm": 0.7038903832435608,
      "learning_rate": 3.4014950368226187e-06,
      "loss": 0.7024,
      "step": 21361
    },
    {
      "epoch": 0.6155131677519737,
      "grad_norm": 0.716918408870697,
      "learning_rate": 3.401052922065675e-06,
      "loss": 0.6885,
      "step": 21362
    },
    {
      "epoch": 0.615541981213623,
      "grad_norm": 0.7162415385246277,
      "learning_rate": 3.4006108212339622e-06,
      "loss": 0.6987,
      "step": 21363
    },
    {
      "epoch": 0.6155707946752723,
      "grad_norm": 0.7283015847206116,
      "learning_rate": 3.4001687343313304e-06,
      "loss": 0.6966,
      "step": 21364
    },
    {
      "epoch": 0.6155996081369216,
      "grad_norm": 0.7199291586875916,
      "learning_rate": 3.3997266613616287e-06,
      "loss": 0.694,
      "step": 21365
    },
    {
      "epoch": 0.6156284215985709,
      "grad_norm": 0.7356033325195312,
      "learning_rate": 3.3992846023287094e-06,
      "loss": 0.7018,
      "step": 21366
    },
    {
      "epoch": 0.6156572350602202,
      "grad_norm": 0.7505224943161011,
      "learning_rate": 3.398842557236421e-06,
      "loss": 0.7117,
      "step": 21367
    },
    {
      "epoch": 0.6156860485218694,
      "grad_norm": 0.7453179359436035,
      "learning_rate": 3.398400526088613e-06,
      "loss": 0.7152,
      "step": 21368
    },
    {
      "epoch": 0.6157148619835187,
      "grad_norm": 0.7097959518432617,
      "learning_rate": 3.3979585088891353e-06,
      "loss": 0.6892,
      "step": 21369
    },
    {
      "epoch": 0.615743675445168,
      "grad_norm": 0.6967355608940125,
      "learning_rate": 3.397516505641838e-06,
      "loss": 0.7152,
      "step": 21370
    },
    {
      "epoch": 0.6157724889068172,
      "grad_norm": 0.7389790415763855,
      "learning_rate": 3.3970745163505682e-06,
      "loss": 0.6908,
      "step": 21371
    },
    {
      "epoch": 0.6158013023684665,
      "grad_norm": 0.7461816072463989,
      "learning_rate": 3.396632541019178e-06,
      "loss": 0.6942,
      "step": 21372
    },
    {
      "epoch": 0.6158301158301158,
      "grad_norm": 0.7334977984428406,
      "learning_rate": 3.396190579651517e-06,
      "loss": 0.6878,
      "step": 21373
    },
    {
      "epoch": 0.6158589292917651,
      "grad_norm": 0.7330628037452698,
      "learning_rate": 3.395748632251432e-06,
      "loss": 0.7108,
      "step": 21374
    },
    {
      "epoch": 0.6158877427534144,
      "grad_norm": 0.7524827718734741,
      "learning_rate": 3.395306698822773e-06,
      "loss": 0.7135,
      "step": 21375
    },
    {
      "epoch": 0.6159165562150637,
      "grad_norm": 0.7104335427284241,
      "learning_rate": 3.394864779369388e-06,
      "loss": 0.7032,
      "step": 21376
    },
    {
      "epoch": 0.615945369676713,
      "grad_norm": 0.7093151807785034,
      "learning_rate": 3.394422873895127e-06,
      "loss": 0.6867,
      "step": 21377
    },
    {
      "epoch": 0.6159741831383623,
      "grad_norm": 0.7579057216644287,
      "learning_rate": 3.3939809824038372e-06,
      "loss": 0.7264,
      "step": 21378
    },
    {
      "epoch": 0.6160029966000116,
      "grad_norm": 0.7209824323654175,
      "learning_rate": 3.3935391048993683e-06,
      "loss": 0.6991,
      "step": 21379
    },
    {
      "epoch": 0.6160318100616609,
      "grad_norm": 0.7094721794128418,
      "learning_rate": 3.3930972413855667e-06,
      "loss": 0.6883,
      "step": 21380
    },
    {
      "epoch": 0.61606062352331,
      "grad_norm": 0.7145448327064514,
      "learning_rate": 3.392655391866282e-06,
      "loss": 0.6619,
      "step": 21381
    },
    {
      "epoch": 0.6160894369849593,
      "grad_norm": 0.739780604839325,
      "learning_rate": 3.392213556345363e-06,
      "loss": 0.6935,
      "step": 21382
    },
    {
      "epoch": 0.6161182504466086,
      "grad_norm": 0.741209864616394,
      "learning_rate": 3.391771734826655e-06,
      "loss": 0.7286,
      "step": 21383
    },
    {
      "epoch": 0.6161470639082579,
      "grad_norm": 0.7307767271995544,
      "learning_rate": 3.391329927314009e-06,
      "loss": 0.7156,
      "step": 21384
    },
    {
      "epoch": 0.6161758773699072,
      "grad_norm": 0.7165288329124451,
      "learning_rate": 3.390888133811271e-06,
      "loss": 0.7172,
      "step": 21385
    },
    {
      "epoch": 0.6162046908315565,
      "grad_norm": 0.7495874762535095,
      "learning_rate": 3.3904463543222876e-06,
      "loss": 0.6926,
      "step": 21386
    },
    {
      "epoch": 0.6162335042932058,
      "grad_norm": 0.7121401429176331,
      "learning_rate": 3.3900045888509092e-06,
      "loss": 0.6986,
      "step": 21387
    },
    {
      "epoch": 0.6162623177548551,
      "grad_norm": 0.7125338315963745,
      "learning_rate": 3.3895628374009815e-06,
      "loss": 0.6987,
      "step": 21388
    },
    {
      "epoch": 0.6162911312165044,
      "grad_norm": 0.7232955694198608,
      "learning_rate": 3.3891210999763514e-06,
      "loss": 0.6927,
      "step": 21389
    },
    {
      "epoch": 0.6163199446781537,
      "grad_norm": 0.7238253355026245,
      "learning_rate": 3.3886793765808655e-06,
      "loss": 0.6978,
      "step": 21390
    },
    {
      "epoch": 0.6163487581398029,
      "grad_norm": 0.7216004133224487,
      "learning_rate": 3.3882376672183727e-06,
      "loss": 0.6939,
      "step": 21391
    },
    {
      "epoch": 0.6163775716014522,
      "grad_norm": 0.7085040807723999,
      "learning_rate": 3.387795971892718e-06,
      "loss": 0.7086,
      "step": 21392
    },
    {
      "epoch": 0.6164063850631015,
      "grad_norm": 0.7601993083953857,
      "learning_rate": 3.387354290607748e-06,
      "loss": 0.7049,
      "step": 21393
    },
    {
      "epoch": 0.6164351985247508,
      "grad_norm": 0.7162604331970215,
      "learning_rate": 3.386912623367311e-06,
      "loss": 0.7066,
      "step": 21394
    },
    {
      "epoch": 0.6164640119864,
      "grad_norm": 0.7067676186561584,
      "learning_rate": 3.3864709701752507e-06,
      "loss": 0.6845,
      "step": 21395
    },
    {
      "epoch": 0.6164928254480493,
      "grad_norm": 0.7018063068389893,
      "learning_rate": 3.386029331035417e-06,
      "loss": 0.6961,
      "step": 21396
    },
    {
      "epoch": 0.6165216389096986,
      "grad_norm": 0.7201600670814514,
      "learning_rate": 3.385587705951654e-06,
      "loss": 0.7111,
      "step": 21397
    },
    {
      "epoch": 0.6165504523713479,
      "grad_norm": 0.689491331577301,
      "learning_rate": 3.385146094927809e-06,
      "loss": 0.6844,
      "step": 21398
    },
    {
      "epoch": 0.6165792658329972,
      "grad_norm": 0.6906113624572754,
      "learning_rate": 3.3847044979677268e-06,
      "loss": 0.697,
      "step": 21399
    },
    {
      "epoch": 0.6166080792946464,
      "grad_norm": 0.7144114375114441,
      "learning_rate": 3.3842629150752544e-06,
      "loss": 0.7045,
      "step": 21400
    },
    {
      "epoch": 0.6166368927562957,
      "grad_norm": 0.7351325750350952,
      "learning_rate": 3.3838213462542363e-06,
      "loss": 0.7067,
      "step": 21401
    },
    {
      "epoch": 0.616665706217945,
      "grad_norm": 0.6865726113319397,
      "learning_rate": 3.3833797915085185e-06,
      "loss": 0.6883,
      "step": 21402
    },
    {
      "epoch": 0.6166945196795943,
      "grad_norm": 0.69725501537323,
      "learning_rate": 3.382938250841946e-06,
      "loss": 0.6777,
      "step": 21403
    },
    {
      "epoch": 0.6167233331412436,
      "grad_norm": 0.7071786522865295,
      "learning_rate": 3.382496724258366e-06,
      "loss": 0.6851,
      "step": 21404
    },
    {
      "epoch": 0.6167521466028929,
      "grad_norm": 0.7174922823905945,
      "learning_rate": 3.3820552117616224e-06,
      "loss": 0.7298,
      "step": 21405
    },
    {
      "epoch": 0.6167809600645422,
      "grad_norm": 0.7242327332496643,
      "learning_rate": 3.3816137133555603e-06,
      "loss": 0.685,
      "step": 21406
    },
    {
      "epoch": 0.6168097735261915,
      "grad_norm": 0.7183778285980225,
      "learning_rate": 3.381172229044024e-06,
      "loss": 0.692,
      "step": 21407
    },
    {
      "epoch": 0.6168385869878407,
      "grad_norm": 0.713703453540802,
      "learning_rate": 3.38073075883086e-06,
      "loss": 0.6959,
      "step": 21408
    },
    {
      "epoch": 0.61686740044949,
      "grad_norm": 0.7123861312866211,
      "learning_rate": 3.380289302719913e-06,
      "loss": 0.6905,
      "step": 21409
    },
    {
      "epoch": 0.6168962139111392,
      "grad_norm": 0.7449146509170532,
      "learning_rate": 3.3798478607150277e-06,
      "loss": 0.7126,
      "step": 21410
    },
    {
      "epoch": 0.6169250273727885,
      "grad_norm": 0.7109245657920837,
      "learning_rate": 3.379406432820047e-06,
      "loss": 0.6985,
      "step": 21411
    },
    {
      "epoch": 0.6169538408344378,
      "grad_norm": 0.7047592401504517,
      "learning_rate": 3.378965019038817e-06,
      "loss": 0.6837,
      "step": 21412
    },
    {
      "epoch": 0.6169826542960871,
      "grad_norm": 0.7148482799530029,
      "learning_rate": 3.3785236193751803e-06,
      "loss": 0.7174,
      "step": 21413
    },
    {
      "epoch": 0.6170114677577364,
      "grad_norm": 0.7046384215354919,
      "learning_rate": 3.378082233832982e-06,
      "loss": 0.6923,
      "step": 21414
    },
    {
      "epoch": 0.6170402812193857,
      "grad_norm": 0.7277722954750061,
      "learning_rate": 3.377640862416067e-06,
      "loss": 0.7162,
      "step": 21415
    },
    {
      "epoch": 0.617069094681035,
      "grad_norm": 0.7205997109413147,
      "learning_rate": 3.377199505128278e-06,
      "loss": 0.7101,
      "step": 21416
    },
    {
      "epoch": 0.6170979081426843,
      "grad_norm": 0.7368155121803284,
      "learning_rate": 3.376758161973459e-06,
      "loss": 0.7,
      "step": 21417
    },
    {
      "epoch": 0.6171267216043336,
      "grad_norm": 0.7572677731513977,
      "learning_rate": 3.3763168329554543e-06,
      "loss": 0.7562,
      "step": 21418
    },
    {
      "epoch": 0.6171555350659829,
      "grad_norm": 0.7246406078338623,
      "learning_rate": 3.3758755180781046e-06,
      "loss": 0.7056,
      "step": 21419
    },
    {
      "epoch": 0.6171843485276322,
      "grad_norm": 0.7377133369445801,
      "learning_rate": 3.375434217345257e-06,
      "loss": 0.7383,
      "step": 21420
    },
    {
      "epoch": 0.6172131619892814,
      "grad_norm": 0.7305034399032593,
      "learning_rate": 3.3749929307607544e-06,
      "loss": 0.7049,
      "step": 21421
    },
    {
      "epoch": 0.6172419754509306,
      "grad_norm": 0.7149844765663147,
      "learning_rate": 3.3745516583284375e-06,
      "loss": 0.687,
      "step": 21422
    },
    {
      "epoch": 0.6172707889125799,
      "grad_norm": 0.7334357500076294,
      "learning_rate": 3.374110400052152e-06,
      "loss": 0.7012,
      "step": 21423
    },
    {
      "epoch": 0.6172996023742292,
      "grad_norm": 0.688181459903717,
      "learning_rate": 3.373669155935738e-06,
      "loss": 0.6983,
      "step": 21424
    },
    {
      "epoch": 0.6173284158358785,
      "grad_norm": 0.7330703735351562,
      "learning_rate": 3.3732279259830416e-06,
      "loss": 0.6913,
      "step": 21425
    },
    {
      "epoch": 0.6173572292975278,
      "grad_norm": 0.7552378177642822,
      "learning_rate": 3.3727867101979037e-06,
      "loss": 0.7126,
      "step": 21426
    },
    {
      "epoch": 0.6173860427591771,
      "grad_norm": 0.7218576073646545,
      "learning_rate": 3.3723455085841667e-06,
      "loss": 0.689,
      "step": 21427
    },
    {
      "epoch": 0.6174148562208264,
      "grad_norm": 0.7474130392074585,
      "learning_rate": 3.3719043211456733e-06,
      "loss": 0.6975,
      "step": 21428
    },
    {
      "epoch": 0.6174436696824757,
      "grad_norm": 0.7358794808387756,
      "learning_rate": 3.371463147886266e-06,
      "loss": 0.6854,
      "step": 21429
    },
    {
      "epoch": 0.617472483144125,
      "grad_norm": 0.7291249632835388,
      "learning_rate": 3.371021988809786e-06,
      "loss": 0.7058,
      "step": 21430
    },
    {
      "epoch": 0.6175012966057742,
      "grad_norm": 0.7206782698631287,
      "learning_rate": 3.3705808439200743e-06,
      "loss": 0.6904,
      "step": 21431
    },
    {
      "epoch": 0.6175301100674235,
      "grad_norm": 0.7254750728607178,
      "learning_rate": 3.3701397132209766e-06,
      "loss": 0.7117,
      "step": 21432
    },
    {
      "epoch": 0.6175589235290728,
      "grad_norm": 0.7227324843406677,
      "learning_rate": 3.3696985967163333e-06,
      "loss": 0.706,
      "step": 21433
    },
    {
      "epoch": 0.6175877369907221,
      "grad_norm": 0.7249598503112793,
      "learning_rate": 3.3692574944099854e-06,
      "loss": 0.6738,
      "step": 21434
    },
    {
      "epoch": 0.6176165504523713,
      "grad_norm": 0.7256566286087036,
      "learning_rate": 3.3688164063057745e-06,
      "loss": 0.7164,
      "step": 21435
    },
    {
      "epoch": 0.6176453639140206,
      "grad_norm": 0.7275571823120117,
      "learning_rate": 3.368375332407542e-06,
      "loss": 0.6878,
      "step": 21436
    },
    {
      "epoch": 0.6176741773756699,
      "grad_norm": 0.7184789180755615,
      "learning_rate": 3.367934272719129e-06,
      "loss": 0.6831,
      "step": 21437
    },
    {
      "epoch": 0.6177029908373192,
      "grad_norm": 0.7515884041786194,
      "learning_rate": 3.367493227244377e-06,
      "loss": 0.7231,
      "step": 21438
    },
    {
      "epoch": 0.6177318042989685,
      "grad_norm": 0.7125594019889832,
      "learning_rate": 3.3670521959871282e-06,
      "loss": 0.676,
      "step": 21439
    },
    {
      "epoch": 0.6177606177606177,
      "grad_norm": 0.6924229860305786,
      "learning_rate": 3.366611178951221e-06,
      "loss": 0.7075,
      "step": 21440
    },
    {
      "epoch": 0.617789431222267,
      "grad_norm": 0.7170688509941101,
      "learning_rate": 3.3661701761404976e-06,
      "loss": 0.6862,
      "step": 21441
    },
    {
      "epoch": 0.6178182446839163,
      "grad_norm": 0.7372862100601196,
      "learning_rate": 3.3657291875587994e-06,
      "loss": 0.6942,
      "step": 21442
    },
    {
      "epoch": 0.6178470581455656,
      "grad_norm": 0.720237672328949,
      "learning_rate": 3.365288213209965e-06,
      "loss": 0.702,
      "step": 21443
    },
    {
      "epoch": 0.6178758716072149,
      "grad_norm": 0.7344251871109009,
      "learning_rate": 3.3648472530978384e-06,
      "loss": 0.6995,
      "step": 21444
    },
    {
      "epoch": 0.6179046850688642,
      "grad_norm": 0.7162559032440186,
      "learning_rate": 3.3644063072262565e-06,
      "loss": 0.7311,
      "step": 21445
    },
    {
      "epoch": 0.6179334985305135,
      "grad_norm": 0.710129976272583,
      "learning_rate": 3.3639653755990596e-06,
      "loss": 0.708,
      "step": 21446
    },
    {
      "epoch": 0.6179623119921628,
      "grad_norm": 0.7093676924705505,
      "learning_rate": 3.3635244582200906e-06,
      "loss": 0.692,
      "step": 21447
    },
    {
      "epoch": 0.6179911254538121,
      "grad_norm": 0.719021737575531,
      "learning_rate": 3.3630835550931874e-06,
      "loss": 0.7169,
      "step": 21448
    },
    {
      "epoch": 0.6180199389154613,
      "grad_norm": 0.7053012251853943,
      "learning_rate": 3.3626426662221896e-06,
      "loss": 0.6904,
      "step": 21449
    },
    {
      "epoch": 0.6180487523771105,
      "grad_norm": 0.7316965460777283,
      "learning_rate": 3.3622017916109378e-06,
      "loss": 0.7011,
      "step": 21450
    },
    {
      "epoch": 0.6180775658387598,
      "grad_norm": 0.7347558736801147,
      "learning_rate": 3.3617609312632714e-06,
      "loss": 0.723,
      "step": 21451
    },
    {
      "epoch": 0.6181063793004091,
      "grad_norm": 0.7365884780883789,
      "learning_rate": 3.3613200851830292e-06,
      "loss": 0.7199,
      "step": 21452
    },
    {
      "epoch": 0.6181351927620584,
      "grad_norm": 0.733008861541748,
      "learning_rate": 3.360879253374051e-06,
      "loss": 0.7262,
      "step": 21453
    },
    {
      "epoch": 0.6181640062237077,
      "grad_norm": 0.7160735726356506,
      "learning_rate": 3.3604384358401765e-06,
      "loss": 0.7146,
      "step": 21454
    },
    {
      "epoch": 0.618192819685357,
      "grad_norm": 0.73243647813797,
      "learning_rate": 3.3599976325852416e-06,
      "loss": 0.7021,
      "step": 21455
    },
    {
      "epoch": 0.6182216331470063,
      "grad_norm": 0.7255484461784363,
      "learning_rate": 3.3595568436130903e-06,
      "loss": 0.7123,
      "step": 21456
    },
    {
      "epoch": 0.6182504466086556,
      "grad_norm": 0.7081074118614197,
      "learning_rate": 3.359116068927559e-06,
      "loss": 0.6998,
      "step": 21457
    },
    {
      "epoch": 0.6182792600703049,
      "grad_norm": 0.719663143157959,
      "learning_rate": 3.3586753085324863e-06,
      "loss": 0.7152,
      "step": 21458
    },
    {
      "epoch": 0.6183080735319542,
      "grad_norm": 0.716444730758667,
      "learning_rate": 3.3582345624317116e-06,
      "loss": 0.6776,
      "step": 21459
    },
    {
      "epoch": 0.6183368869936035,
      "grad_norm": 0.7189323902130127,
      "learning_rate": 3.3577938306290724e-06,
      "loss": 0.6962,
      "step": 21460
    },
    {
      "epoch": 0.6183657004552527,
      "grad_norm": 0.7254061102867126,
      "learning_rate": 3.357353113128407e-06,
      "loss": 0.7321,
      "step": 21461
    },
    {
      "epoch": 0.6183945139169019,
      "grad_norm": 0.728190541267395,
      "learning_rate": 3.3569124099335527e-06,
      "loss": 0.7139,
      "step": 21462
    },
    {
      "epoch": 0.6184233273785512,
      "grad_norm": 0.7103750705718994,
      "learning_rate": 3.3564717210483505e-06,
      "loss": 0.7135,
      "step": 21463
    },
    {
      "epoch": 0.6184521408402005,
      "grad_norm": 0.7046568989753723,
      "learning_rate": 3.3560310464766368e-06,
      "loss": 0.7159,
      "step": 21464
    },
    {
      "epoch": 0.6184809543018498,
      "grad_norm": 0.7715509533882141,
      "learning_rate": 3.3555903862222485e-06,
      "loss": 0.7,
      "step": 21465
    },
    {
      "epoch": 0.6185097677634991,
      "grad_norm": 0.7358019948005676,
      "learning_rate": 3.3551497402890245e-06,
      "loss": 0.703,
      "step": 21466
    },
    {
      "epoch": 0.6185385812251484,
      "grad_norm": 0.7358050346374512,
      "learning_rate": 3.3547091086808004e-06,
      "loss": 0.6856,
      "step": 21467
    },
    {
      "epoch": 0.6185673946867977,
      "grad_norm": 0.7224620580673218,
      "learning_rate": 3.354268491401417e-06,
      "loss": 0.69,
      "step": 21468
    },
    {
      "epoch": 0.618596208148447,
      "grad_norm": 0.7382593154907227,
      "learning_rate": 3.3538278884547092e-06,
      "loss": 0.7058,
      "step": 21469
    },
    {
      "epoch": 0.6186250216100962,
      "grad_norm": 0.7536801695823669,
      "learning_rate": 3.353387299844516e-06,
      "loss": 0.7015,
      "step": 21470
    },
    {
      "epoch": 0.6186538350717455,
      "grad_norm": 0.7300873398780823,
      "learning_rate": 3.352946725574673e-06,
      "loss": 0.682,
      "step": 21471
    },
    {
      "epoch": 0.6186826485333948,
      "grad_norm": 0.7345695495605469,
      "learning_rate": 3.3525061656490177e-06,
      "loss": 0.6962,
      "step": 21472
    },
    {
      "epoch": 0.6187114619950441,
      "grad_norm": 0.7305105328559875,
      "learning_rate": 3.3520656200713865e-06,
      "loss": 0.7145,
      "step": 21473
    },
    {
      "epoch": 0.6187402754566934,
      "grad_norm": 0.7088820338249207,
      "learning_rate": 3.3516250888456165e-06,
      "loss": 0.6909,
      "step": 21474
    },
    {
      "epoch": 0.6187690889183427,
      "grad_norm": 0.7398384809494019,
      "learning_rate": 3.351184571975544e-06,
      "loss": 0.6844,
      "step": 21475
    },
    {
      "epoch": 0.6187979023799919,
      "grad_norm": 0.7466076612472534,
      "learning_rate": 3.3507440694650063e-06,
      "loss": 0.6754,
      "step": 21476
    },
    {
      "epoch": 0.6188267158416412,
      "grad_norm": 0.7355815172195435,
      "learning_rate": 3.3503035813178386e-06,
      "loss": 0.705,
      "step": 21477
    },
    {
      "epoch": 0.6188555293032905,
      "grad_norm": 0.7952907085418701,
      "learning_rate": 3.3498631075378774e-06,
      "loss": 0.6529,
      "step": 21478
    },
    {
      "epoch": 0.6188843427649398,
      "grad_norm": 0.7508665323257446,
      "learning_rate": 3.349422648128957e-06,
      "loss": 0.7085,
      "step": 21479
    },
    {
      "epoch": 0.618913156226589,
      "grad_norm": 0.7331496477127075,
      "learning_rate": 3.348982203094918e-06,
      "loss": 0.6866,
      "step": 21480
    },
    {
      "epoch": 0.6189419696882383,
      "grad_norm": 0.7227949500083923,
      "learning_rate": 3.3485417724395937e-06,
      "loss": 0.7099,
      "step": 21481
    },
    {
      "epoch": 0.6189707831498876,
      "grad_norm": 0.7218408584594727,
      "learning_rate": 3.348101356166819e-06,
      "loss": 0.6856,
      "step": 21482
    },
    {
      "epoch": 0.6189995966115369,
      "grad_norm": 0.7151678800582886,
      "learning_rate": 3.3476609542804308e-06,
      "loss": 0.7051,
      "step": 21483
    },
    {
      "epoch": 0.6190284100731862,
      "grad_norm": 0.7215316295623779,
      "learning_rate": 3.347220566784263e-06,
      "loss": 0.6806,
      "step": 21484
    },
    {
      "epoch": 0.6190572235348355,
      "grad_norm": 0.7213900685310364,
      "learning_rate": 3.346780193682152e-06,
      "loss": 0.704,
      "step": 21485
    },
    {
      "epoch": 0.6190860369964848,
      "grad_norm": 0.7356739044189453,
      "learning_rate": 3.3463398349779338e-06,
      "loss": 0.6838,
      "step": 21486
    },
    {
      "epoch": 0.6191148504581341,
      "grad_norm": 0.7202339768409729,
      "learning_rate": 3.3458994906754415e-06,
      "loss": 0.7105,
      "step": 21487
    },
    {
      "epoch": 0.6191436639197834,
      "grad_norm": 0.7138854265213013,
      "learning_rate": 3.3454591607785127e-06,
      "loss": 0.6907,
      "step": 21488
    },
    {
      "epoch": 0.6191724773814326,
      "grad_norm": 0.7123971581459045,
      "learning_rate": 3.3450188452909793e-06,
      "loss": 0.6902,
      "step": 21489
    },
    {
      "epoch": 0.6192012908430818,
      "grad_norm": 0.702455461025238,
      "learning_rate": 3.344578544216678e-06,
      "loss": 0.6859,
      "step": 21490
    },
    {
      "epoch": 0.6192301043047311,
      "grad_norm": 0.7410439252853394,
      "learning_rate": 3.3441382575594403e-06,
      "loss": 0.7183,
      "step": 21491
    },
    {
      "epoch": 0.6192589177663804,
      "grad_norm": 0.718682050704956,
      "learning_rate": 3.3436979853231063e-06,
      "loss": 0.7147,
      "step": 21492
    },
    {
      "epoch": 0.6192877312280297,
      "grad_norm": 0.7094142436981201,
      "learning_rate": 3.343257727511506e-06,
      "loss": 0.6956,
      "step": 21493
    },
    {
      "epoch": 0.619316544689679,
      "grad_norm": 0.722007691860199,
      "learning_rate": 3.3428174841284753e-06,
      "loss": 0.7076,
      "step": 21494
    },
    {
      "epoch": 0.6193453581513283,
      "grad_norm": 0.7176790833473206,
      "learning_rate": 3.342377255177848e-06,
      "loss": 0.705,
      "step": 21495
    },
    {
      "epoch": 0.6193741716129776,
      "grad_norm": 0.7451490759849548,
      "learning_rate": 3.3419370406634566e-06,
      "loss": 0.7051,
      "step": 21496
    },
    {
      "epoch": 0.6194029850746269,
      "grad_norm": 0.720192551612854,
      "learning_rate": 3.341496840589137e-06,
      "loss": 0.6968,
      "step": 21497
    },
    {
      "epoch": 0.6194317985362762,
      "grad_norm": 0.7150084376335144,
      "learning_rate": 3.341056654958722e-06,
      "loss": 0.7228,
      "step": 21498
    },
    {
      "epoch": 0.6194606119979255,
      "grad_norm": 0.7120417952537537,
      "learning_rate": 3.3406164837760452e-06,
      "loss": 0.7067,
      "step": 21499
    },
    {
      "epoch": 0.6194894254595747,
      "grad_norm": 0.7381386756896973,
      "learning_rate": 3.340176327044939e-06,
      "loss": 0.6921,
      "step": 21500
    },
    {
      "epoch": 0.619518238921224,
      "grad_norm": 0.7372288703918457,
      "learning_rate": 3.339736184769239e-06,
      "loss": 0.6952,
      "step": 21501
    },
    {
      "epoch": 0.6195470523828733,
      "grad_norm": 0.7229928374290466,
      "learning_rate": 3.3392960569527766e-06,
      "loss": 0.7063,
      "step": 21502
    },
    {
      "epoch": 0.6195758658445225,
      "grad_norm": 0.7250674962997437,
      "learning_rate": 3.3388559435993845e-06,
      "loss": 0.7159,
      "step": 21503
    },
    {
      "epoch": 0.6196046793061718,
      "grad_norm": 0.7152246832847595,
      "learning_rate": 3.3384158447128976e-06,
      "loss": 0.6857,
      "step": 21504
    },
    {
      "epoch": 0.6196334927678211,
      "grad_norm": 0.7491960525512695,
      "learning_rate": 3.3379757602971474e-06,
      "loss": 0.6976,
      "step": 21505
    },
    {
      "epoch": 0.6196623062294704,
      "grad_norm": 0.7194316983222961,
      "learning_rate": 3.3375356903559676e-06,
      "loss": 0.6953,
      "step": 21506
    },
    {
      "epoch": 0.6196911196911197,
      "grad_norm": 0.699773907661438,
      "learning_rate": 3.3370956348931893e-06,
      "loss": 0.6976,
      "step": 21507
    },
    {
      "epoch": 0.619719933152769,
      "grad_norm": 0.6978623270988464,
      "learning_rate": 3.3366555939126466e-06,
      "loss": 0.6993,
      "step": 21508
    },
    {
      "epoch": 0.6197487466144183,
      "grad_norm": 0.7328379154205322,
      "learning_rate": 3.3362155674181706e-06,
      "loss": 0.7045,
      "step": 21509
    },
    {
      "epoch": 0.6197775600760675,
      "grad_norm": 0.7739545106887817,
      "learning_rate": 3.335775555413594e-06,
      "loss": 0.704,
      "step": 21510
    },
    {
      "epoch": 0.6198063735377168,
      "grad_norm": 0.7131723761558533,
      "learning_rate": 3.3353355579027493e-06,
      "loss": 0.6997,
      "step": 21511
    },
    {
      "epoch": 0.6198351869993661,
      "grad_norm": 0.7134623527526855,
      "learning_rate": 3.334895574889466e-06,
      "loss": 0.6786,
      "step": 21512
    },
    {
      "epoch": 0.6198640004610154,
      "grad_norm": 0.7223333716392517,
      "learning_rate": 3.3344556063775795e-06,
      "loss": 0.6931,
      "step": 21513
    },
    {
      "epoch": 0.6198928139226647,
      "grad_norm": 0.7338805198669434,
      "learning_rate": 3.334015652370919e-06,
      "loss": 0.7161,
      "step": 21514
    },
    {
      "epoch": 0.619921627384314,
      "grad_norm": 0.7494451403617859,
      "learning_rate": 3.333575712873315e-06,
      "loss": 0.727,
      "step": 21515
    },
    {
      "epoch": 0.6199504408459632,
      "grad_norm": 0.7452044486999512,
      "learning_rate": 3.333135787888603e-06,
      "loss": 0.7103,
      "step": 21516
    },
    {
      "epoch": 0.6199792543076125,
      "grad_norm": 0.6985362768173218,
      "learning_rate": 3.332695877420611e-06,
      "loss": 0.6669,
      "step": 21517
    },
    {
      "epoch": 0.6200080677692618,
      "grad_norm": 0.7406305074691772,
      "learning_rate": 3.3322559814731727e-06,
      "loss": 0.7138,
      "step": 21518
    },
    {
      "epoch": 0.620036881230911,
      "grad_norm": 0.7223654389381409,
      "learning_rate": 3.3318161000501164e-06,
      "loss": 0.6999,
      "step": 21519
    },
    {
      "epoch": 0.6200656946925603,
      "grad_norm": 0.755805253982544,
      "learning_rate": 3.3313762331552748e-06,
      "loss": 0.7076,
      "step": 21520
    },
    {
      "epoch": 0.6200945081542096,
      "grad_norm": 0.7324432134628296,
      "learning_rate": 3.3309363807924777e-06,
      "loss": 0.6947,
      "step": 21521
    },
    {
      "epoch": 0.6201233216158589,
      "grad_norm": 0.7720350623130798,
      "learning_rate": 3.3304965429655567e-06,
      "loss": 0.7094,
      "step": 21522
    },
    {
      "epoch": 0.6201521350775082,
      "grad_norm": 0.7510880827903748,
      "learning_rate": 3.3300567196783416e-06,
      "loss": 0.7139,
      "step": 21523
    },
    {
      "epoch": 0.6201809485391575,
      "grad_norm": 0.7054397463798523,
      "learning_rate": 3.329616910934663e-06,
      "loss": 0.6865,
      "step": 21524
    },
    {
      "epoch": 0.6202097620008068,
      "grad_norm": 0.7304303646087646,
      "learning_rate": 3.3291771167383525e-06,
      "loss": 0.6814,
      "step": 21525
    },
    {
      "epoch": 0.6202385754624561,
      "grad_norm": 0.6939480304718018,
      "learning_rate": 3.3287373370932386e-06,
      "loss": 0.6822,
      "step": 21526
    },
    {
      "epoch": 0.6202673889241054,
      "grad_norm": 0.7514075040817261,
      "learning_rate": 3.3282975720031497e-06,
      "loss": 0.7153,
      "step": 21527
    },
    {
      "epoch": 0.6202962023857547,
      "grad_norm": 0.7539485096931458,
      "learning_rate": 3.3278578214719195e-06,
      "loss": 0.6979,
      "step": 21528
    },
    {
      "epoch": 0.620325015847404,
      "grad_norm": 0.7073322534561157,
      "learning_rate": 3.327418085503377e-06,
      "loss": 0.7187,
      "step": 21529
    },
    {
      "epoch": 0.6203538293090531,
      "grad_norm": 0.7340393662452698,
      "learning_rate": 3.32697836410135e-06,
      "loss": 0.6975,
      "step": 21530
    },
    {
      "epoch": 0.6203826427707024,
      "grad_norm": 0.7041076421737671,
      "learning_rate": 3.3265386572696696e-06,
      "loss": 0.6698,
      "step": 21531
    },
    {
      "epoch": 0.6204114562323517,
      "grad_norm": 0.7090592980384827,
      "learning_rate": 3.3260989650121644e-06,
      "loss": 0.6849,
      "step": 21532
    },
    {
      "epoch": 0.620440269694001,
      "grad_norm": 0.7299776077270508,
      "learning_rate": 3.325659287332664e-06,
      "loss": 0.6951,
      "step": 21533
    },
    {
      "epoch": 0.6204690831556503,
      "grad_norm": 0.7273209095001221,
      "learning_rate": 3.325219624234998e-06,
      "loss": 0.6963,
      "step": 21534
    },
    {
      "epoch": 0.6204978966172996,
      "grad_norm": 0.772649347782135,
      "learning_rate": 3.324779975722994e-06,
      "loss": 0.7409,
      "step": 21535
    },
    {
      "epoch": 0.6205267100789489,
      "grad_norm": 0.7353435158729553,
      "learning_rate": 3.3243403418004828e-06,
      "loss": 0.7157,
      "step": 21536
    },
    {
      "epoch": 0.6205555235405982,
      "grad_norm": 0.7316058874130249,
      "learning_rate": 3.3239007224712916e-06,
      "loss": 0.6996,
      "step": 21537
    },
    {
      "epoch": 0.6205843370022475,
      "grad_norm": 0.7215715646743774,
      "learning_rate": 3.3234611177392483e-06,
      "loss": 0.6907,
      "step": 21538
    },
    {
      "epoch": 0.6206131504638968,
      "grad_norm": 0.7620292901992798,
      "learning_rate": 3.3230215276081833e-06,
      "loss": 0.6997,
      "step": 21539
    },
    {
      "epoch": 0.620641963925546,
      "grad_norm": 0.7011286020278931,
      "learning_rate": 3.3225819520819254e-06,
      "loss": 0.6825,
      "step": 21540
    },
    {
      "epoch": 0.6206707773871953,
      "grad_norm": 0.7146539688110352,
      "learning_rate": 3.322142391164301e-06,
      "loss": 0.7113,
      "step": 21541
    },
    {
      "epoch": 0.6206995908488446,
      "grad_norm": 0.746685266494751,
      "learning_rate": 3.3217028448591405e-06,
      "loss": 0.6841,
      "step": 21542
    },
    {
      "epoch": 0.6207284043104938,
      "grad_norm": 0.7147945165634155,
      "learning_rate": 3.3212633131702685e-06,
      "loss": 0.6823,
      "step": 21543
    },
    {
      "epoch": 0.6207572177721431,
      "grad_norm": 0.7099424004554749,
      "learning_rate": 3.3208237961015164e-06,
      "loss": 0.6973,
      "step": 21544
    },
    {
      "epoch": 0.6207860312337924,
      "grad_norm": 0.7127324342727661,
      "learning_rate": 3.320384293656711e-06,
      "loss": 0.7144,
      "step": 21545
    },
    {
      "epoch": 0.6208148446954417,
      "grad_norm": 0.7243671417236328,
      "learning_rate": 3.319944805839679e-06,
      "loss": 0.6924,
      "step": 21546
    },
    {
      "epoch": 0.620843658157091,
      "grad_norm": 0.752389669418335,
      "learning_rate": 3.3195053326542483e-06,
      "loss": 0.6915,
      "step": 21547
    },
    {
      "epoch": 0.6208724716187403,
      "grad_norm": 0.7534275650978088,
      "learning_rate": 3.319065874104246e-06,
      "loss": 0.6936,
      "step": 21548
    },
    {
      "epoch": 0.6209012850803896,
      "grad_norm": 0.7246181964874268,
      "learning_rate": 3.318626430193499e-06,
      "loss": 0.721,
      "step": 21549
    },
    {
      "epoch": 0.6209300985420388,
      "grad_norm": 0.7222579717636108,
      "learning_rate": 3.318187000925836e-06,
      "loss": 0.7136,
      "step": 21550
    },
    {
      "epoch": 0.6209589120036881,
      "grad_norm": 0.7317949533462524,
      "learning_rate": 3.3177475863050806e-06,
      "loss": 0.6964,
      "step": 21551
    },
    {
      "epoch": 0.6209877254653374,
      "grad_norm": 0.717529833316803,
      "learning_rate": 3.317308186335064e-06,
      "loss": 0.6874,
      "step": 21552
    },
    {
      "epoch": 0.6210165389269867,
      "grad_norm": 0.7355861663818359,
      "learning_rate": 3.3168688010196115e-06,
      "loss": 0.6908,
      "step": 21553
    },
    {
      "epoch": 0.621045352388636,
      "grad_norm": 0.7549183964729309,
      "learning_rate": 3.3164294303625487e-06,
      "loss": 0.7171,
      "step": 21554
    },
    {
      "epoch": 0.6210741658502853,
      "grad_norm": 0.7218003273010254,
      "learning_rate": 3.3159900743677032e-06,
      "loss": 0.6926,
      "step": 21555
    },
    {
      "epoch": 0.6211029793119346,
      "grad_norm": 0.726646900177002,
      "learning_rate": 3.3155507330389004e-06,
      "loss": 0.6988,
      "step": 21556
    },
    {
      "epoch": 0.6211317927735838,
      "grad_norm": 0.715939462184906,
      "learning_rate": 3.315111406379967e-06,
      "loss": 0.712,
      "step": 21557
    },
    {
      "epoch": 0.6211606062352331,
      "grad_norm": 0.7164807915687561,
      "learning_rate": 3.314672094394729e-06,
      "loss": 0.6866,
      "step": 21558
    },
    {
      "epoch": 0.6211894196968824,
      "grad_norm": 0.7274450659751892,
      "learning_rate": 3.3142327970870113e-06,
      "loss": 0.7093,
      "step": 21559
    },
    {
      "epoch": 0.6212182331585316,
      "grad_norm": 0.7446077466011047,
      "learning_rate": 3.3137935144606414e-06,
      "loss": 0.7367,
      "step": 21560
    },
    {
      "epoch": 0.6212470466201809,
      "grad_norm": 0.7067439556121826,
      "learning_rate": 3.3133542465194452e-06,
      "loss": 0.7064,
      "step": 21561
    },
    {
      "epoch": 0.6212758600818302,
      "grad_norm": 0.736194908618927,
      "learning_rate": 3.3129149932672462e-06,
      "loss": 0.7009,
      "step": 21562
    },
    {
      "epoch": 0.6213046735434795,
      "grad_norm": 0.7332513928413391,
      "learning_rate": 3.3124757547078718e-06,
      "loss": 0.699,
      "step": 21563
    },
    {
      "epoch": 0.6213334870051288,
      "grad_norm": 0.6999678015708923,
      "learning_rate": 3.3120365308451457e-06,
      "loss": 0.7008,
      "step": 21564
    },
    {
      "epoch": 0.6213623004667781,
      "grad_norm": 0.7176912426948547,
      "learning_rate": 3.3115973216828954e-06,
      "loss": 0.7111,
      "step": 21565
    },
    {
      "epoch": 0.6213911139284274,
      "grad_norm": 0.7361303567886353,
      "learning_rate": 3.3111581272249444e-06,
      "loss": 0.7235,
      "step": 21566
    },
    {
      "epoch": 0.6214199273900767,
      "grad_norm": 0.7053534984588623,
      "learning_rate": 3.310718947475118e-06,
      "loss": 0.6821,
      "step": 21567
    },
    {
      "epoch": 0.621448740851726,
      "grad_norm": 0.7199810743331909,
      "learning_rate": 3.3102797824372414e-06,
      "loss": 0.7277,
      "step": 21568
    },
    {
      "epoch": 0.6214775543133753,
      "grad_norm": 0.6983734369277954,
      "learning_rate": 3.3098406321151377e-06,
      "loss": 0.6898,
      "step": 21569
    },
    {
      "epoch": 0.6215063677750244,
      "grad_norm": 0.7057565450668335,
      "learning_rate": 3.309401496512633e-06,
      "loss": 0.6706,
      "step": 21570
    },
    {
      "epoch": 0.6215351812366737,
      "grad_norm": 0.7211113572120667,
      "learning_rate": 3.3089623756335514e-06,
      "loss": 0.7105,
      "step": 21571
    },
    {
      "epoch": 0.621563994698323,
      "grad_norm": 0.7062333822250366,
      "learning_rate": 3.308523269481717e-06,
      "loss": 0.6822,
      "step": 21572
    },
    {
      "epoch": 0.6215928081599723,
      "grad_norm": 0.7130873799324036,
      "learning_rate": 3.308084178060954e-06,
      "loss": 0.6867,
      "step": 21573
    },
    {
      "epoch": 0.6216216216216216,
      "grad_norm": 0.7890552282333374,
      "learning_rate": 3.3076451013750847e-06,
      "loss": 0.7152,
      "step": 21574
    },
    {
      "epoch": 0.6216504350832709,
      "grad_norm": 0.7367956042289734,
      "learning_rate": 3.307206039427937e-06,
      "loss": 0.697,
      "step": 21575
    },
    {
      "epoch": 0.6216792485449202,
      "grad_norm": 0.7170146703720093,
      "learning_rate": 3.3067669922233325e-06,
      "loss": 0.6967,
      "step": 21576
    },
    {
      "epoch": 0.6217080620065695,
      "grad_norm": 0.7259535193443298,
      "learning_rate": 3.3063279597650943e-06,
      "loss": 0.6893,
      "step": 21577
    },
    {
      "epoch": 0.6217368754682188,
      "grad_norm": 0.7070180773735046,
      "learning_rate": 3.305888942057047e-06,
      "loss": 0.7125,
      "step": 21578
    },
    {
      "epoch": 0.621765688929868,
      "grad_norm": 0.7600560784339905,
      "learning_rate": 3.3054499391030138e-06,
      "loss": 0.7528,
      "step": 21579
    },
    {
      "epoch": 0.6217945023915173,
      "grad_norm": 0.7070315480232239,
      "learning_rate": 3.305010950906817e-06,
      "loss": 0.6866,
      "step": 21580
    },
    {
      "epoch": 0.6218233158531666,
      "grad_norm": 0.7072806358337402,
      "learning_rate": 3.3045719774722797e-06,
      "loss": 0.6785,
      "step": 21581
    },
    {
      "epoch": 0.6218521293148159,
      "grad_norm": 0.7148173451423645,
      "learning_rate": 3.3041330188032273e-06,
      "loss": 0.7168,
      "step": 21582
    },
    {
      "epoch": 0.6218809427764652,
      "grad_norm": 0.7393915057182312,
      "learning_rate": 3.30369407490348e-06,
      "loss": 0.6924,
      "step": 21583
    },
    {
      "epoch": 0.6219097562381144,
      "grad_norm": 0.7087977528572083,
      "learning_rate": 3.3032551457768626e-06,
      "loss": 0.6648,
      "step": 21584
    },
    {
      "epoch": 0.6219385696997637,
      "grad_norm": 0.7039912939071655,
      "learning_rate": 3.302816231427196e-06,
      "loss": 0.7089,
      "step": 21585
    },
    {
      "epoch": 0.621967383161413,
      "grad_norm": 0.7370463013648987,
      "learning_rate": 3.302377331858302e-06,
      "loss": 0.6948,
      "step": 21586
    },
    {
      "epoch": 0.6219961966230623,
      "grad_norm": 0.7266747951507568,
      "learning_rate": 3.3019384470740056e-06,
      "loss": 0.6772,
      "step": 21587
    },
    {
      "epoch": 0.6220250100847116,
      "grad_norm": 0.7184179425239563,
      "learning_rate": 3.3014995770781287e-06,
      "loss": 0.6901,
      "step": 21588
    },
    {
      "epoch": 0.6220538235463609,
      "grad_norm": 0.7447842359542847,
      "learning_rate": 3.3010607218744927e-06,
      "loss": 0.7197,
      "step": 21589
    },
    {
      "epoch": 0.6220826370080101,
      "grad_norm": 0.7363750338554382,
      "learning_rate": 3.3006218814669187e-06,
      "loss": 0.6971,
      "step": 21590
    },
    {
      "epoch": 0.6221114504696594,
      "grad_norm": 0.7215992212295532,
      "learning_rate": 3.30018305585923e-06,
      "loss": 0.6877,
      "step": 21591
    },
    {
      "epoch": 0.6221402639313087,
      "grad_norm": 0.7229230403900146,
      "learning_rate": 3.2997442450552475e-06,
      "loss": 0.6803,
      "step": 21592
    },
    {
      "epoch": 0.622169077392958,
      "grad_norm": 0.7300748825073242,
      "learning_rate": 3.2993054490587926e-06,
      "loss": 0.6856,
      "step": 21593
    },
    {
      "epoch": 0.6221978908546073,
      "grad_norm": 0.7612304091453552,
      "learning_rate": 3.298866667873688e-06,
      "loss": 0.6829,
      "step": 21594
    },
    {
      "epoch": 0.6222267043162566,
      "grad_norm": 0.7151647806167603,
      "learning_rate": 3.298427901503754e-06,
      "loss": 0.7293,
      "step": 21595
    },
    {
      "epoch": 0.6222555177779059,
      "grad_norm": 0.7665506601333618,
      "learning_rate": 3.2979891499528116e-06,
      "loss": 0.7053,
      "step": 21596
    },
    {
      "epoch": 0.6222843312395551,
      "grad_norm": 0.7313196063041687,
      "learning_rate": 3.2975504132246814e-06,
      "loss": 0.6843,
      "step": 21597
    },
    {
      "epoch": 0.6223131447012044,
      "grad_norm": 0.7080697417259216,
      "learning_rate": 3.297111691323186e-06,
      "loss": 0.6965,
      "step": 21598
    },
    {
      "epoch": 0.6223419581628536,
      "grad_norm": 0.7004972100257874,
      "learning_rate": 3.2966729842521456e-06,
      "loss": 0.7072,
      "step": 21599
    },
    {
      "epoch": 0.6223707716245029,
      "grad_norm": 0.769727885723114,
      "learning_rate": 3.296234292015381e-06,
      "loss": 0.7007,
      "step": 21600
    },
    {
      "epoch": 0.6223995850861522,
      "grad_norm": 0.7192439436912537,
      "learning_rate": 3.2957956146167127e-06,
      "loss": 0.6804,
      "step": 21601
    },
    {
      "epoch": 0.6224283985478015,
      "grad_norm": 0.7690730690956116,
      "learning_rate": 3.2953569520599606e-06,
      "loss": 0.6832,
      "step": 21602
    },
    {
      "epoch": 0.6224572120094508,
      "grad_norm": 0.739507257938385,
      "learning_rate": 3.2949183043489453e-06,
      "loss": 0.7073,
      "step": 21603
    },
    {
      "epoch": 0.6224860254711001,
      "grad_norm": 0.7238263487815857,
      "learning_rate": 3.294479671487488e-06,
      "loss": 0.7053,
      "step": 21604
    },
    {
      "epoch": 0.6225148389327494,
      "grad_norm": 0.7245251536369324,
      "learning_rate": 3.294041053479408e-06,
      "loss": 0.7081,
      "step": 21605
    },
    {
      "epoch": 0.6225436523943987,
      "grad_norm": 0.7122620344161987,
      "learning_rate": 3.2936024503285237e-06,
      "loss": 0.6756,
      "step": 21606
    },
    {
      "epoch": 0.622572465856048,
      "grad_norm": 0.7161053419113159,
      "learning_rate": 3.2931638620386574e-06,
      "loss": 0.6796,
      "step": 21607
    },
    {
      "epoch": 0.6226012793176973,
      "grad_norm": 0.7702913284301758,
      "learning_rate": 3.2927252886136276e-06,
      "loss": 0.7067,
      "step": 21608
    },
    {
      "epoch": 0.6226300927793466,
      "grad_norm": 0.739345908164978,
      "learning_rate": 3.2922867300572537e-06,
      "loss": 0.7078,
      "step": 21609
    },
    {
      "epoch": 0.6226589062409958,
      "grad_norm": 0.7164736390113831,
      "learning_rate": 3.291848186373353e-06,
      "loss": 0.709,
      "step": 21610
    },
    {
      "epoch": 0.622687719702645,
      "grad_norm": 0.7228005528450012,
      "learning_rate": 3.2914096575657495e-06,
      "loss": 0.6984,
      "step": 21611
    },
    {
      "epoch": 0.6227165331642943,
      "grad_norm": 0.7055959105491638,
      "learning_rate": 3.2909711436382596e-06,
      "loss": 0.6761,
      "step": 21612
    },
    {
      "epoch": 0.6227453466259436,
      "grad_norm": 0.7143580317497253,
      "learning_rate": 3.2905326445947026e-06,
      "loss": 0.7167,
      "step": 21613
    },
    {
      "epoch": 0.6227741600875929,
      "grad_norm": 0.7086000442504883,
      "learning_rate": 3.2900941604388982e-06,
      "loss": 0.6934,
      "step": 21614
    },
    {
      "epoch": 0.6228029735492422,
      "grad_norm": 0.7257561087608337,
      "learning_rate": 3.2896556911746636e-06,
      "loss": 0.6727,
      "step": 21615
    },
    {
      "epoch": 0.6228317870108915,
      "grad_norm": 0.7276880145072937,
      "learning_rate": 3.2892172368058185e-06,
      "loss": 0.6989,
      "step": 21616
    },
    {
      "epoch": 0.6228606004725408,
      "grad_norm": 0.727941632270813,
      "learning_rate": 3.2887787973361813e-06,
      "loss": 0.6932,
      "step": 21617
    },
    {
      "epoch": 0.6228894139341901,
      "grad_norm": 0.7174149751663208,
      "learning_rate": 3.2883403727695695e-06,
      "loss": 0.6906,
      "step": 21618
    },
    {
      "epoch": 0.6229182273958394,
      "grad_norm": 0.7068660855293274,
      "learning_rate": 3.287901963109802e-06,
      "loss": 0.6861,
      "step": 21619
    },
    {
      "epoch": 0.6229470408574886,
      "grad_norm": 0.7577401995658875,
      "learning_rate": 3.287463568360697e-06,
      "loss": 0.6998,
      "step": 21620
    },
    {
      "epoch": 0.6229758543191379,
      "grad_norm": 0.7286300659179688,
      "learning_rate": 3.287025188526073e-06,
      "loss": 0.6798,
      "step": 21621
    },
    {
      "epoch": 0.6230046677807872,
      "grad_norm": 0.777375340461731,
      "learning_rate": 3.2865868236097457e-06,
      "loss": 0.6902,
      "step": 21622
    },
    {
      "epoch": 0.6230334812424365,
      "grad_norm": 0.7091740369796753,
      "learning_rate": 3.2861484736155356e-06,
      "loss": 0.7201,
      "step": 21623
    },
    {
      "epoch": 0.6230622947040857,
      "grad_norm": 0.7101147174835205,
      "learning_rate": 3.285710138547259e-06,
      "loss": 0.6815,
      "step": 21624
    },
    {
      "epoch": 0.623091108165735,
      "grad_norm": 0.7342758774757385,
      "learning_rate": 3.285271818408733e-06,
      "loss": 0.6852,
      "step": 21625
    },
    {
      "epoch": 0.6231199216273843,
      "grad_norm": 0.7168891429901123,
      "learning_rate": 3.284833513203776e-06,
      "loss": 0.723,
      "step": 21626
    },
    {
      "epoch": 0.6231487350890336,
      "grad_norm": 0.7303770184516907,
      "learning_rate": 3.284395222936205e-06,
      "loss": 0.6873,
      "step": 21627
    },
    {
      "epoch": 0.6231775485506829,
      "grad_norm": 0.711757481098175,
      "learning_rate": 3.283956947609836e-06,
      "loss": 0.698,
      "step": 21628
    },
    {
      "epoch": 0.6232063620123321,
      "grad_norm": 0.7170360684394836,
      "learning_rate": 3.283518687228487e-06,
      "loss": 0.6958,
      "step": 21629
    },
    {
      "epoch": 0.6232351754739814,
      "grad_norm": 0.7046148777008057,
      "learning_rate": 3.283080441795974e-06,
      "loss": 0.6858,
      "step": 21630
    },
    {
      "epoch": 0.6232639889356307,
      "grad_norm": 0.7209999561309814,
      "learning_rate": 3.282642211316114e-06,
      "loss": 0.701,
      "step": 21631
    },
    {
      "epoch": 0.62329280239728,
      "grad_norm": 0.7505565285682678,
      "learning_rate": 3.2822039957927245e-06,
      "loss": 0.7203,
      "step": 21632
    },
    {
      "epoch": 0.6233216158589293,
      "grad_norm": 0.7074000239372253,
      "learning_rate": 3.2817657952296202e-06,
      "loss": 0.6939,
      "step": 21633
    },
    {
      "epoch": 0.6233504293205786,
      "grad_norm": 0.7235998511314392,
      "learning_rate": 3.2813276096306155e-06,
      "loss": 0.7048,
      "step": 21634
    },
    {
      "epoch": 0.6233792427822279,
      "grad_norm": 0.7058917284011841,
      "learning_rate": 3.2808894389995326e-06,
      "loss": 0.6992,
      "step": 21635
    },
    {
      "epoch": 0.6234080562438772,
      "grad_norm": 0.7013494968414307,
      "learning_rate": 3.2804512833401834e-06,
      "loss": 0.7017,
      "step": 21636
    },
    {
      "epoch": 0.6234368697055265,
      "grad_norm": 0.7100680470466614,
      "learning_rate": 3.2800131426563853e-06,
      "loss": 0.6852,
      "step": 21637
    },
    {
      "epoch": 0.6234656831671757,
      "grad_norm": 0.7101079225540161,
      "learning_rate": 3.279575016951953e-06,
      "loss": 0.6895,
      "step": 21638
    },
    {
      "epoch": 0.623494496628825,
      "grad_norm": 0.7140390872955322,
      "learning_rate": 3.279136906230703e-06,
      "loss": 0.7027,
      "step": 21639
    },
    {
      "epoch": 0.6235233100904742,
      "grad_norm": 0.7357348203659058,
      "learning_rate": 3.2786988104964483e-06,
      "loss": 0.7054,
      "step": 21640
    },
    {
      "epoch": 0.6235521235521235,
      "grad_norm": 0.7356390953063965,
      "learning_rate": 3.278260729753008e-06,
      "loss": 0.7153,
      "step": 21641
    },
    {
      "epoch": 0.6235809370137728,
      "grad_norm": 0.7410369515419006,
      "learning_rate": 3.2778226640041956e-06,
      "loss": 0.7057,
      "step": 21642
    },
    {
      "epoch": 0.6236097504754221,
      "grad_norm": 0.7101755738258362,
      "learning_rate": 3.2773846132538254e-06,
      "loss": 0.6905,
      "step": 21643
    },
    {
      "epoch": 0.6236385639370714,
      "grad_norm": 0.7170816659927368,
      "learning_rate": 3.2769465775057145e-06,
      "loss": 0.6766,
      "step": 21644
    },
    {
      "epoch": 0.6236673773987207,
      "grad_norm": 0.728279173374176,
      "learning_rate": 3.2765085567636756e-06,
      "loss": 0.7097,
      "step": 21645
    },
    {
      "epoch": 0.62369619086037,
      "grad_norm": 0.7337941527366638,
      "learning_rate": 3.2760705510315226e-06,
      "loss": 0.7166,
      "step": 21646
    },
    {
      "epoch": 0.6237250043220193,
      "grad_norm": 0.7114896178245544,
      "learning_rate": 3.2756325603130744e-06,
      "loss": 0.694,
      "step": 21647
    },
    {
      "epoch": 0.6237538177836686,
      "grad_norm": 0.748741865158081,
      "learning_rate": 3.2751945846121425e-06,
      "loss": 0.7281,
      "step": 21648
    },
    {
      "epoch": 0.6237826312453179,
      "grad_norm": 0.717163622379303,
      "learning_rate": 3.274756623932541e-06,
      "loss": 0.702,
      "step": 21649
    },
    {
      "epoch": 0.6238114447069671,
      "grad_norm": 0.711381733417511,
      "learning_rate": 3.2743186782780855e-06,
      "loss": 0.7056,
      "step": 21650
    },
    {
      "epoch": 0.6238402581686163,
      "grad_norm": 0.7294676899909973,
      "learning_rate": 3.2738807476525887e-06,
      "loss": 0.6888,
      "step": 21651
    },
    {
      "epoch": 0.6238690716302656,
      "grad_norm": 0.7332893013954163,
      "learning_rate": 3.2734428320598656e-06,
      "loss": 0.7172,
      "step": 21652
    },
    {
      "epoch": 0.6238978850919149,
      "grad_norm": 0.7249587774276733,
      "learning_rate": 3.273004931503729e-06,
      "loss": 0.708,
      "step": 21653
    },
    {
      "epoch": 0.6239266985535642,
      "grad_norm": 0.7173547744750977,
      "learning_rate": 3.2725670459879932e-06,
      "loss": 0.6993,
      "step": 21654
    },
    {
      "epoch": 0.6239555120152135,
      "grad_norm": 0.7248523831367493,
      "learning_rate": 3.272129175516471e-06,
      "loss": 0.6964,
      "step": 21655
    },
    {
      "epoch": 0.6239843254768628,
      "grad_norm": 0.7086447477340698,
      "learning_rate": 3.2716913200929778e-06,
      "loss": 0.6767,
      "step": 21656
    },
    {
      "epoch": 0.6240131389385121,
      "grad_norm": 0.7090719938278198,
      "learning_rate": 3.2712534797213238e-06,
      "loss": 0.6785,
      "step": 21657
    },
    {
      "epoch": 0.6240419524001614,
      "grad_norm": 0.7231711149215698,
      "learning_rate": 3.270815654405324e-06,
      "loss": 0.6921,
      "step": 21658
    },
    {
      "epoch": 0.6240707658618107,
      "grad_norm": 0.7144010066986084,
      "learning_rate": 3.270377844148792e-06,
      "loss": 0.6813,
      "step": 21659
    },
    {
      "epoch": 0.6240995793234599,
      "grad_norm": 0.7153533697128296,
      "learning_rate": 3.26994004895554e-06,
      "loss": 0.7076,
      "step": 21660
    },
    {
      "epoch": 0.6241283927851092,
      "grad_norm": 0.7218505144119263,
      "learning_rate": 3.2695022688293807e-06,
      "loss": 0.7014,
      "step": 21661
    },
    {
      "epoch": 0.6241572062467585,
      "grad_norm": 0.7319545745849609,
      "learning_rate": 3.2690645037741265e-06,
      "loss": 0.6962,
      "step": 21662
    },
    {
      "epoch": 0.6241860197084078,
      "grad_norm": 0.7300893068313599,
      "learning_rate": 3.2686267537935906e-06,
      "loss": 0.7023,
      "step": 21663
    },
    {
      "epoch": 0.6242148331700571,
      "grad_norm": 0.7170068621635437,
      "learning_rate": 3.2681890188915853e-06,
      "loss": 0.7211,
      "step": 21664
    },
    {
      "epoch": 0.6242436466317063,
      "grad_norm": 0.7024499177932739,
      "learning_rate": 3.267751299071922e-06,
      "loss": 0.6914,
      "step": 21665
    },
    {
      "epoch": 0.6242724600933556,
      "grad_norm": 0.707217812538147,
      "learning_rate": 3.267313594338414e-06,
      "loss": 0.6481,
      "step": 21666
    },
    {
      "epoch": 0.6243012735550049,
      "grad_norm": 0.7038770318031311,
      "learning_rate": 3.266875904694872e-06,
      "loss": 0.6854,
      "step": 21667
    },
    {
      "epoch": 0.6243300870166542,
      "grad_norm": 0.7039754390716553,
      "learning_rate": 3.266438230145108e-06,
      "loss": 0.6923,
      "step": 21668
    },
    {
      "epoch": 0.6243589004783034,
      "grad_norm": 0.7763298749923706,
      "learning_rate": 3.2660005706929343e-06,
      "loss": 0.7137,
      "step": 21669
    },
    {
      "epoch": 0.6243877139399527,
      "grad_norm": 0.7198110818862915,
      "learning_rate": 3.2655629263421608e-06,
      "loss": 0.7029,
      "step": 21670
    },
    {
      "epoch": 0.624416527401602,
      "grad_norm": 0.726824939250946,
      "learning_rate": 3.265125297096602e-06,
      "loss": 0.7212,
      "step": 21671
    },
    {
      "epoch": 0.6244453408632513,
      "grad_norm": 0.709557056427002,
      "learning_rate": 3.2646876829600672e-06,
      "loss": 0.6918,
      "step": 21672
    },
    {
      "epoch": 0.6244741543249006,
      "grad_norm": 0.7705503702163696,
      "learning_rate": 3.2642500839363688e-06,
      "loss": 0.7231,
      "step": 21673
    },
    {
      "epoch": 0.6245029677865499,
      "grad_norm": 0.7477560639381409,
      "learning_rate": 3.2638125000293163e-06,
      "loss": 0.6953,
      "step": 21674
    },
    {
      "epoch": 0.6245317812481992,
      "grad_norm": 0.7134122252464294,
      "learning_rate": 3.263374931242722e-06,
      "loss": 0.6953,
      "step": 21675
    },
    {
      "epoch": 0.6245605947098485,
      "grad_norm": 0.7282791137695312,
      "learning_rate": 3.2629373775803964e-06,
      "loss": 0.7101,
      "step": 21676
    },
    {
      "epoch": 0.6245894081714978,
      "grad_norm": 0.7149183750152588,
      "learning_rate": 3.2624998390461494e-06,
      "loss": 0.6849,
      "step": 21677
    },
    {
      "epoch": 0.624618221633147,
      "grad_norm": 0.7010897397994995,
      "learning_rate": 3.26206231564379e-06,
      "loss": 0.6976,
      "step": 21678
    },
    {
      "epoch": 0.6246470350947962,
      "grad_norm": 0.7075837850570679,
      "learning_rate": 3.2616248073771327e-06,
      "loss": 0.7003,
      "step": 21679
    },
    {
      "epoch": 0.6246758485564455,
      "grad_norm": 0.7198524475097656,
      "learning_rate": 3.261187314249985e-06,
      "loss": 0.6954,
      "step": 21680
    },
    {
      "epoch": 0.6247046620180948,
      "grad_norm": 0.7184561491012573,
      "learning_rate": 3.2607498362661584e-06,
      "loss": 0.7209,
      "step": 21681
    },
    {
      "epoch": 0.6247334754797441,
      "grad_norm": 0.7141388058662415,
      "learning_rate": 3.2603123734294608e-06,
      "loss": 0.699,
      "step": 21682
    },
    {
      "epoch": 0.6247622889413934,
      "grad_norm": 0.7252917885780334,
      "learning_rate": 3.2598749257437025e-06,
      "loss": 0.6825,
      "step": 21683
    },
    {
      "epoch": 0.6247911024030427,
      "grad_norm": 0.7244263291358948,
      "learning_rate": 3.259437493212696e-06,
      "loss": 0.7198,
      "step": 21684
    },
    {
      "epoch": 0.624819915864692,
      "grad_norm": 0.7431215643882751,
      "learning_rate": 3.2590000758402486e-06,
      "loss": 0.7072,
      "step": 21685
    },
    {
      "epoch": 0.6248487293263413,
      "grad_norm": 0.7348049879074097,
      "learning_rate": 3.25856267363017e-06,
      "loss": 0.7252,
      "step": 21686
    },
    {
      "epoch": 0.6248775427879906,
      "grad_norm": 0.7256788611412048,
      "learning_rate": 3.25812528658627e-06,
      "loss": 0.7151,
      "step": 21687
    },
    {
      "epoch": 0.6249063562496399,
      "grad_norm": 0.7232823371887207,
      "learning_rate": 3.257687914712358e-06,
      "loss": 0.7122,
      "step": 21688
    },
    {
      "epoch": 0.6249351697112892,
      "grad_norm": 0.7529544830322266,
      "learning_rate": 3.2572505580122423e-06,
      "loss": 0.7142,
      "step": 21689
    },
    {
      "epoch": 0.6249639831729384,
      "grad_norm": 0.7320842146873474,
      "learning_rate": 3.2568132164897316e-06,
      "loss": 0.7161,
      "step": 21690
    },
    {
      "epoch": 0.6249927966345877,
      "grad_norm": 0.719368577003479,
      "learning_rate": 3.256375890148636e-06,
      "loss": 0.6996,
      "step": 21691
    },
    {
      "epoch": 0.6250216100962369,
      "grad_norm": 0.7676718235015869,
      "learning_rate": 3.2559385789927633e-06,
      "loss": 0.7062,
      "step": 21692
    },
    {
      "epoch": 0.6250504235578862,
      "grad_norm": 0.7393139004707336,
      "learning_rate": 3.2555012830259214e-06,
      "loss": 0.6991,
      "step": 21693
    },
    {
      "epoch": 0.6250792370195355,
      "grad_norm": 0.7202668190002441,
      "learning_rate": 3.2550640022519182e-06,
      "loss": 0.7091,
      "step": 21694
    },
    {
      "epoch": 0.6251080504811848,
      "grad_norm": 0.7327796816825867,
      "learning_rate": 3.254626736674565e-06,
      "loss": 0.6934,
      "step": 21695
    },
    {
      "epoch": 0.6251368639428341,
      "grad_norm": 0.7157266736030579,
      "learning_rate": 3.254189486297668e-06,
      "loss": 0.7115,
      "step": 21696
    },
    {
      "epoch": 0.6251656774044834,
      "grad_norm": 0.7310460805892944,
      "learning_rate": 3.2537522511250354e-06,
      "loss": 0.7103,
      "step": 21697
    },
    {
      "epoch": 0.6251944908661327,
      "grad_norm": 0.7138709425926208,
      "learning_rate": 3.2533150311604757e-06,
      "loss": 0.7115,
      "step": 21698
    },
    {
      "epoch": 0.625223304327782,
      "grad_norm": 0.749581515789032,
      "learning_rate": 3.2528778264077943e-06,
      "loss": 0.7346,
      "step": 21699
    },
    {
      "epoch": 0.6252521177894312,
      "grad_norm": 0.7344729900360107,
      "learning_rate": 3.2524406368708015e-06,
      "loss": 0.7033,
      "step": 21700
    },
    {
      "epoch": 0.6252809312510805,
      "grad_norm": 0.7344440817832947,
      "learning_rate": 3.252003462553305e-06,
      "loss": 0.6945,
      "step": 21701
    },
    {
      "epoch": 0.6253097447127298,
      "grad_norm": 0.7014897465705872,
      "learning_rate": 3.2515663034591094e-06,
      "loss": 0.7006,
      "step": 21702
    },
    {
      "epoch": 0.6253385581743791,
      "grad_norm": 0.7228469252586365,
      "learning_rate": 3.251129159592025e-06,
      "loss": 0.699,
      "step": 21703
    },
    {
      "epoch": 0.6253673716360284,
      "grad_norm": 0.7301709651947021,
      "learning_rate": 3.2506920309558565e-06,
      "loss": 0.7067,
      "step": 21704
    },
    {
      "epoch": 0.6253961850976776,
      "grad_norm": 0.7006132006645203,
      "learning_rate": 3.2502549175544117e-06,
      "loss": 0.7018,
      "step": 21705
    },
    {
      "epoch": 0.6254249985593269,
      "grad_norm": 0.7310788631439209,
      "learning_rate": 3.2498178193914954e-06,
      "loss": 0.7102,
      "step": 21706
    },
    {
      "epoch": 0.6254538120209762,
      "grad_norm": 0.7337926626205444,
      "learning_rate": 3.2493807364709183e-06,
      "loss": 0.6961,
      "step": 21707
    },
    {
      "epoch": 0.6254826254826255,
      "grad_norm": 0.7214386463165283,
      "learning_rate": 3.2489436687964858e-06,
      "loss": 0.7125,
      "step": 21708
    },
    {
      "epoch": 0.6255114389442747,
      "grad_norm": 0.7280741930007935,
      "learning_rate": 3.248506616372003e-06,
      "loss": 0.7021,
      "step": 21709
    },
    {
      "epoch": 0.625540252405924,
      "grad_norm": 0.7184501886367798,
      "learning_rate": 3.2480695792012772e-06,
      "loss": 0.7114,
      "step": 21710
    },
    {
      "epoch": 0.6255690658675733,
      "grad_norm": 0.7147149443626404,
      "learning_rate": 3.2476325572881135e-06,
      "loss": 0.6978,
      "step": 21711
    },
    {
      "epoch": 0.6255978793292226,
      "grad_norm": 0.7177362442016602,
      "learning_rate": 3.2471955506363184e-06,
      "loss": 0.6924,
      "step": 21712
    },
    {
      "epoch": 0.6256266927908719,
      "grad_norm": 0.7155589461326599,
      "learning_rate": 3.2467585592496986e-06,
      "loss": 0.7088,
      "step": 21713
    },
    {
      "epoch": 0.6256555062525212,
      "grad_norm": 0.7258400321006775,
      "learning_rate": 3.246321583132058e-06,
      "loss": 0.7104,
      "step": 21714
    },
    {
      "epoch": 0.6256843197141705,
      "grad_norm": 0.7078902125358582,
      "learning_rate": 3.2458846222872043e-06,
      "loss": 0.6847,
      "step": 21715
    },
    {
      "epoch": 0.6257131331758198,
      "grad_norm": 0.7411177754402161,
      "learning_rate": 3.245447676718941e-06,
      "loss": 0.6904,
      "step": 21716
    },
    {
      "epoch": 0.6257419466374691,
      "grad_norm": 0.7270800471305847,
      "learning_rate": 3.2450107464310755e-06,
      "loss": 0.6983,
      "step": 21717
    },
    {
      "epoch": 0.6257707600991184,
      "grad_norm": 0.7188611030578613,
      "learning_rate": 3.2445738314274106e-06,
      "loss": 0.7035,
      "step": 21718
    },
    {
      "epoch": 0.6257995735607675,
      "grad_norm": 0.7346842288970947,
      "learning_rate": 3.2441369317117544e-06,
      "loss": 0.7025,
      "step": 21719
    },
    {
      "epoch": 0.6258283870224168,
      "grad_norm": 0.7229819893836975,
      "learning_rate": 3.2437000472879097e-06,
      "loss": 0.6998,
      "step": 21720
    },
    {
      "epoch": 0.6258572004840661,
      "grad_norm": 0.7225212454795837,
      "learning_rate": 3.2432631781596814e-06,
      "loss": 0.7027,
      "step": 21721
    },
    {
      "epoch": 0.6258860139457154,
      "grad_norm": 0.7303789258003235,
      "learning_rate": 3.2428263243308755e-06,
      "loss": 0.6887,
      "step": 21722
    },
    {
      "epoch": 0.6259148274073647,
      "grad_norm": 0.7313309907913208,
      "learning_rate": 3.242389485805296e-06,
      "loss": 0.6978,
      "step": 21723
    },
    {
      "epoch": 0.625943640869014,
      "grad_norm": 0.6856455206871033,
      "learning_rate": 3.241952662586747e-06,
      "loss": 0.6906,
      "step": 21724
    },
    {
      "epoch": 0.6259724543306633,
      "grad_norm": 0.7329717874526978,
      "learning_rate": 3.2415158546790328e-06,
      "loss": 0.7358,
      "step": 21725
    },
    {
      "epoch": 0.6260012677923126,
      "grad_norm": 0.7331900000572205,
      "learning_rate": 3.2410790620859576e-06,
      "loss": 0.6726,
      "step": 21726
    },
    {
      "epoch": 0.6260300812539619,
      "grad_norm": 0.7638471126556396,
      "learning_rate": 3.2406422848113263e-06,
      "loss": 0.7239,
      "step": 21727
    },
    {
      "epoch": 0.6260588947156112,
      "grad_norm": 0.7273603677749634,
      "learning_rate": 3.240205522858941e-06,
      "loss": 0.7115,
      "step": 21728
    },
    {
      "epoch": 0.6260877081772604,
      "grad_norm": 0.7452516555786133,
      "learning_rate": 3.239768776232607e-06,
      "loss": 0.7207,
      "step": 21729
    },
    {
      "epoch": 0.6261165216389097,
      "grad_norm": 0.7233152389526367,
      "learning_rate": 3.239332044936125e-06,
      "loss": 0.6945,
      "step": 21730
    },
    {
      "epoch": 0.626145335100559,
      "grad_norm": 0.7509958148002625,
      "learning_rate": 3.2388953289733027e-06,
      "loss": 0.6963,
      "step": 21731
    },
    {
      "epoch": 0.6261741485622082,
      "grad_norm": 0.7209031581878662,
      "learning_rate": 3.2384586283479423e-06,
      "loss": 0.7006,
      "step": 21732
    },
    {
      "epoch": 0.6262029620238575,
      "grad_norm": 0.7317453622817993,
      "learning_rate": 3.2380219430638458e-06,
      "loss": 0.6722,
      "step": 21733
    },
    {
      "epoch": 0.6262317754855068,
      "grad_norm": 0.7064304351806641,
      "learning_rate": 3.237585273124817e-06,
      "loss": 0.6964,
      "step": 21734
    },
    {
      "epoch": 0.6262605889471561,
      "grad_norm": 0.738244891166687,
      "learning_rate": 3.237148618534659e-06,
      "loss": 0.708,
      "step": 21735
    },
    {
      "epoch": 0.6262894024088054,
      "grad_norm": 0.7105793952941895,
      "learning_rate": 3.236711979297174e-06,
      "loss": 0.6996,
      "step": 21736
    },
    {
      "epoch": 0.6263182158704547,
      "grad_norm": 0.7238679528236389,
      "learning_rate": 3.236275355416164e-06,
      "loss": 0.6828,
      "step": 21737
    },
    {
      "epoch": 0.626347029332104,
      "grad_norm": 0.7350130677223206,
      "learning_rate": 3.235838746895434e-06,
      "loss": 0.7131,
      "step": 21738
    },
    {
      "epoch": 0.6263758427937532,
      "grad_norm": 0.7119593024253845,
      "learning_rate": 3.2354021537387847e-06,
      "loss": 0.6942,
      "step": 21739
    },
    {
      "epoch": 0.6264046562554025,
      "grad_norm": 0.7461533546447754,
      "learning_rate": 3.2349655759500187e-06,
      "loss": 0.6827,
      "step": 21740
    },
    {
      "epoch": 0.6264334697170518,
      "grad_norm": 0.718855619430542,
      "learning_rate": 3.2345290135329377e-06,
      "loss": 0.6938,
      "step": 21741
    },
    {
      "epoch": 0.6264622831787011,
      "grad_norm": 0.6897704601287842,
      "learning_rate": 3.2340924664913427e-06,
      "loss": 0.7236,
      "step": 21742
    },
    {
      "epoch": 0.6264910966403504,
      "grad_norm": 0.7060372233390808,
      "learning_rate": 3.2336559348290387e-06,
      "loss": 0.7043,
      "step": 21743
    },
    {
      "epoch": 0.6265199101019997,
      "grad_norm": 0.719954788684845,
      "learning_rate": 3.233219418549826e-06,
      "loss": 0.6973,
      "step": 21744
    },
    {
      "epoch": 0.626548723563649,
      "grad_norm": 0.7146167159080505,
      "learning_rate": 3.2327829176575063e-06,
      "loss": 0.6876,
      "step": 21745
    },
    {
      "epoch": 0.6265775370252982,
      "grad_norm": 0.7396094799041748,
      "learning_rate": 3.2323464321558806e-06,
      "loss": 0.7021,
      "step": 21746
    },
    {
      "epoch": 0.6266063504869475,
      "grad_norm": 0.7089507579803467,
      "learning_rate": 3.2319099620487504e-06,
      "loss": 0.7207,
      "step": 21747
    },
    {
      "epoch": 0.6266351639485968,
      "grad_norm": 0.7220468521118164,
      "learning_rate": 3.2314735073399173e-06,
      "loss": 0.6947,
      "step": 21748
    },
    {
      "epoch": 0.626663977410246,
      "grad_norm": 0.689550518989563,
      "learning_rate": 3.2310370680331815e-06,
      "loss": 0.6824,
      "step": 21749
    },
    {
      "epoch": 0.6266927908718953,
      "grad_norm": 0.7131873369216919,
      "learning_rate": 3.2306006441323445e-06,
      "loss": 0.6876,
      "step": 21750
    },
    {
      "epoch": 0.6267216043335446,
      "grad_norm": 0.7683696746826172,
      "learning_rate": 3.2301642356412077e-06,
      "loss": 0.7096,
      "step": 21751
    },
    {
      "epoch": 0.6267504177951939,
      "grad_norm": 0.6970458030700684,
      "learning_rate": 3.2297278425635714e-06,
      "loss": 0.6872,
      "step": 21752
    },
    {
      "epoch": 0.6267792312568432,
      "grad_norm": 0.7214381694793701,
      "learning_rate": 3.2292914649032354e-06,
      "loss": 0.6935,
      "step": 21753
    },
    {
      "epoch": 0.6268080447184925,
      "grad_norm": 0.7127475142478943,
      "learning_rate": 3.2288551026639985e-06,
      "loss": 0.6982,
      "step": 21754
    },
    {
      "epoch": 0.6268368581801418,
      "grad_norm": 0.7604002356529236,
      "learning_rate": 3.228418755849666e-06,
      "loss": 0.7009,
      "step": 21755
    },
    {
      "epoch": 0.6268656716417911,
      "grad_norm": 0.7241851091384888,
      "learning_rate": 3.2279824244640343e-06,
      "loss": 0.7135,
      "step": 21756
    },
    {
      "epoch": 0.6268944851034404,
      "grad_norm": 0.7193267345428467,
      "learning_rate": 3.227546108510905e-06,
      "loss": 0.7157,
      "step": 21757
    },
    {
      "epoch": 0.6269232985650897,
      "grad_norm": 0.7133544683456421,
      "learning_rate": 3.2271098079940767e-06,
      "loss": 0.6764,
      "step": 21758
    },
    {
      "epoch": 0.6269521120267388,
      "grad_norm": 0.722250759601593,
      "learning_rate": 3.2266735229173492e-06,
      "loss": 0.669,
      "step": 21759
    },
    {
      "epoch": 0.6269809254883881,
      "grad_norm": 0.728356659412384,
      "learning_rate": 3.226237253284523e-06,
      "loss": 0.6976,
      "step": 21760
    },
    {
      "epoch": 0.6270097389500374,
      "grad_norm": 0.7742415070533752,
      "learning_rate": 3.225800999099398e-06,
      "loss": 0.687,
      "step": 21761
    },
    {
      "epoch": 0.6270385524116867,
      "grad_norm": 0.7367445826530457,
      "learning_rate": 3.2253647603657723e-06,
      "loss": 0.7185,
      "step": 21762
    },
    {
      "epoch": 0.627067365873336,
      "grad_norm": 0.6869205832481384,
      "learning_rate": 3.2249285370874463e-06,
      "loss": 0.6848,
      "step": 21763
    },
    {
      "epoch": 0.6270961793349853,
      "grad_norm": 0.7220969200134277,
      "learning_rate": 3.2244923292682175e-06,
      "loss": 0.6761,
      "step": 21764
    },
    {
      "epoch": 0.6271249927966346,
      "grad_norm": 0.7563751339912415,
      "learning_rate": 3.2240561369118863e-06,
      "loss": 0.7083,
      "step": 21765
    },
    {
      "epoch": 0.6271538062582839,
      "grad_norm": 0.7310079336166382,
      "learning_rate": 3.2236199600222477e-06,
      "loss": 0.7028,
      "step": 21766
    },
    {
      "epoch": 0.6271826197199332,
      "grad_norm": 0.7561666369438171,
      "learning_rate": 3.2231837986031063e-06,
      "loss": 0.6999,
      "step": 21767
    },
    {
      "epoch": 0.6272114331815825,
      "grad_norm": 0.7266523241996765,
      "learning_rate": 3.222747652658258e-06,
      "loss": 0.7076,
      "step": 21768
    },
    {
      "epoch": 0.6272402466432317,
      "grad_norm": 0.7083654999732971,
      "learning_rate": 3.2223115221915003e-06,
      "loss": 0.693,
      "step": 21769
    },
    {
      "epoch": 0.627269060104881,
      "grad_norm": 0.7144412994384766,
      "learning_rate": 3.2218754072066327e-06,
      "loss": 0.7037,
      "step": 21770
    },
    {
      "epoch": 0.6272978735665303,
      "grad_norm": 0.7018528580665588,
      "learning_rate": 3.2214393077074535e-06,
      "loss": 0.7066,
      "step": 21771
    },
    {
      "epoch": 0.6273266870281796,
      "grad_norm": 0.7212969660758972,
      "learning_rate": 3.221003223697759e-06,
      "loss": 0.7011,
      "step": 21772
    },
    {
      "epoch": 0.6273555004898288,
      "grad_norm": 0.7322787046432495,
      "learning_rate": 3.220567155181349e-06,
      "loss": 0.7211,
      "step": 21773
    },
    {
      "epoch": 0.6273843139514781,
      "grad_norm": 0.7178192138671875,
      "learning_rate": 3.2201311021620195e-06,
      "loss": 0.7162,
      "step": 21774
    },
    {
      "epoch": 0.6274131274131274,
      "grad_norm": 0.7121596932411194,
      "learning_rate": 3.2196950646435677e-06,
      "loss": 0.6899,
      "step": 21775
    },
    {
      "epoch": 0.6274419408747767,
      "grad_norm": 0.7124951481819153,
      "learning_rate": 3.2192590426297937e-06,
      "loss": 0.701,
      "step": 21776
    },
    {
      "epoch": 0.627470754336426,
      "grad_norm": 0.7282759547233582,
      "learning_rate": 3.2188230361244932e-06,
      "loss": 0.7083,
      "step": 21777
    },
    {
      "epoch": 0.6274995677980753,
      "grad_norm": 0.7302151322364807,
      "learning_rate": 3.2183870451314624e-06,
      "loss": 0.695,
      "step": 21778
    },
    {
      "epoch": 0.6275283812597245,
      "grad_norm": 0.7173922657966614,
      "learning_rate": 3.2179510696545013e-06,
      "loss": 0.7134,
      "step": 21779
    },
    {
      "epoch": 0.6275571947213738,
      "grad_norm": 0.7198131680488586,
      "learning_rate": 3.2175151096974035e-06,
      "loss": 0.7121,
      "step": 21780
    },
    {
      "epoch": 0.6275860081830231,
      "grad_norm": 0.7424886226654053,
      "learning_rate": 3.217079165263968e-06,
      "loss": 0.6999,
      "step": 21781
    },
    {
      "epoch": 0.6276148216446724,
      "grad_norm": 0.7169680595397949,
      "learning_rate": 3.216643236357991e-06,
      "loss": 0.716,
      "step": 21782
    },
    {
      "epoch": 0.6276436351063217,
      "grad_norm": 0.7234470248222351,
      "learning_rate": 3.216207322983268e-06,
      "loss": 0.7042,
      "step": 21783
    },
    {
      "epoch": 0.627672448567971,
      "grad_norm": 0.7192018628120422,
      "learning_rate": 3.215771425143597e-06,
      "loss": 0.7036,
      "step": 21784
    },
    {
      "epoch": 0.6277012620296203,
      "grad_norm": 0.7017211318016052,
      "learning_rate": 3.215335542842773e-06,
      "loss": 0.7143,
      "step": 21785
    },
    {
      "epoch": 0.6277300754912695,
      "grad_norm": 0.7236559391021729,
      "learning_rate": 3.2148996760845923e-06,
      "loss": 0.7183,
      "step": 21786
    },
    {
      "epoch": 0.6277588889529188,
      "grad_norm": 0.7321574091911316,
      "learning_rate": 3.214463824872851e-06,
      "loss": 0.6878,
      "step": 21787
    },
    {
      "epoch": 0.627787702414568,
      "grad_norm": 0.7411795854568481,
      "learning_rate": 3.2140279892113447e-06,
      "loss": 0.678,
      "step": 21788
    },
    {
      "epoch": 0.6278165158762173,
      "grad_norm": 0.7480096817016602,
      "learning_rate": 3.2135921691038694e-06,
      "loss": 0.6795,
      "step": 21789
    },
    {
      "epoch": 0.6278453293378666,
      "grad_norm": 0.7568085193634033,
      "learning_rate": 3.2131563645542193e-06,
      "loss": 0.7193,
      "step": 21790
    },
    {
      "epoch": 0.6278741427995159,
      "grad_norm": 0.713624119758606,
      "learning_rate": 3.2127205755661924e-06,
      "loss": 0.6957,
      "step": 21791
    },
    {
      "epoch": 0.6279029562611652,
      "grad_norm": 0.7238485813140869,
      "learning_rate": 3.2122848021435825e-06,
      "loss": 0.6919,
      "step": 21792
    },
    {
      "epoch": 0.6279317697228145,
      "grad_norm": 0.7245003581047058,
      "learning_rate": 3.211849044290185e-06,
      "loss": 0.7052,
      "step": 21793
    },
    {
      "epoch": 0.6279605831844638,
      "grad_norm": 0.7371245622634888,
      "learning_rate": 3.2114133020097947e-06,
      "loss": 0.7077,
      "step": 21794
    },
    {
      "epoch": 0.6279893966461131,
      "grad_norm": 0.7480674386024475,
      "learning_rate": 3.210977575306208e-06,
      "loss": 0.6811,
      "step": 21795
    },
    {
      "epoch": 0.6280182101077624,
      "grad_norm": 0.7321447730064392,
      "learning_rate": 3.2105418641832164e-06,
      "loss": 0.6905,
      "step": 21796
    },
    {
      "epoch": 0.6280470235694117,
      "grad_norm": 0.7249005436897278,
      "learning_rate": 3.210106168644616e-06,
      "loss": 0.6917,
      "step": 21797
    },
    {
      "epoch": 0.628075837031061,
      "grad_norm": 1.2448562383651733,
      "learning_rate": 3.2096704886942033e-06,
      "loss": 0.7115,
      "step": 21798
    },
    {
      "epoch": 0.6281046504927102,
      "grad_norm": 0.7033872604370117,
      "learning_rate": 3.20923482433577e-06,
      "loss": 0.6834,
      "step": 21799
    },
    {
      "epoch": 0.6281334639543594,
      "grad_norm": 0.743139386177063,
      "learning_rate": 3.208799175573112e-06,
      "loss": 0.7159,
      "step": 21800
    },
    {
      "epoch": 0.6281622774160087,
      "grad_norm": 0.7224945425987244,
      "learning_rate": 3.2083635424100226e-06,
      "loss": 0.699,
      "step": 21801
    },
    {
      "epoch": 0.628191090877658,
      "grad_norm": 0.7117367386817932,
      "learning_rate": 3.2079279248502926e-06,
      "loss": 0.6997,
      "step": 21802
    },
    {
      "epoch": 0.6282199043393073,
      "grad_norm": 0.732175886631012,
      "learning_rate": 3.2074923228977216e-06,
      "loss": 0.7107,
      "step": 21803
    },
    {
      "epoch": 0.6282487178009566,
      "grad_norm": 0.7322614192962646,
      "learning_rate": 3.207056736556101e-06,
      "loss": 0.7168,
      "step": 21804
    },
    {
      "epoch": 0.6282775312626059,
      "grad_norm": 0.7112076878547668,
      "learning_rate": 3.206621165829224e-06,
      "loss": 0.6873,
      "step": 21805
    },
    {
      "epoch": 0.6283063447242552,
      "grad_norm": 0.7389779090881348,
      "learning_rate": 3.206185610720884e-06,
      "loss": 0.672,
      "step": 21806
    },
    {
      "epoch": 0.6283351581859045,
      "grad_norm": 0.7270019054412842,
      "learning_rate": 3.205750071234873e-06,
      "loss": 0.6938,
      "step": 21807
    },
    {
      "epoch": 0.6283639716475538,
      "grad_norm": 0.7156990766525269,
      "learning_rate": 3.205314547374987e-06,
      "loss": 0.7181,
      "step": 21808
    },
    {
      "epoch": 0.628392785109203,
      "grad_norm": 0.7050586938858032,
      "learning_rate": 3.204879039145016e-06,
      "loss": 0.6894,
      "step": 21809
    },
    {
      "epoch": 0.6284215985708523,
      "grad_norm": 0.7226497530937195,
      "learning_rate": 3.2044435465487545e-06,
      "loss": 0.6895,
      "step": 21810
    },
    {
      "epoch": 0.6284504120325016,
      "grad_norm": 0.7268746495246887,
      "learning_rate": 3.204008069589995e-06,
      "loss": 0.6888,
      "step": 21811
    },
    {
      "epoch": 0.6284792254941509,
      "grad_norm": 0.7469171285629272,
      "learning_rate": 3.2035726082725298e-06,
      "loss": 0.7095,
      "step": 21812
    },
    {
      "epoch": 0.6285080389558001,
      "grad_norm": 0.709509015083313,
      "learning_rate": 3.2031371626001506e-06,
      "loss": 0.6789,
      "step": 21813
    },
    {
      "epoch": 0.6285368524174494,
      "grad_norm": 0.7264299392700195,
      "learning_rate": 3.2027017325766496e-06,
      "loss": 0.7031,
      "step": 21814
    },
    {
      "epoch": 0.6285656658790987,
      "grad_norm": 0.7304625511169434,
      "learning_rate": 3.2022663182058216e-06,
      "loss": 0.7021,
      "step": 21815
    },
    {
      "epoch": 0.628594479340748,
      "grad_norm": 0.7120419144630432,
      "learning_rate": 3.201830919491457e-06,
      "loss": 0.697,
      "step": 21816
    },
    {
      "epoch": 0.6286232928023973,
      "grad_norm": 0.7085679769515991,
      "learning_rate": 3.201395536437347e-06,
      "loss": 0.7098,
      "step": 21817
    },
    {
      "epoch": 0.6286521062640466,
      "grad_norm": 0.7211878299713135,
      "learning_rate": 3.2009601690472837e-06,
      "loss": 0.6949,
      "step": 21818
    },
    {
      "epoch": 0.6286809197256958,
      "grad_norm": 0.7112552523612976,
      "learning_rate": 3.2005248173250593e-06,
      "loss": 0.7172,
      "step": 21819
    },
    {
      "epoch": 0.6287097331873451,
      "grad_norm": 0.7564201951026917,
      "learning_rate": 3.2000894812744653e-06,
      "loss": 0.7064,
      "step": 21820
    },
    {
      "epoch": 0.6287385466489944,
      "grad_norm": 0.709735095500946,
      "learning_rate": 3.1996541608992925e-06,
      "loss": 0.6792,
      "step": 21821
    },
    {
      "epoch": 0.6287673601106437,
      "grad_norm": 0.6975453495979309,
      "learning_rate": 3.199218856203332e-06,
      "loss": 0.6934,
      "step": 21822
    },
    {
      "epoch": 0.628796173572293,
      "grad_norm": 0.7227399349212646,
      "learning_rate": 3.1987835671903757e-06,
      "loss": 0.702,
      "step": 21823
    },
    {
      "epoch": 0.6288249870339423,
      "grad_norm": 0.7346916198730469,
      "learning_rate": 3.1983482938642143e-06,
      "loss": 0.726,
      "step": 21824
    },
    {
      "epoch": 0.6288538004955916,
      "grad_norm": 0.7030662894248962,
      "learning_rate": 3.1979130362286356e-06,
      "loss": 0.7096,
      "step": 21825
    },
    {
      "epoch": 0.6288826139572409,
      "grad_norm": 0.7116145491600037,
      "learning_rate": 3.1974777942874347e-06,
      "loss": 0.6878,
      "step": 21826
    },
    {
      "epoch": 0.6289114274188901,
      "grad_norm": 0.7235264182090759,
      "learning_rate": 3.1970425680444005e-06,
      "loss": 0.6884,
      "step": 21827
    },
    {
      "epoch": 0.6289402408805393,
      "grad_norm": 0.7631891965866089,
      "learning_rate": 3.1966073575033234e-06,
      "loss": 0.6889,
      "step": 21828
    },
    {
      "epoch": 0.6289690543421886,
      "grad_norm": 0.7458696961402893,
      "learning_rate": 3.1961721626679935e-06,
      "loss": 0.695,
      "step": 21829
    },
    {
      "epoch": 0.6289978678038379,
      "grad_norm": 0.7191728353500366,
      "learning_rate": 3.195736983542201e-06,
      "loss": 0.7073,
      "step": 21830
    },
    {
      "epoch": 0.6290266812654872,
      "grad_norm": 0.7060241103172302,
      "learning_rate": 3.1953018201297358e-06,
      "loss": 0.6767,
      "step": 21831
    },
    {
      "epoch": 0.6290554947271365,
      "grad_norm": 0.7326345443725586,
      "learning_rate": 3.194866672434387e-06,
      "loss": 0.7028,
      "step": 21832
    },
    {
      "epoch": 0.6290843081887858,
      "grad_norm": 0.733936607837677,
      "learning_rate": 3.1944315404599456e-06,
      "loss": 0.7046,
      "step": 21833
    },
    {
      "epoch": 0.6291131216504351,
      "grad_norm": 0.7569835782051086,
      "learning_rate": 3.1939964242102005e-06,
      "loss": 0.6904,
      "step": 21834
    },
    {
      "epoch": 0.6291419351120844,
      "grad_norm": 0.7509472966194153,
      "learning_rate": 3.193561323688941e-06,
      "loss": 0.6975,
      "step": 21835
    },
    {
      "epoch": 0.6291707485737337,
      "grad_norm": 0.7062505483627319,
      "learning_rate": 3.193126238899956e-06,
      "loss": 0.679,
      "step": 21836
    },
    {
      "epoch": 0.629199562035383,
      "grad_norm": 0.7098217010498047,
      "learning_rate": 3.1926911698470355e-06,
      "loss": 0.7184,
      "step": 21837
    },
    {
      "epoch": 0.6292283754970323,
      "grad_norm": 0.7373673915863037,
      "learning_rate": 3.1922561165339687e-06,
      "loss": 0.6802,
      "step": 21838
    },
    {
      "epoch": 0.6292571889586815,
      "grad_norm": 0.7320712208747864,
      "learning_rate": 3.191821078964543e-06,
      "loss": 0.7011,
      "step": 21839
    },
    {
      "epoch": 0.6292860024203307,
      "grad_norm": 0.725627601146698,
      "learning_rate": 3.19138605714255e-06,
      "loss": 0.6893,
      "step": 21840
    },
    {
      "epoch": 0.62931481588198,
      "grad_norm": 0.7529215216636658,
      "learning_rate": 3.1909510510717755e-06,
      "loss": 0.7196,
      "step": 21841
    },
    {
      "epoch": 0.6293436293436293,
      "grad_norm": 0.7249716520309448,
      "learning_rate": 3.190516060756009e-06,
      "loss": 0.7128,
      "step": 21842
    },
    {
      "epoch": 0.6293724428052786,
      "grad_norm": 0.7621384263038635,
      "learning_rate": 3.1900810861990396e-06,
      "loss": 0.7385,
      "step": 21843
    },
    {
      "epoch": 0.6294012562669279,
      "grad_norm": 0.709578812122345,
      "learning_rate": 3.1896461274046544e-06,
      "loss": 0.6938,
      "step": 21844
    },
    {
      "epoch": 0.6294300697285772,
      "grad_norm": 0.7344944477081299,
      "learning_rate": 3.1892111843766415e-06,
      "loss": 0.7092,
      "step": 21845
    },
    {
      "epoch": 0.6294588831902265,
      "grad_norm": 0.7135345935821533,
      "learning_rate": 3.1887762571187888e-06,
      "loss": 0.702,
      "step": 21846
    },
    {
      "epoch": 0.6294876966518758,
      "grad_norm": 0.7352182269096375,
      "learning_rate": 3.188341345634885e-06,
      "loss": 0.7354,
      "step": 21847
    },
    {
      "epoch": 0.629516510113525,
      "grad_norm": 0.7092639207839966,
      "learning_rate": 3.1879064499287166e-06,
      "loss": 0.7027,
      "step": 21848
    },
    {
      "epoch": 0.6295453235751743,
      "grad_norm": 0.7087957859039307,
      "learning_rate": 3.187471570004069e-06,
      "loss": 0.7078,
      "step": 21849
    },
    {
      "epoch": 0.6295741370368236,
      "grad_norm": 0.7266098856925964,
      "learning_rate": 3.1870367058647357e-06,
      "loss": 0.7088,
      "step": 21850
    },
    {
      "epoch": 0.6296029504984729,
      "grad_norm": 0.7121185660362244,
      "learning_rate": 3.1866018575144996e-06,
      "loss": 0.7084,
      "step": 21851
    },
    {
      "epoch": 0.6296317639601222,
      "grad_norm": 0.7217791676521301,
      "learning_rate": 3.1861670249571484e-06,
      "loss": 0.7158,
      "step": 21852
    },
    {
      "epoch": 0.6296605774217715,
      "grad_norm": 0.7150460481643677,
      "learning_rate": 3.185732208196469e-06,
      "loss": 0.7239,
      "step": 21853
    },
    {
      "epoch": 0.6296893908834207,
      "grad_norm": 0.7049506902694702,
      "learning_rate": 3.185297407236249e-06,
      "loss": 0.7007,
      "step": 21854
    },
    {
      "epoch": 0.62971820434507,
      "grad_norm": 0.6961475610733032,
      "learning_rate": 3.184862622080275e-06,
      "loss": 0.7129,
      "step": 21855
    },
    {
      "epoch": 0.6297470178067193,
      "grad_norm": 0.7118580341339111,
      "learning_rate": 3.1844278527323313e-06,
      "loss": 0.6971,
      "step": 21856
    },
    {
      "epoch": 0.6297758312683686,
      "grad_norm": 0.7335295677185059,
      "learning_rate": 3.183993099196207e-06,
      "loss": 0.7004,
      "step": 21857
    },
    {
      "epoch": 0.6298046447300178,
      "grad_norm": 0.7076334357261658,
      "learning_rate": 3.1835583614756877e-06,
      "loss": 0.7006,
      "step": 21858
    },
    {
      "epoch": 0.6298334581916671,
      "grad_norm": 0.7158050537109375,
      "learning_rate": 3.183123639574559e-06,
      "loss": 0.722,
      "step": 21859
    },
    {
      "epoch": 0.6298622716533164,
      "grad_norm": 0.7232458591461182,
      "learning_rate": 3.1826889334966082e-06,
      "loss": 0.6923,
      "step": 21860
    },
    {
      "epoch": 0.6298910851149657,
      "grad_norm": 0.7216745018959045,
      "learning_rate": 3.1822542432456173e-06,
      "loss": 0.6825,
      "step": 21861
    },
    {
      "epoch": 0.629919898576615,
      "grad_norm": 0.7138550281524658,
      "learning_rate": 3.1818195688253773e-06,
      "loss": 0.7035,
      "step": 21862
    },
    {
      "epoch": 0.6299487120382643,
      "grad_norm": 0.7136277556419373,
      "learning_rate": 3.181384910239671e-06,
      "loss": 0.6917,
      "step": 21863
    },
    {
      "epoch": 0.6299775254999136,
      "grad_norm": 0.7158644199371338,
      "learning_rate": 3.180950267492284e-06,
      "loss": 0.6832,
      "step": 21864
    },
    {
      "epoch": 0.6300063389615629,
      "grad_norm": 0.7199772596359253,
      "learning_rate": 3.1805156405870017e-06,
      "loss": 0.7022,
      "step": 21865
    },
    {
      "epoch": 0.6300351524232122,
      "grad_norm": 0.7283614873886108,
      "learning_rate": 3.1800810295276096e-06,
      "loss": 0.694,
      "step": 21866
    },
    {
      "epoch": 0.6300639658848614,
      "grad_norm": 0.7114160060882568,
      "learning_rate": 3.179646434317893e-06,
      "loss": 0.6817,
      "step": 21867
    },
    {
      "epoch": 0.6300927793465106,
      "grad_norm": 0.7269613146781921,
      "learning_rate": 3.1792118549616357e-06,
      "loss": 0.7109,
      "step": 21868
    },
    {
      "epoch": 0.6301215928081599,
      "grad_norm": 0.7034692168235779,
      "learning_rate": 3.1787772914626234e-06,
      "loss": 0.6875,
      "step": 21869
    },
    {
      "epoch": 0.6301504062698092,
      "grad_norm": 0.7283657193183899,
      "learning_rate": 3.1783427438246402e-06,
      "loss": 0.7235,
      "step": 21870
    },
    {
      "epoch": 0.6301792197314585,
      "grad_norm": 0.7091536521911621,
      "learning_rate": 3.17790821205147e-06,
      "loss": 0.6999,
      "step": 21871
    },
    {
      "epoch": 0.6302080331931078,
      "grad_norm": 0.7250977158546448,
      "learning_rate": 3.177473696146898e-06,
      "loss": 0.707,
      "step": 21872
    },
    {
      "epoch": 0.6302368466547571,
      "grad_norm": 0.7255007028579712,
      "learning_rate": 3.1770391961147074e-06,
      "loss": 0.6958,
      "step": 21873
    },
    {
      "epoch": 0.6302656601164064,
      "grad_norm": 0.7146590948104858,
      "learning_rate": 3.1766047119586846e-06,
      "loss": 0.7221,
      "step": 21874
    },
    {
      "epoch": 0.6302944735780557,
      "grad_norm": 0.7171658873558044,
      "learning_rate": 3.1761702436826115e-06,
      "loss": 0.7075,
      "step": 21875
    },
    {
      "epoch": 0.630323287039705,
      "grad_norm": 0.7385851740837097,
      "learning_rate": 3.175735791290273e-06,
      "loss": 0.7045,
      "step": 21876
    },
    {
      "epoch": 0.6303521005013543,
      "grad_norm": 0.7411518096923828,
      "learning_rate": 3.1753013547854506e-06,
      "loss": 0.6859,
      "step": 21877
    },
    {
      "epoch": 0.6303809139630036,
      "grad_norm": 0.7030602097511292,
      "learning_rate": 3.1748669341719307e-06,
      "loss": 0.7231,
      "step": 21878
    },
    {
      "epoch": 0.6304097274246528,
      "grad_norm": 0.7079483270645142,
      "learning_rate": 3.1744325294534957e-06,
      "loss": 0.701,
      "step": 21879
    },
    {
      "epoch": 0.6304385408863021,
      "grad_norm": 0.7105659246444702,
      "learning_rate": 3.1739981406339284e-06,
      "loss": 0.6921,
      "step": 21880
    },
    {
      "epoch": 0.6304673543479513,
      "grad_norm": 0.7566288113594055,
      "learning_rate": 3.1735637677170112e-06,
      "loss": 0.6935,
      "step": 21881
    },
    {
      "epoch": 0.6304961678096006,
      "grad_norm": 0.7392767071723938,
      "learning_rate": 3.173129410706528e-06,
      "loss": 0.6885,
      "step": 21882
    },
    {
      "epoch": 0.6305249812712499,
      "grad_norm": 0.712049663066864,
      "learning_rate": 3.172695069606262e-06,
      "loss": 0.681,
      "step": 21883
    },
    {
      "epoch": 0.6305537947328992,
      "grad_norm": 0.7251437902450562,
      "learning_rate": 3.172260744419995e-06,
      "loss": 0.7021,
      "step": 21884
    },
    {
      "epoch": 0.6305826081945485,
      "grad_norm": 0.7128678560256958,
      "learning_rate": 3.1718264351515077e-06,
      "loss": 0.7015,
      "step": 21885
    },
    {
      "epoch": 0.6306114216561978,
      "grad_norm": 0.7172044515609741,
      "learning_rate": 3.171392141804587e-06,
      "loss": 0.6724,
      "step": 21886
    },
    {
      "epoch": 0.6306402351178471,
      "grad_norm": 0.7122812271118164,
      "learning_rate": 3.1709578643830125e-06,
      "loss": 0.6918,
      "step": 21887
    },
    {
      "epoch": 0.6306690485794963,
      "grad_norm": 0.7158588767051697,
      "learning_rate": 3.170523602890567e-06,
      "loss": 0.6739,
      "step": 21888
    },
    {
      "epoch": 0.6306978620411456,
      "grad_norm": 0.6956550478935242,
      "learning_rate": 3.1700893573310317e-06,
      "loss": 0.6594,
      "step": 21889
    },
    {
      "epoch": 0.6307266755027949,
      "grad_norm": 0.7353466749191284,
      "learning_rate": 3.169655127708189e-06,
      "loss": 0.6926,
      "step": 21890
    },
    {
      "epoch": 0.6307554889644442,
      "grad_norm": 0.7251282930374146,
      "learning_rate": 3.16922091402582e-06,
      "loss": 0.6877,
      "step": 21891
    },
    {
      "epoch": 0.6307843024260935,
      "grad_norm": 0.7236316204071045,
      "learning_rate": 3.168786716287707e-06,
      "loss": 0.686,
      "step": 21892
    },
    {
      "epoch": 0.6308131158877428,
      "grad_norm": 0.7112581133842468,
      "learning_rate": 3.168352534497632e-06,
      "loss": 0.6754,
      "step": 21893
    },
    {
      "epoch": 0.630841929349392,
      "grad_norm": 0.7171743512153625,
      "learning_rate": 3.167918368659374e-06,
      "loss": 0.7077,
      "step": 21894
    },
    {
      "epoch": 0.6308707428110413,
      "grad_norm": 0.7235990166664124,
      "learning_rate": 3.1674842187767164e-06,
      "loss": 0.6902,
      "step": 21895
    },
    {
      "epoch": 0.6308995562726906,
      "grad_norm": 0.7636733055114746,
      "learning_rate": 3.167050084853439e-06,
      "loss": 0.6853,
      "step": 21896
    },
    {
      "epoch": 0.6309283697343399,
      "grad_norm": 0.699066162109375,
      "learning_rate": 3.166615966893323e-06,
      "loss": 0.6916,
      "step": 21897
    },
    {
      "epoch": 0.6309571831959891,
      "grad_norm": 0.703676700592041,
      "learning_rate": 3.1661818649001497e-06,
      "loss": 0.7192,
      "step": 21898
    },
    {
      "epoch": 0.6309859966576384,
      "grad_norm": 0.7095333337783813,
      "learning_rate": 3.165747778877698e-06,
      "loss": 0.6877,
      "step": 21899
    },
    {
      "epoch": 0.6310148101192877,
      "grad_norm": 0.7417364120483398,
      "learning_rate": 3.1653137088297503e-06,
      "loss": 0.7228,
      "step": 21900
    },
    {
      "epoch": 0.631043623580937,
      "grad_norm": 0.711968183517456,
      "learning_rate": 3.1648796547600868e-06,
      "loss": 0.7024,
      "step": 21901
    },
    {
      "epoch": 0.6310724370425863,
      "grad_norm": 0.7137288451194763,
      "learning_rate": 3.164445616672487e-06,
      "loss": 0.6934,
      "step": 21902
    },
    {
      "epoch": 0.6311012505042356,
      "grad_norm": 0.7123815417289734,
      "learning_rate": 3.1640115945707304e-06,
      "loss": 0.6808,
      "step": 21903
    },
    {
      "epoch": 0.6311300639658849,
      "grad_norm": 0.7013214826583862,
      "learning_rate": 3.1635775884585986e-06,
      "loss": 0.6795,
      "step": 21904
    },
    {
      "epoch": 0.6311588774275342,
      "grad_norm": 0.7233941555023193,
      "learning_rate": 3.1631435983398697e-06,
      "loss": 0.6868,
      "step": 21905
    },
    {
      "epoch": 0.6311876908891835,
      "grad_norm": 0.7115299105644226,
      "learning_rate": 3.162709624218324e-06,
      "loss": 0.7001,
      "step": 21906
    },
    {
      "epoch": 0.6312165043508328,
      "grad_norm": 0.7117083072662354,
      "learning_rate": 3.1622756660977406e-06,
      "loss": 0.7098,
      "step": 21907
    },
    {
      "epoch": 0.6312453178124819,
      "grad_norm": 0.7185761332511902,
      "learning_rate": 3.161841723981899e-06,
      "loss": 0.7092,
      "step": 21908
    },
    {
      "epoch": 0.6312741312741312,
      "grad_norm": 0.7175206542015076,
      "learning_rate": 3.1614077978745765e-06,
      "loss": 0.7273,
      "step": 21909
    },
    {
      "epoch": 0.6313029447357805,
      "grad_norm": 0.7054726481437683,
      "learning_rate": 3.1609738877795563e-06,
      "loss": 0.7049,
      "step": 21910
    },
    {
      "epoch": 0.6313317581974298,
      "grad_norm": 0.6997014880180359,
      "learning_rate": 3.1605399937006154e-06,
      "loss": 0.7071,
      "step": 21911
    },
    {
      "epoch": 0.6313605716590791,
      "grad_norm": 0.7242698073387146,
      "learning_rate": 3.1601061156415325e-06,
      "loss": 0.6986,
      "step": 21912
    },
    {
      "epoch": 0.6313893851207284,
      "grad_norm": 0.7111220955848694,
      "learning_rate": 3.159672253606085e-06,
      "loss": 0.7119,
      "step": 21913
    },
    {
      "epoch": 0.6314181985823777,
      "grad_norm": 0.706965982913971,
      "learning_rate": 3.1592384075980542e-06,
      "loss": 0.6739,
      "step": 21914
    },
    {
      "epoch": 0.631447012044027,
      "grad_norm": 0.7138225436210632,
      "learning_rate": 3.1588045776212145e-06,
      "loss": 0.7018,
      "step": 21915
    },
    {
      "epoch": 0.6314758255056763,
      "grad_norm": 0.7338964343070984,
      "learning_rate": 3.158370763679348e-06,
      "loss": 0.7404,
      "step": 21916
    },
    {
      "epoch": 0.6315046389673256,
      "grad_norm": 0.7208725810050964,
      "learning_rate": 3.157936965776231e-06,
      "loss": 0.7056,
      "step": 21917
    },
    {
      "epoch": 0.6315334524289749,
      "grad_norm": 0.7090713381767273,
      "learning_rate": 3.157503183915642e-06,
      "loss": 0.6935,
      "step": 21918
    },
    {
      "epoch": 0.6315622658906241,
      "grad_norm": 0.7101339101791382,
      "learning_rate": 3.1570694181013577e-06,
      "loss": 0.6801,
      "step": 21919
    },
    {
      "epoch": 0.6315910793522734,
      "grad_norm": 0.7218543887138367,
      "learning_rate": 3.156635668337157e-06,
      "loss": 0.7163,
      "step": 21920
    },
    {
      "epoch": 0.6316198928139226,
      "grad_norm": 0.7175515294075012,
      "learning_rate": 3.1562019346268145e-06,
      "loss": 0.6852,
      "step": 21921
    },
    {
      "epoch": 0.6316487062755719,
      "grad_norm": 0.7235671877861023,
      "learning_rate": 3.1557682169741123e-06,
      "loss": 0.6913,
      "step": 21922
    },
    {
      "epoch": 0.6316775197372212,
      "grad_norm": 0.7177548408508301,
      "learning_rate": 3.155334515382826e-06,
      "loss": 0.6519,
      "step": 21923
    },
    {
      "epoch": 0.6317063331988705,
      "grad_norm": 0.7020424008369446,
      "learning_rate": 3.1549008298567315e-06,
      "loss": 0.6928,
      "step": 21924
    },
    {
      "epoch": 0.6317351466605198,
      "grad_norm": 0.7326717376708984,
      "learning_rate": 3.154467160399606e-06,
      "loss": 0.7108,
      "step": 21925
    },
    {
      "epoch": 0.6317639601221691,
      "grad_norm": 0.7308167815208435,
      "learning_rate": 3.1540335070152274e-06,
      "loss": 0.7138,
      "step": 21926
    },
    {
      "epoch": 0.6317927735838184,
      "grad_norm": 0.7085211873054504,
      "learning_rate": 3.1535998697073712e-06,
      "loss": 0.6851,
      "step": 21927
    },
    {
      "epoch": 0.6318215870454676,
      "grad_norm": 0.7482206225395203,
      "learning_rate": 3.153166248479814e-06,
      "loss": 0.722,
      "step": 21928
    },
    {
      "epoch": 0.6318504005071169,
      "grad_norm": 0.7235016822814941,
      "learning_rate": 3.1527326433363337e-06,
      "loss": 0.699,
      "step": 21929
    },
    {
      "epoch": 0.6318792139687662,
      "grad_norm": 0.7213943600654602,
      "learning_rate": 3.1522990542807046e-06,
      "loss": 0.6968,
      "step": 21930
    },
    {
      "epoch": 0.6319080274304155,
      "grad_norm": 0.7224615812301636,
      "learning_rate": 3.1518654813167042e-06,
      "loss": 0.692,
      "step": 21931
    },
    {
      "epoch": 0.6319368408920648,
      "grad_norm": 0.7263688445091248,
      "learning_rate": 3.151431924448107e-06,
      "loss": 0.6926,
      "step": 21932
    },
    {
      "epoch": 0.6319656543537141,
      "grad_norm": 0.6818132400512695,
      "learning_rate": 3.1509983836786895e-06,
      "loss": 0.6843,
      "step": 21933
    },
    {
      "epoch": 0.6319944678153634,
      "grad_norm": 0.712123453617096,
      "learning_rate": 3.150564859012229e-06,
      "loss": 0.699,
      "step": 21934
    },
    {
      "epoch": 0.6320232812770126,
      "grad_norm": 0.7046458125114441,
      "learning_rate": 3.1501313504525e-06,
      "loss": 0.7014,
      "step": 21935
    },
    {
      "epoch": 0.6320520947386619,
      "grad_norm": 0.7059844732284546,
      "learning_rate": 3.1496978580032773e-06,
      "loss": 0.6764,
      "step": 21936
    },
    {
      "epoch": 0.6320809082003112,
      "grad_norm": 0.7120904326438904,
      "learning_rate": 3.1492643816683356e-06,
      "loss": 0.7137,
      "step": 21937
    },
    {
      "epoch": 0.6321097216619604,
      "grad_norm": 0.7223979830741882,
      "learning_rate": 3.148830921451452e-06,
      "loss": 0.6769,
      "step": 21938
    },
    {
      "epoch": 0.6321385351236097,
      "grad_norm": 0.7095780968666077,
      "learning_rate": 3.1483974773564012e-06,
      "loss": 0.6944,
      "step": 21939
    },
    {
      "epoch": 0.632167348585259,
      "grad_norm": 0.7024148106575012,
      "learning_rate": 3.1479640493869574e-06,
      "loss": 0.6925,
      "step": 21940
    },
    {
      "epoch": 0.6321961620469083,
      "grad_norm": 0.7373760938644409,
      "learning_rate": 3.147530637546895e-06,
      "loss": 0.7094,
      "step": 21941
    },
    {
      "epoch": 0.6322249755085576,
      "grad_norm": 0.7025143504142761,
      "learning_rate": 3.1470972418399894e-06,
      "loss": 0.706,
      "step": 21942
    },
    {
      "epoch": 0.6322537889702069,
      "grad_norm": 0.7276681661605835,
      "learning_rate": 3.146663862270014e-06,
      "loss": 0.698,
      "step": 21943
    },
    {
      "epoch": 0.6322826024318562,
      "grad_norm": 0.7121167182922363,
      "learning_rate": 3.1462304988407437e-06,
      "loss": 0.7162,
      "step": 21944
    },
    {
      "epoch": 0.6323114158935055,
      "grad_norm": 0.7174757719039917,
      "learning_rate": 3.145797151555951e-06,
      "loss": 0.6965,
      "step": 21945
    },
    {
      "epoch": 0.6323402293551548,
      "grad_norm": 0.7156214714050293,
      "learning_rate": 3.1453638204194135e-06,
      "loss": 0.6893,
      "step": 21946
    },
    {
      "epoch": 0.6323690428168041,
      "grad_norm": 0.7203693985939026,
      "learning_rate": 3.1449305054349033e-06,
      "loss": 0.6974,
      "step": 21947
    },
    {
      "epoch": 0.6323978562784532,
      "grad_norm": 0.7542404532432556,
      "learning_rate": 3.144497206606194e-06,
      "loss": 0.7382,
      "step": 21948
    },
    {
      "epoch": 0.6324266697401025,
      "grad_norm": 0.7201029658317566,
      "learning_rate": 3.144063923937059e-06,
      "loss": 0.6852,
      "step": 21949
    },
    {
      "epoch": 0.6324554832017518,
      "grad_norm": 0.7013959288597107,
      "learning_rate": 3.143630657431272e-06,
      "loss": 0.702,
      "step": 21950
    },
    {
      "epoch": 0.6324842966634011,
      "grad_norm": 0.7304877638816833,
      "learning_rate": 3.1431974070926064e-06,
      "loss": 0.7265,
      "step": 21951
    },
    {
      "epoch": 0.6325131101250504,
      "grad_norm": 0.7151381969451904,
      "learning_rate": 3.142764172924835e-06,
      "loss": 0.6948,
      "step": 21952
    },
    {
      "epoch": 0.6325419235866997,
      "grad_norm": 0.7036237716674805,
      "learning_rate": 3.1423309549317305e-06,
      "loss": 0.6863,
      "step": 21953
    },
    {
      "epoch": 0.632570737048349,
      "grad_norm": 0.7546406388282776,
      "learning_rate": 3.141897753117067e-06,
      "loss": 0.7161,
      "step": 21954
    },
    {
      "epoch": 0.6325995505099983,
      "grad_norm": 0.7640329003334045,
      "learning_rate": 3.1414645674846166e-06,
      "loss": 0.7161,
      "step": 21955
    },
    {
      "epoch": 0.6326283639716476,
      "grad_norm": 0.7275037169456482,
      "learning_rate": 3.1410313980381523e-06,
      "loss": 0.6843,
      "step": 21956
    },
    {
      "epoch": 0.6326571774332969,
      "grad_norm": 0.7144694924354553,
      "learning_rate": 3.1405982447814455e-06,
      "loss": 0.6933,
      "step": 21957
    },
    {
      "epoch": 0.6326859908949461,
      "grad_norm": 0.7630108594894409,
      "learning_rate": 3.140165107718269e-06,
      "loss": 0.7125,
      "step": 21958
    },
    {
      "epoch": 0.6327148043565954,
      "grad_norm": 0.7447881698608398,
      "learning_rate": 3.1397319868523955e-06,
      "loss": 0.6829,
      "step": 21959
    },
    {
      "epoch": 0.6327436178182447,
      "grad_norm": 0.7391834259033203,
      "learning_rate": 3.139298882187598e-06,
      "loss": 0.6918,
      "step": 21960
    },
    {
      "epoch": 0.632772431279894,
      "grad_norm": 0.7662880420684814,
      "learning_rate": 3.1388657937276467e-06,
      "loss": 0.6976,
      "step": 21961
    },
    {
      "epoch": 0.6328012447415432,
      "grad_norm": 0.7301923632621765,
      "learning_rate": 3.138432721476314e-06,
      "loss": 0.7077,
      "step": 21962
    },
    {
      "epoch": 0.6328300582031925,
      "grad_norm": 0.7122278213500977,
      "learning_rate": 3.137999665437371e-06,
      "loss": 0.6996,
      "step": 21963
    },
    {
      "epoch": 0.6328588716648418,
      "grad_norm": 0.7387362718582153,
      "learning_rate": 3.13756662561459e-06,
      "loss": 0.7004,
      "step": 21964
    },
    {
      "epoch": 0.6328876851264911,
      "grad_norm": 0.7128443717956543,
      "learning_rate": 3.1371336020117417e-06,
      "loss": 0.703,
      "step": 21965
    },
    {
      "epoch": 0.6329164985881404,
      "grad_norm": 0.738956093788147,
      "learning_rate": 3.136700594632598e-06,
      "loss": 0.6988,
      "step": 21966
    },
    {
      "epoch": 0.6329453120497897,
      "grad_norm": 0.7401301860809326,
      "learning_rate": 3.136267603480929e-06,
      "loss": 0.7169,
      "step": 21967
    },
    {
      "epoch": 0.632974125511439,
      "grad_norm": 0.7038403749465942,
      "learning_rate": 3.135834628560506e-06,
      "loss": 0.71,
      "step": 21968
    },
    {
      "epoch": 0.6330029389730882,
      "grad_norm": 0.6990049481391907,
      "learning_rate": 3.1354016698750982e-06,
      "loss": 0.6924,
      "step": 21969
    },
    {
      "epoch": 0.6330317524347375,
      "grad_norm": 0.7277923822402954,
      "learning_rate": 3.1349687274284803e-06,
      "loss": 0.6948,
      "step": 21970
    },
    {
      "epoch": 0.6330605658963868,
      "grad_norm": 0.7363353371620178,
      "learning_rate": 3.1345358012244197e-06,
      "loss": 0.7184,
      "step": 21971
    },
    {
      "epoch": 0.6330893793580361,
      "grad_norm": 0.7088763117790222,
      "learning_rate": 3.134102891266688e-06,
      "loss": 0.6757,
      "step": 21972
    },
    {
      "epoch": 0.6331181928196854,
      "grad_norm": 0.7432811260223389,
      "learning_rate": 3.1336699975590545e-06,
      "loss": 0.7075,
      "step": 21973
    },
    {
      "epoch": 0.6331470062813347,
      "grad_norm": 0.7128002643585205,
      "learning_rate": 3.1332371201052903e-06,
      "loss": 0.6955,
      "step": 21974
    },
    {
      "epoch": 0.6331758197429839,
      "grad_norm": 0.720504105091095,
      "learning_rate": 3.1328042589091635e-06,
      "loss": 0.708,
      "step": 21975
    },
    {
      "epoch": 0.6332046332046332,
      "grad_norm": 0.7202698588371277,
      "learning_rate": 3.1323714139744456e-06,
      "loss": 0.7289,
      "step": 21976
    },
    {
      "epoch": 0.6332334466662825,
      "grad_norm": 0.7278962731361389,
      "learning_rate": 3.131938585304906e-06,
      "loss": 0.7041,
      "step": 21977
    },
    {
      "epoch": 0.6332622601279317,
      "grad_norm": 0.7132955193519592,
      "learning_rate": 3.131505772904314e-06,
      "loss": 0.7006,
      "step": 21978
    },
    {
      "epoch": 0.633291073589581,
      "grad_norm": 0.7255422472953796,
      "learning_rate": 3.1310729767764393e-06,
      "loss": 0.6861,
      "step": 21979
    },
    {
      "epoch": 0.6333198870512303,
      "grad_norm": 0.9633122682571411,
      "learning_rate": 3.13064019692505e-06,
      "loss": 0.7146,
      "step": 21980
    },
    {
      "epoch": 0.6333487005128796,
      "grad_norm": 0.7095960378646851,
      "learning_rate": 3.1302074333539144e-06,
      "loss": 0.7036,
      "step": 21981
    },
    {
      "epoch": 0.6333775139745289,
      "grad_norm": 0.7054728865623474,
      "learning_rate": 3.129774686066804e-06,
      "loss": 0.6801,
      "step": 21982
    },
    {
      "epoch": 0.6334063274361782,
      "grad_norm": 0.7508555054664612,
      "learning_rate": 3.129341955067488e-06,
      "loss": 0.7113,
      "step": 21983
    },
    {
      "epoch": 0.6334351408978275,
      "grad_norm": 0.7158492803573608,
      "learning_rate": 3.128909240359733e-06,
      "loss": 0.6936,
      "step": 21984
    },
    {
      "epoch": 0.6334639543594768,
      "grad_norm": 0.7188633680343628,
      "learning_rate": 3.1284765419473082e-06,
      "loss": 0.6874,
      "step": 21985
    },
    {
      "epoch": 0.6334927678211261,
      "grad_norm": 0.711347758769989,
      "learning_rate": 3.1280438598339814e-06,
      "loss": 0.6922,
      "step": 21986
    },
    {
      "epoch": 0.6335215812827754,
      "grad_norm": 0.7333440780639648,
      "learning_rate": 3.127611194023522e-06,
      "loss": 0.6761,
      "step": 21987
    },
    {
      "epoch": 0.6335503947444246,
      "grad_norm": 0.7377179265022278,
      "learning_rate": 3.1271785445196966e-06,
      "loss": 0.6989,
      "step": 21988
    },
    {
      "epoch": 0.6335792082060738,
      "grad_norm": 0.7053198218345642,
      "learning_rate": 3.1267459113262744e-06,
      "loss": 0.6967,
      "step": 21989
    },
    {
      "epoch": 0.6336080216677231,
      "grad_norm": 0.7350804805755615,
      "learning_rate": 3.126313294447023e-06,
      "loss": 0.72,
      "step": 21990
    },
    {
      "epoch": 0.6336368351293724,
      "grad_norm": 0.7181355953216553,
      "learning_rate": 3.1258806938857082e-06,
      "loss": 0.6999,
      "step": 21991
    },
    {
      "epoch": 0.6336656485910217,
      "grad_norm": 0.7223294973373413,
      "learning_rate": 3.1254481096461007e-06,
      "loss": 0.6976,
      "step": 21992
    },
    {
      "epoch": 0.633694462052671,
      "grad_norm": 0.710911750793457,
      "learning_rate": 3.125015541731965e-06,
      "loss": 0.7035,
      "step": 21993
    },
    {
      "epoch": 0.6337232755143203,
      "grad_norm": 0.719787061214447,
      "learning_rate": 3.1245829901470704e-06,
      "loss": 0.6959,
      "step": 21994
    },
    {
      "epoch": 0.6337520889759696,
      "grad_norm": 0.7241654396057129,
      "learning_rate": 3.124150454895184e-06,
      "loss": 0.6915,
      "step": 21995
    },
    {
      "epoch": 0.6337809024376189,
      "grad_norm": 0.7169144749641418,
      "learning_rate": 3.1237179359800707e-06,
      "loss": 0.7006,
      "step": 21996
    },
    {
      "epoch": 0.6338097158992682,
      "grad_norm": 0.7162724137306213,
      "learning_rate": 3.123285433405499e-06,
      "loss": 0.7014,
      "step": 21997
    },
    {
      "epoch": 0.6338385293609174,
      "grad_norm": 0.7164425253868103,
      "learning_rate": 3.1228529471752357e-06,
      "loss": 0.7051,
      "step": 21998
    },
    {
      "epoch": 0.6338673428225667,
      "grad_norm": 0.7254040837287903,
      "learning_rate": 3.1224204772930467e-06,
      "loss": 0.6864,
      "step": 21999
    },
    {
      "epoch": 0.633896156284216,
      "grad_norm": 0.7207536101341248,
      "learning_rate": 3.1219880237626975e-06,
      "loss": 0.7194,
      "step": 22000
    },
    {
      "epoch": 0.6339249697458653,
      "grad_norm": 0.7071694731712341,
      "learning_rate": 3.1215555865879564e-06,
      "loss": 0.7022,
      "step": 22001
    },
    {
      "epoch": 0.6339537832075145,
      "grad_norm": 0.6975698471069336,
      "learning_rate": 3.121123165772588e-06,
      "loss": 0.6928,
      "step": 22002
    },
    {
      "epoch": 0.6339825966691638,
      "grad_norm": 0.7257025241851807,
      "learning_rate": 3.120690761320359e-06,
      "loss": 0.6874,
      "step": 22003
    },
    {
      "epoch": 0.6340114101308131,
      "grad_norm": 0.7588615417480469,
      "learning_rate": 3.1202583732350344e-06,
      "loss": 0.6837,
      "step": 22004
    },
    {
      "epoch": 0.6340402235924624,
      "grad_norm": 0.7289250493049622,
      "learning_rate": 3.1198260015203785e-06,
      "loss": 0.7301,
      "step": 22005
    },
    {
      "epoch": 0.6340690370541117,
      "grad_norm": 0.7179392576217651,
      "learning_rate": 3.119393646180161e-06,
      "loss": 0.6943,
      "step": 22006
    },
    {
      "epoch": 0.634097850515761,
      "grad_norm": 0.7204885482788086,
      "learning_rate": 3.1189613072181445e-06,
      "loss": 0.7193,
      "step": 22007
    },
    {
      "epoch": 0.6341266639774102,
      "grad_norm": 0.7213000655174255,
      "learning_rate": 3.118528984638095e-06,
      "loss": 0.6983,
      "step": 22008
    },
    {
      "epoch": 0.6341554774390595,
      "grad_norm": 0.7030182480812073,
      "learning_rate": 3.1180966784437765e-06,
      "loss": 0.6622,
      "step": 22009
    },
    {
      "epoch": 0.6341842909007088,
      "grad_norm": 0.7223100662231445,
      "learning_rate": 3.1176643886389557e-06,
      "loss": 0.6779,
      "step": 22010
    },
    {
      "epoch": 0.6342131043623581,
      "grad_norm": 0.724097490310669,
      "learning_rate": 3.1172321152273954e-06,
      "loss": 0.7131,
      "step": 22011
    },
    {
      "epoch": 0.6342419178240074,
      "grad_norm": 0.7226244807243347,
      "learning_rate": 3.116799858212861e-06,
      "loss": 0.7059,
      "step": 22012
    },
    {
      "epoch": 0.6342707312856567,
      "grad_norm": 0.7042393088340759,
      "learning_rate": 3.116367617599118e-06,
      "loss": 0.6994,
      "step": 22013
    },
    {
      "epoch": 0.634299544747306,
      "grad_norm": 0.7134490013122559,
      "learning_rate": 3.1159353933899304e-06,
      "loss": 0.6829,
      "step": 22014
    },
    {
      "epoch": 0.6343283582089553,
      "grad_norm": 0.7129290103912354,
      "learning_rate": 3.1155031855890624e-06,
      "loss": 0.69,
      "step": 22015
    },
    {
      "epoch": 0.6343571716706045,
      "grad_norm": 0.7133612036705017,
      "learning_rate": 3.115070994200277e-06,
      "loss": 0.6848,
      "step": 22016
    },
    {
      "epoch": 0.6343859851322537,
      "grad_norm": 0.7209770083427429,
      "learning_rate": 3.1146388192273374e-06,
      "loss": 0.6997,
      "step": 22017
    },
    {
      "epoch": 0.634414798593903,
      "grad_norm": 0.724751353263855,
      "learning_rate": 3.1142066606740114e-06,
      "loss": 0.711,
      "step": 22018
    },
    {
      "epoch": 0.6344436120555523,
      "grad_norm": 0.6961643099784851,
      "learning_rate": 3.1137745185440597e-06,
      "loss": 0.7036,
      "step": 22019
    },
    {
      "epoch": 0.6344724255172016,
      "grad_norm": 0.7511668801307678,
      "learning_rate": 3.1133423928412464e-06,
      "loss": 0.7169,
      "step": 22020
    },
    {
      "epoch": 0.6345012389788509,
      "grad_norm": 0.73392653465271,
      "learning_rate": 3.1129102835693353e-06,
      "loss": 0.7062,
      "step": 22021
    },
    {
      "epoch": 0.6345300524405002,
      "grad_norm": 0.7448554635047913,
      "learning_rate": 3.112478190732089e-06,
      "loss": 0.7054,
      "step": 22022
    },
    {
      "epoch": 0.6345588659021495,
      "grad_norm": 0.7083041071891785,
      "learning_rate": 3.1120461143332713e-06,
      "loss": 0.689,
      "step": 22023
    },
    {
      "epoch": 0.6345876793637988,
      "grad_norm": 0.7261013984680176,
      "learning_rate": 3.1116140543766437e-06,
      "loss": 0.7077,
      "step": 22024
    },
    {
      "epoch": 0.6346164928254481,
      "grad_norm": 0.7017223834991455,
      "learning_rate": 3.1111820108659706e-06,
      "loss": 0.7,
      "step": 22025
    },
    {
      "epoch": 0.6346453062870974,
      "grad_norm": 0.7181498408317566,
      "learning_rate": 3.110749983805015e-06,
      "loss": 0.6822,
      "step": 22026
    },
    {
      "epoch": 0.6346741197487467,
      "grad_norm": 0.7237221598625183,
      "learning_rate": 3.110317973197537e-06,
      "loss": 0.683,
      "step": 22027
    },
    {
      "epoch": 0.634702933210396,
      "grad_norm": 0.6980436444282532,
      "learning_rate": 3.1098859790473014e-06,
      "loss": 0.7023,
      "step": 22028
    },
    {
      "epoch": 0.6347317466720451,
      "grad_norm": 0.7113319635391235,
      "learning_rate": 3.1094540013580666e-06,
      "loss": 0.6824,
      "step": 22029
    },
    {
      "epoch": 0.6347605601336944,
      "grad_norm": 0.7471036314964294,
      "learning_rate": 3.1090220401336e-06,
      "loss": 0.6893,
      "step": 22030
    },
    {
      "epoch": 0.6347893735953437,
      "grad_norm": 0.7224972248077393,
      "learning_rate": 3.1085900953776614e-06,
      "loss": 0.6866,
      "step": 22031
    },
    {
      "epoch": 0.634818187056993,
      "grad_norm": 0.7455223202705383,
      "learning_rate": 3.108158167094012e-06,
      "loss": 0.6946,
      "step": 22032
    },
    {
      "epoch": 0.6348470005186423,
      "grad_norm": 0.713943362236023,
      "learning_rate": 3.107726255286414e-06,
      "loss": 0.718,
      "step": 22033
    },
    {
      "epoch": 0.6348758139802916,
      "grad_norm": 0.7289348840713501,
      "learning_rate": 3.107294359958628e-06,
      "loss": 0.7063,
      "step": 22034
    },
    {
      "epoch": 0.6349046274419409,
      "grad_norm": 0.7179174423217773,
      "learning_rate": 3.1068624811144164e-06,
      "loss": 0.696,
      "step": 22035
    },
    {
      "epoch": 0.6349334409035902,
      "grad_norm": 0.6968963146209717,
      "learning_rate": 3.106430618757541e-06,
      "loss": 0.7142,
      "step": 22036
    },
    {
      "epoch": 0.6349622543652395,
      "grad_norm": 0.7021946310997009,
      "learning_rate": 3.1059987728917616e-06,
      "loss": 0.6959,
      "step": 22037
    },
    {
      "epoch": 0.6349910678268887,
      "grad_norm": 0.7254398465156555,
      "learning_rate": 3.1055669435208397e-06,
      "loss": 0.7045,
      "step": 22038
    },
    {
      "epoch": 0.635019881288538,
      "grad_norm": 0.7178260684013367,
      "learning_rate": 3.1051351306485357e-06,
      "loss": 0.7074,
      "step": 22039
    },
    {
      "epoch": 0.6350486947501873,
      "grad_norm": 0.7093742489814758,
      "learning_rate": 3.1047033342786104e-06,
      "loss": 0.704,
      "step": 22040
    },
    {
      "epoch": 0.6350775082118366,
      "grad_norm": 0.7129188776016235,
      "learning_rate": 3.1042715544148225e-06,
      "loss": 0.7084,
      "step": 22041
    },
    {
      "epoch": 0.6351063216734859,
      "grad_norm": 0.7129145264625549,
      "learning_rate": 3.1038397910609375e-06,
      "loss": 0.7005,
      "step": 22042
    },
    {
      "epoch": 0.6351351351351351,
      "grad_norm": 0.7344148755073547,
      "learning_rate": 3.1034080442207116e-06,
      "loss": 0.7042,
      "step": 22043
    },
    {
      "epoch": 0.6351639485967844,
      "grad_norm": 0.7214815020561218,
      "learning_rate": 3.102976313897905e-06,
      "loss": 0.6996,
      "step": 22044
    },
    {
      "epoch": 0.6351927620584337,
      "grad_norm": 0.7285321354866028,
      "learning_rate": 3.1025446000962796e-06,
      "loss": 0.6839,
      "step": 22045
    },
    {
      "epoch": 0.635221575520083,
      "grad_norm": 0.7081379890441895,
      "learning_rate": 3.102112902819593e-06,
      "loss": 0.7107,
      "step": 22046
    },
    {
      "epoch": 0.6352503889817323,
      "grad_norm": 0.7164332866668701,
      "learning_rate": 3.101681222071606e-06,
      "loss": 0.7215,
      "step": 22047
    },
    {
      "epoch": 0.6352792024433815,
      "grad_norm": 0.7107090950012207,
      "learning_rate": 3.1012495578560786e-06,
      "loss": 0.7124,
      "step": 22048
    },
    {
      "epoch": 0.6353080159050308,
      "grad_norm": 0.6985986232757568,
      "learning_rate": 3.100817910176769e-06,
      "loss": 0.684,
      "step": 22049
    },
    {
      "epoch": 0.6353368293666801,
      "grad_norm": 0.6958547830581665,
      "learning_rate": 3.100386279037436e-06,
      "loss": 0.687,
      "step": 22050
    },
    {
      "epoch": 0.6353656428283294,
      "grad_norm": 0.7096691131591797,
      "learning_rate": 3.099954664441841e-06,
      "loss": 0.6983,
      "step": 22051
    },
    {
      "epoch": 0.6353944562899787,
      "grad_norm": 0.7339683771133423,
      "learning_rate": 3.099523066393741e-06,
      "loss": 0.7112,
      "step": 22052
    },
    {
      "epoch": 0.635423269751628,
      "grad_norm": 0.71541827917099,
      "learning_rate": 3.0990914848968945e-06,
      "loss": 0.6991,
      "step": 22053
    },
    {
      "epoch": 0.6354520832132773,
      "grad_norm": 0.7199534177780151,
      "learning_rate": 3.0986599199550615e-06,
      "loss": 0.7036,
      "step": 22054
    },
    {
      "epoch": 0.6354808966749266,
      "grad_norm": 0.7267652153968811,
      "learning_rate": 3.0982283715719996e-06,
      "loss": 0.6843,
      "step": 22055
    },
    {
      "epoch": 0.6355097101365758,
      "grad_norm": 0.7244845628738403,
      "learning_rate": 3.097796839751468e-06,
      "loss": 0.7014,
      "step": 22056
    },
    {
      "epoch": 0.635538523598225,
      "grad_norm": 0.7247459292411804,
      "learning_rate": 3.0973653244972246e-06,
      "loss": 0.6929,
      "step": 22057
    },
    {
      "epoch": 0.6355673370598743,
      "grad_norm": 0.7214023470878601,
      "learning_rate": 3.0969338258130277e-06,
      "loss": 0.7237,
      "step": 22058
    },
    {
      "epoch": 0.6355961505215236,
      "grad_norm": 0.7178046107292175,
      "learning_rate": 3.096502343702634e-06,
      "loss": 0.7014,
      "step": 22059
    },
    {
      "epoch": 0.6356249639831729,
      "grad_norm": 0.7262474298477173,
      "learning_rate": 3.096070878169803e-06,
      "loss": 0.685,
      "step": 22060
    },
    {
      "epoch": 0.6356537774448222,
      "grad_norm": 0.7091329097747803,
      "learning_rate": 3.0956394292182912e-06,
      "loss": 0.6997,
      "step": 22061
    },
    {
      "epoch": 0.6356825909064715,
      "grad_norm": 0.7078421115875244,
      "learning_rate": 3.0952079968518558e-06,
      "loss": 0.677,
      "step": 22062
    },
    {
      "epoch": 0.6357114043681208,
      "grad_norm": 0.7104654312133789,
      "learning_rate": 3.0947765810742548e-06,
      "loss": 0.6811,
      "step": 22063
    },
    {
      "epoch": 0.6357402178297701,
      "grad_norm": 0.738460123538971,
      "learning_rate": 3.0943451818892457e-06,
      "loss": 0.7258,
      "step": 22064
    },
    {
      "epoch": 0.6357690312914194,
      "grad_norm": 0.7472034096717834,
      "learning_rate": 3.0939137993005823e-06,
      "loss": 0.6984,
      "step": 22065
    },
    {
      "epoch": 0.6357978447530687,
      "grad_norm": 0.7508759498596191,
      "learning_rate": 3.0934824333120273e-06,
      "loss": 0.6951,
      "step": 22066
    },
    {
      "epoch": 0.635826658214718,
      "grad_norm": 0.6991212368011475,
      "learning_rate": 3.0930510839273336e-06,
      "loss": 0.6867,
      "step": 22067
    },
    {
      "epoch": 0.6358554716763672,
      "grad_norm": 0.7429501414299011,
      "learning_rate": 3.0926197511502594e-06,
      "loss": 0.7064,
      "step": 22068
    },
    {
      "epoch": 0.6358842851380165,
      "grad_norm": 0.7015599608421326,
      "learning_rate": 3.0921884349845597e-06,
      "loss": 0.6757,
      "step": 22069
    },
    {
      "epoch": 0.6359130985996657,
      "grad_norm": 0.7296705842018127,
      "learning_rate": 3.0917571354339924e-06,
      "loss": 0.6904,
      "step": 22070
    },
    {
      "epoch": 0.635941912061315,
      "grad_norm": 0.7178724408149719,
      "learning_rate": 3.0913258525023127e-06,
      "loss": 0.6736,
      "step": 22071
    },
    {
      "epoch": 0.6359707255229643,
      "grad_norm": 0.7280719876289368,
      "learning_rate": 3.090894586193276e-06,
      "loss": 0.6832,
      "step": 22072
    },
    {
      "epoch": 0.6359995389846136,
      "grad_norm": 0.7105081081390381,
      "learning_rate": 3.0904633365106396e-06,
      "loss": 0.7031,
      "step": 22073
    },
    {
      "epoch": 0.6360283524462629,
      "grad_norm": 0.7448127865791321,
      "learning_rate": 3.0900321034581586e-06,
      "loss": 0.7207,
      "step": 22074
    },
    {
      "epoch": 0.6360571659079122,
      "grad_norm": 0.7416060566902161,
      "learning_rate": 3.089600887039589e-06,
      "loss": 0.7111,
      "step": 22075
    },
    {
      "epoch": 0.6360859793695615,
      "grad_norm": 0.6944217085838318,
      "learning_rate": 3.0891696872586862e-06,
      "loss": 0.7001,
      "step": 22076
    },
    {
      "epoch": 0.6361147928312108,
      "grad_norm": 0.7412465810775757,
      "learning_rate": 3.088738504119203e-06,
      "loss": 0.7089,
      "step": 22077
    },
    {
      "epoch": 0.63614360629286,
      "grad_norm": 0.7052128911018372,
      "learning_rate": 3.088307337624899e-06,
      "loss": 0.703,
      "step": 22078
    },
    {
      "epoch": 0.6361724197545093,
      "grad_norm": 0.7155901193618774,
      "learning_rate": 3.087876187779527e-06,
      "loss": 0.7003,
      "step": 22079
    },
    {
      "epoch": 0.6362012332161586,
      "grad_norm": 0.7297404408454895,
      "learning_rate": 3.087445054586842e-06,
      "loss": 0.7108,
      "step": 22080
    },
    {
      "epoch": 0.6362300466778079,
      "grad_norm": 0.7278196215629578,
      "learning_rate": 3.0870139380505986e-06,
      "loss": 0.6903,
      "step": 22081
    },
    {
      "epoch": 0.6362588601394572,
      "grad_norm": 0.7408389449119568,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.6974,
      "step": 22082
    },
    {
      "epoch": 0.6362876736011064,
      "grad_norm": 0.7288330793380737,
      "learning_rate": 3.0861517549624555e-06,
      "loss": 0.7063,
      "step": 22083
    },
    {
      "epoch": 0.6363164870627557,
      "grad_norm": 0.7169463038444519,
      "learning_rate": 3.0857206884180646e-06,
      "loss": 0.6877,
      "step": 22084
    },
    {
      "epoch": 0.636345300524405,
      "grad_norm": 0.7298044562339783,
      "learning_rate": 3.0852896385451324e-06,
      "loss": 0.6998,
      "step": 22085
    },
    {
      "epoch": 0.6363741139860543,
      "grad_norm": 0.7231197953224182,
      "learning_rate": 3.084858605347414e-06,
      "loss": 0.6995,
      "step": 22086
    },
    {
      "epoch": 0.6364029274477035,
      "grad_norm": 0.7027966976165771,
      "learning_rate": 3.0844275888286623e-06,
      "loss": 0.6992,
      "step": 22087
    },
    {
      "epoch": 0.6364317409093528,
      "grad_norm": 0.72123783826828,
      "learning_rate": 3.0839965889926294e-06,
      "loss": 0.7024,
      "step": 22088
    },
    {
      "epoch": 0.6364605543710021,
      "grad_norm": 0.7078960537910461,
      "learning_rate": 3.083565605843073e-06,
      "loss": 0.6991,
      "step": 22089
    },
    {
      "epoch": 0.6364893678326514,
      "grad_norm": 0.7335478663444519,
      "learning_rate": 3.0831346393837446e-06,
      "loss": 0.7074,
      "step": 22090
    },
    {
      "epoch": 0.6365181812943007,
      "grad_norm": 0.6997430324554443,
      "learning_rate": 3.0827036896183975e-06,
      "loss": 0.7053,
      "step": 22091
    },
    {
      "epoch": 0.63654699475595,
      "grad_norm": 0.7146682739257812,
      "learning_rate": 3.0822727565507844e-06,
      "loss": 0.7214,
      "step": 22092
    },
    {
      "epoch": 0.6365758082175993,
      "grad_norm": 0.7140152454376221,
      "learning_rate": 3.0818418401846574e-06,
      "loss": 0.6914,
      "step": 22093
    },
    {
      "epoch": 0.6366046216792486,
      "grad_norm": 0.7070672512054443,
      "learning_rate": 3.081410940523772e-06,
      "loss": 0.6896,
      "step": 22094
    },
    {
      "epoch": 0.6366334351408979,
      "grad_norm": 0.7136964797973633,
      "learning_rate": 3.080980057571879e-06,
      "loss": 0.7055,
      "step": 22095
    },
    {
      "epoch": 0.6366622486025472,
      "grad_norm": 0.7009682059288025,
      "learning_rate": 3.0805491913327313e-06,
      "loss": 0.6933,
      "step": 22096
    },
    {
      "epoch": 0.6366910620641963,
      "grad_norm": 0.7083982229232788,
      "learning_rate": 3.0801183418100823e-06,
      "loss": 0.7062,
      "step": 22097
    },
    {
      "epoch": 0.6367198755258456,
      "grad_norm": 0.732031524181366,
      "learning_rate": 3.0796875090076824e-06,
      "loss": 0.6771,
      "step": 22098
    },
    {
      "epoch": 0.6367486889874949,
      "grad_norm": 0.7086175084114075,
      "learning_rate": 3.0792566929292844e-06,
      "loss": 0.7006,
      "step": 22099
    },
    {
      "epoch": 0.6367775024491442,
      "grad_norm": 0.7046810388565063,
      "learning_rate": 3.07882589357864e-06,
      "loss": 0.6828,
      "step": 22100
    },
    {
      "epoch": 0.6368063159107935,
      "grad_norm": 0.7296024560928345,
      "learning_rate": 3.0783951109595023e-06,
      "loss": 0.7041,
      "step": 22101
    },
    {
      "epoch": 0.6368351293724428,
      "grad_norm": 0.7066004872322083,
      "learning_rate": 3.0779643450756224e-06,
      "loss": 0.6852,
      "step": 22102
    },
    {
      "epoch": 0.6368639428340921,
      "grad_norm": 0.7392704486846924,
      "learning_rate": 3.077533595930752e-06,
      "loss": 0.7079,
      "step": 22103
    },
    {
      "epoch": 0.6368927562957414,
      "grad_norm": 0.7326787114143372,
      "learning_rate": 3.0771028635286426e-06,
      "loss": 0.6661,
      "step": 22104
    },
    {
      "epoch": 0.6369215697573907,
      "grad_norm": 0.7096793055534363,
      "learning_rate": 3.0766721478730445e-06,
      "loss": 0.7005,
      "step": 22105
    },
    {
      "epoch": 0.63695038321904,
      "grad_norm": 0.6950600147247314,
      "learning_rate": 3.0762414489677096e-06,
      "loss": 0.6795,
      "step": 22106
    },
    {
      "epoch": 0.6369791966806893,
      "grad_norm": 0.7578123211860657,
      "learning_rate": 3.0758107668163883e-06,
      "loss": 0.7384,
      "step": 22107
    },
    {
      "epoch": 0.6370080101423385,
      "grad_norm": 0.7301385998725891,
      "learning_rate": 3.0753801014228324e-06,
      "loss": 0.715,
      "step": 22108
    },
    {
      "epoch": 0.6370368236039878,
      "grad_norm": 0.7008885145187378,
      "learning_rate": 3.0749494527907915e-06,
      "loss": 0.6856,
      "step": 22109
    },
    {
      "epoch": 0.637065637065637,
      "grad_norm": 0.7060184478759766,
      "learning_rate": 3.074518820924015e-06,
      "loss": 0.6878,
      "step": 22110
    },
    {
      "epoch": 0.6370944505272863,
      "grad_norm": 0.7161647081375122,
      "learning_rate": 3.0740882058262567e-06,
      "loss": 0.6985,
      "step": 22111
    },
    {
      "epoch": 0.6371232639889356,
      "grad_norm": 0.7142935395240784,
      "learning_rate": 3.0736576075012635e-06,
      "loss": 0.7161,
      "step": 22112
    },
    {
      "epoch": 0.6371520774505849,
      "grad_norm": 0.7426902651786804,
      "learning_rate": 3.073227025952788e-06,
      "loss": 0.7244,
      "step": 22113
    },
    {
      "epoch": 0.6371808909122342,
      "grad_norm": 0.7349216341972351,
      "learning_rate": 3.072796461184579e-06,
      "loss": 0.6761,
      "step": 22114
    },
    {
      "epoch": 0.6372097043738835,
      "grad_norm": 0.7166930437088013,
      "learning_rate": 3.072365913200386e-06,
      "loss": 0.6967,
      "step": 22115
    },
    {
      "epoch": 0.6372385178355328,
      "grad_norm": 0.7040006518363953,
      "learning_rate": 3.0719353820039586e-06,
      "loss": 0.7053,
      "step": 22116
    },
    {
      "epoch": 0.637267331297182,
      "grad_norm": 0.7176037430763245,
      "learning_rate": 3.0715048675990477e-06,
      "loss": 0.7005,
      "step": 22117
    },
    {
      "epoch": 0.6372961447588313,
      "grad_norm": 0.7192170023918152,
      "learning_rate": 3.0710743699894008e-06,
      "loss": 0.7057,
      "step": 22118
    },
    {
      "epoch": 0.6373249582204806,
      "grad_norm": 0.7487391233444214,
      "learning_rate": 3.070643889178768e-06,
      "loss": 0.7217,
      "step": 22119
    },
    {
      "epoch": 0.6373537716821299,
      "grad_norm": 0.7199576497077942,
      "learning_rate": 3.0702134251708992e-06,
      "loss": 0.6942,
      "step": 22120
    },
    {
      "epoch": 0.6373825851437792,
      "grad_norm": 0.7209006547927856,
      "learning_rate": 3.069782977969542e-06,
      "loss": 0.696,
      "step": 22121
    },
    {
      "epoch": 0.6374113986054285,
      "grad_norm": 0.7153943777084351,
      "learning_rate": 3.0693525475784453e-06,
      "loss": 0.7147,
      "step": 22122
    },
    {
      "epoch": 0.6374402120670778,
      "grad_norm": 0.6999576687812805,
      "learning_rate": 3.068922134001358e-06,
      "loss": 0.6817,
      "step": 22123
    },
    {
      "epoch": 0.637469025528727,
      "grad_norm": 0.712857186794281,
      "learning_rate": 3.0684917372420265e-06,
      "loss": 0.7189,
      "step": 22124
    },
    {
      "epoch": 0.6374978389903763,
      "grad_norm": 0.7149123549461365,
      "learning_rate": 3.0680613573042032e-06,
      "loss": 0.6913,
      "step": 22125
    },
    {
      "epoch": 0.6375266524520256,
      "grad_norm": 0.7275334596633911,
      "learning_rate": 3.067630994191635e-06,
      "loss": 0.701,
      "step": 22126
    },
    {
      "epoch": 0.6375554659136748,
      "grad_norm": 0.7382392883300781,
      "learning_rate": 3.0672006479080684e-06,
      "loss": 0.7195,
      "step": 22127
    },
    {
      "epoch": 0.6375842793753241,
      "grad_norm": 0.7176779508590698,
      "learning_rate": 3.066770318457252e-06,
      "loss": 0.7027,
      "step": 22128
    },
    {
      "epoch": 0.6376130928369734,
      "grad_norm": 0.714706540107727,
      "learning_rate": 3.0663400058429337e-06,
      "loss": 0.7004,
      "step": 22129
    },
    {
      "epoch": 0.6376419062986227,
      "grad_norm": 0.7043589949607849,
      "learning_rate": 3.065909710068861e-06,
      "loss": 0.6831,
      "step": 22130
    },
    {
      "epoch": 0.637670719760272,
      "grad_norm": 0.689273476600647,
      "learning_rate": 3.06547943113878e-06,
      "loss": 0.6897,
      "step": 22131
    },
    {
      "epoch": 0.6376995332219213,
      "grad_norm": 0.7350018620491028,
      "learning_rate": 3.065049169056441e-06,
      "loss": 0.7301,
      "step": 22132
    },
    {
      "epoch": 0.6377283466835706,
      "grad_norm": 0.7501224875450134,
      "learning_rate": 3.06461892382559e-06,
      "loss": 0.7069,
      "step": 22133
    },
    {
      "epoch": 0.6377571601452199,
      "grad_norm": 0.7379206418991089,
      "learning_rate": 3.0641886954499723e-06,
      "loss": 0.6864,
      "step": 22134
    },
    {
      "epoch": 0.6377859736068692,
      "grad_norm": 0.6872711777687073,
      "learning_rate": 3.0637584839333363e-06,
      "loss": 0.6923,
      "step": 22135
    },
    {
      "epoch": 0.6378147870685185,
      "grad_norm": 0.6970989108085632,
      "learning_rate": 3.0633282892794264e-06,
      "loss": 0.6669,
      "step": 22136
    },
    {
      "epoch": 0.6378436005301676,
      "grad_norm": 0.7200850248336792,
      "learning_rate": 3.062898111491993e-06,
      "loss": 0.724,
      "step": 22137
    },
    {
      "epoch": 0.6378724139918169,
      "grad_norm": 1.0053144693374634,
      "learning_rate": 3.062467950574781e-06,
      "loss": 0.6988,
      "step": 22138
    },
    {
      "epoch": 0.6379012274534662,
      "grad_norm": 0.7200734615325928,
      "learning_rate": 3.062037806531536e-06,
      "loss": 0.7063,
      "step": 22139
    },
    {
      "epoch": 0.6379300409151155,
      "grad_norm": 0.7586673498153687,
      "learning_rate": 3.0616076793660047e-06,
      "loss": 0.7094,
      "step": 22140
    },
    {
      "epoch": 0.6379588543767648,
      "grad_norm": 0.7239170074462891,
      "learning_rate": 3.061177569081933e-06,
      "loss": 0.7037,
      "step": 22141
    },
    {
      "epoch": 0.6379876678384141,
      "grad_norm": 0.7304947376251221,
      "learning_rate": 3.060747475683066e-06,
      "loss": 0.7134,
      "step": 22142
    },
    {
      "epoch": 0.6380164813000634,
      "grad_norm": 0.6925340294837952,
      "learning_rate": 3.0603173991731503e-06,
      "loss": 0.6643,
      "step": 22143
    },
    {
      "epoch": 0.6380452947617127,
      "grad_norm": 0.7242363095283508,
      "learning_rate": 3.0598873395559315e-06,
      "loss": 0.7098,
      "step": 22144
    },
    {
      "epoch": 0.638074108223362,
      "grad_norm": 0.7380668520927429,
      "learning_rate": 3.0594572968351544e-06,
      "loss": 0.739,
      "step": 22145
    },
    {
      "epoch": 0.6381029216850113,
      "grad_norm": 0.7193171977996826,
      "learning_rate": 3.059027271014564e-06,
      "loss": 0.6876,
      "step": 22146
    },
    {
      "epoch": 0.6381317351466606,
      "grad_norm": 0.7612470388412476,
      "learning_rate": 3.0585972620979054e-06,
      "loss": 0.6883,
      "step": 22147
    },
    {
      "epoch": 0.6381605486083098,
      "grad_norm": 0.727094829082489,
      "learning_rate": 3.058167270088923e-06,
      "loss": 0.6933,
      "step": 22148
    },
    {
      "epoch": 0.6381893620699591,
      "grad_norm": 0.7434351444244385,
      "learning_rate": 3.0577372949913643e-06,
      "loss": 0.6861,
      "step": 22149
    },
    {
      "epoch": 0.6382181755316084,
      "grad_norm": 0.7137936949729919,
      "learning_rate": 3.0573073368089714e-06,
      "loss": 0.6841,
      "step": 22150
    },
    {
      "epoch": 0.6382469889932576,
      "grad_norm": 0.7025521993637085,
      "learning_rate": 3.05687739554549e-06,
      "loss": 0.6701,
      "step": 22151
    },
    {
      "epoch": 0.6382758024549069,
      "grad_norm": 0.7782064080238342,
      "learning_rate": 3.0564474712046643e-06,
      "loss": 0.6912,
      "step": 22152
    },
    {
      "epoch": 0.6383046159165562,
      "grad_norm": 0.7294300198554993,
      "learning_rate": 3.056017563790238e-06,
      "loss": 0.6868,
      "step": 22153
    },
    {
      "epoch": 0.6383334293782055,
      "grad_norm": 0.7233031988143921,
      "learning_rate": 3.055587673305955e-06,
      "loss": 0.7023,
      "step": 22154
    },
    {
      "epoch": 0.6383622428398548,
      "grad_norm": 0.721646249294281,
      "learning_rate": 3.0551577997555605e-06,
      "loss": 0.6759,
      "step": 22155
    },
    {
      "epoch": 0.6383910563015041,
      "grad_norm": 0.6968464851379395,
      "learning_rate": 3.0547279431427975e-06,
      "loss": 0.6961,
      "step": 22156
    },
    {
      "epoch": 0.6384198697631533,
      "grad_norm": 0.7078377604484558,
      "learning_rate": 3.054298103471409e-06,
      "loss": 0.6911,
      "step": 22157
    },
    {
      "epoch": 0.6384486832248026,
      "grad_norm": 0.7081060409545898,
      "learning_rate": 3.0538682807451394e-06,
      "loss": 0.6952,
      "step": 22158
    },
    {
      "epoch": 0.6384774966864519,
      "grad_norm": 0.7258361577987671,
      "learning_rate": 3.0534384749677316e-06,
      "loss": 0.6989,
      "step": 22159
    },
    {
      "epoch": 0.6385063101481012,
      "grad_norm": 0.7274693250656128,
      "learning_rate": 3.053008686142927e-06,
      "loss": 0.7081,
      "step": 22160
    },
    {
      "epoch": 0.6385351236097505,
      "grad_norm": 0.7075396180152893,
      "learning_rate": 3.0525789142744723e-06,
      "loss": 0.7004,
      "step": 22161
    },
    {
      "epoch": 0.6385639370713998,
      "grad_norm": 0.7077327966690063,
      "learning_rate": 3.052149159366109e-06,
      "loss": 0.6811,
      "step": 22162
    },
    {
      "epoch": 0.6385927505330491,
      "grad_norm": 0.7221527695655823,
      "learning_rate": 3.0517194214215784e-06,
      "loss": 0.7212,
      "step": 22163
    },
    {
      "epoch": 0.6386215639946983,
      "grad_norm": 0.7193131446838379,
      "learning_rate": 3.051289700444625e-06,
      "loss": 0.7111,
      "step": 22164
    },
    {
      "epoch": 0.6386503774563476,
      "grad_norm": 0.7046977281570435,
      "learning_rate": 3.0508599964389897e-06,
      "loss": 0.6904,
      "step": 22165
    },
    {
      "epoch": 0.6386791909179969,
      "grad_norm": 0.7045478224754333,
      "learning_rate": 3.0504303094084153e-06,
      "loss": 0.6869,
      "step": 22166
    },
    {
      "epoch": 0.6387080043796461,
      "grad_norm": 0.7201099991798401,
      "learning_rate": 3.0500006393566452e-06,
      "loss": 0.7113,
      "step": 22167
    },
    {
      "epoch": 0.6387368178412954,
      "grad_norm": 0.7040644288063049,
      "learning_rate": 3.049570986287419e-06,
      "loss": 0.7028,
      "step": 22168
    },
    {
      "epoch": 0.6387656313029447,
      "grad_norm": 0.7327259182929993,
      "learning_rate": 3.0491413502044793e-06,
      "loss": 0.7023,
      "step": 22169
    },
    {
      "epoch": 0.638794444764594,
      "grad_norm": 0.7254106402397156,
      "learning_rate": 3.0487117311115687e-06,
      "loss": 0.6769,
      "step": 22170
    },
    {
      "epoch": 0.6388232582262433,
      "grad_norm": 0.7024914026260376,
      "learning_rate": 3.0482821290124293e-06,
      "loss": 0.6914,
      "step": 22171
    },
    {
      "epoch": 0.6388520716878926,
      "grad_norm": 0.7323818206787109,
      "learning_rate": 3.0478525439108005e-06,
      "loss": 0.6718,
      "step": 22172
    },
    {
      "epoch": 0.6388808851495419,
      "grad_norm": 0.7255306243896484,
      "learning_rate": 3.0474229758104244e-06,
      "loss": 0.7127,
      "step": 22173
    },
    {
      "epoch": 0.6389096986111912,
      "grad_norm": 0.7356736660003662,
      "learning_rate": 3.0469934247150424e-06,
      "loss": 0.6885,
      "step": 22174
    },
    {
      "epoch": 0.6389385120728405,
      "grad_norm": 0.7182745933532715,
      "learning_rate": 3.0465638906283967e-06,
      "loss": 0.6828,
      "step": 22175
    },
    {
      "epoch": 0.6389673255344898,
      "grad_norm": 0.7120933532714844,
      "learning_rate": 3.046134373554226e-06,
      "loss": 0.6977,
      "step": 22176
    },
    {
      "epoch": 0.638996138996139,
      "grad_norm": 0.7121425271034241,
      "learning_rate": 3.045704873496272e-06,
      "loss": 0.6932,
      "step": 22177
    },
    {
      "epoch": 0.6390249524577882,
      "grad_norm": 0.7131081223487854,
      "learning_rate": 3.045275390458275e-06,
      "loss": 0.7083,
      "step": 22178
    },
    {
      "epoch": 0.6390537659194375,
      "grad_norm": 0.7232043147087097,
      "learning_rate": 3.0448459244439755e-06,
      "loss": 0.6832,
      "step": 22179
    },
    {
      "epoch": 0.6390825793810868,
      "grad_norm": 0.7248901128768921,
      "learning_rate": 3.0444164754571134e-06,
      "loss": 0.6911,
      "step": 22180
    },
    {
      "epoch": 0.6391113928427361,
      "grad_norm": 0.7292195558547974,
      "learning_rate": 3.0439870435014284e-06,
      "loss": 0.7053,
      "step": 22181
    },
    {
      "epoch": 0.6391402063043854,
      "grad_norm": 0.7186197638511658,
      "learning_rate": 3.043557628580661e-06,
      "loss": 0.6924,
      "step": 22182
    },
    {
      "epoch": 0.6391690197660347,
      "grad_norm": 0.7214123010635376,
      "learning_rate": 3.043128230698552e-06,
      "loss": 0.6838,
      "step": 22183
    },
    {
      "epoch": 0.639197833227684,
      "grad_norm": 0.7467910051345825,
      "learning_rate": 3.0426988498588373e-06,
      "loss": 0.7078,
      "step": 22184
    },
    {
      "epoch": 0.6392266466893333,
      "grad_norm": 0.7053895592689514,
      "learning_rate": 3.0422694860652615e-06,
      "loss": 0.7061,
      "step": 22185
    },
    {
      "epoch": 0.6392554601509826,
      "grad_norm": 0.7323703169822693,
      "learning_rate": 3.0418401393215602e-06,
      "loss": 0.6876,
      "step": 22186
    },
    {
      "epoch": 0.6392842736126318,
      "grad_norm": 0.7235104441642761,
      "learning_rate": 3.041410809631475e-06,
      "loss": 0.7082,
      "step": 22187
    },
    {
      "epoch": 0.6393130870742811,
      "grad_norm": 0.745268702507019,
      "learning_rate": 3.0409814969987433e-06,
      "loss": 0.7106,
      "step": 22188
    },
    {
      "epoch": 0.6393419005359304,
      "grad_norm": 0.6971004009246826,
      "learning_rate": 3.040552201427105e-06,
      "loss": 0.7072,
      "step": 22189
    },
    {
      "epoch": 0.6393707139975797,
      "grad_norm": 0.7193477153778076,
      "learning_rate": 3.0401229229202965e-06,
      "loss": 0.6835,
      "step": 22190
    },
    {
      "epoch": 0.6393995274592289,
      "grad_norm": 0.7277636528015137,
      "learning_rate": 3.0396936614820592e-06,
      "loss": 0.6924,
      "step": 22191
    },
    {
      "epoch": 0.6394283409208782,
      "grad_norm": 0.7123545408248901,
      "learning_rate": 3.039264417116131e-06,
      "loss": 0.6878,
      "step": 22192
    },
    {
      "epoch": 0.6394571543825275,
      "grad_norm": 0.746929943561554,
      "learning_rate": 3.0388351898262502e-06,
      "loss": 0.6839,
      "step": 22193
    },
    {
      "epoch": 0.6394859678441768,
      "grad_norm": 0.7093217968940735,
      "learning_rate": 3.0384059796161536e-06,
      "loss": 0.6902,
      "step": 22194
    },
    {
      "epoch": 0.6395147813058261,
      "grad_norm": 0.7131466865539551,
      "learning_rate": 3.03797678648958e-06,
      "loss": 0.7239,
      "step": 22195
    },
    {
      "epoch": 0.6395435947674754,
      "grad_norm": 0.7243797779083252,
      "learning_rate": 3.0375476104502656e-06,
      "loss": 0.6905,
      "step": 22196
    },
    {
      "epoch": 0.6395724082291246,
      "grad_norm": 0.7131240367889404,
      "learning_rate": 3.037118451501952e-06,
      "loss": 0.7084,
      "step": 22197
    },
    {
      "epoch": 0.6396012216907739,
      "grad_norm": 0.7262237071990967,
      "learning_rate": 3.0366893096483742e-06,
      "loss": 0.6944,
      "step": 22198
    },
    {
      "epoch": 0.6396300351524232,
      "grad_norm": 0.7138838171958923,
      "learning_rate": 3.0362601848932704e-06,
      "loss": 0.705,
      "step": 22199
    },
    {
      "epoch": 0.6396588486140725,
      "grad_norm": 0.7103431224822998,
      "learning_rate": 3.0358310772403767e-06,
      "loss": 0.6897,
      "step": 22200
    },
    {
      "epoch": 0.6396876620757218,
      "grad_norm": 0.7313738465309143,
      "learning_rate": 3.0354019866934316e-06,
      "loss": 0.7065,
      "step": 22201
    },
    {
      "epoch": 0.6397164755373711,
      "grad_norm": 0.7123204469680786,
      "learning_rate": 3.03497291325617e-06,
      "loss": 0.6994,
      "step": 22202
    },
    {
      "epoch": 0.6397452889990204,
      "grad_norm": 0.7126231789588928,
      "learning_rate": 3.0345438569323315e-06,
      "loss": 0.715,
      "step": 22203
    },
    {
      "epoch": 0.6397741024606697,
      "grad_norm": 0.7074344158172607,
      "learning_rate": 3.0341148177256503e-06,
      "loss": 0.6645,
      "step": 22204
    },
    {
      "epoch": 0.6398029159223189,
      "grad_norm": 0.7103022336959839,
      "learning_rate": 3.033685795639864e-06,
      "loss": 0.7016,
      "step": 22205
    },
    {
      "epoch": 0.6398317293839682,
      "grad_norm": 0.6975786089897156,
      "learning_rate": 3.0332567906787093e-06,
      "loss": 0.6979,
      "step": 22206
    },
    {
      "epoch": 0.6398605428456174,
      "grad_norm": 0.7193869948387146,
      "learning_rate": 3.0328278028459206e-06,
      "loss": 0.7187,
      "step": 22207
    },
    {
      "epoch": 0.6398893563072667,
      "grad_norm": 0.7272768020629883,
      "learning_rate": 3.032398832145235e-06,
      "loss": 0.7064,
      "step": 22208
    },
    {
      "epoch": 0.639918169768916,
      "grad_norm": 0.7216091752052307,
      "learning_rate": 3.03196987858039e-06,
      "loss": 0.7009,
      "step": 22209
    },
    {
      "epoch": 0.6399469832305653,
      "grad_norm": 0.7601068019866943,
      "learning_rate": 3.03154094215512e-06,
      "loss": 0.7239,
      "step": 22210
    },
    {
      "epoch": 0.6399757966922146,
      "grad_norm": 0.7266823649406433,
      "learning_rate": 3.0311120228731612e-06,
      "loss": 0.7133,
      "step": 22211
    },
    {
      "epoch": 0.6400046101538639,
      "grad_norm": 0.7131710052490234,
      "learning_rate": 3.030683120738247e-06,
      "loss": 0.7085,
      "step": 22212
    },
    {
      "epoch": 0.6400334236155132,
      "grad_norm": 0.7042908072471619,
      "learning_rate": 3.0302542357541153e-06,
      "loss": 0.664,
      "step": 22213
    },
    {
      "epoch": 0.6400622370771625,
      "grad_norm": 0.7082758545875549,
      "learning_rate": 3.0298253679245005e-06,
      "loss": 0.679,
      "step": 22214
    },
    {
      "epoch": 0.6400910505388118,
      "grad_norm": 0.7158200740814209,
      "learning_rate": 3.029396517253137e-06,
      "loss": 0.6822,
      "step": 22215
    },
    {
      "epoch": 0.6401198640004611,
      "grad_norm": 0.7054420709609985,
      "learning_rate": 3.0289676837437596e-06,
      "loss": 0.6776,
      "step": 22216
    },
    {
      "epoch": 0.6401486774621103,
      "grad_norm": 0.7170012593269348,
      "learning_rate": 3.0285388674001036e-06,
      "loss": 0.7102,
      "step": 22217
    },
    {
      "epoch": 0.6401774909237595,
      "grad_norm": 0.7381331920623779,
      "learning_rate": 3.0281100682259036e-06,
      "loss": 0.684,
      "step": 22218
    },
    {
      "epoch": 0.6402063043854088,
      "grad_norm": 0.7327461838722229,
      "learning_rate": 3.0276812862248935e-06,
      "loss": 0.7024,
      "step": 22219
    },
    {
      "epoch": 0.6402351178470581,
      "grad_norm": 0.7179707884788513,
      "learning_rate": 3.0272525214008052e-06,
      "loss": 0.6842,
      "step": 22220
    },
    {
      "epoch": 0.6402639313087074,
      "grad_norm": 0.7386965751647949,
      "learning_rate": 3.0268237737573784e-06,
      "loss": 0.7001,
      "step": 22221
    },
    {
      "epoch": 0.6402927447703567,
      "grad_norm": 0.7153424620628357,
      "learning_rate": 3.0263950432983435e-06,
      "loss": 0.7141,
      "step": 22222
    },
    {
      "epoch": 0.640321558232006,
      "grad_norm": 0.7143352627754211,
      "learning_rate": 3.0259663300274354e-06,
      "loss": 0.7158,
      "step": 22223
    },
    {
      "epoch": 0.6403503716936553,
      "grad_norm": 0.7201923131942749,
      "learning_rate": 3.0255376339483865e-06,
      "loss": 0.6831,
      "step": 22224
    },
    {
      "epoch": 0.6403791851553046,
      "grad_norm": 0.7082753777503967,
      "learning_rate": 3.025108955064932e-06,
      "loss": 0.6922,
      "step": 22225
    },
    {
      "epoch": 0.6404079986169539,
      "grad_norm": 0.7224980592727661,
      "learning_rate": 3.024680293380804e-06,
      "loss": 0.703,
      "step": 22226
    },
    {
      "epoch": 0.6404368120786031,
      "grad_norm": 0.7121086716651917,
      "learning_rate": 3.0242516488997357e-06,
      "loss": 0.6809,
      "step": 22227
    },
    {
      "epoch": 0.6404656255402524,
      "grad_norm": 0.7507368326187134,
      "learning_rate": 3.0238230216254594e-06,
      "loss": 0.723,
      "step": 22228
    },
    {
      "epoch": 0.6404944390019017,
      "grad_norm": 0.712198793888092,
      "learning_rate": 3.023394411561711e-06,
      "loss": 0.6706,
      "step": 22229
    },
    {
      "epoch": 0.640523252463551,
      "grad_norm": 0.724593997001648,
      "learning_rate": 3.022965818712221e-06,
      "loss": 0.7074,
      "step": 22230
    },
    {
      "epoch": 0.6405520659252003,
      "grad_norm": 0.7147178053855896,
      "learning_rate": 3.022537243080722e-06,
      "loss": 0.6942,
      "step": 22231
    },
    {
      "epoch": 0.6405808793868495,
      "grad_norm": 0.7328315377235413,
      "learning_rate": 3.022108684670946e-06,
      "loss": 0.7215,
      "step": 22232
    },
    {
      "epoch": 0.6406096928484988,
      "grad_norm": 0.7368596196174622,
      "learning_rate": 3.0216801434866262e-06,
      "loss": 0.7057,
      "step": 22233
    },
    {
      "epoch": 0.6406385063101481,
      "grad_norm": 0.7421675324440002,
      "learning_rate": 3.0212516195314955e-06,
      "loss": 0.6969,
      "step": 22234
    },
    {
      "epoch": 0.6406673197717974,
      "grad_norm": 0.7239561080932617,
      "learning_rate": 3.0208231128092857e-06,
      "loss": 0.6993,
      "step": 22235
    },
    {
      "epoch": 0.6406961332334467,
      "grad_norm": 0.7473401427268982,
      "learning_rate": 3.020394623323728e-06,
      "loss": 0.7097,
      "step": 22236
    },
    {
      "epoch": 0.6407249466950959,
      "grad_norm": 0.7154179811477661,
      "learning_rate": 3.019966151078555e-06,
      "loss": 0.7176,
      "step": 22237
    },
    {
      "epoch": 0.6407537601567452,
      "grad_norm": 0.7132807374000549,
      "learning_rate": 3.0195376960774963e-06,
      "loss": 0.6978,
      "step": 22238
    },
    {
      "epoch": 0.6407825736183945,
      "grad_norm": 0.7285422682762146,
      "learning_rate": 3.0191092583242852e-06,
      "loss": 0.7059,
      "step": 22239
    },
    {
      "epoch": 0.6408113870800438,
      "grad_norm": 0.7255934476852417,
      "learning_rate": 3.0186808378226518e-06,
      "loss": 0.7018,
      "step": 22240
    },
    {
      "epoch": 0.6408402005416931,
      "grad_norm": 0.7270023226737976,
      "learning_rate": 3.018252434576328e-06,
      "loss": 0.7123,
      "step": 22241
    },
    {
      "epoch": 0.6408690140033424,
      "grad_norm": 0.7156025171279907,
      "learning_rate": 3.017824048589044e-06,
      "loss": 0.7116,
      "step": 22242
    },
    {
      "epoch": 0.6408978274649917,
      "grad_norm": 0.7065794467926025,
      "learning_rate": 3.017395679864532e-06,
      "loss": 0.6872,
      "step": 22243
    },
    {
      "epoch": 0.640926640926641,
      "grad_norm": 0.9525474905967712,
      "learning_rate": 3.0169673284065192e-06,
      "loss": 0.6829,
      "step": 22244
    },
    {
      "epoch": 0.6409554543882902,
      "grad_norm": 0.7406242489814758,
      "learning_rate": 3.016538994218741e-06,
      "loss": 0.7358,
      "step": 22245
    },
    {
      "epoch": 0.6409842678499394,
      "grad_norm": 0.7269386649131775,
      "learning_rate": 3.016110677304925e-06,
      "loss": 0.6914,
      "step": 22246
    },
    {
      "epoch": 0.6410130813115887,
      "grad_norm": 0.7136648297309875,
      "learning_rate": 3.0156823776688017e-06,
      "loss": 0.7072,
      "step": 22247
    },
    {
      "epoch": 0.641041894773238,
      "grad_norm": 0.7371382713317871,
      "learning_rate": 3.015254095314102e-06,
      "loss": 0.7245,
      "step": 22248
    },
    {
      "epoch": 0.6410707082348873,
      "grad_norm": 0.7162943482398987,
      "learning_rate": 3.0148258302445543e-06,
      "loss": 0.6772,
      "step": 22249
    },
    {
      "epoch": 0.6410995216965366,
      "grad_norm": 0.691692054271698,
      "learning_rate": 3.014397582463889e-06,
      "loss": 0.6923,
      "step": 22250
    },
    {
      "epoch": 0.6411283351581859,
      "grad_norm": 0.7278086543083191,
      "learning_rate": 3.0139693519758363e-06,
      "loss": 0.6988,
      "step": 22251
    },
    {
      "epoch": 0.6411571486198352,
      "grad_norm": 0.734830379486084,
      "learning_rate": 3.013541138784125e-06,
      "loss": 0.6934,
      "step": 22252
    },
    {
      "epoch": 0.6411859620814845,
      "grad_norm": 0.7347836494445801,
      "learning_rate": 3.0131129428924853e-06,
      "loss": 0.6973,
      "step": 22253
    },
    {
      "epoch": 0.6412147755431338,
      "grad_norm": 0.7534334659576416,
      "learning_rate": 3.0126847643046453e-06,
      "loss": 0.694,
      "step": 22254
    },
    {
      "epoch": 0.6412435890047831,
      "grad_norm": 0.7116450667381287,
      "learning_rate": 3.012256603024335e-06,
      "loss": 0.6793,
      "step": 22255
    },
    {
      "epoch": 0.6412724024664324,
      "grad_norm": 0.7347392439842224,
      "learning_rate": 3.0118284590552805e-06,
      "loss": 0.7252,
      "step": 22256
    },
    {
      "epoch": 0.6413012159280816,
      "grad_norm": 0.7064631581306458,
      "learning_rate": 3.011400332401214e-06,
      "loss": 0.7143,
      "step": 22257
    },
    {
      "epoch": 0.6413300293897308,
      "grad_norm": 0.7236300706863403,
      "learning_rate": 3.0109722230658634e-06,
      "loss": 0.682,
      "step": 22258
    },
    {
      "epoch": 0.6413588428513801,
      "grad_norm": 0.7529134154319763,
      "learning_rate": 3.010544131052956e-06,
      "loss": 0.6964,
      "step": 22259
    },
    {
      "epoch": 0.6413876563130294,
      "grad_norm": 0.7170368432998657,
      "learning_rate": 3.0101160563662206e-06,
      "loss": 0.6929,
      "step": 22260
    },
    {
      "epoch": 0.6414164697746787,
      "grad_norm": 0.7264660000801086,
      "learning_rate": 3.009687999009385e-06,
      "loss": 0.7078,
      "step": 22261
    },
    {
      "epoch": 0.641445283236328,
      "grad_norm": 0.7102206349372864,
      "learning_rate": 3.0092599589861784e-06,
      "loss": 0.6881,
      "step": 22262
    },
    {
      "epoch": 0.6414740966979773,
      "grad_norm": 0.7241936922073364,
      "learning_rate": 3.008831936300326e-06,
      "loss": 0.7108,
      "step": 22263
    },
    {
      "epoch": 0.6415029101596266,
      "grad_norm": 0.7205986380577087,
      "learning_rate": 3.008403930955558e-06,
      "loss": 0.7085,
      "step": 22264
    },
    {
      "epoch": 0.6415317236212759,
      "grad_norm": 0.7342608571052551,
      "learning_rate": 3.007975942955601e-06,
      "loss": 0.7219,
      "step": 22265
    },
    {
      "epoch": 0.6415605370829252,
      "grad_norm": 0.74864262342453,
      "learning_rate": 3.0075479723041805e-06,
      "loss": 0.7102,
      "step": 22266
    },
    {
      "epoch": 0.6415893505445744,
      "grad_norm": 0.7537721991539001,
      "learning_rate": 3.007120019005027e-06,
      "loss": 0.7111,
      "step": 22267
    },
    {
      "epoch": 0.6416181640062237,
      "grad_norm": 0.7323648929595947,
      "learning_rate": 3.0066920830618653e-06,
      "loss": 0.6915,
      "step": 22268
    },
    {
      "epoch": 0.641646977467873,
      "grad_norm": 0.7223057746887207,
      "learning_rate": 3.0062641644784228e-06,
      "loss": 0.6808,
      "step": 22269
    },
    {
      "epoch": 0.6416757909295223,
      "grad_norm": 0.6960482001304626,
      "learning_rate": 3.0058362632584272e-06,
      "loss": 0.6786,
      "step": 22270
    },
    {
      "epoch": 0.6417046043911716,
      "grad_norm": 0.7145681381225586,
      "learning_rate": 3.005408379405603e-06,
      "loss": 0.6926,
      "step": 22271
    },
    {
      "epoch": 0.6417334178528208,
      "grad_norm": 0.7398102879524231,
      "learning_rate": 3.0049805129236798e-06,
      "loss": 0.7054,
      "step": 22272
    },
    {
      "epoch": 0.6417622313144701,
      "grad_norm": 0.7469441890716553,
      "learning_rate": 3.0045526638163813e-06,
      "loss": 0.7169,
      "step": 22273
    },
    {
      "epoch": 0.6417910447761194,
      "grad_norm": 0.7151004672050476,
      "learning_rate": 3.0041248320874346e-06,
      "loss": 0.6878,
      "step": 22274
    },
    {
      "epoch": 0.6418198582377687,
      "grad_norm": 0.7193553447723389,
      "learning_rate": 3.003697017740565e-06,
      "loss": 0.6889,
      "step": 22275
    },
    {
      "epoch": 0.641848671699418,
      "grad_norm": 0.7300666570663452,
      "learning_rate": 3.003269220779499e-06,
      "loss": 0.7195,
      "step": 22276
    },
    {
      "epoch": 0.6418774851610672,
      "grad_norm": 0.7187131643295288,
      "learning_rate": 3.0028414412079624e-06,
      "loss": 0.7189,
      "step": 22277
    },
    {
      "epoch": 0.6419062986227165,
      "grad_norm": 0.7412844896316528,
      "learning_rate": 3.0024136790296805e-06,
      "loss": 0.7115,
      "step": 22278
    },
    {
      "epoch": 0.6419351120843658,
      "grad_norm": 0.7215620279312134,
      "learning_rate": 3.0019859342483782e-06,
      "loss": 0.7017,
      "step": 22279
    },
    {
      "epoch": 0.6419639255460151,
      "grad_norm": 0.7349560856819153,
      "learning_rate": 3.0015582068677797e-06,
      "loss": 0.6962,
      "step": 22280
    },
    {
      "epoch": 0.6419927390076644,
      "grad_norm": 0.7280459403991699,
      "learning_rate": 3.0011304968916127e-06,
      "loss": 0.7173,
      "step": 22281
    },
    {
      "epoch": 0.6420215524693137,
      "grad_norm": 0.7315027713775635,
      "learning_rate": 3.0007028043236012e-06,
      "loss": 0.7045,
      "step": 22282
    },
    {
      "epoch": 0.642050365930963,
      "grad_norm": 0.7550077438354492,
      "learning_rate": 3.0002751291674705e-06,
      "loss": 0.6972,
      "step": 22283
    },
    {
      "epoch": 0.6420791793926123,
      "grad_norm": 0.7015239000320435,
      "learning_rate": 2.999847471426943e-06,
      "loss": 0.6934,
      "step": 22284
    },
    {
      "epoch": 0.6421079928542615,
      "grad_norm": 0.715535044670105,
      "learning_rate": 2.9994198311057458e-06,
      "loss": 0.6837,
      "step": 22285
    },
    {
      "epoch": 0.6421368063159107,
      "grad_norm": 0.7411012649536133,
      "learning_rate": 2.9989922082076015e-06,
      "loss": 0.6971,
      "step": 22286
    },
    {
      "epoch": 0.64216561977756,
      "grad_norm": 0.7444515228271484,
      "learning_rate": 2.9985646027362348e-06,
      "loss": 0.7074,
      "step": 22287
    },
    {
      "epoch": 0.6421944332392093,
      "grad_norm": 0.7362445592880249,
      "learning_rate": 2.998137014695369e-06,
      "loss": 0.6995,
      "step": 22288
    },
    {
      "epoch": 0.6422232467008586,
      "grad_norm": 0.7286790013313293,
      "learning_rate": 2.99770944408873e-06,
      "loss": 0.7065,
      "step": 22289
    },
    {
      "epoch": 0.6422520601625079,
      "grad_norm": 0.7291229367256165,
      "learning_rate": 2.99728189092004e-06,
      "loss": 0.7085,
      "step": 22290
    },
    {
      "epoch": 0.6422808736241572,
      "grad_norm": 0.787103533744812,
      "learning_rate": 2.9968543551930216e-06,
      "loss": 0.7152,
      "step": 22291
    },
    {
      "epoch": 0.6423096870858065,
      "grad_norm": 0.7090399265289307,
      "learning_rate": 2.996426836911399e-06,
      "loss": 0.7113,
      "step": 22292
    },
    {
      "epoch": 0.6423385005474558,
      "grad_norm": 0.7639453411102295,
      "learning_rate": 2.9959993360788964e-06,
      "loss": 0.692,
      "step": 22293
    },
    {
      "epoch": 0.6423673140091051,
      "grad_norm": 0.7302831411361694,
      "learning_rate": 2.995571852699237e-06,
      "loss": 0.6942,
      "step": 22294
    },
    {
      "epoch": 0.6423961274707544,
      "grad_norm": 0.7086853384971619,
      "learning_rate": 2.9951443867761433e-06,
      "loss": 0.6977,
      "step": 22295
    },
    {
      "epoch": 0.6424249409324037,
      "grad_norm": 0.7460373044013977,
      "learning_rate": 2.9947169383133367e-06,
      "loss": 0.6902,
      "step": 22296
    },
    {
      "epoch": 0.642453754394053,
      "grad_norm": 0.7066158652305603,
      "learning_rate": 2.9942895073145423e-06,
      "loss": 0.6808,
      "step": 22297
    },
    {
      "epoch": 0.6424825678557022,
      "grad_norm": 0.7174572944641113,
      "learning_rate": 2.9938620937834808e-06,
      "loss": 0.7291,
      "step": 22298
    },
    {
      "epoch": 0.6425113813173514,
      "grad_norm": 0.7051315903663635,
      "learning_rate": 2.9934346977238748e-06,
      "loss": 0.7061,
      "step": 22299
    },
    {
      "epoch": 0.6425401947790007,
      "grad_norm": 0.7200163006782532,
      "learning_rate": 2.993007319139447e-06,
      "loss": 0.6894,
      "step": 22300
    },
    {
      "epoch": 0.64256900824065,
      "grad_norm": 0.7118121385574341,
      "learning_rate": 2.9925799580339197e-06,
      "loss": 0.6739,
      "step": 22301
    },
    {
      "epoch": 0.6425978217022993,
      "grad_norm": 0.7292795777320862,
      "learning_rate": 2.9921526144110136e-06,
      "loss": 0.6942,
      "step": 22302
    },
    {
      "epoch": 0.6426266351639486,
      "grad_norm": 0.738431990146637,
      "learning_rate": 2.9917252882744517e-06,
      "loss": 0.7116,
      "step": 22303
    },
    {
      "epoch": 0.6426554486255979,
      "grad_norm": 0.7316716313362122,
      "learning_rate": 2.991297979627953e-06,
      "loss": 0.6982,
      "step": 22304
    },
    {
      "epoch": 0.6426842620872472,
      "grad_norm": 0.7964605689048767,
      "learning_rate": 2.990870688475242e-06,
      "loss": 0.702,
      "step": 22305
    },
    {
      "epoch": 0.6427130755488965,
      "grad_norm": 0.7214410305023193,
      "learning_rate": 2.9904434148200396e-06,
      "loss": 0.7339,
      "step": 22306
    },
    {
      "epoch": 0.6427418890105457,
      "grad_norm": 0.7359793782234192,
      "learning_rate": 2.9900161586660658e-06,
      "loss": 0.6983,
      "step": 22307
    },
    {
      "epoch": 0.642770702472195,
      "grad_norm": 0.7213324308395386,
      "learning_rate": 2.9895889200170425e-06,
      "loss": 0.7279,
      "step": 22308
    },
    {
      "epoch": 0.6427995159338443,
      "grad_norm": 0.7292830944061279,
      "learning_rate": 2.989161698876689e-06,
      "loss": 0.6956,
      "step": 22309
    },
    {
      "epoch": 0.6428283293954936,
      "grad_norm": 0.73125821352005,
      "learning_rate": 2.988734495248728e-06,
      "loss": 0.6862,
      "step": 22310
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 0.7064748406410217,
      "learning_rate": 2.988307309136879e-06,
      "loss": 0.6844,
      "step": 22311
    },
    {
      "epoch": 0.6428859563187921,
      "grad_norm": 0.7149837017059326,
      "learning_rate": 2.987880140544862e-06,
      "loss": 0.6975,
      "step": 22312
    },
    {
      "epoch": 0.6429147697804414,
      "grad_norm": 0.70699542760849,
      "learning_rate": 2.9874529894763982e-06,
      "loss": 0.7006,
      "step": 22313
    },
    {
      "epoch": 0.6429435832420907,
      "grad_norm": 0.7099899649620056,
      "learning_rate": 2.9870258559352068e-06,
      "loss": 0.6997,
      "step": 22314
    },
    {
      "epoch": 0.64297239670374,
      "grad_norm": 0.7150484919548035,
      "learning_rate": 2.9865987399250075e-06,
      "loss": 0.7,
      "step": 22315
    },
    {
      "epoch": 0.6430012101653892,
      "grad_norm": 0.7072209715843201,
      "learning_rate": 2.986171641449519e-06,
      "loss": 0.6764,
      "step": 22316
    },
    {
      "epoch": 0.6430300236270385,
      "grad_norm": 0.7541460990905762,
      "learning_rate": 2.9857445605124646e-06,
      "loss": 0.697,
      "step": 22317
    },
    {
      "epoch": 0.6430588370886878,
      "grad_norm": 0.7472043037414551,
      "learning_rate": 2.9853174971175613e-06,
      "loss": 0.6885,
      "step": 22318
    },
    {
      "epoch": 0.6430876505503371,
      "grad_norm": 0.7121455073356628,
      "learning_rate": 2.9848904512685283e-06,
      "loss": 0.695,
      "step": 22319
    },
    {
      "epoch": 0.6431164640119864,
      "grad_norm": 0.7103589177131653,
      "learning_rate": 2.9844634229690854e-06,
      "loss": 0.6939,
      "step": 22320
    },
    {
      "epoch": 0.6431452774736357,
      "grad_norm": 0.7369937300682068,
      "learning_rate": 2.9840364122229514e-06,
      "loss": 0.6906,
      "step": 22321
    },
    {
      "epoch": 0.643174090935285,
      "grad_norm": 0.7089195847511292,
      "learning_rate": 2.983609419033845e-06,
      "loss": 0.6744,
      "step": 22322
    },
    {
      "epoch": 0.6432029043969343,
      "grad_norm": 0.7252972722053528,
      "learning_rate": 2.9831824434054844e-06,
      "loss": 0.7085,
      "step": 22323
    },
    {
      "epoch": 0.6432317178585836,
      "grad_norm": 0.6901273131370544,
      "learning_rate": 2.9827554853415896e-06,
      "loss": 0.7131,
      "step": 22324
    },
    {
      "epoch": 0.6432605313202329,
      "grad_norm": 0.7204646468162537,
      "learning_rate": 2.982328544845877e-06,
      "loss": 0.6961,
      "step": 22325
    },
    {
      "epoch": 0.643289344781882,
      "grad_norm": 0.7126560807228088,
      "learning_rate": 2.9819016219220653e-06,
      "loss": 0.689,
      "step": 22326
    },
    {
      "epoch": 0.6433181582435313,
      "grad_norm": 0.7183108329772949,
      "learning_rate": 2.9814747165738745e-06,
      "loss": 0.6825,
      "step": 22327
    },
    {
      "epoch": 0.6433469717051806,
      "grad_norm": 0.7188299894332886,
      "learning_rate": 2.98104782880502e-06,
      "loss": 0.6943,
      "step": 22328
    },
    {
      "epoch": 0.6433757851668299,
      "grad_norm": 0.7400548458099365,
      "learning_rate": 2.980620958619221e-06,
      "loss": 0.7158,
      "step": 22329
    },
    {
      "epoch": 0.6434045986284792,
      "grad_norm": 0.7048197984695435,
      "learning_rate": 2.9801941060201944e-06,
      "loss": 0.6904,
      "step": 22330
    },
    {
      "epoch": 0.6434334120901285,
      "grad_norm": 0.6983225345611572,
      "learning_rate": 2.9797672710116588e-06,
      "loss": 0.684,
      "step": 22331
    },
    {
      "epoch": 0.6434622255517778,
      "grad_norm": 0.7474731802940369,
      "learning_rate": 2.9793404535973315e-06,
      "loss": 0.6977,
      "step": 22332
    },
    {
      "epoch": 0.6434910390134271,
      "grad_norm": 0.7008471488952637,
      "learning_rate": 2.9789136537809282e-06,
      "loss": 0.6854,
      "step": 22333
    },
    {
      "epoch": 0.6435198524750764,
      "grad_norm": 0.7120786309242249,
      "learning_rate": 2.9784868715661664e-06,
      "loss": 0.6915,
      "step": 22334
    },
    {
      "epoch": 0.6435486659367257,
      "grad_norm": 0.7293053269386292,
      "learning_rate": 2.9780601069567638e-06,
      "loss": 0.6902,
      "step": 22335
    },
    {
      "epoch": 0.643577479398375,
      "grad_norm": 0.7306846380233765,
      "learning_rate": 2.9776333599564366e-06,
      "loss": 0.6893,
      "step": 22336
    },
    {
      "epoch": 0.6436062928600242,
      "grad_norm": 0.7175646424293518,
      "learning_rate": 2.9772066305689005e-06,
      "loss": 0.696,
      "step": 22337
    },
    {
      "epoch": 0.6436351063216735,
      "grad_norm": 0.6998761892318726,
      "learning_rate": 2.976779918797873e-06,
      "loss": 0.6692,
      "step": 22338
    },
    {
      "epoch": 0.6436639197833227,
      "grad_norm": 0.7400509119033813,
      "learning_rate": 2.9763532246470694e-06,
      "loss": 0.7004,
      "step": 22339
    },
    {
      "epoch": 0.643692733244972,
      "grad_norm": 0.7179574966430664,
      "learning_rate": 2.975926548120205e-06,
      "loss": 0.6975,
      "step": 22340
    },
    {
      "epoch": 0.6437215467066213,
      "grad_norm": 0.7288755178451538,
      "learning_rate": 2.9754998892209986e-06,
      "loss": 0.709,
      "step": 22341
    },
    {
      "epoch": 0.6437503601682706,
      "grad_norm": 0.712948203086853,
      "learning_rate": 2.975073247953165e-06,
      "loss": 0.6908,
      "step": 22342
    },
    {
      "epoch": 0.6437791736299199,
      "grad_norm": 0.7291764616966248,
      "learning_rate": 2.9746466243204186e-06,
      "loss": 0.6965,
      "step": 22343
    },
    {
      "epoch": 0.6438079870915692,
      "grad_norm": 0.713074266910553,
      "learning_rate": 2.974220018326476e-06,
      "loss": 0.7015,
      "step": 22344
    },
    {
      "epoch": 0.6438368005532185,
      "grad_norm": 0.8161442875862122,
      "learning_rate": 2.9737934299750514e-06,
      "loss": 0.6902,
      "step": 22345
    },
    {
      "epoch": 0.6438656140148677,
      "grad_norm": 0.7002776861190796,
      "learning_rate": 2.9733668592698612e-06,
      "loss": 0.6602,
      "step": 22346
    },
    {
      "epoch": 0.643894427476517,
      "grad_norm": 0.712103009223938,
      "learning_rate": 2.9729403062146185e-06,
      "loss": 0.7141,
      "step": 22347
    },
    {
      "epoch": 0.6439232409381663,
      "grad_norm": 0.7745777368545532,
      "learning_rate": 2.9725137708130403e-06,
      "loss": 0.6875,
      "step": 22348
    },
    {
      "epoch": 0.6439520543998156,
      "grad_norm": 0.7150496244430542,
      "learning_rate": 2.9720872530688406e-06,
      "loss": 0.7055,
      "step": 22349
    },
    {
      "epoch": 0.6439808678614649,
      "grad_norm": 0.692099392414093,
      "learning_rate": 2.971660752985733e-06,
      "loss": 0.6842,
      "step": 22350
    },
    {
      "epoch": 0.6440096813231142,
      "grad_norm": 0.7093541622161865,
      "learning_rate": 2.9712342705674325e-06,
      "loss": 0.6861,
      "step": 22351
    },
    {
      "epoch": 0.6440384947847635,
      "grad_norm": 0.7194904685020447,
      "learning_rate": 2.970807805817654e-06,
      "loss": 0.677,
      "step": 22352
    },
    {
      "epoch": 0.6440673082464127,
      "grad_norm": 0.7480924129486084,
      "learning_rate": 2.970381358740111e-06,
      "loss": 0.7106,
      "step": 22353
    },
    {
      "epoch": 0.644096121708062,
      "grad_norm": 0.7453442811965942,
      "learning_rate": 2.969954929338518e-06,
      "loss": 0.7188,
      "step": 22354
    },
    {
      "epoch": 0.6441249351697113,
      "grad_norm": 0.7123310565948486,
      "learning_rate": 2.9695285176165876e-06,
      "loss": 0.7083,
      "step": 22355
    },
    {
      "epoch": 0.6441537486313605,
      "grad_norm": 0.7325810194015503,
      "learning_rate": 2.969102123578035e-06,
      "loss": 0.6955,
      "step": 22356
    },
    {
      "epoch": 0.6441825620930098,
      "grad_norm": 0.7183694243431091,
      "learning_rate": 2.9686757472265714e-06,
      "loss": 0.6756,
      "step": 22357
    },
    {
      "epoch": 0.6442113755546591,
      "grad_norm": 0.7028579711914062,
      "learning_rate": 2.9682493885659127e-06,
      "loss": 0.6887,
      "step": 22358
    },
    {
      "epoch": 0.6442401890163084,
      "grad_norm": 0.7228008508682251,
      "learning_rate": 2.9678230475997705e-06,
      "loss": 0.6982,
      "step": 22359
    },
    {
      "epoch": 0.6442690024779577,
      "grad_norm": 0.7098310589790344,
      "learning_rate": 2.967396724331858e-06,
      "loss": 0.7112,
      "step": 22360
    },
    {
      "epoch": 0.644297815939607,
      "grad_norm": 0.7288869023323059,
      "learning_rate": 2.9669704187658887e-06,
      "loss": 0.7136,
      "step": 22361
    },
    {
      "epoch": 0.6443266294012563,
      "grad_norm": 0.7272970676422119,
      "learning_rate": 2.9665441309055745e-06,
      "loss": 0.6761,
      "step": 22362
    },
    {
      "epoch": 0.6443554428629056,
      "grad_norm": 0.7272999286651611,
      "learning_rate": 2.9661178607546257e-06,
      "loss": 0.7031,
      "step": 22363
    },
    {
      "epoch": 0.6443842563245549,
      "grad_norm": 0.733222246170044,
      "learning_rate": 2.9656916083167597e-06,
      "loss": 0.6927,
      "step": 22364
    },
    {
      "epoch": 0.6444130697862042,
      "grad_norm": 0.7379068732261658,
      "learning_rate": 2.965265373595686e-06,
      "loss": 0.7134,
      "step": 22365
    },
    {
      "epoch": 0.6444418832478533,
      "grad_norm": 0.7187634706497192,
      "learning_rate": 2.9648391565951166e-06,
      "loss": 0.7312,
      "step": 22366
    },
    {
      "epoch": 0.6444706967095026,
      "grad_norm": 0.742050290107727,
      "learning_rate": 2.964412957318764e-06,
      "loss": 0.7392,
      "step": 22367
    },
    {
      "epoch": 0.6444995101711519,
      "grad_norm": 0.7223367094993591,
      "learning_rate": 2.963986775770339e-06,
      "loss": 0.7095,
      "step": 22368
    },
    {
      "epoch": 0.6445283236328012,
      "grad_norm": 0.7255659103393555,
      "learning_rate": 2.963560611953555e-06,
      "loss": 0.719,
      "step": 22369
    },
    {
      "epoch": 0.6445571370944505,
      "grad_norm": 0.7783545255661011,
      "learning_rate": 2.9631344658721217e-06,
      "loss": 0.7109,
      "step": 22370
    },
    {
      "epoch": 0.6445859505560998,
      "grad_norm": 0.7187158465385437,
      "learning_rate": 2.9627083375297514e-06,
      "loss": 0.6945,
      "step": 22371
    },
    {
      "epoch": 0.6446147640177491,
      "grad_norm": 0.7099518775939941,
      "learning_rate": 2.9622822269301554e-06,
      "loss": 0.6953,
      "step": 22372
    },
    {
      "epoch": 0.6446435774793984,
      "grad_norm": 0.7412658333778381,
      "learning_rate": 2.9618561340770434e-06,
      "loss": 0.6785,
      "step": 22373
    },
    {
      "epoch": 0.6446723909410477,
      "grad_norm": 0.6888507604598999,
      "learning_rate": 2.961430058974128e-06,
      "loss": 0.7031,
      "step": 22374
    },
    {
      "epoch": 0.644701204402697,
      "grad_norm": 0.7235073447227478,
      "learning_rate": 2.9610040016251164e-06,
      "loss": 0.6972,
      "step": 22375
    },
    {
      "epoch": 0.6447300178643462,
      "grad_norm": 0.7464548349380493,
      "learning_rate": 2.9605779620337237e-06,
      "loss": 0.7094,
      "step": 22376
    },
    {
      "epoch": 0.6447588313259955,
      "grad_norm": 0.7131827473640442,
      "learning_rate": 2.9601519402036583e-06,
      "loss": 0.6982,
      "step": 22377
    },
    {
      "epoch": 0.6447876447876448,
      "grad_norm": 0.7145963311195374,
      "learning_rate": 2.9597259361386304e-06,
      "loss": 0.6979,
      "step": 22378
    },
    {
      "epoch": 0.6448164582492941,
      "grad_norm": 0.7142730355262756,
      "learning_rate": 2.9592999498423505e-06,
      "loss": 0.6854,
      "step": 22379
    },
    {
      "epoch": 0.6448452717109433,
      "grad_norm": 0.7119908332824707,
      "learning_rate": 2.9588739813185273e-06,
      "loss": 0.7083,
      "step": 22380
    },
    {
      "epoch": 0.6448740851725926,
      "grad_norm": 0.7108514904975891,
      "learning_rate": 2.9584480305708717e-06,
      "loss": 0.6995,
      "step": 22381
    },
    {
      "epoch": 0.6449028986342419,
      "grad_norm": 0.716336190700531,
      "learning_rate": 2.9580220976030933e-06,
      "loss": 0.7111,
      "step": 22382
    },
    {
      "epoch": 0.6449317120958912,
      "grad_norm": 0.7111443281173706,
      "learning_rate": 2.9575961824189008e-06,
      "loss": 0.7089,
      "step": 22383
    },
    {
      "epoch": 0.6449605255575405,
      "grad_norm": 0.7037239670753479,
      "learning_rate": 2.957170285022004e-06,
      "loss": 0.7155,
      "step": 22384
    },
    {
      "epoch": 0.6449893390191898,
      "grad_norm": 0.7017484903335571,
      "learning_rate": 2.956744405416111e-06,
      "loss": 0.6867,
      "step": 22385
    },
    {
      "epoch": 0.645018152480839,
      "grad_norm": 0.7102448344230652,
      "learning_rate": 2.9563185436049323e-06,
      "loss": 0.6831,
      "step": 22386
    },
    {
      "epoch": 0.6450469659424883,
      "grad_norm": 0.7161532640457153,
      "learning_rate": 2.9558926995921754e-06,
      "loss": 0.7173,
      "step": 22387
    },
    {
      "epoch": 0.6450757794041376,
      "grad_norm": 0.7218915224075317,
      "learning_rate": 2.955466873381551e-06,
      "loss": 0.7078,
      "step": 22388
    },
    {
      "epoch": 0.6451045928657869,
      "grad_norm": 0.7476140260696411,
      "learning_rate": 2.955041064976766e-06,
      "loss": 0.6868,
      "step": 22389
    },
    {
      "epoch": 0.6451334063274362,
      "grad_norm": 0.7138423323631287,
      "learning_rate": 2.954615274381529e-06,
      "loss": 0.698,
      "step": 22390
    },
    {
      "epoch": 0.6451622197890855,
      "grad_norm": 0.7409082055091858,
      "learning_rate": 2.954189501599548e-06,
      "loss": 0.7258,
      "step": 22391
    },
    {
      "epoch": 0.6451910332507348,
      "grad_norm": 0.7456968426704407,
      "learning_rate": 2.9537637466345327e-06,
      "loss": 0.6999,
      "step": 22392
    },
    {
      "epoch": 0.645219846712384,
      "grad_norm": 0.7258862257003784,
      "learning_rate": 2.953338009490189e-06,
      "loss": 0.7055,
      "step": 22393
    },
    {
      "epoch": 0.6452486601740333,
      "grad_norm": 0.734768807888031,
      "learning_rate": 2.9529122901702244e-06,
      "loss": 0.7149,
      "step": 22394
    },
    {
      "epoch": 0.6452774736356826,
      "grad_norm": 0.704550564289093,
      "learning_rate": 2.952486588678348e-06,
      "loss": 0.6835,
      "step": 22395
    },
    {
      "epoch": 0.6453062870973318,
      "grad_norm": 0.7107054591178894,
      "learning_rate": 2.952060905018268e-06,
      "loss": 0.7049,
      "step": 22396
    },
    {
      "epoch": 0.6453351005589811,
      "grad_norm": 0.7442728281021118,
      "learning_rate": 2.9516352391936888e-06,
      "loss": 0.6927,
      "step": 22397
    },
    {
      "epoch": 0.6453639140206304,
      "grad_norm": 0.7201215028762817,
      "learning_rate": 2.9512095912083192e-06,
      "loss": 0.6779,
      "step": 22398
    },
    {
      "epoch": 0.6453927274822797,
      "grad_norm": 0.7065861225128174,
      "learning_rate": 2.950783961065865e-06,
      "loss": 0.6797,
      "step": 22399
    },
    {
      "epoch": 0.645421540943929,
      "grad_norm": 0.7437618970870972,
      "learning_rate": 2.950358348770035e-06,
      "loss": 0.7125,
      "step": 22400
    },
    {
      "epoch": 0.6454503544055783,
      "grad_norm": 0.7090849876403809,
      "learning_rate": 2.9499327543245355e-06,
      "loss": 0.6991,
      "step": 22401
    },
    {
      "epoch": 0.6454791678672276,
      "grad_norm": 0.7459560036659241,
      "learning_rate": 2.949507177733072e-06,
      "loss": 0.6913,
      "step": 22402
    },
    {
      "epoch": 0.6455079813288769,
      "grad_norm": 0.720725953578949,
      "learning_rate": 2.9490816189993514e-06,
      "loss": 0.6903,
      "step": 22403
    },
    {
      "epoch": 0.6455367947905262,
      "grad_norm": 0.7170193195343018,
      "learning_rate": 2.9486560781270803e-06,
      "loss": 0.6948,
      "step": 22404
    },
    {
      "epoch": 0.6455656082521755,
      "grad_norm": 0.7254860401153564,
      "learning_rate": 2.948230555119963e-06,
      "loss": 0.7217,
      "step": 22405
    },
    {
      "epoch": 0.6455944217138248,
      "grad_norm": 0.7450532913208008,
      "learning_rate": 2.947805049981706e-06,
      "loss": 0.7193,
      "step": 22406
    },
    {
      "epoch": 0.6456232351754739,
      "grad_norm": 0.7204561829566956,
      "learning_rate": 2.9473795627160174e-06,
      "loss": 0.693,
      "step": 22407
    },
    {
      "epoch": 0.6456520486371232,
      "grad_norm": 0.7304114103317261,
      "learning_rate": 2.9469540933266e-06,
      "loss": 0.6841,
      "step": 22408
    },
    {
      "epoch": 0.6456808620987725,
      "grad_norm": 0.7049533724784851,
      "learning_rate": 2.9465286418171603e-06,
      "loss": 0.7008,
      "step": 22409
    },
    {
      "epoch": 0.6457096755604218,
      "grad_norm": 0.7156625986099243,
      "learning_rate": 2.946103208191403e-06,
      "loss": 0.701,
      "step": 22410
    },
    {
      "epoch": 0.6457384890220711,
      "grad_norm": 0.7070661783218384,
      "learning_rate": 2.945677792453032e-06,
      "loss": 0.6909,
      "step": 22411
    },
    {
      "epoch": 0.6457673024837204,
      "grad_norm": 0.7089433670043945,
      "learning_rate": 2.9452523946057563e-06,
      "loss": 0.7056,
      "step": 22412
    },
    {
      "epoch": 0.6457961159453697,
      "grad_norm": 0.7354558706283569,
      "learning_rate": 2.944827014653278e-06,
      "loss": 0.6921,
      "step": 22413
    },
    {
      "epoch": 0.645824929407019,
      "grad_norm": 0.7381880283355713,
      "learning_rate": 2.9444016525993023e-06,
      "loss": 0.7121,
      "step": 22414
    },
    {
      "epoch": 0.6458537428686683,
      "grad_norm": 0.7283166646957397,
      "learning_rate": 2.9439763084475325e-06,
      "loss": 0.7068,
      "step": 22415
    },
    {
      "epoch": 0.6458825563303175,
      "grad_norm": 0.7171803712844849,
      "learning_rate": 2.943550982201675e-06,
      "loss": 0.694,
      "step": 22416
    },
    {
      "epoch": 0.6459113697919668,
      "grad_norm": 0.7262516617774963,
      "learning_rate": 2.9431256738654314e-06,
      "loss": 0.7059,
      "step": 22417
    },
    {
      "epoch": 0.6459401832536161,
      "grad_norm": 0.7202348709106445,
      "learning_rate": 2.942700383442508e-06,
      "loss": 0.6745,
      "step": 22418
    },
    {
      "epoch": 0.6459689967152654,
      "grad_norm": 0.7185025215148926,
      "learning_rate": 2.942275110936608e-06,
      "loss": 0.7075,
      "step": 22419
    },
    {
      "epoch": 0.6459978101769146,
      "grad_norm": 0.7288778424263,
      "learning_rate": 2.941849856351433e-06,
      "loss": 0.67,
      "step": 22420
    },
    {
      "epoch": 0.6460266236385639,
      "grad_norm": 0.7185280323028564,
      "learning_rate": 2.9414246196906905e-06,
      "loss": 0.7129,
      "step": 22421
    },
    {
      "epoch": 0.6460554371002132,
      "grad_norm": 0.7389280200004578,
      "learning_rate": 2.94099940095808e-06,
      "loss": 0.6892,
      "step": 22422
    },
    {
      "epoch": 0.6460842505618625,
      "grad_norm": 0.730855405330658,
      "learning_rate": 2.9405742001573057e-06,
      "loss": 0.6963,
      "step": 22423
    },
    {
      "epoch": 0.6461130640235118,
      "grad_norm": 0.7227170467376709,
      "learning_rate": 2.940149017292073e-06,
      "loss": 0.7029,
      "step": 22424
    },
    {
      "epoch": 0.646141877485161,
      "grad_norm": 0.7169145941734314,
      "learning_rate": 2.939723852366083e-06,
      "loss": 0.7034,
      "step": 22425
    },
    {
      "epoch": 0.6461706909468103,
      "grad_norm": 0.7106072306632996,
      "learning_rate": 2.939298705383039e-06,
      "loss": 0.7028,
      "step": 22426
    },
    {
      "epoch": 0.6461995044084596,
      "grad_norm": 0.7162571549415588,
      "learning_rate": 2.938873576346644e-06,
      "loss": 0.6721,
      "step": 22427
    },
    {
      "epoch": 0.6462283178701089,
      "grad_norm": 0.7481997013092041,
      "learning_rate": 2.938448465260597e-06,
      "loss": 0.7094,
      "step": 22428
    },
    {
      "epoch": 0.6462571313317582,
      "grad_norm": 0.7024563550949097,
      "learning_rate": 2.9380233721286056e-06,
      "loss": 0.6815,
      "step": 22429
    },
    {
      "epoch": 0.6462859447934075,
      "grad_norm": 0.7073099613189697,
      "learning_rate": 2.937598296954369e-06,
      "loss": 0.6831,
      "step": 22430
    },
    {
      "epoch": 0.6463147582550568,
      "grad_norm": 0.7444270849227905,
      "learning_rate": 2.9371732397415897e-06,
      "loss": 0.7027,
      "step": 22431
    },
    {
      "epoch": 0.6463435717167061,
      "grad_norm": 0.7354768514633179,
      "learning_rate": 2.936748200493969e-06,
      "loss": 0.6752,
      "step": 22432
    },
    {
      "epoch": 0.6463723851783554,
      "grad_norm": 0.7366592884063721,
      "learning_rate": 2.936323179215209e-06,
      "loss": 0.7029,
      "step": 22433
    },
    {
      "epoch": 0.6464011986400046,
      "grad_norm": 0.7304686903953552,
      "learning_rate": 2.935898175909012e-06,
      "loss": 0.6764,
      "step": 22434
    },
    {
      "epoch": 0.6464300121016539,
      "grad_norm": 0.7210904359817505,
      "learning_rate": 2.935473190579076e-06,
      "loss": 0.6736,
      "step": 22435
    },
    {
      "epoch": 0.6464588255633031,
      "grad_norm": 0.7291707396507263,
      "learning_rate": 2.9350482232291067e-06,
      "loss": 0.7062,
      "step": 22436
    },
    {
      "epoch": 0.6464876390249524,
      "grad_norm": 0.7258890867233276,
      "learning_rate": 2.9346232738628033e-06,
      "loss": 0.6959,
      "step": 22437
    },
    {
      "epoch": 0.6465164524866017,
      "grad_norm": 0.7065961956977844,
      "learning_rate": 2.9341983424838662e-06,
      "loss": 0.683,
      "step": 22438
    },
    {
      "epoch": 0.646545265948251,
      "grad_norm": 0.7214728593826294,
      "learning_rate": 2.933773429095997e-06,
      "loss": 0.6954,
      "step": 22439
    },
    {
      "epoch": 0.6465740794099003,
      "grad_norm": 0.7282105684280396,
      "learning_rate": 2.9333485337028956e-06,
      "loss": 0.7165,
      "step": 22440
    },
    {
      "epoch": 0.6466028928715496,
      "grad_norm": 0.6801685690879822,
      "learning_rate": 2.9329236563082617e-06,
      "loss": 0.6699,
      "step": 22441
    },
    {
      "epoch": 0.6466317063331989,
      "grad_norm": 0.7227144241333008,
      "learning_rate": 2.9324987969157974e-06,
      "loss": 0.7314,
      "step": 22442
    },
    {
      "epoch": 0.6466605197948482,
      "grad_norm": 0.7320689558982849,
      "learning_rate": 2.932073955529202e-06,
      "loss": 0.6946,
      "step": 22443
    },
    {
      "epoch": 0.6466893332564975,
      "grad_norm": 0.6929979920387268,
      "learning_rate": 2.9316491321521732e-06,
      "loss": 0.7021,
      "step": 22444
    },
    {
      "epoch": 0.6467181467181468,
      "grad_norm": 0.720535397529602,
      "learning_rate": 2.9312243267884157e-06,
      "loss": 0.6808,
      "step": 22445
    },
    {
      "epoch": 0.646746960179796,
      "grad_norm": 0.7575318217277527,
      "learning_rate": 2.9307995394416244e-06,
      "loss": 0.6996,
      "step": 22446
    },
    {
      "epoch": 0.6467757736414452,
      "grad_norm": 0.7185328602790833,
      "learning_rate": 2.9303747701155007e-06,
      "loss": 0.6962,
      "step": 22447
    },
    {
      "epoch": 0.6468045871030945,
      "grad_norm": 0.7246094942092896,
      "learning_rate": 2.9299500188137446e-06,
      "loss": 0.6891,
      "step": 22448
    },
    {
      "epoch": 0.6468334005647438,
      "grad_norm": 0.7493335008621216,
      "learning_rate": 2.9295252855400536e-06,
      "loss": 0.6855,
      "step": 22449
    },
    {
      "epoch": 0.6468622140263931,
      "grad_norm": 0.7404229044914246,
      "learning_rate": 2.929100570298129e-06,
      "loss": 0.6965,
      "step": 22450
    },
    {
      "epoch": 0.6468910274880424,
      "grad_norm": 0.7180913686752319,
      "learning_rate": 2.9286758730916677e-06,
      "loss": 0.6961,
      "step": 22451
    },
    {
      "epoch": 0.6469198409496917,
      "grad_norm": 0.7436396479606628,
      "learning_rate": 2.928251193924369e-06,
      "loss": 0.7092,
      "step": 22452
    },
    {
      "epoch": 0.646948654411341,
      "grad_norm": 0.7081423997879028,
      "learning_rate": 2.927826532799932e-06,
      "loss": 0.7053,
      "step": 22453
    },
    {
      "epoch": 0.6469774678729903,
      "grad_norm": 0.725361704826355,
      "learning_rate": 2.9274018897220533e-06,
      "loss": 0.7223,
      "step": 22454
    },
    {
      "epoch": 0.6470062813346396,
      "grad_norm": 0.705316960811615,
      "learning_rate": 2.926977264694434e-06,
      "loss": 0.6676,
      "step": 22455
    },
    {
      "epoch": 0.6470350947962888,
      "grad_norm": 0.7157213687896729,
      "learning_rate": 2.9265526577207693e-06,
      "loss": 0.6981,
      "step": 22456
    },
    {
      "epoch": 0.6470639082579381,
      "grad_norm": 0.7081534266471863,
      "learning_rate": 2.9261280688047584e-06,
      "loss": 0.7077,
      "step": 22457
    },
    {
      "epoch": 0.6470927217195874,
      "grad_norm": 0.6916560530662537,
      "learning_rate": 2.9257034979500987e-06,
      "loss": 0.6967,
      "step": 22458
    },
    {
      "epoch": 0.6471215351812367,
      "grad_norm": 0.7032702565193176,
      "learning_rate": 2.925278945160487e-06,
      "loss": 0.6866,
      "step": 22459
    },
    {
      "epoch": 0.647150348642886,
      "grad_norm": 0.7261935472488403,
      "learning_rate": 2.9248544104396225e-06,
      "loss": 0.7206,
      "step": 22460
    },
    {
      "epoch": 0.6471791621045352,
      "grad_norm": 0.7103927731513977,
      "learning_rate": 2.9244298937912023e-06,
      "loss": 0.6822,
      "step": 22461
    },
    {
      "epoch": 0.6472079755661845,
      "grad_norm": 0.6860852241516113,
      "learning_rate": 2.924005395218923e-06,
      "loss": 0.6635,
      "step": 22462
    },
    {
      "epoch": 0.6472367890278338,
      "grad_norm": 0.7135419249534607,
      "learning_rate": 2.9235809147264814e-06,
      "loss": 0.6697,
      "step": 22463
    },
    {
      "epoch": 0.6472656024894831,
      "grad_norm": 0.7090957760810852,
      "learning_rate": 2.9231564523175745e-06,
      "loss": 0.699,
      "step": 22464
    },
    {
      "epoch": 0.6472944159511324,
      "grad_norm": 0.7651865482330322,
      "learning_rate": 2.9227320079958987e-06,
      "loss": 0.7399,
      "step": 22465
    },
    {
      "epoch": 0.6473232294127816,
      "grad_norm": 0.7246637940406799,
      "learning_rate": 2.9223075817651492e-06,
      "loss": 0.6847,
      "step": 22466
    },
    {
      "epoch": 0.6473520428744309,
      "grad_norm": 0.7173326015472412,
      "learning_rate": 2.9218831736290255e-06,
      "loss": 0.6981,
      "step": 22467
    },
    {
      "epoch": 0.6473808563360802,
      "grad_norm": 0.7138686776161194,
      "learning_rate": 2.921458783591221e-06,
      "loss": 0.7116,
      "step": 22468
    },
    {
      "epoch": 0.6474096697977295,
      "grad_norm": 0.7121193408966064,
      "learning_rate": 2.9210344116554333e-06,
      "loss": 0.6985,
      "step": 22469
    },
    {
      "epoch": 0.6474384832593788,
      "grad_norm": 0.7659503817558289,
      "learning_rate": 2.920610057825358e-06,
      "loss": 0.6963,
      "step": 22470
    },
    {
      "epoch": 0.6474672967210281,
      "grad_norm": 0.7079156637191772,
      "learning_rate": 2.9201857221046883e-06,
      "loss": 0.6887,
      "step": 22471
    },
    {
      "epoch": 0.6474961101826774,
      "grad_norm": 0.7277490496635437,
      "learning_rate": 2.9197614044971234e-06,
      "loss": 0.6999,
      "step": 22472
    },
    {
      "epoch": 0.6475249236443267,
      "grad_norm": 0.7137355208396912,
      "learning_rate": 2.9193371050063574e-06,
      "loss": 0.7049,
      "step": 22473
    },
    {
      "epoch": 0.6475537371059759,
      "grad_norm": 0.7400793433189392,
      "learning_rate": 2.9189128236360855e-06,
      "loss": 0.6844,
      "step": 22474
    },
    {
      "epoch": 0.6475825505676251,
      "grad_norm": 0.7178335785865784,
      "learning_rate": 2.9184885603900037e-06,
      "loss": 0.6855,
      "step": 22475
    },
    {
      "epoch": 0.6476113640292744,
      "grad_norm": 0.739137589931488,
      "learning_rate": 2.9180643152718035e-06,
      "loss": 0.6984,
      "step": 22476
    },
    {
      "epoch": 0.6476401774909237,
      "grad_norm": 0.7062402963638306,
      "learning_rate": 2.917640088285182e-06,
      "loss": 0.6799,
      "step": 22477
    },
    {
      "epoch": 0.647668990952573,
      "grad_norm": 0.7082576155662537,
      "learning_rate": 2.9172158794338358e-06,
      "loss": 0.7,
      "step": 22478
    },
    {
      "epoch": 0.6476978044142223,
      "grad_norm": 0.7064659595489502,
      "learning_rate": 2.9167916887214553e-06,
      "loss": 0.7052,
      "step": 22479
    },
    {
      "epoch": 0.6477266178758716,
      "grad_norm": 0.7173643112182617,
      "learning_rate": 2.916367516151738e-06,
      "loss": 0.7155,
      "step": 22480
    },
    {
      "epoch": 0.6477554313375209,
      "grad_norm": 0.7141420245170593,
      "learning_rate": 2.9159433617283763e-06,
      "loss": 0.6857,
      "step": 22481
    },
    {
      "epoch": 0.6477842447991702,
      "grad_norm": 0.7102960348129272,
      "learning_rate": 2.915519225455065e-06,
      "loss": 0.675,
      "step": 22482
    },
    {
      "epoch": 0.6478130582608195,
      "grad_norm": 0.7258324027061462,
      "learning_rate": 2.9150951073354945e-06,
      "loss": 0.6979,
      "step": 22483
    },
    {
      "epoch": 0.6478418717224688,
      "grad_norm": 0.740644633769989,
      "learning_rate": 2.914671007373365e-06,
      "loss": 0.7089,
      "step": 22484
    },
    {
      "epoch": 0.6478706851841181,
      "grad_norm": 0.7028132677078247,
      "learning_rate": 2.914246925572365e-06,
      "loss": 0.6897,
      "step": 22485
    },
    {
      "epoch": 0.6478994986457673,
      "grad_norm": 0.7168055772781372,
      "learning_rate": 2.913822861936191e-06,
      "loss": 0.6828,
      "step": 22486
    },
    {
      "epoch": 0.6479283121074166,
      "grad_norm": 0.7168281674385071,
      "learning_rate": 2.9133988164685323e-06,
      "loss": 0.6842,
      "step": 22487
    },
    {
      "epoch": 0.6479571255690658,
      "grad_norm": 0.7299140095710754,
      "learning_rate": 2.9129747891730863e-06,
      "loss": 0.7074,
      "step": 22488
    },
    {
      "epoch": 0.6479859390307151,
      "grad_norm": 0.7068875432014465,
      "learning_rate": 2.9125507800535414e-06,
      "loss": 0.6849,
      "step": 22489
    },
    {
      "epoch": 0.6480147524923644,
      "grad_norm": 0.7083548903465271,
      "learning_rate": 2.9121267891135953e-06,
      "loss": 0.6903,
      "step": 22490
    },
    {
      "epoch": 0.6480435659540137,
      "grad_norm": 0.7243070006370544,
      "learning_rate": 2.911702816356935e-06,
      "loss": 0.7106,
      "step": 22491
    },
    {
      "epoch": 0.648072379415663,
      "grad_norm": 0.7395899891853333,
      "learning_rate": 2.9112788617872577e-06,
      "loss": 0.716,
      "step": 22492
    },
    {
      "epoch": 0.6481011928773123,
      "grad_norm": 0.7459208965301514,
      "learning_rate": 2.910854925408252e-06,
      "loss": 0.6809,
      "step": 22493
    },
    {
      "epoch": 0.6481300063389616,
      "grad_norm": 0.7569031715393066,
      "learning_rate": 2.910431007223611e-06,
      "loss": 0.7045,
      "step": 22494
    },
    {
      "epoch": 0.6481588198006109,
      "grad_norm": 0.7089550495147705,
      "learning_rate": 2.9100071072370275e-06,
      "loss": 0.6899,
      "step": 22495
    },
    {
      "epoch": 0.6481876332622601,
      "grad_norm": 0.7084109783172607,
      "learning_rate": 2.9095832254521937e-06,
      "loss": 0.6757,
      "step": 22496
    },
    {
      "epoch": 0.6482164467239094,
      "grad_norm": 0.7178532481193542,
      "learning_rate": 2.9091593618728e-06,
      "loss": 0.6699,
      "step": 22497
    },
    {
      "epoch": 0.6482452601855587,
      "grad_norm": 0.7096004486083984,
      "learning_rate": 2.9087355165025373e-06,
      "loss": 0.6803,
      "step": 22498
    },
    {
      "epoch": 0.648274073647208,
      "grad_norm": 0.7322142124176025,
      "learning_rate": 2.9083116893451004e-06,
      "loss": 0.7077,
      "step": 22499
    },
    {
      "epoch": 0.6483028871088573,
      "grad_norm": 0.7396312355995178,
      "learning_rate": 2.9078878804041747e-06,
      "loss": 0.731,
      "step": 22500
    },
    {
      "epoch": 0.6483317005705065,
      "grad_norm": 0.7266117334365845,
      "learning_rate": 2.907464089683457e-06,
      "loss": 0.6769,
      "step": 22501
    },
    {
      "epoch": 0.6483605140321558,
      "grad_norm": 0.7211688160896301,
      "learning_rate": 2.907040317186633e-06,
      "loss": 0.7033,
      "step": 22502
    },
    {
      "epoch": 0.6483893274938051,
      "grad_norm": 0.7354827523231506,
      "learning_rate": 2.906616562917398e-06,
      "loss": 0.6876,
      "step": 22503
    },
    {
      "epoch": 0.6484181409554544,
      "grad_norm": 0.7125681638717651,
      "learning_rate": 2.9061928268794377e-06,
      "loss": 0.7251,
      "step": 22504
    },
    {
      "epoch": 0.6484469544171036,
      "grad_norm": 0.742199718952179,
      "learning_rate": 2.905769109076447e-06,
      "loss": 0.6876,
      "step": 22505
    },
    {
      "epoch": 0.6484757678787529,
      "grad_norm": 0.7292689085006714,
      "learning_rate": 2.905345409512112e-06,
      "loss": 0.7041,
      "step": 22506
    },
    {
      "epoch": 0.6485045813404022,
      "grad_norm": 0.7280778288841248,
      "learning_rate": 2.904921728190124e-06,
      "loss": 0.7037,
      "step": 22507
    },
    {
      "epoch": 0.6485333948020515,
      "grad_norm": 0.7080192565917969,
      "learning_rate": 2.904498065114174e-06,
      "loss": 0.7033,
      "step": 22508
    },
    {
      "epoch": 0.6485622082637008,
      "grad_norm": 0.7045794725418091,
      "learning_rate": 2.9040744202879533e-06,
      "loss": 0.6888,
      "step": 22509
    },
    {
      "epoch": 0.6485910217253501,
      "grad_norm": 0.9965324997901917,
      "learning_rate": 2.903650793715147e-06,
      "loss": 0.6922,
      "step": 22510
    },
    {
      "epoch": 0.6486198351869994,
      "grad_norm": 0.7240890860557556,
      "learning_rate": 2.9032271853994487e-06,
      "loss": 0.6909,
      "step": 22511
    },
    {
      "epoch": 0.6486486486486487,
      "grad_norm": 0.7064013481140137,
      "learning_rate": 2.9028035953445438e-06,
      "loss": 0.6893,
      "step": 22512
    },
    {
      "epoch": 0.648677462110298,
      "grad_norm": 0.7123515605926514,
      "learning_rate": 2.902380023554125e-06,
      "loss": 0.671,
      "step": 22513
    },
    {
      "epoch": 0.6487062755719473,
      "grad_norm": 0.7209389805793762,
      "learning_rate": 2.901956470031877e-06,
      "loss": 0.7186,
      "step": 22514
    },
    {
      "epoch": 0.6487350890335964,
      "grad_norm": 0.7169100642204285,
      "learning_rate": 2.901532934781491e-06,
      "loss": 0.6819,
      "step": 22515
    },
    {
      "epoch": 0.6487639024952457,
      "grad_norm": 0.7302496433258057,
      "learning_rate": 2.901109417806658e-06,
      "loss": 0.6822,
      "step": 22516
    },
    {
      "epoch": 0.648792715956895,
      "grad_norm": 0.7125430703163147,
      "learning_rate": 2.9006859191110616e-06,
      "loss": 0.6993,
      "step": 22517
    },
    {
      "epoch": 0.6488215294185443,
      "grad_norm": 0.7272990345954895,
      "learning_rate": 2.9002624386983942e-06,
      "loss": 0.6841,
      "step": 22518
    },
    {
      "epoch": 0.6488503428801936,
      "grad_norm": 0.7134331464767456,
      "learning_rate": 2.8998389765723377e-06,
      "loss": 0.6841,
      "step": 22519
    },
    {
      "epoch": 0.6488791563418429,
      "grad_norm": 0.7177175879478455,
      "learning_rate": 2.8994155327365893e-06,
      "loss": 0.6934,
      "step": 22520
    },
    {
      "epoch": 0.6489079698034922,
      "grad_norm": 0.6991111040115356,
      "learning_rate": 2.8989921071948286e-06,
      "loss": 0.6928,
      "step": 22521
    },
    {
      "epoch": 0.6489367832651415,
      "grad_norm": 0.7191221714019775,
      "learning_rate": 2.898568699950749e-06,
      "loss": 0.689,
      "step": 22522
    },
    {
      "epoch": 0.6489655967267908,
      "grad_norm": 0.7236664295196533,
      "learning_rate": 2.8981453110080332e-06,
      "loss": 0.6945,
      "step": 22523
    },
    {
      "epoch": 0.6489944101884401,
      "grad_norm": 0.7469300627708435,
      "learning_rate": 2.8977219403703725e-06,
      "loss": 0.6926,
      "step": 22524
    },
    {
      "epoch": 0.6490232236500894,
      "grad_norm": 0.7256972789764404,
      "learning_rate": 2.8972985880414506e-06,
      "loss": 0.7232,
      "step": 22525
    },
    {
      "epoch": 0.6490520371117386,
      "grad_norm": 0.713511049747467,
      "learning_rate": 2.896875254024958e-06,
      "loss": 0.7136,
      "step": 22526
    },
    {
      "epoch": 0.6490808505733879,
      "grad_norm": 0.7145788669586182,
      "learning_rate": 2.896451938324577e-06,
      "loss": 0.6719,
      "step": 22527
    },
    {
      "epoch": 0.6491096640350371,
      "grad_norm": 0.7277927398681641,
      "learning_rate": 2.8960286409439996e-06,
      "loss": 0.7146,
      "step": 22528
    },
    {
      "epoch": 0.6491384774966864,
      "grad_norm": 0.7072033882141113,
      "learning_rate": 2.895605361886907e-06,
      "loss": 0.6746,
      "step": 22529
    },
    {
      "epoch": 0.6491672909583357,
      "grad_norm": 0.7217221856117249,
      "learning_rate": 2.8951821011569904e-06,
      "loss": 0.691,
      "step": 22530
    },
    {
      "epoch": 0.649196104419985,
      "grad_norm": 0.7410507798194885,
      "learning_rate": 2.8947588587579296e-06,
      "loss": 0.6757,
      "step": 22531
    },
    {
      "epoch": 0.6492249178816343,
      "grad_norm": 0.7454981207847595,
      "learning_rate": 2.8943356346934187e-06,
      "loss": 0.7086,
      "step": 22532
    },
    {
      "epoch": 0.6492537313432836,
      "grad_norm": 0.721258282661438,
      "learning_rate": 2.8939124289671373e-06,
      "loss": 0.6934,
      "step": 22533
    },
    {
      "epoch": 0.6492825448049329,
      "grad_norm": 0.7254424095153809,
      "learning_rate": 2.8934892415827753e-06,
      "loss": 0.6835,
      "step": 22534
    },
    {
      "epoch": 0.6493113582665822,
      "grad_norm": 0.7245517373085022,
      "learning_rate": 2.8930660725440146e-06,
      "loss": 0.6895,
      "step": 22535
    },
    {
      "epoch": 0.6493401717282314,
      "grad_norm": 0.7346549034118652,
      "learning_rate": 2.892642921854542e-06,
      "loss": 0.6891,
      "step": 22536
    },
    {
      "epoch": 0.6493689851898807,
      "grad_norm": 0.7145788669586182,
      "learning_rate": 2.8922197895180446e-06,
      "loss": 0.6867,
      "step": 22537
    },
    {
      "epoch": 0.64939779865153,
      "grad_norm": 0.7258329391479492,
      "learning_rate": 2.891796675538205e-06,
      "loss": 0.7182,
      "step": 22538
    },
    {
      "epoch": 0.6494266121131793,
      "grad_norm": 0.7283792495727539,
      "learning_rate": 2.8913735799187105e-06,
      "loss": 0.6983,
      "step": 22539
    },
    {
      "epoch": 0.6494554255748286,
      "grad_norm": 0.7444437146186829,
      "learning_rate": 2.8909505026632414e-06,
      "loss": 0.6939,
      "step": 22540
    },
    {
      "epoch": 0.6494842390364779,
      "grad_norm": 0.719904899597168,
      "learning_rate": 2.8905274437754877e-06,
      "loss": 0.7037,
      "step": 22541
    },
    {
      "epoch": 0.6495130524981271,
      "grad_norm": 0.737730860710144,
      "learning_rate": 2.8901044032591295e-06,
      "loss": 0.7206,
      "step": 22542
    },
    {
      "epoch": 0.6495418659597764,
      "grad_norm": 0.7011299133300781,
      "learning_rate": 2.8896813811178525e-06,
      "loss": 0.6854,
      "step": 22543
    },
    {
      "epoch": 0.6495706794214257,
      "grad_norm": 0.7059560418128967,
      "learning_rate": 2.889258377355341e-06,
      "loss": 0.6996,
      "step": 22544
    },
    {
      "epoch": 0.649599492883075,
      "grad_norm": 0.6995071768760681,
      "learning_rate": 2.8888353919752817e-06,
      "loss": 0.6973,
      "step": 22545
    },
    {
      "epoch": 0.6496283063447242,
      "grad_norm": 0.7159459590911865,
      "learning_rate": 2.888412424981353e-06,
      "loss": 0.7069,
      "step": 22546
    },
    {
      "epoch": 0.6496571198063735,
      "grad_norm": 0.728245198726654,
      "learning_rate": 2.8879894763772436e-06,
      "loss": 0.7221,
      "step": 22547
    },
    {
      "epoch": 0.6496859332680228,
      "grad_norm": 0.7268314957618713,
      "learning_rate": 2.8875665461666326e-06,
      "loss": 0.7241,
      "step": 22548
    },
    {
      "epoch": 0.6497147467296721,
      "grad_norm": 0.7456420660018921,
      "learning_rate": 2.8871436343532076e-06,
      "loss": 0.6968,
      "step": 22549
    },
    {
      "epoch": 0.6497435601913214,
      "grad_norm": 0.736823320388794,
      "learning_rate": 2.886720740940647e-06,
      "loss": 0.6969,
      "step": 22550
    },
    {
      "epoch": 0.6497723736529707,
      "grad_norm": 0.6728193759918213,
      "learning_rate": 2.8862978659326383e-06,
      "loss": 0.6685,
      "step": 22551
    },
    {
      "epoch": 0.64980118711462,
      "grad_norm": 0.7052655220031738,
      "learning_rate": 2.8858750093328603e-06,
      "loss": 0.7054,
      "step": 22552
    },
    {
      "epoch": 0.6498300005762693,
      "grad_norm": 1.9643208980560303,
      "learning_rate": 2.885452171144998e-06,
      "loss": 0.7055,
      "step": 22553
    },
    {
      "epoch": 0.6498588140379186,
      "grad_norm": 0.7129037976264954,
      "learning_rate": 2.885029351372735e-06,
      "loss": 0.7343,
      "step": 22554
    },
    {
      "epoch": 0.6498876274995677,
      "grad_norm": 0.728846549987793,
      "learning_rate": 2.884606550019751e-06,
      "loss": 0.7062,
      "step": 22555
    },
    {
      "epoch": 0.649916440961217,
      "grad_norm": 0.7064206004142761,
      "learning_rate": 2.8841837670897283e-06,
      "loss": 0.6889,
      "step": 22556
    },
    {
      "epoch": 0.6499452544228663,
      "grad_norm": 0.7197592854499817,
      "learning_rate": 2.8837610025863505e-06,
      "loss": 0.6866,
      "step": 22557
    },
    {
      "epoch": 0.6499740678845156,
      "grad_norm": 0.7449872493743896,
      "learning_rate": 2.8833382565133005e-06,
      "loss": 0.71,
      "step": 22558
    },
    {
      "epoch": 0.6500028813461649,
      "grad_norm": 0.6860618591308594,
      "learning_rate": 2.8829155288742565e-06,
      "loss": 0.6911,
      "step": 22559
    },
    {
      "epoch": 0.6500316948078142,
      "grad_norm": 0.7320008277893066,
      "learning_rate": 2.882492819672904e-06,
      "loss": 0.7052,
      "step": 22560
    },
    {
      "epoch": 0.6500605082694635,
      "grad_norm": 0.7352539896965027,
      "learning_rate": 2.8820701289129206e-06,
      "loss": 0.7124,
      "step": 22561
    },
    {
      "epoch": 0.6500893217311128,
      "grad_norm": 0.6986526846885681,
      "learning_rate": 2.881647456597991e-06,
      "loss": 0.707,
      "step": 22562
    },
    {
      "epoch": 0.6501181351927621,
      "grad_norm": 0.7260563373565674,
      "learning_rate": 2.8812248027317924e-06,
      "loss": 0.6939,
      "step": 22563
    },
    {
      "epoch": 0.6501469486544114,
      "grad_norm": 0.7194879651069641,
      "learning_rate": 2.8808021673180096e-06,
      "loss": 0.6863,
      "step": 22564
    },
    {
      "epoch": 0.6501757621160607,
      "grad_norm": 0.6978635191917419,
      "learning_rate": 2.8803795503603203e-06,
      "loss": 0.721,
      "step": 22565
    },
    {
      "epoch": 0.6502045755777099,
      "grad_norm": 0.7686407566070557,
      "learning_rate": 2.8799569518624076e-06,
      "loss": 0.7124,
      "step": 22566
    },
    {
      "epoch": 0.6502333890393592,
      "grad_norm": 0.7569689750671387,
      "learning_rate": 2.8795343718279467e-06,
      "loss": 0.6809,
      "step": 22567
    },
    {
      "epoch": 0.6502622025010085,
      "grad_norm": 0.7405811548233032,
      "learning_rate": 2.8791118102606256e-06,
      "loss": 0.7344,
      "step": 22568
    },
    {
      "epoch": 0.6502910159626577,
      "grad_norm": 0.7446985840797424,
      "learning_rate": 2.8786892671641188e-06,
      "loss": 0.7045,
      "step": 22569
    },
    {
      "epoch": 0.650319829424307,
      "grad_norm": 0.7307578325271606,
      "learning_rate": 2.8782667425421094e-06,
      "loss": 0.7067,
      "step": 22570
    },
    {
      "epoch": 0.6503486428859563,
      "grad_norm": 0.7323693633079529,
      "learning_rate": 2.8778442363982746e-06,
      "loss": 0.689,
      "step": 22571
    },
    {
      "epoch": 0.6503774563476056,
      "grad_norm": 0.731625497341156,
      "learning_rate": 2.8774217487362964e-06,
      "loss": 0.6932,
      "step": 22572
    },
    {
      "epoch": 0.6504062698092549,
      "grad_norm": 0.7130777835845947,
      "learning_rate": 2.8769992795598513e-06,
      "loss": 0.6787,
      "step": 22573
    },
    {
      "epoch": 0.6504350832709042,
      "grad_norm": 0.734282374382019,
      "learning_rate": 2.8765768288726205e-06,
      "loss": 0.6962,
      "step": 22574
    },
    {
      "epoch": 0.6504638967325534,
      "grad_norm": 0.7202854156494141,
      "learning_rate": 2.8761543966782848e-06,
      "loss": 0.6867,
      "step": 22575
    },
    {
      "epoch": 0.6504927101942027,
      "grad_norm": 0.7177984714508057,
      "learning_rate": 2.8757319829805186e-06,
      "loss": 0.7113,
      "step": 22576
    },
    {
      "epoch": 0.650521523655852,
      "grad_norm": 0.7097844481468201,
      "learning_rate": 2.8753095877830057e-06,
      "loss": 0.687,
      "step": 22577
    },
    {
      "epoch": 0.6505503371175013,
      "grad_norm": 0.7128317952156067,
      "learning_rate": 2.87488721108942e-06,
      "loss": 0.7197,
      "step": 22578
    },
    {
      "epoch": 0.6505791505791506,
      "grad_norm": 0.7209876775741577,
      "learning_rate": 2.874464852903443e-06,
      "loss": 0.6901,
      "step": 22579
    },
    {
      "epoch": 0.6506079640407999,
      "grad_norm": 0.719853401184082,
      "learning_rate": 2.874042513228752e-06,
      "loss": 0.7012,
      "step": 22580
    },
    {
      "epoch": 0.6506367775024492,
      "grad_norm": 0.7311826944351196,
      "learning_rate": 2.8736201920690276e-06,
      "loss": 0.7196,
      "step": 22581
    },
    {
      "epoch": 0.6506655909640984,
      "grad_norm": 0.7110302448272705,
      "learning_rate": 2.8731978894279445e-06,
      "loss": 0.6855,
      "step": 22582
    },
    {
      "epoch": 0.6506944044257477,
      "grad_norm": 0.7062770128250122,
      "learning_rate": 2.8727756053091824e-06,
      "loss": 0.6822,
      "step": 22583
    },
    {
      "epoch": 0.650723217887397,
      "grad_norm": 0.728874146938324,
      "learning_rate": 2.872353339716417e-06,
      "loss": 0.6723,
      "step": 22584
    },
    {
      "epoch": 0.6507520313490462,
      "grad_norm": 0.7305811643600464,
      "learning_rate": 2.8719310926533293e-06,
      "loss": 0.72,
      "step": 22585
    },
    {
      "epoch": 0.6507808448106955,
      "grad_norm": 0.7223727107048035,
      "learning_rate": 2.871508864123592e-06,
      "loss": 0.6734,
      "step": 22586
    },
    {
      "epoch": 0.6508096582723448,
      "grad_norm": 0.7109036445617676,
      "learning_rate": 2.8710866541308873e-06,
      "loss": 0.7055,
      "step": 22587
    },
    {
      "epoch": 0.6508384717339941,
      "grad_norm": 0.7391239404678345,
      "learning_rate": 2.8706644626788875e-06,
      "loss": 0.7304,
      "step": 22588
    },
    {
      "epoch": 0.6508672851956434,
      "grad_norm": 0.700707197189331,
      "learning_rate": 2.8702422897712735e-06,
      "loss": 0.6813,
      "step": 22589
    },
    {
      "epoch": 0.6508960986572927,
      "grad_norm": 0.7091686725616455,
      "learning_rate": 2.8698201354117185e-06,
      "loss": 0.7042,
      "step": 22590
    },
    {
      "epoch": 0.650924912118942,
      "grad_norm": 0.7424256205558777,
      "learning_rate": 2.8693979996039007e-06,
      "loss": 0.6959,
      "step": 22591
    },
    {
      "epoch": 0.6509537255805913,
      "grad_norm": 0.7226291298866272,
      "learning_rate": 2.868975882351496e-06,
      "loss": 0.7004,
      "step": 22592
    },
    {
      "epoch": 0.6509825390422406,
      "grad_norm": 0.718127429485321,
      "learning_rate": 2.868553783658184e-06,
      "loss": 0.6934,
      "step": 22593
    },
    {
      "epoch": 0.6510113525038899,
      "grad_norm": 0.7230738997459412,
      "learning_rate": 2.868131703527635e-06,
      "loss": 0.6983,
      "step": 22594
    },
    {
      "epoch": 0.6510401659655392,
      "grad_norm": 0.7417945265769958,
      "learning_rate": 2.8677096419635286e-06,
      "loss": 0.7224,
      "step": 22595
    },
    {
      "epoch": 0.6510689794271883,
      "grad_norm": 0.7185912728309631,
      "learning_rate": 2.8672875989695408e-06,
      "loss": 0.6976,
      "step": 22596
    },
    {
      "epoch": 0.6510977928888376,
      "grad_norm": 0.7125122547149658,
      "learning_rate": 2.8668655745493444e-06,
      "loss": 0.7088,
      "step": 22597
    },
    {
      "epoch": 0.6511266063504869,
      "grad_norm": 0.7303187847137451,
      "learning_rate": 2.866443568706619e-06,
      "loss": 0.697,
      "step": 22598
    },
    {
      "epoch": 0.6511554198121362,
      "grad_norm": 0.6910660266876221,
      "learning_rate": 2.866021581445035e-06,
      "loss": 0.686,
      "step": 22599
    },
    {
      "epoch": 0.6511842332737855,
      "grad_norm": 0.725378692150116,
      "learning_rate": 2.8655996127682716e-06,
      "loss": 0.7,
      "step": 22600
    },
    {
      "epoch": 0.6512130467354348,
      "grad_norm": 0.7081360816955566,
      "learning_rate": 2.86517766268e-06,
      "loss": 0.6815,
      "step": 22601
    },
    {
      "epoch": 0.6512418601970841,
      "grad_norm": 0.6871315240859985,
      "learning_rate": 2.864755731183899e-06,
      "loss": 0.6802,
      "step": 22602
    },
    {
      "epoch": 0.6512706736587334,
      "grad_norm": 0.713822603225708,
      "learning_rate": 2.8643338182836378e-06,
      "loss": 0.6906,
      "step": 22603
    },
    {
      "epoch": 0.6512994871203827,
      "grad_norm": 0.7311662435531616,
      "learning_rate": 2.863911923982897e-06,
      "loss": 0.6954,
      "step": 22604
    },
    {
      "epoch": 0.651328300582032,
      "grad_norm": 0.7080343961715698,
      "learning_rate": 2.863490048285347e-06,
      "loss": 0.7025,
      "step": 22605
    },
    {
      "epoch": 0.6513571140436812,
      "grad_norm": 0.7505231499671936,
      "learning_rate": 2.8630681911946643e-06,
      "loss": 0.7191,
      "step": 22606
    },
    {
      "epoch": 0.6513859275053305,
      "grad_norm": 0.7070226073265076,
      "learning_rate": 2.8626463527145196e-06,
      "loss": 0.6925,
      "step": 22607
    },
    {
      "epoch": 0.6514147409669798,
      "grad_norm": 0.7154954671859741,
      "learning_rate": 2.862224532848591e-06,
      "loss": 0.6755,
      "step": 22608
    },
    {
      "epoch": 0.651443554428629,
      "grad_norm": 0.7156786322593689,
      "learning_rate": 2.861802731600547e-06,
      "loss": 0.6899,
      "step": 22609
    },
    {
      "epoch": 0.6514723678902783,
      "grad_norm": 0.7090457677841187,
      "learning_rate": 2.8613809489740665e-06,
      "loss": 0.6947,
      "step": 22610
    },
    {
      "epoch": 0.6515011813519276,
      "grad_norm": 0.7694104313850403,
      "learning_rate": 2.860959184972818e-06,
      "loss": 0.7309,
      "step": 22611
    },
    {
      "epoch": 0.6515299948135769,
      "grad_norm": 0.776171088218689,
      "learning_rate": 2.8605374396004764e-06,
      "loss": 0.6815,
      "step": 22612
    },
    {
      "epoch": 0.6515588082752262,
      "grad_norm": 4.3596577644348145,
      "learning_rate": 2.860115712860717e-06,
      "loss": 0.8628,
      "step": 22613
    },
    {
      "epoch": 0.6515876217368755,
      "grad_norm": 0.7092416286468506,
      "learning_rate": 2.859694004757209e-06,
      "loss": 0.6918,
      "step": 22614
    },
    {
      "epoch": 0.6516164351985247,
      "grad_norm": 0.7288814187049866,
      "learning_rate": 2.8592723152936264e-06,
      "loss": 0.7057,
      "step": 22615
    },
    {
      "epoch": 0.651645248660174,
      "grad_norm": 0.7373493313789368,
      "learning_rate": 2.8588506444736426e-06,
      "loss": 0.6982,
      "step": 22616
    },
    {
      "epoch": 0.6516740621218233,
      "grad_norm": 0.7126783728599548,
      "learning_rate": 2.8584289923009306e-06,
      "loss": 0.6982,
      "step": 22617
    },
    {
      "epoch": 0.6517028755834726,
      "grad_norm": 0.7020149827003479,
      "learning_rate": 2.8580073587791596e-06,
      "loss": 0.6943,
      "step": 22618
    },
    {
      "epoch": 0.6517316890451219,
      "grad_norm": 0.7334895133972168,
      "learning_rate": 2.8575857439120055e-06,
      "loss": 0.7106,
      "step": 22619
    },
    {
      "epoch": 0.6517605025067712,
      "grad_norm": 0.7174863815307617,
      "learning_rate": 2.857164147703135e-06,
      "loss": 0.6734,
      "step": 22620
    },
    {
      "epoch": 0.6517893159684205,
      "grad_norm": 0.6851210594177246,
      "learning_rate": 2.8567425701562256e-06,
      "loss": 0.6839,
      "step": 22621
    },
    {
      "epoch": 0.6518181294300698,
      "grad_norm": 0.7111652493476868,
      "learning_rate": 2.856321011274944e-06,
      "loss": 0.6797,
      "step": 22622
    },
    {
      "epoch": 0.651846942891719,
      "grad_norm": 0.7340456247329712,
      "learning_rate": 2.855899471062965e-06,
      "loss": 0.7103,
      "step": 22623
    },
    {
      "epoch": 0.6518757563533683,
      "grad_norm": 0.7349836230278015,
      "learning_rate": 2.855477949523957e-06,
      "loss": 0.6996,
      "step": 22624
    },
    {
      "epoch": 0.6519045698150175,
      "grad_norm": 0.6897672414779663,
      "learning_rate": 2.855056446661594e-06,
      "loss": 0.6926,
      "step": 22625
    },
    {
      "epoch": 0.6519333832766668,
      "grad_norm": 0.760270357131958,
      "learning_rate": 2.8546349624795406e-06,
      "loss": 0.7186,
      "step": 22626
    },
    {
      "epoch": 0.6519621967383161,
      "grad_norm": 0.7043461799621582,
      "learning_rate": 2.8542134969814765e-06,
      "loss": 0.6903,
      "step": 22627
    },
    {
      "epoch": 0.6519910101999654,
      "grad_norm": 0.7157301902770996,
      "learning_rate": 2.8537920501710658e-06,
      "loss": 0.7209,
      "step": 22628
    },
    {
      "epoch": 0.6520198236616147,
      "grad_norm": 0.9697567224502563,
      "learning_rate": 2.853370622051983e-06,
      "loss": 0.7146,
      "step": 22629
    },
    {
      "epoch": 0.652048637123264,
      "grad_norm": 0.7060866951942444,
      "learning_rate": 2.8529492126278946e-06,
      "loss": 0.6938,
      "step": 22630
    },
    {
      "epoch": 0.6520774505849133,
      "grad_norm": 0.7244524359703064,
      "learning_rate": 2.8525278219024736e-06,
      "loss": 0.7095,
      "step": 22631
    },
    {
      "epoch": 0.6521062640465626,
      "grad_norm": 0.7056377530097961,
      "learning_rate": 2.8521064498793872e-06,
      "loss": 0.7001,
      "step": 22632
    },
    {
      "epoch": 0.6521350775082119,
      "grad_norm": 0.7067700028419495,
      "learning_rate": 2.851685096562306e-06,
      "loss": 0.6824,
      "step": 22633
    },
    {
      "epoch": 0.6521638909698612,
      "grad_norm": 0.7185637950897217,
      "learning_rate": 2.851263761954902e-06,
      "loss": 0.6922,
      "step": 22634
    },
    {
      "epoch": 0.6521927044315105,
      "grad_norm": 0.7084699869155884,
      "learning_rate": 2.8508424460608414e-06,
      "loss": 0.7097,
      "step": 22635
    },
    {
      "epoch": 0.6522215178931596,
      "grad_norm": 0.6886876821517944,
      "learning_rate": 2.850421148883796e-06,
      "loss": 0.6724,
      "step": 22636
    },
    {
      "epoch": 0.6522503313548089,
      "grad_norm": 0.7146195769309998,
      "learning_rate": 2.8499998704274323e-06,
      "loss": 0.7007,
      "step": 22637
    },
    {
      "epoch": 0.6522791448164582,
      "grad_norm": 0.6944780349731445,
      "learning_rate": 2.84957861069542e-06,
      "loss": 0.6788,
      "step": 22638
    },
    {
      "epoch": 0.6523079582781075,
      "grad_norm": 0.7052404880523682,
      "learning_rate": 2.8491573696914288e-06,
      "loss": 0.7083,
      "step": 22639
    },
    {
      "epoch": 0.6523367717397568,
      "grad_norm": 0.7046145796775818,
      "learning_rate": 2.8487361474191283e-06,
      "loss": 0.6668,
      "step": 22640
    },
    {
      "epoch": 0.6523655852014061,
      "grad_norm": 0.7296460270881653,
      "learning_rate": 2.848314943882183e-06,
      "loss": 0.703,
      "step": 22641
    },
    {
      "epoch": 0.6523943986630554,
      "grad_norm": 0.7080866694450378,
      "learning_rate": 2.847893759084267e-06,
      "loss": 0.6855,
      "step": 22642
    },
    {
      "epoch": 0.6524232121247047,
      "grad_norm": 0.7151692509651184,
      "learning_rate": 2.8474725930290425e-06,
      "loss": 0.6831,
      "step": 22643
    },
    {
      "epoch": 0.652452025586354,
      "grad_norm": 0.727632462978363,
      "learning_rate": 2.8470514457201815e-06,
      "loss": 0.7139,
      "step": 22644
    },
    {
      "epoch": 0.6524808390480032,
      "grad_norm": 0.7305262088775635,
      "learning_rate": 2.846630317161349e-06,
      "loss": 0.6876,
      "step": 22645
    },
    {
      "epoch": 0.6525096525096525,
      "grad_norm": 0.7260972261428833,
      "learning_rate": 2.8462092073562155e-06,
      "loss": 0.6858,
      "step": 22646
    },
    {
      "epoch": 0.6525384659713018,
      "grad_norm": 0.7309412360191345,
      "learning_rate": 2.8457881163084445e-06,
      "loss": 0.7093,
      "step": 22647
    },
    {
      "epoch": 0.6525672794329511,
      "grad_norm": 0.7445622086524963,
      "learning_rate": 2.8453670440217084e-06,
      "loss": 0.7186,
      "step": 22648
    },
    {
      "epoch": 0.6525960928946004,
      "grad_norm": 0.7183495759963989,
      "learning_rate": 2.8449459904996692e-06,
      "loss": 0.7,
      "step": 22649
    },
    {
      "epoch": 0.6526249063562496,
      "grad_norm": 0.7213699221611023,
      "learning_rate": 2.8445249557459955e-06,
      "loss": 0.6893,
      "step": 22650
    },
    {
      "epoch": 0.6526537198178989,
      "grad_norm": 0.7236154079437256,
      "learning_rate": 2.844103939764355e-06,
      "loss": 0.6858,
      "step": 22651
    },
    {
      "epoch": 0.6526825332795482,
      "grad_norm": 0.7286574244499207,
      "learning_rate": 2.8436829425584168e-06,
      "loss": 0.7055,
      "step": 22652
    },
    {
      "epoch": 0.6527113467411975,
      "grad_norm": 0.7298591136932373,
      "learning_rate": 2.8432619641318416e-06,
      "loss": 0.7067,
      "step": 22653
    },
    {
      "epoch": 0.6527401602028468,
      "grad_norm": 0.7281907796859741,
      "learning_rate": 2.8428410044882995e-06,
      "loss": 0.7084,
      "step": 22654
    },
    {
      "epoch": 0.652768973664496,
      "grad_norm": 0.7178239822387695,
      "learning_rate": 2.8424200636314577e-06,
      "loss": 0.689,
      "step": 22655
    },
    {
      "epoch": 0.6527977871261453,
      "grad_norm": 0.7227667570114136,
      "learning_rate": 2.8419991415649784e-06,
      "loss": 0.711,
      "step": 22656
    },
    {
      "epoch": 0.6528266005877946,
      "grad_norm": 0.7223313450813293,
      "learning_rate": 2.8415782382925305e-06,
      "loss": 0.6966,
      "step": 22657
    },
    {
      "epoch": 0.6528554140494439,
      "grad_norm": 0.7230850458145142,
      "learning_rate": 2.841157353817777e-06,
      "loss": 0.6963,
      "step": 22658
    },
    {
      "epoch": 0.6528842275110932,
      "grad_norm": 0.7191225290298462,
      "learning_rate": 2.8407364881443878e-06,
      "loss": 0.7022,
      "step": 22659
    },
    {
      "epoch": 0.6529130409727425,
      "grad_norm": 0.7092326879501343,
      "learning_rate": 2.8403156412760222e-06,
      "loss": 0.6996,
      "step": 22660
    },
    {
      "epoch": 0.6529418544343918,
      "grad_norm": 0.7408410310745239,
      "learning_rate": 2.839894813216352e-06,
      "loss": 0.705,
      "step": 22661
    },
    {
      "epoch": 0.6529706678960411,
      "grad_norm": 0.7354315519332886,
      "learning_rate": 2.8394740039690327e-06,
      "loss": 0.7044,
      "step": 22662
    },
    {
      "epoch": 0.6529994813576903,
      "grad_norm": 0.7279836535453796,
      "learning_rate": 2.83905321353774e-06,
      "loss": 0.6914,
      "step": 22663
    },
    {
      "epoch": 0.6530282948193395,
      "grad_norm": 0.739605188369751,
      "learning_rate": 2.838632441926132e-06,
      "loss": 0.6877,
      "step": 22664
    },
    {
      "epoch": 0.6530571082809888,
      "grad_norm": 0.725570797920227,
      "learning_rate": 2.8382116891378764e-06,
      "loss": 0.6846,
      "step": 22665
    },
    {
      "epoch": 0.6530859217426381,
      "grad_norm": 0.7235711812973022,
      "learning_rate": 2.837790955176634e-06,
      "loss": 0.7026,
      "step": 22666
    },
    {
      "epoch": 0.6531147352042874,
      "grad_norm": 0.7198932766914368,
      "learning_rate": 2.8373702400460733e-06,
      "loss": 0.701,
      "step": 22667
    },
    {
      "epoch": 0.6531435486659367,
      "grad_norm": 0.7167763710021973,
      "learning_rate": 2.8369495437498534e-06,
      "loss": 0.717,
      "step": 22668
    },
    {
      "epoch": 0.653172362127586,
      "grad_norm": 0.7365218997001648,
      "learning_rate": 2.8365288662916434e-06,
      "loss": 0.6924,
      "step": 22669
    },
    {
      "epoch": 0.6532011755892353,
      "grad_norm": 0.7184078693389893,
      "learning_rate": 2.836108207675102e-06,
      "loss": 0.7022,
      "step": 22670
    },
    {
      "epoch": 0.6532299890508846,
      "grad_norm": 0.7360439300537109,
      "learning_rate": 2.835687567903895e-06,
      "loss": 0.6933,
      "step": 22671
    },
    {
      "epoch": 0.6532588025125339,
      "grad_norm": 0.7533944845199585,
      "learning_rate": 2.8352669469816874e-06,
      "loss": 0.7007,
      "step": 22672
    },
    {
      "epoch": 0.6532876159741832,
      "grad_norm": 0.735633134841919,
      "learning_rate": 2.8348463449121388e-06,
      "loss": 0.6941,
      "step": 22673
    },
    {
      "epoch": 0.6533164294358325,
      "grad_norm": 0.6992849707603455,
      "learning_rate": 2.8344257616989145e-06,
      "loss": 0.6935,
      "step": 22674
    },
    {
      "epoch": 0.6533452428974817,
      "grad_norm": 0.7392032146453857,
      "learning_rate": 2.8340051973456796e-06,
      "loss": 0.7145,
      "step": 22675
    },
    {
      "epoch": 0.653374056359131,
      "grad_norm": 0.71372389793396,
      "learning_rate": 2.8335846518560913e-06,
      "loss": 0.6749,
      "step": 22676
    },
    {
      "epoch": 0.6534028698207802,
      "grad_norm": 0.720969021320343,
      "learning_rate": 2.833164125233815e-06,
      "loss": 0.6963,
      "step": 22677
    },
    {
      "epoch": 0.6534316832824295,
      "grad_norm": 0.748744010925293,
      "learning_rate": 2.832743617482516e-06,
      "loss": 0.7025,
      "step": 22678
    },
    {
      "epoch": 0.6534604967440788,
      "grad_norm": 0.7443177103996277,
      "learning_rate": 2.8323231286058505e-06,
      "loss": 0.7102,
      "step": 22679
    },
    {
      "epoch": 0.6534893102057281,
      "grad_norm": 0.7289102673530579,
      "learning_rate": 2.8319026586074866e-06,
      "loss": 0.7097,
      "step": 22680
    },
    {
      "epoch": 0.6535181236673774,
      "grad_norm": 0.7323558330535889,
      "learning_rate": 2.8314822074910808e-06,
      "loss": 0.7075,
      "step": 22681
    },
    {
      "epoch": 0.6535469371290267,
      "grad_norm": 0.7143573760986328,
      "learning_rate": 2.8310617752603e-06,
      "loss": 0.7003,
      "step": 22682
    },
    {
      "epoch": 0.653575750590676,
      "grad_norm": 0.7151254415512085,
      "learning_rate": 2.8306413619188e-06,
      "loss": 0.6791,
      "step": 22683
    },
    {
      "epoch": 0.6536045640523253,
      "grad_norm": 0.7241737246513367,
      "learning_rate": 2.8302209674702472e-06,
      "loss": 0.6949,
      "step": 22684
    },
    {
      "epoch": 0.6536333775139745,
      "grad_norm": 0.7289827466011047,
      "learning_rate": 2.829800591918299e-06,
      "loss": 0.7017,
      "step": 22685
    },
    {
      "epoch": 0.6536621909756238,
      "grad_norm": 0.7171176671981812,
      "learning_rate": 2.829380235266619e-06,
      "loss": 0.7038,
      "step": 22686
    },
    {
      "epoch": 0.6536910044372731,
      "grad_norm": 0.7390042543411255,
      "learning_rate": 2.8289598975188657e-06,
      "loss": 0.702,
      "step": 22687
    },
    {
      "epoch": 0.6537198178989224,
      "grad_norm": 0.7096267342567444,
      "learning_rate": 2.828539578678705e-06,
      "loss": 0.6964,
      "step": 22688
    },
    {
      "epoch": 0.6537486313605717,
      "grad_norm": 0.726082444190979,
      "learning_rate": 2.8281192787497914e-06,
      "loss": 0.674,
      "step": 22689
    },
    {
      "epoch": 0.6537774448222209,
      "grad_norm": 0.7173848152160645,
      "learning_rate": 2.8276989977357893e-06,
      "loss": 0.6887,
      "step": 22690
    },
    {
      "epoch": 0.6538062582838702,
      "grad_norm": 0.7304043769836426,
      "learning_rate": 2.8272787356403566e-06,
      "loss": 0.6771,
      "step": 22691
    },
    {
      "epoch": 0.6538350717455195,
      "grad_norm": 0.7176221609115601,
      "learning_rate": 2.826858492467153e-06,
      "loss": 0.6865,
      "step": 22692
    },
    {
      "epoch": 0.6538638852071688,
      "grad_norm": 0.7674608826637268,
      "learning_rate": 2.826438268219842e-06,
      "loss": 0.7176,
      "step": 22693
    },
    {
      "epoch": 0.653892698668818,
      "grad_norm": 0.7560855150222778,
      "learning_rate": 2.826018062902079e-06,
      "loss": 0.7014,
      "step": 22694
    },
    {
      "epoch": 0.6539215121304673,
      "grad_norm": 0.6956045031547546,
      "learning_rate": 2.8255978765175274e-06,
      "loss": 0.6908,
      "step": 22695
    },
    {
      "epoch": 0.6539503255921166,
      "grad_norm": 0.7020297646522522,
      "learning_rate": 2.825177709069843e-06,
      "loss": 0.6873,
      "step": 22696
    },
    {
      "epoch": 0.6539791390537659,
      "grad_norm": 0.7164686918258667,
      "learning_rate": 2.824757560562688e-06,
      "loss": 0.6955,
      "step": 22697
    },
    {
      "epoch": 0.6540079525154152,
      "grad_norm": 0.7128530144691467,
      "learning_rate": 2.824337430999717e-06,
      "loss": 0.6587,
      "step": 22698
    },
    {
      "epoch": 0.6540367659770645,
      "grad_norm": 0.7280944585800171,
      "learning_rate": 2.823917320384596e-06,
      "loss": 0.7147,
      "step": 22699
    },
    {
      "epoch": 0.6540655794387138,
      "grad_norm": 0.7263684868812561,
      "learning_rate": 2.8234972287209773e-06,
      "loss": 0.6893,
      "step": 22700
    },
    {
      "epoch": 0.6540943929003631,
      "grad_norm": 0.7346769571304321,
      "learning_rate": 2.823077156012524e-06,
      "loss": 0.6923,
      "step": 22701
    },
    {
      "epoch": 0.6541232063620124,
      "grad_norm": 0.73627108335495,
      "learning_rate": 2.822657102262891e-06,
      "loss": 0.7118,
      "step": 22702
    },
    {
      "epoch": 0.6541520198236617,
      "grad_norm": 0.7607216835021973,
      "learning_rate": 2.8222370674757404e-06,
      "loss": 0.6907,
      "step": 22703
    },
    {
      "epoch": 0.6541808332853108,
      "grad_norm": 0.7347227334976196,
      "learning_rate": 2.821817051654725e-06,
      "loss": 0.7145,
      "step": 22704
    },
    {
      "epoch": 0.6542096467469601,
      "grad_norm": 0.7201369404792786,
      "learning_rate": 2.8213970548035085e-06,
      "loss": 0.6947,
      "step": 22705
    },
    {
      "epoch": 0.6542384602086094,
      "grad_norm": 0.7604854106903076,
      "learning_rate": 2.820977076925744e-06,
      "loss": 0.699,
      "step": 22706
    },
    {
      "epoch": 0.6542672736702587,
      "grad_norm": 0.6985368728637695,
      "learning_rate": 2.8205571180250923e-06,
      "loss": 0.639,
      "step": 22707
    },
    {
      "epoch": 0.654296087131908,
      "grad_norm": 0.7494000792503357,
      "learning_rate": 2.8201371781052077e-06,
      "loss": 0.7051,
      "step": 22708
    },
    {
      "epoch": 0.6543249005935573,
      "grad_norm": 0.7408681511878967,
      "learning_rate": 2.819717257169749e-06,
      "loss": 0.7018,
      "step": 22709
    },
    {
      "epoch": 0.6543537140552066,
      "grad_norm": 0.7425481677055359,
      "learning_rate": 2.8192973552223733e-06,
      "loss": 0.6967,
      "step": 22710
    },
    {
      "epoch": 0.6543825275168559,
      "grad_norm": 0.7363964319229126,
      "learning_rate": 2.8188774722667407e-06,
      "loss": 0.667,
      "step": 22711
    },
    {
      "epoch": 0.6544113409785052,
      "grad_norm": 0.7426129579544067,
      "learning_rate": 2.8184576083065024e-06,
      "loss": 0.7068,
      "step": 22712
    },
    {
      "epoch": 0.6544401544401545,
      "grad_norm": 0.720854640007019,
      "learning_rate": 2.8180377633453195e-06,
      "loss": 0.7014,
      "step": 22713
    },
    {
      "epoch": 0.6544689679018038,
      "grad_norm": 0.7194874882698059,
      "learning_rate": 2.8176179373868447e-06,
      "loss": 0.6828,
      "step": 22714
    },
    {
      "epoch": 0.654497781363453,
      "grad_norm": 0.7344549298286438,
      "learning_rate": 2.8171981304347364e-06,
      "loss": 0.7027,
      "step": 22715
    },
    {
      "epoch": 0.6545265948251023,
      "grad_norm": 0.7317225933074951,
      "learning_rate": 2.8167783424926516e-06,
      "loss": 0.7224,
      "step": 22716
    },
    {
      "epoch": 0.6545554082867515,
      "grad_norm": 0.7475550770759583,
      "learning_rate": 2.8163585735642428e-06,
      "loss": 0.7246,
      "step": 22717
    },
    {
      "epoch": 0.6545842217484008,
      "grad_norm": 0.7184105515480042,
      "learning_rate": 2.8159388236531716e-06,
      "loss": 0.7053,
      "step": 22718
    },
    {
      "epoch": 0.6546130352100501,
      "grad_norm": 0.7333577871322632,
      "learning_rate": 2.8155190927630867e-06,
      "loss": 0.715,
      "step": 22719
    },
    {
      "epoch": 0.6546418486716994,
      "grad_norm": 0.7170262336730957,
      "learning_rate": 2.815099380897649e-06,
      "loss": 0.7195,
      "step": 22720
    },
    {
      "epoch": 0.6546706621333487,
      "grad_norm": 0.7251346707344055,
      "learning_rate": 2.81467968806051e-06,
      "loss": 0.7085,
      "step": 22721
    },
    {
      "epoch": 0.654699475594998,
      "grad_norm": 0.734847366809845,
      "learning_rate": 2.8142600142553257e-06,
      "loss": 0.7004,
      "step": 22722
    },
    {
      "epoch": 0.6547282890566473,
      "grad_norm": 0.6988894939422607,
      "learning_rate": 2.8138403594857523e-06,
      "loss": 0.6881,
      "step": 22723
    },
    {
      "epoch": 0.6547571025182966,
      "grad_norm": 0.7190127372741699,
      "learning_rate": 2.8134207237554457e-06,
      "loss": 0.7232,
      "step": 22724
    },
    {
      "epoch": 0.6547859159799458,
      "grad_norm": 0.7280476093292236,
      "learning_rate": 2.813001107068057e-06,
      "loss": 0.6894,
      "step": 22725
    },
    {
      "epoch": 0.6548147294415951,
      "grad_norm": 0.6991137266159058,
      "learning_rate": 2.812581509427245e-06,
      "loss": 0.6699,
      "step": 22726
    },
    {
      "epoch": 0.6548435429032444,
      "grad_norm": 0.7064347863197327,
      "learning_rate": 2.8121619308366578e-06,
      "loss": 0.6695,
      "step": 22727
    },
    {
      "epoch": 0.6548723563648937,
      "grad_norm": 0.7360261082649231,
      "learning_rate": 2.8117423712999566e-06,
      "loss": 0.6978,
      "step": 22728
    },
    {
      "epoch": 0.654901169826543,
      "grad_norm": 0.7286866903305054,
      "learning_rate": 2.8113228308207894e-06,
      "loss": 0.6834,
      "step": 22729
    },
    {
      "epoch": 0.6549299832881923,
      "grad_norm": 0.7237406373023987,
      "learning_rate": 2.8109033094028125e-06,
      "loss": 0.7107,
      "step": 22730
    },
    {
      "epoch": 0.6549587967498415,
      "grad_norm": 0.7338764071464539,
      "learning_rate": 2.8104838070496813e-06,
      "loss": 0.702,
      "step": 22731
    },
    {
      "epoch": 0.6549876102114908,
      "grad_norm": 0.7163276076316833,
      "learning_rate": 2.810064323765046e-06,
      "loss": 0.7074,
      "step": 22732
    },
    {
      "epoch": 0.6550164236731401,
      "grad_norm": 0.7390914559364319,
      "learning_rate": 2.8096448595525627e-06,
      "loss": 0.6818,
      "step": 22733
    },
    {
      "epoch": 0.6550452371347893,
      "grad_norm": 0.7104548811912537,
      "learning_rate": 2.8092254144158815e-06,
      "loss": 0.6968,
      "step": 22734
    },
    {
      "epoch": 0.6550740505964386,
      "grad_norm": 0.6951335072517395,
      "learning_rate": 2.8088059883586573e-06,
      "loss": 0.6869,
      "step": 22735
    },
    {
      "epoch": 0.6551028640580879,
      "grad_norm": 0.7293068766593933,
      "learning_rate": 2.808386581384541e-06,
      "loss": 0.6915,
      "step": 22736
    },
    {
      "epoch": 0.6551316775197372,
      "grad_norm": 0.7257567644119263,
      "learning_rate": 2.80796719349719e-06,
      "loss": 0.6824,
      "step": 22737
    },
    {
      "epoch": 0.6551604909813865,
      "grad_norm": 0.7109553813934326,
      "learning_rate": 2.807547824700252e-06,
      "loss": 0.6979,
      "step": 22738
    },
    {
      "epoch": 0.6551893044430358,
      "grad_norm": 0.7156885266304016,
      "learning_rate": 2.8071284749973817e-06,
      "loss": 0.7016,
      "step": 22739
    },
    {
      "epoch": 0.6552181179046851,
      "grad_norm": 0.7069185972213745,
      "learning_rate": 2.806709144392229e-06,
      "loss": 0.6993,
      "step": 22740
    },
    {
      "epoch": 0.6552469313663344,
      "grad_norm": 0.7314323782920837,
      "learning_rate": 2.806289832888449e-06,
      "loss": 0.6888,
      "step": 22741
    },
    {
      "epoch": 0.6552757448279837,
      "grad_norm": 0.78468257188797,
      "learning_rate": 2.805870540489689e-06,
      "loss": 0.7301,
      "step": 22742
    },
    {
      "epoch": 0.655304558289633,
      "grad_norm": 0.7080190181732178,
      "learning_rate": 2.8054512671996063e-06,
      "loss": 0.6889,
      "step": 22743
    },
    {
      "epoch": 0.6553333717512821,
      "grad_norm": 0.7071105241775513,
      "learning_rate": 2.805032013021847e-06,
      "loss": 0.6991,
      "step": 22744
    },
    {
      "epoch": 0.6553621852129314,
      "grad_norm": 0.7191044092178345,
      "learning_rate": 2.8046127779600664e-06,
      "loss": 0.7108,
      "step": 22745
    },
    {
      "epoch": 0.6553909986745807,
      "grad_norm": 0.706304132938385,
      "learning_rate": 2.8041935620179104e-06,
      "loss": 0.6945,
      "step": 22746
    },
    {
      "epoch": 0.65541981213623,
      "grad_norm": 0.7155190706253052,
      "learning_rate": 2.803774365199037e-06,
      "loss": 0.7009,
      "step": 22747
    },
    {
      "epoch": 0.6554486255978793,
      "grad_norm": 0.738361120223999,
      "learning_rate": 2.803355187507092e-06,
      "loss": 0.6859,
      "step": 22748
    },
    {
      "epoch": 0.6554774390595286,
      "grad_norm": 0.7192994952201843,
      "learning_rate": 2.8029360289457285e-06,
      "loss": 0.6935,
      "step": 22749
    },
    {
      "epoch": 0.6555062525211779,
      "grad_norm": 0.7150110602378845,
      "learning_rate": 2.802516889518595e-06,
      "loss": 0.7097,
      "step": 22750
    },
    {
      "epoch": 0.6555350659828272,
      "grad_norm": 0.7156615853309631,
      "learning_rate": 2.8020977692293415e-06,
      "loss": 0.7202,
      "step": 22751
    },
    {
      "epoch": 0.6555638794444765,
      "grad_norm": 0.737762451171875,
      "learning_rate": 2.801678668081622e-06,
      "loss": 0.7247,
      "step": 22752
    },
    {
      "epoch": 0.6555926929061258,
      "grad_norm": 0.7501963376998901,
      "learning_rate": 2.8012595860790827e-06,
      "loss": 0.7153,
      "step": 22753
    },
    {
      "epoch": 0.655621506367775,
      "grad_norm": 0.7399746179580688,
      "learning_rate": 2.8008405232253756e-06,
      "loss": 0.7001,
      "step": 22754
    },
    {
      "epoch": 0.6556503198294243,
      "grad_norm": 0.7536689639091492,
      "learning_rate": 2.800421479524147e-06,
      "loss": 0.6862,
      "step": 22755
    },
    {
      "epoch": 0.6556791332910736,
      "grad_norm": 0.7213069200515747,
      "learning_rate": 2.8000024549790507e-06,
      "loss": 0.6941,
      "step": 22756
    },
    {
      "epoch": 0.6557079467527229,
      "grad_norm": 0.7232292294502258,
      "learning_rate": 2.799583449593732e-06,
      "loss": 0.6967,
      "step": 22757
    },
    {
      "epoch": 0.6557367602143721,
      "grad_norm": 0.7084513902664185,
      "learning_rate": 2.7991644633718417e-06,
      "loss": 0.6846,
      "step": 22758
    },
    {
      "epoch": 0.6557655736760214,
      "grad_norm": 0.7283313274383545,
      "learning_rate": 2.798745496317029e-06,
      "loss": 0.6828,
      "step": 22759
    },
    {
      "epoch": 0.6557943871376707,
      "grad_norm": 0.7155045866966248,
      "learning_rate": 2.798326548432944e-06,
      "loss": 0.6945,
      "step": 22760
    },
    {
      "epoch": 0.65582320059932,
      "grad_norm": 0.7452946305274963,
      "learning_rate": 2.7979076197232324e-06,
      "loss": 0.7216,
      "step": 22761
    },
    {
      "epoch": 0.6558520140609693,
      "grad_norm": 0.7049785256385803,
      "learning_rate": 2.797488710191546e-06,
      "loss": 0.7027,
      "step": 22762
    },
    {
      "epoch": 0.6558808275226186,
      "grad_norm": 0.713913083076477,
      "learning_rate": 2.7970698198415295e-06,
      "loss": 0.6742,
      "step": 22763
    },
    {
      "epoch": 0.6559096409842678,
      "grad_norm": 0.6975282430648804,
      "learning_rate": 2.7966509486768346e-06,
      "loss": 0.6818,
      "step": 22764
    },
    {
      "epoch": 0.6559384544459171,
      "grad_norm": 0.7190948724746704,
      "learning_rate": 2.7962320967011057e-06,
      "loss": 0.7065,
      "step": 22765
    },
    {
      "epoch": 0.6559672679075664,
      "grad_norm": 0.6998895406723022,
      "learning_rate": 2.7958132639179935e-06,
      "loss": 0.6781,
      "step": 22766
    },
    {
      "epoch": 0.6559960813692157,
      "grad_norm": 0.7159742116928101,
      "learning_rate": 2.7953944503311432e-06,
      "loss": 0.6736,
      "step": 22767
    },
    {
      "epoch": 0.656024894830865,
      "grad_norm": 0.7308415174484253,
      "learning_rate": 2.794975655944203e-06,
      "loss": 0.7145,
      "step": 22768
    },
    {
      "epoch": 0.6560537082925143,
      "grad_norm": 0.7247582674026489,
      "learning_rate": 2.7945568807608225e-06,
      "loss": 0.7048,
      "step": 22769
    },
    {
      "epoch": 0.6560825217541636,
      "grad_norm": 0.7321133017539978,
      "learning_rate": 2.7941381247846455e-06,
      "loss": 0.7025,
      "step": 22770
    },
    {
      "epoch": 0.6561113352158128,
      "grad_norm": 0.7117292284965515,
      "learning_rate": 2.7937193880193205e-06,
      "loss": 0.6827,
      "step": 22771
    },
    {
      "epoch": 0.6561401486774621,
      "grad_norm": 0.6910458207130432,
      "learning_rate": 2.793300670468495e-06,
      "loss": 0.6941,
      "step": 22772
    },
    {
      "epoch": 0.6561689621391114,
      "grad_norm": 0.708335280418396,
      "learning_rate": 2.7928819721358135e-06,
      "loss": 0.6963,
      "step": 22773
    },
    {
      "epoch": 0.6561977756007606,
      "grad_norm": 0.7133706212043762,
      "learning_rate": 2.792463293024924e-06,
      "loss": 0.6976,
      "step": 22774
    },
    {
      "epoch": 0.6562265890624099,
      "grad_norm": 0.7204100489616394,
      "learning_rate": 2.792044633139474e-06,
      "loss": 0.6856,
      "step": 22775
    },
    {
      "epoch": 0.6562554025240592,
      "grad_norm": 0.7100081443786621,
      "learning_rate": 2.7916259924831064e-06,
      "loss": 0.6949,
      "step": 22776
    },
    {
      "epoch": 0.6562842159857085,
      "grad_norm": 0.6945374011993408,
      "learning_rate": 2.7912073710594704e-06,
      "loss": 0.6914,
      "step": 22777
    },
    {
      "epoch": 0.6563130294473578,
      "grad_norm": 0.7057849168777466,
      "learning_rate": 2.7907887688722084e-06,
      "loss": 0.6951,
      "step": 22778
    },
    {
      "epoch": 0.6563418429090071,
      "grad_norm": 0.7209963202476501,
      "learning_rate": 2.79037018592497e-06,
      "loss": 0.7044,
      "step": 22779
    },
    {
      "epoch": 0.6563706563706564,
      "grad_norm": 0.7131214141845703,
      "learning_rate": 2.7899516222213962e-06,
      "loss": 0.7179,
      "step": 22780
    },
    {
      "epoch": 0.6563994698323057,
      "grad_norm": 0.7106849551200867,
      "learning_rate": 2.789533077765137e-06,
      "loss": 0.7051,
      "step": 22781
    },
    {
      "epoch": 0.656428283293955,
      "grad_norm": 0.7112374901771545,
      "learning_rate": 2.7891145525598307e-06,
      "loss": 0.688,
      "step": 22782
    },
    {
      "epoch": 0.6564570967556043,
      "grad_norm": 0.7096062302589417,
      "learning_rate": 2.7886960466091305e-06,
      "loss": 0.6939,
      "step": 22783
    },
    {
      "epoch": 0.6564859102172536,
      "grad_norm": 0.6980929374694824,
      "learning_rate": 2.7882775599166756e-06,
      "loss": 0.6548,
      "step": 22784
    },
    {
      "epoch": 0.6565147236789027,
      "grad_norm": 0.7284151315689087,
      "learning_rate": 2.7878590924861138e-06,
      "loss": 0.6936,
      "step": 22785
    },
    {
      "epoch": 0.656543537140552,
      "grad_norm": 0.730606198310852,
      "learning_rate": 2.7874406443210867e-06,
      "loss": 0.7061,
      "step": 22786
    },
    {
      "epoch": 0.6565723506022013,
      "grad_norm": 0.7160446047782898,
      "learning_rate": 2.7870222154252414e-06,
      "loss": 0.7032,
      "step": 22787
    },
    {
      "epoch": 0.6566011640638506,
      "grad_norm": 0.7753499150276184,
      "learning_rate": 2.7866038058022184e-06,
      "loss": 0.6836,
      "step": 22788
    },
    {
      "epoch": 0.6566299775254999,
      "grad_norm": 0.7031781077384949,
      "learning_rate": 2.7861854154556633e-06,
      "loss": 0.685,
      "step": 22789
    },
    {
      "epoch": 0.6566587909871492,
      "grad_norm": 0.7336552143096924,
      "learning_rate": 2.7857670443892226e-06,
      "loss": 0.712,
      "step": 22790
    },
    {
      "epoch": 0.6566876044487985,
      "grad_norm": 0.7146158218383789,
      "learning_rate": 2.7853486926065356e-06,
      "loss": 0.7007,
      "step": 22791
    },
    {
      "epoch": 0.6567164179104478,
      "grad_norm": 0.7156935930252075,
      "learning_rate": 2.784930360111249e-06,
      "loss": 0.6808,
      "step": 22792
    },
    {
      "epoch": 0.6567452313720971,
      "grad_norm": 0.7095080018043518,
      "learning_rate": 2.784512046907003e-06,
      "loss": 0.6755,
      "step": 22793
    },
    {
      "epoch": 0.6567740448337464,
      "grad_norm": 0.6918689608573914,
      "learning_rate": 2.784093752997443e-06,
      "loss": 0.6834,
      "step": 22794
    },
    {
      "epoch": 0.6568028582953956,
      "grad_norm": 0.745624840259552,
      "learning_rate": 2.783675478386211e-06,
      "loss": 0.7286,
      "step": 22795
    },
    {
      "epoch": 0.6568316717570449,
      "grad_norm": 0.7034336924552917,
      "learning_rate": 2.7832572230769517e-06,
      "loss": 0.7003,
      "step": 22796
    },
    {
      "epoch": 0.6568604852186942,
      "grad_norm": 0.7242170572280884,
      "learning_rate": 2.7828389870733047e-06,
      "loss": 0.7007,
      "step": 22797
    },
    {
      "epoch": 0.6568892986803434,
      "grad_norm": 0.7267312407493591,
      "learning_rate": 2.782420770378915e-06,
      "loss": 0.6904,
      "step": 22798
    },
    {
      "epoch": 0.6569181121419927,
      "grad_norm": 0.722649872303009,
      "learning_rate": 2.782002572997422e-06,
      "loss": 0.7001,
      "step": 22799
    },
    {
      "epoch": 0.656946925603642,
      "grad_norm": 0.710422694683075,
      "learning_rate": 2.7815843949324716e-06,
      "loss": 0.7056,
      "step": 22800
    },
    {
      "epoch": 0.6569757390652913,
      "grad_norm": 0.7374314069747925,
      "learning_rate": 2.7811662361877013e-06,
      "loss": 0.6898,
      "step": 22801
    },
    {
      "epoch": 0.6570045525269406,
      "grad_norm": 0.704156756401062,
      "learning_rate": 2.780748096766758e-06,
      "loss": 0.6972,
      "step": 22802
    },
    {
      "epoch": 0.6570333659885899,
      "grad_norm": 0.710726261138916,
      "learning_rate": 2.780329976673277e-06,
      "loss": 0.6914,
      "step": 22803
    },
    {
      "epoch": 0.6570621794502391,
      "grad_norm": 0.7016999125480652,
      "learning_rate": 2.7799118759109054e-06,
      "loss": 0.671,
      "step": 22804
    },
    {
      "epoch": 0.6570909929118884,
      "grad_norm": 0.7350764274597168,
      "learning_rate": 2.77949379448328e-06,
      "loss": 0.6952,
      "step": 22805
    },
    {
      "epoch": 0.6571198063735377,
      "grad_norm": 0.7255440950393677,
      "learning_rate": 2.7790757323940442e-06,
      "loss": 0.7188,
      "step": 22806
    },
    {
      "epoch": 0.657148619835187,
      "grad_norm": 0.7234339118003845,
      "learning_rate": 2.7786576896468387e-06,
      "loss": 0.6881,
      "step": 22807
    },
    {
      "epoch": 0.6571774332968363,
      "grad_norm": 0.7224119901657104,
      "learning_rate": 2.778239666245306e-06,
      "loss": 0.7175,
      "step": 22808
    },
    {
      "epoch": 0.6572062467584856,
      "grad_norm": 0.7530018091201782,
      "learning_rate": 2.777821662193083e-06,
      "loss": 0.685,
      "step": 22809
    },
    {
      "epoch": 0.6572350602201349,
      "grad_norm": 0.697856068611145,
      "learning_rate": 2.777403677493814e-06,
      "loss": 0.6769,
      "step": 22810
    },
    {
      "epoch": 0.6572638736817842,
      "grad_norm": 0.7317840456962585,
      "learning_rate": 2.776985712151135e-06,
      "loss": 0.7116,
      "step": 22811
    },
    {
      "epoch": 0.6572926871434334,
      "grad_norm": 0.7447496652603149,
      "learning_rate": 2.7765677661686887e-06,
      "loss": 0.668,
      "step": 22812
    },
    {
      "epoch": 0.6573215006050827,
      "grad_norm": 0.7317749857902527,
      "learning_rate": 2.776149839550116e-06,
      "loss": 0.7127,
      "step": 22813
    },
    {
      "epoch": 0.6573503140667319,
      "grad_norm": 0.7879668474197388,
      "learning_rate": 2.775731932299053e-06,
      "loss": 0.709,
      "step": 22814
    },
    {
      "epoch": 0.6573791275283812,
      "grad_norm": 0.7217226624488831,
      "learning_rate": 2.775314044419144e-06,
      "loss": 0.685,
      "step": 22815
    },
    {
      "epoch": 0.6574079409900305,
      "grad_norm": 0.7445046305656433,
      "learning_rate": 2.7748961759140236e-06,
      "loss": 0.6722,
      "step": 22816
    },
    {
      "epoch": 0.6574367544516798,
      "grad_norm": 0.7350589632987976,
      "learning_rate": 2.7744783267873356e-06,
      "loss": 0.7222,
      "step": 22817
    },
    {
      "epoch": 0.6574655679133291,
      "grad_norm": 0.7162060141563416,
      "learning_rate": 2.774060497042712e-06,
      "loss": 0.708,
      "step": 22818
    },
    {
      "epoch": 0.6574943813749784,
      "grad_norm": 0.7045892477035522,
      "learning_rate": 2.773642686683801e-06,
      "loss": 0.7251,
      "step": 22819
    },
    {
      "epoch": 0.6575231948366277,
      "grad_norm": 0.6981503367424011,
      "learning_rate": 2.7732248957142336e-06,
      "loss": 0.6705,
      "step": 22820
    },
    {
      "epoch": 0.657552008298277,
      "grad_norm": 0.749329149723053,
      "learning_rate": 2.7728071241376542e-06,
      "loss": 0.7141,
      "step": 22821
    },
    {
      "epoch": 0.6575808217599263,
      "grad_norm": 0.7312412261962891,
      "learning_rate": 2.772389371957696e-06,
      "loss": 0.684,
      "step": 22822
    },
    {
      "epoch": 0.6576096352215756,
      "grad_norm": 0.7325813174247742,
      "learning_rate": 2.7719716391780015e-06,
      "loss": 0.7089,
      "step": 22823
    },
    {
      "epoch": 0.6576384486832249,
      "grad_norm": 0.7342609763145447,
      "learning_rate": 2.7715539258022057e-06,
      "loss": 0.6908,
      "step": 22824
    },
    {
      "epoch": 0.657667262144874,
      "grad_norm": 0.7120869755744934,
      "learning_rate": 2.7711362318339487e-06,
      "loss": 0.6792,
      "step": 22825
    },
    {
      "epoch": 0.6576960756065233,
      "grad_norm": 0.6852805614471436,
      "learning_rate": 2.7707185572768657e-06,
      "loss": 0.6859,
      "step": 22826
    },
    {
      "epoch": 0.6577248890681726,
      "grad_norm": 0.690207839012146,
      "learning_rate": 2.770300902134595e-06,
      "loss": 0.6946,
      "step": 22827
    },
    {
      "epoch": 0.6577537025298219,
      "grad_norm": 0.7686630487442017,
      "learning_rate": 2.7698832664107767e-06,
      "loss": 0.735,
      "step": 22828
    },
    {
      "epoch": 0.6577825159914712,
      "grad_norm": 0.7004894614219666,
      "learning_rate": 2.7694656501090445e-06,
      "loss": 0.6999,
      "step": 22829
    },
    {
      "epoch": 0.6578113294531205,
      "grad_norm": 0.7037214040756226,
      "learning_rate": 2.7690480532330355e-06,
      "loss": 0.718,
      "step": 22830
    },
    {
      "epoch": 0.6578401429147698,
      "grad_norm": 0.7154520153999329,
      "learning_rate": 2.7686304757863915e-06,
      "loss": 0.6963,
      "step": 22831
    },
    {
      "epoch": 0.6578689563764191,
      "grad_norm": 0.7015055418014526,
      "learning_rate": 2.7682129177727424e-06,
      "loss": 0.6648,
      "step": 22832
    },
    {
      "epoch": 0.6578977698380684,
      "grad_norm": 0.7331849336624146,
      "learning_rate": 2.767795379195728e-06,
      "loss": 0.712,
      "step": 22833
    },
    {
      "epoch": 0.6579265832997176,
      "grad_norm": 0.7243738174438477,
      "learning_rate": 2.7673778600589862e-06,
      "loss": 0.6898,
      "step": 22834
    },
    {
      "epoch": 0.6579553967613669,
      "grad_norm": 0.7312775254249573,
      "learning_rate": 2.766960360366149e-06,
      "loss": 0.6909,
      "step": 22835
    },
    {
      "epoch": 0.6579842102230162,
      "grad_norm": 0.7198591232299805,
      "learning_rate": 2.7665428801208583e-06,
      "loss": 0.6984,
      "step": 22836
    },
    {
      "epoch": 0.6580130236846655,
      "grad_norm": 0.7218391299247742,
      "learning_rate": 2.7661254193267434e-06,
      "loss": 0.7078,
      "step": 22837
    },
    {
      "epoch": 0.6580418371463148,
      "grad_norm": 0.7104736566543579,
      "learning_rate": 2.7657079779874453e-06,
      "loss": 0.6945,
      "step": 22838
    },
    {
      "epoch": 0.658070650607964,
      "grad_norm": 0.7031086683273315,
      "learning_rate": 2.765290556106595e-06,
      "loss": 0.6801,
      "step": 22839
    },
    {
      "epoch": 0.6580994640696133,
      "grad_norm": 0.7619794607162476,
      "learning_rate": 2.7648731536878326e-06,
      "loss": 0.6873,
      "step": 22840
    },
    {
      "epoch": 0.6581282775312626,
      "grad_norm": 0.7194691896438599,
      "learning_rate": 2.7644557707347886e-06,
      "loss": 0.7032,
      "step": 22841
    },
    {
      "epoch": 0.6581570909929119,
      "grad_norm": 0.7306931614875793,
      "learning_rate": 2.7640384072510993e-06,
      "loss": 0.697,
      "step": 22842
    },
    {
      "epoch": 0.6581859044545612,
      "grad_norm": 0.726832389831543,
      "learning_rate": 2.7636210632404015e-06,
      "loss": 0.6953,
      "step": 22843
    },
    {
      "epoch": 0.6582147179162104,
      "grad_norm": 0.7081723213195801,
      "learning_rate": 2.7632037387063305e-06,
      "loss": 0.7073,
      "step": 22844
    },
    {
      "epoch": 0.6582435313778597,
      "grad_norm": 0.7316102981567383,
      "learning_rate": 2.7627864336525163e-06,
      "loss": 0.7166,
      "step": 22845
    },
    {
      "epoch": 0.658272344839509,
      "grad_norm": 0.718069314956665,
      "learning_rate": 2.7623691480825983e-06,
      "loss": 0.6907,
      "step": 22846
    },
    {
      "epoch": 0.6583011583011583,
      "grad_norm": 0.722997784614563,
      "learning_rate": 2.7619518820002055e-06,
      "loss": 0.7032,
      "step": 22847
    },
    {
      "epoch": 0.6583299717628076,
      "grad_norm": 0.732225775718689,
      "learning_rate": 2.7615346354089777e-06,
      "loss": 0.7011,
      "step": 22848
    },
    {
      "epoch": 0.6583587852244569,
      "grad_norm": 0.7254973649978638,
      "learning_rate": 2.7611174083125426e-06,
      "loss": 0.6826,
      "step": 22849
    },
    {
      "epoch": 0.6583875986861062,
      "grad_norm": 0.7150200605392456,
      "learning_rate": 2.760700200714538e-06,
      "loss": 0.6688,
      "step": 22850
    },
    {
      "epoch": 0.6584164121477555,
      "grad_norm": 0.7066197991371155,
      "learning_rate": 2.760283012618597e-06,
      "loss": 0.6912,
      "step": 22851
    },
    {
      "epoch": 0.6584452256094047,
      "grad_norm": 0.7428976893424988,
      "learning_rate": 2.7598658440283504e-06,
      "loss": 0.7259,
      "step": 22852
    },
    {
      "epoch": 0.658474039071054,
      "grad_norm": 0.7299926280975342,
      "learning_rate": 2.759448694947435e-06,
      "loss": 0.7148,
      "step": 22853
    },
    {
      "epoch": 0.6585028525327032,
      "grad_norm": 0.7244488596916199,
      "learning_rate": 2.759031565379478e-06,
      "loss": 0.6941,
      "step": 22854
    },
    {
      "epoch": 0.6585316659943525,
      "grad_norm": 0.7041251063346863,
      "learning_rate": 2.7586144553281196e-06,
      "loss": 0.6745,
      "step": 22855
    },
    {
      "epoch": 0.6585604794560018,
      "grad_norm": 0.72491455078125,
      "learning_rate": 2.758197364796987e-06,
      "loss": 0.7102,
      "step": 22856
    },
    {
      "epoch": 0.6585892929176511,
      "grad_norm": 0.7317894697189331,
      "learning_rate": 2.757780293789716e-06,
      "loss": 0.6995,
      "step": 22857
    },
    {
      "epoch": 0.6586181063793004,
      "grad_norm": 0.7241318225860596,
      "learning_rate": 2.7573632423099356e-06,
      "loss": 0.6904,
      "step": 22858
    },
    {
      "epoch": 0.6586469198409497,
      "grad_norm": 0.7055306434631348,
      "learning_rate": 2.7569462103612815e-06,
      "loss": 0.6971,
      "step": 22859
    },
    {
      "epoch": 0.658675733302599,
      "grad_norm": 0.7035240530967712,
      "learning_rate": 2.756529197947382e-06,
      "loss": 0.6997,
      "step": 22860
    },
    {
      "epoch": 0.6587045467642483,
      "grad_norm": 0.7197920083999634,
      "learning_rate": 2.7561122050718715e-06,
      "loss": 0.7021,
      "step": 22861
    },
    {
      "epoch": 0.6587333602258976,
      "grad_norm": 0.7229263782501221,
      "learning_rate": 2.7556952317383797e-06,
      "loss": 0.7043,
      "step": 22862
    },
    {
      "epoch": 0.6587621736875469,
      "grad_norm": 0.7386353611946106,
      "learning_rate": 2.7552782779505404e-06,
      "loss": 0.7069,
      "step": 22863
    },
    {
      "epoch": 0.6587909871491961,
      "grad_norm": 0.7256569862365723,
      "learning_rate": 2.7548613437119813e-06,
      "loss": 0.7023,
      "step": 22864
    },
    {
      "epoch": 0.6588198006108454,
      "grad_norm": 0.7048722505569458,
      "learning_rate": 2.7544444290263363e-06,
      "loss": 0.726,
      "step": 22865
    },
    {
      "epoch": 0.6588486140724946,
      "grad_norm": 0.7360844016075134,
      "learning_rate": 2.7540275338972345e-06,
      "loss": 0.6915,
      "step": 22866
    },
    {
      "epoch": 0.6588774275341439,
      "grad_norm": 0.733614444732666,
      "learning_rate": 2.7536106583283105e-06,
      "loss": 0.6987,
      "step": 22867
    },
    {
      "epoch": 0.6589062409957932,
      "grad_norm": 0.7240250110626221,
      "learning_rate": 2.7531938023231896e-06,
      "loss": 0.7005,
      "step": 22868
    },
    {
      "epoch": 0.6589350544574425,
      "grad_norm": 0.7335927486419678,
      "learning_rate": 2.7527769658855076e-06,
      "loss": 0.6891,
      "step": 22869
    },
    {
      "epoch": 0.6589638679190918,
      "grad_norm": 0.6975252032279968,
      "learning_rate": 2.752360149018889e-06,
      "loss": 0.6885,
      "step": 22870
    },
    {
      "epoch": 0.6589926813807411,
      "grad_norm": 0.7280068397521973,
      "learning_rate": 2.7519433517269665e-06,
      "loss": 0.7195,
      "step": 22871
    },
    {
      "epoch": 0.6590214948423904,
      "grad_norm": 0.7325619459152222,
      "learning_rate": 2.751526574013373e-06,
      "loss": 0.6791,
      "step": 22872
    },
    {
      "epoch": 0.6590503083040397,
      "grad_norm": 0.7157217264175415,
      "learning_rate": 2.751109815881733e-06,
      "loss": 0.6773,
      "step": 22873
    },
    {
      "epoch": 0.659079121765689,
      "grad_norm": 0.714078962802887,
      "learning_rate": 2.75069307733568e-06,
      "loss": 0.686,
      "step": 22874
    },
    {
      "epoch": 0.6591079352273382,
      "grad_norm": 0.7231417894363403,
      "learning_rate": 2.7502763583788407e-06,
      "loss": 0.7055,
      "step": 22875
    },
    {
      "epoch": 0.6591367486889875,
      "grad_norm": 0.7265710234642029,
      "learning_rate": 2.7498596590148464e-06,
      "loss": 0.7124,
      "step": 22876
    },
    {
      "epoch": 0.6591655621506368,
      "grad_norm": 0.699799656867981,
      "learning_rate": 2.7494429792473222e-06,
      "loss": 0.6826,
      "step": 22877
    },
    {
      "epoch": 0.6591943756122861,
      "grad_norm": 0.7044080495834351,
      "learning_rate": 2.749026319079903e-06,
      "loss": 0.6831,
      "step": 22878
    },
    {
      "epoch": 0.6592231890739353,
      "grad_norm": 0.705109715461731,
      "learning_rate": 2.748609678516213e-06,
      "loss": 0.6709,
      "step": 22879
    },
    {
      "epoch": 0.6592520025355846,
      "grad_norm": 0.7411872744560242,
      "learning_rate": 2.7481930575598836e-06,
      "loss": 0.6942,
      "step": 22880
    },
    {
      "epoch": 0.6592808159972339,
      "grad_norm": 0.7185786962509155,
      "learning_rate": 2.7477764562145402e-06,
      "loss": 0.6947,
      "step": 22881
    },
    {
      "epoch": 0.6593096294588832,
      "grad_norm": 0.74315345287323,
      "learning_rate": 2.7473598744838147e-06,
      "loss": 0.7033,
      "step": 22882
    },
    {
      "epoch": 0.6593384429205325,
      "grad_norm": 0.7426128387451172,
      "learning_rate": 2.7469433123713302e-06,
      "loss": 0.7143,
      "step": 22883
    },
    {
      "epoch": 0.6593672563821817,
      "grad_norm": 0.7183735966682434,
      "learning_rate": 2.7465267698807198e-06,
      "loss": 0.7227,
      "step": 22884
    },
    {
      "epoch": 0.659396069843831,
      "grad_norm": 0.6963945031166077,
      "learning_rate": 2.746110247015606e-06,
      "loss": 0.6957,
      "step": 22885
    },
    {
      "epoch": 0.6594248833054803,
      "grad_norm": 0.7001025080680847,
      "learning_rate": 2.745693743779621e-06,
      "loss": 0.6718,
      "step": 22886
    },
    {
      "epoch": 0.6594536967671296,
      "grad_norm": 0.6934295892715454,
      "learning_rate": 2.745277260176388e-06,
      "loss": 0.6813,
      "step": 22887
    },
    {
      "epoch": 0.6594825102287789,
      "grad_norm": 0.7483130097389221,
      "learning_rate": 2.7448607962095363e-06,
      "loss": 0.7091,
      "step": 22888
    },
    {
      "epoch": 0.6595113236904282,
      "grad_norm": 0.7308632731437683,
      "learning_rate": 2.7444443518826934e-06,
      "loss": 0.7014,
      "step": 22889
    },
    {
      "epoch": 0.6595401371520775,
      "grad_norm": 0.7089500427246094,
      "learning_rate": 2.744027927199486e-06,
      "loss": 0.6941,
      "step": 22890
    },
    {
      "epoch": 0.6595689506137268,
      "grad_norm": 0.7302184700965881,
      "learning_rate": 2.7436115221635396e-06,
      "loss": 0.6764,
      "step": 22891
    },
    {
      "epoch": 0.6595977640753761,
      "grad_norm": 0.7240081429481506,
      "learning_rate": 2.743195136778481e-06,
      "loss": 0.6977,
      "step": 22892
    },
    {
      "epoch": 0.6596265775370252,
      "grad_norm": 0.7383553385734558,
      "learning_rate": 2.7427787710479382e-06,
      "loss": 0.7247,
      "step": 22893
    },
    {
      "epoch": 0.6596553909986745,
      "grad_norm": 0.7187661528587341,
      "learning_rate": 2.7423624249755345e-06,
      "loss": 0.683,
      "step": 22894
    },
    {
      "epoch": 0.6596842044603238,
      "grad_norm": 0.7336071133613586,
      "learning_rate": 2.7419460985648987e-06,
      "loss": 0.7042,
      "step": 22895
    },
    {
      "epoch": 0.6597130179219731,
      "grad_norm": 0.6965324282646179,
      "learning_rate": 2.7415297918196536e-06,
      "loss": 0.6867,
      "step": 22896
    },
    {
      "epoch": 0.6597418313836224,
      "grad_norm": 0.7244847416877747,
      "learning_rate": 2.741113504743428e-06,
      "loss": 0.6886,
      "step": 22897
    },
    {
      "epoch": 0.6597706448452717,
      "grad_norm": 0.7003315091133118,
      "learning_rate": 2.7406972373398443e-06,
      "loss": 0.6944,
      "step": 22898
    },
    {
      "epoch": 0.659799458306921,
      "grad_norm": 0.7040152549743652,
      "learning_rate": 2.7402809896125304e-06,
      "loss": 0.6744,
      "step": 22899
    },
    {
      "epoch": 0.6598282717685703,
      "grad_norm": 0.7131291031837463,
      "learning_rate": 2.739864761565109e-06,
      "loss": 0.6952,
      "step": 22900
    },
    {
      "epoch": 0.6598570852302196,
      "grad_norm": 0.7041209936141968,
      "learning_rate": 2.739448553201206e-06,
      "loss": 0.6933,
      "step": 22901
    },
    {
      "epoch": 0.6598858986918689,
      "grad_norm": 0.7302857041358948,
      "learning_rate": 2.739032364524446e-06,
      "loss": 0.7167,
      "step": 22902
    },
    {
      "epoch": 0.6599147121535182,
      "grad_norm": 0.7344827651977539,
      "learning_rate": 2.738616195538456e-06,
      "loss": 0.703,
      "step": 22903
    },
    {
      "epoch": 0.6599435256151674,
      "grad_norm": 0.700796902179718,
      "learning_rate": 2.7382000462468566e-06,
      "loss": 0.6961,
      "step": 22904
    },
    {
      "epoch": 0.6599723390768167,
      "grad_norm": 0.7034772038459778,
      "learning_rate": 2.7377839166532762e-06,
      "loss": 0.7085,
      "step": 22905
    },
    {
      "epoch": 0.6600011525384659,
      "grad_norm": 0.723028302192688,
      "learning_rate": 2.737367806761334e-06,
      "loss": 0.7168,
      "step": 22906
    },
    {
      "epoch": 0.6600299660001152,
      "grad_norm": 0.7082751989364624,
      "learning_rate": 2.7369517165746585e-06,
      "loss": 0.7145,
      "step": 22907
    },
    {
      "epoch": 0.6600587794617645,
      "grad_norm": 0.7216241955757141,
      "learning_rate": 2.736535646096869e-06,
      "loss": 0.6933,
      "step": 22908
    },
    {
      "epoch": 0.6600875929234138,
      "grad_norm": 0.7132461071014404,
      "learning_rate": 2.7361195953315918e-06,
      "loss": 0.6825,
      "step": 22909
    },
    {
      "epoch": 0.6601164063850631,
      "grad_norm": 0.7138887047767639,
      "learning_rate": 2.7357035642824513e-06,
      "loss": 0.7052,
      "step": 22910
    },
    {
      "epoch": 0.6601452198467124,
      "grad_norm": 0.7517318725585938,
      "learning_rate": 2.735287552953068e-06,
      "loss": 0.7058,
      "step": 22911
    },
    {
      "epoch": 0.6601740333083617,
      "grad_norm": 0.7182313799858093,
      "learning_rate": 2.734871561347068e-06,
      "loss": 0.6893,
      "step": 22912
    },
    {
      "epoch": 0.660202846770011,
      "grad_norm": 0.6929961442947388,
      "learning_rate": 2.7344555894680687e-06,
      "loss": 0.692,
      "step": 22913
    },
    {
      "epoch": 0.6602316602316602,
      "grad_norm": 0.7276204228401184,
      "learning_rate": 2.7340396373197e-06,
      "loss": 0.7154,
      "step": 22914
    },
    {
      "epoch": 0.6602604736933095,
      "grad_norm": 0.7286370992660522,
      "learning_rate": 2.733623704905579e-06,
      "loss": 0.7078,
      "step": 22915
    },
    {
      "epoch": 0.6602892871549588,
      "grad_norm": 0.7246015071868896,
      "learning_rate": 2.733207792229332e-06,
      "loss": 0.6964,
      "step": 22916
    },
    {
      "epoch": 0.6603181006166081,
      "grad_norm": 0.6857815384864807,
      "learning_rate": 2.732791899294577e-06,
      "loss": 0.6843,
      "step": 22917
    },
    {
      "epoch": 0.6603469140782574,
      "grad_norm": 0.7225633859634399,
      "learning_rate": 2.732376026104941e-06,
      "loss": 0.6762,
      "step": 22918
    },
    {
      "epoch": 0.6603757275399067,
      "grad_norm": 0.7122883796691895,
      "learning_rate": 2.7319601726640397e-06,
      "loss": 0.6902,
      "step": 22919
    },
    {
      "epoch": 0.6604045410015559,
      "grad_norm": 0.7549934387207031,
      "learning_rate": 2.7315443389755013e-06,
      "loss": 0.6974,
      "step": 22920
    },
    {
      "epoch": 0.6604333544632052,
      "grad_norm": 0.7544528245925903,
      "learning_rate": 2.731128525042942e-06,
      "loss": 0.7263,
      "step": 22921
    },
    {
      "epoch": 0.6604621679248545,
      "grad_norm": 0.7229333519935608,
      "learning_rate": 2.7307127308699865e-06,
      "loss": 0.6875,
      "step": 22922
    },
    {
      "epoch": 0.6604909813865038,
      "grad_norm": 0.7116197943687439,
      "learning_rate": 2.7302969564602535e-06,
      "loss": 0.6816,
      "step": 22923
    },
    {
      "epoch": 0.660519794848153,
      "grad_norm": 0.7159953713417053,
      "learning_rate": 2.729881201817364e-06,
      "loss": 0.6899,
      "step": 22924
    },
    {
      "epoch": 0.6605486083098023,
      "grad_norm": 0.7178232073783875,
      "learning_rate": 2.7294654669449404e-06,
      "loss": 0.7064,
      "step": 22925
    },
    {
      "epoch": 0.6605774217714516,
      "grad_norm": 0.7277717590332031,
      "learning_rate": 2.7290497518466054e-06,
      "loss": 0.6924,
      "step": 22926
    },
    {
      "epoch": 0.6606062352331009,
      "grad_norm": 0.7304313778877258,
      "learning_rate": 2.7286340565259743e-06,
      "loss": 0.713,
      "step": 22927
    },
    {
      "epoch": 0.6606350486947502,
      "grad_norm": 0.7239370346069336,
      "learning_rate": 2.7282183809866724e-06,
      "loss": 0.7116,
      "step": 22928
    },
    {
      "epoch": 0.6606638621563995,
      "grad_norm": 0.7283070683479309,
      "learning_rate": 2.7278027252323157e-06,
      "loss": 0.6964,
      "step": 22929
    },
    {
      "epoch": 0.6606926756180488,
      "grad_norm": 0.7240017652511597,
      "learning_rate": 2.7273870892665257e-06,
      "loss": 0.6908,
      "step": 22930
    },
    {
      "epoch": 0.6607214890796981,
      "grad_norm": 0.7167237997055054,
      "learning_rate": 2.7269714730929242e-06,
      "loss": 0.6939,
      "step": 22931
    },
    {
      "epoch": 0.6607503025413474,
      "grad_norm": 0.7040475010871887,
      "learning_rate": 2.726555876715127e-06,
      "loss": 0.6912,
      "step": 22932
    },
    {
      "epoch": 0.6607791160029965,
      "grad_norm": 0.7972272634506226,
      "learning_rate": 2.726140300136757e-06,
      "loss": 0.7055,
      "step": 22933
    },
    {
      "epoch": 0.6608079294646458,
      "grad_norm": 0.7196207642555237,
      "learning_rate": 2.7257247433614307e-06,
      "loss": 0.7037,
      "step": 22934
    },
    {
      "epoch": 0.6608367429262951,
      "grad_norm": 0.7370268106460571,
      "learning_rate": 2.72530920639277e-06,
      "loss": 0.6639,
      "step": 22935
    },
    {
      "epoch": 0.6608655563879444,
      "grad_norm": 0.7257287502288818,
      "learning_rate": 2.7248936892343908e-06,
      "loss": 0.6909,
      "step": 22936
    },
    {
      "epoch": 0.6608943698495937,
      "grad_norm": 0.7280731201171875,
      "learning_rate": 2.7244781918899133e-06,
      "loss": 0.6914,
      "step": 22937
    },
    {
      "epoch": 0.660923183311243,
      "grad_norm": 0.7317912578582764,
      "learning_rate": 2.724062714362955e-06,
      "loss": 0.69,
      "step": 22938
    },
    {
      "epoch": 0.6609519967728923,
      "grad_norm": 0.7406340837478638,
      "learning_rate": 2.7236472566571383e-06,
      "loss": 0.7004,
      "step": 22939
    },
    {
      "epoch": 0.6609808102345416,
      "grad_norm": 0.7187806963920593,
      "learning_rate": 2.7232318187760766e-06,
      "loss": 0.7207,
      "step": 22940
    },
    {
      "epoch": 0.6610096236961909,
      "grad_norm": 0.7233579754829407,
      "learning_rate": 2.7228164007233913e-06,
      "loss": 0.7094,
      "step": 22941
    },
    {
      "epoch": 0.6610384371578402,
      "grad_norm": 0.7181544899940491,
      "learning_rate": 2.722401002502697e-06,
      "loss": 0.6842,
      "step": 22942
    },
    {
      "epoch": 0.6610672506194895,
      "grad_norm": 0.7516617178916931,
      "learning_rate": 2.721985624117615e-06,
      "loss": 0.716,
      "step": 22943
    },
    {
      "epoch": 0.6610960640811387,
      "grad_norm": 0.7294523119926453,
      "learning_rate": 2.721570265571759e-06,
      "loss": 0.6947,
      "step": 22944
    },
    {
      "epoch": 0.661124877542788,
      "grad_norm": 0.7047020196914673,
      "learning_rate": 2.721154926868751e-06,
      "loss": 0.6939,
      "step": 22945
    },
    {
      "epoch": 0.6611536910044373,
      "grad_norm": 0.7480795979499817,
      "learning_rate": 2.720739608012203e-06,
      "loss": 0.7303,
      "step": 22946
    },
    {
      "epoch": 0.6611825044660865,
      "grad_norm": 0.732564389705658,
      "learning_rate": 2.7203243090057353e-06,
      "loss": 0.6991,
      "step": 22947
    },
    {
      "epoch": 0.6612113179277358,
      "grad_norm": 0.749659538269043,
      "learning_rate": 2.719909029852965e-06,
      "loss": 0.7023,
      "step": 22948
    },
    {
      "epoch": 0.6612401313893851,
      "grad_norm": 0.7092633247375488,
      "learning_rate": 2.7194937705575065e-06,
      "loss": 0.7373,
      "step": 22949
    },
    {
      "epoch": 0.6612689448510344,
      "grad_norm": 0.7062575817108154,
      "learning_rate": 2.7190785311229774e-06,
      "loss": 0.6603,
      "step": 22950
    },
    {
      "epoch": 0.6612977583126837,
      "grad_norm": 0.7176477313041687,
      "learning_rate": 2.718663311552994e-06,
      "loss": 0.6929,
      "step": 22951
    },
    {
      "epoch": 0.661326571774333,
      "grad_norm": 0.7452471852302551,
      "learning_rate": 2.718248111851175e-06,
      "loss": 0.6744,
      "step": 22952
    },
    {
      "epoch": 0.6613553852359823,
      "grad_norm": 0.7325173020362854,
      "learning_rate": 2.7178329320211314e-06,
      "loss": 0.7036,
      "step": 22953
    },
    {
      "epoch": 0.6613841986976315,
      "grad_norm": 0.7474783062934875,
      "learning_rate": 2.717417772066484e-06,
      "loss": 0.689,
      "step": 22954
    },
    {
      "epoch": 0.6614130121592808,
      "grad_norm": 0.7293702960014343,
      "learning_rate": 2.7170026319908442e-06,
      "loss": 0.6854,
      "step": 22955
    },
    {
      "epoch": 0.6614418256209301,
      "grad_norm": 0.79616779088974,
      "learning_rate": 2.7165875117978314e-06,
      "loss": 0.6769,
      "step": 22956
    },
    {
      "epoch": 0.6614706390825794,
      "grad_norm": 0.7436279058456421,
      "learning_rate": 2.716172411491057e-06,
      "loss": 0.7229,
      "step": 22957
    },
    {
      "epoch": 0.6614994525442287,
      "grad_norm": 0.7422657012939453,
      "learning_rate": 2.7157573310741393e-06,
      "loss": 0.7084,
      "step": 22958
    },
    {
      "epoch": 0.661528266005878,
      "grad_norm": 0.7320012450218201,
      "learning_rate": 2.7153422705506903e-06,
      "loss": 0.6831,
      "step": 22959
    },
    {
      "epoch": 0.6615570794675272,
      "grad_norm": 0.7269938588142395,
      "learning_rate": 2.7149272299243277e-06,
      "loss": 0.7009,
      "step": 22960
    },
    {
      "epoch": 0.6615858929291765,
      "grad_norm": 0.6868999600410461,
      "learning_rate": 2.7145122091986616e-06,
      "loss": 0.6598,
      "step": 22961
    },
    {
      "epoch": 0.6616147063908258,
      "grad_norm": 0.7173488736152649,
      "learning_rate": 2.7140972083773125e-06,
      "loss": 0.7016,
      "step": 22962
    },
    {
      "epoch": 0.661643519852475,
      "grad_norm": 0.7339988946914673,
      "learning_rate": 2.7136822274638903e-06,
      "loss": 0.6874,
      "step": 22963
    },
    {
      "epoch": 0.6616723333141243,
      "grad_norm": 0.7350561618804932,
      "learning_rate": 2.713267266462012e-06,
      "loss": 0.6997,
      "step": 22964
    },
    {
      "epoch": 0.6617011467757736,
      "grad_norm": 0.7209306955337524,
      "learning_rate": 2.712852325375288e-06,
      "loss": 0.7019,
      "step": 22965
    },
    {
      "epoch": 0.6617299602374229,
      "grad_norm": 0.7049630284309387,
      "learning_rate": 2.7124374042073365e-06,
      "loss": 0.7216,
      "step": 22966
    },
    {
      "epoch": 0.6617587736990722,
      "grad_norm": 0.7042686343193054,
      "learning_rate": 2.7120225029617662e-06,
      "loss": 0.654,
      "step": 22967
    },
    {
      "epoch": 0.6617875871607215,
      "grad_norm": 0.7039917707443237,
      "learning_rate": 2.711607621642193e-06,
      "loss": 0.6972,
      "step": 22968
    },
    {
      "epoch": 0.6618164006223708,
      "grad_norm": 0.7395591139793396,
      "learning_rate": 2.7111927602522316e-06,
      "loss": 0.7127,
      "step": 22969
    },
    {
      "epoch": 0.6618452140840201,
      "grad_norm": 0.7462002635002136,
      "learning_rate": 2.7107779187954907e-06,
      "loss": 0.7334,
      "step": 22970
    },
    {
      "epoch": 0.6618740275456694,
      "grad_norm": 0.7483238577842712,
      "learning_rate": 2.710363097275588e-06,
      "loss": 0.7057,
      "step": 22971
    },
    {
      "epoch": 0.6619028410073187,
      "grad_norm": 0.7204499840736389,
      "learning_rate": 2.709948295696132e-06,
      "loss": 0.6758,
      "step": 22972
    },
    {
      "epoch": 0.661931654468968,
      "grad_norm": 0.7518163919448853,
      "learning_rate": 2.709533514060738e-06,
      "loss": 0.7063,
      "step": 22973
    },
    {
      "epoch": 0.6619604679306171,
      "grad_norm": 0.7213732600212097,
      "learning_rate": 2.7091187523730167e-06,
      "loss": 0.694,
      "step": 22974
    },
    {
      "epoch": 0.6619892813922664,
      "grad_norm": 0.7113245129585266,
      "learning_rate": 2.708704010636583e-06,
      "loss": 0.6636,
      "step": 22975
    },
    {
      "epoch": 0.6620180948539157,
      "grad_norm": 0.7236950993537903,
      "learning_rate": 2.708289288855045e-06,
      "loss": 0.7088,
      "step": 22976
    },
    {
      "epoch": 0.662046908315565,
      "grad_norm": 0.6990618705749512,
      "learning_rate": 2.7078745870320185e-06,
      "loss": 0.6632,
      "step": 22977
    },
    {
      "epoch": 0.6620757217772143,
      "grad_norm": 0.7124999165534973,
      "learning_rate": 2.707459905171111e-06,
      "loss": 0.7016,
      "step": 22978
    },
    {
      "epoch": 0.6621045352388636,
      "grad_norm": 0.6984218955039978,
      "learning_rate": 2.7070452432759375e-06,
      "loss": 0.6734,
      "step": 22979
    },
    {
      "epoch": 0.6621333487005129,
      "grad_norm": 0.6997480988502502,
      "learning_rate": 2.7066306013501067e-06,
      "loss": 0.6987,
      "step": 22980
    },
    {
      "epoch": 0.6621621621621622,
      "grad_norm": 0.7183563709259033,
      "learning_rate": 2.706215979397232e-06,
      "loss": 0.7203,
      "step": 22981
    },
    {
      "epoch": 0.6621909756238115,
      "grad_norm": 0.7097045183181763,
      "learning_rate": 2.7058013774209222e-06,
      "loss": 0.6851,
      "step": 22982
    },
    {
      "epoch": 0.6622197890854608,
      "grad_norm": 0.7231066823005676,
      "learning_rate": 2.7053867954247904e-06,
      "loss": 0.7104,
      "step": 22983
    },
    {
      "epoch": 0.66224860254711,
      "grad_norm": 0.7083967328071594,
      "learning_rate": 2.704972233412444e-06,
      "loss": 0.6844,
      "step": 22984
    },
    {
      "epoch": 0.6622774160087593,
      "grad_norm": 0.7380768656730652,
      "learning_rate": 2.704557691387495e-06,
      "loss": 0.7169,
      "step": 22985
    },
    {
      "epoch": 0.6623062294704086,
      "grad_norm": 0.7264997959136963,
      "learning_rate": 2.704143169353554e-06,
      "loss": 0.7086,
      "step": 22986
    },
    {
      "epoch": 0.6623350429320578,
      "grad_norm": 0.8570841550827026,
      "learning_rate": 2.703728667314233e-06,
      "loss": 0.6935,
      "step": 22987
    },
    {
      "epoch": 0.6623638563937071,
      "grad_norm": 0.7154909372329712,
      "learning_rate": 2.703314185273138e-06,
      "loss": 0.6945,
      "step": 22988
    },
    {
      "epoch": 0.6623926698553564,
      "grad_norm": 0.7065746188163757,
      "learning_rate": 2.7028997232338806e-06,
      "loss": 0.7093,
      "step": 22989
    },
    {
      "epoch": 0.6624214833170057,
      "grad_norm": 0.7336387038230896,
      "learning_rate": 2.7024852812000724e-06,
      "loss": 0.7023,
      "step": 22990
    },
    {
      "epoch": 0.662450296778655,
      "grad_norm": 0.7250612378120422,
      "learning_rate": 2.7020708591753187e-06,
      "loss": 0.6823,
      "step": 22991
    },
    {
      "epoch": 0.6624791102403043,
      "grad_norm": 0.7218239307403564,
      "learning_rate": 2.701656457163232e-06,
      "loss": 0.6979,
      "step": 22992
    },
    {
      "epoch": 0.6625079237019535,
      "grad_norm": 0.7248391509056091,
      "learning_rate": 2.701242075167419e-06,
      "loss": 0.6602,
      "step": 22993
    },
    {
      "epoch": 0.6625367371636028,
      "grad_norm": 0.7180044054985046,
      "learning_rate": 2.700827713191492e-06,
      "loss": 0.69,
      "step": 22994
    },
    {
      "epoch": 0.6625655506252521,
      "grad_norm": 0.7458279728889465,
      "learning_rate": 2.7004133712390542e-06,
      "loss": 0.7245,
      "step": 22995
    },
    {
      "epoch": 0.6625943640869014,
      "grad_norm": 0.7168967127799988,
      "learning_rate": 2.6999990493137196e-06,
      "loss": 0.6892,
      "step": 22996
    },
    {
      "epoch": 0.6626231775485507,
      "grad_norm": 0.7306206226348877,
      "learning_rate": 2.6995847474190897e-06,
      "loss": 0.6903,
      "step": 22997
    },
    {
      "epoch": 0.6626519910102,
      "grad_norm": 0.7192161679267883,
      "learning_rate": 2.6991704655587815e-06,
      "loss": 0.7184,
      "step": 22998
    },
    {
      "epoch": 0.6626808044718493,
      "grad_norm": 0.7140172719955444,
      "learning_rate": 2.698756203736397e-06,
      "loss": 0.6964,
      "step": 22999
    },
    {
      "epoch": 0.6627096179334986,
      "grad_norm": 0.7098427414894104,
      "learning_rate": 2.698341961955547e-06,
      "loss": 0.6922,
      "step": 23000
    },
    {
      "epoch": 0.6627384313951478,
      "grad_norm": 0.7239034175872803,
      "learning_rate": 2.6979277402198354e-06,
      "loss": 0.6831,
      "step": 23001
    },
    {
      "epoch": 0.662767244856797,
      "grad_norm": 0.7014628052711487,
      "learning_rate": 2.697513538532874e-06,
      "loss": 0.675,
      "step": 23002
    },
    {
      "epoch": 0.6627960583184463,
      "grad_norm": 0.7068554759025574,
      "learning_rate": 2.6970993568982668e-06,
      "loss": 0.7012,
      "step": 23003
    },
    {
      "epoch": 0.6628248717800956,
      "grad_norm": 0.7492464184761047,
      "learning_rate": 2.6966851953196238e-06,
      "loss": 0.7161,
      "step": 23004
    },
    {
      "epoch": 0.6628536852417449,
      "grad_norm": 0.703435480594635,
      "learning_rate": 2.696271053800548e-06,
      "loss": 0.6618,
      "step": 23005
    },
    {
      "epoch": 0.6628824987033942,
      "grad_norm": 0.7369054555892944,
      "learning_rate": 2.6958569323446487e-06,
      "loss": 0.692,
      "step": 23006
    },
    {
      "epoch": 0.6629113121650435,
      "grad_norm": 0.7675592303276062,
      "learning_rate": 2.6954428309555343e-06,
      "loss": 0.704,
      "step": 23007
    },
    {
      "epoch": 0.6629401256266928,
      "grad_norm": 0.701747715473175,
      "learning_rate": 2.6950287496368066e-06,
      "loss": 0.6754,
      "step": 23008
    },
    {
      "epoch": 0.6629689390883421,
      "grad_norm": 0.7322805523872375,
      "learning_rate": 2.6946146883920754e-06,
      "loss": 0.6822,
      "step": 23009
    },
    {
      "epoch": 0.6629977525499914,
      "grad_norm": 0.7183442711830139,
      "learning_rate": 2.6942006472249447e-06,
      "loss": 0.7023,
      "step": 23010
    },
    {
      "epoch": 0.6630265660116407,
      "grad_norm": 0.7291457056999207,
      "learning_rate": 2.693786626139024e-06,
      "loss": 0.6785,
      "step": 23011
    },
    {
      "epoch": 0.66305537947329,
      "grad_norm": 0.7280843257904053,
      "learning_rate": 2.6933726251379144e-06,
      "loss": 0.66,
      "step": 23012
    },
    {
      "epoch": 0.6630841929349393,
      "grad_norm": 0.7047858238220215,
      "learning_rate": 2.692958644225225e-06,
      "loss": 0.7087,
      "step": 23013
    },
    {
      "epoch": 0.6631130063965884,
      "grad_norm": 0.7311607003211975,
      "learning_rate": 2.692544683404558e-06,
      "loss": 0.7093,
      "step": 23014
    },
    {
      "epoch": 0.6631418198582377,
      "grad_norm": 0.7283296585083008,
      "learning_rate": 2.692130742679522e-06,
      "loss": 0.6974,
      "step": 23015
    },
    {
      "epoch": 0.663170633319887,
      "grad_norm": 0.7371519804000854,
      "learning_rate": 2.6917168220537182e-06,
      "loss": 0.6897,
      "step": 23016
    },
    {
      "epoch": 0.6631994467815363,
      "grad_norm": 0.7317221760749817,
      "learning_rate": 2.691302921530755e-06,
      "loss": 0.6949,
      "step": 23017
    },
    {
      "epoch": 0.6632282602431856,
      "grad_norm": 0.7152863144874573,
      "learning_rate": 2.690889041114234e-06,
      "loss": 0.6774,
      "step": 23018
    },
    {
      "epoch": 0.6632570737048349,
      "grad_norm": 0.7105162739753723,
      "learning_rate": 2.6904751808077633e-06,
      "loss": 0.6919,
      "step": 23019
    },
    {
      "epoch": 0.6632858871664842,
      "grad_norm": 0.7132821679115295,
      "learning_rate": 2.690061340614943e-06,
      "loss": 0.6797,
      "step": 23020
    },
    {
      "epoch": 0.6633147006281335,
      "grad_norm": 0.7446514964103699,
      "learning_rate": 2.6896475205393796e-06,
      "loss": 0.6978,
      "step": 23021
    },
    {
      "epoch": 0.6633435140897828,
      "grad_norm": 0.7310215830802917,
      "learning_rate": 2.6892337205846763e-06,
      "loss": 0.7063,
      "step": 23022
    },
    {
      "epoch": 0.663372327551432,
      "grad_norm": 0.7117496132850647,
      "learning_rate": 2.6888199407544394e-06,
      "loss": 0.6947,
      "step": 23023
    },
    {
      "epoch": 0.6634011410130813,
      "grad_norm": 0.7135781049728394,
      "learning_rate": 2.688406181052269e-06,
      "loss": 0.6672,
      "step": 23024
    },
    {
      "epoch": 0.6634299544747306,
      "grad_norm": 0.735160231590271,
      "learning_rate": 2.6879924414817715e-06,
      "loss": 0.6876,
      "step": 23025
    },
    {
      "epoch": 0.6634587679363799,
      "grad_norm": 0.7100611925125122,
      "learning_rate": 2.6875787220465466e-06,
      "loss": 0.6951,
      "step": 23026
    },
    {
      "epoch": 0.6634875813980292,
      "grad_norm": 0.7217330932617188,
      "learning_rate": 2.6871650227502e-06,
      "loss": 0.7167,
      "step": 23027
    },
    {
      "epoch": 0.6635163948596784,
      "grad_norm": 0.6985123157501221,
      "learning_rate": 2.6867513435963354e-06,
      "loss": 0.6792,
      "step": 23028
    },
    {
      "epoch": 0.6635452083213277,
      "grad_norm": 0.7412122488021851,
      "learning_rate": 2.686337684588553e-06,
      "loss": 0.7217,
      "step": 23029
    },
    {
      "epoch": 0.663574021782977,
      "grad_norm": 0.7153928279876709,
      "learning_rate": 2.685924045730458e-06,
      "loss": 0.6958,
      "step": 23030
    },
    {
      "epoch": 0.6636028352446263,
      "grad_norm": 0.7205520272254944,
      "learning_rate": 2.685510427025649e-06,
      "loss": 0.6933,
      "step": 23031
    },
    {
      "epoch": 0.6636316487062756,
      "grad_norm": 0.7085193395614624,
      "learning_rate": 2.6850968284777336e-06,
      "loss": 0.6769,
      "step": 23032
    },
    {
      "epoch": 0.6636604621679248,
      "grad_norm": 0.7246890664100647,
      "learning_rate": 2.6846832500903063e-06,
      "loss": 0.7053,
      "step": 23033
    },
    {
      "epoch": 0.6636892756295741,
      "grad_norm": 0.7373705506324768,
      "learning_rate": 2.684269691866977e-06,
      "loss": 0.7236,
      "step": 23034
    },
    {
      "epoch": 0.6637180890912234,
      "grad_norm": 0.7168190479278564,
      "learning_rate": 2.683856153811342e-06,
      "loss": 0.7171,
      "step": 23035
    },
    {
      "epoch": 0.6637469025528727,
      "grad_norm": 0.7282664179801941,
      "learning_rate": 2.6834426359270066e-06,
      "loss": 0.6803,
      "step": 23036
    },
    {
      "epoch": 0.663775716014522,
      "grad_norm": 0.74726402759552,
      "learning_rate": 2.6830291382175685e-06,
      "loss": 0.712,
      "step": 23037
    },
    {
      "epoch": 0.6638045294761713,
      "grad_norm": 0.7286896109580994,
      "learning_rate": 2.682615660686632e-06,
      "loss": 0.6997,
      "step": 23038
    },
    {
      "epoch": 0.6638333429378206,
      "grad_norm": 0.7302253246307373,
      "learning_rate": 2.6822022033377946e-06,
      "loss": 0.6885,
      "step": 23039
    },
    {
      "epoch": 0.6638621563994699,
      "grad_norm": 0.7156780958175659,
      "learning_rate": 2.681788766174661e-06,
      "loss": 0.6758,
      "step": 23040
    },
    {
      "epoch": 0.6638909698611191,
      "grad_norm": 0.7175264954566956,
      "learning_rate": 2.6813753492008287e-06,
      "loss": 0.6678,
      "step": 23041
    },
    {
      "epoch": 0.6639197833227684,
      "grad_norm": 0.7153940796852112,
      "learning_rate": 2.6809619524199004e-06,
      "loss": 0.6892,
      "step": 23042
    },
    {
      "epoch": 0.6639485967844176,
      "grad_norm": 0.7208943367004395,
      "learning_rate": 2.6805485758354736e-06,
      "loss": 0.6921,
      "step": 23043
    },
    {
      "epoch": 0.6639774102460669,
      "grad_norm": 0.7075685858726501,
      "learning_rate": 2.68013521945115e-06,
      "loss": 0.6897,
      "step": 23044
    },
    {
      "epoch": 0.6640062237077162,
      "grad_norm": 0.722197413444519,
      "learning_rate": 2.6797218832705294e-06,
      "loss": 0.6964,
      "step": 23045
    },
    {
      "epoch": 0.6640350371693655,
      "grad_norm": 0.7028859257698059,
      "learning_rate": 2.6793085672972147e-06,
      "loss": 0.7073,
      "step": 23046
    },
    {
      "epoch": 0.6640638506310148,
      "grad_norm": 0.7241083383560181,
      "learning_rate": 2.6788952715348e-06,
      "loss": 0.7046,
      "step": 23047
    },
    {
      "epoch": 0.6640926640926641,
      "grad_norm": 0.7195364236831665,
      "learning_rate": 2.678481995986887e-06,
      "loss": 0.6835,
      "step": 23048
    },
    {
      "epoch": 0.6641214775543134,
      "grad_norm": 0.7043675184249878,
      "learning_rate": 2.6780687406570773e-06,
      "loss": 0.6527,
      "step": 23049
    },
    {
      "epoch": 0.6641502910159627,
      "grad_norm": 0.7238985896110535,
      "learning_rate": 2.677655505548966e-06,
      "loss": 0.6971,
      "step": 23050
    },
    {
      "epoch": 0.664179104477612,
      "grad_norm": 0.7477269768714905,
      "learning_rate": 2.6772422906661557e-06,
      "loss": 0.6886,
      "step": 23051
    },
    {
      "epoch": 0.6642079179392613,
      "grad_norm": 0.7268052697181702,
      "learning_rate": 2.6768290960122413e-06,
      "loss": 0.7123,
      "step": 23052
    },
    {
      "epoch": 0.6642367314009106,
      "grad_norm": 0.7179288864135742,
      "learning_rate": 2.676415921590825e-06,
      "loss": 0.6921,
      "step": 23053
    },
    {
      "epoch": 0.6642655448625598,
      "grad_norm": 0.7373196482658386,
      "learning_rate": 2.6760027674055008e-06,
      "loss": 0.6832,
      "step": 23054
    },
    {
      "epoch": 0.664294358324209,
      "grad_norm": 0.7172558307647705,
      "learning_rate": 2.675589633459872e-06,
      "loss": 0.6742,
      "step": 23055
    },
    {
      "epoch": 0.6643231717858583,
      "grad_norm": 0.733320415019989,
      "learning_rate": 2.6751765197575307e-06,
      "loss": 0.7001,
      "step": 23056
    },
    {
      "epoch": 0.6643519852475076,
      "grad_norm": 0.7131529450416565,
      "learning_rate": 2.6747634263020794e-06,
      "loss": 0.6821,
      "step": 23057
    },
    {
      "epoch": 0.6643807987091569,
      "grad_norm": 0.7189638614654541,
      "learning_rate": 2.6743503530971136e-06,
      "loss": 0.7252,
      "step": 23058
    },
    {
      "epoch": 0.6644096121708062,
      "grad_norm": 0.7232959866523743,
      "learning_rate": 2.6739373001462336e-06,
      "loss": 0.6797,
      "step": 23059
    },
    {
      "epoch": 0.6644384256324555,
      "grad_norm": 0.6919270753860474,
      "learning_rate": 2.6735242674530325e-06,
      "loss": 0.664,
      "step": 23060
    },
    {
      "epoch": 0.6644672390941048,
      "grad_norm": 0.6982417106628418,
      "learning_rate": 2.673111255021111e-06,
      "loss": 0.6821,
      "step": 23061
    },
    {
      "epoch": 0.6644960525557541,
      "grad_norm": 0.7173218131065369,
      "learning_rate": 2.672698262854063e-06,
      "loss": 0.7049,
      "step": 23062
    },
    {
      "epoch": 0.6645248660174033,
      "grad_norm": 0.7155652046203613,
      "learning_rate": 2.6722852909554888e-06,
      "loss": 0.6929,
      "step": 23063
    },
    {
      "epoch": 0.6645536794790526,
      "grad_norm": 0.7197914123535156,
      "learning_rate": 2.6718723393289803e-06,
      "loss": 0.6617,
      "step": 23064
    },
    {
      "epoch": 0.6645824929407019,
      "grad_norm": 0.7571334838867188,
      "learning_rate": 2.671459407978137e-06,
      "loss": 0.6871,
      "step": 23065
    },
    {
      "epoch": 0.6646113064023512,
      "grad_norm": 0.7404542565345764,
      "learning_rate": 2.6710464969065565e-06,
      "loss": 0.7155,
      "step": 23066
    },
    {
      "epoch": 0.6646401198640005,
      "grad_norm": 0.7461989521980286,
      "learning_rate": 2.670633606117831e-06,
      "loss": 0.6916,
      "step": 23067
    },
    {
      "epoch": 0.6646689333256497,
      "grad_norm": 0.7420341968536377,
      "learning_rate": 2.67022073561556e-06,
      "loss": 0.7146,
      "step": 23068
    },
    {
      "epoch": 0.664697746787299,
      "grad_norm": 0.7277034521102905,
      "learning_rate": 2.6698078854033325e-06,
      "loss": 0.6951,
      "step": 23069
    },
    {
      "epoch": 0.6647265602489483,
      "grad_norm": 0.7344915270805359,
      "learning_rate": 2.6693950554847537e-06,
      "loss": 0.7025,
      "step": 23070
    },
    {
      "epoch": 0.6647553737105976,
      "grad_norm": 0.7177934050559998,
      "learning_rate": 2.668982245863413e-06,
      "loss": 0.6694,
      "step": 23071
    },
    {
      "epoch": 0.6647841871722469,
      "grad_norm": 0.7316264510154724,
      "learning_rate": 2.668569456542908e-06,
      "loss": 0.7083,
      "step": 23072
    },
    {
      "epoch": 0.6648130006338961,
      "grad_norm": 0.7229687571525574,
      "learning_rate": 2.6681566875268306e-06,
      "loss": 0.7146,
      "step": 23073
    },
    {
      "epoch": 0.6648418140955454,
      "grad_norm": 0.7137765288352966,
      "learning_rate": 2.66774393881878e-06,
      "loss": 0.6861,
      "step": 23074
    },
    {
      "epoch": 0.6648706275571947,
      "grad_norm": 0.7412994503974915,
      "learning_rate": 2.6673312104223452e-06,
      "loss": 0.7174,
      "step": 23075
    },
    {
      "epoch": 0.664899441018844,
      "grad_norm": 0.7282285690307617,
      "learning_rate": 2.6669185023411267e-06,
      "loss": 0.7139,
      "step": 23076
    },
    {
      "epoch": 0.6649282544804933,
      "grad_norm": 0.7246278524398804,
      "learning_rate": 2.666505814578713e-06,
      "loss": 0.6947,
      "step": 23077
    },
    {
      "epoch": 0.6649570679421426,
      "grad_norm": 0.7223573923110962,
      "learning_rate": 2.6660931471387038e-06,
      "loss": 0.6868,
      "step": 23078
    },
    {
      "epoch": 0.6649858814037919,
      "grad_norm": 0.7189881801605225,
      "learning_rate": 2.6656805000246878e-06,
      "loss": 0.6917,
      "step": 23079
    },
    {
      "epoch": 0.6650146948654412,
      "grad_norm": 0.7365918755531311,
      "learning_rate": 2.665267873240263e-06,
      "loss": 0.7045,
      "step": 23080
    },
    {
      "epoch": 0.6650435083270905,
      "grad_norm": 0.732033908367157,
      "learning_rate": 2.664855266789018e-06,
      "loss": 0.7209,
      "step": 23081
    },
    {
      "epoch": 0.6650723217887397,
      "grad_norm": 0.6980786323547363,
      "learning_rate": 2.6644426806745526e-06,
      "loss": 0.6662,
      "step": 23082
    },
    {
      "epoch": 0.6651011352503889,
      "grad_norm": 0.7165144681930542,
      "learning_rate": 2.6640301149004545e-06,
      "loss": 0.697,
      "step": 23083
    },
    {
      "epoch": 0.6651299487120382,
      "grad_norm": 0.7300407290458679,
      "learning_rate": 2.6636175694703213e-06,
      "loss": 0.6883,
      "step": 23084
    },
    {
      "epoch": 0.6651587621736875,
      "grad_norm": 0.7088125348091125,
      "learning_rate": 2.6632050443877423e-06,
      "loss": 0.6725,
      "step": 23085
    },
    {
      "epoch": 0.6651875756353368,
      "grad_norm": 0.724163830280304,
      "learning_rate": 2.6627925396563102e-06,
      "loss": 0.6939,
      "step": 23086
    },
    {
      "epoch": 0.6652163890969861,
      "grad_norm": 0.7290457487106323,
      "learning_rate": 2.6623800552796215e-06,
      "loss": 0.6953,
      "step": 23087
    },
    {
      "epoch": 0.6652452025586354,
      "grad_norm": 0.7103719711303711,
      "learning_rate": 2.6619675912612635e-06,
      "loss": 0.6875,
      "step": 23088
    },
    {
      "epoch": 0.6652740160202847,
      "grad_norm": 0.7245286107063293,
      "learning_rate": 2.661555147604833e-06,
      "loss": 0.7036,
      "step": 23089
    },
    {
      "epoch": 0.665302829481934,
      "grad_norm": 0.7225470542907715,
      "learning_rate": 2.6611427243139166e-06,
      "loss": 0.6901,
      "step": 23090
    },
    {
      "epoch": 0.6653316429435833,
      "grad_norm": 0.7116026282310486,
      "learning_rate": 2.6607303213921123e-06,
      "loss": 0.6993,
      "step": 23091
    },
    {
      "epoch": 0.6653604564052326,
      "grad_norm": 0.7014891505241394,
      "learning_rate": 2.6603179388430055e-06,
      "loss": 0.6981,
      "step": 23092
    },
    {
      "epoch": 0.6653892698668818,
      "grad_norm": 0.727014422416687,
      "learning_rate": 2.6599055766701907e-06,
      "loss": 0.6913,
      "step": 23093
    },
    {
      "epoch": 0.6654180833285311,
      "grad_norm": 0.7127260565757751,
      "learning_rate": 2.6594932348772597e-06,
      "loss": 0.6964,
      "step": 23094
    },
    {
      "epoch": 0.6654468967901803,
      "grad_norm": 0.7222851514816284,
      "learning_rate": 2.6590809134678046e-06,
      "loss": 0.6745,
      "step": 23095
    },
    {
      "epoch": 0.6654757102518296,
      "grad_norm": 0.7170175909996033,
      "learning_rate": 2.6586686124454132e-06,
      "loss": 0.6955,
      "step": 23096
    },
    {
      "epoch": 0.6655045237134789,
      "grad_norm": 0.6944438815116882,
      "learning_rate": 2.658256331813679e-06,
      "loss": 0.6763,
      "step": 23097
    },
    {
      "epoch": 0.6655333371751282,
      "grad_norm": 0.7530845403671265,
      "learning_rate": 2.6578440715761896e-06,
      "loss": 0.7203,
      "step": 23098
    },
    {
      "epoch": 0.6655621506367775,
      "grad_norm": 0.7038272619247437,
      "learning_rate": 2.657431831736539e-06,
      "loss": 0.7039,
      "step": 23099
    },
    {
      "epoch": 0.6655909640984268,
      "grad_norm": 0.738524854183197,
      "learning_rate": 2.6570196122983133e-06,
      "loss": 0.6843,
      "step": 23100
    },
    {
      "epoch": 0.6656197775600761,
      "grad_norm": 0.7268379926681519,
      "learning_rate": 2.6566074132651067e-06,
      "loss": 0.6992,
      "step": 23101
    },
    {
      "epoch": 0.6656485910217254,
      "grad_norm": 0.7042130827903748,
      "learning_rate": 2.656195234640505e-06,
      "loss": 0.6851,
      "step": 23102
    },
    {
      "epoch": 0.6656774044833746,
      "grad_norm": 0.6963449716567993,
      "learning_rate": 2.6557830764281003e-06,
      "loss": 0.6811,
      "step": 23103
    },
    {
      "epoch": 0.6657062179450239,
      "grad_norm": 0.7039430737495422,
      "learning_rate": 2.655370938631483e-06,
      "loss": 0.7034,
      "step": 23104
    },
    {
      "epoch": 0.6657350314066732,
      "grad_norm": 0.6788386702537537,
      "learning_rate": 2.6549588212542386e-06,
      "loss": 0.6727,
      "step": 23105
    },
    {
      "epoch": 0.6657638448683225,
      "grad_norm": 0.7133219838142395,
      "learning_rate": 2.65454672429996e-06,
      "loss": 0.7051,
      "step": 23106
    },
    {
      "epoch": 0.6657926583299718,
      "grad_norm": 0.7087177634239197,
      "learning_rate": 2.654134647772234e-06,
      "loss": 0.6928,
      "step": 23107
    },
    {
      "epoch": 0.6658214717916211,
      "grad_norm": 0.7223413586616516,
      "learning_rate": 2.6537225916746524e-06,
      "loss": 0.6987,
      "step": 23108
    },
    {
      "epoch": 0.6658502852532703,
      "grad_norm": 0.7300898432731628,
      "learning_rate": 2.6533105560107995e-06,
      "loss": 0.677,
      "step": 23109
    },
    {
      "epoch": 0.6658790987149196,
      "grad_norm": 0.716342568397522,
      "learning_rate": 2.6528985407842685e-06,
      "loss": 0.6839,
      "step": 23110
    },
    {
      "epoch": 0.6659079121765689,
      "grad_norm": 0.7323417663574219,
      "learning_rate": 2.652486545998642e-06,
      "loss": 0.7102,
      "step": 23111
    },
    {
      "epoch": 0.6659367256382182,
      "grad_norm": 0.713132381439209,
      "learning_rate": 2.6520745716575137e-06,
      "loss": 0.6904,
      "step": 23112
    },
    {
      "epoch": 0.6659655390998674,
      "grad_norm": 0.7169703841209412,
      "learning_rate": 2.6516626177644665e-06,
      "loss": 0.725,
      "step": 23113
    },
    {
      "epoch": 0.6659943525615167,
      "grad_norm": 0.7475017309188843,
      "learning_rate": 2.6512506843230924e-06,
      "loss": 0.7079,
      "step": 23114
    },
    {
      "epoch": 0.666023166023166,
      "grad_norm": 0.7140738368034363,
      "learning_rate": 2.6508387713369754e-06,
      "loss": 0.6941,
      "step": 23115
    },
    {
      "epoch": 0.6660519794848153,
      "grad_norm": 0.7182012796401978,
      "learning_rate": 2.6504268788097054e-06,
      "loss": 0.7046,
      "step": 23116
    },
    {
      "epoch": 0.6660807929464646,
      "grad_norm": 0.6977487206459045,
      "learning_rate": 2.650015006744866e-06,
      "loss": 0.6919,
      "step": 23117
    },
    {
      "epoch": 0.6661096064081139,
      "grad_norm": 0.7093857526779175,
      "learning_rate": 2.6496031551460498e-06,
      "loss": 0.707,
      "step": 23118
    },
    {
      "epoch": 0.6661384198697632,
      "grad_norm": 0.7103422284126282,
      "learning_rate": 2.649191324016839e-06,
      "loss": 0.6764,
      "step": 23119
    },
    {
      "epoch": 0.6661672333314125,
      "grad_norm": 0.756391167640686,
      "learning_rate": 2.648779513360823e-06,
      "loss": 0.7335,
      "step": 23120
    },
    {
      "epoch": 0.6661960467930618,
      "grad_norm": 0.7067237496376038,
      "learning_rate": 2.648367723181586e-06,
      "loss": 0.6886,
      "step": 23121
    },
    {
      "epoch": 0.666224860254711,
      "grad_norm": 0.706544041633606,
      "learning_rate": 2.647955953482717e-06,
      "loss": 0.6887,
      "step": 23122
    },
    {
      "epoch": 0.6662536737163602,
      "grad_norm": 0.7195912003517151,
      "learning_rate": 2.6475442042677983e-06,
      "loss": 0.683,
      "step": 23123
    },
    {
      "epoch": 0.6662824871780095,
      "grad_norm": 0.7193734645843506,
      "learning_rate": 2.6471324755404182e-06,
      "loss": 0.7122,
      "step": 23124
    },
    {
      "epoch": 0.6663113006396588,
      "grad_norm": 0.7224399447441101,
      "learning_rate": 2.646720767304164e-06,
      "loss": 0.7074,
      "step": 23125
    },
    {
      "epoch": 0.6663401141013081,
      "grad_norm": 0.7247359752655029,
      "learning_rate": 2.646309079562618e-06,
      "loss": 0.6988,
      "step": 23126
    },
    {
      "epoch": 0.6663689275629574,
      "grad_norm": 0.7200589179992676,
      "learning_rate": 2.6458974123193686e-06,
      "loss": 0.7019,
      "step": 23127
    },
    {
      "epoch": 0.6663977410246067,
      "grad_norm": 0.710221529006958,
      "learning_rate": 2.6454857655779973e-06,
      "loss": 0.6821,
      "step": 23128
    },
    {
      "epoch": 0.666426554486256,
      "grad_norm": 0.6998841166496277,
      "learning_rate": 2.645074139342092e-06,
      "loss": 0.6669,
      "step": 23129
    },
    {
      "epoch": 0.6664553679479053,
      "grad_norm": 0.701969563961029,
      "learning_rate": 2.6446625336152367e-06,
      "loss": 0.7099,
      "step": 23130
    },
    {
      "epoch": 0.6664841814095546,
      "grad_norm": 0.7179821729660034,
      "learning_rate": 2.644250948401018e-06,
      "loss": 0.711,
      "step": 23131
    },
    {
      "epoch": 0.6665129948712039,
      "grad_norm": 0.7194944024085999,
      "learning_rate": 2.6438393837030173e-06,
      "loss": 0.7114,
      "step": 23132
    },
    {
      "epoch": 0.6665418083328531,
      "grad_norm": 0.7225069999694824,
      "learning_rate": 2.6434278395248215e-06,
      "loss": 0.6775,
      "step": 23133
    },
    {
      "epoch": 0.6665706217945024,
      "grad_norm": 0.751196026802063,
      "learning_rate": 2.6430163158700116e-06,
      "loss": 0.7077,
      "step": 23134
    },
    {
      "epoch": 0.6665994352561517,
      "grad_norm": 0.7030013203620911,
      "learning_rate": 2.6426048127421753e-06,
      "loss": 0.7015,
      "step": 23135
    },
    {
      "epoch": 0.6666282487178009,
      "grad_norm": 0.7164328098297119,
      "learning_rate": 2.642193330144893e-06,
      "loss": 0.7092,
      "step": 23136
    },
    {
      "epoch": 0.6666570621794502,
      "grad_norm": 0.7825586199760437,
      "learning_rate": 2.641781868081752e-06,
      "loss": 0.7018,
      "step": 23137
    },
    {
      "epoch": 0.6666858756410995,
      "grad_norm": 0.7250924110412598,
      "learning_rate": 2.641370426556331e-06,
      "loss": 0.6927,
      "step": 23138
    },
    {
      "epoch": 0.6667146891027488,
      "grad_norm": 0.7329827547073364,
      "learning_rate": 2.640959005572219e-06,
      "loss": 0.6985,
      "step": 23139
    },
    {
      "epoch": 0.6667435025643981,
      "grad_norm": 0.7346090078353882,
      "learning_rate": 2.6405476051329903e-06,
      "loss": 0.6901,
      "step": 23140
    },
    {
      "epoch": 0.6667723160260474,
      "grad_norm": 0.7292937636375427,
      "learning_rate": 2.6401362252422387e-06,
      "loss": 0.6943,
      "step": 23141
    },
    {
      "epoch": 0.6668011294876967,
      "grad_norm": 0.7628280520439148,
      "learning_rate": 2.639724865903539e-06,
      "loss": 0.6848,
      "step": 23142
    },
    {
      "epoch": 0.6668299429493459,
      "grad_norm": 0.7077272534370422,
      "learning_rate": 2.6393135271204785e-06,
      "loss": 0.7085,
      "step": 23143
    },
    {
      "epoch": 0.6668587564109952,
      "grad_norm": 0.7127943634986877,
      "learning_rate": 2.638902208896635e-06,
      "loss": 0.6986,
      "step": 23144
    },
    {
      "epoch": 0.6668875698726445,
      "grad_norm": 0.7049962878227234,
      "learning_rate": 2.638490911235593e-06,
      "loss": 0.6775,
      "step": 23145
    },
    {
      "epoch": 0.6669163833342938,
      "grad_norm": 0.7702306509017944,
      "learning_rate": 2.6380796341409363e-06,
      "loss": 0.677,
      "step": 23146
    },
    {
      "epoch": 0.6669451967959431,
      "grad_norm": 0.7429682612419128,
      "learning_rate": 2.6376683776162437e-06,
      "loss": 0.6968,
      "step": 23147
    },
    {
      "epoch": 0.6669740102575924,
      "grad_norm": 0.7354841232299805,
      "learning_rate": 2.6372571416650996e-06,
      "loss": 0.6924,
      "step": 23148
    },
    {
      "epoch": 0.6670028237192416,
      "grad_norm": 0.7280813455581665,
      "learning_rate": 2.636845926291082e-06,
      "loss": 0.6899,
      "step": 23149
    },
    {
      "epoch": 0.6670316371808909,
      "grad_norm": 0.7063299417495728,
      "learning_rate": 2.6364347314977756e-06,
      "loss": 0.6801,
      "step": 23150
    },
    {
      "epoch": 0.6670604506425402,
      "grad_norm": 0.7436345815658569,
      "learning_rate": 2.6360235572887587e-06,
      "loss": 0.6896,
      "step": 23151
    },
    {
      "epoch": 0.6670892641041894,
      "grad_norm": 0.7154443264007568,
      "learning_rate": 2.6356124036676136e-06,
      "loss": 0.7101,
      "step": 23152
    },
    {
      "epoch": 0.6671180775658387,
      "grad_norm": 0.7210056781768799,
      "learning_rate": 2.6352012706379204e-06,
      "loss": 0.7014,
      "step": 23153
    },
    {
      "epoch": 0.667146891027488,
      "grad_norm": 0.7281453609466553,
      "learning_rate": 2.6347901582032627e-06,
      "loss": 0.6916,
      "step": 23154
    },
    {
      "epoch": 0.6671757044891373,
      "grad_norm": 0.7487739324569702,
      "learning_rate": 2.6343790663672157e-06,
      "loss": 0.7071,
      "step": 23155
    },
    {
      "epoch": 0.6672045179507866,
      "grad_norm": 0.8025892972946167,
      "learning_rate": 2.633967995133365e-06,
      "loss": 0.6961,
      "step": 23156
    },
    {
      "epoch": 0.6672333314124359,
      "grad_norm": 0.7637860774993896,
      "learning_rate": 2.6335569445052854e-06,
      "loss": 0.7167,
      "step": 23157
    },
    {
      "epoch": 0.6672621448740852,
      "grad_norm": 0.7256913781166077,
      "learning_rate": 2.6331459144865613e-06,
      "loss": 0.7188,
      "step": 23158
    },
    {
      "epoch": 0.6672909583357345,
      "grad_norm": 0.7240177392959595,
      "learning_rate": 2.632734905080768e-06,
      "loss": 0.7064,
      "step": 23159
    },
    {
      "epoch": 0.6673197717973838,
      "grad_norm": 0.7274985313415527,
      "learning_rate": 2.6323239162914904e-06,
      "loss": 0.6979,
      "step": 23160
    },
    {
      "epoch": 0.6673485852590331,
      "grad_norm": 0.7134769558906555,
      "learning_rate": 2.6319129481223015e-06,
      "loss": 0.6981,
      "step": 23161
    },
    {
      "epoch": 0.6673773987206824,
      "grad_norm": 0.7595308423042297,
      "learning_rate": 2.6315020005767843e-06,
      "loss": 0.7084,
      "step": 23162
    },
    {
      "epoch": 0.6674062121823315,
      "grad_norm": 0.7316113114356995,
      "learning_rate": 2.631091073658519e-06,
      "loss": 0.6766,
      "step": 23163
    },
    {
      "epoch": 0.6674350256439808,
      "grad_norm": 0.709034264087677,
      "learning_rate": 2.6306801673710803e-06,
      "loss": 0.681,
      "step": 23164
    },
    {
      "epoch": 0.6674638391056301,
      "grad_norm": 0.719948947429657,
      "learning_rate": 2.6302692817180488e-06,
      "loss": 0.7012,
      "step": 23165
    },
    {
      "epoch": 0.6674926525672794,
      "grad_norm": 0.7270098924636841,
      "learning_rate": 2.629858416703005e-06,
      "loss": 0.7135,
      "step": 23166
    },
    {
      "epoch": 0.6675214660289287,
      "grad_norm": 0.7298475503921509,
      "learning_rate": 2.6294475723295233e-06,
      "loss": 0.6896,
      "step": 23167
    },
    {
      "epoch": 0.667550279490578,
      "grad_norm": 0.7360813617706299,
      "learning_rate": 2.629036748601183e-06,
      "loss": 0.7053,
      "step": 23168
    },
    {
      "epoch": 0.6675790929522273,
      "grad_norm": 0.706122100353241,
      "learning_rate": 2.6286259455215645e-06,
      "loss": 0.6868,
      "step": 23169
    },
    {
      "epoch": 0.6676079064138766,
      "grad_norm": 0.7149116396903992,
      "learning_rate": 2.628215163094242e-06,
      "loss": 0.6918,
      "step": 23170
    },
    {
      "epoch": 0.6676367198755259,
      "grad_norm": 0.7047092318534851,
      "learning_rate": 2.6278044013227965e-06,
      "loss": 0.703,
      "step": 23171
    },
    {
      "epoch": 0.6676655333371752,
      "grad_norm": 0.7176594734191895,
      "learning_rate": 2.6273936602108014e-06,
      "loss": 0.7144,
      "step": 23172
    },
    {
      "epoch": 0.6676943467988244,
      "grad_norm": 0.7194720506668091,
      "learning_rate": 2.6269829397618376e-06,
      "loss": 0.7008,
      "step": 23173
    },
    {
      "epoch": 0.6677231602604737,
      "grad_norm": 0.7111002802848816,
      "learning_rate": 2.626572239979478e-06,
      "loss": 0.6928,
      "step": 23174
    },
    {
      "epoch": 0.667751973722123,
      "grad_norm": 0.7047692537307739,
      "learning_rate": 2.626161560867304e-06,
      "loss": 0.6746,
      "step": 23175
    },
    {
      "epoch": 0.6677807871837722,
      "grad_norm": 0.7175498604774475,
      "learning_rate": 2.625750902428886e-06,
      "loss": 0.7115,
      "step": 23176
    },
    {
      "epoch": 0.6678096006454215,
      "grad_norm": 0.713606059551239,
      "learning_rate": 2.6253402646678084e-06,
      "loss": 0.7085,
      "step": 23177
    },
    {
      "epoch": 0.6678384141070708,
      "grad_norm": 0.7158448100090027,
      "learning_rate": 2.6249296475876407e-06,
      "loss": 0.707,
      "step": 23178
    },
    {
      "epoch": 0.6678672275687201,
      "grad_norm": 0.7027333378791809,
      "learning_rate": 2.6245190511919637e-06,
      "loss": 0.6806,
      "step": 23179
    },
    {
      "epoch": 0.6678960410303694,
      "grad_norm": 0.6944326162338257,
      "learning_rate": 2.6241084754843495e-06,
      "loss": 0.6906,
      "step": 23180
    },
    {
      "epoch": 0.6679248544920187,
      "grad_norm": 0.7042502164840698,
      "learning_rate": 2.6236979204683767e-06,
      "loss": 0.7011,
      "step": 23181
    },
    {
      "epoch": 0.667953667953668,
      "grad_norm": 0.7364858388900757,
      "learning_rate": 2.623287386147618e-06,
      "loss": 0.713,
      "step": 23182
    },
    {
      "epoch": 0.6679824814153172,
      "grad_norm": 0.9628608226776123,
      "learning_rate": 2.6228768725256503e-06,
      "loss": 0.6753,
      "step": 23183
    },
    {
      "epoch": 0.6680112948769665,
      "grad_norm": 0.7415916323661804,
      "learning_rate": 2.6224663796060512e-06,
      "loss": 0.7269,
      "step": 23184
    },
    {
      "epoch": 0.6680401083386158,
      "grad_norm": 0.9809389114379883,
      "learning_rate": 2.622055907392391e-06,
      "loss": 0.6887,
      "step": 23185
    },
    {
      "epoch": 0.6680689218002651,
      "grad_norm": 0.7203863859176636,
      "learning_rate": 2.6216454558882486e-06,
      "loss": 0.7142,
      "step": 23186
    },
    {
      "epoch": 0.6680977352619144,
      "grad_norm": 0.7410458326339722,
      "learning_rate": 2.6212350250971952e-06,
      "loss": 0.6982,
      "step": 23187
    },
    {
      "epoch": 0.6681265487235637,
      "grad_norm": 0.7316604256629944,
      "learning_rate": 2.6208246150228064e-06,
      "loss": 0.691,
      "step": 23188
    },
    {
      "epoch": 0.668155362185213,
      "grad_norm": 0.7190736532211304,
      "learning_rate": 2.6204142256686573e-06,
      "loss": 0.7118,
      "step": 23189
    },
    {
      "epoch": 0.6681841756468622,
      "grad_norm": 0.7429199814796448,
      "learning_rate": 2.620003857038323e-06,
      "loss": 0.6803,
      "step": 23190
    },
    {
      "epoch": 0.6682129891085115,
      "grad_norm": 0.7259296774864197,
      "learning_rate": 2.619593509135374e-06,
      "loss": 0.7102,
      "step": 23191
    },
    {
      "epoch": 0.6682418025701607,
      "grad_norm": 0.7227638959884644,
      "learning_rate": 2.6191831819633883e-06,
      "loss": 0.7095,
      "step": 23192
    },
    {
      "epoch": 0.66827061603181,
      "grad_norm": 0.7225174307823181,
      "learning_rate": 2.6187728755259357e-06,
      "loss": 0.6996,
      "step": 23193
    },
    {
      "epoch": 0.6682994294934593,
      "grad_norm": 0.7335081696510315,
      "learning_rate": 2.618362589826592e-06,
      "loss": 0.68,
      "step": 23194
    },
    {
      "epoch": 0.6683282429551086,
      "grad_norm": 0.7150768637657166,
      "learning_rate": 2.6179523248689275e-06,
      "loss": 0.7195,
      "step": 23195
    },
    {
      "epoch": 0.6683570564167579,
      "grad_norm": 0.7128056883811951,
      "learning_rate": 2.61754208065652e-06,
      "loss": 0.6959,
      "step": 23196
    },
    {
      "epoch": 0.6683858698784072,
      "grad_norm": 0.7310868501663208,
      "learning_rate": 2.6171318571929365e-06,
      "loss": 0.6835,
      "step": 23197
    },
    {
      "epoch": 0.6684146833400565,
      "grad_norm": 0.6919721961021423,
      "learning_rate": 2.616721654481755e-06,
      "loss": 0.692,
      "step": 23198
    },
    {
      "epoch": 0.6684434968017058,
      "grad_norm": 0.7167902588844299,
      "learning_rate": 2.616311472526543e-06,
      "loss": 0.7066,
      "step": 23199
    },
    {
      "epoch": 0.6684723102633551,
      "grad_norm": 0.7242329716682434,
      "learning_rate": 2.6159013113308756e-06,
      "loss": 0.7132,
      "step": 23200
    },
    {
      "epoch": 0.6685011237250044,
      "grad_norm": 0.7120326161384583,
      "learning_rate": 2.6154911708983243e-06,
      "loss": 0.6627,
      "step": 23201
    },
    {
      "epoch": 0.6685299371866537,
      "grad_norm": 0.7227637767791748,
      "learning_rate": 2.6150810512324637e-06,
      "loss": 0.6923,
      "step": 23202
    },
    {
      "epoch": 0.6685587506483028,
      "grad_norm": 0.725396454334259,
      "learning_rate": 2.6146709523368606e-06,
      "loss": 0.6982,
      "step": 23203
    },
    {
      "epoch": 0.6685875641099521,
      "grad_norm": 0.7042524218559265,
      "learning_rate": 2.614260874215092e-06,
      "loss": 0.6902,
      "step": 23204
    },
    {
      "epoch": 0.6686163775716014,
      "grad_norm": 0.7480495572090149,
      "learning_rate": 2.613850816870723e-06,
      "loss": 0.7001,
      "step": 23205
    },
    {
      "epoch": 0.6686451910332507,
      "grad_norm": 0.7261307239532471,
      "learning_rate": 2.6134407803073293e-06,
      "loss": 0.7049,
      "step": 23206
    },
    {
      "epoch": 0.6686740044949,
      "grad_norm": 0.7049626111984253,
      "learning_rate": 2.613030764528482e-06,
      "loss": 0.6955,
      "step": 23207
    },
    {
      "epoch": 0.6687028179565493,
      "grad_norm": 0.7165685296058655,
      "learning_rate": 2.6126207695377486e-06,
      "loss": 0.6907,
      "step": 23208
    },
    {
      "epoch": 0.6687316314181986,
      "grad_norm": 0.7409979701042175,
      "learning_rate": 2.6122107953387044e-06,
      "loss": 0.6962,
      "step": 23209
    },
    {
      "epoch": 0.6687604448798479,
      "grad_norm": 0.7250872254371643,
      "learning_rate": 2.611800841934915e-06,
      "loss": 0.6974,
      "step": 23210
    },
    {
      "epoch": 0.6687892583414972,
      "grad_norm": 0.7544768452644348,
      "learning_rate": 2.6113909093299545e-06,
      "loss": 0.6838,
      "step": 23211
    },
    {
      "epoch": 0.6688180718031465,
      "grad_norm": 0.7180935144424438,
      "learning_rate": 2.610980997527388e-06,
      "loss": 0.6972,
      "step": 23212
    },
    {
      "epoch": 0.6688468852647957,
      "grad_norm": 0.7372117638587952,
      "learning_rate": 2.6105711065307924e-06,
      "loss": 0.7408,
      "step": 23213
    },
    {
      "epoch": 0.668875698726445,
      "grad_norm": 0.7294164896011353,
      "learning_rate": 2.6101612363437323e-06,
      "loss": 0.7166,
      "step": 23214
    },
    {
      "epoch": 0.6689045121880943,
      "grad_norm": 0.7299211621284485,
      "learning_rate": 2.6097513869697803e-06,
      "loss": 0.7002,
      "step": 23215
    },
    {
      "epoch": 0.6689333256497436,
      "grad_norm": 0.7026171684265137,
      "learning_rate": 2.6093415584125024e-06,
      "loss": 0.6867,
      "step": 23216
    },
    {
      "epoch": 0.6689621391113928,
      "grad_norm": 0.7173524498939514,
      "learning_rate": 2.6089317506754722e-06,
      "loss": 0.6962,
      "step": 23217
    },
    {
      "epoch": 0.6689909525730421,
      "grad_norm": 0.7152454257011414,
      "learning_rate": 2.6085219637622545e-06,
      "loss": 0.6978,
      "step": 23218
    },
    {
      "epoch": 0.6690197660346914,
      "grad_norm": 0.710821807384491,
      "learning_rate": 2.608112197676421e-06,
      "loss": 0.6852,
      "step": 23219
    },
    {
      "epoch": 0.6690485794963407,
      "grad_norm": 0.7371066212654114,
      "learning_rate": 2.6077024524215383e-06,
      "loss": 0.7238,
      "step": 23220
    },
    {
      "epoch": 0.66907739295799,
      "grad_norm": 0.7241291403770447,
      "learning_rate": 2.607292728001175e-06,
      "loss": 0.699,
      "step": 23221
    },
    {
      "epoch": 0.6691062064196392,
      "grad_norm": 0.7295987010002136,
      "learning_rate": 2.6068830244189025e-06,
      "loss": 0.7081,
      "step": 23222
    },
    {
      "epoch": 0.6691350198812885,
      "grad_norm": 0.7133105993270874,
      "learning_rate": 2.606473341678285e-06,
      "loss": 0.677,
      "step": 23223
    },
    {
      "epoch": 0.6691638333429378,
      "grad_norm": 0.7133324146270752,
      "learning_rate": 2.6060636797828914e-06,
      "loss": 0.6966,
      "step": 23224
    },
    {
      "epoch": 0.6691926468045871,
      "grad_norm": 0.7166175842285156,
      "learning_rate": 2.605654038736293e-06,
      "loss": 0.7075,
      "step": 23225
    },
    {
      "epoch": 0.6692214602662364,
      "grad_norm": 0.7088900208473206,
      "learning_rate": 2.605244418542051e-06,
      "loss": 0.7013,
      "step": 23226
    },
    {
      "epoch": 0.6692502737278857,
      "grad_norm": 0.731758713722229,
      "learning_rate": 2.6048348192037375e-06,
      "loss": 0.702,
      "step": 23227
    },
    {
      "epoch": 0.669279087189535,
      "grad_norm": 0.7202879786491394,
      "learning_rate": 2.6044252407249193e-06,
      "loss": 0.7125,
      "step": 23228
    },
    {
      "epoch": 0.6693079006511843,
      "grad_norm": 0.7306976914405823,
      "learning_rate": 2.6040156831091616e-06,
      "loss": 0.7205,
      "step": 23229
    },
    {
      "epoch": 0.6693367141128335,
      "grad_norm": 0.7281272411346436,
      "learning_rate": 2.603606146360034e-06,
      "loss": 0.6986,
      "step": 23230
    },
    {
      "epoch": 0.6693655275744828,
      "grad_norm": 0.7073789834976196,
      "learning_rate": 2.6031966304810992e-06,
      "loss": 0.6906,
      "step": 23231
    },
    {
      "epoch": 0.669394341036132,
      "grad_norm": 0.7202585935592651,
      "learning_rate": 2.602787135475928e-06,
      "loss": 0.7068,
      "step": 23232
    },
    {
      "epoch": 0.6694231544977813,
      "grad_norm": 0.7135648131370544,
      "learning_rate": 2.6023776613480823e-06,
      "loss": 0.6894,
      "step": 23233
    },
    {
      "epoch": 0.6694519679594306,
      "grad_norm": 0.7167870402336121,
      "learning_rate": 2.601968208101132e-06,
      "loss": 0.6903,
      "step": 23234
    },
    {
      "epoch": 0.6694807814210799,
      "grad_norm": 0.745973527431488,
      "learning_rate": 2.6015587757386397e-06,
      "loss": 0.7097,
      "step": 23235
    },
    {
      "epoch": 0.6695095948827292,
      "grad_norm": 0.7404240369796753,
      "learning_rate": 2.6011493642641734e-06,
      "loss": 0.7086,
      "step": 23236
    },
    {
      "epoch": 0.6695384083443785,
      "grad_norm": 0.717036247253418,
      "learning_rate": 2.600739973681298e-06,
      "loss": 0.6982,
      "step": 23237
    },
    {
      "epoch": 0.6695672218060278,
      "grad_norm": 0.7127031087875366,
      "learning_rate": 2.6003306039935803e-06,
      "loss": 0.6917,
      "step": 23238
    },
    {
      "epoch": 0.6695960352676771,
      "grad_norm": 0.7142426371574402,
      "learning_rate": 2.5999212552045834e-06,
      "loss": 0.7192,
      "step": 23239
    },
    {
      "epoch": 0.6696248487293264,
      "grad_norm": 0.7296170592308044,
      "learning_rate": 2.599511927317874e-06,
      "loss": 0.7126,
      "step": 23240
    },
    {
      "epoch": 0.6696536621909757,
      "grad_norm": 0.7124989032745361,
      "learning_rate": 2.599102620337015e-06,
      "loss": 0.6963,
      "step": 23241
    },
    {
      "epoch": 0.669682475652625,
      "grad_norm": 0.7243794202804565,
      "learning_rate": 2.598693334265572e-06,
      "loss": 0.6886,
      "step": 23242
    },
    {
      "epoch": 0.6697112891142742,
      "grad_norm": 0.7500365376472473,
      "learning_rate": 2.5982840691071114e-06,
      "loss": 0.7106,
      "step": 23243
    },
    {
      "epoch": 0.6697401025759234,
      "grad_norm": 0.6947289109230042,
      "learning_rate": 2.597874824865194e-06,
      "loss": 0.6733,
      "step": 23244
    },
    {
      "epoch": 0.6697689160375727,
      "grad_norm": 0.7189366221427917,
      "learning_rate": 2.5974656015433875e-06,
      "loss": 0.7099,
      "step": 23245
    },
    {
      "epoch": 0.669797729499222,
      "grad_norm": 0.7386396527290344,
      "learning_rate": 2.5970563991452513e-06,
      "loss": 0.7137,
      "step": 23246
    },
    {
      "epoch": 0.6698265429608713,
      "grad_norm": 0.7327574491500854,
      "learning_rate": 2.596647217674355e-06,
      "loss": 0.7275,
      "step": 23247
    },
    {
      "epoch": 0.6698553564225206,
      "grad_norm": 0.7161776423454285,
      "learning_rate": 2.596238057134254e-06,
      "loss": 0.7024,
      "step": 23248
    },
    {
      "epoch": 0.6698841698841699,
      "grad_norm": 0.7078805565834045,
      "learning_rate": 2.595828917528521e-06,
      "loss": 0.6976,
      "step": 23249
    },
    {
      "epoch": 0.6699129833458192,
      "grad_norm": 0.7282742261886597,
      "learning_rate": 2.595419798860713e-06,
      "loss": 0.7021,
      "step": 23250
    },
    {
      "epoch": 0.6699417968074685,
      "grad_norm": 0.7409786581993103,
      "learning_rate": 2.5950107011343965e-06,
      "loss": 0.6894,
      "step": 23251
    },
    {
      "epoch": 0.6699706102691178,
      "grad_norm": 0.7246981263160706,
      "learning_rate": 2.5946016243531315e-06,
      "loss": 0.7122,
      "step": 23252
    },
    {
      "epoch": 0.669999423730767,
      "grad_norm": 0.7114278078079224,
      "learning_rate": 2.594192568520483e-06,
      "loss": 0.7167,
      "step": 23253
    },
    {
      "epoch": 0.6700282371924163,
      "grad_norm": 0.7357845306396484,
      "learning_rate": 2.5937835336400104e-06,
      "loss": 0.6847,
      "step": 23254
    },
    {
      "epoch": 0.6700570506540656,
      "grad_norm": 0.7148006558418274,
      "learning_rate": 2.59337451971528e-06,
      "loss": 0.6821,
      "step": 23255
    },
    {
      "epoch": 0.6700858641157149,
      "grad_norm": 0.7309713959693909,
      "learning_rate": 2.59296552674985e-06,
      "loss": 0.6957,
      "step": 23256
    },
    {
      "epoch": 0.6701146775773641,
      "grad_norm": 0.726606011390686,
      "learning_rate": 2.592556554747285e-06,
      "loss": 0.6812,
      "step": 23257
    },
    {
      "epoch": 0.6701434910390134,
      "grad_norm": 0.7187551259994507,
      "learning_rate": 2.592147603711145e-06,
      "loss": 0.6912,
      "step": 23258
    },
    {
      "epoch": 0.6701723045006627,
      "grad_norm": 0.7289296388626099,
      "learning_rate": 2.5917386736449924e-06,
      "loss": 0.7018,
      "step": 23259
    },
    {
      "epoch": 0.670201117962312,
      "grad_norm": 0.7241860628128052,
      "learning_rate": 2.5913297645523883e-06,
      "loss": 0.7202,
      "step": 23260
    },
    {
      "epoch": 0.6702299314239613,
      "grad_norm": 0.7194662690162659,
      "learning_rate": 2.5909208764368966e-06,
      "loss": 0.6877,
      "step": 23261
    },
    {
      "epoch": 0.6702587448856105,
      "grad_norm": 0.746570885181427,
      "learning_rate": 2.590512009302073e-06,
      "loss": 0.6929,
      "step": 23262
    },
    {
      "epoch": 0.6702875583472598,
      "grad_norm": 0.7223972082138062,
      "learning_rate": 2.590103163151484e-06,
      "loss": 0.699,
      "step": 23263
    },
    {
      "epoch": 0.6703163718089091,
      "grad_norm": 0.7331482768058777,
      "learning_rate": 2.5896943379886844e-06,
      "loss": 0.6819,
      "step": 23264
    },
    {
      "epoch": 0.6703451852705584,
      "grad_norm": 0.7281532883644104,
      "learning_rate": 2.5892855338172384e-06,
      "loss": 0.7073,
      "step": 23265
    },
    {
      "epoch": 0.6703739987322077,
      "grad_norm": 0.746325671672821,
      "learning_rate": 2.5888767506407076e-06,
      "loss": 0.7102,
      "step": 23266
    },
    {
      "epoch": 0.670402812193857,
      "grad_norm": 0.7331082820892334,
      "learning_rate": 2.5884679884626483e-06,
      "loss": 0.6885,
      "step": 23267
    },
    {
      "epoch": 0.6704316256555063,
      "grad_norm": 0.7364136576652527,
      "learning_rate": 2.588059247286624e-06,
      "loss": 0.6919,
      "step": 23268
    },
    {
      "epoch": 0.6704604391171556,
      "grad_norm": 0.722676694393158,
      "learning_rate": 2.58765052711619e-06,
      "loss": 0.6916,
      "step": 23269
    },
    {
      "epoch": 0.6704892525788049,
      "grad_norm": 0.8235422372817993,
      "learning_rate": 2.5872418279549113e-06,
      "loss": 0.6903,
      "step": 23270
    },
    {
      "epoch": 0.670518066040454,
      "grad_norm": 0.7265548706054688,
      "learning_rate": 2.5868331498063413e-06,
      "loss": 0.7055,
      "step": 23271
    },
    {
      "epoch": 0.6705468795021033,
      "grad_norm": 0.7156409621238708,
      "learning_rate": 2.5864244926740432e-06,
      "loss": 0.7249,
      "step": 23272
    },
    {
      "epoch": 0.6705756929637526,
      "grad_norm": 0.7516221404075623,
      "learning_rate": 2.586015856561575e-06,
      "loss": 0.7001,
      "step": 23273
    },
    {
      "epoch": 0.6706045064254019,
      "grad_norm": 0.7165597081184387,
      "learning_rate": 2.5856072414724976e-06,
      "loss": 0.6812,
      "step": 23274
    },
    {
      "epoch": 0.6706333198870512,
      "grad_norm": 0.7215070128440857,
      "learning_rate": 2.585198647410365e-06,
      "loss": 0.7086,
      "step": 23275
    },
    {
      "epoch": 0.6706621333487005,
      "grad_norm": 0.7231096625328064,
      "learning_rate": 2.5847900743787415e-06,
      "loss": 0.713,
      "step": 23276
    },
    {
      "epoch": 0.6706909468103498,
      "grad_norm": 0.7452691793441772,
      "learning_rate": 2.5843815223811787e-06,
      "loss": 0.7144,
      "step": 23277
    },
    {
      "epoch": 0.6707197602719991,
      "grad_norm": 0.7266173958778381,
      "learning_rate": 2.5839729914212406e-06,
      "loss": 0.7147,
      "step": 23278
    },
    {
      "epoch": 0.6707485737336484,
      "grad_norm": 0.7196162939071655,
      "learning_rate": 2.5835644815024815e-06,
      "loss": 0.6705,
      "step": 23279
    },
    {
      "epoch": 0.6707773871952977,
      "grad_norm": 0.7075401544570923,
      "learning_rate": 2.5831559926284588e-06,
      "loss": 0.697,
      "step": 23280
    },
    {
      "epoch": 0.670806200656947,
      "grad_norm": 0.7090088129043579,
      "learning_rate": 2.5827475248027347e-06,
      "loss": 0.7001,
      "step": 23281
    },
    {
      "epoch": 0.6708350141185963,
      "grad_norm": 0.7253144979476929,
      "learning_rate": 2.5823390780288604e-06,
      "loss": 0.724,
      "step": 23282
    },
    {
      "epoch": 0.6708638275802455,
      "grad_norm": 0.7172539234161377,
      "learning_rate": 2.581930652310398e-06,
      "loss": 0.6814,
      "step": 23283
    },
    {
      "epoch": 0.6708926410418947,
      "grad_norm": 0.7504076957702637,
      "learning_rate": 2.5815222476509013e-06,
      "loss": 0.6906,
      "step": 23284
    },
    {
      "epoch": 0.670921454503544,
      "grad_norm": 0.7059284448623657,
      "learning_rate": 2.5811138640539272e-06,
      "loss": 0.7117,
      "step": 23285
    },
    {
      "epoch": 0.6709502679651933,
      "grad_norm": 0.7337180376052856,
      "learning_rate": 2.580705501523034e-06,
      "loss": 0.7093,
      "step": 23286
    },
    {
      "epoch": 0.6709790814268426,
      "grad_norm": 0.736531674861908,
      "learning_rate": 2.5802971600617787e-06,
      "loss": 0.7149,
      "step": 23287
    },
    {
      "epoch": 0.6710078948884919,
      "grad_norm": 0.7272952198982239,
      "learning_rate": 2.5798888396737155e-06,
      "loss": 0.7015,
      "step": 23288
    },
    {
      "epoch": 0.6710367083501412,
      "grad_norm": 0.7134860754013062,
      "learning_rate": 2.5794805403624024e-06,
      "loss": 0.7118,
      "step": 23289
    },
    {
      "epoch": 0.6710655218117905,
      "grad_norm": 0.7449773550033569,
      "learning_rate": 2.5790722621313926e-06,
      "loss": 0.7139,
      "step": 23290
    },
    {
      "epoch": 0.6710943352734398,
      "grad_norm": 0.7206482887268066,
      "learning_rate": 2.578664004984245e-06,
      "loss": 0.693,
      "step": 23291
    },
    {
      "epoch": 0.671123148735089,
      "grad_norm": 0.7450682520866394,
      "learning_rate": 2.5782557689245115e-06,
      "loss": 0.7022,
      "step": 23292
    },
    {
      "epoch": 0.6711519621967383,
      "grad_norm": 0.7447513341903687,
      "learning_rate": 2.5778475539557515e-06,
      "loss": 0.6952,
      "step": 23293
    },
    {
      "epoch": 0.6711807756583876,
      "grad_norm": 0.716517448425293,
      "learning_rate": 2.577439360081516e-06,
      "loss": 0.6998,
      "step": 23294
    },
    {
      "epoch": 0.6712095891200369,
      "grad_norm": 0.7126280069351196,
      "learning_rate": 2.5770311873053645e-06,
      "loss": 0.6833,
      "step": 23295
    },
    {
      "epoch": 0.6712384025816862,
      "grad_norm": 0.7109774351119995,
      "learning_rate": 2.5766230356308446e-06,
      "loss": 0.7168,
      "step": 23296
    },
    {
      "epoch": 0.6712672160433355,
      "grad_norm": 0.7924176454544067,
      "learning_rate": 2.5762149050615198e-06,
      "loss": 0.6702,
      "step": 23297
    },
    {
      "epoch": 0.6712960295049847,
      "grad_norm": 0.7310839295387268,
      "learning_rate": 2.575806795600938e-06,
      "loss": 0.7092,
      "step": 23298
    },
    {
      "epoch": 0.671324842966634,
      "grad_norm": 0.7472106218338013,
      "learning_rate": 2.575398707252658e-06,
      "loss": 0.7223,
      "step": 23299
    },
    {
      "epoch": 0.6713536564282833,
      "grad_norm": 0.6901611089706421,
      "learning_rate": 2.5749906400202295e-06,
      "loss": 0.7002,
      "step": 23300
    },
    {
      "epoch": 0.6713824698899326,
      "grad_norm": 0.7095348834991455,
      "learning_rate": 2.574582593907211e-06,
      "loss": 0.6753,
      "step": 23301
    },
    {
      "epoch": 0.6714112833515818,
      "grad_norm": 0.7313450574874878,
      "learning_rate": 2.5741745689171503e-06,
      "loss": 0.6622,
      "step": 23302
    },
    {
      "epoch": 0.6714400968132311,
      "grad_norm": 0.7236016988754272,
      "learning_rate": 2.5737665650536053e-06,
      "loss": 0.7136,
      "step": 23303
    },
    {
      "epoch": 0.6714689102748804,
      "grad_norm": 0.7457404732704163,
      "learning_rate": 2.5733585823201303e-06,
      "loss": 0.6721,
      "step": 23304
    },
    {
      "epoch": 0.6714977237365297,
      "grad_norm": 0.7162338495254517,
      "learning_rate": 2.572950620720273e-06,
      "loss": 0.7091,
      "step": 23305
    },
    {
      "epoch": 0.671526537198179,
      "grad_norm": 0.7061218023300171,
      "learning_rate": 2.5725426802575926e-06,
      "loss": 0.6975,
      "step": 23306
    },
    {
      "epoch": 0.6715553506598283,
      "grad_norm": 0.7150806188583374,
      "learning_rate": 2.5721347609356366e-06,
      "loss": 0.6851,
      "step": 23307
    },
    {
      "epoch": 0.6715841641214776,
      "grad_norm": 0.7388721108436584,
      "learning_rate": 2.57172686275796e-06,
      "loss": 0.6918,
      "step": 23308
    },
    {
      "epoch": 0.6716129775831269,
      "grad_norm": 0.708504319190979,
      "learning_rate": 2.571318985728115e-06,
      "loss": 0.6905,
      "step": 23309
    },
    {
      "epoch": 0.6716417910447762,
      "grad_norm": 0.7355492115020752,
      "learning_rate": 2.570911129849656e-06,
      "loss": 0.6702,
      "step": 23310
    },
    {
      "epoch": 0.6716706045064254,
      "grad_norm": 0.7051775455474854,
      "learning_rate": 2.5705032951261304e-06,
      "loss": 0.7042,
      "step": 23311
    },
    {
      "epoch": 0.6716994179680746,
      "grad_norm": 0.7319245338439941,
      "learning_rate": 2.5700954815610947e-06,
      "loss": 0.6828,
      "step": 23312
    },
    {
      "epoch": 0.6717282314297239,
      "grad_norm": 0.7090073823928833,
      "learning_rate": 2.569687689158097e-06,
      "loss": 0.6977,
      "step": 23313
    },
    {
      "epoch": 0.6717570448913732,
      "grad_norm": 0.7141073346138,
      "learning_rate": 2.5692799179206905e-06,
      "loss": 0.7009,
      "step": 23314
    },
    {
      "epoch": 0.6717858583530225,
      "grad_norm": 0.7226869463920593,
      "learning_rate": 2.568872167852425e-06,
      "loss": 0.7066,
      "step": 23315
    },
    {
      "epoch": 0.6718146718146718,
      "grad_norm": 0.6933479905128479,
      "learning_rate": 2.5684644389568554e-06,
      "loss": 0.6702,
      "step": 23316
    },
    {
      "epoch": 0.6718434852763211,
      "grad_norm": 0.7207790613174438,
      "learning_rate": 2.568056731237527e-06,
      "loss": 0.6937,
      "step": 23317
    },
    {
      "epoch": 0.6718722987379704,
      "grad_norm": 0.7221535444259644,
      "learning_rate": 2.567649044697994e-06,
      "loss": 0.7001,
      "step": 23318
    },
    {
      "epoch": 0.6719011121996197,
      "grad_norm": 0.7344037294387817,
      "learning_rate": 2.567241379341808e-06,
      "loss": 0.6732,
      "step": 23319
    },
    {
      "epoch": 0.671929925661269,
      "grad_norm": 0.7069208025932312,
      "learning_rate": 2.566833735172516e-06,
      "loss": 0.6979,
      "step": 23320
    },
    {
      "epoch": 0.6719587391229183,
      "grad_norm": 0.7141858339309692,
      "learning_rate": 2.5664261121936694e-06,
      "loss": 0.683,
      "step": 23321
    },
    {
      "epoch": 0.6719875525845675,
      "grad_norm": 0.706868588924408,
      "learning_rate": 2.566018510408821e-06,
      "loss": 0.6725,
      "step": 23322
    },
    {
      "epoch": 0.6720163660462168,
      "grad_norm": 0.7380217909812927,
      "learning_rate": 2.565610929821516e-06,
      "loss": 0.6889,
      "step": 23323
    },
    {
      "epoch": 0.672045179507866,
      "grad_norm": 0.7179675698280334,
      "learning_rate": 2.5652033704353065e-06,
      "loss": 0.6989,
      "step": 23324
    },
    {
      "epoch": 0.6720739929695153,
      "grad_norm": 0.7140435576438904,
      "learning_rate": 2.5647958322537437e-06,
      "loss": 0.7059,
      "step": 23325
    },
    {
      "epoch": 0.6721028064311646,
      "grad_norm": 0.7326292991638184,
      "learning_rate": 2.5643883152803726e-06,
      "loss": 0.6859,
      "step": 23326
    },
    {
      "epoch": 0.6721316198928139,
      "grad_norm": 0.7357228398323059,
      "learning_rate": 2.563980819518747e-06,
      "loss": 0.6909,
      "step": 23327
    },
    {
      "epoch": 0.6721604333544632,
      "grad_norm": 0.7215965986251831,
      "learning_rate": 2.5635733449724114e-06,
      "loss": 0.6929,
      "step": 23328
    },
    {
      "epoch": 0.6721892468161125,
      "grad_norm": 0.708558976650238,
      "learning_rate": 2.563165891644918e-06,
      "loss": 0.6967,
      "step": 23329
    },
    {
      "epoch": 0.6722180602777618,
      "grad_norm": 0.7257212400436401,
      "learning_rate": 2.5627584595398124e-06,
      "loss": 0.7198,
      "step": 23330
    },
    {
      "epoch": 0.672246873739411,
      "grad_norm": 0.7300313115119934,
      "learning_rate": 2.5623510486606456e-06,
      "loss": 0.7217,
      "step": 23331
    },
    {
      "epoch": 0.6722756872010603,
      "grad_norm": 0.7242940068244934,
      "learning_rate": 2.5619436590109607e-06,
      "loss": 0.6927,
      "step": 23332
    },
    {
      "epoch": 0.6723045006627096,
      "grad_norm": 0.7161828279495239,
      "learning_rate": 2.561536290594313e-06,
      "loss": 0.6659,
      "step": 23333
    },
    {
      "epoch": 0.6723333141243589,
      "grad_norm": 0.714023768901825,
      "learning_rate": 2.561128943414245e-06,
      "loss": 0.6913,
      "step": 23334
    },
    {
      "epoch": 0.6723621275860082,
      "grad_norm": 0.7125647068023682,
      "learning_rate": 2.5607216174743082e-06,
      "loss": 0.6984,
      "step": 23335
    },
    {
      "epoch": 0.6723909410476575,
      "grad_norm": 0.7218614220619202,
      "learning_rate": 2.5603143127780456e-06,
      "loss": 0.689,
      "step": 23336
    },
    {
      "epoch": 0.6724197545093068,
      "grad_norm": 0.7285768389701843,
      "learning_rate": 2.5599070293290086e-06,
      "loss": 0.7151,
      "step": 23337
    },
    {
      "epoch": 0.672448567970956,
      "grad_norm": 0.718970537185669,
      "learning_rate": 2.55949976713074e-06,
      "loss": 0.7095,
      "step": 23338
    },
    {
      "epoch": 0.6724773814326053,
      "grad_norm": 0.7293382883071899,
      "learning_rate": 2.5590925261867906e-06,
      "loss": 0.6947,
      "step": 23339
    },
    {
      "epoch": 0.6725061948942546,
      "grad_norm": 0.7240163683891296,
      "learning_rate": 2.5586853065007032e-06,
      "loss": 0.6996,
      "step": 23340
    },
    {
      "epoch": 0.6725350083559039,
      "grad_norm": 0.7135798931121826,
      "learning_rate": 2.5582781080760266e-06,
      "loss": 0.6842,
      "step": 23341
    },
    {
      "epoch": 0.6725638218175531,
      "grad_norm": 0.7146496772766113,
      "learning_rate": 2.5578709309163087e-06,
      "loss": 0.6891,
      "step": 23342
    },
    {
      "epoch": 0.6725926352792024,
      "grad_norm": 0.7436023950576782,
      "learning_rate": 2.5574637750250917e-06,
      "loss": 0.7078,
      "step": 23343
    },
    {
      "epoch": 0.6726214487408517,
      "grad_norm": 0.7026958465576172,
      "learning_rate": 2.557056640405924e-06,
      "loss": 0.6842,
      "step": 23344
    },
    {
      "epoch": 0.672650262202501,
      "grad_norm": 0.7236756086349487,
      "learning_rate": 2.556649527062351e-06,
      "loss": 0.7216,
      "step": 23345
    },
    {
      "epoch": 0.6726790756641503,
      "grad_norm": 0.7356566786766052,
      "learning_rate": 2.5562424349979198e-06,
      "loss": 0.6935,
      "step": 23346
    },
    {
      "epoch": 0.6727078891257996,
      "grad_norm": 0.7275036573410034,
      "learning_rate": 2.5558353642161722e-06,
      "loss": 0.7173,
      "step": 23347
    },
    {
      "epoch": 0.6727367025874489,
      "grad_norm": 0.7264057397842407,
      "learning_rate": 2.555428314720657e-06,
      "loss": 0.6959,
      "step": 23348
    },
    {
      "epoch": 0.6727655160490982,
      "grad_norm": 0.6930751800537109,
      "learning_rate": 2.555021286514916e-06,
      "loss": 0.6912,
      "step": 23349
    },
    {
      "epoch": 0.6727943295107475,
      "grad_norm": 0.7020395994186401,
      "learning_rate": 2.5546142796024974e-06,
      "loss": 0.6951,
      "step": 23350
    },
    {
      "epoch": 0.6728231429723966,
      "grad_norm": 0.7228519320487976,
      "learning_rate": 2.5542072939869417e-06,
      "loss": 0.6979,
      "step": 23351
    },
    {
      "epoch": 0.6728519564340459,
      "grad_norm": 0.7306907176971436,
      "learning_rate": 2.5538003296717973e-06,
      "loss": 0.695,
      "step": 23352
    },
    {
      "epoch": 0.6728807698956952,
      "grad_norm": 0.7280241250991821,
      "learning_rate": 2.553393386660605e-06,
      "loss": 0.6809,
      "step": 23353
    },
    {
      "epoch": 0.6729095833573445,
      "grad_norm": 0.7230938076972961,
      "learning_rate": 2.552986464956913e-06,
      "loss": 0.6764,
      "step": 23354
    },
    {
      "epoch": 0.6729383968189938,
      "grad_norm": 0.7083497643470764,
      "learning_rate": 2.5525795645642602e-06,
      "loss": 0.6957,
      "step": 23355
    },
    {
      "epoch": 0.6729672102806431,
      "grad_norm": 0.724979043006897,
      "learning_rate": 2.5521726854861938e-06,
      "loss": 0.7016,
      "step": 23356
    },
    {
      "epoch": 0.6729960237422924,
      "grad_norm": 0.756855845451355,
      "learning_rate": 2.551765827726256e-06,
      "loss": 0.7006,
      "step": 23357
    },
    {
      "epoch": 0.6730248372039417,
      "grad_norm": 0.7656822800636292,
      "learning_rate": 2.5513589912879923e-06,
      "loss": 0.7085,
      "step": 23358
    },
    {
      "epoch": 0.673053650665591,
      "grad_norm": 0.7410157322883606,
      "learning_rate": 2.550952176174941e-06,
      "loss": 0.6853,
      "step": 23359
    },
    {
      "epoch": 0.6730824641272403,
      "grad_norm": 0.7317848205566406,
      "learning_rate": 2.5505453823906512e-06,
      "loss": 0.7035,
      "step": 23360
    },
    {
      "epoch": 0.6731112775888896,
      "grad_norm": 0.7211939692497253,
      "learning_rate": 2.5501386099386604e-06,
      "loss": 0.6821,
      "step": 23361
    },
    {
      "epoch": 0.6731400910505388,
      "grad_norm": 0.7185123562812805,
      "learning_rate": 2.549731858822514e-06,
      "loss": 0.7045,
      "step": 23362
    },
    {
      "epoch": 0.6731689045121881,
      "grad_norm": 0.7342597246170044,
      "learning_rate": 2.549325129045755e-06,
      "loss": 0.7073,
      "step": 23363
    },
    {
      "epoch": 0.6731977179738374,
      "grad_norm": 0.7252818942070007,
      "learning_rate": 2.548918420611922e-06,
      "loss": 0.6756,
      "step": 23364
    },
    {
      "epoch": 0.6732265314354866,
      "grad_norm": 0.7396842241287231,
      "learning_rate": 2.5485117335245624e-06,
      "loss": 0.6824,
      "step": 23365
    },
    {
      "epoch": 0.6732553448971359,
      "grad_norm": 0.7286996841430664,
      "learning_rate": 2.548105067787212e-06,
      "loss": 0.7028,
      "step": 23366
    },
    {
      "epoch": 0.6732841583587852,
      "grad_norm": 0.7156690359115601,
      "learning_rate": 2.547698423403418e-06,
      "loss": 0.6883,
      "step": 23367
    },
    {
      "epoch": 0.6733129718204345,
      "grad_norm": 0.714967668056488,
      "learning_rate": 2.5472918003767154e-06,
      "loss": 0.6802,
      "step": 23368
    },
    {
      "epoch": 0.6733417852820838,
      "grad_norm": 0.7299681901931763,
      "learning_rate": 2.5468851987106535e-06,
      "loss": 0.7124,
      "step": 23369
    },
    {
      "epoch": 0.6733705987437331,
      "grad_norm": 0.731229841709137,
      "learning_rate": 2.5464786184087665e-06,
      "loss": 0.7049,
      "step": 23370
    },
    {
      "epoch": 0.6733994122053824,
      "grad_norm": 0.7482807636260986,
      "learning_rate": 2.5460720594746015e-06,
      "loss": 0.6947,
      "step": 23371
    },
    {
      "epoch": 0.6734282256670316,
      "grad_norm": 0.7547958493232727,
      "learning_rate": 2.5456655219116923e-06,
      "loss": 0.6889,
      "step": 23372
    },
    {
      "epoch": 0.6734570391286809,
      "grad_norm": 0.7278934121131897,
      "learning_rate": 2.545259005723586e-06,
      "loss": 0.7071,
      "step": 23373
    },
    {
      "epoch": 0.6734858525903302,
      "grad_norm": 0.7228738069534302,
      "learning_rate": 2.544852510913818e-06,
      "loss": 0.7084,
      "step": 23374
    },
    {
      "epoch": 0.6735146660519795,
      "grad_norm": 0.716887354850769,
      "learning_rate": 2.5444460374859315e-06,
      "loss": 0.6735,
      "step": 23375
    },
    {
      "epoch": 0.6735434795136288,
      "grad_norm": 0.7232420444488525,
      "learning_rate": 2.5440395854434637e-06,
      "loss": 0.7054,
      "step": 23376
    },
    {
      "epoch": 0.6735722929752781,
      "grad_norm": 0.7552514672279358,
      "learning_rate": 2.5436331547899585e-06,
      "loss": 0.6899,
      "step": 23377
    },
    {
      "epoch": 0.6736011064369273,
      "grad_norm": 0.726109504699707,
      "learning_rate": 2.5432267455289505e-06,
      "loss": 0.6856,
      "step": 23378
    },
    {
      "epoch": 0.6736299198985766,
      "grad_norm": 0.7100540995597839,
      "learning_rate": 2.542820357663982e-06,
      "loss": 0.6797,
      "step": 23379
    },
    {
      "epoch": 0.6736587333602259,
      "grad_norm": 0.7162044644355774,
      "learning_rate": 2.542413991198591e-06,
      "loss": 0.7234,
      "step": 23380
    },
    {
      "epoch": 0.6736875468218751,
      "grad_norm": 0.7237941026687622,
      "learning_rate": 2.5420076461363206e-06,
      "loss": 0.6666,
      "step": 23381
    },
    {
      "epoch": 0.6737163602835244,
      "grad_norm": 1.8203513622283936,
      "learning_rate": 2.5416013224807046e-06,
      "loss": 0.6968,
      "step": 23382
    },
    {
      "epoch": 0.6737451737451737,
      "grad_norm": 0.735607922077179,
      "learning_rate": 2.541195020235283e-06,
      "loss": 0.7125,
      "step": 23383
    },
    {
      "epoch": 0.673773987206823,
      "grad_norm": 0.7244418263435364,
      "learning_rate": 2.540788739403597e-06,
      "loss": 0.6752,
      "step": 23384
    },
    {
      "epoch": 0.6738028006684723,
      "grad_norm": 0.7203262448310852,
      "learning_rate": 2.5403824799891807e-06,
      "loss": 0.6997,
      "step": 23385
    },
    {
      "epoch": 0.6738316141301216,
      "grad_norm": 0.7409917116165161,
      "learning_rate": 2.5399762419955766e-06,
      "loss": 0.6976,
      "step": 23386
    },
    {
      "epoch": 0.6738604275917709,
      "grad_norm": 0.7201744318008423,
      "learning_rate": 2.539570025426318e-06,
      "loss": 0.711,
      "step": 23387
    },
    {
      "epoch": 0.6738892410534202,
      "grad_norm": 0.7187026739120483,
      "learning_rate": 2.5391638302849466e-06,
      "loss": 0.7002,
      "step": 23388
    },
    {
      "epoch": 0.6739180545150695,
      "grad_norm": 0.7268617749214172,
      "learning_rate": 2.538757656574997e-06,
      "loss": 0.702,
      "step": 23389
    },
    {
      "epoch": 0.6739468679767188,
      "grad_norm": 0.769397497177124,
      "learning_rate": 2.5383515043000097e-06,
      "loss": 0.6823,
      "step": 23390
    },
    {
      "epoch": 0.6739756814383681,
      "grad_norm": 0.7492691278457642,
      "learning_rate": 2.5379453734635172e-06,
      "loss": 0.7156,
      "step": 23391
    },
    {
      "epoch": 0.6740044949000172,
      "grad_norm": 0.7049899697303772,
      "learning_rate": 2.5375392640690607e-06,
      "loss": 0.6821,
      "step": 23392
    },
    {
      "epoch": 0.6740333083616665,
      "grad_norm": 0.7292248606681824,
      "learning_rate": 2.5371331761201744e-06,
      "loss": 0.6916,
      "step": 23393
    },
    {
      "epoch": 0.6740621218233158,
      "grad_norm": 0.7055718302726746,
      "learning_rate": 2.5367271096203993e-06,
      "loss": 0.6708,
      "step": 23394
    },
    {
      "epoch": 0.6740909352849651,
      "grad_norm": 0.7314881086349487,
      "learning_rate": 2.536321064573266e-06,
      "loss": 0.6862,
      "step": 23395
    },
    {
      "epoch": 0.6741197487466144,
      "grad_norm": 0.699143648147583,
      "learning_rate": 2.5359150409823156e-06,
      "loss": 0.7014,
      "step": 23396
    },
    {
      "epoch": 0.6741485622082637,
      "grad_norm": 0.7471304535865784,
      "learning_rate": 2.5355090388510806e-06,
      "loss": 0.6826,
      "step": 23397
    },
    {
      "epoch": 0.674177375669913,
      "grad_norm": 0.7160915732383728,
      "learning_rate": 2.5351030581830993e-06,
      "loss": 0.7101,
      "step": 23398
    },
    {
      "epoch": 0.6742061891315623,
      "grad_norm": 0.7396613955497742,
      "learning_rate": 2.5346970989819055e-06,
      "loss": 0.6902,
      "step": 23399
    },
    {
      "epoch": 0.6742350025932116,
      "grad_norm": 0.7318152189254761,
      "learning_rate": 2.5342911612510348e-06,
      "loss": 0.6937,
      "step": 23400
    },
    {
      "epoch": 0.6742638160548609,
      "grad_norm": 0.7051456570625305,
      "learning_rate": 2.5338852449940248e-06,
      "loss": 0.7036,
      "step": 23401
    },
    {
      "epoch": 0.6742926295165101,
      "grad_norm": 0.7324182391166687,
      "learning_rate": 2.533479350214408e-06,
      "loss": 0.705,
      "step": 23402
    },
    {
      "epoch": 0.6743214429781594,
      "grad_norm": 0.7130029797554016,
      "learning_rate": 2.5330734769157196e-06,
      "loss": 0.6765,
      "step": 23403
    },
    {
      "epoch": 0.6743502564398087,
      "grad_norm": 0.7255778312683105,
      "learning_rate": 2.532667625101496e-06,
      "loss": 0.7126,
      "step": 23404
    },
    {
      "epoch": 0.6743790699014579,
      "grad_norm": 0.7366347908973694,
      "learning_rate": 2.5322617947752727e-06,
      "loss": 0.7062,
      "step": 23405
    },
    {
      "epoch": 0.6744078833631072,
      "grad_norm": 0.7387695908546448,
      "learning_rate": 2.53185598594058e-06,
      "loss": 0.6895,
      "step": 23406
    },
    {
      "epoch": 0.6744366968247565,
      "grad_norm": 0.7270444631576538,
      "learning_rate": 2.5314501986009565e-06,
      "loss": 0.7243,
      "step": 23407
    },
    {
      "epoch": 0.6744655102864058,
      "grad_norm": 0.7455862164497375,
      "learning_rate": 2.5310444327599328e-06,
      "loss": 0.7157,
      "step": 23408
    },
    {
      "epoch": 0.6744943237480551,
      "grad_norm": 0.7040755748748779,
      "learning_rate": 2.5306386884210454e-06,
      "loss": 0.6952,
      "step": 23409
    },
    {
      "epoch": 0.6745231372097044,
      "grad_norm": 0.7034858465194702,
      "learning_rate": 2.5302329655878243e-06,
      "loss": 0.6956,
      "step": 23410
    },
    {
      "epoch": 0.6745519506713537,
      "grad_norm": 0.7221071124076843,
      "learning_rate": 2.5298272642638077e-06,
      "loss": 0.7003,
      "step": 23411
    },
    {
      "epoch": 0.6745807641330029,
      "grad_norm": 0.755078136920929,
      "learning_rate": 2.529421584452525e-06,
      "loss": 0.703,
      "step": 23412
    },
    {
      "epoch": 0.6746095775946522,
      "grad_norm": 0.7195826768875122,
      "learning_rate": 2.5290159261575115e-06,
      "loss": 0.6934,
      "step": 23413
    },
    {
      "epoch": 0.6746383910563015,
      "grad_norm": 0.7240790724754333,
      "learning_rate": 2.528610289382298e-06,
      "loss": 0.7151,
      "step": 23414
    },
    {
      "epoch": 0.6746672045179508,
      "grad_norm": 0.7232722640037537,
      "learning_rate": 2.5282046741304177e-06,
      "loss": 0.6945,
      "step": 23415
    },
    {
      "epoch": 0.6746960179796001,
      "grad_norm": 0.7319983839988708,
      "learning_rate": 2.527799080405404e-06,
      "loss": 0.7064,
      "step": 23416
    },
    {
      "epoch": 0.6747248314412494,
      "grad_norm": 0.7180188894271851,
      "learning_rate": 2.5273935082107912e-06,
      "loss": 0.7061,
      "step": 23417
    },
    {
      "epoch": 0.6747536449028987,
      "grad_norm": 0.7215464115142822,
      "learning_rate": 2.526987957550107e-06,
      "loss": 0.6665,
      "step": 23418
    },
    {
      "epoch": 0.6747824583645479,
      "grad_norm": 0.7085050940513611,
      "learning_rate": 2.526582428426888e-06,
      "loss": 0.6964,
      "step": 23419
    },
    {
      "epoch": 0.6748112718261972,
      "grad_norm": 0.733849823474884,
      "learning_rate": 2.5261769208446614e-06,
      "loss": 0.6893,
      "step": 23420
    },
    {
      "epoch": 0.6748400852878464,
      "grad_norm": 0.7158271670341492,
      "learning_rate": 2.5257714348069606e-06,
      "loss": 0.6943,
      "step": 23421
    },
    {
      "epoch": 0.6748688987494957,
      "grad_norm": 0.7368897199630737,
      "learning_rate": 2.5253659703173193e-06,
      "loss": 0.7095,
      "step": 23422
    },
    {
      "epoch": 0.674897712211145,
      "grad_norm": 0.7147136330604553,
      "learning_rate": 2.5249605273792654e-06,
      "loss": 0.6983,
      "step": 23423
    },
    {
      "epoch": 0.6749265256727943,
      "grad_norm": 0.7133565545082092,
      "learning_rate": 2.5245551059963323e-06,
      "loss": 0.6875,
      "step": 23424
    },
    {
      "epoch": 0.6749553391344436,
      "grad_norm": 0.7033573389053345,
      "learning_rate": 2.524149706172048e-06,
      "loss": 0.6756,
      "step": 23425
    },
    {
      "epoch": 0.6749841525960929,
      "grad_norm": 0.7243462800979614,
      "learning_rate": 2.523744327909947e-06,
      "loss": 0.6911,
      "step": 23426
    },
    {
      "epoch": 0.6750129660577422,
      "grad_norm": 0.716707706451416,
      "learning_rate": 2.5233389712135536e-06,
      "loss": 0.6916,
      "step": 23427
    },
    {
      "epoch": 0.6750417795193915,
      "grad_norm": 0.7197026014328003,
      "learning_rate": 2.5229336360864055e-06,
      "loss": 0.7426,
      "step": 23428
    },
    {
      "epoch": 0.6750705929810408,
      "grad_norm": 0.7302544713020325,
      "learning_rate": 2.5225283225320273e-06,
      "loss": 0.7233,
      "step": 23429
    },
    {
      "epoch": 0.6750994064426901,
      "grad_norm": 0.7147642970085144,
      "learning_rate": 2.5221230305539535e-06,
      "loss": 0.6767,
      "step": 23430
    },
    {
      "epoch": 0.6751282199043394,
      "grad_norm": 0.7393271327018738,
      "learning_rate": 2.5217177601557088e-06,
      "loss": 0.6899,
      "step": 23431
    },
    {
      "epoch": 0.6751570333659885,
      "grad_norm": 0.7135877013206482,
      "learning_rate": 2.5213125113408266e-06,
      "loss": 0.684,
      "step": 23432
    },
    {
      "epoch": 0.6751858468276378,
      "grad_norm": 0.7264853119850159,
      "learning_rate": 2.520907284112833e-06,
      "loss": 0.6975,
      "step": 23433
    },
    {
      "epoch": 0.6752146602892871,
      "grad_norm": 0.7073830962181091,
      "learning_rate": 2.520502078475261e-06,
      "loss": 0.697,
      "step": 23434
    },
    {
      "epoch": 0.6752434737509364,
      "grad_norm": 0.7357376217842102,
      "learning_rate": 2.520096894431635e-06,
      "loss": 0.709,
      "step": 23435
    },
    {
      "epoch": 0.6752722872125857,
      "grad_norm": 0.713974118232727,
      "learning_rate": 2.519691731985488e-06,
      "loss": 0.678,
      "step": 23436
    },
    {
      "epoch": 0.675301100674235,
      "grad_norm": 0.7207241654396057,
      "learning_rate": 2.519286591140345e-06,
      "loss": 0.6978,
      "step": 23437
    },
    {
      "epoch": 0.6753299141358843,
      "grad_norm": 0.730689287185669,
      "learning_rate": 2.5188814718997357e-06,
      "loss": 0.7075,
      "step": 23438
    },
    {
      "epoch": 0.6753587275975336,
      "grad_norm": 0.7178241014480591,
      "learning_rate": 2.5184763742671882e-06,
      "loss": 0.6893,
      "step": 23439
    },
    {
      "epoch": 0.6753875410591829,
      "grad_norm": 0.7229242324829102,
      "learning_rate": 2.5180712982462337e-06,
      "loss": 0.7009,
      "step": 23440
    },
    {
      "epoch": 0.6754163545208322,
      "grad_norm": 0.7666977643966675,
      "learning_rate": 2.5176662438403944e-06,
      "loss": 0.7319,
      "step": 23441
    },
    {
      "epoch": 0.6754451679824814,
      "grad_norm": 0.7172539234161377,
      "learning_rate": 2.5172612110532013e-06,
      "loss": 0.6979,
      "step": 23442
    },
    {
      "epoch": 0.6754739814441307,
      "grad_norm": 0.7146351933479309,
      "learning_rate": 2.5168561998881823e-06,
      "loss": 0.7038,
      "step": 23443
    },
    {
      "epoch": 0.67550279490578,
      "grad_norm": 0.7185565233230591,
      "learning_rate": 2.5164512103488626e-06,
      "loss": 0.6895,
      "step": 23444
    },
    {
      "epoch": 0.6755316083674293,
      "grad_norm": 0.7092505097389221,
      "learning_rate": 2.5160462424387723e-06,
      "loss": 0.7056,
      "step": 23445
    },
    {
      "epoch": 0.6755604218290785,
      "grad_norm": 0.7659671902656555,
      "learning_rate": 2.515641296161433e-06,
      "loss": 0.7137,
      "step": 23446
    },
    {
      "epoch": 0.6755892352907278,
      "grad_norm": 0.7612408399581909,
      "learning_rate": 2.5152363715203776e-06,
      "loss": 0.6951,
      "step": 23447
    },
    {
      "epoch": 0.6756180487523771,
      "grad_norm": 0.7164121270179749,
      "learning_rate": 2.5148314685191273e-06,
      "loss": 0.6885,
      "step": 23448
    },
    {
      "epoch": 0.6756468622140264,
      "grad_norm": 0.745380163192749,
      "learning_rate": 2.514426587161213e-06,
      "loss": 0.6672,
      "step": 23449
    },
    {
      "epoch": 0.6756756756756757,
      "grad_norm": 0.7048839926719666,
      "learning_rate": 2.5140217274501554e-06,
      "loss": 0.6952,
      "step": 23450
    },
    {
      "epoch": 0.675704489137325,
      "grad_norm": 0.7468568086624146,
      "learning_rate": 2.513616889389484e-06,
      "loss": 0.7019,
      "step": 23451
    },
    {
      "epoch": 0.6757333025989742,
      "grad_norm": 0.7232785224914551,
      "learning_rate": 2.5132120729827248e-06,
      "loss": 0.6882,
      "step": 23452
    },
    {
      "epoch": 0.6757621160606235,
      "grad_norm": 0.7131996154785156,
      "learning_rate": 2.5128072782334036e-06,
      "loss": 0.6993,
      "step": 23453
    },
    {
      "epoch": 0.6757909295222728,
      "grad_norm": 0.7408028841018677,
      "learning_rate": 2.512402505145043e-06,
      "loss": 0.6873,
      "step": 23454
    },
    {
      "epoch": 0.6758197429839221,
      "grad_norm": 0.7027917504310608,
      "learning_rate": 2.5119977537211715e-06,
      "loss": 0.7025,
      "step": 23455
    },
    {
      "epoch": 0.6758485564455714,
      "grad_norm": 0.7343908548355103,
      "learning_rate": 2.5115930239653114e-06,
      "loss": 0.7013,
      "step": 23456
    },
    {
      "epoch": 0.6758773699072207,
      "grad_norm": 0.7220861911773682,
      "learning_rate": 2.5111883158809903e-06,
      "loss": 0.693,
      "step": 23457
    },
    {
      "epoch": 0.67590618336887,
      "grad_norm": 0.7601746916770935,
      "learning_rate": 2.510783629471728e-06,
      "loss": 0.7023,
      "step": 23458
    },
    {
      "epoch": 0.6759349968305192,
      "grad_norm": 0.7041060328483582,
      "learning_rate": 2.510378964741053e-06,
      "loss": 0.693,
      "step": 23459
    },
    {
      "epoch": 0.6759638102921685,
      "grad_norm": 0.7170623540878296,
      "learning_rate": 2.5099743216924905e-06,
      "loss": 0.7172,
      "step": 23460
    },
    {
      "epoch": 0.6759926237538177,
      "grad_norm": 0.7211065292358398,
      "learning_rate": 2.5095697003295605e-06,
      "loss": 0.7185,
      "step": 23461
    },
    {
      "epoch": 0.676021437215467,
      "grad_norm": 0.7166973948478699,
      "learning_rate": 2.5091651006557904e-06,
      "loss": 0.7181,
      "step": 23462
    },
    {
      "epoch": 0.6760502506771163,
      "grad_norm": 0.7467391490936279,
      "learning_rate": 2.508760522674699e-06,
      "loss": 0.6885,
      "step": 23463
    },
    {
      "epoch": 0.6760790641387656,
      "grad_norm": 0.7282037138938904,
      "learning_rate": 2.508355966389817e-06,
      "loss": 0.6828,
      "step": 23464
    },
    {
      "epoch": 0.6761078776004149,
      "grad_norm": 0.7110658288002014,
      "learning_rate": 2.5079514318046615e-06,
      "loss": 0.6993,
      "step": 23465
    },
    {
      "epoch": 0.6761366910620642,
      "grad_norm": 0.7894166707992554,
      "learning_rate": 2.5075469189227597e-06,
      "loss": 0.6722,
      "step": 23466
    },
    {
      "epoch": 0.6761655045237135,
      "grad_norm": 0.7104441523551941,
      "learning_rate": 2.5071424277476315e-06,
      "loss": 0.6599,
      "step": 23467
    },
    {
      "epoch": 0.6761943179853628,
      "grad_norm": 0.6977922320365906,
      "learning_rate": 2.506737958282802e-06,
      "loss": 0.6818,
      "step": 23468
    },
    {
      "epoch": 0.6762231314470121,
      "grad_norm": 0.7311710119247437,
      "learning_rate": 2.506333510531791e-06,
      "loss": 0.6835,
      "step": 23469
    },
    {
      "epoch": 0.6762519449086614,
      "grad_norm": 0.7959649562835693,
      "learning_rate": 2.5059290844981245e-06,
      "loss": 0.6981,
      "step": 23470
    },
    {
      "epoch": 0.6762807583703107,
      "grad_norm": 0.7492028474807739,
      "learning_rate": 2.50552468018532e-06,
      "loss": 0.6996,
      "step": 23471
    },
    {
      "epoch": 0.6763095718319599,
      "grad_norm": 0.7268466353416443,
      "learning_rate": 2.5051202975969043e-06,
      "loss": 0.6941,
      "step": 23472
    },
    {
      "epoch": 0.6763383852936091,
      "grad_norm": 0.7172577977180481,
      "learning_rate": 2.504715936736395e-06,
      "loss": 0.6802,
      "step": 23473
    },
    {
      "epoch": 0.6763671987552584,
      "grad_norm": 0.7163684964179993,
      "learning_rate": 2.5043115976073166e-06,
      "loss": 0.6924,
      "step": 23474
    },
    {
      "epoch": 0.6763960122169077,
      "grad_norm": 0.7461585998535156,
      "learning_rate": 2.5039072802131863e-06,
      "loss": 0.6851,
      "step": 23475
    },
    {
      "epoch": 0.676424825678557,
      "grad_norm": 0.7425785064697266,
      "learning_rate": 2.503502984557532e-06,
      "loss": 0.6865,
      "step": 23476
    },
    {
      "epoch": 0.6764536391402063,
      "grad_norm": 0.7430095076560974,
      "learning_rate": 2.50309871064387e-06,
      "loss": 0.6771,
      "step": 23477
    },
    {
      "epoch": 0.6764824526018556,
      "grad_norm": 0.7451969385147095,
      "learning_rate": 2.5026944584757227e-06,
      "loss": 0.6983,
      "step": 23478
    },
    {
      "epoch": 0.6765112660635049,
      "grad_norm": 0.7308139204978943,
      "learning_rate": 2.5022902280566096e-06,
      "loss": 0.7089,
      "step": 23479
    },
    {
      "epoch": 0.6765400795251542,
      "grad_norm": 0.7625160217285156,
      "learning_rate": 2.5018860193900514e-06,
      "loss": 0.6977,
      "step": 23480
    },
    {
      "epoch": 0.6765688929868034,
      "grad_norm": 0.7185850143432617,
      "learning_rate": 2.501481832479571e-06,
      "loss": 0.6888,
      "step": 23481
    },
    {
      "epoch": 0.6765977064484527,
      "grad_norm": 0.7457637190818787,
      "learning_rate": 2.5010776673286836e-06,
      "loss": 0.6653,
      "step": 23482
    },
    {
      "epoch": 0.676626519910102,
      "grad_norm": 0.7196564078330994,
      "learning_rate": 2.500673523940914e-06,
      "loss": 0.6989,
      "step": 23483
    },
    {
      "epoch": 0.6766553333717513,
      "grad_norm": 0.7385735511779785,
      "learning_rate": 2.5002694023197776e-06,
      "loss": 0.7072,
      "step": 23484
    },
    {
      "epoch": 0.6766841468334006,
      "grad_norm": 0.7122330069541931,
      "learning_rate": 2.499865302468798e-06,
      "loss": 0.6489,
      "step": 23485
    },
    {
      "epoch": 0.6767129602950498,
      "grad_norm": 0.749677836894989,
      "learning_rate": 2.49946122439149e-06,
      "loss": 0.7081,
      "step": 23486
    },
    {
      "epoch": 0.6767417737566991,
      "grad_norm": 0.756994903087616,
      "learning_rate": 2.4990571680913754e-06,
      "loss": 0.7064,
      "step": 23487
    },
    {
      "epoch": 0.6767705872183484,
      "grad_norm": 0.708730936050415,
      "learning_rate": 2.4986531335719734e-06,
      "loss": 0.6766,
      "step": 23488
    },
    {
      "epoch": 0.6767994006799977,
      "grad_norm": 0.7203859090805054,
      "learning_rate": 2.4982491208368037e-06,
      "loss": 0.6936,
      "step": 23489
    },
    {
      "epoch": 0.676828214141647,
      "grad_norm": 0.7323285937309265,
      "learning_rate": 2.4978451298893814e-06,
      "loss": 0.6942,
      "step": 23490
    },
    {
      "epoch": 0.6768570276032962,
      "grad_norm": 0.7000338435173035,
      "learning_rate": 2.4974411607332284e-06,
      "loss": 0.6927,
      "step": 23491
    },
    {
      "epoch": 0.6768858410649455,
      "grad_norm": 0.747142493724823,
      "learning_rate": 2.4970372133718595e-06,
      "loss": 0.698,
      "step": 23492
    },
    {
      "epoch": 0.6769146545265948,
      "grad_norm": 0.7177867293357849,
      "learning_rate": 2.4966332878087963e-06,
      "loss": 0.6941,
      "step": 23493
    },
    {
      "epoch": 0.6769434679882441,
      "grad_norm": 0.7177350521087646,
      "learning_rate": 2.496229384047553e-06,
      "loss": 0.7014,
      "step": 23494
    },
    {
      "epoch": 0.6769722814498934,
      "grad_norm": 0.7313885688781738,
      "learning_rate": 2.4958255020916505e-06,
      "loss": 0.6797,
      "step": 23495
    },
    {
      "epoch": 0.6770010949115427,
      "grad_norm": 0.7256617546081543,
      "learning_rate": 2.495421641944603e-06,
      "loss": 0.68,
      "step": 23496
    },
    {
      "epoch": 0.677029908373192,
      "grad_norm": 0.7110991477966309,
      "learning_rate": 2.495017803609929e-06,
      "loss": 0.679,
      "step": 23497
    },
    {
      "epoch": 0.6770587218348413,
      "grad_norm": 0.7157652378082275,
      "learning_rate": 2.4946139870911486e-06,
      "loss": 0.6854,
      "step": 23498
    },
    {
      "epoch": 0.6770875352964906,
      "grad_norm": 0.7449844479560852,
      "learning_rate": 2.4942101923917734e-06,
      "loss": 0.7254,
      "step": 23499
    },
    {
      "epoch": 0.6771163487581398,
      "grad_norm": 0.7175201773643494,
      "learning_rate": 2.4938064195153224e-06,
      "loss": 0.6666,
      "step": 23500
    },
    {
      "epoch": 0.677145162219789,
      "grad_norm": 0.7116124629974365,
      "learning_rate": 2.493402668465312e-06,
      "loss": 0.687,
      "step": 23501
    },
    {
      "epoch": 0.6771739756814383,
      "grad_norm": 0.7219972610473633,
      "learning_rate": 2.4929989392452615e-06,
      "loss": 0.6748,
      "step": 23502
    },
    {
      "epoch": 0.6772027891430876,
      "grad_norm": 0.725010871887207,
      "learning_rate": 2.4925952318586816e-06,
      "loss": 0.698,
      "step": 23503
    },
    {
      "epoch": 0.6772316026047369,
      "grad_norm": 0.7172830104827881,
      "learning_rate": 2.492191546309093e-06,
      "loss": 0.6646,
      "step": 23504
    },
    {
      "epoch": 0.6772604160663862,
      "grad_norm": 0.7250449061393738,
      "learning_rate": 2.491787882600007e-06,
      "loss": 0.6902,
      "step": 23505
    },
    {
      "epoch": 0.6772892295280355,
      "grad_norm": 0.7025745511054993,
      "learning_rate": 2.491384240734943e-06,
      "loss": 0.6751,
      "step": 23506
    },
    {
      "epoch": 0.6773180429896848,
      "grad_norm": 0.7208526730537415,
      "learning_rate": 2.4909806207174127e-06,
      "loss": 0.6808,
      "step": 23507
    },
    {
      "epoch": 0.6773468564513341,
      "grad_norm": 0.7330303192138672,
      "learning_rate": 2.490577022550935e-06,
      "loss": 0.6711,
      "step": 23508
    },
    {
      "epoch": 0.6773756699129834,
      "grad_norm": 0.7702812552452087,
      "learning_rate": 2.4901734462390216e-06,
      "loss": 0.7218,
      "step": 23509
    },
    {
      "epoch": 0.6774044833746327,
      "grad_norm": 0.7206306457519531,
      "learning_rate": 2.48976989178519e-06,
      "loss": 0.6816,
      "step": 23510
    },
    {
      "epoch": 0.677433296836282,
      "grad_norm": 0.727613091468811,
      "learning_rate": 2.489366359192949e-06,
      "loss": 0.6973,
      "step": 23511
    },
    {
      "epoch": 0.6774621102979312,
      "grad_norm": 0.738925039768219,
      "learning_rate": 2.4889628484658217e-06,
      "loss": 0.701,
      "step": 23512
    },
    {
      "epoch": 0.6774909237595804,
      "grad_norm": 0.7127599120140076,
      "learning_rate": 2.488559359607316e-06,
      "loss": 0.6968,
      "step": 23513
    },
    {
      "epoch": 0.6775197372212297,
      "grad_norm": 0.7132371664047241,
      "learning_rate": 2.48815589262095e-06,
      "loss": 0.6884,
      "step": 23514
    },
    {
      "epoch": 0.677548550682879,
      "grad_norm": 0.7351928949356079,
      "learning_rate": 2.487752447510233e-06,
      "loss": 0.6898,
      "step": 23515
    },
    {
      "epoch": 0.6775773641445283,
      "grad_norm": 0.7088222503662109,
      "learning_rate": 2.4873490242786826e-06,
      "loss": 0.6995,
      "step": 23516
    },
    {
      "epoch": 0.6776061776061776,
      "grad_norm": 0.7018744349479675,
      "learning_rate": 2.4869456229298084e-06,
      "loss": 0.707,
      "step": 23517
    },
    {
      "epoch": 0.6776349910678269,
      "grad_norm": 0.7113491296768188,
      "learning_rate": 2.486542243467126e-06,
      "loss": 0.7266,
      "step": 23518
    },
    {
      "epoch": 0.6776638045294762,
      "grad_norm": 0.734339714050293,
      "learning_rate": 2.4861388858941497e-06,
      "loss": 0.673,
      "step": 23519
    },
    {
      "epoch": 0.6776926179911255,
      "grad_norm": 0.739465057849884,
      "learning_rate": 2.4857355502143893e-06,
      "loss": 0.691,
      "step": 23520
    },
    {
      "epoch": 0.6777214314527747,
      "grad_norm": 0.7544865608215332,
      "learning_rate": 2.48533223643136e-06,
      "loss": 0.6943,
      "step": 23521
    },
    {
      "epoch": 0.677750244914424,
      "grad_norm": 0.7623985409736633,
      "learning_rate": 2.4849289445485725e-06,
      "loss": 0.6904,
      "step": 23522
    },
    {
      "epoch": 0.6777790583760733,
      "grad_norm": 0.7307782173156738,
      "learning_rate": 2.484525674569539e-06,
      "loss": 0.6964,
      "step": 23523
    },
    {
      "epoch": 0.6778078718377226,
      "grad_norm": 0.7235532999038696,
      "learning_rate": 2.4841224264977728e-06,
      "loss": 0.6987,
      "step": 23524
    },
    {
      "epoch": 0.6778366852993719,
      "grad_norm": 0.711754560470581,
      "learning_rate": 2.4837192003367865e-06,
      "loss": 0.6835,
      "step": 23525
    },
    {
      "epoch": 0.6778654987610212,
      "grad_norm": 0.7251524329185486,
      "learning_rate": 2.483315996090089e-06,
      "loss": 0.6792,
      "step": 23526
    },
    {
      "epoch": 0.6778943122226704,
      "grad_norm": 0.7134549617767334,
      "learning_rate": 2.4829128137611957e-06,
      "loss": 0.6804,
      "step": 23527
    },
    {
      "epoch": 0.6779231256843197,
      "grad_norm": 0.7324515581130981,
      "learning_rate": 2.482509653353613e-06,
      "loss": 0.7084,
      "step": 23528
    },
    {
      "epoch": 0.677951939145969,
      "grad_norm": 0.7613412141799927,
      "learning_rate": 2.4821065148708575e-06,
      "loss": 0.7317,
      "step": 23529
    },
    {
      "epoch": 0.6779807526076183,
      "grad_norm": 0.7422110438346863,
      "learning_rate": 2.4817033983164347e-06,
      "loss": 0.7105,
      "step": 23530
    },
    {
      "epoch": 0.6780095660692675,
      "grad_norm": 0.8989660739898682,
      "learning_rate": 2.48130030369386e-06,
      "loss": 0.7101,
      "step": 23531
    },
    {
      "epoch": 0.6780383795309168,
      "grad_norm": 0.7051500082015991,
      "learning_rate": 2.48089723100664e-06,
      "loss": 0.706,
      "step": 23532
    },
    {
      "epoch": 0.6780671929925661,
      "grad_norm": 0.7403827905654907,
      "learning_rate": 2.480494180258289e-06,
      "loss": 0.7216,
      "step": 23533
    },
    {
      "epoch": 0.6780960064542154,
      "grad_norm": 0.699359118938446,
      "learning_rate": 2.4800911514523126e-06,
      "loss": 0.6899,
      "step": 23534
    },
    {
      "epoch": 0.6781248199158647,
      "grad_norm": 0.6882683634757996,
      "learning_rate": 2.4796881445922238e-06,
      "loss": 0.6882,
      "step": 23535
    },
    {
      "epoch": 0.678153633377514,
      "grad_norm": 0.7520551085472107,
      "learning_rate": 2.4792851596815314e-06,
      "loss": 0.6993,
      "step": 23536
    },
    {
      "epoch": 0.6781824468391633,
      "grad_norm": 0.724724292755127,
      "learning_rate": 2.478882196723748e-06,
      "loss": 0.7216,
      "step": 23537
    },
    {
      "epoch": 0.6782112603008126,
      "grad_norm": 0.7115222215652466,
      "learning_rate": 2.478479255722378e-06,
      "loss": 0.6918,
      "step": 23538
    },
    {
      "epoch": 0.6782400737624619,
      "grad_norm": 0.7086887955665588,
      "learning_rate": 2.4780763366809333e-06,
      "loss": 0.7033,
      "step": 23539
    },
    {
      "epoch": 0.678268887224111,
      "grad_norm": 0.7081864476203918,
      "learning_rate": 2.477673439602925e-06,
      "loss": 0.6932,
      "step": 23540
    },
    {
      "epoch": 0.6782977006857603,
      "grad_norm": 0.7103584408760071,
      "learning_rate": 2.4772705644918576e-06,
      "loss": 0.6754,
      "step": 23541
    },
    {
      "epoch": 0.6783265141474096,
      "grad_norm": 0.7454468011856079,
      "learning_rate": 2.4768677113512436e-06,
      "loss": 0.7058,
      "step": 23542
    },
    {
      "epoch": 0.6783553276090589,
      "grad_norm": 0.7040948271751404,
      "learning_rate": 2.4764648801845875e-06,
      "loss": 0.6953,
      "step": 23543
    },
    {
      "epoch": 0.6783841410707082,
      "grad_norm": 0.7073121666908264,
      "learning_rate": 2.4760620709954015e-06,
      "loss": 0.6892,
      "step": 23544
    },
    {
      "epoch": 0.6784129545323575,
      "grad_norm": 0.7605248689651489,
      "learning_rate": 2.47565928378719e-06,
      "loss": 0.7211,
      "step": 23545
    },
    {
      "epoch": 0.6784417679940068,
      "grad_norm": 0.6974915862083435,
      "learning_rate": 2.475256518563465e-06,
      "loss": 0.69,
      "step": 23546
    },
    {
      "epoch": 0.6784705814556561,
      "grad_norm": 0.7068099975585938,
      "learning_rate": 2.474853775327728e-06,
      "loss": 0.7055,
      "step": 23547
    },
    {
      "epoch": 0.6784993949173054,
      "grad_norm": 0.7424934506416321,
      "learning_rate": 2.4744510540834936e-06,
      "loss": 0.6995,
      "step": 23548
    },
    {
      "epoch": 0.6785282083789547,
      "grad_norm": 0.7037780284881592,
      "learning_rate": 2.4740483548342646e-06,
      "loss": 0.6896,
      "step": 23549
    },
    {
      "epoch": 0.678557021840604,
      "grad_norm": 0.7155177593231201,
      "learning_rate": 2.473645677583551e-06,
      "loss": 0.6874,
      "step": 23550
    },
    {
      "epoch": 0.6785858353022532,
      "grad_norm": 0.7241472005844116,
      "learning_rate": 2.473243022334857e-06,
      "loss": 0.6772,
      "step": 23551
    },
    {
      "epoch": 0.6786146487639025,
      "grad_norm": 0.7277387976646423,
      "learning_rate": 2.4728403890916915e-06,
      "loss": 0.7008,
      "step": 23552
    },
    {
      "epoch": 0.6786434622255518,
      "grad_norm": 0.7160487771034241,
      "learning_rate": 2.472437777857558e-06,
      "loss": 0.7141,
      "step": 23553
    },
    {
      "epoch": 0.678672275687201,
      "grad_norm": 0.7082672119140625,
      "learning_rate": 2.472035188635967e-06,
      "loss": 0.6928,
      "step": 23554
    },
    {
      "epoch": 0.6787010891488503,
      "grad_norm": 0.726702094078064,
      "learning_rate": 2.471632621430421e-06,
      "loss": 0.7056,
      "step": 23555
    },
    {
      "epoch": 0.6787299026104996,
      "grad_norm": 0.6938260197639465,
      "learning_rate": 2.471230076244427e-06,
      "loss": 0.6669,
      "step": 23556
    },
    {
      "epoch": 0.6787587160721489,
      "grad_norm": 0.7086886167526245,
      "learning_rate": 2.470827553081493e-06,
      "loss": 0.7001,
      "step": 23557
    },
    {
      "epoch": 0.6787875295337982,
      "grad_norm": 0.6950708031654358,
      "learning_rate": 2.4704250519451207e-06,
      "loss": 0.6659,
      "step": 23558
    },
    {
      "epoch": 0.6788163429954475,
      "grad_norm": 0.7566670775413513,
      "learning_rate": 2.4700225728388178e-06,
      "loss": 0.6813,
      "step": 23559
    },
    {
      "epoch": 0.6788451564570968,
      "grad_norm": 0.7135833501815796,
      "learning_rate": 2.4696201157660895e-06,
      "loss": 0.6927,
      "step": 23560
    },
    {
      "epoch": 0.678873969918746,
      "grad_norm": 0.7003036737442017,
      "learning_rate": 2.469217680730442e-06,
      "loss": 0.6889,
      "step": 23561
    },
    {
      "epoch": 0.6789027833803953,
      "grad_norm": 0.7265549302101135,
      "learning_rate": 2.468815267735377e-06,
      "loss": 0.6979,
      "step": 23562
    },
    {
      "epoch": 0.6789315968420446,
      "grad_norm": 0.7155527472496033,
      "learning_rate": 2.468412876784403e-06,
      "loss": 0.6952,
      "step": 23563
    },
    {
      "epoch": 0.6789604103036939,
      "grad_norm": 0.6913748979568481,
      "learning_rate": 2.4680105078810197e-06,
      "loss": 0.6735,
      "step": 23564
    },
    {
      "epoch": 0.6789892237653432,
      "grad_norm": 0.7064865231513977,
      "learning_rate": 2.4676081610287358e-06,
      "loss": 0.691,
      "step": 23565
    },
    {
      "epoch": 0.6790180372269925,
      "grad_norm": 0.7142331004142761,
      "learning_rate": 2.4672058362310514e-06,
      "loss": 0.678,
      "step": 23566
    },
    {
      "epoch": 0.6790468506886417,
      "grad_norm": 0.7471378445625305,
      "learning_rate": 2.466803533491474e-06,
      "loss": 0.7389,
      "step": 23567
    },
    {
      "epoch": 0.679075664150291,
      "grad_norm": 0.7340097427368164,
      "learning_rate": 2.4664012528135045e-06,
      "loss": 0.7164,
      "step": 23568
    },
    {
      "epoch": 0.6791044776119403,
      "grad_norm": 0.7123980522155762,
      "learning_rate": 2.4659989942006483e-06,
      "loss": 0.6713,
      "step": 23569
    },
    {
      "epoch": 0.6791332910735896,
      "grad_norm": 0.7135636210441589,
      "learning_rate": 2.4655967576564066e-06,
      "loss": 0.6907,
      "step": 23570
    },
    {
      "epoch": 0.6791621045352388,
      "grad_norm": 0.725257933139801,
      "learning_rate": 2.465194543184284e-06,
      "loss": 0.7107,
      "step": 23571
    },
    {
      "epoch": 0.6791909179968881,
      "grad_norm": 0.7507457137107849,
      "learning_rate": 2.464792350787782e-06,
      "loss": 0.7007,
      "step": 23572
    },
    {
      "epoch": 0.6792197314585374,
      "grad_norm": 0.8321290016174316,
      "learning_rate": 2.464390180470407e-06,
      "loss": 0.6876,
      "step": 23573
    },
    {
      "epoch": 0.6792485449201867,
      "grad_norm": 0.7090574502944946,
      "learning_rate": 2.4639880322356567e-06,
      "loss": 0.6886,
      "step": 23574
    },
    {
      "epoch": 0.679277358381836,
      "grad_norm": 0.7252680063247681,
      "learning_rate": 2.4635859060870372e-06,
      "loss": 0.6907,
      "step": 23575
    },
    {
      "epoch": 0.6793061718434853,
      "grad_norm": 0.7393798232078552,
      "learning_rate": 2.4631838020280473e-06,
      "loss": 0.6965,
      "step": 23576
    },
    {
      "epoch": 0.6793349853051346,
      "grad_norm": 0.7375816106796265,
      "learning_rate": 2.4627817200621906e-06,
      "loss": 0.6952,
      "step": 23577
    },
    {
      "epoch": 0.6793637987667839,
      "grad_norm": 0.7569211721420288,
      "learning_rate": 2.4623796601929713e-06,
      "loss": 0.6708,
      "step": 23578
    },
    {
      "epoch": 0.6793926122284332,
      "grad_norm": 0.7036262154579163,
      "learning_rate": 2.4619776224238856e-06,
      "loss": 0.6796,
      "step": 23579
    },
    {
      "epoch": 0.6794214256900825,
      "grad_norm": 0.7103201150894165,
      "learning_rate": 2.46157560675844e-06,
      "loss": 0.6961,
      "step": 23580
    },
    {
      "epoch": 0.6794502391517316,
      "grad_norm": 0.7103011608123779,
      "learning_rate": 2.461173613200132e-06,
      "loss": 0.7155,
      "step": 23581
    },
    {
      "epoch": 0.6794790526133809,
      "grad_norm": 0.7090175151824951,
      "learning_rate": 2.4607716417524653e-06,
      "loss": 0.7257,
      "step": 23582
    },
    {
      "epoch": 0.6795078660750302,
      "grad_norm": 0.7227737903594971,
      "learning_rate": 2.4603696924189352e-06,
      "loss": 0.7029,
      "step": 23583
    },
    {
      "epoch": 0.6795366795366795,
      "grad_norm": 0.7154294848442078,
      "learning_rate": 2.4599677652030508e-06,
      "loss": 0.6887,
      "step": 23584
    },
    {
      "epoch": 0.6795654929983288,
      "grad_norm": 0.7545790076255798,
      "learning_rate": 2.4595658601083054e-06,
      "loss": 0.7125,
      "step": 23585
    },
    {
      "epoch": 0.6795943064599781,
      "grad_norm": 0.7462210655212402,
      "learning_rate": 2.459163977138204e-06,
      "loss": 0.7109,
      "step": 23586
    },
    {
      "epoch": 0.6796231199216274,
      "grad_norm": 0.7129663228988647,
      "learning_rate": 2.4587621162962433e-06,
      "loss": 0.6804,
      "step": 23587
    },
    {
      "epoch": 0.6796519333832767,
      "grad_norm": 0.7231006622314453,
      "learning_rate": 2.4583602775859256e-06,
      "loss": 0.7055,
      "step": 23588
    },
    {
      "epoch": 0.679680746844926,
      "grad_norm": 0.8366995453834534,
      "learning_rate": 2.457958461010747e-06,
      "loss": 0.6735,
      "step": 23589
    },
    {
      "epoch": 0.6797095603065753,
      "grad_norm": 0.7485207915306091,
      "learning_rate": 2.457556666574212e-06,
      "loss": 0.7268,
      "step": 23590
    },
    {
      "epoch": 0.6797383737682245,
      "grad_norm": 0.7229446172714233,
      "learning_rate": 2.457154894279814e-06,
      "loss": 0.6833,
      "step": 23591
    },
    {
      "epoch": 0.6797671872298738,
      "grad_norm": 0.7079241871833801,
      "learning_rate": 2.456753144131057e-06,
      "loss": 0.7021,
      "step": 23592
    },
    {
      "epoch": 0.6797960006915231,
      "grad_norm": 0.7448685765266418,
      "learning_rate": 2.456351416131436e-06,
      "loss": 0.6986,
      "step": 23593
    },
    {
      "epoch": 0.6798248141531723,
      "grad_norm": 0.7309920191764832,
      "learning_rate": 2.4559497102844516e-06,
      "loss": 0.6971,
      "step": 23594
    },
    {
      "epoch": 0.6798536276148216,
      "grad_norm": 0.7089516520500183,
      "learning_rate": 2.455548026593602e-06,
      "loss": 0.6863,
      "step": 23595
    },
    {
      "epoch": 0.6798824410764709,
      "grad_norm": 0.722042977809906,
      "learning_rate": 2.4551463650623873e-06,
      "loss": 0.7069,
      "step": 23596
    },
    {
      "epoch": 0.6799112545381202,
      "grad_norm": 0.731174647808075,
      "learning_rate": 2.4547447256943026e-06,
      "loss": 0.6911,
      "step": 23597
    },
    {
      "epoch": 0.6799400679997695,
      "grad_norm": 0.7180007100105286,
      "learning_rate": 2.4543431084928466e-06,
      "loss": 0.6944,
      "step": 23598
    },
    {
      "epoch": 0.6799688814614188,
      "grad_norm": 0.7034518718719482,
      "learning_rate": 2.453941513461519e-06,
      "loss": 0.6941,
      "step": 23599
    },
    {
      "epoch": 0.679997694923068,
      "grad_norm": 0.7147948741912842,
      "learning_rate": 2.453539940603814e-06,
      "loss": 0.6726,
      "step": 23600
    },
    {
      "epoch": 0.6800265083847173,
      "grad_norm": 0.732047975063324,
      "learning_rate": 2.4531383899232325e-06,
      "loss": 0.7043,
      "step": 23601
    },
    {
      "epoch": 0.6800553218463666,
      "grad_norm": 0.7300956845283508,
      "learning_rate": 2.4527368614232683e-06,
      "loss": 0.6965,
      "step": 23602
    },
    {
      "epoch": 0.6800841353080159,
      "grad_norm": 0.713272213935852,
      "learning_rate": 2.452335355107421e-06,
      "loss": 0.696,
      "step": 23603
    },
    {
      "epoch": 0.6801129487696652,
      "grad_norm": 0.7168956398963928,
      "learning_rate": 2.451933870979184e-06,
      "loss": 0.6925,
      "step": 23604
    },
    {
      "epoch": 0.6801417622313145,
      "grad_norm": 0.7450692653656006,
      "learning_rate": 2.4515324090420585e-06,
      "loss": 0.7159,
      "step": 23605
    },
    {
      "epoch": 0.6801705756929638,
      "grad_norm": 0.7053653597831726,
      "learning_rate": 2.4511309692995355e-06,
      "loss": 0.7073,
      "step": 23606
    },
    {
      "epoch": 0.6801993891546131,
      "grad_norm": 0.7249984741210938,
      "learning_rate": 2.450729551755114e-06,
      "loss": 0.6946,
      "step": 23607
    },
    {
      "epoch": 0.6802282026162623,
      "grad_norm": 0.7005087733268738,
      "learning_rate": 2.45032815641229e-06,
      "loss": 0.691,
      "step": 23608
    },
    {
      "epoch": 0.6802570160779116,
      "grad_norm": 0.7123035192489624,
      "learning_rate": 2.449926783274561e-06,
      "loss": 0.7094,
      "step": 23609
    },
    {
      "epoch": 0.6802858295395608,
      "grad_norm": 0.7217819690704346,
      "learning_rate": 2.4495254323454183e-06,
      "loss": 0.7066,
      "step": 23610
    },
    {
      "epoch": 0.6803146430012101,
      "grad_norm": 0.7160211205482483,
      "learning_rate": 2.4491241036283615e-06,
      "loss": 0.6891,
      "step": 23611
    },
    {
      "epoch": 0.6803434564628594,
      "grad_norm": 0.700669527053833,
      "learning_rate": 2.4487227971268825e-06,
      "loss": 0.7078,
      "step": 23612
    },
    {
      "epoch": 0.6803722699245087,
      "grad_norm": 0.7034903764724731,
      "learning_rate": 2.4483215128444787e-06,
      "loss": 0.6793,
      "step": 23613
    },
    {
      "epoch": 0.680401083386158,
      "grad_norm": 0.7099391222000122,
      "learning_rate": 2.447920250784642e-06,
      "loss": 0.7109,
      "step": 23614
    },
    {
      "epoch": 0.6804298968478073,
      "grad_norm": 0.7188441753387451,
      "learning_rate": 2.447519010950869e-06,
      "loss": 0.7071,
      "step": 23615
    },
    {
      "epoch": 0.6804587103094566,
      "grad_norm": 0.7104893326759338,
      "learning_rate": 2.447117793346656e-06,
      "loss": 0.7187,
      "step": 23616
    },
    {
      "epoch": 0.6804875237711059,
      "grad_norm": 0.7124400734901428,
      "learning_rate": 2.4467165979754925e-06,
      "loss": 0.7087,
      "step": 23617
    },
    {
      "epoch": 0.6805163372327552,
      "grad_norm": 0.7224041223526001,
      "learning_rate": 2.446315424840878e-06,
      "loss": 0.6976,
      "step": 23618
    },
    {
      "epoch": 0.6805451506944045,
      "grad_norm": 0.7437769770622253,
      "learning_rate": 2.4459142739463003e-06,
      "loss": 0.6794,
      "step": 23619
    },
    {
      "epoch": 0.6805739641560538,
      "grad_norm": 0.6990332007408142,
      "learning_rate": 2.445513145295257e-06,
      "loss": 0.6814,
      "step": 23620
    },
    {
      "epoch": 0.6806027776177029,
      "grad_norm": 0.7185156345367432,
      "learning_rate": 2.4451120388912404e-06,
      "loss": 0.6926,
      "step": 23621
    },
    {
      "epoch": 0.6806315910793522,
      "grad_norm": 0.7140899896621704,
      "learning_rate": 2.4447109547377456e-06,
      "loss": 0.686,
      "step": 23622
    },
    {
      "epoch": 0.6806604045410015,
      "grad_norm": 0.7180702686309814,
      "learning_rate": 2.4443098928382623e-06,
      "loss": 0.6864,
      "step": 23623
    },
    {
      "epoch": 0.6806892180026508,
      "grad_norm": 0.7151591777801514,
      "learning_rate": 2.4439088531962874e-06,
      "loss": 0.6772,
      "step": 23624
    },
    {
      "epoch": 0.6807180314643001,
      "grad_norm": 0.7314831018447876,
      "learning_rate": 2.443507835815309e-06,
      "loss": 0.682,
      "step": 23625
    },
    {
      "epoch": 0.6807468449259494,
      "grad_norm": 0.7358783483505249,
      "learning_rate": 2.443106840698824e-06,
      "loss": 0.6935,
      "step": 23626
    },
    {
      "epoch": 0.6807756583875987,
      "grad_norm": 0.7523512244224548,
      "learning_rate": 2.44270586785032e-06,
      "loss": 0.6962,
      "step": 23627
    },
    {
      "epoch": 0.680804471849248,
      "grad_norm": 0.6948211789131165,
      "learning_rate": 2.4423049172732937e-06,
      "loss": 0.703,
      "step": 23628
    },
    {
      "epoch": 0.6808332853108973,
      "grad_norm": 0.7013964653015137,
      "learning_rate": 2.4419039889712325e-06,
      "loss": 0.6908,
      "step": 23629
    },
    {
      "epoch": 0.6808620987725466,
      "grad_norm": 0.742231547832489,
      "learning_rate": 2.441503082947633e-06,
      "loss": 0.7229,
      "step": 23630
    },
    {
      "epoch": 0.6808909122341958,
      "grad_norm": 0.7178170084953308,
      "learning_rate": 2.44110219920598e-06,
      "loss": 0.6826,
      "step": 23631
    },
    {
      "epoch": 0.6809197256958451,
      "grad_norm": 0.7191375494003296,
      "learning_rate": 2.440701337749773e-06,
      "loss": 0.7073,
      "step": 23632
    },
    {
      "epoch": 0.6809485391574944,
      "grad_norm": 0.7107396125793457,
      "learning_rate": 2.440300498582496e-06,
      "loss": 0.6759,
      "step": 23633
    },
    {
      "epoch": 0.6809773526191437,
      "grad_norm": 0.7448018789291382,
      "learning_rate": 2.4398996817076458e-06,
      "loss": 0.6865,
      "step": 23634
    },
    {
      "epoch": 0.6810061660807929,
      "grad_norm": 0.74024498462677,
      "learning_rate": 2.4394988871287083e-06,
      "loss": 0.7095,
      "step": 23635
    },
    {
      "epoch": 0.6810349795424422,
      "grad_norm": 0.7268757820129395,
      "learning_rate": 2.4390981148491753e-06,
      "loss": 0.6974,
      "step": 23636
    },
    {
      "epoch": 0.6810637930040915,
      "grad_norm": 0.746446430683136,
      "learning_rate": 2.4386973648725403e-06,
      "loss": 0.6967,
      "step": 23637
    },
    {
      "epoch": 0.6810926064657408,
      "grad_norm": 0.7367163300514221,
      "learning_rate": 2.438296637202289e-06,
      "loss": 0.6861,
      "step": 23638
    },
    {
      "epoch": 0.6811214199273901,
      "grad_norm": 0.7372339367866516,
      "learning_rate": 2.437895931841915e-06,
      "loss": 0.6991,
      "step": 23639
    },
    {
      "epoch": 0.6811502333890394,
      "grad_norm": 0.7272555232048035,
      "learning_rate": 2.4374952487949037e-06,
      "loss": 0.6877,
      "step": 23640
    },
    {
      "epoch": 0.6811790468506886,
      "grad_norm": 0.6996456980705261,
      "learning_rate": 2.43709458806475e-06,
      "loss": 0.68,
      "step": 23641
    },
    {
      "epoch": 0.6812078603123379,
      "grad_norm": 0.7264438271522522,
      "learning_rate": 2.436693949654938e-06,
      "loss": 0.6923,
      "step": 23642
    },
    {
      "epoch": 0.6812366737739872,
      "grad_norm": 0.7372094392776489,
      "learning_rate": 2.43629333356896e-06,
      "loss": 0.6881,
      "step": 23643
    },
    {
      "epoch": 0.6812654872356365,
      "grad_norm": 0.775196373462677,
      "learning_rate": 2.4358927398103033e-06,
      "loss": 0.678,
      "step": 23644
    },
    {
      "epoch": 0.6812943006972858,
      "grad_norm": 0.721952497959137,
      "learning_rate": 2.43549216838246e-06,
      "loss": 0.7103,
      "step": 23645
    },
    {
      "epoch": 0.6813231141589351,
      "grad_norm": 0.7065948843955994,
      "learning_rate": 2.4350916192889144e-06,
      "loss": 0.6901,
      "step": 23646
    },
    {
      "epoch": 0.6813519276205844,
      "grad_norm": 0.7197771072387695,
      "learning_rate": 2.434691092533159e-06,
      "loss": 0.6978,
      "step": 23647
    },
    {
      "epoch": 0.6813807410822336,
      "grad_norm": 0.7356934547424316,
      "learning_rate": 2.434290588118678e-06,
      "loss": 0.7179,
      "step": 23648
    },
    {
      "epoch": 0.6814095545438829,
      "grad_norm": 0.7370237112045288,
      "learning_rate": 2.4338901060489627e-06,
      "loss": 0.6934,
      "step": 23649
    },
    {
      "epoch": 0.6814383680055321,
      "grad_norm": 0.7186567187309265,
      "learning_rate": 2.4334896463274977e-06,
      "loss": 0.6862,
      "step": 23650
    },
    {
      "epoch": 0.6814671814671814,
      "grad_norm": 0.7219485640525818,
      "learning_rate": 2.4330892089577747e-06,
      "loss": 0.6799,
      "step": 23651
    },
    {
      "epoch": 0.6814959949288307,
      "grad_norm": 0.7246645092964172,
      "learning_rate": 2.4326887939432757e-06,
      "loss": 0.6746,
      "step": 23652
    },
    {
      "epoch": 0.68152480839048,
      "grad_norm": 0.7400345206260681,
      "learning_rate": 2.432288401287492e-06,
      "loss": 0.6944,
      "step": 23653
    },
    {
      "epoch": 0.6815536218521293,
      "grad_norm": 0.7324612736701965,
      "learning_rate": 2.431888030993911e-06,
      "loss": 0.7101,
      "step": 23654
    },
    {
      "epoch": 0.6815824353137786,
      "grad_norm": 0.7025284171104431,
      "learning_rate": 2.4314876830660163e-06,
      "loss": 0.6889,
      "step": 23655
    },
    {
      "epoch": 0.6816112487754279,
      "grad_norm": 0.7102782726287842,
      "learning_rate": 2.4310873575072957e-06,
      "loss": 0.6857,
      "step": 23656
    },
    {
      "epoch": 0.6816400622370772,
      "grad_norm": 0.7225756049156189,
      "learning_rate": 2.430687054321239e-06,
      "loss": 0.7042,
      "step": 23657
    },
    {
      "epoch": 0.6816688756987265,
      "grad_norm": 0.744532585144043,
      "learning_rate": 2.430286773511327e-06,
      "loss": 0.6865,
      "step": 23658
    },
    {
      "epoch": 0.6816976891603758,
      "grad_norm": 0.7123438119888306,
      "learning_rate": 2.4298865150810493e-06,
      "loss": 0.6797,
      "step": 23659
    },
    {
      "epoch": 0.681726502622025,
      "grad_norm": 0.7319261431694031,
      "learning_rate": 2.429486279033892e-06,
      "loss": 0.6813,
      "step": 23660
    },
    {
      "epoch": 0.6817553160836743,
      "grad_norm": 0.7375983595848083,
      "learning_rate": 2.429086065373338e-06,
      "loss": 0.7185,
      "step": 23661
    },
    {
      "epoch": 0.6817841295453235,
      "grad_norm": 0.7285767197608948,
      "learning_rate": 2.4286858741028767e-06,
      "loss": 0.6785,
      "step": 23662
    },
    {
      "epoch": 0.6818129430069728,
      "grad_norm": 0.7564827799797058,
      "learning_rate": 2.428285705225989e-06,
      "loss": 0.7436,
      "step": 23663
    },
    {
      "epoch": 0.6818417564686221,
      "grad_norm": 0.7192716598510742,
      "learning_rate": 2.4278855587461636e-06,
      "loss": 0.6813,
      "step": 23664
    },
    {
      "epoch": 0.6818705699302714,
      "grad_norm": 0.7278667092323303,
      "learning_rate": 2.427485434666882e-06,
      "loss": 0.6994,
      "step": 23665
    },
    {
      "epoch": 0.6818993833919207,
      "grad_norm": 0.7323088645935059,
      "learning_rate": 2.4270853329916305e-06,
      "loss": 0.6828,
      "step": 23666
    },
    {
      "epoch": 0.68192819685357,
      "grad_norm": 0.7075616717338562,
      "learning_rate": 2.4266852537238944e-06,
      "loss": 0.6888,
      "step": 23667
    },
    {
      "epoch": 0.6819570103152193,
      "grad_norm": 0.7411883473396301,
      "learning_rate": 2.4262851968671585e-06,
      "loss": 0.7006,
      "step": 23668
    },
    {
      "epoch": 0.6819858237768686,
      "grad_norm": 0.7322559952735901,
      "learning_rate": 2.425885162424905e-06,
      "loss": 0.6887,
      "step": 23669
    },
    {
      "epoch": 0.6820146372385179,
      "grad_norm": 0.7148874998092651,
      "learning_rate": 2.4254851504006195e-06,
      "loss": 0.6961,
      "step": 23670
    },
    {
      "epoch": 0.6820434507001671,
      "grad_norm": 0.6763729453086853,
      "learning_rate": 2.4250851607977825e-06,
      "loss": 0.6702,
      "step": 23671
    },
    {
      "epoch": 0.6820722641618164,
      "grad_norm": 0.7243652939796448,
      "learning_rate": 2.4246851936198827e-06,
      "loss": 0.7095,
      "step": 23672
    },
    {
      "epoch": 0.6821010776234657,
      "grad_norm": 0.7132010459899902,
      "learning_rate": 2.4242852488703984e-06,
      "loss": 0.7075,
      "step": 23673
    },
    {
      "epoch": 0.682129891085115,
      "grad_norm": 0.709908664226532,
      "learning_rate": 2.4238853265528145e-06,
      "loss": 0.7234,
      "step": 23674
    },
    {
      "epoch": 0.6821587045467642,
      "grad_norm": 0.7224316000938416,
      "learning_rate": 2.423485426670616e-06,
      "loss": 0.7098,
      "step": 23675
    },
    {
      "epoch": 0.6821875180084135,
      "grad_norm": 0.6934329867362976,
      "learning_rate": 2.423085549227282e-06,
      "loss": 0.7014,
      "step": 23676
    },
    {
      "epoch": 0.6822163314700628,
      "grad_norm": 0.6869717836380005,
      "learning_rate": 2.422685694226299e-06,
      "loss": 0.6612,
      "step": 23677
    },
    {
      "epoch": 0.6822451449317121,
      "grad_norm": 0.7135652899742126,
      "learning_rate": 2.4222858616711452e-06,
      "loss": 0.6984,
      "step": 23678
    },
    {
      "epoch": 0.6822739583933614,
      "grad_norm": 0.7369760870933533,
      "learning_rate": 2.4218860515653055e-06,
      "loss": 0.7051,
      "step": 23679
    },
    {
      "epoch": 0.6823027718550106,
      "grad_norm": 0.730174720287323,
      "learning_rate": 2.42148626391226e-06,
      "loss": 0.7211,
      "step": 23680
    },
    {
      "epoch": 0.6823315853166599,
      "grad_norm": 0.7208313941955566,
      "learning_rate": 2.421086498715494e-06,
      "loss": 0.7,
      "step": 23681
    },
    {
      "epoch": 0.6823603987783092,
      "grad_norm": 0.7069339156150818,
      "learning_rate": 2.4206867559784847e-06,
      "loss": 0.7088,
      "step": 23682
    },
    {
      "epoch": 0.6823892122399585,
      "grad_norm": 0.704464852809906,
      "learning_rate": 2.4202870357047175e-06,
      "loss": 0.6852,
      "step": 23683
    },
    {
      "epoch": 0.6824180257016078,
      "grad_norm": 0.7281095385551453,
      "learning_rate": 2.419887337897669e-06,
      "loss": 0.7006,
      "step": 23684
    },
    {
      "epoch": 0.6824468391632571,
      "grad_norm": 0.7382953763008118,
      "learning_rate": 2.419487662560825e-06,
      "loss": 0.6854,
      "step": 23685
    },
    {
      "epoch": 0.6824756526249064,
      "grad_norm": 0.6984303593635559,
      "learning_rate": 2.4190880096976615e-06,
      "loss": 0.6921,
      "step": 23686
    },
    {
      "epoch": 0.6825044660865557,
      "grad_norm": 0.7303811311721802,
      "learning_rate": 2.4186883793116638e-06,
      "loss": 0.6651,
      "step": 23687
    },
    {
      "epoch": 0.682533279548205,
      "grad_norm": 0.7047393321990967,
      "learning_rate": 2.418288771406308e-06,
      "loss": 0.6814,
      "step": 23688
    },
    {
      "epoch": 0.6825620930098542,
      "grad_norm": 0.7296433448791504,
      "learning_rate": 2.4178891859850784e-06,
      "loss": 0.6874,
      "step": 23689
    },
    {
      "epoch": 0.6825909064715034,
      "grad_norm": 0.736650824546814,
      "learning_rate": 2.4174896230514482e-06,
      "loss": 0.7034,
      "step": 23690
    },
    {
      "epoch": 0.6826197199331527,
      "grad_norm": 0.723358690738678,
      "learning_rate": 2.4170900826089065e-06,
      "loss": 0.6852,
      "step": 23691
    },
    {
      "epoch": 0.682648533394802,
      "grad_norm": 0.7069502472877502,
      "learning_rate": 2.4166905646609256e-06,
      "loss": 0.7128,
      "step": 23692
    },
    {
      "epoch": 0.6826773468564513,
      "grad_norm": 0.743150532245636,
      "learning_rate": 2.41629106921099e-06,
      "loss": 0.712,
      "step": 23693
    },
    {
      "epoch": 0.6827061603181006,
      "grad_norm": 0.7316293120384216,
      "learning_rate": 2.4158915962625743e-06,
      "loss": 0.6949,
      "step": 23694
    },
    {
      "epoch": 0.6827349737797499,
      "grad_norm": 0.7321550250053406,
      "learning_rate": 2.4154921458191615e-06,
      "loss": 0.7211,
      "step": 23695
    },
    {
      "epoch": 0.6827637872413992,
      "grad_norm": 0.702784538269043,
      "learning_rate": 2.4150927178842267e-06,
      "loss": 0.6828,
      "step": 23696
    },
    {
      "epoch": 0.6827926007030485,
      "grad_norm": 0.7040635943412781,
      "learning_rate": 2.4146933124612503e-06,
      "loss": 0.68,
      "step": 23697
    },
    {
      "epoch": 0.6828214141646978,
      "grad_norm": 0.7042173743247986,
      "learning_rate": 2.414293929553713e-06,
      "loss": 0.6775,
      "step": 23698
    },
    {
      "epoch": 0.6828502276263471,
      "grad_norm": 0.6974636912345886,
      "learning_rate": 2.4138945691650884e-06,
      "loss": 0.6993,
      "step": 23699
    },
    {
      "epoch": 0.6828790410879964,
      "grad_norm": 0.7172880172729492,
      "learning_rate": 2.413495231298859e-06,
      "loss": 0.6915,
      "step": 23700
    },
    {
      "epoch": 0.6829078545496456,
      "grad_norm": 0.6949900984764099,
      "learning_rate": 2.4130959159584986e-06,
      "loss": 0.687,
      "step": 23701
    },
    {
      "epoch": 0.6829366680112948,
      "grad_norm": 0.7199325561523438,
      "learning_rate": 2.4126966231474874e-06,
      "loss": 0.6882,
      "step": 23702
    },
    {
      "epoch": 0.6829654814729441,
      "grad_norm": 0.7338899970054626,
      "learning_rate": 2.4122973528693018e-06,
      "loss": 0.7134,
      "step": 23703
    },
    {
      "epoch": 0.6829942949345934,
      "grad_norm": 0.7224417924880981,
      "learning_rate": 2.4118981051274217e-06,
      "loss": 0.7144,
      "step": 23704
    },
    {
      "epoch": 0.6830231083962427,
      "grad_norm": 0.7174481153488159,
      "learning_rate": 2.411498879925319e-06,
      "loss": 0.6975,
      "step": 23705
    },
    {
      "epoch": 0.683051921857892,
      "grad_norm": 0.7110719084739685,
      "learning_rate": 2.4110996772664762e-06,
      "loss": 0.7115,
      "step": 23706
    },
    {
      "epoch": 0.6830807353195413,
      "grad_norm": 0.7174387574195862,
      "learning_rate": 2.410700497154366e-06,
      "loss": 0.6853,
      "step": 23707
    },
    {
      "epoch": 0.6831095487811906,
      "grad_norm": 0.7147629261016846,
      "learning_rate": 2.410301339592467e-06,
      "loss": 0.7039,
      "step": 23708
    },
    {
      "epoch": 0.6831383622428399,
      "grad_norm": 0.7040325403213501,
      "learning_rate": 2.4099022045842528e-06,
      "loss": 0.6963,
      "step": 23709
    },
    {
      "epoch": 0.6831671757044891,
      "grad_norm": 0.7203571796417236,
      "learning_rate": 2.4095030921332037e-06,
      "loss": 0.6982,
      "step": 23710
    },
    {
      "epoch": 0.6831959891661384,
      "grad_norm": 0.708807110786438,
      "learning_rate": 2.409104002242791e-06,
      "loss": 0.7045,
      "step": 23711
    },
    {
      "epoch": 0.6832248026277877,
      "grad_norm": 0.7252302169799805,
      "learning_rate": 2.408704934916492e-06,
      "loss": 0.6987,
      "step": 23712
    },
    {
      "epoch": 0.683253616089437,
      "grad_norm": 0.7103949189186096,
      "learning_rate": 2.4083058901577847e-06,
      "loss": 0.6981,
      "step": 23713
    },
    {
      "epoch": 0.6832824295510863,
      "grad_norm": 0.7216910123825073,
      "learning_rate": 2.407906867970141e-06,
      "loss": 0.7164,
      "step": 23714
    },
    {
      "epoch": 0.6833112430127356,
      "grad_norm": 0.7134481072425842,
      "learning_rate": 2.4075078683570375e-06,
      "loss": 0.6765,
      "step": 23715
    },
    {
      "epoch": 0.6833400564743848,
      "grad_norm": 0.7059482336044312,
      "learning_rate": 2.4071088913219504e-06,
      "loss": 0.688,
      "step": 23716
    },
    {
      "epoch": 0.6833688699360341,
      "grad_norm": 0.7033352851867676,
      "learning_rate": 2.4067099368683505e-06,
      "loss": 0.6656,
      "step": 23717
    },
    {
      "epoch": 0.6833976833976834,
      "grad_norm": 0.7298787832260132,
      "learning_rate": 2.406311004999715e-06,
      "loss": 0.7017,
      "step": 23718
    },
    {
      "epoch": 0.6834264968593327,
      "grad_norm": 0.7176156044006348,
      "learning_rate": 2.4059120957195202e-06,
      "loss": 0.6687,
      "step": 23719
    },
    {
      "epoch": 0.683455310320982,
      "grad_norm": 0.7221193909645081,
      "learning_rate": 2.405513209031236e-06,
      "loss": 0.6956,
      "step": 23720
    },
    {
      "epoch": 0.6834841237826312,
      "grad_norm": 0.7282077074050903,
      "learning_rate": 2.40511434493834e-06,
      "loss": 0.6839,
      "step": 23721
    },
    {
      "epoch": 0.6835129372442805,
      "grad_norm": 0.7351460456848145,
      "learning_rate": 2.404715503444302e-06,
      "loss": 0.6996,
      "step": 23722
    },
    {
      "epoch": 0.6835417507059298,
      "grad_norm": 0.7248837947845459,
      "learning_rate": 2.4043166845525996e-06,
      "loss": 0.7083,
      "step": 23723
    },
    {
      "epoch": 0.6835705641675791,
      "grad_norm": 0.703167200088501,
      "learning_rate": 2.4039178882667024e-06,
      "loss": 0.7266,
      "step": 23724
    },
    {
      "epoch": 0.6835993776292284,
      "grad_norm": 0.7262743711471558,
      "learning_rate": 2.4035191145900872e-06,
      "loss": 0.6767,
      "step": 23725
    },
    {
      "epoch": 0.6836281910908777,
      "grad_norm": 0.7082343101501465,
      "learning_rate": 2.4031203635262207e-06,
      "loss": 0.6799,
      "step": 23726
    },
    {
      "epoch": 0.683657004552527,
      "grad_norm": 0.7330306768417358,
      "learning_rate": 2.4027216350785844e-06,
      "loss": 0.7022,
      "step": 23727
    },
    {
      "epoch": 0.6836858180141763,
      "grad_norm": 0.7414679527282715,
      "learning_rate": 2.4023229292506437e-06,
      "loss": 0.699,
      "step": 23728
    },
    {
      "epoch": 0.6837146314758255,
      "grad_norm": 0.6881287693977356,
      "learning_rate": 2.4019242460458765e-06,
      "loss": 0.6681,
      "step": 23729
    },
    {
      "epoch": 0.6837434449374747,
      "grad_norm": 0.7270877957344055,
      "learning_rate": 2.4015255854677487e-06,
      "loss": 0.7103,
      "step": 23730
    },
    {
      "epoch": 0.683772258399124,
      "grad_norm": 0.7116888165473938,
      "learning_rate": 2.4011269475197384e-06,
      "loss": 0.7091,
      "step": 23731
    },
    {
      "epoch": 0.6838010718607733,
      "grad_norm": 0.7272321581840515,
      "learning_rate": 2.4007283322053128e-06,
      "loss": 0.7177,
      "step": 23732
    },
    {
      "epoch": 0.6838298853224226,
      "grad_norm": 0.7102634906768799,
      "learning_rate": 2.4003297395279447e-06,
      "loss": 0.7033,
      "step": 23733
    },
    {
      "epoch": 0.6838586987840719,
      "grad_norm": 0.7269845008850098,
      "learning_rate": 2.3999311694911076e-06,
      "loss": 0.7034,
      "step": 23734
    },
    {
      "epoch": 0.6838875122457212,
      "grad_norm": 0.717636227607727,
      "learning_rate": 2.399532622098269e-06,
      "loss": 0.7013,
      "step": 23735
    },
    {
      "epoch": 0.6839163257073705,
      "grad_norm": 0.7203472852706909,
      "learning_rate": 2.3991340973529035e-06,
      "loss": 0.7026,
      "step": 23736
    },
    {
      "epoch": 0.6839451391690198,
      "grad_norm": 0.8774959444999695,
      "learning_rate": 2.3987355952584785e-06,
      "loss": 0.6873,
      "step": 23737
    },
    {
      "epoch": 0.6839739526306691,
      "grad_norm": 0.7183813452720642,
      "learning_rate": 2.398337115818466e-06,
      "loss": 0.7041,
      "step": 23738
    },
    {
      "epoch": 0.6840027660923184,
      "grad_norm": 0.7193628549575806,
      "learning_rate": 2.397938659036337e-06,
      "loss": 0.7126,
      "step": 23739
    },
    {
      "epoch": 0.6840315795539677,
      "grad_norm": 0.7196312546730042,
      "learning_rate": 2.3975402249155623e-06,
      "loss": 0.7143,
      "step": 23740
    },
    {
      "epoch": 0.6840603930156169,
      "grad_norm": 0.6923837661743164,
      "learning_rate": 2.3971418134596092e-06,
      "loss": 0.6791,
      "step": 23741
    },
    {
      "epoch": 0.6840892064772662,
      "grad_norm": 0.726430356502533,
      "learning_rate": 2.3967434246719506e-06,
      "loss": 0.6724,
      "step": 23742
    },
    {
      "epoch": 0.6841180199389154,
      "grad_norm": 0.7089890837669373,
      "learning_rate": 2.3963450585560534e-06,
      "loss": 0.7043,
      "step": 23743
    },
    {
      "epoch": 0.6841468334005647,
      "grad_norm": 0.7161300182342529,
      "learning_rate": 2.395946715115389e-06,
      "loss": 0.6967,
      "step": 23744
    },
    {
      "epoch": 0.684175646862214,
      "grad_norm": 0.7228603363037109,
      "learning_rate": 2.3955483943534243e-06,
      "loss": 0.6911,
      "step": 23745
    },
    {
      "epoch": 0.6842044603238633,
      "grad_norm": 0.7872854471206665,
      "learning_rate": 2.3951500962736312e-06,
      "loss": 0.7216,
      "step": 23746
    },
    {
      "epoch": 0.6842332737855126,
      "grad_norm": 0.7254598140716553,
      "learning_rate": 2.3947518208794752e-06,
      "loss": 0.691,
      "step": 23747
    },
    {
      "epoch": 0.6842620872471619,
      "grad_norm": 0.7387152910232544,
      "learning_rate": 2.3943535681744282e-06,
      "loss": 0.7396,
      "step": 23748
    },
    {
      "epoch": 0.6842909007088112,
      "grad_norm": 0.7155687212944031,
      "learning_rate": 2.393955338161955e-06,
      "loss": 0.7059,
      "step": 23749
    },
    {
      "epoch": 0.6843197141704604,
      "grad_norm": 0.7420839071273804,
      "learning_rate": 2.3935571308455256e-06,
      "loss": 0.6914,
      "step": 23750
    },
    {
      "epoch": 0.6843485276321097,
      "grad_norm": 0.746583878993988,
      "learning_rate": 2.393158946228609e-06,
      "loss": 0.7109,
      "step": 23751
    },
    {
      "epoch": 0.684377341093759,
      "grad_norm": 0.7304702997207642,
      "learning_rate": 2.3927607843146734e-06,
      "loss": 0.6982,
      "step": 23752
    },
    {
      "epoch": 0.6844061545554083,
      "grad_norm": 0.6937216520309448,
      "learning_rate": 2.3923626451071828e-06,
      "loss": 0.6771,
      "step": 23753
    },
    {
      "epoch": 0.6844349680170576,
      "grad_norm": 0.7115468978881836,
      "learning_rate": 2.3919645286096094e-06,
      "loss": 0.6897,
      "step": 23754
    },
    {
      "epoch": 0.6844637814787069,
      "grad_norm": 0.7132738828659058,
      "learning_rate": 2.391566434825415e-06,
      "loss": 0.6921,
      "step": 23755
    },
    {
      "epoch": 0.6844925949403561,
      "grad_norm": 0.7141363024711609,
      "learning_rate": 2.3911683637580705e-06,
      "loss": 0.6882,
      "step": 23756
    },
    {
      "epoch": 0.6845214084020054,
      "grad_norm": 0.7230284810066223,
      "learning_rate": 2.3907703154110435e-06,
      "loss": 0.6663,
      "step": 23757
    },
    {
      "epoch": 0.6845502218636547,
      "grad_norm": 0.718441903591156,
      "learning_rate": 2.3903722897877963e-06,
      "loss": 0.7116,
      "step": 23758
    },
    {
      "epoch": 0.684579035325304,
      "grad_norm": 0.709378182888031,
      "learning_rate": 2.3899742868918004e-06,
      "loss": 0.6894,
      "step": 23759
    },
    {
      "epoch": 0.6846078487869532,
      "grad_norm": 0.7349686622619629,
      "learning_rate": 2.3895763067265164e-06,
      "loss": 0.6894,
      "step": 23760
    },
    {
      "epoch": 0.6846366622486025,
      "grad_norm": 0.704602837562561,
      "learning_rate": 2.3891783492954163e-06,
      "loss": 0.6805,
      "step": 23761
    },
    {
      "epoch": 0.6846654757102518,
      "grad_norm": 0.726008415222168,
      "learning_rate": 2.388780414601959e-06,
      "loss": 0.6793,
      "step": 23762
    },
    {
      "epoch": 0.6846942891719011,
      "grad_norm": 0.7064697742462158,
      "learning_rate": 2.3883825026496172e-06,
      "loss": 0.6792,
      "step": 23763
    },
    {
      "epoch": 0.6847231026335504,
      "grad_norm": 0.7423151731491089,
      "learning_rate": 2.3879846134418516e-06,
      "loss": 0.712,
      "step": 23764
    },
    {
      "epoch": 0.6847519160951997,
      "grad_norm": 0.7193893194198608,
      "learning_rate": 2.3875867469821307e-06,
      "loss": 0.695,
      "step": 23765
    },
    {
      "epoch": 0.684780729556849,
      "grad_norm": 0.72491455078125,
      "learning_rate": 2.387188903273916e-06,
      "loss": 0.7138,
      "step": 23766
    },
    {
      "epoch": 0.6848095430184983,
      "grad_norm": 0.7140735387802124,
      "learning_rate": 2.386791082320676e-06,
      "loss": 0.6953,
      "step": 23767
    },
    {
      "epoch": 0.6848383564801476,
      "grad_norm": 0.7135543823242188,
      "learning_rate": 2.3863932841258716e-06,
      "loss": 0.6865,
      "step": 23768
    },
    {
      "epoch": 0.6848671699417969,
      "grad_norm": 0.7000893950462341,
      "learning_rate": 2.385995508692971e-06,
      "loss": 0.6714,
      "step": 23769
    },
    {
      "epoch": 0.684895983403446,
      "grad_norm": 0.7470740079879761,
      "learning_rate": 2.385597756025434e-06,
      "loss": 0.72,
      "step": 23770
    },
    {
      "epoch": 0.6849247968650953,
      "grad_norm": 0.7343244552612305,
      "learning_rate": 2.385200026126728e-06,
      "loss": 0.7139,
      "step": 23771
    },
    {
      "epoch": 0.6849536103267446,
      "grad_norm": 0.7138302326202393,
      "learning_rate": 2.3848023190003173e-06,
      "loss": 0.692,
      "step": 23772
    },
    {
      "epoch": 0.6849824237883939,
      "grad_norm": 0.731232762336731,
      "learning_rate": 2.384404634649662e-06,
      "loss": 0.7025,
      "step": 23773
    },
    {
      "epoch": 0.6850112372500432,
      "grad_norm": 0.7116025686264038,
      "learning_rate": 2.384006973078228e-06,
      "loss": 0.6702,
      "step": 23774
    },
    {
      "epoch": 0.6850400507116925,
      "grad_norm": 0.704591691493988,
      "learning_rate": 2.38360933428948e-06,
      "loss": 0.6757,
      "step": 23775
    },
    {
      "epoch": 0.6850688641733418,
      "grad_norm": 0.7415496706962585,
      "learning_rate": 2.3832117182868776e-06,
      "loss": 0.6659,
      "step": 23776
    },
    {
      "epoch": 0.6850976776349911,
      "grad_norm": 0.7442739605903625,
      "learning_rate": 2.382814125073885e-06,
      "loss": 0.6935,
      "step": 23777
    },
    {
      "epoch": 0.6851264910966404,
      "grad_norm": 0.7412011623382568,
      "learning_rate": 2.3824165546539674e-06,
      "loss": 0.7255,
      "step": 23778
    },
    {
      "epoch": 0.6851553045582897,
      "grad_norm": 0.7268147468566895,
      "learning_rate": 2.3820190070305832e-06,
      "loss": 0.7116,
      "step": 23779
    },
    {
      "epoch": 0.685184118019939,
      "grad_norm": 0.7244207262992859,
      "learning_rate": 2.3816214822071974e-06,
      "loss": 0.6967,
      "step": 23780
    },
    {
      "epoch": 0.6852129314815882,
      "grad_norm": 0.7252218723297119,
      "learning_rate": 2.3812239801872704e-06,
      "loss": 0.6872,
      "step": 23781
    },
    {
      "epoch": 0.6852417449432375,
      "grad_norm": 0.7240880131721497,
      "learning_rate": 2.380826500974266e-06,
      "loss": 0.6961,
      "step": 23782
    },
    {
      "epoch": 0.6852705584048867,
      "grad_norm": 0.704704225063324,
      "learning_rate": 2.3804290445716426e-06,
      "loss": 0.6621,
      "step": 23783
    },
    {
      "epoch": 0.685299371866536,
      "grad_norm": 0.7030459642410278,
      "learning_rate": 2.380031610982866e-06,
      "loss": 0.7193,
      "step": 23784
    },
    {
      "epoch": 0.6853281853281853,
      "grad_norm": 0.720199465751648,
      "learning_rate": 2.3796342002113928e-06,
      "loss": 0.6949,
      "step": 23785
    },
    {
      "epoch": 0.6853569987898346,
      "grad_norm": 0.6907097101211548,
      "learning_rate": 2.3792368122606856e-06,
      "loss": 0.6572,
      "step": 23786
    },
    {
      "epoch": 0.6853858122514839,
      "grad_norm": 0.7164276838302612,
      "learning_rate": 2.378839447134207e-06,
      "loss": 0.7185,
      "step": 23787
    },
    {
      "epoch": 0.6854146257131332,
      "grad_norm": 0.7262961864471436,
      "learning_rate": 2.378442104835418e-06,
      "loss": 0.6897,
      "step": 23788
    },
    {
      "epoch": 0.6854434391747825,
      "grad_norm": 0.7085074782371521,
      "learning_rate": 2.378044785367776e-06,
      "loss": 0.6888,
      "step": 23789
    },
    {
      "epoch": 0.6854722526364317,
      "grad_norm": 0.6994157433509827,
      "learning_rate": 2.3776474887347445e-06,
      "loss": 0.6931,
      "step": 23790
    },
    {
      "epoch": 0.685501066098081,
      "grad_norm": 0.7164839506149292,
      "learning_rate": 2.3772502149397802e-06,
      "loss": 0.6916,
      "step": 23791
    },
    {
      "epoch": 0.6855298795597303,
      "grad_norm": 0.7292670607566833,
      "learning_rate": 2.376852963986347e-06,
      "loss": 0.703,
      "step": 23792
    },
    {
      "epoch": 0.6855586930213796,
      "grad_norm": 0.7152995467185974,
      "learning_rate": 2.3764557358779e-06,
      "loss": 0.7116,
      "step": 23793
    },
    {
      "epoch": 0.6855875064830289,
      "grad_norm": 0.7015390396118164,
      "learning_rate": 2.3760585306179013e-06,
      "loss": 0.672,
      "step": 23794
    },
    {
      "epoch": 0.6856163199446782,
      "grad_norm": 0.6880284547805786,
      "learning_rate": 2.3756613482098117e-06,
      "loss": 0.7031,
      "step": 23795
    },
    {
      "epoch": 0.6856451334063275,
      "grad_norm": 0.7298834323883057,
      "learning_rate": 2.375264188657086e-06,
      "loss": 0.7204,
      "step": 23796
    },
    {
      "epoch": 0.6856739468679767,
      "grad_norm": 0.7225733399391174,
      "learning_rate": 2.3748670519631878e-06,
      "loss": 0.7063,
      "step": 23797
    },
    {
      "epoch": 0.685702760329626,
      "grad_norm": 0.7650822401046753,
      "learning_rate": 2.3744699381315695e-06,
      "loss": 0.6733,
      "step": 23798
    },
    {
      "epoch": 0.6857315737912753,
      "grad_norm": 0.7000763416290283,
      "learning_rate": 2.374072847165697e-06,
      "loss": 0.699,
      "step": 23799
    },
    {
      "epoch": 0.6857603872529245,
      "grad_norm": 0.7242120504379272,
      "learning_rate": 2.3736757790690236e-06,
      "loss": 0.6799,
      "step": 23800
    },
    {
      "epoch": 0.6857892007145738,
      "grad_norm": 0.6970241665840149,
      "learning_rate": 2.3732787338450103e-06,
      "loss": 0.6797,
      "step": 23801
    },
    {
      "epoch": 0.6858180141762231,
      "grad_norm": 0.7087633013725281,
      "learning_rate": 2.372881711497112e-06,
      "loss": 0.6866,
      "step": 23802
    },
    {
      "epoch": 0.6858468276378724,
      "grad_norm": 0.7087277173995972,
      "learning_rate": 2.372484712028789e-06,
      "loss": 0.6793,
      "step": 23803
    },
    {
      "epoch": 0.6858756410995217,
      "grad_norm": 0.7257556915283203,
      "learning_rate": 2.372087735443496e-06,
      "loss": 0.6974,
      "step": 23804
    },
    {
      "epoch": 0.685904454561171,
      "grad_norm": 0.7215595245361328,
      "learning_rate": 2.3716907817446944e-06,
      "loss": 0.6956,
      "step": 23805
    },
    {
      "epoch": 0.6859332680228203,
      "grad_norm": 0.7377644777297974,
      "learning_rate": 2.3712938509358363e-06,
      "loss": 0.7125,
      "step": 23806
    },
    {
      "epoch": 0.6859620814844696,
      "grad_norm": 0.7257322072982788,
      "learning_rate": 2.370896943020383e-06,
      "loss": 0.6764,
      "step": 23807
    },
    {
      "epoch": 0.6859908949461189,
      "grad_norm": 0.720483124256134,
      "learning_rate": 2.3705000580017872e-06,
      "loss": 0.6753,
      "step": 23808
    },
    {
      "epoch": 0.6860197084077682,
      "grad_norm": 0.7114745378494263,
      "learning_rate": 2.3701031958835073e-06,
      "loss": 0.685,
      "step": 23809
    },
    {
      "epoch": 0.6860485218694173,
      "grad_norm": 0.7139617204666138,
      "learning_rate": 2.369706356669e-06,
      "loss": 0.6867,
      "step": 23810
    },
    {
      "epoch": 0.6860773353310666,
      "grad_norm": 0.7059417963027954,
      "learning_rate": 2.3693095403617227e-06,
      "loss": 0.6926,
      "step": 23811
    },
    {
      "epoch": 0.6861061487927159,
      "grad_norm": 0.7115556597709656,
      "learning_rate": 2.3689127469651273e-06,
      "loss": 0.7039,
      "step": 23812
    },
    {
      "epoch": 0.6861349622543652,
      "grad_norm": 0.7165939211845398,
      "learning_rate": 2.3685159764826743e-06,
      "loss": 0.7043,
      "step": 23813
    },
    {
      "epoch": 0.6861637757160145,
      "grad_norm": 3.2069108486175537,
      "learning_rate": 2.3681192289178145e-06,
      "loss": 0.6936,
      "step": 23814
    },
    {
      "epoch": 0.6861925891776638,
      "grad_norm": 0.7186664342880249,
      "learning_rate": 2.3677225042740055e-06,
      "loss": 0.6968,
      "step": 23815
    },
    {
      "epoch": 0.6862214026393131,
      "grad_norm": 0.7273721694946289,
      "learning_rate": 2.3673258025547036e-06,
      "loss": 0.7263,
      "step": 23816
    },
    {
      "epoch": 0.6862502161009624,
      "grad_norm": 0.7354809641838074,
      "learning_rate": 2.3669291237633605e-06,
      "loss": 0.6841,
      "step": 23817
    },
    {
      "epoch": 0.6862790295626117,
      "grad_norm": 0.7141467332839966,
      "learning_rate": 2.3665324679034346e-06,
      "loss": 0.6699,
      "step": 23818
    },
    {
      "epoch": 0.686307843024261,
      "grad_norm": 0.7269234657287598,
      "learning_rate": 2.3661358349783763e-06,
      "loss": 0.6831,
      "step": 23819
    },
    {
      "epoch": 0.6863366564859102,
      "grad_norm": 0.7153235077857971,
      "learning_rate": 2.365739224991644e-06,
      "loss": 0.6866,
      "step": 23820
    },
    {
      "epoch": 0.6863654699475595,
      "grad_norm": 0.7055639028549194,
      "learning_rate": 2.365342637946688e-06,
      "loss": 0.6856,
      "step": 23821
    },
    {
      "epoch": 0.6863942834092088,
      "grad_norm": 0.7084670066833496,
      "learning_rate": 2.3649460738469637e-06,
      "loss": 0.6773,
      "step": 23822
    },
    {
      "epoch": 0.6864230968708581,
      "grad_norm": 0.7291578054428101,
      "learning_rate": 2.364549532695925e-06,
      "loss": 0.7038,
      "step": 23823
    },
    {
      "epoch": 0.6864519103325073,
      "grad_norm": 0.7315319180488586,
      "learning_rate": 2.3641530144970277e-06,
      "loss": 0.6671,
      "step": 23824
    },
    {
      "epoch": 0.6864807237941566,
      "grad_norm": 0.7254832983016968,
      "learning_rate": 2.3637565192537203e-06,
      "loss": 0.6959,
      "step": 23825
    },
    {
      "epoch": 0.6865095372558059,
      "grad_norm": 0.7288594245910645,
      "learning_rate": 2.3633600469694608e-06,
      "loss": 0.6932,
      "step": 23826
    },
    {
      "epoch": 0.6865383507174552,
      "grad_norm": 0.7697188854217529,
      "learning_rate": 2.3629635976476974e-06,
      "loss": 0.6838,
      "step": 23827
    },
    {
      "epoch": 0.6865671641791045,
      "grad_norm": 0.7427257895469666,
      "learning_rate": 2.3625671712918868e-06,
      "loss": 0.6962,
      "step": 23828
    },
    {
      "epoch": 0.6865959776407538,
      "grad_norm": 0.7369797825813293,
      "learning_rate": 2.3621707679054778e-06,
      "loss": 0.6991,
      "step": 23829
    },
    {
      "epoch": 0.686624791102403,
      "grad_norm": 0.7337762117385864,
      "learning_rate": 2.361774387491927e-06,
      "loss": 0.6695,
      "step": 23830
    },
    {
      "epoch": 0.6866536045640523,
      "grad_norm": 0.7161297798156738,
      "learning_rate": 2.361378030054681e-06,
      "loss": 0.6961,
      "step": 23831
    },
    {
      "epoch": 0.6866824180257016,
      "grad_norm": 0.7480742335319519,
      "learning_rate": 2.360981695597196e-06,
      "loss": 0.6724,
      "step": 23832
    },
    {
      "epoch": 0.6867112314873509,
      "grad_norm": 0.7112650275230408,
      "learning_rate": 2.3605853841229238e-06,
      "loss": 0.6997,
      "step": 23833
    },
    {
      "epoch": 0.6867400449490002,
      "grad_norm": 0.7165928483009338,
      "learning_rate": 2.3601890956353118e-06,
      "loss": 0.7016,
      "step": 23834
    },
    {
      "epoch": 0.6867688584106495,
      "grad_norm": 0.7094783782958984,
      "learning_rate": 2.3597928301378147e-06,
      "loss": 0.7053,
      "step": 23835
    },
    {
      "epoch": 0.6867976718722988,
      "grad_norm": 0.7298287153244019,
      "learning_rate": 2.3593965876338824e-06,
      "loss": 0.706,
      "step": 23836
    },
    {
      "epoch": 0.686826485333948,
      "grad_norm": 0.719342052936554,
      "learning_rate": 2.3590003681269684e-06,
      "loss": 0.7081,
      "step": 23837
    },
    {
      "epoch": 0.6868552987955973,
      "grad_norm": 0.7224518656730652,
      "learning_rate": 2.3586041716205196e-06,
      "loss": 0.6977,
      "step": 23838
    },
    {
      "epoch": 0.6868841122572465,
      "grad_norm": 0.7414172887802124,
      "learning_rate": 2.3582079981179897e-06,
      "loss": 0.7156,
      "step": 23839
    },
    {
      "epoch": 0.6869129257188958,
      "grad_norm": 0.7191956639289856,
      "learning_rate": 2.357811847622825e-06,
      "loss": 0.6968,
      "step": 23840
    },
    {
      "epoch": 0.6869417391805451,
      "grad_norm": 0.726184070110321,
      "learning_rate": 2.35741572013848e-06,
      "loss": 0.687,
      "step": 23841
    },
    {
      "epoch": 0.6869705526421944,
      "grad_norm": 0.7079347968101501,
      "learning_rate": 2.3570196156684006e-06,
      "loss": 0.6657,
      "step": 23842
    },
    {
      "epoch": 0.6869993661038437,
      "grad_norm": 0.7349410653114319,
      "learning_rate": 2.3566235342160405e-06,
      "loss": 0.7024,
      "step": 23843
    },
    {
      "epoch": 0.687028179565493,
      "grad_norm": 0.7069993019104004,
      "learning_rate": 2.356227475784845e-06,
      "loss": 0.6957,
      "step": 23844
    },
    {
      "epoch": 0.6870569930271423,
      "grad_norm": 0.7268518805503845,
      "learning_rate": 2.3558314403782668e-06,
      "loss": 0.7126,
      "step": 23845
    },
    {
      "epoch": 0.6870858064887916,
      "grad_norm": 0.7277959585189819,
      "learning_rate": 2.35543542799975e-06,
      "loss": 0.683,
      "step": 23846
    },
    {
      "epoch": 0.6871146199504409,
      "grad_norm": 0.7273065447807312,
      "learning_rate": 2.3550394386527507e-06,
      "loss": 0.6893,
      "step": 23847
    },
    {
      "epoch": 0.6871434334120902,
      "grad_norm": 0.7251391410827637,
      "learning_rate": 2.3546434723407122e-06,
      "loss": 0.6939,
      "step": 23848
    },
    {
      "epoch": 0.6871722468737395,
      "grad_norm": 0.7402777671813965,
      "learning_rate": 2.3542475290670865e-06,
      "loss": 0.6861,
      "step": 23849
    },
    {
      "epoch": 0.6872010603353887,
      "grad_norm": 0.7454902529716492,
      "learning_rate": 2.353851608835318e-06,
      "loss": 0.7067,
      "step": 23850
    },
    {
      "epoch": 0.6872298737970379,
      "grad_norm": 0.7156445384025574,
      "learning_rate": 2.3534557116488594e-06,
      "loss": 0.6981,
      "step": 23851
    },
    {
      "epoch": 0.6872586872586872,
      "grad_norm": 0.719767153263092,
      "learning_rate": 2.3530598375111533e-06,
      "loss": 0.6659,
      "step": 23852
    },
    {
      "epoch": 0.6872875007203365,
      "grad_norm": 0.7252017855644226,
      "learning_rate": 2.35266398642565e-06,
      "loss": 0.6898,
      "step": 23853
    },
    {
      "epoch": 0.6873163141819858,
      "grad_norm": 0.7326147556304932,
      "learning_rate": 2.3522681583957997e-06,
      "loss": 0.7014,
      "step": 23854
    },
    {
      "epoch": 0.6873451276436351,
      "grad_norm": 0.7370016574859619,
      "learning_rate": 2.3518723534250444e-06,
      "loss": 0.6994,
      "step": 23855
    },
    {
      "epoch": 0.6873739411052844,
      "grad_norm": 0.6940099596977234,
      "learning_rate": 2.351476571516836e-06,
      "loss": 0.6781,
      "step": 23856
    },
    {
      "epoch": 0.6874027545669337,
      "grad_norm": 0.7108246684074402,
      "learning_rate": 2.351080812674617e-06,
      "loss": 0.6853,
      "step": 23857
    },
    {
      "epoch": 0.687431568028583,
      "grad_norm": 0.7009439468383789,
      "learning_rate": 2.3506850769018363e-06,
      "loss": 0.6878,
      "step": 23858
    },
    {
      "epoch": 0.6874603814902323,
      "grad_norm": 0.711505651473999,
      "learning_rate": 2.3502893642019396e-06,
      "loss": 0.7053,
      "step": 23859
    },
    {
      "epoch": 0.6874891949518815,
      "grad_norm": 0.718109667301178,
      "learning_rate": 2.349893674578377e-06,
      "loss": 0.6597,
      "step": 23860
    },
    {
      "epoch": 0.6875180084135308,
      "grad_norm": 0.7218914031982422,
      "learning_rate": 2.3494980080345887e-06,
      "loss": 0.7052,
      "step": 23861
    },
    {
      "epoch": 0.6875468218751801,
      "grad_norm": 0.6894137859344482,
      "learning_rate": 2.3491023645740247e-06,
      "loss": 0.6989,
      "step": 23862
    },
    {
      "epoch": 0.6875756353368294,
      "grad_norm": 0.75926673412323,
      "learning_rate": 2.3487067442001276e-06,
      "loss": 0.7052,
      "step": 23863
    },
    {
      "epoch": 0.6876044487984786,
      "grad_norm": 0.7125607132911682,
      "learning_rate": 2.3483111469163465e-06,
      "loss": 0.6962,
      "step": 23864
    },
    {
      "epoch": 0.6876332622601279,
      "grad_norm": 0.7342005968093872,
      "learning_rate": 2.3479155727261225e-06,
      "loss": 0.6926,
      "step": 23865
    },
    {
      "epoch": 0.6876620757217772,
      "grad_norm": 0.7116212248802185,
      "learning_rate": 2.347520021632905e-06,
      "loss": 0.7077,
      "step": 23866
    },
    {
      "epoch": 0.6876908891834265,
      "grad_norm": 0.7237202525138855,
      "learning_rate": 2.3471244936401343e-06,
      "loss": 0.6986,
      "step": 23867
    },
    {
      "epoch": 0.6877197026450758,
      "grad_norm": 0.710662305355072,
      "learning_rate": 2.3467289887512594e-06,
      "loss": 0.6834,
      "step": 23868
    },
    {
      "epoch": 0.687748516106725,
      "grad_norm": 0.7217308878898621,
      "learning_rate": 2.346333506969721e-06,
      "loss": 0.6783,
      "step": 23869
    },
    {
      "epoch": 0.6877773295683743,
      "grad_norm": 0.7324057221412659,
      "learning_rate": 2.345938048298965e-06,
      "loss": 0.7058,
      "step": 23870
    },
    {
      "epoch": 0.6878061430300236,
      "grad_norm": 0.7299630641937256,
      "learning_rate": 2.345542612742435e-06,
      "loss": 0.6969,
      "step": 23871
    },
    {
      "epoch": 0.6878349564916729,
      "grad_norm": 0.7350186109542847,
      "learning_rate": 2.345147200303578e-06,
      "loss": 0.7015,
      "step": 23872
    },
    {
      "epoch": 0.6878637699533222,
      "grad_norm": 0.723944365978241,
      "learning_rate": 2.344751810985833e-06,
      "loss": 0.6997,
      "step": 23873
    },
    {
      "epoch": 0.6878925834149715,
      "grad_norm": 0.729313850402832,
      "learning_rate": 2.344356444792645e-06,
      "loss": 0.6862,
      "step": 23874
    },
    {
      "epoch": 0.6879213968766208,
      "grad_norm": 0.6868787407875061,
      "learning_rate": 2.34396110172746e-06,
      "loss": 0.6874,
      "step": 23875
    },
    {
      "epoch": 0.6879502103382701,
      "grad_norm": 0.711232602596283,
      "learning_rate": 2.343565781793716e-06,
      "loss": 0.6852,
      "step": 23876
    },
    {
      "epoch": 0.6879790237999194,
      "grad_norm": 0.7110705375671387,
      "learning_rate": 2.3431704849948617e-06,
      "loss": 0.7139,
      "step": 23877
    },
    {
      "epoch": 0.6880078372615686,
      "grad_norm": 0.7354482412338257,
      "learning_rate": 2.342775211334334e-06,
      "loss": 0.7113,
      "step": 23878
    },
    {
      "epoch": 0.6880366507232178,
      "grad_norm": 0.7056974768638611,
      "learning_rate": 2.34237996081558e-06,
      "loss": 0.7154,
      "step": 23879
    },
    {
      "epoch": 0.6880654641848671,
      "grad_norm": 0.7275897860527039,
      "learning_rate": 2.3419847334420383e-06,
      "loss": 0.6859,
      "step": 23880
    },
    {
      "epoch": 0.6880942776465164,
      "grad_norm": 0.6885539293289185,
      "learning_rate": 2.3415895292171536e-06,
      "loss": 0.6747,
      "step": 23881
    },
    {
      "epoch": 0.6881230911081657,
      "grad_norm": 0.7505030632019043,
      "learning_rate": 2.341194348144364e-06,
      "loss": 0.703,
      "step": 23882
    },
    {
      "epoch": 0.688151904569815,
      "grad_norm": 0.7186885476112366,
      "learning_rate": 2.3407991902271166e-06,
      "loss": 0.698,
      "step": 23883
    },
    {
      "epoch": 0.6881807180314643,
      "grad_norm": 0.7343882918357849,
      "learning_rate": 2.3404040554688485e-06,
      "loss": 0.6965,
      "step": 23884
    },
    {
      "epoch": 0.6882095314931136,
      "grad_norm": 0.7128956317901611,
      "learning_rate": 2.3400089438730046e-06,
      "loss": 0.684,
      "step": 23885
    },
    {
      "epoch": 0.6882383449547629,
      "grad_norm": 0.7357690930366516,
      "learning_rate": 2.3396138554430215e-06,
      "loss": 0.7057,
      "step": 23886
    },
    {
      "epoch": 0.6882671584164122,
      "grad_norm": 0.748363733291626,
      "learning_rate": 2.3392187901823443e-06,
      "loss": 0.6982,
      "step": 23887
    },
    {
      "epoch": 0.6882959718780615,
      "grad_norm": 0.7425631880760193,
      "learning_rate": 2.33882374809441e-06,
      "loss": 0.6968,
      "step": 23888
    },
    {
      "epoch": 0.6883247853397108,
      "grad_norm": 0.7304093241691589,
      "learning_rate": 2.3384287291826625e-06,
      "loss": 0.7037,
      "step": 23889
    },
    {
      "epoch": 0.68835359880136,
      "grad_norm": 0.6981497406959534,
      "learning_rate": 2.3380337334505374e-06,
      "loss": 0.6872,
      "step": 23890
    },
    {
      "epoch": 0.6883824122630092,
      "grad_norm": 0.7321070432662964,
      "learning_rate": 2.3376387609014784e-06,
      "loss": 0.6846,
      "step": 23891
    },
    {
      "epoch": 0.6884112257246585,
      "grad_norm": 0.7319753170013428,
      "learning_rate": 2.337243811538926e-06,
      "loss": 0.6998,
      "step": 23892
    },
    {
      "epoch": 0.6884400391863078,
      "grad_norm": 0.7301638126373291,
      "learning_rate": 2.3368488853663168e-06,
      "loss": 0.7137,
      "step": 23893
    },
    {
      "epoch": 0.6884688526479571,
      "grad_norm": 0.7236790060997009,
      "learning_rate": 2.3364539823870912e-06,
      "loss": 0.684,
      "step": 23894
    },
    {
      "epoch": 0.6884976661096064,
      "grad_norm": 0.7108282446861267,
      "learning_rate": 2.3360591026046892e-06,
      "loss": 0.6605,
      "step": 23895
    },
    {
      "epoch": 0.6885264795712557,
      "grad_norm": 0.7256035804748535,
      "learning_rate": 2.335664246022551e-06,
      "loss": 0.7232,
      "step": 23896
    },
    {
      "epoch": 0.688555293032905,
      "grad_norm": 0.7024908661842346,
      "learning_rate": 2.3352694126441117e-06,
      "loss": 0.7053,
      "step": 23897
    },
    {
      "epoch": 0.6885841064945543,
      "grad_norm": 0.7257930636405945,
      "learning_rate": 2.334874602472814e-06,
      "loss": 0.6953,
      "step": 23898
    },
    {
      "epoch": 0.6886129199562036,
      "grad_norm": 0.7348644137382507,
      "learning_rate": 2.334479815512093e-06,
      "loss": 0.6861,
      "step": 23899
    },
    {
      "epoch": 0.6886417334178528,
      "grad_norm": 0.7127801179885864,
      "learning_rate": 2.3340850517653903e-06,
      "loss": 0.6786,
      "step": 23900
    },
    {
      "epoch": 0.6886705468795021,
      "grad_norm": 0.7261855006217957,
      "learning_rate": 2.333690311236139e-06,
      "loss": 0.7141,
      "step": 23901
    },
    {
      "epoch": 0.6886993603411514,
      "grad_norm": 0.7061572670936584,
      "learning_rate": 2.3332955939277825e-06,
      "loss": 0.6896,
      "step": 23902
    },
    {
      "epoch": 0.6887281738028007,
      "grad_norm": 0.7323287725448608,
      "learning_rate": 2.3329008998437534e-06,
      "loss": 0.7156,
      "step": 23903
    },
    {
      "epoch": 0.68875698726445,
      "grad_norm": 0.717994213104248,
      "learning_rate": 2.3325062289874933e-06,
      "loss": 0.7041,
      "step": 23904
    },
    {
      "epoch": 0.6887858007260992,
      "grad_norm": 0.7425099611282349,
      "learning_rate": 2.3321115813624357e-06,
      "loss": 0.7166,
      "step": 23905
    },
    {
      "epoch": 0.6888146141877485,
      "grad_norm": 0.7289308309555054,
      "learning_rate": 2.3317169569720187e-06,
      "loss": 0.7159,
      "step": 23906
    },
    {
      "epoch": 0.6888434276493978,
      "grad_norm": 0.7080803513526917,
      "learning_rate": 2.33132235581968e-06,
      "loss": 0.6733,
      "step": 23907
    },
    {
      "epoch": 0.6888722411110471,
      "grad_norm": 0.7330138087272644,
      "learning_rate": 2.3309277779088577e-06,
      "loss": 0.6968,
      "step": 23908
    },
    {
      "epoch": 0.6889010545726963,
      "grad_norm": 0.7294937968254089,
      "learning_rate": 2.330533223242985e-06,
      "loss": 0.6807,
      "step": 23909
    },
    {
      "epoch": 0.6889298680343456,
      "grad_norm": 0.7156116366386414,
      "learning_rate": 2.3301386918255007e-06,
      "loss": 0.6796,
      "step": 23910
    },
    {
      "epoch": 0.6889586814959949,
      "grad_norm": 0.7379085421562195,
      "learning_rate": 2.3297441836598384e-06,
      "loss": 0.7134,
      "step": 23911
    },
    {
      "epoch": 0.6889874949576442,
      "grad_norm": 0.7460336089134216,
      "learning_rate": 2.329349698749435e-06,
      "loss": 0.714,
      "step": 23912
    },
    {
      "epoch": 0.6890163084192935,
      "grad_norm": 0.7309703826904297,
      "learning_rate": 2.328955237097727e-06,
      "loss": 0.7028,
      "step": 23913
    },
    {
      "epoch": 0.6890451218809428,
      "grad_norm": 0.7211859226226807,
      "learning_rate": 2.328560798708147e-06,
      "loss": 0.7078,
      "step": 23914
    },
    {
      "epoch": 0.6890739353425921,
      "grad_norm": 0.744367778301239,
      "learning_rate": 2.3281663835841344e-06,
      "loss": 0.6812,
      "step": 23915
    },
    {
      "epoch": 0.6891027488042414,
      "grad_norm": 0.7570493221282959,
      "learning_rate": 2.3277719917291197e-06,
      "loss": 0.6761,
      "step": 23916
    },
    {
      "epoch": 0.6891315622658907,
      "grad_norm": 0.7229127287864685,
      "learning_rate": 2.3273776231465415e-06,
      "loss": 0.688,
      "step": 23917
    },
    {
      "epoch": 0.6891603757275399,
      "grad_norm": 0.742103099822998,
      "learning_rate": 2.3269832778398286e-06,
      "loss": 0.7064,
      "step": 23918
    },
    {
      "epoch": 0.6891891891891891,
      "grad_norm": 0.7260946035385132,
      "learning_rate": 2.3265889558124238e-06,
      "loss": 0.7042,
      "step": 23919
    },
    {
      "epoch": 0.6892180026508384,
      "grad_norm": 0.7351951599121094,
      "learning_rate": 2.3261946570677544e-06,
      "loss": 0.6931,
      "step": 23920
    },
    {
      "epoch": 0.6892468161124877,
      "grad_norm": 0.7263466715812683,
      "learning_rate": 2.3258003816092583e-06,
      "loss": 0.6929,
      "step": 23921
    },
    {
      "epoch": 0.689275629574137,
      "grad_norm": 0.7362737655639648,
      "learning_rate": 2.3254061294403664e-06,
      "loss": 0.6876,
      "step": 23922
    },
    {
      "epoch": 0.6893044430357863,
      "grad_norm": 0.706195056438446,
      "learning_rate": 2.325011900564515e-06,
      "loss": 0.6786,
      "step": 23923
    },
    {
      "epoch": 0.6893332564974356,
      "grad_norm": 0.7121372222900391,
      "learning_rate": 2.324617694985134e-06,
      "loss": 0.6876,
      "step": 23924
    },
    {
      "epoch": 0.6893620699590849,
      "grad_norm": 0.7242713570594788,
      "learning_rate": 2.3242235127056605e-06,
      "loss": 0.705,
      "step": 23925
    },
    {
      "epoch": 0.6893908834207342,
      "grad_norm": 0.7316738367080688,
      "learning_rate": 2.323829353729524e-06,
      "loss": 0.7125,
      "step": 23926
    },
    {
      "epoch": 0.6894196968823835,
      "grad_norm": 0.7600468397140503,
      "learning_rate": 2.32343521806016e-06,
      "loss": 0.6885,
      "step": 23927
    },
    {
      "epoch": 0.6894485103440328,
      "grad_norm": 0.7065054178237915,
      "learning_rate": 2.3230411057009977e-06,
      "loss": 0.6726,
      "step": 23928
    },
    {
      "epoch": 0.689477323805682,
      "grad_norm": 0.7188458442687988,
      "learning_rate": 2.3226470166554717e-06,
      "loss": 0.6847,
      "step": 23929
    },
    {
      "epoch": 0.6895061372673313,
      "grad_norm": 0.7254883050918579,
      "learning_rate": 2.322252950927013e-06,
      "loss": 0.7046,
      "step": 23930
    },
    {
      "epoch": 0.6895349507289806,
      "grad_norm": 0.7228883504867554,
      "learning_rate": 2.321858908519057e-06,
      "loss": 0.6867,
      "step": 23931
    },
    {
      "epoch": 0.6895637641906298,
      "grad_norm": 0.7256489992141724,
      "learning_rate": 2.3214648894350308e-06,
      "loss": 0.6944,
      "step": 23932
    },
    {
      "epoch": 0.6895925776522791,
      "grad_norm": 0.7175007462501526,
      "learning_rate": 2.321070893678368e-06,
      "loss": 0.6765,
      "step": 23933
    },
    {
      "epoch": 0.6896213911139284,
      "grad_norm": 0.7050480246543884,
      "learning_rate": 2.320676921252501e-06,
      "loss": 0.6947,
      "step": 23934
    },
    {
      "epoch": 0.6896502045755777,
      "grad_norm": 0.7027987837791443,
      "learning_rate": 2.3202829721608584e-06,
      "loss": 0.7002,
      "step": 23935
    },
    {
      "epoch": 0.689679018037227,
      "grad_norm": 0.7363063097000122,
      "learning_rate": 2.3198890464068743e-06,
      "loss": 0.6897,
      "step": 23936
    },
    {
      "epoch": 0.6897078314988763,
      "grad_norm": 0.7595824599266052,
      "learning_rate": 2.319495143993975e-06,
      "loss": 0.713,
      "step": 23937
    },
    {
      "epoch": 0.6897366449605256,
      "grad_norm": 0.702953577041626,
      "learning_rate": 2.3191012649255956e-06,
      "loss": 0.6851,
      "step": 23938
    },
    {
      "epoch": 0.6897654584221748,
      "grad_norm": 0.7227708101272583,
      "learning_rate": 2.3187074092051624e-06,
      "loss": 0.6833,
      "step": 23939
    },
    {
      "epoch": 0.6897942718838241,
      "grad_norm": 0.7110292315483093,
      "learning_rate": 2.318313576836109e-06,
      "loss": 0.697,
      "step": 23940
    },
    {
      "epoch": 0.6898230853454734,
      "grad_norm": 0.7133133411407471,
      "learning_rate": 2.3179197678218605e-06,
      "loss": 0.6977,
      "step": 23941
    },
    {
      "epoch": 0.6898518988071227,
      "grad_norm": 0.7053151726722717,
      "learning_rate": 2.317525982165853e-06,
      "loss": 0.7072,
      "step": 23942
    },
    {
      "epoch": 0.689880712268772,
      "grad_norm": 0.7151147127151489,
      "learning_rate": 2.3171322198715112e-06,
      "loss": 0.6746,
      "step": 23943
    },
    {
      "epoch": 0.6899095257304213,
      "grad_norm": 0.7101579904556274,
      "learning_rate": 2.316738480942268e-06,
      "loss": 0.6994,
      "step": 23944
    },
    {
      "epoch": 0.6899383391920705,
      "grad_norm": 0.7307154536247253,
      "learning_rate": 2.3163447653815475e-06,
      "loss": 0.6836,
      "step": 23945
    },
    {
      "epoch": 0.6899671526537198,
      "grad_norm": 0.734239399433136,
      "learning_rate": 2.315951073192784e-06,
      "loss": 0.7027,
      "step": 23946
    },
    {
      "epoch": 0.6899959661153691,
      "grad_norm": 0.735342264175415,
      "learning_rate": 2.3155574043794017e-06,
      "loss": 0.7195,
      "step": 23947
    },
    {
      "epoch": 0.6900247795770184,
      "grad_norm": 0.7085250616073608,
      "learning_rate": 2.315163758944832e-06,
      "loss": 0.6901,
      "step": 23948
    },
    {
      "epoch": 0.6900535930386676,
      "grad_norm": 0.7099911570549011,
      "learning_rate": 2.314770136892501e-06,
      "loss": 0.7131,
      "step": 23949
    },
    {
      "epoch": 0.6900824065003169,
      "grad_norm": 0.7109401822090149,
      "learning_rate": 2.314376538225838e-06,
      "loss": 0.6995,
      "step": 23950
    },
    {
      "epoch": 0.6901112199619662,
      "grad_norm": 0.7155207991600037,
      "learning_rate": 2.3139829629482725e-06,
      "loss": 0.704,
      "step": 23951
    },
    {
      "epoch": 0.6901400334236155,
      "grad_norm": 0.7220737934112549,
      "learning_rate": 2.3135894110632285e-06,
      "loss": 0.6984,
      "step": 23952
    },
    {
      "epoch": 0.6901688468852648,
      "grad_norm": 0.7331327795982361,
      "learning_rate": 2.3131958825741344e-06,
      "loss": 0.6978,
      "step": 23953
    },
    {
      "epoch": 0.6901976603469141,
      "grad_norm": 0.7429999113082886,
      "learning_rate": 2.3128023774844196e-06,
      "loss": 0.6979,
      "step": 23954
    },
    {
      "epoch": 0.6902264738085634,
      "grad_norm": 0.7137117981910706,
      "learning_rate": 2.312408895797511e-06,
      "loss": 0.6939,
      "step": 23955
    },
    {
      "epoch": 0.6902552872702127,
      "grad_norm": 0.7142159938812256,
      "learning_rate": 2.312015437516833e-06,
      "loss": 0.6994,
      "step": 23956
    },
    {
      "epoch": 0.690284100731862,
      "grad_norm": 0.8506179451942444,
      "learning_rate": 2.3116220026458148e-06,
      "loss": 0.6774,
      "step": 23957
    },
    {
      "epoch": 0.6903129141935113,
      "grad_norm": 0.7457646727561951,
      "learning_rate": 2.31122859118788e-06,
      "loss": 0.6765,
      "step": 23958
    },
    {
      "epoch": 0.6903417276551604,
      "grad_norm": 0.7381770014762878,
      "learning_rate": 2.3108352031464582e-06,
      "loss": 0.6978,
      "step": 23959
    },
    {
      "epoch": 0.6903705411168097,
      "grad_norm": 0.719132661819458,
      "learning_rate": 2.3104418385249714e-06,
      "loss": 0.7032,
      "step": 23960
    },
    {
      "epoch": 0.690399354578459,
      "grad_norm": 0.7084947228431702,
      "learning_rate": 2.31004849732685e-06,
      "loss": 0.6848,
      "step": 23961
    },
    {
      "epoch": 0.6904281680401083,
      "grad_norm": 0.7443594336509705,
      "learning_rate": 2.3096551795555154e-06,
      "loss": 0.701,
      "step": 23962
    },
    {
      "epoch": 0.6904569815017576,
      "grad_norm": 0.7429308891296387,
      "learning_rate": 2.309261885214396e-06,
      "loss": 0.6915,
      "step": 23963
    },
    {
      "epoch": 0.6904857949634069,
      "grad_norm": 0.7407526969909668,
      "learning_rate": 2.308868614306915e-06,
      "loss": 0.713,
      "step": 23964
    },
    {
      "epoch": 0.6905146084250562,
      "grad_norm": 0.6948366165161133,
      "learning_rate": 2.3084753668364973e-06,
      "loss": 0.669,
      "step": 23965
    },
    {
      "epoch": 0.6905434218867055,
      "grad_norm": 0.7033445239067078,
      "learning_rate": 2.308082142806569e-06,
      "loss": 0.6678,
      "step": 23966
    },
    {
      "epoch": 0.6905722353483548,
      "grad_norm": 0.7037070393562317,
      "learning_rate": 2.3076889422205562e-06,
      "loss": 0.6685,
      "step": 23967
    },
    {
      "epoch": 0.6906010488100041,
      "grad_norm": 0.7137101888656616,
      "learning_rate": 2.3072957650818794e-06,
      "loss": 0.6779,
      "step": 23968
    },
    {
      "epoch": 0.6906298622716533,
      "grad_norm": 0.6930518746376038,
      "learning_rate": 2.306902611393967e-06,
      "loss": 0.6781,
      "step": 23969
    },
    {
      "epoch": 0.6906586757333026,
      "grad_norm": 0.7231758236885071,
      "learning_rate": 2.3065094811602386e-06,
      "loss": 0.704,
      "step": 23970
    },
    {
      "epoch": 0.6906874891949519,
      "grad_norm": 0.7475666403770447,
      "learning_rate": 2.30611637438412e-06,
      "loss": 0.7128,
      "step": 23971
    },
    {
      "epoch": 0.6907163026566011,
      "grad_norm": 0.7348684072494507,
      "learning_rate": 2.3057232910690363e-06,
      "loss": 0.699,
      "step": 23972
    },
    {
      "epoch": 0.6907451161182504,
      "grad_norm": 0.697600245475769,
      "learning_rate": 2.3053302312184086e-06,
      "loss": 0.6954,
      "step": 23973
    },
    {
      "epoch": 0.6907739295798997,
      "grad_norm": 0.7242128849029541,
      "learning_rate": 2.3049371948356617e-06,
      "loss": 0.7013,
      "step": 23974
    },
    {
      "epoch": 0.690802743041549,
      "grad_norm": 0.7145766019821167,
      "learning_rate": 2.304544181924217e-06,
      "loss": 0.7002,
      "step": 23975
    },
    {
      "epoch": 0.6908315565031983,
      "grad_norm": 0.7223550081253052,
      "learning_rate": 2.304151192487499e-06,
      "loss": 0.6994,
      "step": 23976
    },
    {
      "epoch": 0.6908603699648476,
      "grad_norm": 0.726998507976532,
      "learning_rate": 2.303758226528926e-06,
      "loss": 0.6788,
      "step": 23977
    },
    {
      "epoch": 0.6908891834264969,
      "grad_norm": 0.7143818140029907,
      "learning_rate": 2.303365284051927e-06,
      "loss": 0.686,
      "step": 23978
    },
    {
      "epoch": 0.6909179968881461,
      "grad_norm": 0.7629100680351257,
      "learning_rate": 2.302972365059919e-06,
      "loss": 0.7133,
      "step": 23979
    },
    {
      "epoch": 0.6909468103497954,
      "grad_norm": 0.7197325825691223,
      "learning_rate": 2.3025794695563276e-06,
      "loss": 0.6777,
      "step": 23980
    },
    {
      "epoch": 0.6909756238114447,
      "grad_norm": 0.7270020842552185,
      "learning_rate": 2.3021865975445708e-06,
      "loss": 0.6919,
      "step": 23981
    },
    {
      "epoch": 0.691004437273094,
      "grad_norm": 0.7319133877754211,
      "learning_rate": 2.3017937490280734e-06,
      "loss": 0.7167,
      "step": 23982
    },
    {
      "epoch": 0.6910332507347433,
      "grad_norm": 0.7389987707138062,
      "learning_rate": 2.3014009240102534e-06,
      "loss": 0.6992,
      "step": 23983
    },
    {
      "epoch": 0.6910620641963926,
      "grad_norm": 0.7125537395477295,
      "learning_rate": 2.3010081224945365e-06,
      "loss": 0.6808,
      "step": 23984
    },
    {
      "epoch": 0.6910908776580419,
      "grad_norm": 0.7553341388702393,
      "learning_rate": 2.3006153444843383e-06,
      "loss": 0.7306,
      "step": 23985
    },
    {
      "epoch": 0.6911196911196911,
      "grad_norm": 0.7217632532119751,
      "learning_rate": 2.300222589983084e-06,
      "loss": 0.6906,
      "step": 23986
    },
    {
      "epoch": 0.6911485045813404,
      "grad_norm": 0.7217146754264832,
      "learning_rate": 2.2998298589941898e-06,
      "loss": 0.6991,
      "step": 23987
    },
    {
      "epoch": 0.6911773180429897,
      "grad_norm": 0.7117321491241455,
      "learning_rate": 2.2994371515210787e-06,
      "loss": 0.6881,
      "step": 23988
    },
    {
      "epoch": 0.6912061315046389,
      "grad_norm": 0.7180972099304199,
      "learning_rate": 2.2990444675671707e-06,
      "loss": 0.7193,
      "step": 23989
    },
    {
      "epoch": 0.6912349449662882,
      "grad_norm": 0.7193886041641235,
      "learning_rate": 2.2986518071358873e-06,
      "loss": 0.6833,
      "step": 23990
    },
    {
      "epoch": 0.6912637584279375,
      "grad_norm": 0.7380875945091248,
      "learning_rate": 2.298259170230644e-06,
      "loss": 0.702,
      "step": 23991
    },
    {
      "epoch": 0.6912925718895868,
      "grad_norm": 0.7214244604110718,
      "learning_rate": 2.2978665568548624e-06,
      "loss": 0.6715,
      "step": 23992
    },
    {
      "epoch": 0.6913213853512361,
      "grad_norm": 0.716084361076355,
      "learning_rate": 2.2974739670119645e-06,
      "loss": 0.7101,
      "step": 23993
    },
    {
      "epoch": 0.6913501988128854,
      "grad_norm": 0.7249174118041992,
      "learning_rate": 2.297081400705365e-06,
      "loss": 0.7184,
      "step": 23994
    },
    {
      "epoch": 0.6913790122745347,
      "grad_norm": 0.7175437211990356,
      "learning_rate": 2.2966888579384855e-06,
      "loss": 0.67,
      "step": 23995
    },
    {
      "epoch": 0.691407825736184,
      "grad_norm": 0.7146183252334595,
      "learning_rate": 2.2962963387147425e-06,
      "loss": 0.6731,
      "step": 23996
    },
    {
      "epoch": 0.6914366391978333,
      "grad_norm": 0.7304428219795227,
      "learning_rate": 2.2959038430375574e-06,
      "loss": 0.7053,
      "step": 23997
    },
    {
      "epoch": 0.6914654526594826,
      "grad_norm": 0.7427365779876709,
      "learning_rate": 2.295511370910345e-06,
      "loss": 0.7035,
      "step": 23998
    },
    {
      "epoch": 0.6914942661211317,
      "grad_norm": 0.7216109037399292,
      "learning_rate": 2.295118922336527e-06,
      "loss": 0.6915,
      "step": 23999
    },
    {
      "epoch": 0.691523079582781,
      "grad_norm": 0.7147195339202881,
      "learning_rate": 2.294726497319517e-06,
      "loss": 0.6938,
      "step": 24000
    },
    {
      "epoch": 0.6915518930444303,
      "grad_norm": 0.7227882146835327,
      "learning_rate": 2.294334095862735e-06,
      "loss": 0.6813,
      "step": 24001
    },
    {
      "epoch": 0.6915807065060796,
      "grad_norm": 0.7273170351982117,
      "learning_rate": 2.293941717969599e-06,
      "loss": 0.6713,
      "step": 24002
    },
    {
      "epoch": 0.6916095199677289,
      "grad_norm": 0.7428483366966248,
      "learning_rate": 2.2935493636435267e-06,
      "loss": 0.6902,
      "step": 24003
    },
    {
      "epoch": 0.6916383334293782,
      "grad_norm": 0.6993556618690491,
      "learning_rate": 2.2931570328879327e-06,
      "loss": 0.6853,
      "step": 24004
    },
    {
      "epoch": 0.6916671468910275,
      "grad_norm": 0.7172604203224182,
      "learning_rate": 2.2927647257062367e-06,
      "loss": 0.6992,
      "step": 24005
    },
    {
      "epoch": 0.6916959603526768,
      "grad_norm": 0.6939291954040527,
      "learning_rate": 2.2923724421018515e-06,
      "loss": 0.6855,
      "step": 24006
    },
    {
      "epoch": 0.6917247738143261,
      "grad_norm": 0.7259841561317444,
      "learning_rate": 2.2919801820781982e-06,
      "loss": 0.6927,
      "step": 24007
    },
    {
      "epoch": 0.6917535872759754,
      "grad_norm": 0.7250761389732361,
      "learning_rate": 2.291587945638688e-06,
      "loss": 0.6901,
      "step": 24008
    },
    {
      "epoch": 0.6917824007376246,
      "grad_norm": 0.7241511940956116,
      "learning_rate": 2.2911957327867402e-06,
      "loss": 0.6937,
      "step": 24009
    },
    {
      "epoch": 0.6918112141992739,
      "grad_norm": 0.7051132917404175,
      "learning_rate": 2.2908035435257707e-06,
      "loss": 0.6896,
      "step": 24010
    },
    {
      "epoch": 0.6918400276609232,
      "grad_norm": 0.6997987627983093,
      "learning_rate": 2.290411377859193e-06,
      "loss": 0.6817,
      "step": 24011
    },
    {
      "epoch": 0.6918688411225725,
      "grad_norm": 0.721137285232544,
      "learning_rate": 2.290019235790425e-06,
      "loss": 0.7132,
      "step": 24012
    },
    {
      "epoch": 0.6918976545842217,
      "grad_norm": 0.7022355198860168,
      "learning_rate": 2.2896271173228784e-06,
      "loss": 0.7111,
      "step": 24013
    },
    {
      "epoch": 0.691926468045871,
      "grad_norm": 0.7246968150138855,
      "learning_rate": 2.289235022459971e-06,
      "loss": 0.6975,
      "step": 24014
    },
    {
      "epoch": 0.6919552815075203,
      "grad_norm": 0.7118400931358337,
      "learning_rate": 2.288842951205116e-06,
      "loss": 0.704,
      "step": 24015
    },
    {
      "epoch": 0.6919840949691696,
      "grad_norm": 0.7097131609916687,
      "learning_rate": 2.2884509035617307e-06,
      "loss": 0.6835,
      "step": 24016
    },
    {
      "epoch": 0.6920129084308189,
      "grad_norm": 0.7248460054397583,
      "learning_rate": 2.288058879533226e-06,
      "loss": 0.6919,
      "step": 24017
    },
    {
      "epoch": 0.6920417218924682,
      "grad_norm": 0.7265620827674866,
      "learning_rate": 2.2876668791230196e-06,
      "loss": 0.7261,
      "step": 24018
    },
    {
      "epoch": 0.6920705353541174,
      "grad_norm": 0.7227942943572998,
      "learning_rate": 2.2872749023345207e-06,
      "loss": 0.703,
      "step": 24019
    },
    {
      "epoch": 0.6920993488157667,
      "grad_norm": 0.7312762141227722,
      "learning_rate": 2.286882949171148e-06,
      "loss": 0.685,
      "step": 24020
    },
    {
      "epoch": 0.692128162277416,
      "grad_norm": 0.7205495834350586,
      "learning_rate": 2.2864910196363104e-06,
      "loss": 0.7006,
      "step": 24021
    },
    {
      "epoch": 0.6921569757390653,
      "grad_norm": 0.7114118933677673,
      "learning_rate": 2.286099113733426e-06,
      "loss": 0.6825,
      "step": 24022
    },
    {
      "epoch": 0.6921857892007146,
      "grad_norm": 0.7093347311019897,
      "learning_rate": 2.285707231465903e-06,
      "loss": 0.6996,
      "step": 24023
    },
    {
      "epoch": 0.6922146026623639,
      "grad_norm": 0.7360148429870605,
      "learning_rate": 2.285315372837159e-06,
      "loss": 0.6981,
      "step": 24024
    },
    {
      "epoch": 0.6922434161240132,
      "grad_norm": 0.7346146106719971,
      "learning_rate": 2.2849235378506003e-06,
      "loss": 0.7032,
      "step": 24025
    },
    {
      "epoch": 0.6922722295856624,
      "grad_norm": 0.7145547866821289,
      "learning_rate": 2.2845317265096474e-06,
      "loss": 0.6865,
      "step": 24026
    },
    {
      "epoch": 0.6923010430473117,
      "grad_norm": 0.7168949246406555,
      "learning_rate": 2.284139938817707e-06,
      "loss": 0.7051,
      "step": 24027
    },
    {
      "epoch": 0.692329856508961,
      "grad_norm": 0.7094737887382507,
      "learning_rate": 2.283748174778195e-06,
      "loss": 0.6773,
      "step": 24028
    },
    {
      "epoch": 0.6923586699706102,
      "grad_norm": 0.7107676863670349,
      "learning_rate": 2.2833564343945185e-06,
      "loss": 0.6789,
      "step": 24029
    },
    {
      "epoch": 0.6923874834322595,
      "grad_norm": 0.7245588898658752,
      "learning_rate": 2.282964717670092e-06,
      "loss": 0.7206,
      "step": 24030
    },
    {
      "epoch": 0.6924162968939088,
      "grad_norm": 0.7163394093513489,
      "learning_rate": 2.282573024608329e-06,
      "loss": 0.6907,
      "step": 24031
    },
    {
      "epoch": 0.6924451103555581,
      "grad_norm": 0.7060021162033081,
      "learning_rate": 2.2821813552126364e-06,
      "loss": 0.6885,
      "step": 24032
    },
    {
      "epoch": 0.6924739238172074,
      "grad_norm": 0.7265362739562988,
      "learning_rate": 2.2817897094864296e-06,
      "loss": 0.677,
      "step": 24033
    },
    {
      "epoch": 0.6925027372788567,
      "grad_norm": 0.7317925095558167,
      "learning_rate": 2.281398087433115e-06,
      "loss": 0.7062,
      "step": 24034
    },
    {
      "epoch": 0.692531550740506,
      "grad_norm": 0.7091281414031982,
      "learning_rate": 2.2810064890561066e-06,
      "loss": 0.6898,
      "step": 24035
    },
    {
      "epoch": 0.6925603642021553,
      "grad_norm": 0.7047039866447449,
      "learning_rate": 2.280614914358813e-06,
      "loss": 0.7054,
      "step": 24036
    },
    {
      "epoch": 0.6925891776638046,
      "grad_norm": 0.7107895612716675,
      "learning_rate": 2.2802233633446445e-06,
      "loss": 0.7036,
      "step": 24037
    },
    {
      "epoch": 0.6926179911254539,
      "grad_norm": 0.706443727016449,
      "learning_rate": 2.279831836017012e-06,
      "loss": 0.6753,
      "step": 24038
    },
    {
      "epoch": 0.6926468045871031,
      "grad_norm": 0.7447088360786438,
      "learning_rate": 2.2794403323793267e-06,
      "loss": 0.7124,
      "step": 24039
    },
    {
      "epoch": 0.6926756180487523,
      "grad_norm": 0.6968512535095215,
      "learning_rate": 2.2790488524349953e-06,
      "loss": 0.6646,
      "step": 24040
    },
    {
      "epoch": 0.6927044315104016,
      "grad_norm": 0.7211509346961975,
      "learning_rate": 2.2786573961874293e-06,
      "loss": 0.7165,
      "step": 24041
    },
    {
      "epoch": 0.6927332449720509,
      "grad_norm": 0.7185069918632507,
      "learning_rate": 2.2782659636400356e-06,
      "loss": 0.6799,
      "step": 24042
    },
    {
      "epoch": 0.6927620584337002,
      "grad_norm": 0.7223743796348572,
      "learning_rate": 2.2778745547962267e-06,
      "loss": 0.6892,
      "step": 24043
    },
    {
      "epoch": 0.6927908718953495,
      "grad_norm": 0.7125169038772583,
      "learning_rate": 2.2774831696594076e-06,
      "loss": 0.6875,
      "step": 24044
    },
    {
      "epoch": 0.6928196853569988,
      "grad_norm": 0.7130874395370483,
      "learning_rate": 2.2770918082329897e-06,
      "loss": 0.6843,
      "step": 24045
    },
    {
      "epoch": 0.6928484988186481,
      "grad_norm": 0.7301998138427734,
      "learning_rate": 2.276700470520379e-06,
      "loss": 0.7272,
      "step": 24046
    },
    {
      "epoch": 0.6928773122802974,
      "grad_norm": 0.7942342758178711,
      "learning_rate": 2.2763091565249846e-06,
      "loss": 0.7006,
      "step": 24047
    },
    {
      "epoch": 0.6929061257419467,
      "grad_norm": 0.7453740835189819,
      "learning_rate": 2.275917866250217e-06,
      "loss": 0.6966,
      "step": 24048
    },
    {
      "epoch": 0.692934939203596,
      "grad_norm": 0.7196796536445618,
      "learning_rate": 2.275526599699479e-06,
      "loss": 0.6764,
      "step": 24049
    },
    {
      "epoch": 0.6929637526652452,
      "grad_norm": 0.7077918648719788,
      "learning_rate": 2.2751353568761814e-06,
      "loss": 0.6939,
      "step": 24050
    },
    {
      "epoch": 0.6929925661268945,
      "grad_norm": 0.7205641865730286,
      "learning_rate": 2.274744137783731e-06,
      "loss": 0.7097,
      "step": 24051
    },
    {
      "epoch": 0.6930213795885438,
      "grad_norm": 0.7293069958686829,
      "learning_rate": 2.2743529424255366e-06,
      "loss": 0.7072,
      "step": 24052
    },
    {
      "epoch": 0.693050193050193,
      "grad_norm": 0.7115052342414856,
      "learning_rate": 2.273961770805002e-06,
      "loss": 0.6952,
      "step": 24053
    },
    {
      "epoch": 0.6930790065118423,
      "grad_norm": 0.7321434020996094,
      "learning_rate": 2.2735706229255365e-06,
      "loss": 0.693,
      "step": 24054
    },
    {
      "epoch": 0.6931078199734916,
      "grad_norm": 0.7141863703727722,
      "learning_rate": 2.2731794987905444e-06,
      "loss": 0.6751,
      "step": 24055
    },
    {
      "epoch": 0.6931366334351409,
      "grad_norm": 0.7369338274002075,
      "learning_rate": 2.2727883984034343e-06,
      "loss": 0.702,
      "step": 24056
    },
    {
      "epoch": 0.6931654468967902,
      "grad_norm": 0.7238418459892273,
      "learning_rate": 2.272397321767609e-06,
      "loss": 0.6795,
      "step": 24057
    },
    {
      "epoch": 0.6931942603584395,
      "grad_norm": 0.7087958455085754,
      "learning_rate": 2.272006268886479e-06,
      "loss": 0.6828,
      "step": 24058
    },
    {
      "epoch": 0.6932230738200887,
      "grad_norm": 0.7391760945320129,
      "learning_rate": 2.271615239763445e-06,
      "loss": 0.7084,
      "step": 24059
    },
    {
      "epoch": 0.693251887281738,
      "grad_norm": 0.7344071269035339,
      "learning_rate": 2.271224234401917e-06,
      "loss": 0.6975,
      "step": 24060
    },
    {
      "epoch": 0.6932807007433873,
      "grad_norm": 0.7169910669326782,
      "learning_rate": 2.270833252805294e-06,
      "loss": 0.6661,
      "step": 24061
    },
    {
      "epoch": 0.6933095142050366,
      "grad_norm": 0.7146108746528625,
      "learning_rate": 2.2704422949769896e-06,
      "loss": 0.6851,
      "step": 24062
    },
    {
      "epoch": 0.6933383276666859,
      "grad_norm": 0.7303915619850159,
      "learning_rate": 2.2700513609204015e-06,
      "loss": 0.7279,
      "step": 24063
    },
    {
      "epoch": 0.6933671411283352,
      "grad_norm": 0.7203507423400879,
      "learning_rate": 2.2696604506389398e-06,
      "loss": 0.7029,
      "step": 24064
    },
    {
      "epoch": 0.6933959545899845,
      "grad_norm": 0.7381927371025085,
      "learning_rate": 2.269269564136004e-06,
      "loss": 0.6916,
      "step": 24065
    },
    {
      "epoch": 0.6934247680516338,
      "grad_norm": 0.7246184349060059,
      "learning_rate": 2.2688787014150027e-06,
      "loss": 0.6717,
      "step": 24066
    },
    {
      "epoch": 0.693453581513283,
      "grad_norm": 0.7058734893798828,
      "learning_rate": 2.2684878624793353e-06,
      "loss": 0.6988,
      "step": 24067
    },
    {
      "epoch": 0.6934823949749322,
      "grad_norm": 0.7254440188407898,
      "learning_rate": 2.2680970473324087e-06,
      "loss": 0.7322,
      "step": 24068
    },
    {
      "epoch": 0.6935112084365815,
      "grad_norm": 0.7009137272834778,
      "learning_rate": 2.2677062559776275e-06,
      "loss": 0.706,
      "step": 24069
    },
    {
      "epoch": 0.6935400218982308,
      "grad_norm": 0.7225383520126343,
      "learning_rate": 2.2673154884183917e-06,
      "loss": 0.6868,
      "step": 24070
    },
    {
      "epoch": 0.6935688353598801,
      "grad_norm": 0.7278606295585632,
      "learning_rate": 2.2669247446581084e-06,
      "loss": 0.6968,
      "step": 24071
    },
    {
      "epoch": 0.6935976488215294,
      "grad_norm": 0.7329275012016296,
      "learning_rate": 2.2665340247001765e-06,
      "loss": 0.6775,
      "step": 24072
    },
    {
      "epoch": 0.6936264622831787,
      "grad_norm": 0.7187570333480835,
      "learning_rate": 2.2661433285480005e-06,
      "loss": 0.6913,
      "step": 24073
    },
    {
      "epoch": 0.693655275744828,
      "grad_norm": 0.689945638179779,
      "learning_rate": 2.2657526562049834e-06,
      "loss": 0.6998,
      "step": 24074
    },
    {
      "epoch": 0.6936840892064773,
      "grad_norm": 0.7212340831756592,
      "learning_rate": 2.26536200767453e-06,
      "loss": 0.6915,
      "step": 24075
    },
    {
      "epoch": 0.6937129026681266,
      "grad_norm": 0.7388603091239929,
      "learning_rate": 2.2649713829600374e-06,
      "loss": 0.7164,
      "step": 24076
    },
    {
      "epoch": 0.6937417161297759,
      "grad_norm": 0.7505958080291748,
      "learning_rate": 2.2645807820649114e-06,
      "loss": 0.7208,
      "step": 24077
    },
    {
      "epoch": 0.6937705295914252,
      "grad_norm": 0.7022356986999512,
      "learning_rate": 2.2641902049925514e-06,
      "loss": 0.6742,
      "step": 24078
    },
    {
      "epoch": 0.6937993430530744,
      "grad_norm": 0.7008862495422363,
      "learning_rate": 2.2637996517463613e-06,
      "loss": 0.7133,
      "step": 24079
    },
    {
      "epoch": 0.6938281565147236,
      "grad_norm": 0.7740370035171509,
      "learning_rate": 2.263409122329739e-06,
      "loss": 0.6791,
      "step": 24080
    },
    {
      "epoch": 0.6938569699763729,
      "grad_norm": 0.7283598184585571,
      "learning_rate": 2.2630186167460894e-06,
      "loss": 0.7071,
      "step": 24081
    },
    {
      "epoch": 0.6938857834380222,
      "grad_norm": 0.707648754119873,
      "learning_rate": 2.2626281349988106e-06,
      "loss": 0.6787,
      "step": 24082
    },
    {
      "epoch": 0.6939145968996715,
      "grad_norm": 0.7294678688049316,
      "learning_rate": 2.262237677091306e-06,
      "loss": 0.7131,
      "step": 24083
    },
    {
      "epoch": 0.6939434103613208,
      "grad_norm": 0.7114375233650208,
      "learning_rate": 2.2618472430269717e-06,
      "loss": 0.6919,
      "step": 24084
    },
    {
      "epoch": 0.6939722238229701,
      "grad_norm": 0.7199472188949585,
      "learning_rate": 2.2614568328092116e-06,
      "loss": 0.6805,
      "step": 24085
    },
    {
      "epoch": 0.6940010372846194,
      "grad_norm": 0.7254707217216492,
      "learning_rate": 2.261066446441425e-06,
      "loss": 0.7044,
      "step": 24086
    },
    {
      "epoch": 0.6940298507462687,
      "grad_norm": 0.7210438251495361,
      "learning_rate": 2.260676083927013e-06,
      "loss": 0.7075,
      "step": 24087
    },
    {
      "epoch": 0.694058664207918,
      "grad_norm": 0.7230954170227051,
      "learning_rate": 2.2602857452693726e-06,
      "loss": 0.7004,
      "step": 24088
    },
    {
      "epoch": 0.6940874776695672,
      "grad_norm": 0.7470828294754028,
      "learning_rate": 2.2598954304719044e-06,
      "loss": 0.6981,
      "step": 24089
    },
    {
      "epoch": 0.6941162911312165,
      "grad_norm": 0.7494809031486511,
      "learning_rate": 2.25950513953801e-06,
      "loss": 0.6846,
      "step": 24090
    },
    {
      "epoch": 0.6941451045928658,
      "grad_norm": 0.6986843943595886,
      "learning_rate": 2.2591148724710844e-06,
      "loss": 0.6876,
      "step": 24091
    },
    {
      "epoch": 0.6941739180545151,
      "grad_norm": 0.7385454177856445,
      "learning_rate": 2.25872462927453e-06,
      "loss": 0.7135,
      "step": 24092
    },
    {
      "epoch": 0.6942027315161644,
      "grad_norm": 0.7093808054924011,
      "learning_rate": 2.2583344099517426e-06,
      "loss": 0.6893,
      "step": 24093
    },
    {
      "epoch": 0.6942315449778136,
      "grad_norm": 0.7158203721046448,
      "learning_rate": 2.257944214506123e-06,
      "loss": 0.6775,
      "step": 24094
    },
    {
      "epoch": 0.6942603584394629,
      "grad_norm": 0.7028533816337585,
      "learning_rate": 2.257554042941067e-06,
      "loss": 0.6547,
      "step": 24095
    },
    {
      "epoch": 0.6942891719011122,
      "grad_norm": 0.7515764832496643,
      "learning_rate": 2.257163895259975e-06,
      "loss": 0.7315,
      "step": 24096
    },
    {
      "epoch": 0.6943179853627615,
      "grad_norm": 0.7301256060600281,
      "learning_rate": 2.2567737714662404e-06,
      "loss": 0.7109,
      "step": 24097
    },
    {
      "epoch": 0.6943467988244107,
      "grad_norm": 0.7267493009567261,
      "learning_rate": 2.2563836715632674e-06,
      "loss": 0.7272,
      "step": 24098
    },
    {
      "epoch": 0.69437561228606,
      "grad_norm": 0.7310618162155151,
      "learning_rate": 2.2559935955544485e-06,
      "loss": 0.7043,
      "step": 24099
    },
    {
      "epoch": 0.6944044257477093,
      "grad_norm": 0.7231849431991577,
      "learning_rate": 2.255603543443184e-06,
      "loss": 0.7034,
      "step": 24100
    },
    {
      "epoch": 0.6944332392093586,
      "grad_norm": 0.7740909457206726,
      "learning_rate": 2.2552135152328676e-06,
      "loss": 0.6869,
      "step": 24101
    },
    {
      "epoch": 0.6944620526710079,
      "grad_norm": 0.737349808216095,
      "learning_rate": 2.2548235109268996e-06,
      "loss": 0.7092,
      "step": 24102
    },
    {
      "epoch": 0.6944908661326572,
      "grad_norm": 0.760615885257721,
      "learning_rate": 2.2544335305286725e-06,
      "loss": 0.7081,
      "step": 24103
    },
    {
      "epoch": 0.6945196795943065,
      "grad_norm": 0.7418018579483032,
      "learning_rate": 2.2540435740415866e-06,
      "loss": 0.6935,
      "step": 24104
    },
    {
      "epoch": 0.6945484930559558,
      "grad_norm": 0.7241223454475403,
      "learning_rate": 2.2536536414690347e-06,
      "loss": 0.7045,
      "step": 24105
    },
    {
      "epoch": 0.6945773065176051,
      "grad_norm": 0.7292837500572205,
      "learning_rate": 2.253263732814414e-06,
      "loss": 0.7016,
      "step": 24106
    },
    {
      "epoch": 0.6946061199792543,
      "grad_norm": 0.7074645757675171,
      "learning_rate": 2.2528738480811225e-06,
      "loss": 0.6792,
      "step": 24107
    },
    {
      "epoch": 0.6946349334409035,
      "grad_norm": 0.7276313900947571,
      "learning_rate": 2.252483987272552e-06,
      "loss": 0.7005,
      "step": 24108
    },
    {
      "epoch": 0.6946637469025528,
      "grad_norm": 0.7241643071174622,
      "learning_rate": 2.252094150392099e-06,
      "loss": 0.6777,
      "step": 24109
    },
    {
      "epoch": 0.6946925603642021,
      "grad_norm": 0.7391055822372437,
      "learning_rate": 2.251704337443161e-06,
      "loss": 0.72,
      "step": 24110
    },
    {
      "epoch": 0.6947213738258514,
      "grad_norm": 0.7207005023956299,
      "learning_rate": 2.2513145484291287e-06,
      "loss": 0.7024,
      "step": 24111
    },
    {
      "epoch": 0.6947501872875007,
      "grad_norm": 0.7274391055107117,
      "learning_rate": 2.2509247833533996e-06,
      "loss": 0.7027,
      "step": 24112
    },
    {
      "epoch": 0.69477900074915,
      "grad_norm": 0.707024872303009,
      "learning_rate": 2.2505350422193694e-06,
      "loss": 0.6861,
      "step": 24113
    },
    {
      "epoch": 0.6948078142107993,
      "grad_norm": 0.731296956539154,
      "learning_rate": 2.2501453250304282e-06,
      "loss": 0.679,
      "step": 24114
    },
    {
      "epoch": 0.6948366276724486,
      "grad_norm": 0.7468540072441101,
      "learning_rate": 2.249755631789975e-06,
      "loss": 0.706,
      "step": 24115
    },
    {
      "epoch": 0.6948654411340979,
      "grad_norm": 0.7158873677253723,
      "learning_rate": 2.2493659625013986e-06,
      "loss": 0.6793,
      "step": 24116
    },
    {
      "epoch": 0.6948942545957472,
      "grad_norm": 0.7086527943611145,
      "learning_rate": 2.2489763171680966e-06,
      "loss": 0.6863,
      "step": 24117
    },
    {
      "epoch": 0.6949230680573965,
      "grad_norm": 0.7228140234947205,
      "learning_rate": 2.24858669579346e-06,
      "loss": 0.6819,
      "step": 24118
    },
    {
      "epoch": 0.6949518815190457,
      "grad_norm": 0.7378847599029541,
      "learning_rate": 2.248197098380884e-06,
      "loss": 0.6804,
      "step": 24119
    },
    {
      "epoch": 0.694980694980695,
      "grad_norm": 0.6999446749687195,
      "learning_rate": 2.247807524933759e-06,
      "loss": 0.6689,
      "step": 24120
    },
    {
      "epoch": 0.6950095084423442,
      "grad_norm": 0.7111849784851074,
      "learning_rate": 2.247417975455479e-06,
      "loss": 0.678,
      "step": 24121
    },
    {
      "epoch": 0.6950383219039935,
      "grad_norm": 0.7369922399520874,
      "learning_rate": 2.2470284499494367e-06,
      "loss": 0.6772,
      "step": 24122
    },
    {
      "epoch": 0.6950671353656428,
      "grad_norm": 0.7227230668067932,
      "learning_rate": 2.2466389484190264e-06,
      "loss": 0.6945,
      "step": 24123
    },
    {
      "epoch": 0.6950959488272921,
      "grad_norm": 0.7244171500205994,
      "learning_rate": 2.246249470867637e-06,
      "loss": 0.711,
      "step": 24124
    },
    {
      "epoch": 0.6951247622889414,
      "grad_norm": 0.7178106904029846,
      "learning_rate": 2.245860017298664e-06,
      "loss": 0.6921,
      "step": 24125
    },
    {
      "epoch": 0.6951535757505907,
      "grad_norm": 0.7120188474655151,
      "learning_rate": 2.2454705877154947e-06,
      "loss": 0.7141,
      "step": 24126
    },
    {
      "epoch": 0.69518238921224,
      "grad_norm": 0.7274985313415527,
      "learning_rate": 2.2450811821215234e-06,
      "loss": 0.7021,
      "step": 24127
    },
    {
      "epoch": 0.6952112026738893,
      "grad_norm": 0.7251147031784058,
      "learning_rate": 2.244691800520143e-06,
      "loss": 0.6952,
      "step": 24128
    },
    {
      "epoch": 0.6952400161355385,
      "grad_norm": 0.7274340391159058,
      "learning_rate": 2.244302442914741e-06,
      "loss": 0.693,
      "step": 24129
    },
    {
      "epoch": 0.6952688295971878,
      "grad_norm": 0.7172275185585022,
      "learning_rate": 2.243913109308711e-06,
      "loss": 0.6968,
      "step": 24130
    },
    {
      "epoch": 0.6952976430588371,
      "grad_norm": 0.7260661125183105,
      "learning_rate": 2.243523799705442e-06,
      "loss": 0.6899,
      "step": 24131
    },
    {
      "epoch": 0.6953264565204864,
      "grad_norm": 0.6985270977020264,
      "learning_rate": 2.2431345141083264e-06,
      "loss": 0.686,
      "step": 24132
    },
    {
      "epoch": 0.6953552699821357,
      "grad_norm": 0.6962634921073914,
      "learning_rate": 2.24274525252075e-06,
      "loss": 0.6775,
      "step": 24133
    },
    {
      "epoch": 0.6953840834437849,
      "grad_norm": 0.7236718535423279,
      "learning_rate": 2.2423560149461098e-06,
      "loss": 0.7125,
      "step": 24134
    },
    {
      "epoch": 0.6954128969054342,
      "grad_norm": 0.6991856694221497,
      "learning_rate": 2.2419668013877904e-06,
      "loss": 0.6714,
      "step": 24135
    },
    {
      "epoch": 0.6954417103670835,
      "grad_norm": 0.7257030010223389,
      "learning_rate": 2.241577611849185e-06,
      "loss": 0.661,
      "step": 24136
    },
    {
      "epoch": 0.6954705238287328,
      "grad_norm": 0.7130317687988281,
      "learning_rate": 2.2411884463336785e-06,
      "loss": 0.6858,
      "step": 24137
    },
    {
      "epoch": 0.695499337290382,
      "grad_norm": 0.7266800403594971,
      "learning_rate": 2.2407993048446657e-06,
      "loss": 0.707,
      "step": 24138
    },
    {
      "epoch": 0.6955281507520313,
      "grad_norm": 0.7417541146278381,
      "learning_rate": 2.2404101873855304e-06,
      "loss": 0.6803,
      "step": 24139
    },
    {
      "epoch": 0.6955569642136806,
      "grad_norm": 0.7066653966903687,
      "learning_rate": 2.2400210939596658e-06,
      "loss": 0.6627,
      "step": 24140
    },
    {
      "epoch": 0.6955857776753299,
      "grad_norm": 0.7276080250740051,
      "learning_rate": 2.2396320245704566e-06,
      "loss": 0.6986,
      "step": 24141
    },
    {
      "epoch": 0.6956145911369792,
      "grad_norm": 0.7400978207588196,
      "learning_rate": 2.239242979221295e-06,
      "loss": 0.6842,
      "step": 24142
    },
    {
      "epoch": 0.6956434045986285,
      "grad_norm": 0.716445803642273,
      "learning_rate": 2.2388539579155656e-06,
      "loss": 0.6907,
      "step": 24143
    },
    {
      "epoch": 0.6956722180602778,
      "grad_norm": 0.7107324600219727,
      "learning_rate": 2.238464960656658e-06,
      "loss": 0.6818,
      "step": 24144
    },
    {
      "epoch": 0.6957010315219271,
      "grad_norm": 0.7056323289871216,
      "learning_rate": 2.2380759874479598e-06,
      "loss": 0.6764,
      "step": 24145
    },
    {
      "epoch": 0.6957298449835764,
      "grad_norm": 0.7269574999809265,
      "learning_rate": 2.2376870382928608e-06,
      "loss": 0.707,
      "step": 24146
    },
    {
      "epoch": 0.6957586584452257,
      "grad_norm": 0.7133417725563049,
      "learning_rate": 2.2372981131947447e-06,
      "loss": 0.6904,
      "step": 24147
    },
    {
      "epoch": 0.6957874719068748,
      "grad_norm": 0.7301615476608276,
      "learning_rate": 2.236909212157002e-06,
      "loss": 0.6709,
      "step": 24148
    },
    {
      "epoch": 0.6958162853685241,
      "grad_norm": 0.7118933200836182,
      "learning_rate": 2.236520335183016e-06,
      "loss": 0.7118,
      "step": 24149
    },
    {
      "epoch": 0.6958450988301734,
      "grad_norm": 0.7101120352745056,
      "learning_rate": 2.236131482276176e-06,
      "loss": 0.6996,
      "step": 24150
    },
    {
      "epoch": 0.6958739122918227,
      "grad_norm": 0.7098739743232727,
      "learning_rate": 2.235742653439869e-06,
      "loss": 0.6845,
      "step": 24151
    },
    {
      "epoch": 0.695902725753472,
      "grad_norm": 0.7282634377479553,
      "learning_rate": 2.235353848677479e-06,
      "loss": 0.681,
      "step": 24152
    },
    {
      "epoch": 0.6959315392151213,
      "grad_norm": 0.7364455461502075,
      "learning_rate": 2.234965067992395e-06,
      "loss": 0.7101,
      "step": 24153
    },
    {
      "epoch": 0.6959603526767706,
      "grad_norm": 0.7238950133323669,
      "learning_rate": 2.234576311388e-06,
      "loss": 0.717,
      "step": 24154
    },
    {
      "epoch": 0.6959891661384199,
      "grad_norm": 0.7354992032051086,
      "learning_rate": 2.2341875788676813e-06,
      "loss": 0.6816,
      "step": 24155
    },
    {
      "epoch": 0.6960179796000692,
      "grad_norm": 0.7146230340003967,
      "learning_rate": 2.2337988704348236e-06,
      "loss": 0.6982,
      "step": 24156
    },
    {
      "epoch": 0.6960467930617185,
      "grad_norm": 0.7075104713439941,
      "learning_rate": 2.233410186092812e-06,
      "loss": 0.676,
      "step": 24157
    },
    {
      "epoch": 0.6960756065233678,
      "grad_norm": 0.7092946767807007,
      "learning_rate": 2.2330215258450315e-06,
      "loss": 0.6713,
      "step": 24158
    },
    {
      "epoch": 0.696104419985017,
      "grad_norm": 0.7362231016159058,
      "learning_rate": 2.23263288969487e-06,
      "loss": 0.6867,
      "step": 24159
    },
    {
      "epoch": 0.6961332334466663,
      "grad_norm": 0.7326217293739319,
      "learning_rate": 2.232244277645708e-06,
      "loss": 0.6699,
      "step": 24160
    },
    {
      "epoch": 0.6961620469083155,
      "grad_norm": 0.8363730311393738,
      "learning_rate": 2.231855689700933e-06,
      "loss": 0.6917,
      "step": 24161
    },
    {
      "epoch": 0.6961908603699648,
      "grad_norm": 0.7089260816574097,
      "learning_rate": 2.2314671258639264e-06,
      "loss": 0.6882,
      "step": 24162
    },
    {
      "epoch": 0.6962196738316141,
      "grad_norm": 0.7576859593391418,
      "learning_rate": 2.231078586138075e-06,
      "loss": 0.7393,
      "step": 24163
    },
    {
      "epoch": 0.6962484872932634,
      "grad_norm": 0.7374047040939331,
      "learning_rate": 2.23069007052676e-06,
      "loss": 0.7013,
      "step": 24164
    },
    {
      "epoch": 0.6962773007549127,
      "grad_norm": 0.714327871799469,
      "learning_rate": 2.230301579033366e-06,
      "loss": 0.6802,
      "step": 24165
    },
    {
      "epoch": 0.696306114216562,
      "grad_norm": 0.7386523485183716,
      "learning_rate": 2.2299131116612776e-06,
      "loss": 0.701,
      "step": 24166
    },
    {
      "epoch": 0.6963349276782113,
      "grad_norm": 0.7188432216644287,
      "learning_rate": 2.229524668413876e-06,
      "loss": 0.7095,
      "step": 24167
    },
    {
      "epoch": 0.6963637411398605,
      "grad_norm": 0.723828136920929,
      "learning_rate": 2.229136249294547e-06,
      "loss": 0.7177,
      "step": 24168
    },
    {
      "epoch": 0.6963925546015098,
      "grad_norm": 0.7164544463157654,
      "learning_rate": 2.228747854306669e-06,
      "loss": 0.6786,
      "step": 24169
    },
    {
      "epoch": 0.6964213680631591,
      "grad_norm": 0.7279087901115417,
      "learning_rate": 2.228359483453627e-06,
      "loss": 0.6737,
      "step": 24170
    },
    {
      "epoch": 0.6964501815248084,
      "grad_norm": 0.743137001991272,
      "learning_rate": 2.2279711367388033e-06,
      "loss": 0.7063,
      "step": 24171
    },
    {
      "epoch": 0.6964789949864577,
      "grad_norm": 0.7153415679931641,
      "learning_rate": 2.227582814165582e-06,
      "loss": 0.6937,
      "step": 24172
    },
    {
      "epoch": 0.696507808448107,
      "grad_norm": 0.7195417881011963,
      "learning_rate": 2.227194515737341e-06,
      "loss": 0.7026,
      "step": 24173
    },
    {
      "epoch": 0.6965366219097563,
      "grad_norm": 0.7142753005027771,
      "learning_rate": 2.226806241457466e-06,
      "loss": 0.6991,
      "step": 24174
    },
    {
      "epoch": 0.6965654353714055,
      "grad_norm": 0.7360350489616394,
      "learning_rate": 2.226417991329334e-06,
      "loss": 0.6756,
      "step": 24175
    },
    {
      "epoch": 0.6965942488330548,
      "grad_norm": 0.7352983355522156,
      "learning_rate": 2.2260297653563318e-06,
      "loss": 0.7044,
      "step": 24176
    },
    {
      "epoch": 0.696623062294704,
      "grad_norm": 0.7465786933898926,
      "learning_rate": 2.2256415635418343e-06,
      "loss": 0.713,
      "step": 24177
    },
    {
      "epoch": 0.6966518757563533,
      "grad_norm": 0.7260675430297852,
      "learning_rate": 2.2252533858892278e-06,
      "loss": 0.7029,
      "step": 24178
    },
    {
      "epoch": 0.6966806892180026,
      "grad_norm": 0.7216808795928955,
      "learning_rate": 2.224865232401889e-06,
      "loss": 0.6728,
      "step": 24179
    },
    {
      "epoch": 0.6967095026796519,
      "grad_norm": 0.7191162109375,
      "learning_rate": 2.224477103083201e-06,
      "loss": 0.7064,
      "step": 24180
    },
    {
      "epoch": 0.6967383161413012,
      "grad_norm": 0.717287003993988,
      "learning_rate": 2.2240889979365404e-06,
      "loss": 0.6951,
      "step": 24181
    },
    {
      "epoch": 0.6967671296029505,
      "grad_norm": 0.7292394042015076,
      "learning_rate": 2.2237009169652922e-06,
      "loss": 0.7041,
      "step": 24182
    },
    {
      "epoch": 0.6967959430645998,
      "grad_norm": 0.7231507301330566,
      "learning_rate": 2.223312860172833e-06,
      "loss": 0.6943,
      "step": 24183
    },
    {
      "epoch": 0.6968247565262491,
      "grad_norm": 0.7186278104782104,
      "learning_rate": 2.222924827562544e-06,
      "loss": 0.6873,
      "step": 24184
    },
    {
      "epoch": 0.6968535699878984,
      "grad_norm": 0.8643543124198914,
      "learning_rate": 2.2225368191378023e-06,
      "loss": 0.753,
      "step": 24185
    },
    {
      "epoch": 0.6968823834495477,
      "grad_norm": 0.7013949155807495,
      "learning_rate": 2.2221488349019903e-06,
      "loss": 0.6917,
      "step": 24186
    },
    {
      "epoch": 0.696911196911197,
      "grad_norm": 0.7203569412231445,
      "learning_rate": 2.2217608748584834e-06,
      "loss": 0.6858,
      "step": 24187
    },
    {
      "epoch": 0.6969400103728461,
      "grad_norm": 0.743351936340332,
      "learning_rate": 2.221372939010662e-06,
      "loss": 0.7089,
      "step": 24188
    },
    {
      "epoch": 0.6969688238344954,
      "grad_norm": 0.7104344964027405,
      "learning_rate": 2.220985027361906e-06,
      "loss": 0.6978,
      "step": 24189
    },
    {
      "epoch": 0.6969976372961447,
      "grad_norm": 0.7247240543365479,
      "learning_rate": 2.2205971399155912e-06,
      "loss": 0.7041,
      "step": 24190
    },
    {
      "epoch": 0.697026450757794,
      "grad_norm": 0.7357364892959595,
      "learning_rate": 2.220209276675098e-06,
      "loss": 0.6868,
      "step": 24191
    },
    {
      "epoch": 0.6970552642194433,
      "grad_norm": 0.7034575343132019,
      "learning_rate": 2.2198214376438006e-06,
      "loss": 0.6727,
      "step": 24192
    },
    {
      "epoch": 0.6970840776810926,
      "grad_norm": 0.7077149152755737,
      "learning_rate": 2.219433622825082e-06,
      "loss": 0.7112,
      "step": 24193
    },
    {
      "epoch": 0.6971128911427419,
      "grad_norm": 0.711320161819458,
      "learning_rate": 2.219045832222316e-06,
      "loss": 0.6977,
      "step": 24194
    },
    {
      "epoch": 0.6971417046043912,
      "grad_norm": 0.7226985692977905,
      "learning_rate": 2.2186580658388822e-06,
      "loss": 0.7007,
      "step": 24195
    },
    {
      "epoch": 0.6971705180660405,
      "grad_norm": 0.7110329270362854,
      "learning_rate": 2.218270323678155e-06,
      "loss": 0.7042,
      "step": 24196
    },
    {
      "epoch": 0.6971993315276898,
      "grad_norm": 0.7135975360870361,
      "learning_rate": 2.217882605743514e-06,
      "loss": 0.6899,
      "step": 24197
    },
    {
      "epoch": 0.697228144989339,
      "grad_norm": 0.708970308303833,
      "learning_rate": 2.2174949120383326e-06,
      "loss": 0.6843,
      "step": 24198
    },
    {
      "epoch": 0.6972569584509883,
      "grad_norm": 0.6833356022834778,
      "learning_rate": 2.2171072425659905e-06,
      "loss": 0.6495,
      "step": 24199
    },
    {
      "epoch": 0.6972857719126376,
      "grad_norm": 0.6990953087806702,
      "learning_rate": 2.216719597329861e-06,
      "loss": 0.6944,
      "step": 24200
    },
    {
      "epoch": 0.6973145853742869,
      "grad_norm": 0.7243633270263672,
      "learning_rate": 2.2163319763333234e-06,
      "loss": 0.7274,
      "step": 24201
    },
    {
      "epoch": 0.6973433988359361,
      "grad_norm": 0.712345540523529,
      "learning_rate": 2.2159443795797496e-06,
      "loss": 0.716,
      "step": 24202
    },
    {
      "epoch": 0.6973722122975854,
      "grad_norm": 0.7172257304191589,
      "learning_rate": 2.2155568070725176e-06,
      "loss": 0.6817,
      "step": 24203
    },
    {
      "epoch": 0.6974010257592347,
      "grad_norm": 0.7245849967002869,
      "learning_rate": 2.215169258815002e-06,
      "loss": 0.7078,
      "step": 24204
    },
    {
      "epoch": 0.697429839220884,
      "grad_norm": 0.7269096970558167,
      "learning_rate": 2.2147817348105804e-06,
      "loss": 0.693,
      "step": 24205
    },
    {
      "epoch": 0.6974586526825333,
      "grad_norm": 0.7152659893035889,
      "learning_rate": 2.214394235062624e-06,
      "loss": 0.6804,
      "step": 24206
    },
    {
      "epoch": 0.6974874661441826,
      "grad_norm": 0.7068305611610413,
      "learning_rate": 2.2140067595745106e-06,
      "loss": 0.6793,
      "step": 24207
    },
    {
      "epoch": 0.6975162796058318,
      "grad_norm": 0.7079188227653503,
      "learning_rate": 2.213619308349612e-06,
      "loss": 0.6915,
      "step": 24208
    },
    {
      "epoch": 0.6975450930674811,
      "grad_norm": 0.7235120534896851,
      "learning_rate": 2.2132318813913036e-06,
      "loss": 0.6825,
      "step": 24209
    },
    {
      "epoch": 0.6975739065291304,
      "grad_norm": 0.7186506986618042,
      "learning_rate": 2.212844478702962e-06,
      "loss": 0.6909,
      "step": 24210
    },
    {
      "epoch": 0.6976027199907797,
      "grad_norm": 0.7048314809799194,
      "learning_rate": 2.2124571002879563e-06,
      "loss": 0.698,
      "step": 24211
    },
    {
      "epoch": 0.697631533452429,
      "grad_norm": 0.7211605906486511,
      "learning_rate": 2.212069746149665e-06,
      "loss": 0.6789,
      "step": 24212
    },
    {
      "epoch": 0.6976603469140783,
      "grad_norm": 0.730522871017456,
      "learning_rate": 2.2116824162914574e-06,
      "loss": 0.6976,
      "step": 24213
    },
    {
      "epoch": 0.6976891603757276,
      "grad_norm": 0.6967996954917908,
      "learning_rate": 2.2112951107167103e-06,
      "loss": 0.6845,
      "step": 24214
    },
    {
      "epoch": 0.6977179738373768,
      "grad_norm": 0.7002112865447998,
      "learning_rate": 2.2109078294287937e-06,
      "loss": 0.6818,
      "step": 24215
    },
    {
      "epoch": 0.6977467872990261,
      "grad_norm": 0.6870675683021545,
      "learning_rate": 2.2105205724310813e-06,
      "loss": 0.6807,
      "step": 24216
    },
    {
      "epoch": 0.6977756007606754,
      "grad_norm": 0.7425279021263123,
      "learning_rate": 2.210133339726947e-06,
      "loss": 0.6857,
      "step": 24217
    },
    {
      "epoch": 0.6978044142223246,
      "grad_norm": 0.7362670302391052,
      "learning_rate": 2.2097461313197638e-06,
      "loss": 0.6535,
      "step": 24218
    },
    {
      "epoch": 0.6978332276839739,
      "grad_norm": 0.7229413390159607,
      "learning_rate": 2.2093589472129007e-06,
      "loss": 0.7064,
      "step": 24219
    },
    {
      "epoch": 0.6978620411456232,
      "grad_norm": 0.7154456973075867,
      "learning_rate": 2.2089717874097333e-06,
      "loss": 0.693,
      "step": 24220
    },
    {
      "epoch": 0.6978908546072725,
      "grad_norm": 0.7317591309547424,
      "learning_rate": 2.20858465191363e-06,
      "loss": 0.7062,
      "step": 24221
    },
    {
      "epoch": 0.6979196680689218,
      "grad_norm": 0.7185149192810059,
      "learning_rate": 2.2081975407279654e-06,
      "loss": 0.6943,
      "step": 24222
    },
    {
      "epoch": 0.6979484815305711,
      "grad_norm": 0.7104181051254272,
      "learning_rate": 2.2078104538561085e-06,
      "loss": 0.7005,
      "step": 24223
    },
    {
      "epoch": 0.6979772949922204,
      "grad_norm": 0.7195807695388794,
      "learning_rate": 2.2074233913014303e-06,
      "loss": 0.6855,
      "step": 24224
    },
    {
      "epoch": 0.6980061084538697,
      "grad_norm": 0.7255443930625916,
      "learning_rate": 2.2070363530673055e-06,
      "loss": 0.7082,
      "step": 24225
    },
    {
      "epoch": 0.698034921915519,
      "grad_norm": 0.7274132966995239,
      "learning_rate": 2.2066493391570997e-06,
      "loss": 0.6909,
      "step": 24226
    },
    {
      "epoch": 0.6980637353771683,
      "grad_norm": 0.702307403087616,
      "learning_rate": 2.206262349574188e-06,
      "loss": 0.6851,
      "step": 24227
    },
    {
      "epoch": 0.6980925488388176,
      "grad_norm": 0.7239433526992798,
      "learning_rate": 2.2058753843219365e-06,
      "loss": 0.701,
      "step": 24228
    },
    {
      "epoch": 0.6981213623004667,
      "grad_norm": 0.7143436074256897,
      "learning_rate": 2.2054884434037175e-06,
      "loss": 0.684,
      "step": 24229
    },
    {
      "epoch": 0.698150175762116,
      "grad_norm": 0.7046546339988708,
      "learning_rate": 2.205101526822901e-06,
      "loss": 0.7003,
      "step": 24230
    },
    {
      "epoch": 0.6981789892237653,
      "grad_norm": 0.7077057361602783,
      "learning_rate": 2.2047146345828575e-06,
      "loss": 0.6982,
      "step": 24231
    },
    {
      "epoch": 0.6982078026854146,
      "grad_norm": 0.724226713180542,
      "learning_rate": 2.204327766686954e-06,
      "loss": 0.6774,
      "step": 24232
    },
    {
      "epoch": 0.6982366161470639,
      "grad_norm": 0.7402669191360474,
      "learning_rate": 2.2039409231385624e-06,
      "loss": 0.6944,
      "step": 24233
    },
    {
      "epoch": 0.6982654296087132,
      "grad_norm": 0.7212876677513123,
      "learning_rate": 2.2035541039410485e-06,
      "loss": 0.6829,
      "step": 24234
    },
    {
      "epoch": 0.6982942430703625,
      "grad_norm": 0.7420119047164917,
      "learning_rate": 2.203167309097785e-06,
      "loss": 0.6815,
      "step": 24235
    },
    {
      "epoch": 0.6983230565320118,
      "grad_norm": 0.7216356992721558,
      "learning_rate": 2.2027805386121366e-06,
      "loss": 0.6832,
      "step": 24236
    },
    {
      "epoch": 0.6983518699936611,
      "grad_norm": 0.7012999653816223,
      "learning_rate": 2.202393792487475e-06,
      "loss": 0.6713,
      "step": 24237
    },
    {
      "epoch": 0.6983806834553103,
      "grad_norm": 0.7393693327903748,
      "learning_rate": 2.202007070727165e-06,
      "loss": 0.6994,
      "step": 24238
    },
    {
      "epoch": 0.6984094969169596,
      "grad_norm": 0.7234730124473572,
      "learning_rate": 2.2016203733345788e-06,
      "loss": 0.7127,
      "step": 24239
    },
    {
      "epoch": 0.6984383103786089,
      "grad_norm": 0.7370679378509521,
      "learning_rate": 2.201233700313078e-06,
      "loss": 0.6838,
      "step": 24240
    },
    {
      "epoch": 0.6984671238402582,
      "grad_norm": 0.7013280987739563,
      "learning_rate": 2.2008470516660375e-06,
      "loss": 0.6816,
      "step": 24241
    },
    {
      "epoch": 0.6984959373019074,
      "grad_norm": 0.7217843532562256,
      "learning_rate": 2.200460427396819e-06,
      "loss": 0.6858,
      "step": 24242
    },
    {
      "epoch": 0.6985247507635567,
      "grad_norm": 0.7162925601005554,
      "learning_rate": 2.2000738275087936e-06,
      "loss": 0.6902,
      "step": 24243
    },
    {
      "epoch": 0.698553564225206,
      "grad_norm": 0.6986884474754333,
      "learning_rate": 2.1996872520053247e-06,
      "loss": 0.6718,
      "step": 24244
    },
    {
      "epoch": 0.6985823776868553,
      "grad_norm": 0.7030417919158936,
      "learning_rate": 2.199300700889782e-06,
      "loss": 0.6884,
      "step": 24245
    },
    {
      "epoch": 0.6986111911485046,
      "grad_norm": 0.7168347239494324,
      "learning_rate": 2.198914174165529e-06,
      "loss": 0.7056,
      "step": 24246
    },
    {
      "epoch": 0.6986400046101539,
      "grad_norm": 0.7063633799552917,
      "learning_rate": 2.1985276718359333e-06,
      "loss": 0.6995,
      "step": 24247
    },
    {
      "epoch": 0.6986688180718031,
      "grad_norm": 0.719785213470459,
      "learning_rate": 2.1981411939043627e-06,
      "loss": 0.6951,
      "step": 24248
    },
    {
      "epoch": 0.6986976315334524,
      "grad_norm": 0.6961022615432739,
      "learning_rate": 2.1977547403741807e-06,
      "loss": 0.6934,
      "step": 24249
    },
    {
      "epoch": 0.6987264449951017,
      "grad_norm": 0.7108792662620544,
      "learning_rate": 2.1973683112487544e-06,
      "loss": 0.6771,
      "step": 24250
    },
    {
      "epoch": 0.698755258456751,
      "grad_norm": 0.7112727761268616,
      "learning_rate": 2.1969819065314472e-06,
      "loss": 0.6755,
      "step": 24251
    },
    {
      "epoch": 0.6987840719184003,
      "grad_norm": 0.7226611375808716,
      "learning_rate": 2.1965955262256254e-06,
      "loss": 0.7072,
      "step": 24252
    },
    {
      "epoch": 0.6988128853800496,
      "grad_norm": 0.7335023880004883,
      "learning_rate": 2.1962091703346545e-06,
      "loss": 0.7337,
      "step": 24253
    },
    {
      "epoch": 0.6988416988416989,
      "grad_norm": 0.7702564001083374,
      "learning_rate": 2.1958228388619003e-06,
      "loss": 0.7081,
      "step": 24254
    },
    {
      "epoch": 0.6988705123033482,
      "grad_norm": 0.7241091132164001,
      "learning_rate": 2.1954365318107247e-06,
      "loss": 0.663,
      "step": 24255
    },
    {
      "epoch": 0.6988993257649974,
      "grad_norm": 0.7226550579071045,
      "learning_rate": 2.1950502491844946e-06,
      "loss": 0.6578,
      "step": 24256
    },
    {
      "epoch": 0.6989281392266466,
      "grad_norm": 0.720967710018158,
      "learning_rate": 2.1946639909865714e-06,
      "loss": 0.6868,
      "step": 24257
    },
    {
      "epoch": 0.6989569526882959,
      "grad_norm": 0.7063256502151489,
      "learning_rate": 2.194277757220322e-06,
      "loss": 0.656,
      "step": 24258
    },
    {
      "epoch": 0.6989857661499452,
      "grad_norm": 0.7271559238433838,
      "learning_rate": 2.193891547889108e-06,
      "loss": 0.707,
      "step": 24259
    },
    {
      "epoch": 0.6990145796115945,
      "grad_norm": 0.7029381394386292,
      "learning_rate": 2.193505362996294e-06,
      "loss": 0.6675,
      "step": 24260
    },
    {
      "epoch": 0.6990433930732438,
      "grad_norm": 0.7038237452507019,
      "learning_rate": 2.193119202545242e-06,
      "loss": 0.6904,
      "step": 24261
    },
    {
      "epoch": 0.6990722065348931,
      "grad_norm": 0.7373114228248596,
      "learning_rate": 2.192733066539315e-06,
      "loss": 0.6868,
      "step": 24262
    },
    {
      "epoch": 0.6991010199965424,
      "grad_norm": 0.7208968997001648,
      "learning_rate": 2.1923469549818794e-06,
      "loss": 0.6692,
      "step": 24263
    },
    {
      "epoch": 0.6991298334581917,
      "grad_norm": 0.7077429294586182,
      "learning_rate": 2.191960867876293e-06,
      "loss": 0.707,
      "step": 24264
    },
    {
      "epoch": 0.699158646919841,
      "grad_norm": 0.7193606495857239,
      "learning_rate": 2.19157480522592e-06,
      "loss": 0.7147,
      "step": 24265
    },
    {
      "epoch": 0.6991874603814903,
      "grad_norm": 0.719319760799408,
      "learning_rate": 2.191188767034125e-06,
      "loss": 0.6869,
      "step": 24266
    },
    {
      "epoch": 0.6992162738431396,
      "grad_norm": 0.7210947275161743,
      "learning_rate": 2.190802753304267e-06,
      "loss": 0.6778,
      "step": 24267
    },
    {
      "epoch": 0.6992450873047888,
      "grad_norm": 0.7118632197380066,
      "learning_rate": 2.1904167640397085e-06,
      "loss": 0.6711,
      "step": 24268
    },
    {
      "epoch": 0.699273900766438,
      "grad_norm": 0.6989747881889343,
      "learning_rate": 2.190030799243813e-06,
      "loss": 0.6694,
      "step": 24269
    },
    {
      "epoch": 0.6993027142280873,
      "grad_norm": 0.7235652804374695,
      "learning_rate": 2.189644858919938e-06,
      "loss": 0.6849,
      "step": 24270
    },
    {
      "epoch": 0.6993315276897366,
      "grad_norm": 0.6945896148681641,
      "learning_rate": 2.18925894307145e-06,
      "loss": 0.6786,
      "step": 24271
    },
    {
      "epoch": 0.6993603411513859,
      "grad_norm": 0.7350858449935913,
      "learning_rate": 2.188873051701704e-06,
      "loss": 0.7003,
      "step": 24272
    },
    {
      "epoch": 0.6993891546130352,
      "grad_norm": 0.7127845883369446,
      "learning_rate": 2.188487184814066e-06,
      "loss": 0.6988,
      "step": 24273
    },
    {
      "epoch": 0.6994179680746845,
      "grad_norm": 0.7489179968833923,
      "learning_rate": 2.1881013424118925e-06,
      "loss": 0.707,
      "step": 24274
    },
    {
      "epoch": 0.6994467815363338,
      "grad_norm": 0.7322676777839661,
      "learning_rate": 2.1877155244985465e-06,
      "loss": 0.6851,
      "step": 24275
    },
    {
      "epoch": 0.6994755949979831,
      "grad_norm": 0.7354908585548401,
      "learning_rate": 2.187329731077384e-06,
      "loss": 0.7055,
      "step": 24276
    },
    {
      "epoch": 0.6995044084596324,
      "grad_norm": 0.7294327020645142,
      "learning_rate": 2.186943962151771e-06,
      "loss": 0.7141,
      "step": 24277
    },
    {
      "epoch": 0.6995332219212816,
      "grad_norm": 0.7031390070915222,
      "learning_rate": 2.186558217725063e-06,
      "loss": 0.6954,
      "step": 24278
    },
    {
      "epoch": 0.6995620353829309,
      "grad_norm": 0.717883288860321,
      "learning_rate": 2.1861724978006223e-06,
      "loss": 0.7028,
      "step": 24279
    },
    {
      "epoch": 0.6995908488445802,
      "grad_norm": 0.7098450660705566,
      "learning_rate": 2.185786802381804e-06,
      "loss": 0.7042,
      "step": 24280
    },
    {
      "epoch": 0.6996196623062295,
      "grad_norm": 0.6885015964508057,
      "learning_rate": 2.1854011314719713e-06,
      "loss": 0.6839,
      "step": 24281
    },
    {
      "epoch": 0.6996484757678788,
      "grad_norm": 0.7476516962051392,
      "learning_rate": 2.1850154850744796e-06,
      "loss": 0.6948,
      "step": 24282
    },
    {
      "epoch": 0.699677289229528,
      "grad_norm": 0.7039051651954651,
      "learning_rate": 2.184629863192691e-06,
      "loss": 0.6913,
      "step": 24283
    },
    {
      "epoch": 0.6997061026911773,
      "grad_norm": 0.7213229537010193,
      "learning_rate": 2.1842442658299595e-06,
      "loss": 0.6805,
      "step": 24284
    },
    {
      "epoch": 0.6997349161528266,
      "grad_norm": 0.7232910394668579,
      "learning_rate": 2.1838586929896467e-06,
      "loss": 0.6643,
      "step": 24285
    },
    {
      "epoch": 0.6997637296144759,
      "grad_norm": 0.714693546295166,
      "learning_rate": 2.1834731446751106e-06,
      "loss": 0.7368,
      "step": 24286
    },
    {
      "epoch": 0.6997925430761252,
      "grad_norm": 0.7170575857162476,
      "learning_rate": 2.1830876208897064e-06,
      "loss": 0.6955,
      "step": 24287
    },
    {
      "epoch": 0.6998213565377744,
      "grad_norm": 0.7045593857765198,
      "learning_rate": 2.182702121636793e-06,
      "loss": 0.6977,
      "step": 24288
    },
    {
      "epoch": 0.6998501699994237,
      "grad_norm": 0.6991647481918335,
      "learning_rate": 2.1823166469197277e-06,
      "loss": 0.6896,
      "step": 24289
    },
    {
      "epoch": 0.699878983461073,
      "grad_norm": 0.7023872137069702,
      "learning_rate": 2.18193119674187e-06,
      "loss": 0.6956,
      "step": 24290
    },
    {
      "epoch": 0.6999077969227223,
      "grad_norm": 0.7578942775726318,
      "learning_rate": 2.1815457711065723e-06,
      "loss": 0.6987,
      "step": 24291
    },
    {
      "epoch": 0.6999366103843716,
      "grad_norm": 0.714721143245697,
      "learning_rate": 2.1811603700171956e-06,
      "loss": 0.6857,
      "step": 24292
    },
    {
      "epoch": 0.6999654238460209,
      "grad_norm": 0.7401191592216492,
      "learning_rate": 2.1807749934770928e-06,
      "loss": 0.7006,
      "step": 24293
    },
    {
      "epoch": 0.6999942373076702,
      "grad_norm": 0.7125264406204224,
      "learning_rate": 2.180389641489623e-06,
      "loss": 0.6983,
      "step": 24294
    },
    {
      "epoch": 0.7000230507693195,
      "grad_norm": 0.7134762406349182,
      "learning_rate": 2.180004314058139e-06,
      "loss": 0.6985,
      "step": 24295
    },
    {
      "epoch": 0.7000518642309687,
      "grad_norm": 0.7109701037406921,
      "learning_rate": 2.1796190111860007e-06,
      "loss": 0.7112,
      "step": 24296
    },
    {
      "epoch": 0.700080677692618,
      "grad_norm": 0.7180153727531433,
      "learning_rate": 2.179233732876559e-06,
      "loss": 0.6856,
      "step": 24297
    },
    {
      "epoch": 0.7001094911542672,
      "grad_norm": 0.7163897752761841,
      "learning_rate": 2.178848479133174e-06,
      "loss": 0.6836,
      "step": 24298
    },
    {
      "epoch": 0.7001383046159165,
      "grad_norm": 0.7434770464897156,
      "learning_rate": 2.178463249959196e-06,
      "loss": 0.6858,
      "step": 24299
    },
    {
      "epoch": 0.7001671180775658,
      "grad_norm": 0.7265965938568115,
      "learning_rate": 2.1780780453579835e-06,
      "loss": 0.663,
      "step": 24300
    },
    {
      "epoch": 0.7001959315392151,
      "grad_norm": 0.7176247835159302,
      "learning_rate": 2.1776928653328905e-06,
      "loss": 0.6905,
      "step": 24301
    },
    {
      "epoch": 0.7002247450008644,
      "grad_norm": 0.7210660576820374,
      "learning_rate": 2.1773077098872723e-06,
      "loss": 0.6874,
      "step": 24302
    },
    {
      "epoch": 0.7002535584625137,
      "grad_norm": 0.7265874743461609,
      "learning_rate": 2.1769225790244815e-06,
      "loss": 0.6886,
      "step": 24303
    },
    {
      "epoch": 0.700282371924163,
      "grad_norm": 0.7181962132453918,
      "learning_rate": 2.176537472747874e-06,
      "loss": 0.6926,
      "step": 24304
    },
    {
      "epoch": 0.7003111853858123,
      "grad_norm": 0.7083588242530823,
      "learning_rate": 2.176152391060801e-06,
      "loss": 0.684,
      "step": 24305
    },
    {
      "epoch": 0.7003399988474616,
      "grad_norm": 0.7043052911758423,
      "learning_rate": 2.175767333966618e-06,
      "loss": 0.6683,
      "step": 24306
    },
    {
      "epoch": 0.7003688123091109,
      "grad_norm": 0.7137129902839661,
      "learning_rate": 2.1753823014686803e-06,
      "loss": 0.6973,
      "step": 24307
    },
    {
      "epoch": 0.7003976257707601,
      "grad_norm": 0.716093897819519,
      "learning_rate": 2.174997293570337e-06,
      "loss": 0.6943,
      "step": 24308
    },
    {
      "epoch": 0.7004264392324094,
      "grad_norm": 0.7306383848190308,
      "learning_rate": 2.1746123102749455e-06,
      "loss": 0.6907,
      "step": 24309
    },
    {
      "epoch": 0.7004552526940586,
      "grad_norm": 0.6940576434135437,
      "learning_rate": 2.1742273515858542e-06,
      "loss": 0.6842,
      "step": 24310
    },
    {
      "epoch": 0.7004840661557079,
      "grad_norm": 0.731387197971344,
      "learning_rate": 2.1738424175064197e-06,
      "loss": 0.6922,
      "step": 24311
    },
    {
      "epoch": 0.7005128796173572,
      "grad_norm": 0.7154773473739624,
      "learning_rate": 2.1734575080399893e-06,
      "loss": 0.7127,
      "step": 24312
    },
    {
      "epoch": 0.7005416930790065,
      "grad_norm": 0.7051751017570496,
      "learning_rate": 2.1730726231899225e-06,
      "loss": 0.6769,
      "step": 24313
    },
    {
      "epoch": 0.7005705065406558,
      "grad_norm": 0.7132088541984558,
      "learning_rate": 2.172687762959565e-06,
      "loss": 0.6973,
      "step": 24314
    },
    {
      "epoch": 0.7005993200023051,
      "grad_norm": 0.6932479739189148,
      "learning_rate": 2.172302927352272e-06,
      "loss": 0.6937,
      "step": 24315
    },
    {
      "epoch": 0.7006281334639544,
      "grad_norm": 0.7184317111968994,
      "learning_rate": 2.171918116371393e-06,
      "loss": 0.7006,
      "step": 24316
    },
    {
      "epoch": 0.7006569469256037,
      "grad_norm": 0.7368355989456177,
      "learning_rate": 2.1715333300202816e-06,
      "loss": 0.6686,
      "step": 24317
    },
    {
      "epoch": 0.7006857603872529,
      "grad_norm": 0.7113174796104431,
      "learning_rate": 2.171148568302286e-06,
      "loss": 0.6944,
      "step": 24318
    },
    {
      "epoch": 0.7007145738489022,
      "grad_norm": 0.7121967673301697,
      "learning_rate": 2.1707638312207603e-06,
      "loss": 0.6874,
      "step": 24319
    },
    {
      "epoch": 0.7007433873105515,
      "grad_norm": 0.7192184925079346,
      "learning_rate": 2.1703791187790508e-06,
      "loss": 0.6916,
      "step": 24320
    },
    {
      "epoch": 0.7007722007722008,
      "grad_norm": 0.7020397782325745,
      "learning_rate": 2.1699944309805134e-06,
      "loss": 0.7004,
      "step": 24321
    },
    {
      "epoch": 0.7008010142338501,
      "grad_norm": 0.7284576296806335,
      "learning_rate": 2.1696097678284934e-06,
      "loss": 0.6932,
      "step": 24322
    },
    {
      "epoch": 0.7008298276954993,
      "grad_norm": 0.7119918465614319,
      "learning_rate": 2.1692251293263422e-06,
      "loss": 0.6537,
      "step": 24323
    },
    {
      "epoch": 0.7008586411571486,
      "grad_norm": 0.6953808665275574,
      "learning_rate": 2.168840515477411e-06,
      "loss": 0.6647,
      "step": 24324
    },
    {
      "epoch": 0.7008874546187979,
      "grad_norm": 0.7258304357528687,
      "learning_rate": 2.168455926285051e-06,
      "loss": 0.7084,
      "step": 24325
    },
    {
      "epoch": 0.7009162680804472,
      "grad_norm": 0.7145716547966003,
      "learning_rate": 2.1680713617526072e-06,
      "loss": 0.6809,
      "step": 24326
    },
    {
      "epoch": 0.7009450815420964,
      "grad_norm": 0.7310694456100464,
      "learning_rate": 2.1676868218834308e-06,
      "loss": 0.712,
      "step": 24327
    },
    {
      "epoch": 0.7009738950037457,
      "grad_norm": 0.7203277945518494,
      "learning_rate": 2.167302306680873e-06,
      "loss": 0.6953,
      "step": 24328
    },
    {
      "epoch": 0.701002708465395,
      "grad_norm": 0.7195767760276794,
      "learning_rate": 2.1669178161482784e-06,
      "loss": 0.7195,
      "step": 24329
    },
    {
      "epoch": 0.7010315219270443,
      "grad_norm": 0.7152245044708252,
      "learning_rate": 2.1665333502889996e-06,
      "loss": 0.676,
      "step": 24330
    },
    {
      "epoch": 0.7010603353886936,
      "grad_norm": 0.7239017486572266,
      "learning_rate": 2.1661489091063807e-06,
      "loss": 0.7162,
      "step": 24331
    },
    {
      "epoch": 0.7010891488503429,
      "grad_norm": 0.705288827419281,
      "learning_rate": 2.165764492603774e-06,
      "loss": 0.6787,
      "step": 24332
    },
    {
      "epoch": 0.7011179623119922,
      "grad_norm": 0.7120196223258972,
      "learning_rate": 2.165380100784524e-06,
      "loss": 0.6913,
      "step": 24333
    },
    {
      "epoch": 0.7011467757736415,
      "grad_norm": 0.7149800062179565,
      "learning_rate": 2.1649957336519813e-06,
      "loss": 0.6857,
      "step": 24334
    },
    {
      "epoch": 0.7011755892352908,
      "grad_norm": 0.7322036027908325,
      "learning_rate": 2.16461139120949e-06,
      "loss": 0.7158,
      "step": 24335
    },
    {
      "epoch": 0.7012044026969401,
      "grad_norm": 0.712825357913971,
      "learning_rate": 2.1642270734603986e-06,
      "loss": 0.6561,
      "step": 24336
    },
    {
      "epoch": 0.7012332161585892,
      "grad_norm": 0.7190744876861572,
      "learning_rate": 2.1638427804080553e-06,
      "loss": 0.6898,
      "step": 24337
    },
    {
      "epoch": 0.7012620296202385,
      "grad_norm": 0.7240719199180603,
      "learning_rate": 2.1634585120558076e-06,
      "loss": 0.6892,
      "step": 24338
    },
    {
      "epoch": 0.7012908430818878,
      "grad_norm": 0.7428410053253174,
      "learning_rate": 2.1630742684069993e-06,
      "loss": 0.6876,
      "step": 24339
    },
    {
      "epoch": 0.7013196565435371,
      "grad_norm": 0.7199839949607849,
      "learning_rate": 2.16269004946498e-06,
      "loss": 0.6916,
      "step": 24340
    },
    {
      "epoch": 0.7013484700051864,
      "grad_norm": 0.7180384993553162,
      "learning_rate": 2.1623058552330922e-06,
      "loss": 0.6806,
      "step": 24341
    },
    {
      "epoch": 0.7013772834668357,
      "grad_norm": 0.7094944715499878,
      "learning_rate": 2.161921685714685e-06,
      "loss": 0.692,
      "step": 24342
    },
    {
      "epoch": 0.701406096928485,
      "grad_norm": 0.7105748653411865,
      "learning_rate": 2.161537540913102e-06,
      "loss": 0.6847,
      "step": 24343
    },
    {
      "epoch": 0.7014349103901343,
      "grad_norm": 0.7039834260940552,
      "learning_rate": 2.161153420831689e-06,
      "loss": 0.6949,
      "step": 24344
    },
    {
      "epoch": 0.7014637238517836,
      "grad_norm": 0.6964978575706482,
      "learning_rate": 2.1607693254737944e-06,
      "loss": 0.6698,
      "step": 24345
    },
    {
      "epoch": 0.7014925373134329,
      "grad_norm": 0.7176089286804199,
      "learning_rate": 2.1603852548427586e-06,
      "loss": 0.6785,
      "step": 24346
    },
    {
      "epoch": 0.7015213507750822,
      "grad_norm": 0.7181714177131653,
      "learning_rate": 2.1600012089419297e-06,
      "loss": 0.7303,
      "step": 24347
    },
    {
      "epoch": 0.7015501642367314,
      "grad_norm": 0.7235158681869507,
      "learning_rate": 2.1596171877746487e-06,
      "loss": 0.7033,
      "step": 24348
    },
    {
      "epoch": 0.7015789776983807,
      "grad_norm": 0.734409511089325,
      "learning_rate": 2.159233191344266e-06,
      "loss": 0.6813,
      "step": 24349
    },
    {
      "epoch": 0.7016077911600299,
      "grad_norm": 0.731755793094635,
      "learning_rate": 2.1588492196541207e-06,
      "loss": 0.6972,
      "step": 24350
    },
    {
      "epoch": 0.7016366046216792,
      "grad_norm": 0.7273039221763611,
      "learning_rate": 2.15846527270756e-06,
      "loss": 0.6965,
      "step": 24351
    },
    {
      "epoch": 0.7016654180833285,
      "grad_norm": 0.716162919998169,
      "learning_rate": 2.1580813505079246e-06,
      "loss": 0.6948,
      "step": 24352
    },
    {
      "epoch": 0.7016942315449778,
      "grad_norm": 0.7047131657600403,
      "learning_rate": 2.1576974530585622e-06,
      "loss": 0.6789,
      "step": 24353
    },
    {
      "epoch": 0.7017230450066271,
      "grad_norm": 0.7374721765518188,
      "learning_rate": 2.1573135803628115e-06,
      "loss": 0.6838,
      "step": 24354
    },
    {
      "epoch": 0.7017518584682764,
      "grad_norm": 0.7236908674240112,
      "learning_rate": 2.156929732424019e-06,
      "loss": 0.6767,
      "step": 24355
    },
    {
      "epoch": 0.7017806719299257,
      "grad_norm": 0.729882538318634,
      "learning_rate": 2.1565459092455256e-06,
      "loss": 0.7288,
      "step": 24356
    },
    {
      "epoch": 0.701809485391575,
      "grad_norm": 0.71905118227005,
      "learning_rate": 2.156162110830676e-06,
      "loss": 0.6856,
      "step": 24357
    },
    {
      "epoch": 0.7018382988532242,
      "grad_norm": 0.705941915512085,
      "learning_rate": 2.1557783371828107e-06,
      "loss": 0.6859,
      "step": 24358
    },
    {
      "epoch": 0.7018671123148735,
      "grad_norm": 0.7119441032409668,
      "learning_rate": 2.1553945883052724e-06,
      "loss": 0.6827,
      "step": 24359
    },
    {
      "epoch": 0.7018959257765228,
      "grad_norm": 0.7024694681167603,
      "learning_rate": 2.155010864201404e-06,
      "loss": 0.6748,
      "step": 24360
    },
    {
      "epoch": 0.7019247392381721,
      "grad_norm": 0.7059592008590698,
      "learning_rate": 2.1546271648745483e-06,
      "loss": 0.6828,
      "step": 24361
    },
    {
      "epoch": 0.7019535526998214,
      "grad_norm": 0.7059249877929688,
      "learning_rate": 2.154243490328044e-06,
      "loss": 0.6898,
      "step": 24362
    },
    {
      "epoch": 0.7019823661614707,
      "grad_norm": 0.7313376069068909,
      "learning_rate": 2.153859840565236e-06,
      "loss": 0.6987,
      "step": 24363
    },
    {
      "epoch": 0.7020111796231199,
      "grad_norm": 0.7099308371543884,
      "learning_rate": 2.153476215589462e-06,
      "loss": 0.6965,
      "step": 24364
    },
    {
      "epoch": 0.7020399930847692,
      "grad_norm": 0.7123441100120544,
      "learning_rate": 2.153092615404065e-06,
      "loss": 0.7034,
      "step": 24365
    },
    {
      "epoch": 0.7020688065464185,
      "grad_norm": 0.7491445541381836,
      "learning_rate": 2.152709040012387e-06,
      "loss": 0.7168,
      "step": 24366
    },
    {
      "epoch": 0.7020976200080677,
      "grad_norm": 0.7066776752471924,
      "learning_rate": 2.152325489417765e-06,
      "loss": 0.6604,
      "step": 24367
    },
    {
      "epoch": 0.702126433469717,
      "grad_norm": 0.7014785408973694,
      "learning_rate": 2.1519419636235433e-06,
      "loss": 0.6812,
      "step": 24368
    },
    {
      "epoch": 0.7021552469313663,
      "grad_norm": 0.7417953014373779,
      "learning_rate": 2.151558462633058e-06,
      "loss": 0.704,
      "step": 24369
    },
    {
      "epoch": 0.7021840603930156,
      "grad_norm": 0.7048850655555725,
      "learning_rate": 2.1511749864496533e-06,
      "loss": 0.6513,
      "step": 24370
    },
    {
      "epoch": 0.7022128738546649,
      "grad_norm": 0.7045262455940247,
      "learning_rate": 2.1507915350766646e-06,
      "loss": 0.6797,
      "step": 24371
    },
    {
      "epoch": 0.7022416873163142,
      "grad_norm": 0.7144119739532471,
      "learning_rate": 2.1504081085174338e-06,
      "loss": 0.6843,
      "step": 24372
    },
    {
      "epoch": 0.7022705007779635,
      "grad_norm": 0.7175770998001099,
      "learning_rate": 2.1500247067752992e-06,
      "loss": 0.6904,
      "step": 24373
    },
    {
      "epoch": 0.7022993142396128,
      "grad_norm": 0.7095559239387512,
      "learning_rate": 2.149641329853603e-06,
      "loss": 0.6793,
      "step": 24374
    },
    {
      "epoch": 0.7023281277012621,
      "grad_norm": 0.7108833193778992,
      "learning_rate": 2.1492579777556794e-06,
      "loss": 0.6781,
      "step": 24375
    },
    {
      "epoch": 0.7023569411629114,
      "grad_norm": 0.7439630031585693,
      "learning_rate": 2.1488746504848705e-06,
      "loss": 0.6925,
      "step": 24376
    },
    {
      "epoch": 0.7023857546245605,
      "grad_norm": 0.7137324810028076,
      "learning_rate": 2.1484913480445122e-06,
      "loss": 0.6851,
      "step": 24377
    },
    {
      "epoch": 0.7024145680862098,
      "grad_norm": 0.7038294672966003,
      "learning_rate": 2.1481080704379453e-06,
      "loss": 0.694,
      "step": 24378
    },
    {
      "epoch": 0.7024433815478591,
      "grad_norm": 0.725746750831604,
      "learning_rate": 2.1477248176685046e-06,
      "loss": 0.683,
      "step": 24379
    },
    {
      "epoch": 0.7024721950095084,
      "grad_norm": 0.7525771856307983,
      "learning_rate": 2.147341589739531e-06,
      "loss": 0.6878,
      "step": 24380
    },
    {
      "epoch": 0.7025010084711577,
      "grad_norm": 0.8706995844841003,
      "learning_rate": 2.146958386654359e-06,
      "loss": 0.7178,
      "step": 24381
    },
    {
      "epoch": 0.702529821932807,
      "grad_norm": 0.7078317403793335,
      "learning_rate": 2.146575208416328e-06,
      "loss": 0.7094,
      "step": 24382
    },
    {
      "epoch": 0.7025586353944563,
      "grad_norm": 0.7228491306304932,
      "learning_rate": 2.146192055028775e-06,
      "loss": 0.6746,
      "step": 24383
    },
    {
      "epoch": 0.7025874488561056,
      "grad_norm": 0.7140562534332275,
      "learning_rate": 2.1458089264950355e-06,
      "loss": 0.6999,
      "step": 24384
    },
    {
      "epoch": 0.7026162623177549,
      "grad_norm": 0.7181406617164612,
      "learning_rate": 2.145425822818447e-06,
      "loss": 0.704,
      "step": 24385
    },
    {
      "epoch": 0.7026450757794042,
      "grad_norm": 0.7330302596092224,
      "learning_rate": 2.1450427440023457e-06,
      "loss": 0.7085,
      "step": 24386
    },
    {
      "epoch": 0.7026738892410535,
      "grad_norm": 0.7328001260757446,
      "learning_rate": 2.1446596900500703e-06,
      "loss": 0.692,
      "step": 24387
    },
    {
      "epoch": 0.7027027027027027,
      "grad_norm": 0.7064943909645081,
      "learning_rate": 2.1442766609649535e-06,
      "loss": 0.6987,
      "step": 24388
    },
    {
      "epoch": 0.702731516164352,
      "grad_norm": 0.7292013168334961,
      "learning_rate": 2.1438936567503333e-06,
      "loss": 0.706,
      "step": 24389
    },
    {
      "epoch": 0.7027603296260013,
      "grad_norm": 0.7151600122451782,
      "learning_rate": 2.1435106774095425e-06,
      "loss": 0.6969,
      "step": 24390
    },
    {
      "epoch": 0.7027891430876505,
      "grad_norm": 0.7155031561851501,
      "learning_rate": 2.14312772294592e-06,
      "loss": 0.6923,
      "step": 24391
    },
    {
      "epoch": 0.7028179565492998,
      "grad_norm": 0.7032210826873779,
      "learning_rate": 2.142744793362798e-06,
      "loss": 0.6931,
      "step": 24392
    },
    {
      "epoch": 0.7028467700109491,
      "grad_norm": 0.7104364633560181,
      "learning_rate": 2.1423618886635146e-06,
      "loss": 0.6822,
      "step": 24393
    },
    {
      "epoch": 0.7028755834725984,
      "grad_norm": 0.7095369100570679,
      "learning_rate": 2.1419790088514e-06,
      "loss": 0.7081,
      "step": 24394
    },
    {
      "epoch": 0.7029043969342477,
      "grad_norm": 0.7092453241348267,
      "learning_rate": 2.141596153929793e-06,
      "loss": 0.712,
      "step": 24395
    },
    {
      "epoch": 0.702933210395897,
      "grad_norm": 0.7240855097770691,
      "learning_rate": 2.1412133239020233e-06,
      "loss": 0.6614,
      "step": 24396
    },
    {
      "epoch": 0.7029620238575462,
      "grad_norm": 0.7142029404640198,
      "learning_rate": 2.140830518771431e-06,
      "loss": 0.6887,
      "step": 24397
    },
    {
      "epoch": 0.7029908373191955,
      "grad_norm": 0.7038455605506897,
      "learning_rate": 2.1404477385413447e-06,
      "loss": 0.7165,
      "step": 24398
    },
    {
      "epoch": 0.7030196507808448,
      "grad_norm": 0.6946907639503479,
      "learning_rate": 2.140064983215102e-06,
      "loss": 0.6938,
      "step": 24399
    },
    {
      "epoch": 0.7030484642424941,
      "grad_norm": 0.7103226780891418,
      "learning_rate": 2.1396822527960327e-06,
      "loss": 0.6874,
      "step": 24400
    },
    {
      "epoch": 0.7030772777041434,
      "grad_norm": 1.0893983840942383,
      "learning_rate": 2.139299547287474e-06,
      "loss": 0.6843,
      "step": 24401
    },
    {
      "epoch": 0.7031060911657927,
      "grad_norm": 0.7522521615028381,
      "learning_rate": 2.138916866692754e-06,
      "loss": 0.6952,
      "step": 24402
    },
    {
      "epoch": 0.703134904627442,
      "grad_norm": 0.713068425655365,
      "learning_rate": 2.1385342110152085e-06,
      "loss": 0.693,
      "step": 24403
    },
    {
      "epoch": 0.7031637180890912,
      "grad_norm": 0.7267213463783264,
      "learning_rate": 2.138151580258171e-06,
      "loss": 0.692,
      "step": 24404
    },
    {
      "epoch": 0.7031925315507405,
      "grad_norm": 0.713037371635437,
      "learning_rate": 2.1377689744249712e-06,
      "loss": 0.6652,
      "step": 24405
    },
    {
      "epoch": 0.7032213450123898,
      "grad_norm": 0.7116928100585938,
      "learning_rate": 2.1373863935189438e-06,
      "loss": 0.6488,
      "step": 24406
    },
    {
      "epoch": 0.703250158474039,
      "grad_norm": 0.696212649345398,
      "learning_rate": 2.1370038375434177e-06,
      "loss": 0.6678,
      "step": 24407
    },
    {
      "epoch": 0.7032789719356883,
      "grad_norm": 0.7002179622650146,
      "learning_rate": 2.136621306501726e-06,
      "loss": 0.7134,
      "step": 24408
    },
    {
      "epoch": 0.7033077853973376,
      "grad_norm": 0.7215494513511658,
      "learning_rate": 2.1362388003972005e-06,
      "loss": 0.7009,
      "step": 24409
    },
    {
      "epoch": 0.7033365988589869,
      "grad_norm": 0.7384467720985413,
      "learning_rate": 2.135856319233175e-06,
      "loss": 0.6904,
      "step": 24410
    },
    {
      "epoch": 0.7033654123206362,
      "grad_norm": 0.7191171050071716,
      "learning_rate": 2.135473863012975e-06,
      "loss": 0.6975,
      "step": 24411
    },
    {
      "epoch": 0.7033942257822855,
      "grad_norm": 0.7205935120582581,
      "learning_rate": 2.1350914317399363e-06,
      "loss": 0.696,
      "step": 24412
    },
    {
      "epoch": 0.7034230392439348,
      "grad_norm": 0.7303277254104614,
      "learning_rate": 2.1347090254173857e-06,
      "loss": 0.715,
      "step": 24413
    },
    {
      "epoch": 0.7034518527055841,
      "grad_norm": 0.7256940603256226,
      "learning_rate": 2.134326644048657e-06,
      "loss": 0.6736,
      "step": 24414
    },
    {
      "epoch": 0.7034806661672334,
      "grad_norm": 0.6971396803855896,
      "learning_rate": 2.133944287637076e-06,
      "loss": 0.7056,
      "step": 24415
    },
    {
      "epoch": 0.7035094796288827,
      "grad_norm": 0.7142674922943115,
      "learning_rate": 2.133561956185978e-06,
      "loss": 0.6975,
      "step": 24416
    },
    {
      "epoch": 0.7035382930905318,
      "grad_norm": 0.7378345131874084,
      "learning_rate": 2.133179649698688e-06,
      "loss": 0.6957,
      "step": 24417
    },
    {
      "epoch": 0.7035671065521811,
      "grad_norm": 0.7163604497909546,
      "learning_rate": 2.1327973681785396e-06,
      "loss": 0.6877,
      "step": 24418
    },
    {
      "epoch": 0.7035959200138304,
      "grad_norm": 0.69956374168396,
      "learning_rate": 2.1324151116288576e-06,
      "loss": 0.6794,
      "step": 24419
    },
    {
      "epoch": 0.7036247334754797,
      "grad_norm": 0.7350134253501892,
      "learning_rate": 2.132032880052974e-06,
      "loss": 0.6954,
      "step": 24420
    },
    {
      "epoch": 0.703653546937129,
      "grad_norm": 0.7278799414634705,
      "learning_rate": 2.1316506734542163e-06,
      "loss": 0.7099,
      "step": 24421
    },
    {
      "epoch": 0.7036823603987783,
      "grad_norm": 0.7446362376213074,
      "learning_rate": 2.131268491835916e-06,
      "loss": 0.718,
      "step": 24422
    },
    {
      "epoch": 0.7037111738604276,
      "grad_norm": 0.7475868463516235,
      "learning_rate": 2.130886335201398e-06,
      "loss": 0.7039,
      "step": 24423
    },
    {
      "epoch": 0.7037399873220769,
      "grad_norm": 0.7199313640594482,
      "learning_rate": 2.130504203553992e-06,
      "loss": 0.6886,
      "step": 24424
    },
    {
      "epoch": 0.7037688007837262,
      "grad_norm": 0.7289949059486389,
      "learning_rate": 2.1301220968970276e-06,
      "loss": 0.7246,
      "step": 24425
    },
    {
      "epoch": 0.7037976142453755,
      "grad_norm": 0.7217675447463989,
      "learning_rate": 2.1297400152338286e-06,
      "loss": 0.7177,
      "step": 24426
    },
    {
      "epoch": 0.7038264277070247,
      "grad_norm": 0.730342447757721,
      "learning_rate": 2.1293579585677277e-06,
      "loss": 0.7159,
      "step": 24427
    },
    {
      "epoch": 0.703855241168674,
      "grad_norm": 0.6980987787246704,
      "learning_rate": 2.1289759269020465e-06,
      "loss": 0.6692,
      "step": 24428
    },
    {
      "epoch": 0.7038840546303233,
      "grad_norm": 0.7041776776313782,
      "learning_rate": 2.1285939202401175e-06,
      "loss": 0.689,
      "step": 24429
    },
    {
      "epoch": 0.7039128680919726,
      "grad_norm": 0.7081274390220642,
      "learning_rate": 2.128211938585263e-06,
      "loss": 0.6926,
      "step": 24430
    },
    {
      "epoch": 0.7039416815536218,
      "grad_norm": 0.7016440033912659,
      "learning_rate": 2.127829981940814e-06,
      "loss": 0.688,
      "step": 24431
    },
    {
      "epoch": 0.7039704950152711,
      "grad_norm": 0.7407227754592896,
      "learning_rate": 2.1274480503100905e-06,
      "loss": 0.7178,
      "step": 24432
    },
    {
      "epoch": 0.7039993084769204,
      "grad_norm": 0.7202747464179993,
      "learning_rate": 2.1270661436964267e-06,
      "loss": 0.6887,
      "step": 24433
    },
    {
      "epoch": 0.7040281219385697,
      "grad_norm": 0.7003260254859924,
      "learning_rate": 2.1266842621031434e-06,
      "loss": 0.6676,
      "step": 24434
    },
    {
      "epoch": 0.704056935400219,
      "grad_norm": 0.7323969602584839,
      "learning_rate": 2.1263024055335698e-06,
      "loss": 0.6993,
      "step": 24435
    },
    {
      "epoch": 0.7040857488618683,
      "grad_norm": 0.7464050650596619,
      "learning_rate": 2.125920573991027e-06,
      "loss": 0.7019,
      "step": 24436
    },
    {
      "epoch": 0.7041145623235175,
      "grad_norm": 0.701213538646698,
      "learning_rate": 2.125538767478846e-06,
      "loss": 0.6648,
      "step": 24437
    },
    {
      "epoch": 0.7041433757851668,
      "grad_norm": 0.7178856730461121,
      "learning_rate": 2.1251569860003465e-06,
      "loss": 0.6856,
      "step": 24438
    },
    {
      "epoch": 0.7041721892468161,
      "grad_norm": 0.7094391584396362,
      "learning_rate": 2.1247752295588573e-06,
      "loss": 0.6745,
      "step": 24439
    },
    {
      "epoch": 0.7042010027084654,
      "grad_norm": 0.721714198589325,
      "learning_rate": 2.1243934981577005e-06,
      "loss": 0.6636,
      "step": 24440
    },
    {
      "epoch": 0.7042298161701147,
      "grad_norm": 0.7301375269889832,
      "learning_rate": 2.1240117918002014e-06,
      "loss": 0.7019,
      "step": 24441
    },
    {
      "epoch": 0.704258629631764,
      "grad_norm": 0.7264122366905212,
      "learning_rate": 2.1236301104896867e-06,
      "loss": 0.6779,
      "step": 24442
    },
    {
      "epoch": 0.7042874430934133,
      "grad_norm": 0.7239410877227783,
      "learning_rate": 2.123248454229476e-06,
      "loss": 0.7136,
      "step": 24443
    },
    {
      "epoch": 0.7043162565550625,
      "grad_norm": 0.7313833236694336,
      "learning_rate": 2.1228668230228966e-06,
      "loss": 0.705,
      "step": 24444
    },
    {
      "epoch": 0.7043450700167118,
      "grad_norm": 0.7469211220741272,
      "learning_rate": 2.12248521687327e-06,
      "loss": 0.7248,
      "step": 24445
    },
    {
      "epoch": 0.704373883478361,
      "grad_norm": 0.7268261313438416,
      "learning_rate": 2.122103635783923e-06,
      "loss": 0.6966,
      "step": 24446
    },
    {
      "epoch": 0.7044026969400103,
      "grad_norm": 0.7402873039245605,
      "learning_rate": 2.121722079758175e-06,
      "loss": 0.7089,
      "step": 24447
    },
    {
      "epoch": 0.7044315104016596,
      "grad_norm": 0.6958139538764954,
      "learning_rate": 2.1213405487993522e-06,
      "loss": 0.7147,
      "step": 24448
    },
    {
      "epoch": 0.7044603238633089,
      "grad_norm": 0.6980804204940796,
      "learning_rate": 2.1209590429107734e-06,
      "loss": 0.6815,
      "step": 24449
    },
    {
      "epoch": 0.7044891373249582,
      "grad_norm": 0.7253727316856384,
      "learning_rate": 2.1205775620957653e-06,
      "loss": 0.6892,
      "step": 24450
    },
    {
      "epoch": 0.7045179507866075,
      "grad_norm": 0.7206321358680725,
      "learning_rate": 2.1201961063576466e-06,
      "loss": 0.7093,
      "step": 24451
    },
    {
      "epoch": 0.7045467642482568,
      "grad_norm": 0.7122585773468018,
      "learning_rate": 2.119814675699743e-06,
      "loss": 0.686,
      "step": 24452
    },
    {
      "epoch": 0.7045755777099061,
      "grad_norm": 0.7208706736564636,
      "learning_rate": 2.119433270125372e-06,
      "loss": 0.6947,
      "step": 24453
    },
    {
      "epoch": 0.7046043911715554,
      "grad_norm": 0.6961820721626282,
      "learning_rate": 2.1190518896378603e-06,
      "loss": 0.6897,
      "step": 24454
    },
    {
      "epoch": 0.7046332046332047,
      "grad_norm": 0.7078885436058044,
      "learning_rate": 2.118670534240523e-06,
      "loss": 0.6658,
      "step": 24455
    },
    {
      "epoch": 0.704662018094854,
      "grad_norm": 0.7326499223709106,
      "learning_rate": 2.118289203936688e-06,
      "loss": 0.6807,
      "step": 24456
    },
    {
      "epoch": 0.7046908315565032,
      "grad_norm": 0.7288954854011536,
      "learning_rate": 2.117907898729672e-06,
      "loss": 0.7121,
      "step": 24457
    },
    {
      "epoch": 0.7047196450181524,
      "grad_norm": 0.7081519365310669,
      "learning_rate": 2.117526618622799e-06,
      "loss": 0.6838,
      "step": 24458
    },
    {
      "epoch": 0.7047484584798017,
      "grad_norm": 0.7142632007598877,
      "learning_rate": 2.117145363619386e-06,
      "loss": 0.685,
      "step": 24459
    },
    {
      "epoch": 0.704777271941451,
      "grad_norm": 0.7328436970710754,
      "learning_rate": 2.1167641337227567e-06,
      "loss": 0.6874,
      "step": 24460
    },
    {
      "epoch": 0.7048060854031003,
      "grad_norm": 0.7493531107902527,
      "learning_rate": 2.116382928936228e-06,
      "loss": 0.6999,
      "step": 24461
    },
    {
      "epoch": 0.7048348988647496,
      "grad_norm": 0.7359805703163147,
      "learning_rate": 2.1160017492631215e-06,
      "loss": 0.6974,
      "step": 24462
    },
    {
      "epoch": 0.7048637123263989,
      "grad_norm": 0.7124572396278381,
      "learning_rate": 2.115620594706758e-06,
      "loss": 0.6806,
      "step": 24463
    },
    {
      "epoch": 0.7048925257880482,
      "grad_norm": 0.7193517088890076,
      "learning_rate": 2.115239465270455e-06,
      "loss": 0.6787,
      "step": 24464
    },
    {
      "epoch": 0.7049213392496975,
      "grad_norm": 0.6988184452056885,
      "learning_rate": 2.1148583609575335e-06,
      "loss": 0.683,
      "step": 24465
    },
    {
      "epoch": 0.7049501527113468,
      "grad_norm": 0.7410684823989868,
      "learning_rate": 2.1144772817713106e-06,
      "loss": 0.7086,
      "step": 24466
    },
    {
      "epoch": 0.704978966172996,
      "grad_norm": 0.7051290273666382,
      "learning_rate": 2.1140962277151052e-06,
      "loss": 0.6945,
      "step": 24467
    },
    {
      "epoch": 0.7050077796346453,
      "grad_norm": 0.7205786108970642,
      "learning_rate": 2.113715198792237e-06,
      "loss": 0.6969,
      "step": 24468
    },
    {
      "epoch": 0.7050365930962946,
      "grad_norm": 0.7311135530471802,
      "learning_rate": 2.1133341950060264e-06,
      "loss": 0.6941,
      "step": 24469
    },
    {
      "epoch": 0.7050654065579439,
      "grad_norm": 0.6831105351448059,
      "learning_rate": 2.112953216359788e-06,
      "loss": 0.6611,
      "step": 24470
    },
    {
      "epoch": 0.7050942200195931,
      "grad_norm": 0.7195982336997986,
      "learning_rate": 2.112572262856842e-06,
      "loss": 0.69,
      "step": 24471
    },
    {
      "epoch": 0.7051230334812424,
      "grad_norm": 0.7169312238693237,
      "learning_rate": 2.1121913345005037e-06,
      "loss": 0.7015,
      "step": 24472
    },
    {
      "epoch": 0.7051518469428917,
      "grad_norm": 0.7070823311805725,
      "learning_rate": 2.1118104312940933e-06,
      "loss": 0.6831,
      "step": 24473
    },
    {
      "epoch": 0.705180660404541,
      "grad_norm": 0.7354649901390076,
      "learning_rate": 2.1114295532409264e-06,
      "loss": 0.7098,
      "step": 24474
    },
    {
      "epoch": 0.7052094738661903,
      "grad_norm": 0.9323503971099854,
      "learning_rate": 2.1110487003443213e-06,
      "loss": 0.6921,
      "step": 24475
    },
    {
      "epoch": 0.7052382873278396,
      "grad_norm": 0.7224454283714294,
      "learning_rate": 2.110667872607593e-06,
      "loss": 0.6918,
      "step": 24476
    },
    {
      "epoch": 0.7052671007894888,
      "grad_norm": 0.7169390320777893,
      "learning_rate": 2.1102870700340607e-06,
      "loss": 0.6882,
      "step": 24477
    },
    {
      "epoch": 0.7052959142511381,
      "grad_norm": 0.7203691005706787,
      "learning_rate": 2.1099062926270376e-06,
      "loss": 0.688,
      "step": 24478
    },
    {
      "epoch": 0.7053247277127874,
      "grad_norm": 0.7163476943969727,
      "learning_rate": 2.1095255403898416e-06,
      "loss": 0.6936,
      "step": 24479
    },
    {
      "epoch": 0.7053535411744367,
      "grad_norm": 0.7365590929985046,
      "learning_rate": 2.1091448133257884e-06,
      "loss": 0.7158,
      "step": 24480
    },
    {
      "epoch": 0.705382354636086,
      "grad_norm": 0.719213604927063,
      "learning_rate": 2.108764111438197e-06,
      "loss": 0.693,
      "step": 24481
    },
    {
      "epoch": 0.7054111680977353,
      "grad_norm": 0.7149966955184937,
      "learning_rate": 2.1083834347303773e-06,
      "loss": 0.6986,
      "step": 24482
    },
    {
      "epoch": 0.7054399815593846,
      "grad_norm": 0.7183074355125427,
      "learning_rate": 2.1080027832056478e-06,
      "loss": 0.6838,
      "step": 24483
    },
    {
      "epoch": 0.7054687950210339,
      "grad_norm": 0.7043167948722839,
      "learning_rate": 2.1076221568673243e-06,
      "loss": 0.6768,
      "step": 24484
    },
    {
      "epoch": 0.7054976084826831,
      "grad_norm": 0.7239335775375366,
      "learning_rate": 2.1072415557187193e-06,
      "loss": 0.6873,
      "step": 24485
    },
    {
      "epoch": 0.7055264219443323,
      "grad_norm": 0.709929883480072,
      "learning_rate": 2.10686097976315e-06,
      "loss": 0.6751,
      "step": 24486
    },
    {
      "epoch": 0.7055552354059816,
      "grad_norm": 0.7133051156997681,
      "learning_rate": 2.1064804290039274e-06,
      "loss": 0.7017,
      "step": 24487
    },
    {
      "epoch": 0.7055840488676309,
      "grad_norm": 0.7137471437454224,
      "learning_rate": 2.1060999034443706e-06,
      "loss": 0.6804,
      "step": 24488
    },
    {
      "epoch": 0.7056128623292802,
      "grad_norm": 0.7328081727027893,
      "learning_rate": 2.105719403087789e-06,
      "loss": 0.6916,
      "step": 24489
    },
    {
      "epoch": 0.7056416757909295,
      "grad_norm": 0.7124301791191101,
      "learning_rate": 2.1053389279374985e-06,
      "loss": 0.6817,
      "step": 24490
    },
    {
      "epoch": 0.7056704892525788,
      "grad_norm": 0.7222074866294861,
      "learning_rate": 2.10495847799681e-06,
      "loss": 0.6792,
      "step": 24491
    },
    {
      "epoch": 0.7056993027142281,
      "grad_norm": 0.730484664440155,
      "learning_rate": 2.1045780532690423e-06,
      "loss": 0.6867,
      "step": 24492
    },
    {
      "epoch": 0.7057281161758774,
      "grad_norm": 0.7196969389915466,
      "learning_rate": 2.1041976537575036e-06,
      "loss": 0.7036,
      "step": 24493
    },
    {
      "epoch": 0.7057569296375267,
      "grad_norm": 0.7290801405906677,
      "learning_rate": 2.103817279465511e-06,
      "loss": 0.6866,
      "step": 24494
    },
    {
      "epoch": 0.705785743099176,
      "grad_norm": 0.7335218191146851,
      "learning_rate": 2.1034369303963724e-06,
      "loss": 0.6929,
      "step": 24495
    },
    {
      "epoch": 0.7058145565608253,
      "grad_norm": 0.7470656037330627,
      "learning_rate": 2.103056606553405e-06,
      "loss": 0.7274,
      "step": 24496
    },
    {
      "epoch": 0.7058433700224745,
      "grad_norm": 0.7118573188781738,
      "learning_rate": 2.102676307939916e-06,
      "loss": 0.6849,
      "step": 24497
    },
    {
      "epoch": 0.7058721834841237,
      "grad_norm": 0.7301774024963379,
      "learning_rate": 2.1022960345592226e-06,
      "loss": 0.7065,
      "step": 24498
    },
    {
      "epoch": 0.705900996945773,
      "grad_norm": 0.699296236038208,
      "learning_rate": 2.1019157864146316e-06,
      "loss": 0.6905,
      "step": 24499
    },
    {
      "epoch": 0.7059298104074223,
      "grad_norm": 0.6935881972312927,
      "learning_rate": 2.1015355635094573e-06,
      "loss": 0.6788,
      "step": 24500
    },
    {
      "epoch": 0.7059586238690716,
      "grad_norm": 0.6960645318031311,
      "learning_rate": 2.1011553658470135e-06,
      "loss": 0.6575,
      "step": 24501
    },
    {
      "epoch": 0.7059874373307209,
      "grad_norm": 0.7247999906539917,
      "learning_rate": 2.1007751934306057e-06,
      "loss": 0.6726,
      "step": 24502
    },
    {
      "epoch": 0.7060162507923702,
      "grad_norm": 0.7147618532180786,
      "learning_rate": 2.1003950462635487e-06,
      "loss": 0.7098,
      "step": 24503
    },
    {
      "epoch": 0.7060450642540195,
      "grad_norm": 0.7260116338729858,
      "learning_rate": 2.100014924349152e-06,
      "loss": 0.6889,
      "step": 24504
    },
    {
      "epoch": 0.7060738777156688,
      "grad_norm": 0.698300302028656,
      "learning_rate": 2.099634827690728e-06,
      "loss": 0.6752,
      "step": 24505
    },
    {
      "epoch": 0.706102691177318,
      "grad_norm": 0.7348496913909912,
      "learning_rate": 2.0992547562915836e-06,
      "loss": 0.694,
      "step": 24506
    },
    {
      "epoch": 0.7061315046389673,
      "grad_norm": 0.7049810290336609,
      "learning_rate": 2.0988747101550327e-06,
      "loss": 0.6832,
      "step": 24507
    },
    {
      "epoch": 0.7061603181006166,
      "grad_norm": 0.7096641659736633,
      "learning_rate": 2.098494689284381e-06,
      "loss": 0.6845,
      "step": 24508
    },
    {
      "epoch": 0.7061891315622659,
      "grad_norm": 0.7557408809661865,
      "learning_rate": 2.0981146936829416e-06,
      "loss": 0.7101,
      "step": 24509
    },
    {
      "epoch": 0.7062179450239152,
      "grad_norm": 0.7570043802261353,
      "learning_rate": 2.097734723354021e-06,
      "loss": 0.6986,
      "step": 24510
    },
    {
      "epoch": 0.7062467584855645,
      "grad_norm": 0.7174320220947266,
      "learning_rate": 2.0973547783009314e-06,
      "loss": 0.7028,
      "step": 24511
    },
    {
      "epoch": 0.7062755719472137,
      "grad_norm": 0.7330262660980225,
      "learning_rate": 2.0969748585269782e-06,
      "loss": 0.7007,
      "step": 24512
    },
    {
      "epoch": 0.706304385408863,
      "grad_norm": 0.7098080515861511,
      "learning_rate": 2.0965949640354737e-06,
      "loss": 0.6952,
      "step": 24513
    },
    {
      "epoch": 0.7063331988705123,
      "grad_norm": 0.7415139675140381,
      "learning_rate": 2.0962150948297233e-06,
      "loss": 0.706,
      "step": 24514
    },
    {
      "epoch": 0.7063620123321616,
      "grad_norm": 0.7113625407218933,
      "learning_rate": 2.095835250913037e-06,
      "loss": 0.6614,
      "step": 24515
    },
    {
      "epoch": 0.7063908257938109,
      "grad_norm": 0.7116649746894836,
      "learning_rate": 2.0954554322887216e-06,
      "loss": 0.7034,
      "step": 24516
    },
    {
      "epoch": 0.7064196392554601,
      "grad_norm": 0.7066122889518738,
      "learning_rate": 2.0950756389600885e-06,
      "loss": 0.6894,
      "step": 24517
    },
    {
      "epoch": 0.7064484527171094,
      "grad_norm": 0.7073690295219421,
      "learning_rate": 2.0946958709304404e-06,
      "loss": 0.7006,
      "step": 24518
    },
    {
      "epoch": 0.7064772661787587,
      "grad_norm": 0.7345945835113525,
      "learning_rate": 2.0943161282030893e-06,
      "loss": 0.6934,
      "step": 24519
    },
    {
      "epoch": 0.706506079640408,
      "grad_norm": 0.7138713002204895,
      "learning_rate": 2.093936410781338e-06,
      "loss": 0.6943,
      "step": 24520
    },
    {
      "epoch": 0.7065348931020573,
      "grad_norm": 0.7244747877120972,
      "learning_rate": 2.093556718668496e-06,
      "loss": 0.6876,
      "step": 24521
    },
    {
      "epoch": 0.7065637065637066,
      "grad_norm": 0.7330285906791687,
      "learning_rate": 2.093177051867871e-06,
      "loss": 0.6933,
      "step": 24522
    },
    {
      "epoch": 0.7065925200253559,
      "grad_norm": 0.7353419661521912,
      "learning_rate": 2.092797410382766e-06,
      "loss": 0.7026,
      "step": 24523
    },
    {
      "epoch": 0.7066213334870052,
      "grad_norm": 0.7373613119125366,
      "learning_rate": 2.092417794216492e-06,
      "loss": 0.6914,
      "step": 24524
    },
    {
      "epoch": 0.7066501469486544,
      "grad_norm": 0.7144920229911804,
      "learning_rate": 2.09203820337235e-06,
      "loss": 0.7032,
      "step": 24525
    },
    {
      "epoch": 0.7066789604103036,
      "grad_norm": 0.7122079133987427,
      "learning_rate": 2.0916586378536504e-06,
      "loss": 0.6736,
      "step": 24526
    },
    {
      "epoch": 0.7067077738719529,
      "grad_norm": 0.7175590991973877,
      "learning_rate": 2.091279097663693e-06,
      "loss": 0.6815,
      "step": 24527
    },
    {
      "epoch": 0.7067365873336022,
      "grad_norm": 0.6937167048454285,
      "learning_rate": 2.090899582805791e-06,
      "loss": 0.6877,
      "step": 24528
    },
    {
      "epoch": 0.7067654007952515,
      "grad_norm": 0.7403408288955688,
      "learning_rate": 2.0905200932832437e-06,
      "loss": 0.7171,
      "step": 24529
    },
    {
      "epoch": 0.7067942142569008,
      "grad_norm": 0.7699558734893799,
      "learning_rate": 2.09014062909936e-06,
      "loss": 0.7106,
      "step": 24530
    },
    {
      "epoch": 0.7068230277185501,
      "grad_norm": 0.7319220304489136,
      "learning_rate": 2.08976119025744e-06,
      "loss": 0.6949,
      "step": 24531
    },
    {
      "epoch": 0.7068518411801994,
      "grad_norm": 0.7103826403617859,
      "learning_rate": 2.0893817767607934e-06,
      "loss": 0.6987,
      "step": 24532
    },
    {
      "epoch": 0.7068806546418487,
      "grad_norm": 0.7350884675979614,
      "learning_rate": 2.08900238861272e-06,
      "loss": 0.6845,
      "step": 24533
    },
    {
      "epoch": 0.706909468103498,
      "grad_norm": 0.7184628844261169,
      "learning_rate": 2.0886230258165273e-06,
      "loss": 0.7091,
      "step": 24534
    },
    {
      "epoch": 0.7069382815651473,
      "grad_norm": 0.7113044261932373,
      "learning_rate": 2.0882436883755162e-06,
      "loss": 0.7102,
      "step": 24535
    },
    {
      "epoch": 0.7069670950267966,
      "grad_norm": 0.7266822457313538,
      "learning_rate": 2.087864376292994e-06,
      "loss": 0.7148,
      "step": 24536
    },
    {
      "epoch": 0.7069959084884458,
      "grad_norm": 0.7118000984191895,
      "learning_rate": 2.0874850895722596e-06,
      "loss": 0.6946,
      "step": 24537
    },
    {
      "epoch": 0.7070247219500951,
      "grad_norm": 0.7215744853019714,
      "learning_rate": 2.087105828216619e-06,
      "loss": 0.6813,
      "step": 24538
    },
    {
      "epoch": 0.7070535354117443,
      "grad_norm": 0.7054808735847473,
      "learning_rate": 2.086726592229375e-06,
      "loss": 0.6772,
      "step": 24539
    },
    {
      "epoch": 0.7070823488733936,
      "grad_norm": 0.7474992871284485,
      "learning_rate": 2.0863473816138313e-06,
      "loss": 0.7324,
      "step": 24540
    },
    {
      "epoch": 0.7071111623350429,
      "grad_norm": 0.694837212562561,
      "learning_rate": 2.085968196373288e-06,
      "loss": 0.689,
      "step": 24541
    },
    {
      "epoch": 0.7071399757966922,
      "grad_norm": 0.7378778457641602,
      "learning_rate": 2.085589036511049e-06,
      "loss": 0.6979,
      "step": 24542
    },
    {
      "epoch": 0.7071687892583415,
      "grad_norm": 0.7178202271461487,
      "learning_rate": 2.085209902030418e-06,
      "loss": 0.6742,
      "step": 24543
    },
    {
      "epoch": 0.7071976027199908,
      "grad_norm": 0.7288993000984192,
      "learning_rate": 2.084830792934692e-06,
      "loss": 0.6815,
      "step": 24544
    },
    {
      "epoch": 0.7072264161816401,
      "grad_norm": 0.7082340121269226,
      "learning_rate": 2.0844517092271786e-06,
      "loss": 0.6896,
      "step": 24545
    },
    {
      "epoch": 0.7072552296432894,
      "grad_norm": 0.7388027310371399,
      "learning_rate": 2.0840726509111747e-06,
      "loss": 0.7174,
      "step": 24546
    },
    {
      "epoch": 0.7072840431049386,
      "grad_norm": 0.7074882388114929,
      "learning_rate": 2.0836936179899842e-06,
      "loss": 0.6918,
      "step": 24547
    },
    {
      "epoch": 0.7073128565665879,
      "grad_norm": 0.7200679183006287,
      "learning_rate": 2.0833146104669056e-06,
      "loss": 0.7026,
      "step": 24548
    },
    {
      "epoch": 0.7073416700282372,
      "grad_norm": 0.6955813765525818,
      "learning_rate": 2.082935628345243e-06,
      "loss": 0.6594,
      "step": 24549
    },
    {
      "epoch": 0.7073704834898865,
      "grad_norm": 0.7152919769287109,
      "learning_rate": 2.082556671628293e-06,
      "loss": 0.7067,
      "step": 24550
    },
    {
      "epoch": 0.7073992969515358,
      "grad_norm": 0.7273561954498291,
      "learning_rate": 2.0821777403193584e-06,
      "loss": 0.6961,
      "step": 24551
    },
    {
      "epoch": 0.707428110413185,
      "grad_norm": 0.7248396873474121,
      "learning_rate": 2.081798834421739e-06,
      "loss": 0.6796,
      "step": 24552
    },
    {
      "epoch": 0.7074569238748343,
      "grad_norm": 0.7387272119522095,
      "learning_rate": 2.081419953938736e-06,
      "loss": 0.7049,
      "step": 24553
    },
    {
      "epoch": 0.7074857373364836,
      "grad_norm": 0.7205221652984619,
      "learning_rate": 2.081041098873646e-06,
      "loss": 0.6993,
      "step": 24554
    },
    {
      "epoch": 0.7075145507981329,
      "grad_norm": 0.715965986251831,
      "learning_rate": 2.0806622692297717e-06,
      "loss": 0.6664,
      "step": 24555
    },
    {
      "epoch": 0.7075433642597821,
      "grad_norm": 0.7344255447387695,
      "learning_rate": 2.08028346501041e-06,
      "loss": 0.6768,
      "step": 24556
    },
    {
      "epoch": 0.7075721777214314,
      "grad_norm": 0.7046992182731628,
      "learning_rate": 2.079904686218861e-06,
      "loss": 0.6817,
      "step": 24557
    },
    {
      "epoch": 0.7076009911830807,
      "grad_norm": 0.7151898145675659,
      "learning_rate": 2.0795259328584224e-06,
      "loss": 0.6675,
      "step": 24558
    },
    {
      "epoch": 0.70762980464473,
      "grad_norm": 0.7057310938835144,
      "learning_rate": 2.079147204932393e-06,
      "loss": 0.6881,
      "step": 24559
    },
    {
      "epoch": 0.7076586181063793,
      "grad_norm": 0.7178118228912354,
      "learning_rate": 2.078768502444074e-06,
      "loss": 0.6607,
      "step": 24560
    },
    {
      "epoch": 0.7076874315680286,
      "grad_norm": 0.7149766683578491,
      "learning_rate": 2.078389825396759e-06,
      "loss": 0.7039,
      "step": 24561
    },
    {
      "epoch": 0.7077162450296779,
      "grad_norm": 0.7153722047805786,
      "learning_rate": 2.0780111737937495e-06,
      "loss": 0.7164,
      "step": 24562
    },
    {
      "epoch": 0.7077450584913272,
      "grad_norm": 0.7276490926742554,
      "learning_rate": 2.077632547638341e-06,
      "loss": 0.7085,
      "step": 24563
    },
    {
      "epoch": 0.7077738719529765,
      "grad_norm": 0.7148424386978149,
      "learning_rate": 2.0772539469338305e-06,
      "loss": 0.7032,
      "step": 24564
    },
    {
      "epoch": 0.7078026854146258,
      "grad_norm": 0.7026948928833008,
      "learning_rate": 2.076875371683517e-06,
      "loss": 0.6767,
      "step": 24565
    },
    {
      "epoch": 0.707831498876275,
      "grad_norm": 0.7270007729530334,
      "learning_rate": 2.076496821890699e-06,
      "loss": 0.7031,
      "step": 24566
    },
    {
      "epoch": 0.7078603123379242,
      "grad_norm": 0.7251538038253784,
      "learning_rate": 2.076118297558669e-06,
      "loss": 0.7007,
      "step": 24567
    },
    {
      "epoch": 0.7078891257995735,
      "grad_norm": 0.7266044020652771,
      "learning_rate": 2.075739798690728e-06,
      "loss": 0.7089,
      "step": 24568
    },
    {
      "epoch": 0.7079179392612228,
      "grad_norm": 0.7261672019958496,
      "learning_rate": 2.0753613252901684e-06,
      "loss": 0.7112,
      "step": 24569
    },
    {
      "epoch": 0.7079467527228721,
      "grad_norm": 0.7174795269966125,
      "learning_rate": 2.0749828773602898e-06,
      "loss": 0.7116,
      "step": 24570
    },
    {
      "epoch": 0.7079755661845214,
      "grad_norm": 0.7204919457435608,
      "learning_rate": 2.074604454904384e-06,
      "loss": 0.7143,
      "step": 24571
    },
    {
      "epoch": 0.7080043796461707,
      "grad_norm": 0.7177718877792358,
      "learning_rate": 2.074226057925752e-06,
      "loss": 0.7252,
      "step": 24572
    },
    {
      "epoch": 0.70803319310782,
      "grad_norm": 1.7960938215255737,
      "learning_rate": 2.0738476864276834e-06,
      "loss": 0.7065,
      "step": 24573
    },
    {
      "epoch": 0.7080620065694693,
      "grad_norm": 0.7085882425308228,
      "learning_rate": 2.0734693404134786e-06,
      "loss": 0.6681,
      "step": 24574
    },
    {
      "epoch": 0.7080908200311186,
      "grad_norm": 0.6938328742980957,
      "learning_rate": 2.0730910198864274e-06,
      "loss": 0.69,
      "step": 24575
    },
    {
      "epoch": 0.7081196334927679,
      "grad_norm": 0.713084876537323,
      "learning_rate": 2.0727127248498303e-06,
      "loss": 0.694,
      "step": 24576
    },
    {
      "epoch": 0.7081484469544171,
      "grad_norm": 0.694017767906189,
      "learning_rate": 2.072334455306978e-06,
      "loss": 0.6903,
      "step": 24577
    },
    {
      "epoch": 0.7081772604160664,
      "grad_norm": 0.691087007522583,
      "learning_rate": 2.0719562112611673e-06,
      "loss": 0.6854,
      "step": 24578
    },
    {
      "epoch": 0.7082060738777156,
      "grad_norm": 0.729677677154541,
      "learning_rate": 2.0715779927156895e-06,
      "loss": 0.6811,
      "step": 24579
    },
    {
      "epoch": 0.7082348873393649,
      "grad_norm": 0.7149507403373718,
      "learning_rate": 2.0711997996738397e-06,
      "loss": 0.7047,
      "step": 24580
    },
    {
      "epoch": 0.7082637008010142,
      "grad_norm": 0.7591188549995422,
      "learning_rate": 2.070821632138914e-06,
      "loss": 0.7088,
      "step": 24581
    },
    {
      "epoch": 0.7082925142626635,
      "grad_norm": 0.727938175201416,
      "learning_rate": 2.0704434901142018e-06,
      "loss": 0.6852,
      "step": 24582
    },
    {
      "epoch": 0.7083213277243128,
      "grad_norm": 0.7214000821113586,
      "learning_rate": 2.070065373603e-06,
      "loss": 0.702,
      "step": 24583
    },
    {
      "epoch": 0.7083501411859621,
      "grad_norm": 0.7110441327095032,
      "learning_rate": 2.0696872826085988e-06,
      "loss": 0.6683,
      "step": 24584
    },
    {
      "epoch": 0.7083789546476114,
      "grad_norm": 0.6947131156921387,
      "learning_rate": 2.069309217134294e-06,
      "loss": 0.6846,
      "step": 24585
    },
    {
      "epoch": 0.7084077681092606,
      "grad_norm": 0.7209503054618835,
      "learning_rate": 2.068931177183374e-06,
      "loss": 0.6921,
      "step": 24586
    },
    {
      "epoch": 0.7084365815709099,
      "grad_norm": 0.7084079384803772,
      "learning_rate": 2.068553162759133e-06,
      "loss": 0.678,
      "step": 24587
    },
    {
      "epoch": 0.7084653950325592,
      "grad_norm": 0.7252660393714905,
      "learning_rate": 2.0681751738648643e-06,
      "loss": 0.7182,
      "step": 24588
    },
    {
      "epoch": 0.7084942084942085,
      "grad_norm": 0.7181573510169983,
      "learning_rate": 2.0677972105038603e-06,
      "loss": 0.692,
      "step": 24589
    },
    {
      "epoch": 0.7085230219558578,
      "grad_norm": 0.7382926344871521,
      "learning_rate": 2.06741927267941e-06,
      "loss": 0.6983,
      "step": 24590
    },
    {
      "epoch": 0.7085518354175071,
      "grad_norm": 0.7350069284439087,
      "learning_rate": 2.067041360394808e-06,
      "loss": 0.6719,
      "step": 24591
    },
    {
      "epoch": 0.7085806488791564,
      "grad_norm": 0.7163324952125549,
      "learning_rate": 2.0666634736533424e-06,
      "loss": 0.7027,
      "step": 24592
    },
    {
      "epoch": 0.7086094623408056,
      "grad_norm": 0.7184645533561707,
      "learning_rate": 2.0662856124583074e-06,
      "loss": 0.6729,
      "step": 24593
    },
    {
      "epoch": 0.7086382758024549,
      "grad_norm": 0.7123807668685913,
      "learning_rate": 2.0659077768129897e-06,
      "loss": 0.6976,
      "step": 24594
    },
    {
      "epoch": 0.7086670892641042,
      "grad_norm": 0.7058371901512146,
      "learning_rate": 2.0655299667206845e-06,
      "loss": 0.6782,
      "step": 24595
    },
    {
      "epoch": 0.7086959027257534,
      "grad_norm": 0.7267059683799744,
      "learning_rate": 2.0651521821846783e-06,
      "loss": 0.6794,
      "step": 24596
    },
    {
      "epoch": 0.7087247161874027,
      "grad_norm": 0.7234236598014832,
      "learning_rate": 2.0647744232082627e-06,
      "loss": 0.6896,
      "step": 24597
    },
    {
      "epoch": 0.708753529649052,
      "grad_norm": 0.7339318990707397,
      "learning_rate": 2.0643966897947293e-06,
      "loss": 0.7156,
      "step": 24598
    },
    {
      "epoch": 0.7087823431107013,
      "grad_norm": 0.7209595441818237,
      "learning_rate": 2.064018981947364e-06,
      "loss": 0.7017,
      "step": 24599
    },
    {
      "epoch": 0.7088111565723506,
      "grad_norm": 0.6934888958930969,
      "learning_rate": 2.0636412996694597e-06,
      "loss": 0.6739,
      "step": 24600
    },
    {
      "epoch": 0.7088399700339999,
      "grad_norm": 0.6992164850234985,
      "learning_rate": 2.0632636429643053e-06,
      "loss": 0.6865,
      "step": 24601
    },
    {
      "epoch": 0.7088687834956492,
      "grad_norm": 0.6958311200141907,
      "learning_rate": 2.062886011835187e-06,
      "loss": 0.684,
      "step": 24602
    },
    {
      "epoch": 0.7088975969572985,
      "grad_norm": 0.7123088240623474,
      "learning_rate": 2.0625084062853963e-06,
      "loss": 0.6801,
      "step": 24603
    },
    {
      "epoch": 0.7089264104189478,
      "grad_norm": 0.7261667847633362,
      "learning_rate": 2.062130826318222e-06,
      "loss": 0.7094,
      "step": 24604
    },
    {
      "epoch": 0.7089552238805971,
      "grad_norm": 0.7204053997993469,
      "learning_rate": 2.06175327193695e-06,
      "loss": 0.6813,
      "step": 24605
    },
    {
      "epoch": 0.7089840373422462,
      "grad_norm": 0.6991671919822693,
      "learning_rate": 2.0613757431448715e-06,
      "loss": 0.7022,
      "step": 24606
    },
    {
      "epoch": 0.7090128508038955,
      "grad_norm": 0.7378076910972595,
      "learning_rate": 2.060998239945271e-06,
      "loss": 0.7062,
      "step": 24607
    },
    {
      "epoch": 0.7090416642655448,
      "grad_norm": 0.6993286609649658,
      "learning_rate": 2.0606207623414398e-06,
      "loss": 0.684,
      "step": 24608
    },
    {
      "epoch": 0.7090704777271941,
      "grad_norm": 0.7023473381996155,
      "learning_rate": 2.0602433103366617e-06,
      "loss": 0.7032,
      "step": 24609
    },
    {
      "epoch": 0.7090992911888434,
      "grad_norm": 0.7181193232536316,
      "learning_rate": 2.0598658839342265e-06,
      "loss": 0.693,
      "step": 24610
    },
    {
      "epoch": 0.7091281046504927,
      "grad_norm": 0.7251780033111572,
      "learning_rate": 2.0594884831374177e-06,
      "loss": 0.6927,
      "step": 24611
    },
    {
      "epoch": 0.709156918112142,
      "grad_norm": 0.7356610298156738,
      "learning_rate": 2.0591111079495287e-06,
      "loss": 0.7232,
      "step": 24612
    },
    {
      "epoch": 0.7091857315737913,
      "grad_norm": 0.7231043577194214,
      "learning_rate": 2.05873375837384e-06,
      "loss": 0.6822,
      "step": 24613
    },
    {
      "epoch": 0.7092145450354406,
      "grad_norm": 0.7130229473114014,
      "learning_rate": 2.0583564344136415e-06,
      "loss": 0.6802,
      "step": 24614
    },
    {
      "epoch": 0.7092433584970899,
      "grad_norm": 0.7554346323013306,
      "learning_rate": 2.057979136072216e-06,
      "loss": 0.6768,
      "step": 24615
    },
    {
      "epoch": 0.7092721719587392,
      "grad_norm": 0.7287392020225525,
      "learning_rate": 2.0576018633528537e-06,
      "loss": 0.6996,
      "step": 24616
    },
    {
      "epoch": 0.7093009854203884,
      "grad_norm": 0.7147942781448364,
      "learning_rate": 2.0572246162588354e-06,
      "loss": 0.6674,
      "step": 24617
    },
    {
      "epoch": 0.7093297988820377,
      "grad_norm": 0.7182298898696899,
      "learning_rate": 2.0568473947934498e-06,
      "loss": 0.7031,
      "step": 24618
    },
    {
      "epoch": 0.709358612343687,
      "grad_norm": 0.7235636711120605,
      "learning_rate": 2.056470198959982e-06,
      "loss": 0.6764,
      "step": 24619
    },
    {
      "epoch": 0.7093874258053362,
      "grad_norm": 0.7044389247894287,
      "learning_rate": 2.0560930287617158e-06,
      "loss": 0.6753,
      "step": 24620
    },
    {
      "epoch": 0.7094162392669855,
      "grad_norm": 0.7063313722610474,
      "learning_rate": 2.0557158842019374e-06,
      "loss": 0.685,
      "step": 24621
    },
    {
      "epoch": 0.7094450527286348,
      "grad_norm": 0.7216196656227112,
      "learning_rate": 2.055338765283929e-06,
      "loss": 0.697,
      "step": 24622
    },
    {
      "epoch": 0.7094738661902841,
      "grad_norm": 0.7554308176040649,
      "learning_rate": 2.054961672010977e-06,
      "loss": 0.6877,
      "step": 24623
    },
    {
      "epoch": 0.7095026796519334,
      "grad_norm": 0.7155599594116211,
      "learning_rate": 2.0545846043863637e-06,
      "loss": 0.6815,
      "step": 24624
    },
    {
      "epoch": 0.7095314931135827,
      "grad_norm": 0.7116370797157288,
      "learning_rate": 2.054207562413376e-06,
      "loss": 0.6862,
      "step": 24625
    },
    {
      "epoch": 0.709560306575232,
      "grad_norm": 0.7226035594940186,
      "learning_rate": 2.0538305460952944e-06,
      "loss": 0.6827,
      "step": 24626
    },
    {
      "epoch": 0.7095891200368812,
      "grad_norm": 0.7128154039382935,
      "learning_rate": 2.0534535554354055e-06,
      "loss": 0.7017,
      "step": 24627
    },
    {
      "epoch": 0.7096179334985305,
      "grad_norm": 0.7412245869636536,
      "learning_rate": 2.053076590436989e-06,
      "loss": 0.6875,
      "step": 24628
    },
    {
      "epoch": 0.7096467469601798,
      "grad_norm": 0.7258695960044861,
      "learning_rate": 2.0526996511033308e-06,
      "loss": 0.7157,
      "step": 24629
    },
    {
      "epoch": 0.7096755604218291,
      "grad_norm": 0.7146799564361572,
      "learning_rate": 2.0523227374377107e-06,
      "loss": 0.6775,
      "step": 24630
    },
    {
      "epoch": 0.7097043738834784,
      "grad_norm": 0.717501699924469,
      "learning_rate": 2.0519458494434154e-06,
      "loss": 0.6721,
      "step": 24631
    },
    {
      "epoch": 0.7097331873451277,
      "grad_norm": 0.7213974595069885,
      "learning_rate": 2.0515689871237225e-06,
      "loss": 0.7006,
      "step": 24632
    },
    {
      "epoch": 0.7097620008067769,
      "grad_norm": 0.740943968296051,
      "learning_rate": 2.051192150481918e-06,
      "loss": 0.6856,
      "step": 24633
    },
    {
      "epoch": 0.7097908142684262,
      "grad_norm": 0.7208804488182068,
      "learning_rate": 2.0508153395212808e-06,
      "loss": 0.7004,
      "step": 24634
    },
    {
      "epoch": 0.7098196277300755,
      "grad_norm": 0.7600710391998291,
      "learning_rate": 2.050438554245094e-06,
      "loss": 0.7074,
      "step": 24635
    },
    {
      "epoch": 0.7098484411917247,
      "grad_norm": 0.7075461745262146,
      "learning_rate": 2.0500617946566383e-06,
      "loss": 0.6854,
      "step": 24636
    },
    {
      "epoch": 0.709877254653374,
      "grad_norm": 0.7249077558517456,
      "learning_rate": 2.0496850607591974e-06,
      "loss": 0.7002,
      "step": 24637
    },
    {
      "epoch": 0.7099060681150233,
      "grad_norm": 0.7254974246025085,
      "learning_rate": 2.0493083525560486e-06,
      "loss": 0.703,
      "step": 24638
    },
    {
      "epoch": 0.7099348815766726,
      "grad_norm": 0.7145038843154907,
      "learning_rate": 2.0489316700504767e-06,
      "loss": 0.6709,
      "step": 24639
    },
    {
      "epoch": 0.7099636950383219,
      "grad_norm": 0.7295873165130615,
      "learning_rate": 2.0485550132457576e-06,
      "loss": 0.6956,
      "step": 24640
    },
    {
      "epoch": 0.7099925084999712,
      "grad_norm": 0.7728344798088074,
      "learning_rate": 2.0481783821451745e-06,
      "loss": 0.691,
      "step": 24641
    },
    {
      "epoch": 0.7100213219616205,
      "grad_norm": 0.7321401834487915,
      "learning_rate": 2.0478017767520088e-06,
      "loss": 0.7022,
      "step": 24642
    },
    {
      "epoch": 0.7100501354232698,
      "grad_norm": 0.7350188493728638,
      "learning_rate": 2.047425197069536e-06,
      "loss": 0.6808,
      "step": 24643
    },
    {
      "epoch": 0.7100789488849191,
      "grad_norm": 0.7036651372909546,
      "learning_rate": 2.0470486431010404e-06,
      "loss": 0.6753,
      "step": 24644
    },
    {
      "epoch": 0.7101077623465684,
      "grad_norm": 0.7021858096122742,
      "learning_rate": 2.0466721148497978e-06,
      "loss": 0.6762,
      "step": 24645
    },
    {
      "epoch": 0.7101365758082177,
      "grad_norm": 0.722571074962616,
      "learning_rate": 2.0462956123190904e-06,
      "loss": 0.6996,
      "step": 24646
    },
    {
      "epoch": 0.7101653892698668,
      "grad_norm": 0.7364031672477722,
      "learning_rate": 2.0459191355121917e-06,
      "loss": 0.6992,
      "step": 24647
    },
    {
      "epoch": 0.7101942027315161,
      "grad_norm": 0.7546426653862,
      "learning_rate": 2.045542684432388e-06,
      "loss": 0.698,
      "step": 24648
    },
    {
      "epoch": 0.7102230161931654,
      "grad_norm": 0.730503261089325,
      "learning_rate": 2.0451662590829525e-06,
      "loss": 0.6953,
      "step": 24649
    },
    {
      "epoch": 0.7102518296548147,
      "grad_norm": 0.7054967880249023,
      "learning_rate": 2.0447898594671667e-06,
      "loss": 0.6801,
      "step": 24650
    },
    {
      "epoch": 0.710280643116464,
      "grad_norm": 0.7011505961418152,
      "learning_rate": 2.044413485588306e-06,
      "loss": 0.7008,
      "step": 24651
    },
    {
      "epoch": 0.7103094565781133,
      "grad_norm": 0.717646062374115,
      "learning_rate": 2.0440371374496506e-06,
      "loss": 0.6743,
      "step": 24652
    },
    {
      "epoch": 0.7103382700397626,
      "grad_norm": 0.7281455993652344,
      "learning_rate": 2.043660815054475e-06,
      "loss": 0.6956,
      "step": 24653
    },
    {
      "epoch": 0.7103670835014119,
      "grad_norm": 0.7132112383842468,
      "learning_rate": 2.04328451840606e-06,
      "loss": 0.6873,
      "step": 24654
    },
    {
      "epoch": 0.7103958969630612,
      "grad_norm": 0.7120976448059082,
      "learning_rate": 2.04290824750768e-06,
      "loss": 0.6774,
      "step": 24655
    },
    {
      "epoch": 0.7104247104247104,
      "grad_norm": 0.7591051459312439,
      "learning_rate": 2.042532002362613e-06,
      "loss": 0.7034,
      "step": 24656
    },
    {
      "epoch": 0.7104535238863597,
      "grad_norm": 0.7333466410636902,
      "learning_rate": 2.0421557829741375e-06,
      "loss": 0.6897,
      "step": 24657
    },
    {
      "epoch": 0.710482337348009,
      "grad_norm": 0.749148964881897,
      "learning_rate": 2.041779589345527e-06,
      "loss": 0.7131,
      "step": 24658
    },
    {
      "epoch": 0.7105111508096583,
      "grad_norm": 0.7219021916389465,
      "learning_rate": 2.041403421480058e-06,
      "loss": 0.7039,
      "step": 24659
    },
    {
      "epoch": 0.7105399642713075,
      "grad_norm": 0.7220301032066345,
      "learning_rate": 2.0410272793810102e-06,
      "loss": 0.6953,
      "step": 24660
    },
    {
      "epoch": 0.7105687777329568,
      "grad_norm": 0.7296332716941833,
      "learning_rate": 2.0406511630516557e-06,
      "loss": 0.6978,
      "step": 24661
    },
    {
      "epoch": 0.7105975911946061,
      "grad_norm": 0.7117089629173279,
      "learning_rate": 2.0402750724952707e-06,
      "loss": 0.695,
      "step": 24662
    },
    {
      "epoch": 0.7106264046562554,
      "grad_norm": 0.7229097485542297,
      "learning_rate": 2.039899007715133e-06,
      "loss": 0.6963,
      "step": 24663
    },
    {
      "epoch": 0.7106552181179047,
      "grad_norm": 0.7188385128974915,
      "learning_rate": 2.0395229687145142e-06,
      "loss": 0.6672,
      "step": 24664
    },
    {
      "epoch": 0.710684031579554,
      "grad_norm": 0.716949999332428,
      "learning_rate": 2.0391469554966923e-06,
      "loss": 0.6815,
      "step": 24665
    },
    {
      "epoch": 0.7107128450412032,
      "grad_norm": 0.7102901935577393,
      "learning_rate": 2.03877096806494e-06,
      "loss": 0.6786,
      "step": 24666
    },
    {
      "epoch": 0.7107416585028525,
      "grad_norm": 0.7456071376800537,
      "learning_rate": 2.038395006422533e-06,
      "loss": 0.7159,
      "step": 24667
    },
    {
      "epoch": 0.7107704719645018,
      "grad_norm": 0.7266349196434021,
      "learning_rate": 2.038019070572744e-06,
      "loss": 0.7068,
      "step": 24668
    },
    {
      "epoch": 0.7107992854261511,
      "grad_norm": 0.7424684166908264,
      "learning_rate": 2.0376431605188492e-06,
      "loss": 0.695,
      "step": 24669
    },
    {
      "epoch": 0.7108280988878004,
      "grad_norm": 0.7471131682395935,
      "learning_rate": 2.03726727626412e-06,
      "loss": 0.7007,
      "step": 24670
    },
    {
      "epoch": 0.7108569123494497,
      "grad_norm": 0.7081283330917358,
      "learning_rate": 2.0368914178118303e-06,
      "loss": 0.679,
      "step": 24671
    },
    {
      "epoch": 0.710885725811099,
      "grad_norm": 0.7001433968544006,
      "learning_rate": 2.0365155851652553e-06,
      "loss": 0.706,
      "step": 24672
    },
    {
      "epoch": 0.7109145392727483,
      "grad_norm": 0.7132948637008667,
      "learning_rate": 2.0361397783276683e-06,
      "loss": 0.6899,
      "step": 24673
    },
    {
      "epoch": 0.7109433527343975,
      "grad_norm": 0.7006312608718872,
      "learning_rate": 2.0357639973023397e-06,
      "loss": 0.6795,
      "step": 24674
    },
    {
      "epoch": 0.7109721661960468,
      "grad_norm": 0.7011454105377197,
      "learning_rate": 2.0353882420925446e-06,
      "loss": 0.6778,
      "step": 24675
    },
    {
      "epoch": 0.711000979657696,
      "grad_norm": 0.8115861415863037,
      "learning_rate": 2.0350125127015534e-06,
      "loss": 0.6689,
      "step": 24676
    },
    {
      "epoch": 0.7110297931193453,
      "grad_norm": 0.7639732956886292,
      "learning_rate": 2.034636809132639e-06,
      "loss": 0.7105,
      "step": 24677
    },
    {
      "epoch": 0.7110586065809946,
      "grad_norm": 0.7236916422843933,
      "learning_rate": 2.034261131389075e-06,
      "loss": 0.6862,
      "step": 24678
    },
    {
      "epoch": 0.7110874200426439,
      "grad_norm": 0.7249773740768433,
      "learning_rate": 2.0338854794741314e-06,
      "loss": 0.6883,
      "step": 24679
    },
    {
      "epoch": 0.7111162335042932,
      "grad_norm": 0.7124369144439697,
      "learning_rate": 2.0335098533910808e-06,
      "loss": 0.7117,
      "step": 24680
    },
    {
      "epoch": 0.7111450469659425,
      "grad_norm": 0.7096071243286133,
      "learning_rate": 2.033134253143193e-06,
      "loss": 0.6886,
      "step": 24681
    },
    {
      "epoch": 0.7111738604275918,
      "grad_norm": 0.7228993773460388,
      "learning_rate": 2.032758678733741e-06,
      "loss": 0.672,
      "step": 24682
    },
    {
      "epoch": 0.7112026738892411,
      "grad_norm": 0.7102055549621582,
      "learning_rate": 2.032383130165992e-06,
      "loss": 0.6772,
      "step": 24683
    },
    {
      "epoch": 0.7112314873508904,
      "grad_norm": 0.7301743030548096,
      "learning_rate": 2.0320076074432226e-06,
      "loss": 0.7011,
      "step": 24684
    },
    {
      "epoch": 0.7112603008125397,
      "grad_norm": 0.7455933690071106,
      "learning_rate": 2.0316321105686983e-06,
      "loss": 0.6972,
      "step": 24685
    },
    {
      "epoch": 0.711289114274189,
      "grad_norm": 0.7383061647415161,
      "learning_rate": 2.031256639545693e-06,
      "loss": 0.6952,
      "step": 24686
    },
    {
      "epoch": 0.7113179277358381,
      "grad_norm": 0.7341257333755493,
      "learning_rate": 2.0308811943774727e-06,
      "loss": 0.7013,
      "step": 24687
    },
    {
      "epoch": 0.7113467411974874,
      "grad_norm": 0.7105768918991089,
      "learning_rate": 2.030505775067311e-06,
      "loss": 0.6834,
      "step": 24688
    },
    {
      "epoch": 0.7113755546591367,
      "grad_norm": 0.7198821306228638,
      "learning_rate": 2.0301303816184736e-06,
      "loss": 0.6978,
      "step": 24689
    },
    {
      "epoch": 0.711404368120786,
      "grad_norm": 0.7081692814826965,
      "learning_rate": 2.029755014034234e-06,
      "loss": 0.6911,
      "step": 24690
    },
    {
      "epoch": 0.7114331815824353,
      "grad_norm": 0.722739040851593,
      "learning_rate": 2.029379672317857e-06,
      "loss": 0.6903,
      "step": 24691
    },
    {
      "epoch": 0.7114619950440846,
      "grad_norm": 0.6954250335693359,
      "learning_rate": 2.0290043564726153e-06,
      "loss": 0.673,
      "step": 24692
    },
    {
      "epoch": 0.7114908085057339,
      "grad_norm": 0.7340771555900574,
      "learning_rate": 2.028629066501774e-06,
      "loss": 0.6926,
      "step": 24693
    },
    {
      "epoch": 0.7115196219673832,
      "grad_norm": 0.7043043375015259,
      "learning_rate": 2.0282538024086037e-06,
      "loss": 0.6919,
      "step": 24694
    },
    {
      "epoch": 0.7115484354290325,
      "grad_norm": 0.7126848697662354,
      "learning_rate": 2.0278785641963717e-06,
      "loss": 0.6979,
      "step": 24695
    },
    {
      "epoch": 0.7115772488906817,
      "grad_norm": 0.7328564524650574,
      "learning_rate": 2.0275033518683477e-06,
      "loss": 0.688,
      "step": 24696
    },
    {
      "epoch": 0.711606062352331,
      "grad_norm": 0.7291196584701538,
      "learning_rate": 2.027128165427797e-06,
      "loss": 0.6959,
      "step": 24697
    },
    {
      "epoch": 0.7116348758139803,
      "grad_norm": 0.7145729660987854,
      "learning_rate": 2.02675300487799e-06,
      "loss": 0.6747,
      "step": 24698
    },
    {
      "epoch": 0.7116636892756296,
      "grad_norm": 0.7042490839958191,
      "learning_rate": 2.0263778702221896e-06,
      "loss": 0.7054,
      "step": 24699
    },
    {
      "epoch": 0.7116925027372789,
      "grad_norm": 0.7184917330741882,
      "learning_rate": 2.026002761463666e-06,
      "loss": 0.668,
      "step": 24700
    },
    {
      "epoch": 0.7117213161989281,
      "grad_norm": 0.7310255765914917,
      "learning_rate": 2.0256276786056874e-06,
      "loss": 0.6924,
      "step": 24701
    },
    {
      "epoch": 0.7117501296605774,
      "grad_norm": 0.7359933853149414,
      "learning_rate": 2.025252621651516e-06,
      "loss": 0.707,
      "step": 24702
    },
    {
      "epoch": 0.7117789431222267,
      "grad_norm": 0.7237229347229004,
      "learning_rate": 2.0248775906044233e-06,
      "loss": 0.6744,
      "step": 24703
    },
    {
      "epoch": 0.711807756583876,
      "grad_norm": 0.721134603023529,
      "learning_rate": 2.0245025854676704e-06,
      "loss": 0.6925,
      "step": 24704
    },
    {
      "epoch": 0.7118365700455253,
      "grad_norm": 0.7377936840057373,
      "learning_rate": 2.0241276062445276e-06,
      "loss": 0.6649,
      "step": 24705
    },
    {
      "epoch": 0.7118653835071745,
      "grad_norm": 0.7099849581718445,
      "learning_rate": 2.023752652938256e-06,
      "loss": 0.6763,
      "step": 24706
    },
    {
      "epoch": 0.7118941969688238,
      "grad_norm": 0.7507058382034302,
      "learning_rate": 2.023377725552125e-06,
      "loss": 0.6971,
      "step": 24707
    },
    {
      "epoch": 0.7119230104304731,
      "grad_norm": 0.7116756439208984,
      "learning_rate": 2.023002824089397e-06,
      "loss": 0.6894,
      "step": 24708
    },
    {
      "epoch": 0.7119518238921224,
      "grad_norm": 0.714474081993103,
      "learning_rate": 2.0226279485533413e-06,
      "loss": 0.6652,
      "step": 24709
    },
    {
      "epoch": 0.7119806373537717,
      "grad_norm": 0.7567136287689209,
      "learning_rate": 2.022253098947218e-06,
      "loss": 0.7039,
      "step": 24710
    },
    {
      "epoch": 0.712009450815421,
      "grad_norm": 0.7254711985588074,
      "learning_rate": 2.021878275274295e-06,
      "loss": 0.6828,
      "step": 24711
    },
    {
      "epoch": 0.7120382642770703,
      "grad_norm": 0.7295833826065063,
      "learning_rate": 2.0215034775378336e-06,
      "loss": 0.6939,
      "step": 24712
    },
    {
      "epoch": 0.7120670777387196,
      "grad_norm": 0.7268215417861938,
      "learning_rate": 2.021128705741101e-06,
      "loss": 0.6838,
      "step": 24713
    },
    {
      "epoch": 0.7120958912003688,
      "grad_norm": 0.7403587102890015,
      "learning_rate": 2.0207539598873582e-06,
      "loss": 0.6999,
      "step": 24714
    },
    {
      "epoch": 0.712124704662018,
      "grad_norm": 0.7304959893226624,
      "learning_rate": 2.02037923997987e-06,
      "loss": 0.6995,
      "step": 24715
    },
    {
      "epoch": 0.7121535181236673,
      "grad_norm": 0.7270345687866211,
      "learning_rate": 2.020004546021902e-06,
      "loss": 0.7015,
      "step": 24716
    },
    {
      "epoch": 0.7121823315853166,
      "grad_norm": 0.7028122544288635,
      "learning_rate": 2.019629878016714e-06,
      "loss": 0.7087,
      "step": 24717
    },
    {
      "epoch": 0.7122111450469659,
      "grad_norm": 0.7268419861793518,
      "learning_rate": 2.0192552359675698e-06,
      "loss": 0.7198,
      "step": 24718
    },
    {
      "epoch": 0.7122399585086152,
      "grad_norm": 0.7245154976844788,
      "learning_rate": 2.018880619877735e-06,
      "loss": 0.6914,
      "step": 24719
    },
    {
      "epoch": 0.7122687719702645,
      "grad_norm": 0.7214730978012085,
      "learning_rate": 2.018506029750468e-06,
      "loss": 0.6921,
      "step": 24720
    },
    {
      "epoch": 0.7122975854319138,
      "grad_norm": 0.7123430967330933,
      "learning_rate": 2.018131465589034e-06,
      "loss": 0.7118,
      "step": 24721
    },
    {
      "epoch": 0.7123263988935631,
      "grad_norm": 0.7328504323959351,
      "learning_rate": 2.0177569273966944e-06,
      "loss": 0.696,
      "step": 24722
    },
    {
      "epoch": 0.7123552123552124,
      "grad_norm": 0.7158498167991638,
      "learning_rate": 2.0173824151767103e-06,
      "loss": 0.6883,
      "step": 24723
    },
    {
      "epoch": 0.7123840258168617,
      "grad_norm": 0.689728319644928,
      "learning_rate": 2.017007928932345e-06,
      "loss": 0.7077,
      "step": 24724
    },
    {
      "epoch": 0.712412839278511,
      "grad_norm": 0.719184398651123,
      "learning_rate": 2.016633468666857e-06,
      "loss": 0.7112,
      "step": 24725
    },
    {
      "epoch": 0.7124416527401602,
      "grad_norm": 0.7027393579483032,
      "learning_rate": 2.0162590343835114e-06,
      "loss": 0.6801,
      "step": 24726
    },
    {
      "epoch": 0.7124704662018095,
      "grad_norm": 0.756413459777832,
      "learning_rate": 2.0158846260855647e-06,
      "loss": 0.6904,
      "step": 24727
    },
    {
      "epoch": 0.7124992796634587,
      "grad_norm": 0.7200604677200317,
      "learning_rate": 2.015510243776282e-06,
      "loss": 0.704,
      "step": 24728
    },
    {
      "epoch": 0.712528093125108,
      "grad_norm": 0.7184860110282898,
      "learning_rate": 2.01513588745892e-06,
      "loss": 0.6667,
      "step": 24729
    },
    {
      "epoch": 0.7125569065867573,
      "grad_norm": 0.7246115207672119,
      "learning_rate": 2.0147615571367403e-06,
      "loss": 0.6683,
      "step": 24730
    },
    {
      "epoch": 0.7125857200484066,
      "grad_norm": 0.7119277715682983,
      "learning_rate": 2.014387252813003e-06,
      "loss": 0.6908,
      "step": 24731
    },
    {
      "epoch": 0.7126145335100559,
      "grad_norm": 0.7340152263641357,
      "learning_rate": 2.0140129744909703e-06,
      "loss": 0.6912,
      "step": 24732
    },
    {
      "epoch": 0.7126433469717052,
      "grad_norm": 0.706537663936615,
      "learning_rate": 2.0136387221738986e-06,
      "loss": 0.6957,
      "step": 24733
    },
    {
      "epoch": 0.7126721604333545,
      "grad_norm": 0.725760817527771,
      "learning_rate": 2.0132644958650492e-06,
      "loss": 0.6879,
      "step": 24734
    },
    {
      "epoch": 0.7127009738950038,
      "grad_norm": 0.705125093460083,
      "learning_rate": 2.012890295567679e-06,
      "loss": 0.6643,
      "step": 24735
    },
    {
      "epoch": 0.712729787356653,
      "grad_norm": 0.720054030418396,
      "learning_rate": 2.0125161212850504e-06,
      "loss": 0.7044,
      "step": 24736
    },
    {
      "epoch": 0.7127586008183023,
      "grad_norm": 0.7331181168556213,
      "learning_rate": 2.012141973020417e-06,
      "loss": 0.6972,
      "step": 24737
    },
    {
      "epoch": 0.7127874142799516,
      "grad_norm": 0.6958508491516113,
      "learning_rate": 2.0117678507770418e-06,
      "loss": 0.6437,
      "step": 24738
    },
    {
      "epoch": 0.7128162277416009,
      "grad_norm": 0.7265718579292297,
      "learning_rate": 2.0113937545581824e-06,
      "loss": 0.7032,
      "step": 24739
    },
    {
      "epoch": 0.7128450412032502,
      "grad_norm": 0.7214533090591431,
      "learning_rate": 2.0110196843670945e-06,
      "loss": 0.6933,
      "step": 24740
    },
    {
      "epoch": 0.7128738546648994,
      "grad_norm": 0.710601270198822,
      "learning_rate": 2.0106456402070384e-06,
      "loss": 0.6887,
      "step": 24741
    },
    {
      "epoch": 0.7129026681265487,
      "grad_norm": 0.7336897253990173,
      "learning_rate": 2.0102716220812675e-06,
      "loss": 0.6931,
      "step": 24742
    },
    {
      "epoch": 0.712931481588198,
      "grad_norm": 0.7464935779571533,
      "learning_rate": 2.0098976299930455e-06,
      "loss": 0.7087,
      "step": 24743
    },
    {
      "epoch": 0.7129602950498473,
      "grad_norm": 0.7126590013504028,
      "learning_rate": 2.0095236639456246e-06,
      "loss": 0.713,
      "step": 24744
    },
    {
      "epoch": 0.7129891085114965,
      "grad_norm": 0.7195475697517395,
      "learning_rate": 2.009149723942265e-06,
      "loss": 0.6953,
      "step": 24745
    },
    {
      "epoch": 0.7130179219731458,
      "grad_norm": 0.7507834434509277,
      "learning_rate": 2.008775809986219e-06,
      "loss": 0.7042,
      "step": 24746
    },
    {
      "epoch": 0.7130467354347951,
      "grad_norm": 0.7242614030838013,
      "learning_rate": 2.008401922080748e-06,
      "loss": 0.6921,
      "step": 24747
    },
    {
      "epoch": 0.7130755488964444,
      "grad_norm": 0.7580089569091797,
      "learning_rate": 2.008028060229104e-06,
      "loss": 0.6978,
      "step": 24748
    },
    {
      "epoch": 0.7131043623580937,
      "grad_norm": 0.7252565622329712,
      "learning_rate": 2.0076542244345464e-06,
      "loss": 0.685,
      "step": 24749
    },
    {
      "epoch": 0.713133175819743,
      "grad_norm": 0.728904664516449,
      "learning_rate": 2.0072804147003274e-06,
      "loss": 0.7165,
      "step": 24750
    },
    {
      "epoch": 0.7131619892813923,
      "grad_norm": 0.7261961698532104,
      "learning_rate": 2.0069066310297065e-06,
      "loss": 0.6756,
      "step": 24751
    },
    {
      "epoch": 0.7131908027430416,
      "grad_norm": 0.712332546710968,
      "learning_rate": 2.0065328734259342e-06,
      "loss": 0.6779,
      "step": 24752
    },
    {
      "epoch": 0.7132196162046909,
      "grad_norm": 0.7268822193145752,
      "learning_rate": 2.0061591418922694e-06,
      "loss": 0.6713,
      "step": 24753
    },
    {
      "epoch": 0.7132484296663402,
      "grad_norm": 0.7202244997024536,
      "learning_rate": 2.0057854364319644e-06,
      "loss": 0.6712,
      "step": 24754
    },
    {
      "epoch": 0.7132772431279893,
      "grad_norm": 0.7247198224067688,
      "learning_rate": 2.005411757048278e-06,
      "loss": 0.6849,
      "step": 24755
    },
    {
      "epoch": 0.7133060565896386,
      "grad_norm": 0.7066340446472168,
      "learning_rate": 2.0050381037444595e-06,
      "loss": 0.6692,
      "step": 24756
    },
    {
      "epoch": 0.7133348700512879,
      "grad_norm": 0.7110073566436768,
      "learning_rate": 2.004664476523767e-06,
      "loss": 0.6689,
      "step": 24757
    },
    {
      "epoch": 0.7133636835129372,
      "grad_norm": 0.7180758714675903,
      "learning_rate": 2.0042908753894506e-06,
      "loss": 0.686,
      "step": 24758
    },
    {
      "epoch": 0.7133924969745865,
      "grad_norm": 0.7111905217170715,
      "learning_rate": 2.003917300344767e-06,
      "loss": 0.6658,
      "step": 24759
    },
    {
      "epoch": 0.7134213104362358,
      "grad_norm": 0.6991621255874634,
      "learning_rate": 2.0035437513929695e-06,
      "loss": 0.6786,
      "step": 24760
    },
    {
      "epoch": 0.7134501238978851,
      "grad_norm": 0.7256681323051453,
      "learning_rate": 2.0031702285373096e-06,
      "loss": 0.6924,
      "step": 24761
    },
    {
      "epoch": 0.7134789373595344,
      "grad_norm": 0.7429186701774597,
      "learning_rate": 2.0027967317810425e-06,
      "loss": 0.7107,
      "step": 24762
    },
    {
      "epoch": 0.7135077508211837,
      "grad_norm": 0.7572370767593384,
      "learning_rate": 2.002423261127418e-06,
      "loss": 0.7044,
      "step": 24763
    },
    {
      "epoch": 0.713536564282833,
      "grad_norm": 0.7111419439315796,
      "learning_rate": 2.0020498165796928e-06,
      "loss": 0.6836,
      "step": 24764
    },
    {
      "epoch": 0.7135653777444823,
      "grad_norm": 0.719502329826355,
      "learning_rate": 2.001676398141114e-06,
      "loss": 0.6999,
      "step": 24765
    },
    {
      "epoch": 0.7135941912061315,
      "grad_norm": 0.7226057052612305,
      "learning_rate": 2.0013030058149376e-06,
      "loss": 0.6802,
      "step": 24766
    },
    {
      "epoch": 0.7136230046677808,
      "grad_norm": 0.7586529850959778,
      "learning_rate": 2.0009296396044137e-06,
      "loss": 0.7058,
      "step": 24767
    },
    {
      "epoch": 0.71365181812943,
      "grad_norm": 0.7239702939987183,
      "learning_rate": 2.000556299512797e-06,
      "loss": 0.6877,
      "step": 24768
    },
    {
      "epoch": 0.7136806315910793,
      "grad_norm": 0.720009982585907,
      "learning_rate": 2.0001829855433347e-06,
      "loss": 0.7057,
      "step": 24769
    },
    {
      "epoch": 0.7137094450527286,
      "grad_norm": 0.7042438983917236,
      "learning_rate": 1.9998096976992814e-06,
      "loss": 0.7311,
      "step": 24770
    },
    {
      "epoch": 0.7137382585143779,
      "grad_norm": 0.7045918703079224,
      "learning_rate": 1.9994364359838847e-06,
      "loss": 0.707,
      "step": 24771
    },
    {
      "epoch": 0.7137670719760272,
      "grad_norm": 0.7293068766593933,
      "learning_rate": 1.9990632004003992e-06,
      "loss": 0.6809,
      "step": 24772
    },
    {
      "epoch": 0.7137958854376765,
      "grad_norm": 0.7199869155883789,
      "learning_rate": 1.9986899909520713e-06,
      "loss": 0.6906,
      "step": 24773
    },
    {
      "epoch": 0.7138246988993258,
      "grad_norm": 0.6950744986534119,
      "learning_rate": 1.9983168076421557e-06,
      "loss": 0.6813,
      "step": 24774
    },
    {
      "epoch": 0.713853512360975,
      "grad_norm": 0.7181453704833984,
      "learning_rate": 1.9979436504738977e-06,
      "loss": 0.7131,
      "step": 24775
    },
    {
      "epoch": 0.7138823258226243,
      "grad_norm": 0.7444166541099548,
      "learning_rate": 1.9975705194505496e-06,
      "loss": 0.7009,
      "step": 24776
    },
    {
      "epoch": 0.7139111392842736,
      "grad_norm": 0.7555431723594666,
      "learning_rate": 1.997197414575363e-06,
      "loss": 0.7032,
      "step": 24777
    },
    {
      "epoch": 0.7139399527459229,
      "grad_norm": 0.7321972846984863,
      "learning_rate": 1.9968243358515836e-06,
      "loss": 0.6554,
      "step": 24778
    },
    {
      "epoch": 0.7139687662075722,
      "grad_norm": 0.7163708806037903,
      "learning_rate": 1.9964512832824616e-06,
      "loss": 0.683,
      "step": 24779
    },
    {
      "epoch": 0.7139975796692215,
      "grad_norm": 0.6979038715362549,
      "learning_rate": 1.9960782568712472e-06,
      "loss": 0.6608,
      "step": 24780
    },
    {
      "epoch": 0.7140263931308708,
      "grad_norm": 0.701294481754303,
      "learning_rate": 1.99570525662119e-06,
      "loss": 0.679,
      "step": 24781
    },
    {
      "epoch": 0.71405520659252,
      "grad_norm": 0.7280226945877075,
      "learning_rate": 1.9953322825355342e-06,
      "loss": 0.6975,
      "step": 24782
    },
    {
      "epoch": 0.7140840200541693,
      "grad_norm": 0.7305853962898254,
      "learning_rate": 1.9949593346175332e-06,
      "loss": 0.6906,
      "step": 24783
    },
    {
      "epoch": 0.7141128335158186,
      "grad_norm": 0.7200559377670288,
      "learning_rate": 1.9945864128704307e-06,
      "loss": 0.6814,
      "step": 24784
    },
    {
      "epoch": 0.7141416469774678,
      "grad_norm": 0.7171362638473511,
      "learning_rate": 1.994213517297478e-06,
      "loss": 0.6945,
      "step": 24785
    },
    {
      "epoch": 0.7141704604391171,
      "grad_norm": 0.7295272350311279,
      "learning_rate": 1.9938406479019186e-06,
      "loss": 0.7087,
      "step": 24786
    },
    {
      "epoch": 0.7141992739007664,
      "grad_norm": 0.6853070259094238,
      "learning_rate": 1.9934678046870044e-06,
      "loss": 0.6672,
      "step": 24787
    },
    {
      "epoch": 0.7142280873624157,
      "grad_norm": 0.715916633605957,
      "learning_rate": 1.993094987655978e-06,
      "loss": 0.6855,
      "step": 24788
    },
    {
      "epoch": 0.714256900824065,
      "grad_norm": 0.7262297868728638,
      "learning_rate": 1.9927221968120908e-06,
      "loss": 0.6828,
      "step": 24789
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.724811315536499,
      "learning_rate": 1.992349432158583e-06,
      "loss": 0.6755,
      "step": 24790
    },
    {
      "epoch": 0.7143145277473636,
      "grad_norm": 0.7046515941619873,
      "learning_rate": 1.991976693698709e-06,
      "loss": 0.6582,
      "step": 24791
    },
    {
      "epoch": 0.7143433412090129,
      "grad_norm": 0.7018823623657227,
      "learning_rate": 1.991603981435709e-06,
      "loss": 0.6638,
      "step": 24792
    },
    {
      "epoch": 0.7143721546706622,
      "grad_norm": 0.7250913977622986,
      "learning_rate": 1.9912312953728328e-06,
      "loss": 0.6903,
      "step": 24793
    },
    {
      "epoch": 0.7144009681323115,
      "grad_norm": 0.7271173596382141,
      "learning_rate": 1.9908586355133226e-06,
      "loss": 0.6964,
      "step": 24794
    },
    {
      "epoch": 0.7144297815939606,
      "grad_norm": 0.6980763673782349,
      "learning_rate": 1.9904860018604277e-06,
      "loss": 0.6928,
      "step": 24795
    },
    {
      "epoch": 0.7144585950556099,
      "grad_norm": 0.7223705053329468,
      "learning_rate": 1.990113394417389e-06,
      "loss": 0.699,
      "step": 24796
    },
    {
      "epoch": 0.7144874085172592,
      "grad_norm": 0.7055701613426208,
      "learning_rate": 1.9897408131874547e-06,
      "loss": 0.6809,
      "step": 24797
    },
    {
      "epoch": 0.7145162219789085,
      "grad_norm": 0.7660709619522095,
      "learning_rate": 1.98936825817387e-06,
      "loss": 0.6869,
      "step": 24798
    },
    {
      "epoch": 0.7145450354405578,
      "grad_norm": 0.7148029208183289,
      "learning_rate": 1.988995729379876e-06,
      "loss": 0.6865,
      "step": 24799
    },
    {
      "epoch": 0.7145738489022071,
      "grad_norm": 0.7004271745681763,
      "learning_rate": 1.9886232268087217e-06,
      "loss": 0.7023,
      "step": 24800
    },
    {
      "epoch": 0.7146026623638564,
      "grad_norm": 0.7217936515808105,
      "learning_rate": 1.988250750463647e-06,
      "loss": 0.6765,
      "step": 24801
    },
    {
      "epoch": 0.7146314758255057,
      "grad_norm": 0.6874313354492188,
      "learning_rate": 1.9878783003478973e-06,
      "loss": 0.6764,
      "step": 24802
    },
    {
      "epoch": 0.714660289287155,
      "grad_norm": 0.7141547799110413,
      "learning_rate": 1.987505876464717e-06,
      "loss": 0.6888,
      "step": 24803
    },
    {
      "epoch": 0.7146891027488043,
      "grad_norm": 0.711356520652771,
      "learning_rate": 1.987133478817351e-06,
      "loss": 0.7057,
      "step": 24804
    },
    {
      "epoch": 0.7147179162104536,
      "grad_norm": 0.7181177139282227,
      "learning_rate": 1.986761107409038e-06,
      "loss": 0.6999,
      "step": 24805
    },
    {
      "epoch": 0.7147467296721028,
      "grad_norm": 0.6979102492332458,
      "learning_rate": 1.9863887622430263e-06,
      "loss": 0.6998,
      "step": 24806
    },
    {
      "epoch": 0.7147755431337521,
      "grad_norm": 0.7388159036636353,
      "learning_rate": 1.9860164433225537e-06,
      "loss": 0.6967,
      "step": 24807
    },
    {
      "epoch": 0.7148043565954014,
      "grad_norm": 0.7190250754356384,
      "learning_rate": 1.9856441506508673e-06,
      "loss": 0.6997,
      "step": 24808
    },
    {
      "epoch": 0.7148331700570506,
      "grad_norm": 0.7155390381813049,
      "learning_rate": 1.9852718842312048e-06,
      "loss": 0.678,
      "step": 24809
    },
    {
      "epoch": 0.7148619835186999,
      "grad_norm": 0.7351670861244202,
      "learning_rate": 1.984899644066812e-06,
      "loss": 0.6742,
      "step": 24810
    },
    {
      "epoch": 0.7148907969803492,
      "grad_norm": 0.7050330638885498,
      "learning_rate": 1.9845274301609286e-06,
      "loss": 0.6723,
      "step": 24811
    },
    {
      "epoch": 0.7149196104419985,
      "grad_norm": 0.6931192874908447,
      "learning_rate": 1.984155242516798e-06,
      "loss": 0.6663,
      "step": 24812
    },
    {
      "epoch": 0.7149484239036478,
      "grad_norm": 0.7134386897087097,
      "learning_rate": 1.983783081137659e-06,
      "loss": 0.6879,
      "step": 24813
    },
    {
      "epoch": 0.7149772373652971,
      "grad_norm": 0.7326571345329285,
      "learning_rate": 1.9834109460267536e-06,
      "loss": 0.6791,
      "step": 24814
    },
    {
      "epoch": 0.7150060508269463,
      "grad_norm": 0.7152656316757202,
      "learning_rate": 1.9830388371873244e-06,
      "loss": 0.6921,
      "step": 24815
    },
    {
      "epoch": 0.7150348642885956,
      "grad_norm": 0.6988546252250671,
      "learning_rate": 1.982666754622612e-06,
      "loss": 0.6889,
      "step": 24816
    },
    {
      "epoch": 0.7150636777502449,
      "grad_norm": 0.7412351965904236,
      "learning_rate": 1.982294698335854e-06,
      "loss": 0.7094,
      "step": 24817
    },
    {
      "epoch": 0.7150924912118942,
      "grad_norm": 0.7212167978286743,
      "learning_rate": 1.9819226683302934e-06,
      "loss": 0.7118,
      "step": 24818
    },
    {
      "epoch": 0.7151213046735435,
      "grad_norm": 0.718571126461029,
      "learning_rate": 1.981550664609171e-06,
      "loss": 0.7089,
      "step": 24819
    },
    {
      "epoch": 0.7151501181351928,
      "grad_norm": 0.7314990758895874,
      "learning_rate": 1.9811786871757234e-06,
      "loss": 0.7049,
      "step": 24820
    },
    {
      "epoch": 0.7151789315968421,
      "grad_norm": 0.7338142395019531,
      "learning_rate": 1.9808067360331927e-06,
      "loss": 0.6757,
      "step": 24821
    },
    {
      "epoch": 0.7152077450584913,
      "grad_norm": 0.7127306461334229,
      "learning_rate": 1.980434811184816e-06,
      "loss": 0.7066,
      "step": 24822
    },
    {
      "epoch": 0.7152365585201406,
      "grad_norm": 0.7165841460227966,
      "learning_rate": 1.980062912633835e-06,
      "loss": 0.7057,
      "step": 24823
    },
    {
      "epoch": 0.7152653719817899,
      "grad_norm": 0.7121061682701111,
      "learning_rate": 1.9796910403834857e-06,
      "loss": 0.6858,
      "step": 24824
    },
    {
      "epoch": 0.7152941854434391,
      "grad_norm": 0.6729356050491333,
      "learning_rate": 1.97931919443701e-06,
      "loss": 0.6609,
      "step": 24825
    },
    {
      "epoch": 0.7153229989050884,
      "grad_norm": 0.7211169600486755,
      "learning_rate": 1.978947374797641e-06,
      "loss": 0.6871,
      "step": 24826
    },
    {
      "epoch": 0.7153518123667377,
      "grad_norm": 0.7544628977775574,
      "learning_rate": 1.978575581468624e-06,
      "loss": 0.6975,
      "step": 24827
    },
    {
      "epoch": 0.715380625828387,
      "grad_norm": 0.728692352771759,
      "learning_rate": 1.978203814453192e-06,
      "loss": 0.7037,
      "step": 24828
    },
    {
      "epoch": 0.7154094392900363,
      "grad_norm": 0.7080541849136353,
      "learning_rate": 1.977832073754585e-06,
      "loss": 0.6999,
      "step": 24829
    },
    {
      "epoch": 0.7154382527516856,
      "grad_norm": 0.7234447002410889,
      "learning_rate": 1.977460359376038e-06,
      "loss": 0.7034,
      "step": 24830
    },
    {
      "epoch": 0.7154670662133349,
      "grad_norm": 0.7030031085014343,
      "learning_rate": 1.977088671320792e-06,
      "loss": 0.6763,
      "step": 24831
    },
    {
      "epoch": 0.7154958796749842,
      "grad_norm": 0.7010145783424377,
      "learning_rate": 1.97671700959208e-06,
      "loss": 0.6427,
      "step": 24832
    },
    {
      "epoch": 0.7155246931366335,
      "grad_norm": 0.7605682611465454,
      "learning_rate": 1.9763453741931422e-06,
      "loss": 0.6924,
      "step": 24833
    },
    {
      "epoch": 0.7155535065982828,
      "grad_norm": 0.7196662425994873,
      "learning_rate": 1.975973765127212e-06,
      "loss": 0.686,
      "step": 24834
    },
    {
      "epoch": 0.715582320059932,
      "grad_norm": 0.6890138983726501,
      "learning_rate": 1.9756021823975272e-06,
      "loss": 0.6765,
      "step": 24835
    },
    {
      "epoch": 0.7156111335215812,
      "grad_norm": 0.7158272862434387,
      "learning_rate": 1.975230626007326e-06,
      "loss": 0.6802,
      "step": 24836
    },
    {
      "epoch": 0.7156399469832305,
      "grad_norm": 0.7066608667373657,
      "learning_rate": 1.974859095959841e-06,
      "loss": 0.6881,
      "step": 24837
    },
    {
      "epoch": 0.7156687604448798,
      "grad_norm": 0.7367346286773682,
      "learning_rate": 1.974487592258309e-06,
      "loss": 0.7159,
      "step": 24838
    },
    {
      "epoch": 0.7156975739065291,
      "grad_norm": 0.7274715900421143,
      "learning_rate": 1.9741161149059658e-06,
      "loss": 0.7019,
      "step": 24839
    },
    {
      "epoch": 0.7157263873681784,
      "grad_norm": 0.7339811325073242,
      "learning_rate": 1.973744663906048e-06,
      "loss": 0.6831,
      "step": 24840
    },
    {
      "epoch": 0.7157552008298277,
      "grad_norm": 0.7095144391059875,
      "learning_rate": 1.973373239261787e-06,
      "loss": 0.6838,
      "step": 24841
    },
    {
      "epoch": 0.715784014291477,
      "grad_norm": 0.7285321354866028,
      "learning_rate": 1.9730018409764218e-06,
      "loss": 0.7151,
      "step": 24842
    },
    {
      "epoch": 0.7158128277531263,
      "grad_norm": 0.7294058799743652,
      "learning_rate": 1.972630469053183e-06,
      "loss": 0.7124,
      "step": 24843
    },
    {
      "epoch": 0.7158416412147756,
      "grad_norm": 0.7090948820114136,
      "learning_rate": 1.9722591234953074e-06,
      "loss": 0.6736,
      "step": 24844
    },
    {
      "epoch": 0.7158704546764248,
      "grad_norm": 0.7196934819221497,
      "learning_rate": 1.9718878043060274e-06,
      "loss": 0.6791,
      "step": 24845
    },
    {
      "epoch": 0.7158992681380741,
      "grad_norm": 0.7326611876487732,
      "learning_rate": 1.971516511488579e-06,
      "loss": 0.6989,
      "step": 24846
    },
    {
      "epoch": 0.7159280815997234,
      "grad_norm": 0.7276600003242493,
      "learning_rate": 1.971145245046193e-06,
      "loss": 0.6726,
      "step": 24847
    },
    {
      "epoch": 0.7159568950613727,
      "grad_norm": 0.725693941116333,
      "learning_rate": 1.9707740049821055e-06,
      "loss": 0.6795,
      "step": 24848
    },
    {
      "epoch": 0.7159857085230219,
      "grad_norm": 0.7291510701179504,
      "learning_rate": 1.9704027912995467e-06,
      "loss": 0.6923,
      "step": 24849
    },
    {
      "epoch": 0.7160145219846712,
      "grad_norm": 0.7145373225212097,
      "learning_rate": 1.9700316040017513e-06,
      "loss": 0.6905,
      "step": 24850
    },
    {
      "epoch": 0.7160433354463205,
      "grad_norm": 0.7071726322174072,
      "learning_rate": 1.9696604430919515e-06,
      "loss": 0.6763,
      "step": 24851
    },
    {
      "epoch": 0.7160721489079698,
      "grad_norm": 0.7060620784759521,
      "learning_rate": 1.9692893085733817e-06,
      "loss": 0.6847,
      "step": 24852
    },
    {
      "epoch": 0.7161009623696191,
      "grad_norm": 0.7197756767272949,
      "learning_rate": 1.9689182004492708e-06,
      "loss": 0.6677,
      "step": 24853
    },
    {
      "epoch": 0.7161297758312684,
      "grad_norm": 0.7164196968078613,
      "learning_rate": 1.9685471187228544e-06,
      "loss": 0.6926,
      "step": 24854
    },
    {
      "epoch": 0.7161585892929176,
      "grad_norm": 0.7230624556541443,
      "learning_rate": 1.9681760633973603e-06,
      "loss": 0.6754,
      "step": 24855
    },
    {
      "epoch": 0.7161874027545669,
      "grad_norm": 0.7174599170684814,
      "learning_rate": 1.967805034476022e-06,
      "loss": 0.6732,
      "step": 24856
    },
    {
      "epoch": 0.7162162162162162,
      "grad_norm": 0.7155083417892456,
      "learning_rate": 1.967434031962072e-06,
      "loss": 0.6853,
      "step": 24857
    },
    {
      "epoch": 0.7162450296778655,
      "grad_norm": 0.7087476849555969,
      "learning_rate": 1.967063055858739e-06,
      "loss": 0.6798,
      "step": 24858
    },
    {
      "epoch": 0.7162738431395148,
      "grad_norm": 0.7224438190460205,
      "learning_rate": 1.9666921061692564e-06,
      "loss": 0.6955,
      "step": 24859
    },
    {
      "epoch": 0.7163026566011641,
      "grad_norm": 0.7320347428321838,
      "learning_rate": 1.966321182896852e-06,
      "loss": 0.6815,
      "step": 24860
    },
    {
      "epoch": 0.7163314700628134,
      "grad_norm": 0.7172545790672302,
      "learning_rate": 1.9659502860447587e-06,
      "loss": 0.6905,
      "step": 24861
    },
    {
      "epoch": 0.7163602835244627,
      "grad_norm": 0.7106466293334961,
      "learning_rate": 1.9655794156162023e-06,
      "loss": 0.6856,
      "step": 24862
    },
    {
      "epoch": 0.7163890969861119,
      "grad_norm": 0.6965265870094299,
      "learning_rate": 1.965208571614419e-06,
      "loss": 0.6865,
      "step": 24863
    },
    {
      "epoch": 0.7164179104477612,
      "grad_norm": 0.7404602766036987,
      "learning_rate": 1.9648377540426343e-06,
      "loss": 0.7311,
      "step": 24864
    },
    {
      "epoch": 0.7164467239094104,
      "grad_norm": 0.7190604209899902,
      "learning_rate": 1.9644669629040804e-06,
      "loss": 0.6945,
      "step": 24865
    },
    {
      "epoch": 0.7164755373710597,
      "grad_norm": 0.7703462839126587,
      "learning_rate": 1.9640961982019826e-06,
      "loss": 0.6993,
      "step": 24866
    },
    {
      "epoch": 0.716504350832709,
      "grad_norm": 0.7288333773612976,
      "learning_rate": 1.963725459939574e-06,
      "loss": 0.6648,
      "step": 24867
    },
    {
      "epoch": 0.7165331642943583,
      "grad_norm": 0.7031846642494202,
      "learning_rate": 1.96335474812008e-06,
      "loss": 0.6832,
      "step": 24868
    },
    {
      "epoch": 0.7165619777560076,
      "grad_norm": 0.7056843638420105,
      "learning_rate": 1.962984062746732e-06,
      "loss": 0.6858,
      "step": 24869
    },
    {
      "epoch": 0.7165907912176569,
      "grad_norm": 0.7338714599609375,
      "learning_rate": 1.962613403822755e-06,
      "loss": 0.6849,
      "step": 24870
    },
    {
      "epoch": 0.7166196046793062,
      "grad_norm": 0.7147071957588196,
      "learning_rate": 1.9622427713513813e-06,
      "loss": 0.6852,
      "step": 24871
    },
    {
      "epoch": 0.7166484181409555,
      "grad_norm": 0.7278233766555786,
      "learning_rate": 1.9618721653358345e-06,
      "loss": 0.6684,
      "step": 24872
    },
    {
      "epoch": 0.7166772316026048,
      "grad_norm": 0.723865807056427,
      "learning_rate": 1.961501585779344e-06,
      "loss": 0.6849,
      "step": 24873
    },
    {
      "epoch": 0.7167060450642541,
      "grad_norm": 3.008971691131592,
      "learning_rate": 1.9611310326851375e-06,
      "loss": 0.7414,
      "step": 24874
    },
    {
      "epoch": 0.7167348585259034,
      "grad_norm": 0.7328056693077087,
      "learning_rate": 1.9607605060564433e-06,
      "loss": 0.6773,
      "step": 24875
    },
    {
      "epoch": 0.7167636719875525,
      "grad_norm": 0.7050742506980896,
      "learning_rate": 1.960390005896486e-06,
      "loss": 0.6787,
      "step": 24876
    },
    {
      "epoch": 0.7167924854492018,
      "grad_norm": 0.7268694043159485,
      "learning_rate": 1.9600195322084925e-06,
      "loss": 0.6802,
      "step": 24877
    },
    {
      "epoch": 0.7168212989108511,
      "grad_norm": 0.7468422651290894,
      "learning_rate": 1.959649084995692e-06,
      "loss": 0.7169,
      "step": 24878
    },
    {
      "epoch": 0.7168501123725004,
      "grad_norm": 0.7085443139076233,
      "learning_rate": 1.9592786642613064e-06,
      "loss": 0.6771,
      "step": 24879
    },
    {
      "epoch": 0.7168789258341497,
      "grad_norm": 0.7229885458946228,
      "learning_rate": 1.9589082700085666e-06,
      "loss": 0.7016,
      "step": 24880
    },
    {
      "epoch": 0.716907739295799,
      "grad_norm": 0.7559266686439514,
      "learning_rate": 1.9585379022406938e-06,
      "loss": 0.6782,
      "step": 24881
    },
    {
      "epoch": 0.7169365527574483,
      "grad_norm": 0.724478542804718,
      "learning_rate": 1.9581675609609176e-06,
      "loss": 0.6994,
      "step": 24882
    },
    {
      "epoch": 0.7169653662190976,
      "grad_norm": 0.7248393893241882,
      "learning_rate": 1.9577972461724585e-06,
      "loss": 0.6946,
      "step": 24883
    },
    {
      "epoch": 0.7169941796807469,
      "grad_norm": 0.7203525304794312,
      "learning_rate": 1.957426957878547e-06,
      "loss": 0.6956,
      "step": 24884
    },
    {
      "epoch": 0.7170229931423961,
      "grad_norm": 0.7226667404174805,
      "learning_rate": 1.9570566960824027e-06,
      "loss": 0.6908,
      "step": 24885
    },
    {
      "epoch": 0.7170518066040454,
      "grad_norm": 0.7183972597122192,
      "learning_rate": 1.956686460787254e-06,
      "loss": 0.6865,
      "step": 24886
    },
    {
      "epoch": 0.7170806200656947,
      "grad_norm": 0.7442984580993652,
      "learning_rate": 1.956316251996323e-06,
      "loss": 0.6768,
      "step": 24887
    },
    {
      "epoch": 0.717109433527344,
      "grad_norm": 0.7108609676361084,
      "learning_rate": 1.955946069712837e-06,
      "loss": 0.6894,
      "step": 24888
    },
    {
      "epoch": 0.7171382469889933,
      "grad_norm": 0.7262668609619141,
      "learning_rate": 1.9555759139400153e-06,
      "loss": 0.6946,
      "step": 24889
    },
    {
      "epoch": 0.7171670604506425,
      "grad_norm": 0.7163413166999817,
      "learning_rate": 1.9552057846810866e-06,
      "loss": 0.6835,
      "step": 24890
    },
    {
      "epoch": 0.7171958739122918,
      "grad_norm": 0.7069113254547119,
      "learning_rate": 1.95483568193927e-06,
      "loss": 0.6835,
      "step": 24891
    },
    {
      "epoch": 0.7172246873739411,
      "grad_norm": 0.7431157827377319,
      "learning_rate": 1.954465605717792e-06,
      "loss": 0.685,
      "step": 24892
    },
    {
      "epoch": 0.7172535008355904,
      "grad_norm": 0.7242009043693542,
      "learning_rate": 1.9540955560198727e-06,
      "loss": 0.6884,
      "step": 24893
    },
    {
      "epoch": 0.7172823142972397,
      "grad_norm": 0.7074158787727356,
      "learning_rate": 1.9537255328487367e-06,
      "loss": 0.6915,
      "step": 24894
    },
    {
      "epoch": 0.717311127758889,
      "grad_norm": 0.7436858415603638,
      "learning_rate": 1.953355536207607e-06,
      "loss": 0.7096,
      "step": 24895
    },
    {
      "epoch": 0.7173399412205382,
      "grad_norm": 0.7128943204879761,
      "learning_rate": 1.952985566099704e-06,
      "loss": 0.6783,
      "step": 24896
    },
    {
      "epoch": 0.7173687546821875,
      "grad_norm": 0.697420597076416,
      "learning_rate": 1.9526156225282524e-06,
      "loss": 0.6841,
      "step": 24897
    },
    {
      "epoch": 0.7173975681438368,
      "grad_norm": 0.6960832476615906,
      "learning_rate": 1.9522457054964683e-06,
      "loss": 0.6863,
      "step": 24898
    },
    {
      "epoch": 0.7174263816054861,
      "grad_norm": 0.7257786989212036,
      "learning_rate": 1.9518758150075817e-06,
      "loss": 0.6941,
      "step": 24899
    },
    {
      "epoch": 0.7174551950671354,
      "grad_norm": 0.7281875610351562,
      "learning_rate": 1.9515059510648077e-06,
      "loss": 0.7206,
      "step": 24900
    },
    {
      "epoch": 0.7174840085287847,
      "grad_norm": 0.7062786221504211,
      "learning_rate": 1.9511361136713713e-06,
      "loss": 0.6995,
      "step": 24901
    },
    {
      "epoch": 0.717512821990434,
      "grad_norm": 0.7131301164627075,
      "learning_rate": 1.9507663028304904e-06,
      "loss": 0.6602,
      "step": 24902
    },
    {
      "epoch": 0.7175416354520832,
      "grad_norm": 0.7188485264778137,
      "learning_rate": 1.950396518545388e-06,
      "loss": 0.6774,
      "step": 24903
    },
    {
      "epoch": 0.7175704489137325,
      "grad_norm": 0.7296265363693237,
      "learning_rate": 1.9500267608192812e-06,
      "loss": 0.7184,
      "step": 24904
    },
    {
      "epoch": 0.7175992623753817,
      "grad_norm": 0.7253453731536865,
      "learning_rate": 1.949657029655395e-06,
      "loss": 0.6968,
      "step": 24905
    },
    {
      "epoch": 0.717628075837031,
      "grad_norm": 0.7190273404121399,
      "learning_rate": 1.949287325056945e-06,
      "loss": 0.7019,
      "step": 24906
    },
    {
      "epoch": 0.7176568892986803,
      "grad_norm": 0.7278390526771545,
      "learning_rate": 1.948917647027155e-06,
      "loss": 0.6798,
      "step": 24907
    },
    {
      "epoch": 0.7176857027603296,
      "grad_norm": 0.7110560536384583,
      "learning_rate": 1.94854799556924e-06,
      "loss": 0.6753,
      "step": 24908
    },
    {
      "epoch": 0.7177145162219789,
      "grad_norm": 0.7051630020141602,
      "learning_rate": 1.948178370686423e-06,
      "loss": 0.693,
      "step": 24909
    },
    {
      "epoch": 0.7177433296836282,
      "grad_norm": 0.7389899492263794,
      "learning_rate": 1.947808772381919e-06,
      "loss": 0.7177,
      "step": 24910
    },
    {
      "epoch": 0.7177721431452775,
      "grad_norm": 0.7173779606819153,
      "learning_rate": 1.947439200658952e-06,
      "loss": 0.6953,
      "step": 24911
    },
    {
      "epoch": 0.7178009566069268,
      "grad_norm": 0.6895724534988403,
      "learning_rate": 1.9470696555207373e-06,
      "loss": 0.7154,
      "step": 24912
    },
    {
      "epoch": 0.7178297700685761,
      "grad_norm": 0.7389928102493286,
      "learning_rate": 1.9467001369704956e-06,
      "loss": 0.6957,
      "step": 24913
    },
    {
      "epoch": 0.7178585835302254,
      "grad_norm": 0.7287164926528931,
      "learning_rate": 1.9463306450114417e-06,
      "loss": 0.6906,
      "step": 24914
    },
    {
      "epoch": 0.7178873969918746,
      "grad_norm": 0.7231810092926025,
      "learning_rate": 1.945961179646795e-06,
      "loss": 0.6922,
      "step": 24915
    },
    {
      "epoch": 0.7179162104535239,
      "grad_norm": 0.6854225397109985,
      "learning_rate": 1.945591740879776e-06,
      "loss": 0.6765,
      "step": 24916
    },
    {
      "epoch": 0.7179450239151731,
      "grad_norm": 0.7359741926193237,
      "learning_rate": 1.9452223287135974e-06,
      "loss": 0.7035,
      "step": 24917
    },
    {
      "epoch": 0.7179738373768224,
      "grad_norm": 0.7068819403648376,
      "learning_rate": 1.9448529431514807e-06,
      "loss": 0.7012,
      "step": 24918
    },
    {
      "epoch": 0.7180026508384717,
      "grad_norm": 0.7257919907569885,
      "learning_rate": 1.9444835841966386e-06,
      "loss": 0.6819,
      "step": 24919
    },
    {
      "epoch": 0.718031464300121,
      "grad_norm": 0.7313904762268066,
      "learning_rate": 1.944114251852292e-06,
      "loss": 0.7032,
      "step": 24920
    },
    {
      "epoch": 0.7180602777617703,
      "grad_norm": 0.7303802371025085,
      "learning_rate": 1.943744946121654e-06,
      "loss": 0.7001,
      "step": 24921
    },
    {
      "epoch": 0.7180890912234196,
      "grad_norm": 0.7374057769775391,
      "learning_rate": 1.9433756670079424e-06,
      "loss": 0.7089,
      "step": 24922
    },
    {
      "epoch": 0.7181179046850689,
      "grad_norm": 0.7062659859657288,
      "learning_rate": 1.943006414514373e-06,
      "loss": 0.6973,
      "step": 24923
    },
    {
      "epoch": 0.7181467181467182,
      "grad_norm": 0.7271531820297241,
      "learning_rate": 1.9426371886441635e-06,
      "loss": 0.7007,
      "step": 24924
    },
    {
      "epoch": 0.7181755316083674,
      "grad_norm": 0.7382854223251343,
      "learning_rate": 1.9422679894005266e-06,
      "loss": 0.6942,
      "step": 24925
    },
    {
      "epoch": 0.7182043450700167,
      "grad_norm": 0.7099286317825317,
      "learning_rate": 1.9418988167866805e-06,
      "loss": 0.6855,
      "step": 24926
    },
    {
      "epoch": 0.718233158531666,
      "grad_norm": 0.722679078578949,
      "learning_rate": 1.941529670805837e-06,
      "loss": 0.6943,
      "step": 24927
    },
    {
      "epoch": 0.7182619719933153,
      "grad_norm": 0.734661877155304,
      "learning_rate": 1.9411605514612136e-06,
      "loss": 0.6924,
      "step": 24928
    },
    {
      "epoch": 0.7182907854549646,
      "grad_norm": 0.7020886540412903,
      "learning_rate": 1.9407914587560233e-06,
      "loss": 0.6777,
      "step": 24929
    },
    {
      "epoch": 0.7183195989166138,
      "grad_norm": 0.7437604069709778,
      "learning_rate": 1.940422392693483e-06,
      "loss": 0.7017,
      "step": 24930
    },
    {
      "epoch": 0.7183484123782631,
      "grad_norm": 0.7215863466262817,
      "learning_rate": 1.9400533532768027e-06,
      "loss": 0.6959,
      "step": 24931
    },
    {
      "epoch": 0.7183772258399124,
      "grad_norm": 0.7255247235298157,
      "learning_rate": 1.9396843405091996e-06,
      "loss": 0.6792,
      "step": 24932
    },
    {
      "epoch": 0.7184060393015617,
      "grad_norm": 0.7162168025970459,
      "learning_rate": 1.939315354393888e-06,
      "loss": 0.6898,
      "step": 24933
    },
    {
      "epoch": 0.718434852763211,
      "grad_norm": 0.7354459762573242,
      "learning_rate": 1.938946394934079e-06,
      "loss": 0.6896,
      "step": 24934
    },
    {
      "epoch": 0.7184636662248602,
      "grad_norm": 0.738520085811615,
      "learning_rate": 1.938577462132986e-06,
      "loss": 0.6943,
      "step": 24935
    },
    {
      "epoch": 0.7184924796865095,
      "grad_norm": 0.7335724830627441,
      "learning_rate": 1.9382085559938235e-06,
      "loss": 0.7073,
      "step": 24936
    },
    {
      "epoch": 0.7185212931481588,
      "grad_norm": 0.7008414268493652,
      "learning_rate": 1.937839676519806e-06,
      "loss": 0.6775,
      "step": 24937
    },
    {
      "epoch": 0.7185501066098081,
      "grad_norm": 0.7301973104476929,
      "learning_rate": 1.9374708237141413e-06,
      "loss": 0.7078,
      "step": 24938
    },
    {
      "epoch": 0.7185789200714574,
      "grad_norm": 0.7068396210670471,
      "learning_rate": 1.9371019975800465e-06,
      "loss": 0.6932,
      "step": 24939
    },
    {
      "epoch": 0.7186077335331067,
      "grad_norm": 0.715846836566925,
      "learning_rate": 1.93673319812073e-06,
      "loss": 0.6738,
      "step": 24940
    },
    {
      "epoch": 0.718636546994756,
      "grad_norm": 0.7428473830223083,
      "learning_rate": 1.9363644253394063e-06,
      "loss": 0.725,
      "step": 24941
    },
    {
      "epoch": 0.7186653604564053,
      "grad_norm": 0.7375968098640442,
      "learning_rate": 1.935995679239285e-06,
      "loss": 0.7172,
      "step": 24942
    },
    {
      "epoch": 0.7186941739180546,
      "grad_norm": 0.7053338885307312,
      "learning_rate": 1.93562695982358e-06,
      "loss": 0.6846,
      "step": 24943
    },
    {
      "epoch": 0.7187229873797037,
      "grad_norm": 0.7446231245994568,
      "learning_rate": 1.935258267095499e-06,
      "loss": 0.6977,
      "step": 24944
    },
    {
      "epoch": 0.718751800841353,
      "grad_norm": 0.716153085231781,
      "learning_rate": 1.934889601058256e-06,
      "loss": 0.6864,
      "step": 24945
    },
    {
      "epoch": 0.7187806143030023,
      "grad_norm": 0.7037632465362549,
      "learning_rate": 1.934520961715058e-06,
      "loss": 0.6673,
      "step": 24946
    },
    {
      "epoch": 0.7188094277646516,
      "grad_norm": 0.7223188281059265,
      "learning_rate": 1.934152349069121e-06,
      "loss": 0.7078,
      "step": 24947
    },
    {
      "epoch": 0.7188382412263009,
      "grad_norm": 0.7152453064918518,
      "learning_rate": 1.9337837631236512e-06,
      "loss": 0.712,
      "step": 24948
    },
    {
      "epoch": 0.7188670546879502,
      "grad_norm": 0.6931784749031067,
      "learning_rate": 1.933415203881861e-06,
      "loss": 0.6724,
      "step": 24949
    },
    {
      "epoch": 0.7188958681495995,
      "grad_norm": 0.7224819660186768,
      "learning_rate": 1.933046671346957e-06,
      "loss": 0.6907,
      "step": 24950
    },
    {
      "epoch": 0.7189246816112488,
      "grad_norm": 0.7209732532501221,
      "learning_rate": 1.9326781655221526e-06,
      "loss": 0.6787,
      "step": 24951
    },
    {
      "epoch": 0.7189534950728981,
      "grad_norm": 0.7034627795219421,
      "learning_rate": 1.9323096864106537e-06,
      "loss": 0.6559,
      "step": 24952
    },
    {
      "epoch": 0.7189823085345474,
      "grad_norm": 0.7595667839050293,
      "learning_rate": 1.9319412340156705e-06,
      "loss": 0.7138,
      "step": 24953
    },
    {
      "epoch": 0.7190111219961967,
      "grad_norm": 0.7440003156661987,
      "learning_rate": 1.9315728083404144e-06,
      "loss": 0.6851,
      "step": 24954
    },
    {
      "epoch": 0.719039935457846,
      "grad_norm": 0.7338368892669678,
      "learning_rate": 1.9312044093880898e-06,
      "loss": 0.718,
      "step": 24955
    },
    {
      "epoch": 0.7190687489194952,
      "grad_norm": 0.6995752453804016,
      "learning_rate": 1.930836037161909e-06,
      "loss": 0.6766,
      "step": 24956
    },
    {
      "epoch": 0.7190975623811444,
      "grad_norm": 0.7259618639945984,
      "learning_rate": 1.930467691665076e-06,
      "loss": 0.6849,
      "step": 24957
    },
    {
      "epoch": 0.7191263758427937,
      "grad_norm": 0.737903892993927,
      "learning_rate": 1.9300993729008015e-06,
      "loss": 0.7044,
      "step": 24958
    },
    {
      "epoch": 0.719155189304443,
      "grad_norm": 0.7379418611526489,
      "learning_rate": 1.929731080872293e-06,
      "loss": 0.6597,
      "step": 24959
    },
    {
      "epoch": 0.7191840027660923,
      "grad_norm": 0.7092871069908142,
      "learning_rate": 1.9293628155827587e-06,
      "loss": 0.683,
      "step": 24960
    },
    {
      "epoch": 0.7192128162277416,
      "grad_norm": 0.7377827763557434,
      "learning_rate": 1.9289945770354036e-06,
      "loss": 0.6821,
      "step": 24961
    },
    {
      "epoch": 0.7192416296893909,
      "grad_norm": 0.7352903485298157,
      "learning_rate": 1.928626365233437e-06,
      "loss": 0.6902,
      "step": 24962
    },
    {
      "epoch": 0.7192704431510402,
      "grad_norm": 0.7155812382698059,
      "learning_rate": 1.9282581801800633e-06,
      "loss": 0.6599,
      "step": 24963
    },
    {
      "epoch": 0.7192992566126895,
      "grad_norm": 0.7327709794044495,
      "learning_rate": 1.9278900218784917e-06,
      "loss": 0.6848,
      "step": 24964
    },
    {
      "epoch": 0.7193280700743387,
      "grad_norm": 0.7175082564353943,
      "learning_rate": 1.9275218903319253e-06,
      "loss": 0.6619,
      "step": 24965
    },
    {
      "epoch": 0.719356883535988,
      "grad_norm": 0.7024570107460022,
      "learning_rate": 1.9271537855435736e-06,
      "loss": 0.6799,
      "step": 24966
    },
    {
      "epoch": 0.7193856969976373,
      "grad_norm": 0.6850122809410095,
      "learning_rate": 1.926785707516639e-06,
      "loss": 0.677,
      "step": 24967
    },
    {
      "epoch": 0.7194145104592866,
      "grad_norm": 0.7183818221092224,
      "learning_rate": 1.92641765625433e-06,
      "loss": 0.6896,
      "step": 24968
    },
    {
      "epoch": 0.7194433239209359,
      "grad_norm": 0.7479971647262573,
      "learning_rate": 1.92604963175985e-06,
      "loss": 0.6973,
      "step": 24969
    },
    {
      "epoch": 0.7194721373825852,
      "grad_norm": 0.7003495097160339,
      "learning_rate": 1.9256816340364043e-06,
      "loss": 0.6691,
      "step": 24970
    },
    {
      "epoch": 0.7195009508442344,
      "grad_norm": 0.71480792760849,
      "learning_rate": 1.9253136630871988e-06,
      "loss": 0.6939,
      "step": 24971
    },
    {
      "epoch": 0.7195297643058837,
      "grad_norm": 0.7062976956367493,
      "learning_rate": 1.924945718915439e-06,
      "loss": 0.6741,
      "step": 24972
    },
    {
      "epoch": 0.719558577767533,
      "grad_norm": 0.705915093421936,
      "learning_rate": 1.9245778015243265e-06,
      "loss": 0.6819,
      "step": 24973
    },
    {
      "epoch": 0.7195873912291822,
      "grad_norm": 0.7176575660705566,
      "learning_rate": 1.9242099109170668e-06,
      "loss": 0.6943,
      "step": 24974
    },
    {
      "epoch": 0.7196162046908315,
      "grad_norm": 0.7174461483955383,
      "learning_rate": 1.9238420470968665e-06,
      "loss": 0.7134,
      "step": 24975
    },
    {
      "epoch": 0.7196450181524808,
      "grad_norm": 0.734977126121521,
      "learning_rate": 1.9234742100669246e-06,
      "loss": 0.712,
      "step": 24976
    },
    {
      "epoch": 0.7196738316141301,
      "grad_norm": 0.7034088969230652,
      "learning_rate": 1.9231063998304488e-06,
      "loss": 0.6835,
      "step": 24977
    },
    {
      "epoch": 0.7197026450757794,
      "grad_norm": 0.7336205840110779,
      "learning_rate": 1.922738616390639e-06,
      "loss": 0.7116,
      "step": 24978
    },
    {
      "epoch": 0.7197314585374287,
      "grad_norm": 0.7138072848320007,
      "learning_rate": 1.9223708597507014e-06,
      "loss": 0.67,
      "step": 24979
    },
    {
      "epoch": 0.719760271999078,
      "grad_norm": 0.7079371213912964,
      "learning_rate": 1.9220031299138353e-06,
      "loss": 0.678,
      "step": 24980
    },
    {
      "epoch": 0.7197890854607273,
      "grad_norm": 0.7187792062759399,
      "learning_rate": 1.921635426883245e-06,
      "loss": 0.7001,
      "step": 24981
    },
    {
      "epoch": 0.7198178989223766,
      "grad_norm": 0.7320442795753479,
      "learning_rate": 1.9212677506621323e-06,
      "loss": 0.7167,
      "step": 24982
    },
    {
      "epoch": 0.7198467123840259,
      "grad_norm": 0.7254796028137207,
      "learning_rate": 1.920900101253703e-06,
      "loss": 0.7073,
      "step": 24983
    },
    {
      "epoch": 0.719875525845675,
      "grad_norm": 0.6909791827201843,
      "learning_rate": 1.9205324786611528e-06,
      "loss": 0.6633,
      "step": 24984
    },
    {
      "epoch": 0.7199043393073243,
      "grad_norm": 0.7164907455444336,
      "learning_rate": 1.920164882887689e-06,
      "loss": 0.6966,
      "step": 24985
    },
    {
      "epoch": 0.7199331527689736,
      "grad_norm": 0.7166745662689209,
      "learning_rate": 1.9197973139365086e-06,
      "loss": 0.6902,
      "step": 24986
    },
    {
      "epoch": 0.7199619662306229,
      "grad_norm": 0.7133803367614746,
      "learning_rate": 1.919429771810816e-06,
      "loss": 0.6716,
      "step": 24987
    },
    {
      "epoch": 0.7199907796922722,
      "grad_norm": 0.74749356508255,
      "learning_rate": 1.9190622565138083e-06,
      "loss": 0.6964,
      "step": 24988
    },
    {
      "epoch": 0.7200195931539215,
      "grad_norm": 0.7085139751434326,
      "learning_rate": 1.9186947680486904e-06,
      "loss": 0.6932,
      "step": 24989
    },
    {
      "epoch": 0.7200484066155708,
      "grad_norm": 0.6989524960517883,
      "learning_rate": 1.9183273064186597e-06,
      "loss": 0.6805,
      "step": 24990
    },
    {
      "epoch": 0.7200772200772201,
      "grad_norm": 0.7308291792869568,
      "learning_rate": 1.917959871626918e-06,
      "loss": 0.6794,
      "step": 24991
    },
    {
      "epoch": 0.7201060335388694,
      "grad_norm": 0.7103824615478516,
      "learning_rate": 1.9175924636766658e-06,
      "loss": 0.6874,
      "step": 24992
    },
    {
      "epoch": 0.7201348470005187,
      "grad_norm": 0.7436970472335815,
      "learning_rate": 1.9172250825711007e-06,
      "loss": 0.6875,
      "step": 24993
    },
    {
      "epoch": 0.720163660462168,
      "grad_norm": 0.7157236933708191,
      "learning_rate": 1.916857728313423e-06,
      "loss": 0.6941,
      "step": 24994
    },
    {
      "epoch": 0.7201924739238172,
      "grad_norm": 0.7329537272453308,
      "learning_rate": 1.916490400906833e-06,
      "loss": 0.678,
      "step": 24995
    },
    {
      "epoch": 0.7202212873854665,
      "grad_norm": 0.7301455140113831,
      "learning_rate": 1.9161231003545313e-06,
      "loss": 0.6786,
      "step": 24996
    },
    {
      "epoch": 0.7202501008471158,
      "grad_norm": 0.7441198229789734,
      "learning_rate": 1.915755826659712e-06,
      "loss": 0.6758,
      "step": 24997
    },
    {
      "epoch": 0.720278914308765,
      "grad_norm": 0.7397142648696899,
      "learning_rate": 1.9153885798255785e-06,
      "loss": 0.7019,
      "step": 24998
    },
    {
      "epoch": 0.7203077277704143,
      "grad_norm": 0.7102134823799133,
      "learning_rate": 1.9150213598553254e-06,
      "loss": 0.7037,
      "step": 24999
    },
    {
      "epoch": 0.7203365412320636,
      "grad_norm": 0.7274644374847412,
      "learning_rate": 1.9146541667521534e-06,
      "loss": 0.7127,
      "step": 25000
    },
    {
      "epoch": 0.7203653546937129,
      "grad_norm": 0.7100554704666138,
      "learning_rate": 1.914287000519258e-06,
      "loss": 0.6937,
      "step": 25001
    },
    {
      "epoch": 0.7203941681553622,
      "grad_norm": 0.7046732902526855,
      "learning_rate": 1.91391986115984e-06,
      "loss": 0.6837,
      "step": 25002
    },
    {
      "epoch": 0.7204229816170115,
      "grad_norm": 0.7051274180412292,
      "learning_rate": 1.913552748677094e-06,
      "loss": 0.6887,
      "step": 25003
    },
    {
      "epoch": 0.7204517950786608,
      "grad_norm": 0.7061405777931213,
      "learning_rate": 1.9131856630742195e-06,
      "loss": 0.6891,
      "step": 25004
    },
    {
      "epoch": 0.72048060854031,
      "grad_norm": 0.7065639495849609,
      "learning_rate": 1.912818604354409e-06,
      "loss": 0.6899,
      "step": 25005
    },
    {
      "epoch": 0.7205094220019593,
      "grad_norm": 0.7374297976493835,
      "learning_rate": 1.9124515725208655e-06,
      "loss": 0.6922,
      "step": 25006
    },
    {
      "epoch": 0.7205382354636086,
      "grad_norm": 0.7003012895584106,
      "learning_rate": 1.9120845675767806e-06,
      "loss": 0.6896,
      "step": 25007
    },
    {
      "epoch": 0.7205670489252579,
      "grad_norm": 0.6951101422309875,
      "learning_rate": 1.9117175895253544e-06,
      "loss": 0.6775,
      "step": 25008
    },
    {
      "epoch": 0.7205958623869072,
      "grad_norm": 0.6938262581825256,
      "learning_rate": 1.9113506383697785e-06,
      "loss": 0.6971,
      "step": 25009
    },
    {
      "epoch": 0.7206246758485565,
      "grad_norm": 0.714573323726654,
      "learning_rate": 1.910983714113253e-06,
      "loss": 0.671,
      "step": 25010
    },
    {
      "epoch": 0.7206534893102057,
      "grad_norm": 0.7291960716247559,
      "learning_rate": 1.91061681675897e-06,
      "loss": 0.6944,
      "step": 25011
    },
    {
      "epoch": 0.720682302771855,
      "grad_norm": 0.7060542106628418,
      "learning_rate": 1.9102499463101264e-06,
      "loss": 0.6765,
      "step": 25012
    },
    {
      "epoch": 0.7207111162335043,
      "grad_norm": 0.721737265586853,
      "learning_rate": 1.9098831027699182e-06,
      "loss": 0.7084,
      "step": 25013
    },
    {
      "epoch": 0.7207399296951535,
      "grad_norm": 0.6998351216316223,
      "learning_rate": 1.9095162861415385e-06,
      "loss": 0.7076,
      "step": 25014
    },
    {
      "epoch": 0.7207687431568028,
      "grad_norm": 0.7965431809425354,
      "learning_rate": 1.9091494964281832e-06,
      "loss": 0.6638,
      "step": 25015
    },
    {
      "epoch": 0.7207975566184521,
      "grad_norm": 0.7118651866912842,
      "learning_rate": 1.908782733633045e-06,
      "loss": 0.6736,
      "step": 25016
    },
    {
      "epoch": 0.7208263700801014,
      "grad_norm": 0.7245211601257324,
      "learning_rate": 1.908415997759319e-06,
      "loss": 0.6733,
      "step": 25017
    },
    {
      "epoch": 0.7208551835417507,
      "grad_norm": 0.7120020389556885,
      "learning_rate": 1.908049288810199e-06,
      "loss": 0.6909,
      "step": 25018
    },
    {
      "epoch": 0.7208839970034,
      "grad_norm": 0.7161368131637573,
      "learning_rate": 1.907682606788881e-06,
      "loss": 0.7112,
      "step": 25019
    },
    {
      "epoch": 0.7209128104650493,
      "grad_norm": 0.7070009708404541,
      "learning_rate": 1.907315951698554e-06,
      "loss": 0.6716,
      "step": 25020
    },
    {
      "epoch": 0.7209416239266986,
      "grad_norm": 0.7240282893180847,
      "learning_rate": 1.9069493235424158e-06,
      "loss": 0.6855,
      "step": 25021
    },
    {
      "epoch": 0.7209704373883479,
      "grad_norm": 0.7547810077667236,
      "learning_rate": 1.906582722323655e-06,
      "loss": 0.699,
      "step": 25022
    },
    {
      "epoch": 0.7209992508499972,
      "grad_norm": 0.7266780734062195,
      "learning_rate": 1.9062161480454683e-06,
      "loss": 0.6855,
      "step": 25023
    },
    {
      "epoch": 0.7210280643116465,
      "grad_norm": 0.7047845721244812,
      "learning_rate": 1.9058496007110443e-06,
      "loss": 0.6753,
      "step": 25024
    },
    {
      "epoch": 0.7210568777732956,
      "grad_norm": 0.7169877886772156,
      "learning_rate": 1.9054830803235795e-06,
      "loss": 0.6967,
      "step": 25025
    },
    {
      "epoch": 0.7210856912349449,
      "grad_norm": 0.7247229814529419,
      "learning_rate": 1.9051165868862614e-06,
      "loss": 0.7007,
      "step": 25026
    },
    {
      "epoch": 0.7211145046965942,
      "grad_norm": 0.7300410866737366,
      "learning_rate": 1.904750120402286e-06,
      "loss": 0.6844,
      "step": 25027
    },
    {
      "epoch": 0.7211433181582435,
      "grad_norm": 0.7150269746780396,
      "learning_rate": 1.9043836808748417e-06,
      "loss": 0.6826,
      "step": 25028
    },
    {
      "epoch": 0.7211721316198928,
      "grad_norm": 0.7050626277923584,
      "learning_rate": 1.9040172683071207e-06,
      "loss": 0.6693,
      "step": 25029
    },
    {
      "epoch": 0.7212009450815421,
      "grad_norm": 2.8175203800201416,
      "learning_rate": 1.9036508827023142e-06,
      "loss": 0.7493,
      "step": 25030
    },
    {
      "epoch": 0.7212297585431914,
      "grad_norm": 0.7210967540740967,
      "learning_rate": 1.9032845240636156e-06,
      "loss": 0.704,
      "step": 25031
    },
    {
      "epoch": 0.7212585720048407,
      "grad_norm": 0.7354099154472351,
      "learning_rate": 1.9029181923942108e-06,
      "loss": 0.6963,
      "step": 25032
    },
    {
      "epoch": 0.72128738546649,
      "grad_norm": 0.728354275226593,
      "learning_rate": 1.9025518876972926e-06,
      "loss": 0.6945,
      "step": 25033
    },
    {
      "epoch": 0.7213161989281393,
      "grad_norm": 0.7086248397827148,
      "learning_rate": 1.9021856099760532e-06,
      "loss": 0.6872,
      "step": 25034
    },
    {
      "epoch": 0.7213450123897885,
      "grad_norm": 0.7195020318031311,
      "learning_rate": 1.9018193592336786e-06,
      "loss": 0.693,
      "step": 25035
    },
    {
      "epoch": 0.7213738258514378,
      "grad_norm": 0.7116740942001343,
      "learning_rate": 1.9014531354733623e-06,
      "loss": 0.6828,
      "step": 25036
    },
    {
      "epoch": 0.7214026393130871,
      "grad_norm": 0.7186055183410645,
      "learning_rate": 1.9010869386982899e-06,
      "loss": 0.7014,
      "step": 25037
    },
    {
      "epoch": 0.7214314527747363,
      "grad_norm": 0.7215104103088379,
      "learning_rate": 1.9007207689116542e-06,
      "loss": 0.7036,
      "step": 25038
    },
    {
      "epoch": 0.7214602662363856,
      "grad_norm": 0.7306539416313171,
      "learning_rate": 1.900354626116641e-06,
      "loss": 0.6919,
      "step": 25039
    },
    {
      "epoch": 0.7214890796980349,
      "grad_norm": 0.715873122215271,
      "learning_rate": 1.8999885103164416e-06,
      "loss": 0.6932,
      "step": 25040
    },
    {
      "epoch": 0.7215178931596842,
      "grad_norm": 0.7276315689086914,
      "learning_rate": 1.8996224215142406e-06,
      "loss": 0.7112,
      "step": 25041
    },
    {
      "epoch": 0.7215467066213335,
      "grad_norm": 0.7172256708145142,
      "learning_rate": 1.8992563597132324e-06,
      "loss": 0.6939,
      "step": 25042
    },
    {
      "epoch": 0.7215755200829828,
      "grad_norm": 0.7361436486244202,
      "learning_rate": 1.8988903249165995e-06,
      "loss": 0.6818,
      "step": 25043
    },
    {
      "epoch": 0.721604333544632,
      "grad_norm": 0.7272612452507019,
      "learning_rate": 1.898524317127534e-06,
      "loss": 0.7102,
      "step": 25044
    },
    {
      "epoch": 0.7216331470062813,
      "grad_norm": 0.7528884410858154,
      "learning_rate": 1.8981583363492196e-06,
      "loss": 0.6959,
      "step": 25045
    },
    {
      "epoch": 0.7216619604679306,
      "grad_norm": 0.7075713872909546,
      "learning_rate": 1.8977923825848472e-06,
      "loss": 0.6703,
      "step": 25046
    },
    {
      "epoch": 0.7216907739295799,
      "grad_norm": 0.7068674564361572,
      "learning_rate": 1.8974264558376e-06,
      "loss": 0.6978,
      "step": 25047
    },
    {
      "epoch": 0.7217195873912292,
      "grad_norm": 0.7147291898727417,
      "learning_rate": 1.8970605561106685e-06,
      "loss": 0.6816,
      "step": 25048
    },
    {
      "epoch": 0.7217484008528785,
      "grad_norm": 0.707581639289856,
      "learning_rate": 1.896694683407237e-06,
      "loss": 0.6877,
      "step": 25049
    },
    {
      "epoch": 0.7217772143145278,
      "grad_norm": 0.7152453064918518,
      "learning_rate": 1.8963288377304916e-06,
      "loss": 0.6921,
      "step": 25050
    },
    {
      "epoch": 0.7218060277761771,
      "grad_norm": 0.7108516097068787,
      "learning_rate": 1.895963019083622e-06,
      "loss": 0.6855,
      "step": 25051
    },
    {
      "epoch": 0.7218348412378263,
      "grad_norm": 0.7235193252563477,
      "learning_rate": 1.8955972274698093e-06,
      "loss": 0.6931,
      "step": 25052
    },
    {
      "epoch": 0.7218636546994756,
      "grad_norm": 0.7135761380195618,
      "learning_rate": 1.895231462892242e-06,
      "loss": 0.6906,
      "step": 25053
    },
    {
      "epoch": 0.7218924681611248,
      "grad_norm": 0.7209702730178833,
      "learning_rate": 1.8948657253541065e-06,
      "loss": 0.6974,
      "step": 25054
    },
    {
      "epoch": 0.7219212816227741,
      "grad_norm": 0.7333497405052185,
      "learning_rate": 1.8945000148585847e-06,
      "loss": 0.7053,
      "step": 25055
    },
    {
      "epoch": 0.7219500950844234,
      "grad_norm": 0.7344601154327393,
      "learning_rate": 1.8941343314088634e-06,
      "loss": 0.6966,
      "step": 25056
    },
    {
      "epoch": 0.7219789085460727,
      "grad_norm": 0.7017200589179993,
      "learning_rate": 1.8937686750081291e-06,
      "loss": 0.6992,
      "step": 25057
    },
    {
      "epoch": 0.722007722007722,
      "grad_norm": 0.7248305082321167,
      "learning_rate": 1.8934030456595626e-06,
      "loss": 0.7043,
      "step": 25058
    },
    {
      "epoch": 0.7220365354693713,
      "grad_norm": 0.7333755493164062,
      "learning_rate": 1.8930374433663522e-06,
      "loss": 0.7175,
      "step": 25059
    },
    {
      "epoch": 0.7220653489310206,
      "grad_norm": 0.7191557288169861,
      "learning_rate": 1.892671868131678e-06,
      "loss": 0.6748,
      "step": 25060
    },
    {
      "epoch": 0.7220941623926699,
      "grad_norm": 0.7007588148117065,
      "learning_rate": 1.8923063199587276e-06,
      "loss": 0.6833,
      "step": 25061
    },
    {
      "epoch": 0.7221229758543192,
      "grad_norm": 0.6974173188209534,
      "learning_rate": 1.8919407988506805e-06,
      "loss": 0.6647,
      "step": 25062
    },
    {
      "epoch": 0.7221517893159685,
      "grad_norm": 0.7100550532341003,
      "learning_rate": 1.8915753048107238e-06,
      "loss": 0.6987,
      "step": 25063
    },
    {
      "epoch": 0.7221806027776178,
      "grad_norm": 0.707555890083313,
      "learning_rate": 1.8912098378420374e-06,
      "loss": 0.7145,
      "step": 25064
    },
    {
      "epoch": 0.7222094162392669,
      "grad_norm": 0.7633688449859619,
      "learning_rate": 1.890844397947806e-06,
      "loss": 0.6896,
      "step": 25065
    },
    {
      "epoch": 0.7222382297009162,
      "grad_norm": 0.7378700375556946,
      "learning_rate": 1.8904789851312112e-06,
      "loss": 0.7162,
      "step": 25066
    },
    {
      "epoch": 0.7222670431625655,
      "grad_norm": 0.6888260841369629,
      "learning_rate": 1.8901135993954384e-06,
      "loss": 0.6647,
      "step": 25067
    },
    {
      "epoch": 0.7222958566242148,
      "grad_norm": 0.7362227439880371,
      "learning_rate": 1.8897482407436652e-06,
      "loss": 0.6933,
      "step": 25068
    },
    {
      "epoch": 0.7223246700858641,
      "grad_norm": 0.715563952922821,
      "learning_rate": 1.8893829091790778e-06,
      "loss": 0.6602,
      "step": 25069
    },
    {
      "epoch": 0.7223534835475134,
      "grad_norm": 0.7330856919288635,
      "learning_rate": 1.8890176047048535e-06,
      "loss": 0.6961,
      "step": 25070
    },
    {
      "epoch": 0.7223822970091627,
      "grad_norm": 0.7348607182502747,
      "learning_rate": 1.8886523273241764e-06,
      "loss": 0.6951,
      "step": 25071
    },
    {
      "epoch": 0.722411110470812,
      "grad_norm": 0.6969715356826782,
      "learning_rate": 1.888287077040229e-06,
      "loss": 0.6856,
      "step": 25072
    },
    {
      "epoch": 0.7224399239324613,
      "grad_norm": 0.7045347690582275,
      "learning_rate": 1.8879218538561889e-06,
      "loss": 0.6839,
      "step": 25073
    },
    {
      "epoch": 0.7224687373941105,
      "grad_norm": 0.7073053121566772,
      "learning_rate": 1.8875566577752403e-06,
      "loss": 0.6613,
      "step": 25074
    },
    {
      "epoch": 0.7224975508557598,
      "grad_norm": 0.7096446752548218,
      "learning_rate": 1.88719148880056e-06,
      "loss": 0.6459,
      "step": 25075
    },
    {
      "epoch": 0.7225263643174091,
      "grad_norm": 0.7534395456314087,
      "learning_rate": 1.886826346935332e-06,
      "loss": 0.7176,
      "step": 25076
    },
    {
      "epoch": 0.7225551777790584,
      "grad_norm": 0.7541497349739075,
      "learning_rate": 1.8864612321827313e-06,
      "loss": 0.6899,
      "step": 25077
    },
    {
      "epoch": 0.7225839912407077,
      "grad_norm": 0.7172791361808777,
      "learning_rate": 1.886096144545944e-06,
      "loss": 0.69,
      "step": 25078
    },
    {
      "epoch": 0.7226128047023569,
      "grad_norm": 0.7167650461196899,
      "learning_rate": 1.8857310840281445e-06,
      "loss": 0.7133,
      "step": 25079
    },
    {
      "epoch": 0.7226416181640062,
      "grad_norm": 0.7323018312454224,
      "learning_rate": 1.8853660506325166e-06,
      "loss": 0.6916,
      "step": 25080
    },
    {
      "epoch": 0.7226704316256555,
      "grad_norm": 0.7567322254180908,
      "learning_rate": 1.8850010443622346e-06,
      "loss": 0.6947,
      "step": 25081
    },
    {
      "epoch": 0.7226992450873048,
      "grad_norm": 0.7326058745384216,
      "learning_rate": 1.8846360652204814e-06,
      "loss": 0.7073,
      "step": 25082
    },
    {
      "epoch": 0.722728058548954,
      "grad_norm": 0.7111480832099915,
      "learning_rate": 1.8842711132104325e-06,
      "loss": 0.6728,
      "step": 25083
    },
    {
      "epoch": 0.7227568720106033,
      "grad_norm": 0.7058554887771606,
      "learning_rate": 1.8839061883352689e-06,
      "loss": 0.6852,
      "step": 25084
    },
    {
      "epoch": 0.7227856854722526,
      "grad_norm": 0.7416487336158752,
      "learning_rate": 1.8835412905981659e-06,
      "loss": 0.7003,
      "step": 25085
    },
    {
      "epoch": 0.7228144989339019,
      "grad_norm": 0.712216854095459,
      "learning_rate": 1.883176420002305e-06,
      "loss": 0.6741,
      "step": 25086
    },
    {
      "epoch": 0.7228433123955512,
      "grad_norm": 0.7378414273262024,
      "learning_rate": 1.8828115765508597e-06,
      "loss": 0.693,
      "step": 25087
    },
    {
      "epoch": 0.7228721258572005,
      "grad_norm": 0.719697117805481,
      "learning_rate": 1.8824467602470092e-06,
      "loss": 0.6666,
      "step": 25088
    },
    {
      "epoch": 0.7229009393188498,
      "grad_norm": 0.6999768018722534,
      "learning_rate": 1.8820819710939314e-06,
      "loss": 0.6776,
      "step": 25089
    },
    {
      "epoch": 0.7229297527804991,
      "grad_norm": 0.7466703057289124,
      "learning_rate": 1.881717209094805e-06,
      "loss": 0.699,
      "step": 25090
    },
    {
      "epoch": 0.7229585662421484,
      "grad_norm": 0.7469785809516907,
      "learning_rate": 1.8813524742528021e-06,
      "loss": 0.7228,
      "step": 25091
    },
    {
      "epoch": 0.7229873797037976,
      "grad_norm": 0.7362471222877502,
      "learning_rate": 1.8809877665711034e-06,
      "loss": 0.6924,
      "step": 25092
    },
    {
      "epoch": 0.7230161931654469,
      "grad_norm": 0.7330949902534485,
      "learning_rate": 1.8806230860528823e-06,
      "loss": 0.6677,
      "step": 25093
    },
    {
      "epoch": 0.7230450066270961,
      "grad_norm": 0.7234723567962646,
      "learning_rate": 1.8802584327013158e-06,
      "loss": 0.676,
      "step": 25094
    },
    {
      "epoch": 0.7230738200887454,
      "grad_norm": 0.701335608959198,
      "learning_rate": 1.879893806519581e-06,
      "loss": 0.6936,
      "step": 25095
    },
    {
      "epoch": 0.7231026335503947,
      "grad_norm": 0.7420838475227356,
      "learning_rate": 1.8795292075108506e-06,
      "loss": 0.7023,
      "step": 25096
    },
    {
      "epoch": 0.723131447012044,
      "grad_norm": 0.715775191783905,
      "learning_rate": 1.8791646356783032e-06,
      "loss": 0.6915,
      "step": 25097
    },
    {
      "epoch": 0.7231602604736933,
      "grad_norm": 0.7131571173667908,
      "learning_rate": 1.8788000910251102e-06,
      "loss": 0.6841,
      "step": 25098
    },
    {
      "epoch": 0.7231890739353426,
      "grad_norm": 0.7062501907348633,
      "learning_rate": 1.8784355735544507e-06,
      "loss": 0.6782,
      "step": 25099
    },
    {
      "epoch": 0.7232178873969919,
      "grad_norm": 0.7535465359687805,
      "learning_rate": 1.8780710832694949e-06,
      "loss": 0.6981,
      "step": 25100
    },
    {
      "epoch": 0.7232467008586412,
      "grad_norm": 0.7340343594551086,
      "learning_rate": 1.877706620173419e-06,
      "loss": 0.6882,
      "step": 25101
    },
    {
      "epoch": 0.7232755143202905,
      "grad_norm": 0.7199569344520569,
      "learning_rate": 1.8773421842693973e-06,
      "loss": 0.6792,
      "step": 25102
    },
    {
      "epoch": 0.7233043277819398,
      "grad_norm": 0.7189928293228149,
      "learning_rate": 1.8769777755606055e-06,
      "loss": 0.6911,
      "step": 25103
    },
    {
      "epoch": 0.723333141243589,
      "grad_norm": 0.7303962111473083,
      "learning_rate": 1.8766133940502134e-06,
      "loss": 0.6995,
      "step": 25104
    },
    {
      "epoch": 0.7233619547052383,
      "grad_norm": 0.7199637293815613,
      "learning_rate": 1.8762490397413985e-06,
      "loss": 0.6979,
      "step": 25105
    },
    {
      "epoch": 0.7233907681668875,
      "grad_norm": 0.7316281795501709,
      "learning_rate": 1.8758847126373304e-06,
      "loss": 0.6926,
      "step": 25106
    },
    {
      "epoch": 0.7234195816285368,
      "grad_norm": 0.7160825133323669,
      "learning_rate": 1.8755204127411848e-06,
      "loss": 0.6845,
      "step": 25107
    },
    {
      "epoch": 0.7234483950901861,
      "grad_norm": 0.707787275314331,
      "learning_rate": 1.8751561400561314e-06,
      "loss": 0.671,
      "step": 25108
    },
    {
      "epoch": 0.7234772085518354,
      "grad_norm": 0.7244834303855896,
      "learning_rate": 1.8747918945853439e-06,
      "loss": 0.706,
      "step": 25109
    },
    {
      "epoch": 0.7235060220134847,
      "grad_norm": 0.7257046103477478,
      "learning_rate": 1.874427676331997e-06,
      "loss": 0.6896,
      "step": 25110
    },
    {
      "epoch": 0.723534835475134,
      "grad_norm": 0.7205327749252319,
      "learning_rate": 1.874063485299259e-06,
      "loss": 0.687,
      "step": 25111
    },
    {
      "epoch": 0.7235636489367833,
      "grad_norm": 0.7161186933517456,
      "learning_rate": 1.873699321490305e-06,
      "loss": 0.7004,
      "step": 25112
    },
    {
      "epoch": 0.7235924623984326,
      "grad_norm": 0.7119237184524536,
      "learning_rate": 1.8733351849083025e-06,
      "loss": 0.6908,
      "step": 25113
    },
    {
      "epoch": 0.7236212758600818,
      "grad_norm": 0.7288590669631958,
      "learning_rate": 1.8729710755564256e-06,
      "loss": 0.712,
      "step": 25114
    },
    {
      "epoch": 0.7236500893217311,
      "grad_norm": 0.7047142386436462,
      "learning_rate": 1.8726069934378448e-06,
      "loss": 0.6777,
      "step": 25115
    },
    {
      "epoch": 0.7236789027833804,
      "grad_norm": 0.7136920690536499,
      "learning_rate": 1.8722429385557323e-06,
      "loss": 0.6668,
      "step": 25116
    },
    {
      "epoch": 0.7237077162450297,
      "grad_norm": 0.7194629311561584,
      "learning_rate": 1.8718789109132556e-06,
      "loss": 0.688,
      "step": 25117
    },
    {
      "epoch": 0.723736529706679,
      "grad_norm": 0.7330127358436584,
      "learning_rate": 1.8715149105135883e-06,
      "loss": 0.696,
      "step": 25118
    },
    {
      "epoch": 0.7237653431683282,
      "grad_norm": 0.7207164168357849,
      "learning_rate": 1.8711509373598969e-06,
      "loss": 0.6876,
      "step": 25119
    },
    {
      "epoch": 0.7237941566299775,
      "grad_norm": 0.7135283946990967,
      "learning_rate": 1.870786991455355e-06,
      "loss": 0.6701,
      "step": 25120
    },
    {
      "epoch": 0.7238229700916268,
      "grad_norm": 0.6872463822364807,
      "learning_rate": 1.8704230728031287e-06,
      "loss": 0.6598,
      "step": 25121
    },
    {
      "epoch": 0.7238517835532761,
      "grad_norm": 0.7412446737289429,
      "learning_rate": 1.8700591814063907e-06,
      "loss": 0.692,
      "step": 25122
    },
    {
      "epoch": 0.7238805970149254,
      "grad_norm": 0.70963454246521,
      "learning_rate": 1.8696953172683063e-06,
      "loss": 0.699,
      "step": 25123
    },
    {
      "epoch": 0.7239094104765746,
      "grad_norm": 0.7085332274436951,
      "learning_rate": 1.8693314803920487e-06,
      "loss": 0.6871,
      "step": 25124
    },
    {
      "epoch": 0.7239382239382239,
      "grad_norm": 0.7159886956214905,
      "learning_rate": 1.8689676707807808e-06,
      "loss": 0.6877,
      "step": 25125
    },
    {
      "epoch": 0.7239670373998732,
      "grad_norm": 0.7203993797302246,
      "learning_rate": 1.8686038884376777e-06,
      "loss": 0.6766,
      "step": 25126
    },
    {
      "epoch": 0.7239958508615225,
      "grad_norm": 0.7200666666030884,
      "learning_rate": 1.8682401333659034e-06,
      "loss": 0.6955,
      "step": 25127
    },
    {
      "epoch": 0.7240246643231718,
      "grad_norm": 0.7045425176620483,
      "learning_rate": 1.8678764055686283e-06,
      "loss": 0.6802,
      "step": 25128
    },
    {
      "epoch": 0.7240534777848211,
      "grad_norm": 0.7013364434242249,
      "learning_rate": 1.867512705049017e-06,
      "loss": 0.6953,
      "step": 25129
    },
    {
      "epoch": 0.7240822912464704,
      "grad_norm": 0.7120028734207153,
      "learning_rate": 1.867149031810241e-06,
      "loss": 0.6761,
      "step": 25130
    },
    {
      "epoch": 0.7241111047081197,
      "grad_norm": 0.7226235866546631,
      "learning_rate": 1.8667853858554625e-06,
      "loss": 0.6906,
      "step": 25131
    },
    {
      "epoch": 0.724139918169769,
      "grad_norm": 0.7259361147880554,
      "learning_rate": 1.8664217671878521e-06,
      "loss": 0.6589,
      "step": 25132
    },
    {
      "epoch": 0.7241687316314181,
      "grad_norm": 0.7113955020904541,
      "learning_rate": 1.8660581758105767e-06,
      "loss": 0.6799,
      "step": 25133
    },
    {
      "epoch": 0.7241975450930674,
      "grad_norm": 0.7336143255233765,
      "learning_rate": 1.8656946117268005e-06,
      "loss": 0.6944,
      "step": 25134
    },
    {
      "epoch": 0.7242263585547167,
      "grad_norm": 0.7101426720619202,
      "learning_rate": 1.8653310749396924e-06,
      "loss": 0.6901,
      "step": 25135
    },
    {
      "epoch": 0.724255172016366,
      "grad_norm": 0.7303134202957153,
      "learning_rate": 1.8649675654524157e-06,
      "loss": 0.6891,
      "step": 25136
    },
    {
      "epoch": 0.7242839854780153,
      "grad_norm": 0.7144711017608643,
      "learning_rate": 1.8646040832681378e-06,
      "loss": 0.6855,
      "step": 25137
    },
    {
      "epoch": 0.7243127989396646,
      "grad_norm": 0.7053363919258118,
      "learning_rate": 1.8642406283900238e-06,
      "loss": 0.6674,
      "step": 25138
    },
    {
      "epoch": 0.7243416124013139,
      "grad_norm": 0.7391722202301025,
      "learning_rate": 1.8638772008212414e-06,
      "loss": 0.6938,
      "step": 25139
    },
    {
      "epoch": 0.7243704258629632,
      "grad_norm": 0.7030418515205383,
      "learning_rate": 1.8635138005649516e-06,
      "loss": 0.6688,
      "step": 25140
    },
    {
      "epoch": 0.7243992393246125,
      "grad_norm": 0.7333713173866272,
      "learning_rate": 1.8631504276243229e-06,
      "loss": 0.692,
      "step": 25141
    },
    {
      "epoch": 0.7244280527862618,
      "grad_norm": 0.7095006704330444,
      "learning_rate": 1.8627870820025169e-06,
      "loss": 0.6844,
      "step": 25142
    },
    {
      "epoch": 0.7244568662479111,
      "grad_norm": 0.7727128267288208,
      "learning_rate": 1.8624237637027003e-06,
      "loss": 0.7033,
      "step": 25143
    },
    {
      "epoch": 0.7244856797095603,
      "grad_norm": 0.7207925915718079,
      "learning_rate": 1.862060472728035e-06,
      "loss": 0.684,
      "step": 25144
    },
    {
      "epoch": 0.7245144931712096,
      "grad_norm": 0.7280327081680298,
      "learning_rate": 1.8616972090816881e-06,
      "loss": 0.6703,
      "step": 25145
    },
    {
      "epoch": 0.7245433066328588,
      "grad_norm": 0.6973268985748291,
      "learning_rate": 1.8613339727668194e-06,
      "loss": 0.6792,
      "step": 25146
    },
    {
      "epoch": 0.7245721200945081,
      "grad_norm": 0.704461932182312,
      "learning_rate": 1.8609707637865948e-06,
      "loss": 0.685,
      "step": 25147
    },
    {
      "epoch": 0.7246009335561574,
      "grad_norm": 0.7568005919456482,
      "learning_rate": 1.8606075821441783e-06,
      "loss": 0.7148,
      "step": 25148
    },
    {
      "epoch": 0.7246297470178067,
      "grad_norm": 0.711615800857544,
      "learning_rate": 1.8602444278427295e-06,
      "loss": 0.6913,
      "step": 25149
    },
    {
      "epoch": 0.724658560479456,
      "grad_norm": 0.7292187213897705,
      "learning_rate": 1.8598813008854134e-06,
      "loss": 0.6952,
      "step": 25150
    },
    {
      "epoch": 0.7246873739411053,
      "grad_norm": 0.72020423412323,
      "learning_rate": 1.8595182012753943e-06,
      "loss": 0.6789,
      "step": 25151
    },
    {
      "epoch": 0.7247161874027546,
      "grad_norm": 0.7415705919265747,
      "learning_rate": 1.8591551290158299e-06,
      "loss": 0.7009,
      "step": 25152
    },
    {
      "epoch": 0.7247450008644039,
      "grad_norm": 0.7021457552909851,
      "learning_rate": 1.858792084109885e-06,
      "loss": 0.6859,
      "step": 25153
    },
    {
      "epoch": 0.7247738143260531,
      "grad_norm": 0.740300178527832,
      "learning_rate": 1.858429066560723e-06,
      "loss": 0.6955,
      "step": 25154
    },
    {
      "epoch": 0.7248026277877024,
      "grad_norm": 0.7375465035438538,
      "learning_rate": 1.8580660763715008e-06,
      "loss": 0.6789,
      "step": 25155
    },
    {
      "epoch": 0.7248314412493517,
      "grad_norm": 0.7292055487632751,
      "learning_rate": 1.8577031135453848e-06,
      "loss": 0.6732,
      "step": 25156
    },
    {
      "epoch": 0.724860254711001,
      "grad_norm": 0.7131723165512085,
      "learning_rate": 1.8573401780855315e-06,
      "loss": 0.7039,
      "step": 25157
    },
    {
      "epoch": 0.7248890681726503,
      "grad_norm": 0.7126054763793945,
      "learning_rate": 1.8569772699951055e-06,
      "loss": 0.6702,
      "step": 25158
    },
    {
      "epoch": 0.7249178816342996,
      "grad_norm": 0.7058071494102478,
      "learning_rate": 1.8566143892772643e-06,
      "loss": 0.6876,
      "step": 25159
    },
    {
      "epoch": 0.7249466950959488,
      "grad_norm": 0.7035775780677795,
      "learning_rate": 1.8562515359351707e-06,
      "loss": 0.674,
      "step": 25160
    },
    {
      "epoch": 0.7249755085575981,
      "grad_norm": 0.707365095615387,
      "learning_rate": 1.8558887099719802e-06,
      "loss": 0.6855,
      "step": 25161
    },
    {
      "epoch": 0.7250043220192474,
      "grad_norm": 0.7353331446647644,
      "learning_rate": 1.8555259113908597e-06,
      "loss": 0.725,
      "step": 25162
    },
    {
      "epoch": 0.7250331354808967,
      "grad_norm": 0.7300958037376404,
      "learning_rate": 1.8551631401949639e-06,
      "loss": 0.6933,
      "step": 25163
    },
    {
      "epoch": 0.7250619489425459,
      "grad_norm": 0.729671835899353,
      "learning_rate": 1.8548003963874544e-06,
      "loss": 0.7025,
      "step": 25164
    },
    {
      "epoch": 0.7250907624041952,
      "grad_norm": 0.7146118879318237,
      "learning_rate": 1.8544376799714881e-06,
      "loss": 0.6834,
      "step": 25165
    },
    {
      "epoch": 0.7251195758658445,
      "grad_norm": 0.7168354988098145,
      "learning_rate": 1.854074990950227e-06,
      "loss": 0.6823,
      "step": 25166
    },
    {
      "epoch": 0.7251483893274938,
      "grad_norm": 0.7198038697242737,
      "learning_rate": 1.853712329326826e-06,
      "loss": 0.6923,
      "step": 25167
    },
    {
      "epoch": 0.7251772027891431,
      "grad_norm": 0.7088112235069275,
      "learning_rate": 1.8533496951044456e-06,
      "loss": 0.6848,
      "step": 25168
    },
    {
      "epoch": 0.7252060162507924,
      "grad_norm": 0.7639641165733337,
      "learning_rate": 1.8529870882862455e-06,
      "loss": 0.6968,
      "step": 25169
    },
    {
      "epoch": 0.7252348297124417,
      "grad_norm": 0.7241339087486267,
      "learning_rate": 1.8526245088753802e-06,
      "loss": 0.6856,
      "step": 25170
    },
    {
      "epoch": 0.725263643174091,
      "grad_norm": 0.7033758759498596,
      "learning_rate": 1.8522619568750112e-06,
      "loss": 0.6783,
      "step": 25171
    },
    {
      "epoch": 0.7252924566357403,
      "grad_norm": 0.7236292362213135,
      "learning_rate": 1.8518994322882922e-06,
      "loss": 0.7116,
      "step": 25172
    },
    {
      "epoch": 0.7253212700973894,
      "grad_norm": 0.713525116443634,
      "learning_rate": 1.851536935118382e-06,
      "loss": 0.6994,
      "step": 25173
    },
    {
      "epoch": 0.7253500835590387,
      "grad_norm": 0.7155084013938904,
      "learning_rate": 1.8511744653684377e-06,
      "loss": 0.7032,
      "step": 25174
    },
    {
      "epoch": 0.725378897020688,
      "grad_norm": 0.7328453660011292,
      "learning_rate": 1.8508120230416181e-06,
      "loss": 0.7046,
      "step": 25175
    },
    {
      "epoch": 0.7254077104823373,
      "grad_norm": 0.7266165018081665,
      "learning_rate": 1.8504496081410762e-06,
      "loss": 0.6831,
      "step": 25176
    },
    {
      "epoch": 0.7254365239439866,
      "grad_norm": 0.7134408950805664,
      "learning_rate": 1.850087220669971e-06,
      "loss": 0.6681,
      "step": 25177
    },
    {
      "epoch": 0.7254653374056359,
      "grad_norm": 0.7219741344451904,
      "learning_rate": 1.849724860631456e-06,
      "loss": 0.6834,
      "step": 25178
    },
    {
      "epoch": 0.7254941508672852,
      "grad_norm": 0.7511993050575256,
      "learning_rate": 1.8493625280286897e-06,
      "loss": 0.7046,
      "step": 25179
    },
    {
      "epoch": 0.7255229643289345,
      "grad_norm": 0.7407367825508118,
      "learning_rate": 1.8490002228648253e-06,
      "loss": 0.6675,
      "step": 25180
    },
    {
      "epoch": 0.7255517777905838,
      "grad_norm": 0.8244193196296692,
      "learning_rate": 1.8486379451430197e-06,
      "loss": 0.6875,
      "step": 25181
    },
    {
      "epoch": 0.7255805912522331,
      "grad_norm": 0.7287607789039612,
      "learning_rate": 1.8482756948664265e-06,
      "loss": 0.672,
      "step": 25182
    },
    {
      "epoch": 0.7256094047138824,
      "grad_norm": 0.7228326797485352,
      "learning_rate": 1.847913472038203e-06,
      "loss": 0.6947,
      "step": 25183
    },
    {
      "epoch": 0.7256382181755316,
      "grad_norm": 0.7219559550285339,
      "learning_rate": 1.8475512766615005e-06,
      "loss": 0.6723,
      "step": 25184
    },
    {
      "epoch": 0.7256670316371809,
      "grad_norm": 0.700169563293457,
      "learning_rate": 1.847189108739475e-06,
      "loss": 0.7011,
      "step": 25185
    },
    {
      "epoch": 0.7256958450988302,
      "grad_norm": 0.718533992767334,
      "learning_rate": 1.8468269682752809e-06,
      "loss": 0.6924,
      "step": 25186
    },
    {
      "epoch": 0.7257246585604794,
      "grad_norm": 0.7145099639892578,
      "learning_rate": 1.8464648552720737e-06,
      "loss": 0.7137,
      "step": 25187
    },
    {
      "epoch": 0.7257534720221287,
      "grad_norm": 0.7423404455184937,
      "learning_rate": 1.846102769733003e-06,
      "loss": 0.6931,
      "step": 25188
    },
    {
      "epoch": 0.725782285483778,
      "grad_norm": 0.740084171295166,
      "learning_rate": 1.8457407116612265e-06,
      "loss": 0.7003,
      "step": 25189
    },
    {
      "epoch": 0.7258110989454273,
      "grad_norm": 0.723395586013794,
      "learning_rate": 1.8453786810598933e-06,
      "loss": 0.6812,
      "step": 25190
    },
    {
      "epoch": 0.7258399124070766,
      "grad_norm": 0.7130138278007507,
      "learning_rate": 1.8450166779321589e-06,
      "loss": 0.6701,
      "step": 25191
    },
    {
      "epoch": 0.7258687258687259,
      "grad_norm": 0.717089831829071,
      "learning_rate": 1.8446547022811767e-06,
      "loss": 0.6729,
      "step": 25192
    },
    {
      "epoch": 0.7258975393303752,
      "grad_norm": 0.7439119219779968,
      "learning_rate": 1.8442927541100963e-06,
      "loss": 0.6966,
      "step": 25193
    },
    {
      "epoch": 0.7259263527920244,
      "grad_norm": 0.7271906733512878,
      "learning_rate": 1.8439308334220729e-06,
      "loss": 0.6819,
      "step": 25194
    },
    {
      "epoch": 0.7259551662536737,
      "grad_norm": 0.7402828335762024,
      "learning_rate": 1.8435689402202555e-06,
      "loss": 0.6785,
      "step": 25195
    },
    {
      "epoch": 0.725983979715323,
      "grad_norm": 0.7458587288856506,
      "learning_rate": 1.843207074507799e-06,
      "loss": 0.7146,
      "step": 25196
    },
    {
      "epoch": 0.7260127931769723,
      "grad_norm": 0.7373050451278687,
      "learning_rate": 1.8428452362878501e-06,
      "loss": 0.7102,
      "step": 25197
    },
    {
      "epoch": 0.7260416066386216,
      "grad_norm": 0.7287267446517944,
      "learning_rate": 1.8424834255635664e-06,
      "loss": 0.7077,
      "step": 25198
    },
    {
      "epoch": 0.7260704201002709,
      "grad_norm": 0.7158463597297668,
      "learning_rate": 1.842121642338094e-06,
      "loss": 0.6627,
      "step": 25199
    },
    {
      "epoch": 0.7260992335619201,
      "grad_norm": 0.7179487943649292,
      "learning_rate": 1.8417598866145868e-06,
      "loss": 0.7086,
      "step": 25200
    },
    {
      "epoch": 0.7261280470235694,
      "grad_norm": 0.7279611825942993,
      "learning_rate": 1.8413981583961926e-06,
      "loss": 0.6792,
      "step": 25201
    },
    {
      "epoch": 0.7261568604852187,
      "grad_norm": 0.7465738654136658,
      "learning_rate": 1.8410364576860645e-06,
      "loss": 0.667,
      "step": 25202
    },
    {
      "epoch": 0.726185673946868,
      "grad_norm": 0.7099936604499817,
      "learning_rate": 1.8406747844873495e-06,
      "loss": 0.6966,
      "step": 25203
    },
    {
      "epoch": 0.7262144874085172,
      "grad_norm": 0.7074798345565796,
      "learning_rate": 1.8403131388032009e-06,
      "loss": 0.6663,
      "step": 25204
    },
    {
      "epoch": 0.7262433008701665,
      "grad_norm": 0.7201412916183472,
      "learning_rate": 1.839951520636764e-06,
      "loss": 0.7001,
      "step": 25205
    },
    {
      "epoch": 0.7262721143318158,
      "grad_norm": 0.7036045789718628,
      "learning_rate": 1.8395899299911907e-06,
      "loss": 0.6777,
      "step": 25206
    },
    {
      "epoch": 0.7263009277934651,
      "grad_norm": 0.7079740762710571,
      "learning_rate": 1.8392283668696315e-06,
      "loss": 0.6702,
      "step": 25207
    },
    {
      "epoch": 0.7263297412551144,
      "grad_norm": 0.7218616604804993,
      "learning_rate": 1.8388668312752323e-06,
      "loss": 0.6822,
      "step": 25208
    },
    {
      "epoch": 0.7263585547167637,
      "grad_norm": 0.7016682624816895,
      "learning_rate": 1.8385053232111426e-06,
      "loss": 0.6963,
      "step": 25209
    },
    {
      "epoch": 0.726387368178413,
      "grad_norm": 0.7093663215637207,
      "learning_rate": 1.838143842680513e-06,
      "loss": 0.6625,
      "step": 25210
    },
    {
      "epoch": 0.7264161816400623,
      "grad_norm": 0.6983906626701355,
      "learning_rate": 1.8377823896864882e-06,
      "loss": 0.6457,
      "step": 25211
    },
    {
      "epoch": 0.7264449951017116,
      "grad_norm": 0.7156877517700195,
      "learning_rate": 1.8374209642322177e-06,
      "loss": 0.6769,
      "step": 25212
    },
    {
      "epoch": 0.7264738085633609,
      "grad_norm": 0.7131766676902771,
      "learning_rate": 1.8370595663208506e-06,
      "loss": 0.6699,
      "step": 25213
    },
    {
      "epoch": 0.72650262202501,
      "grad_norm": 0.7101571559906006,
      "learning_rate": 1.836698195955532e-06,
      "loss": 0.6852,
      "step": 25214
    },
    {
      "epoch": 0.7265314354866593,
      "grad_norm": 0.7028458714485168,
      "learning_rate": 1.836336853139411e-06,
      "loss": 0.6791,
      "step": 25215
    },
    {
      "epoch": 0.7265602489483086,
      "grad_norm": 0.7162619233131409,
      "learning_rate": 1.8359755378756317e-06,
      "loss": 0.6952,
      "step": 25216
    },
    {
      "epoch": 0.7265890624099579,
      "grad_norm": 0.7238685488700867,
      "learning_rate": 1.8356142501673446e-06,
      "loss": 0.6999,
      "step": 25217
    },
    {
      "epoch": 0.7266178758716072,
      "grad_norm": 0.7350056767463684,
      "learning_rate": 1.8352529900176924e-06,
      "loss": 0.674,
      "step": 25218
    },
    {
      "epoch": 0.7266466893332565,
      "grad_norm": 0.7020804286003113,
      "learning_rate": 1.8348917574298248e-06,
      "loss": 0.6743,
      "step": 25219
    },
    {
      "epoch": 0.7266755027949058,
      "grad_norm": 0.7360683679580688,
      "learning_rate": 1.8345305524068846e-06,
      "loss": 0.6983,
      "step": 25220
    },
    {
      "epoch": 0.7267043162565551,
      "grad_norm": 0.7286463975906372,
      "learning_rate": 1.8341693749520185e-06,
      "loss": 0.6945,
      "step": 25221
    },
    {
      "epoch": 0.7267331297182044,
      "grad_norm": 0.7229282855987549,
      "learning_rate": 1.8338082250683725e-06,
      "loss": 0.6812,
      "step": 25222
    },
    {
      "epoch": 0.7267619431798537,
      "grad_norm": 0.7102741599082947,
      "learning_rate": 1.8334471027590938e-06,
      "loss": 0.6902,
      "step": 25223
    },
    {
      "epoch": 0.7267907566415029,
      "grad_norm": 0.7133212089538574,
      "learning_rate": 1.833086008027324e-06,
      "loss": 0.6697,
      "step": 25224
    },
    {
      "epoch": 0.7268195701031522,
      "grad_norm": 0.73778235912323,
      "learning_rate": 1.8327249408762104e-06,
      "loss": 0.7112,
      "step": 25225
    },
    {
      "epoch": 0.7268483835648015,
      "grad_norm": 0.7187771797180176,
      "learning_rate": 1.8323639013088952e-06,
      "loss": 0.7041,
      "step": 25226
    },
    {
      "epoch": 0.7268771970264507,
      "grad_norm": 0.7303265929222107,
      "learning_rate": 1.8320028893285251e-06,
      "loss": 0.6815,
      "step": 25227
    },
    {
      "epoch": 0.7269060104881,
      "grad_norm": 0.7334094047546387,
      "learning_rate": 1.8316419049382416e-06,
      "loss": 0.7021,
      "step": 25228
    },
    {
      "epoch": 0.7269348239497493,
      "grad_norm": 0.7067137956619263,
      "learning_rate": 1.8312809481411897e-06,
      "loss": 0.7001,
      "step": 25229
    },
    {
      "epoch": 0.7269636374113986,
      "grad_norm": 0.7106738686561584,
      "learning_rate": 1.8309200189405152e-06,
      "loss": 0.6893,
      "step": 25230
    },
    {
      "epoch": 0.7269924508730479,
      "grad_norm": 0.7258710265159607,
      "learning_rate": 1.8305591173393577e-06,
      "loss": 0.6827,
      "step": 25231
    },
    {
      "epoch": 0.7270212643346972,
      "grad_norm": 0.710247814655304,
      "learning_rate": 1.8301982433408638e-06,
      "loss": 0.6971,
      "step": 25232
    },
    {
      "epoch": 0.7270500777963465,
      "grad_norm": 0.7167842388153076,
      "learning_rate": 1.829837396948171e-06,
      "loss": 0.6523,
      "step": 25233
    },
    {
      "epoch": 0.7270788912579957,
      "grad_norm": 0.7169893383979797,
      "learning_rate": 1.8294765781644285e-06,
      "loss": 0.7028,
      "step": 25234
    },
    {
      "epoch": 0.727107704719645,
      "grad_norm": 0.7058070302009583,
      "learning_rate": 1.8291157869927746e-06,
      "loss": 0.6654,
      "step": 25235
    },
    {
      "epoch": 0.7271365181812943,
      "grad_norm": 0.7231160998344421,
      "learning_rate": 1.8287550234363539e-06,
      "loss": 0.6878,
      "step": 25236
    },
    {
      "epoch": 0.7271653316429436,
      "grad_norm": 0.707424521446228,
      "learning_rate": 1.828394287498305e-06,
      "loss": 0.6775,
      "step": 25237
    },
    {
      "epoch": 0.7271941451045929,
      "grad_norm": 0.7183484435081482,
      "learning_rate": 1.8280335791817733e-06,
      "loss": 0.6755,
      "step": 25238
    },
    {
      "epoch": 0.7272229585662422,
      "grad_norm": 0.7170640826225281,
      "learning_rate": 1.8276728984898968e-06,
      "loss": 0.7246,
      "step": 25239
    },
    {
      "epoch": 0.7272517720278915,
      "grad_norm": 0.7136106491088867,
      "learning_rate": 1.8273122454258197e-06,
      "loss": 0.6883,
      "step": 25240
    },
    {
      "epoch": 0.7272805854895407,
      "grad_norm": 0.6940943598747253,
      "learning_rate": 1.8269516199926796e-06,
      "loss": 0.6704,
      "step": 25241
    },
    {
      "epoch": 0.72730939895119,
      "grad_norm": 0.7204453349113464,
      "learning_rate": 1.8265910221936206e-06,
      "loss": 0.6952,
      "step": 25242
    },
    {
      "epoch": 0.7273382124128392,
      "grad_norm": 0.7167473435401917,
      "learning_rate": 1.8262304520317804e-06,
      "loss": 0.6754,
      "step": 25243
    },
    {
      "epoch": 0.7273670258744885,
      "grad_norm": 0.7014899849891663,
      "learning_rate": 1.8258699095103e-06,
      "loss": 0.6815,
      "step": 25244
    },
    {
      "epoch": 0.7273958393361378,
      "grad_norm": 0.7206301093101501,
      "learning_rate": 1.8255093946323193e-06,
      "loss": 0.6855,
      "step": 25245
    },
    {
      "epoch": 0.7274246527977871,
      "grad_norm": 0.7086716890335083,
      "learning_rate": 1.8251489074009805e-06,
      "loss": 0.6768,
      "step": 25246
    },
    {
      "epoch": 0.7274534662594364,
      "grad_norm": 0.7101688385009766,
      "learning_rate": 1.8247884478194195e-06,
      "loss": 0.677,
      "step": 25247
    },
    {
      "epoch": 0.7274822797210857,
      "grad_norm": 0.7042746543884277,
      "learning_rate": 1.8244280158907785e-06,
      "loss": 0.6737,
      "step": 25248
    },
    {
      "epoch": 0.727511093182735,
      "grad_norm": 0.7182536125183105,
      "learning_rate": 1.8240676116181938e-06,
      "loss": 0.6891,
      "step": 25249
    },
    {
      "epoch": 0.7275399066443843,
      "grad_norm": 0.7691565155982971,
      "learning_rate": 1.823707235004805e-06,
      "loss": 0.6948,
      "step": 25250
    },
    {
      "epoch": 0.7275687201060336,
      "grad_norm": 0.7234476804733276,
      "learning_rate": 1.823346886053753e-06,
      "loss": 0.7158,
      "step": 25251
    },
    {
      "epoch": 0.7275975335676829,
      "grad_norm": 0.7105529308319092,
      "learning_rate": 1.8229865647681722e-06,
      "loss": 0.6802,
      "step": 25252
    },
    {
      "epoch": 0.7276263470293322,
      "grad_norm": 0.7510053515434265,
      "learning_rate": 1.8226262711512043e-06,
      "loss": 0.6875,
      "step": 25253
    },
    {
      "epoch": 0.7276551604909813,
      "grad_norm": 0.7420360445976257,
      "learning_rate": 1.8222660052059837e-06,
      "loss": 0.6813,
      "step": 25254
    },
    {
      "epoch": 0.7276839739526306,
      "grad_norm": 0.7255150079727173,
      "learning_rate": 1.8219057669356515e-06,
      "loss": 0.7113,
      "step": 25255
    },
    {
      "epoch": 0.7277127874142799,
      "grad_norm": 0.7166736125946045,
      "learning_rate": 1.8215455563433399e-06,
      "loss": 0.6854,
      "step": 25256
    },
    {
      "epoch": 0.7277416008759292,
      "grad_norm": 0.7082299590110779,
      "learning_rate": 1.8211853734321922e-06,
      "loss": 0.681,
      "step": 25257
    },
    {
      "epoch": 0.7277704143375785,
      "grad_norm": 0.7054564952850342,
      "learning_rate": 1.8208252182053405e-06,
      "loss": 0.6795,
      "step": 25258
    },
    {
      "epoch": 0.7277992277992278,
      "grad_norm": 0.7078080177307129,
      "learning_rate": 1.820465090665925e-06,
      "loss": 0.6703,
      "step": 25259
    },
    {
      "epoch": 0.7278280412608771,
      "grad_norm": 0.7214711308479309,
      "learning_rate": 1.8201049908170787e-06,
      "loss": 0.6959,
      "step": 25260
    },
    {
      "epoch": 0.7278568547225264,
      "grad_norm": 0.7173570394515991,
      "learning_rate": 1.819744918661941e-06,
      "loss": 0.7001,
      "step": 25261
    },
    {
      "epoch": 0.7278856681841757,
      "grad_norm": 0.7299048900604248,
      "learning_rate": 1.8193848742036435e-06,
      "loss": 0.686,
      "step": 25262
    },
    {
      "epoch": 0.727914481645825,
      "grad_norm": 0.7034804224967957,
      "learning_rate": 1.8190248574453267e-06,
      "loss": 0.6919,
      "step": 25263
    },
    {
      "epoch": 0.7279432951074742,
      "grad_norm": 0.7362469434738159,
      "learning_rate": 1.8186648683901214e-06,
      "loss": 0.6884,
      "step": 25264
    },
    {
      "epoch": 0.7279721085691235,
      "grad_norm": 0.7333921194076538,
      "learning_rate": 1.8183049070411674e-06,
      "loss": 0.6733,
      "step": 25265
    },
    {
      "epoch": 0.7280009220307728,
      "grad_norm": 0.7087094783782959,
      "learning_rate": 1.8179449734015946e-06,
      "loss": 0.7029,
      "step": 25266
    },
    {
      "epoch": 0.7280297354924221,
      "grad_norm": 0.7164771556854248,
      "learning_rate": 1.8175850674745404e-06,
      "loss": 0.6725,
      "step": 25267
    },
    {
      "epoch": 0.7280585489540713,
      "grad_norm": 0.7292649745941162,
      "learning_rate": 1.8172251892631392e-06,
      "loss": 0.68,
      "step": 25268
    },
    {
      "epoch": 0.7280873624157206,
      "grad_norm": 0.7404031157493591,
      "learning_rate": 1.8168653387705266e-06,
      "loss": 0.684,
      "step": 25269
    },
    {
      "epoch": 0.7281161758773699,
      "grad_norm": 0.7191637754440308,
      "learning_rate": 1.8165055159998335e-06,
      "loss": 0.7125,
      "step": 25270
    },
    {
      "epoch": 0.7281449893390192,
      "grad_norm": 0.7297422289848328,
      "learning_rate": 1.8161457209541944e-06,
      "loss": 0.669,
      "step": 25271
    },
    {
      "epoch": 0.7281738028006685,
      "grad_norm": 0.726229190826416,
      "learning_rate": 1.8157859536367456e-06,
      "loss": 0.6781,
      "step": 25272
    },
    {
      "epoch": 0.7282026162623177,
      "grad_norm": 0.7295852899551392,
      "learning_rate": 1.8154262140506162e-06,
      "loss": 0.6925,
      "step": 25273
    },
    {
      "epoch": 0.728231429723967,
      "grad_norm": 0.7045154571533203,
      "learning_rate": 1.8150665021989427e-06,
      "loss": 0.6825,
      "step": 25274
    },
    {
      "epoch": 0.7282602431856163,
      "grad_norm": 0.7228100299835205,
      "learning_rate": 1.8147068180848548e-06,
      "loss": 0.6938,
      "step": 25275
    },
    {
      "epoch": 0.7282890566472656,
      "grad_norm": 0.7399854063987732,
      "learning_rate": 1.8143471617114878e-06,
      "loss": 0.7151,
      "step": 25276
    },
    {
      "epoch": 0.7283178701089149,
      "grad_norm": 0.7347914576530457,
      "learning_rate": 1.8139875330819707e-06,
      "loss": 0.6861,
      "step": 25277
    },
    {
      "epoch": 0.7283466835705642,
      "grad_norm": 0.7249792218208313,
      "learning_rate": 1.8136279321994394e-06,
      "loss": 0.6823,
      "step": 25278
    },
    {
      "epoch": 0.7283754970322135,
      "grad_norm": 0.720513641834259,
      "learning_rate": 1.8132683590670219e-06,
      "loss": 0.7063,
      "step": 25279
    },
    {
      "epoch": 0.7284043104938628,
      "grad_norm": 0.6999866366386414,
      "learning_rate": 1.8129088136878515e-06,
      "loss": 0.6671,
      "step": 25280
    },
    {
      "epoch": 0.728433123955512,
      "grad_norm": 0.7104081511497498,
      "learning_rate": 1.8125492960650586e-06,
      "loss": 0.6899,
      "step": 25281
    },
    {
      "epoch": 0.7284619374171613,
      "grad_norm": 0.7040360569953918,
      "learning_rate": 1.812189806201778e-06,
      "loss": 0.6855,
      "step": 25282
    },
    {
      "epoch": 0.7284907508788105,
      "grad_norm": 0.7263453006744385,
      "learning_rate": 1.8118303441011348e-06,
      "loss": 0.708,
      "step": 25283
    },
    {
      "epoch": 0.7285195643404598,
      "grad_norm": 0.7216576933860779,
      "learning_rate": 1.811470909766264e-06,
      "loss": 0.6781,
      "step": 25284
    },
    {
      "epoch": 0.7285483778021091,
      "grad_norm": 0.6935247182846069,
      "learning_rate": 1.8111115032002935e-06,
      "loss": 0.7155,
      "step": 25285
    },
    {
      "epoch": 0.7285771912637584,
      "grad_norm": 0.7027208805084229,
      "learning_rate": 1.8107521244063548e-06,
      "loss": 0.6974,
      "step": 25286
    },
    {
      "epoch": 0.7286060047254077,
      "grad_norm": 0.7159631252288818,
      "learning_rate": 1.8103927733875758e-06,
      "loss": 0.6791,
      "step": 25287
    },
    {
      "epoch": 0.728634818187057,
      "grad_norm": 0.7195504903793335,
      "learning_rate": 1.8100334501470873e-06,
      "loss": 0.6894,
      "step": 25288
    },
    {
      "epoch": 0.7286636316487063,
      "grad_norm": 0.7063275575637817,
      "learning_rate": 1.8096741546880204e-06,
      "loss": 0.6786,
      "step": 25289
    },
    {
      "epoch": 0.7286924451103556,
      "grad_norm": 0.7326130867004395,
      "learning_rate": 1.8093148870135003e-06,
      "loss": 0.6956,
      "step": 25290
    },
    {
      "epoch": 0.7287212585720049,
      "grad_norm": 0.7102136611938477,
      "learning_rate": 1.8089556471266601e-06,
      "loss": 0.6798,
      "step": 25291
    },
    {
      "epoch": 0.7287500720336542,
      "grad_norm": 0.7252016067504883,
      "learning_rate": 1.8085964350306228e-06,
      "loss": 0.688,
      "step": 25292
    },
    {
      "epoch": 0.7287788854953035,
      "grad_norm": 0.7190676927566528,
      "learning_rate": 1.808237250728524e-06,
      "loss": 0.6959,
      "step": 25293
    },
    {
      "epoch": 0.7288076989569527,
      "grad_norm": 0.7145550847053528,
      "learning_rate": 1.8078780942234863e-06,
      "loss": 0.6823,
      "step": 25294
    },
    {
      "epoch": 0.7288365124186019,
      "grad_norm": 0.7062216401100159,
      "learning_rate": 1.807518965518641e-06,
      "loss": 0.6782,
      "step": 25295
    },
    {
      "epoch": 0.7288653258802512,
      "grad_norm": 0.7476263046264648,
      "learning_rate": 1.8071598646171123e-06,
      "loss": 0.7115,
      "step": 25296
    },
    {
      "epoch": 0.7288941393419005,
      "grad_norm": 0.7489199042320251,
      "learning_rate": 1.8068007915220315e-06,
      "loss": 0.7058,
      "step": 25297
    },
    {
      "epoch": 0.7289229528035498,
      "grad_norm": 0.7192673683166504,
      "learning_rate": 1.8064417462365225e-06,
      "loss": 0.711,
      "step": 25298
    },
    {
      "epoch": 0.7289517662651991,
      "grad_norm": 0.723037838935852,
      "learning_rate": 1.8060827287637145e-06,
      "loss": 0.6941,
      "step": 25299
    },
    {
      "epoch": 0.7289805797268484,
      "grad_norm": 0.6986650228500366,
      "learning_rate": 1.8057237391067316e-06,
      "loss": 0.6539,
      "step": 25300
    },
    {
      "epoch": 0.7290093931884977,
      "grad_norm": 0.7168740630149841,
      "learning_rate": 1.8053647772687038e-06,
      "loss": 0.7207,
      "step": 25301
    },
    {
      "epoch": 0.729038206650147,
      "grad_norm": 0.7217769026756287,
      "learning_rate": 1.8050058432527534e-06,
      "loss": 0.688,
      "step": 25302
    },
    {
      "epoch": 0.7290670201117962,
      "grad_norm": 0.7085407376289368,
      "learning_rate": 1.8046469370620102e-06,
      "loss": 0.6847,
      "step": 25303
    },
    {
      "epoch": 0.7290958335734455,
      "grad_norm": 0.7314993739128113,
      "learning_rate": 1.8042880586995943e-06,
      "loss": 0.6797,
      "step": 25304
    },
    {
      "epoch": 0.7291246470350948,
      "grad_norm": 0.7175490856170654,
      "learning_rate": 1.8039292081686388e-06,
      "loss": 0.6839,
      "step": 25305
    },
    {
      "epoch": 0.7291534604967441,
      "grad_norm": 0.7231070399284363,
      "learning_rate": 1.8035703854722624e-06,
      "loss": 0.6911,
      "step": 25306
    },
    {
      "epoch": 0.7291822739583934,
      "grad_norm": 0.7179484963417053,
      "learning_rate": 1.803211590613595e-06,
      "loss": 0.6839,
      "step": 25307
    },
    {
      "epoch": 0.7292110874200426,
      "grad_norm": 0.7082943320274353,
      "learning_rate": 1.8028528235957577e-06,
      "loss": 0.6976,
      "step": 25308
    },
    {
      "epoch": 0.7292399008816919,
      "grad_norm": 0.7137569189071655,
      "learning_rate": 1.8024940844218758e-06,
      "loss": 0.707,
      "step": 25309
    },
    {
      "epoch": 0.7292687143433412,
      "grad_norm": 0.7442719340324402,
      "learning_rate": 1.8021353730950758e-06,
      "loss": 0.6906,
      "step": 25310
    },
    {
      "epoch": 0.7292975278049905,
      "grad_norm": 0.7224791646003723,
      "learning_rate": 1.801776689618478e-06,
      "loss": 0.6685,
      "step": 25311
    },
    {
      "epoch": 0.7293263412666398,
      "grad_norm": 0.7408201098442078,
      "learning_rate": 1.8014180339952102e-06,
      "loss": 0.684,
      "step": 25312
    },
    {
      "epoch": 0.729355154728289,
      "grad_norm": 0.7084918022155762,
      "learning_rate": 1.8010594062283926e-06,
      "loss": 0.6967,
      "step": 25313
    },
    {
      "epoch": 0.7293839681899383,
      "grad_norm": 0.7172513604164124,
      "learning_rate": 1.8007008063211512e-06,
      "loss": 0.7008,
      "step": 25314
    },
    {
      "epoch": 0.7294127816515876,
      "grad_norm": 0.7059162855148315,
      "learning_rate": 1.8003422342766059e-06,
      "loss": 0.6846,
      "step": 25315
    },
    {
      "epoch": 0.7294415951132369,
      "grad_norm": 0.7112867832183838,
      "learning_rate": 1.7999836900978813e-06,
      "loss": 0.6824,
      "step": 25316
    },
    {
      "epoch": 0.7294704085748862,
      "grad_norm": 0.6893907189369202,
      "learning_rate": 1.7996251737880998e-06,
      "loss": 0.6964,
      "step": 25317
    },
    {
      "epoch": 0.7294992220365355,
      "grad_norm": 0.7049188613891602,
      "learning_rate": 1.7992666853503854e-06,
      "loss": 0.6813,
      "step": 25318
    },
    {
      "epoch": 0.7295280354981848,
      "grad_norm": 0.7013640999794006,
      "learning_rate": 1.798908224787857e-06,
      "loss": 0.6972,
      "step": 25319
    },
    {
      "epoch": 0.7295568489598341,
      "grad_norm": 0.7082501649856567,
      "learning_rate": 1.79854979210364e-06,
      "loss": 0.6859,
      "step": 25320
    },
    {
      "epoch": 0.7295856624214834,
      "grad_norm": 0.7106777429580688,
      "learning_rate": 1.7981913873008521e-06,
      "loss": 0.6818,
      "step": 25321
    },
    {
      "epoch": 0.7296144758831326,
      "grad_norm": 0.6981722712516785,
      "learning_rate": 1.7978330103826186e-06,
      "loss": 0.6791,
      "step": 25322
    },
    {
      "epoch": 0.7296432893447818,
      "grad_norm": 0.7324439883232117,
      "learning_rate": 1.7974746613520566e-06,
      "loss": 0.6979,
      "step": 25323
    },
    {
      "epoch": 0.7296721028064311,
      "grad_norm": 0.7301058173179626,
      "learning_rate": 1.7971163402122905e-06,
      "loss": 0.6969,
      "step": 25324
    },
    {
      "epoch": 0.7297009162680804,
      "grad_norm": 0.7143253684043884,
      "learning_rate": 1.796758046966438e-06,
      "loss": 0.6919,
      "step": 25325
    },
    {
      "epoch": 0.7297297297297297,
      "grad_norm": 0.7078993916511536,
      "learning_rate": 1.7963997816176204e-06,
      "loss": 0.6926,
      "step": 25326
    },
    {
      "epoch": 0.729758543191379,
      "grad_norm": 0.7025889754295349,
      "learning_rate": 1.7960415441689604e-06,
      "loss": 0.6926,
      "step": 25327
    },
    {
      "epoch": 0.7297873566530283,
      "grad_norm": 0.7192100882530212,
      "learning_rate": 1.7956833346235741e-06,
      "loss": 0.7052,
      "step": 25328
    },
    {
      "epoch": 0.7298161701146776,
      "grad_norm": 0.708816409111023,
      "learning_rate": 1.7953251529845828e-06,
      "loss": 0.6706,
      "step": 25329
    },
    {
      "epoch": 0.7298449835763269,
      "grad_norm": 0.7286770343780518,
      "learning_rate": 1.7949669992551056e-06,
      "loss": 0.706,
      "step": 25330
    },
    {
      "epoch": 0.7298737970379762,
      "grad_norm": 0.7175610661506653,
      "learning_rate": 1.7946088734382638e-06,
      "loss": 0.6898,
      "step": 25331
    },
    {
      "epoch": 0.7299026104996255,
      "grad_norm": 0.7166604995727539,
      "learning_rate": 1.794250775537173e-06,
      "loss": 0.6866,
      "step": 25332
    },
    {
      "epoch": 0.7299314239612748,
      "grad_norm": 0.7953423857688904,
      "learning_rate": 1.7938927055549548e-06,
      "loss": 0.6891,
      "step": 25333
    },
    {
      "epoch": 0.729960237422924,
      "grad_norm": 0.7426555752754211,
      "learning_rate": 1.7935346634947248e-06,
      "loss": 0.6909,
      "step": 25334
    },
    {
      "epoch": 0.7299890508845732,
      "grad_norm": 0.7222840785980225,
      "learning_rate": 1.7931766493596037e-06,
      "loss": 0.7252,
      "step": 25335
    },
    {
      "epoch": 0.7300178643462225,
      "grad_norm": 0.7316310405731201,
      "learning_rate": 1.7928186631527072e-06,
      "loss": 0.6845,
      "step": 25336
    },
    {
      "epoch": 0.7300466778078718,
      "grad_norm": 0.7245233654975891,
      "learning_rate": 1.792460704877156e-06,
      "loss": 0.6857,
      "step": 25337
    },
    {
      "epoch": 0.7300754912695211,
      "grad_norm": 0.6990588307380676,
      "learning_rate": 1.792102774536063e-06,
      "loss": 0.6856,
      "step": 25338
    },
    {
      "epoch": 0.7301043047311704,
      "grad_norm": 0.7239516973495483,
      "learning_rate": 1.791744872132551e-06,
      "loss": 0.6794,
      "step": 25339
    },
    {
      "epoch": 0.7301331181928197,
      "grad_norm": 0.7209356427192688,
      "learning_rate": 1.7913869976697302e-06,
      "loss": 0.6751,
      "step": 25340
    },
    {
      "epoch": 0.730161931654469,
      "grad_norm": 0.7033998370170593,
      "learning_rate": 1.7910291511507245e-06,
      "loss": 0.6725,
      "step": 25341
    },
    {
      "epoch": 0.7301907451161183,
      "grad_norm": 0.7091752886772156,
      "learning_rate": 1.790671332578645e-06,
      "loss": 0.6916,
      "step": 25342
    },
    {
      "epoch": 0.7302195585777675,
      "grad_norm": 0.7054109573364258,
      "learning_rate": 1.7903135419566125e-06,
      "loss": 0.6834,
      "step": 25343
    },
    {
      "epoch": 0.7302483720394168,
      "grad_norm": 0.7082759141921997,
      "learning_rate": 1.7899557792877386e-06,
      "loss": 0.6746,
      "step": 25344
    },
    {
      "epoch": 0.7302771855010661,
      "grad_norm": 0.737899661064148,
      "learning_rate": 1.7895980445751427e-06,
      "loss": 0.6975,
      "step": 25345
    },
    {
      "epoch": 0.7303059989627154,
      "grad_norm": 0.7369727492332458,
      "learning_rate": 1.7892403378219365e-06,
      "loss": 0.6833,
      "step": 25346
    },
    {
      "epoch": 0.7303348124243647,
      "grad_norm": 0.7311269044876099,
      "learning_rate": 1.788882659031238e-06,
      "loss": 0.6991,
      "step": 25347
    },
    {
      "epoch": 0.730363625886014,
      "grad_norm": 0.722373366355896,
      "learning_rate": 1.7885250082061627e-06,
      "loss": 0.6761,
      "step": 25348
    },
    {
      "epoch": 0.7303924393476632,
      "grad_norm": 0.8797628879547119,
      "learning_rate": 1.7881673853498232e-06,
      "loss": 0.694,
      "step": 25349
    },
    {
      "epoch": 0.7304212528093125,
      "grad_norm": 0.7297316193580627,
      "learning_rate": 1.7878097904653364e-06,
      "loss": 0.6989,
      "step": 25350
    },
    {
      "epoch": 0.7304500662709618,
      "grad_norm": 0.7323988676071167,
      "learning_rate": 1.7874522235558133e-06,
      "loss": 0.7121,
      "step": 25351
    },
    {
      "epoch": 0.730478879732611,
      "grad_norm": 0.6924667358398438,
      "learning_rate": 1.7870946846243708e-06,
      "loss": 0.6772,
      "step": 25352
    },
    {
      "epoch": 0.7305076931942603,
      "grad_norm": 0.7167013883590698,
      "learning_rate": 1.7867371736741212e-06,
      "loss": 0.6781,
      "step": 25353
    },
    {
      "epoch": 0.7305365066559096,
      "grad_norm": 0.7596858143806458,
      "learning_rate": 1.786379690708181e-06,
      "loss": 0.7048,
      "step": 25354
    },
    {
      "epoch": 0.7305653201175589,
      "grad_norm": 0.7266009449958801,
      "learning_rate": 1.7860222357296592e-06,
      "loss": 0.6815,
      "step": 25355
    },
    {
      "epoch": 0.7305941335792082,
      "grad_norm": 0.6988731622695923,
      "learning_rate": 1.7856648087416721e-06,
      "loss": 0.6934,
      "step": 25356
    },
    {
      "epoch": 0.7306229470408575,
      "grad_norm": 0.7206021547317505,
      "learning_rate": 1.7853074097473306e-06,
      "loss": 0.7059,
      "step": 25357
    },
    {
      "epoch": 0.7306517605025068,
      "grad_norm": 0.7160583138465881,
      "learning_rate": 1.7849500387497492e-06,
      "loss": 0.6904,
      "step": 25358
    },
    {
      "epoch": 0.7306805739641561,
      "grad_norm": 0.7230913043022156,
      "learning_rate": 1.7845926957520377e-06,
      "loss": 0.7088,
      "step": 25359
    },
    {
      "epoch": 0.7307093874258054,
      "grad_norm": 0.731865406036377,
      "learning_rate": 1.784235380757311e-06,
      "loss": 0.6468,
      "step": 25360
    },
    {
      "epoch": 0.7307382008874547,
      "grad_norm": 0.7255080342292786,
      "learning_rate": 1.7838780937686778e-06,
      "loss": 0.7102,
      "step": 25361
    },
    {
      "epoch": 0.7307670143491038,
      "grad_norm": 0.7102993726730347,
      "learning_rate": 1.7835208347892536e-06,
      "loss": 0.6875,
      "step": 25362
    },
    {
      "epoch": 0.7307958278107531,
      "grad_norm": 0.7511783838272095,
      "learning_rate": 1.7831636038221456e-06,
      "loss": 0.7033,
      "step": 25363
    },
    {
      "epoch": 0.7308246412724024,
      "grad_norm": 0.7185407280921936,
      "learning_rate": 1.782806400870467e-06,
      "loss": 0.6804,
      "step": 25364
    },
    {
      "epoch": 0.7308534547340517,
      "grad_norm": 0.748916506767273,
      "learning_rate": 1.7824492259373288e-06,
      "loss": 0.7274,
      "step": 25365
    },
    {
      "epoch": 0.730882268195701,
      "grad_norm": 0.722653329372406,
      "learning_rate": 1.7820920790258428e-06,
      "loss": 0.6918,
      "step": 25366
    },
    {
      "epoch": 0.7309110816573503,
      "grad_norm": 0.760021984577179,
      "learning_rate": 1.7817349601391165e-06,
      "loss": 0.7039,
      "step": 25367
    },
    {
      "epoch": 0.7309398951189996,
      "grad_norm": 0.6936745047569275,
      "learning_rate": 1.781377869280262e-06,
      "loss": 0.6807,
      "step": 25368
    },
    {
      "epoch": 0.7309687085806489,
      "grad_norm": 0.7204429507255554,
      "learning_rate": 1.7810208064523898e-06,
      "loss": 0.6861,
      "step": 25369
    },
    {
      "epoch": 0.7309975220422982,
      "grad_norm": 0.7167443037033081,
      "learning_rate": 1.7806637716586072e-06,
      "loss": 0.6818,
      "step": 25370
    },
    {
      "epoch": 0.7310263355039475,
      "grad_norm": 0.714072585105896,
      "learning_rate": 1.7803067649020267e-06,
      "loss": 0.68,
      "step": 25371
    },
    {
      "epoch": 0.7310551489655968,
      "grad_norm": 0.7110313177108765,
      "learning_rate": 1.7799497861857546e-06,
      "loss": 0.7132,
      "step": 25372
    },
    {
      "epoch": 0.731083962427246,
      "grad_norm": 0.7206441760063171,
      "learning_rate": 1.7795928355129022e-06,
      "loss": 0.6845,
      "step": 25373
    },
    {
      "epoch": 0.7311127758888953,
      "grad_norm": 0.7421890497207642,
      "learning_rate": 1.7792359128865749e-06,
      "loss": 0.666,
      "step": 25374
    },
    {
      "epoch": 0.7311415893505446,
      "grad_norm": 1.0588315725326538,
      "learning_rate": 1.7788790183098852e-06,
      "loss": 0.6676,
      "step": 25375
    },
    {
      "epoch": 0.7311704028121938,
      "grad_norm": 0.7253909111022949,
      "learning_rate": 1.7785221517859364e-06,
      "loss": 0.6938,
      "step": 25376
    },
    {
      "epoch": 0.7311992162738431,
      "grad_norm": 0.7547322511672974,
      "learning_rate": 1.7781653133178422e-06,
      "loss": 0.6933,
      "step": 25377
    },
    {
      "epoch": 0.7312280297354924,
      "grad_norm": 0.7186493277549744,
      "learning_rate": 1.7778085029087061e-06,
      "loss": 0.692,
      "step": 25378
    },
    {
      "epoch": 0.7312568431971417,
      "grad_norm": 0.7104902863502502,
      "learning_rate": 1.7774517205616382e-06,
      "loss": 0.6688,
      "step": 25379
    },
    {
      "epoch": 0.731285656658791,
      "grad_norm": 0.7087825536727905,
      "learning_rate": 1.7770949662797433e-06,
      "loss": 0.7017,
      "step": 25380
    },
    {
      "epoch": 0.7313144701204403,
      "grad_norm": 0.7234501838684082,
      "learning_rate": 1.7767382400661303e-06,
      "loss": 0.6915,
      "step": 25381
    },
    {
      "epoch": 0.7313432835820896,
      "grad_norm": 0.7214551568031311,
      "learning_rate": 1.7763815419239045e-06,
      "loss": 0.6709,
      "step": 25382
    },
    {
      "epoch": 0.7313720970437388,
      "grad_norm": 0.7166338562965393,
      "learning_rate": 1.776024871856174e-06,
      "loss": 0.6974,
      "step": 25383
    },
    {
      "epoch": 0.7314009105053881,
      "grad_norm": 0.7086072564125061,
      "learning_rate": 1.7756682298660427e-06,
      "loss": 0.6814,
      "step": 25384
    },
    {
      "epoch": 0.7314297239670374,
      "grad_norm": 0.6935688853263855,
      "learning_rate": 1.7753116159566181e-06,
      "loss": 0.6512,
      "step": 25385
    },
    {
      "epoch": 0.7314585374286867,
      "grad_norm": 0.7470173835754395,
      "learning_rate": 1.7749550301310075e-06,
      "loss": 0.6935,
      "step": 25386
    },
    {
      "epoch": 0.731487350890336,
      "grad_norm": 0.7237868905067444,
      "learning_rate": 1.7745984723923127e-06,
      "loss": 0.6923,
      "step": 25387
    },
    {
      "epoch": 0.7315161643519853,
      "grad_norm": 0.720099687576294,
      "learning_rate": 1.7742419427436413e-06,
      "loss": 0.6656,
      "step": 25388
    },
    {
      "epoch": 0.7315449778136345,
      "grad_norm": 0.7064680457115173,
      "learning_rate": 1.773885441188098e-06,
      "loss": 0.6861,
      "step": 25389
    },
    {
      "epoch": 0.7315737912752838,
      "grad_norm": 0.709965169429779,
      "learning_rate": 1.7735289677287886e-06,
      "loss": 0.6743,
      "step": 25390
    },
    {
      "epoch": 0.7316026047369331,
      "grad_norm": 0.7235297560691833,
      "learning_rate": 1.7731725223688156e-06,
      "loss": 0.6663,
      "step": 25391
    },
    {
      "epoch": 0.7316314181985824,
      "grad_norm": 0.7238336801528931,
      "learning_rate": 1.7728161051112858e-06,
      "loss": 0.6878,
      "step": 25392
    },
    {
      "epoch": 0.7316602316602316,
      "grad_norm": 0.6989381909370422,
      "learning_rate": 1.7724597159592998e-06,
      "loss": 0.68,
      "step": 25393
    },
    {
      "epoch": 0.7316890451218809,
      "grad_norm": 0.7377497553825378,
      "learning_rate": 1.7721033549159656e-06,
      "loss": 0.6928,
      "step": 25394
    },
    {
      "epoch": 0.7317178585835302,
      "grad_norm": 0.6925659775733948,
      "learning_rate": 1.7717470219843824e-06,
      "loss": 0.6873,
      "step": 25395
    },
    {
      "epoch": 0.7317466720451795,
      "grad_norm": 0.7543690204620361,
      "learning_rate": 1.771390717167657e-06,
      "loss": 0.7095,
      "step": 25396
    },
    {
      "epoch": 0.7317754855068288,
      "grad_norm": 0.7264520525932312,
      "learning_rate": 1.7710344404688896e-06,
      "loss": 0.7007,
      "step": 25397
    },
    {
      "epoch": 0.7318042989684781,
      "grad_norm": 0.738893985748291,
      "learning_rate": 1.7706781918911865e-06,
      "loss": 0.7113,
      "step": 25398
    },
    {
      "epoch": 0.7318331124301274,
      "grad_norm": 0.7188284397125244,
      "learning_rate": 1.7703219714376458e-06,
      "loss": 0.6878,
      "step": 25399
    },
    {
      "epoch": 0.7318619258917767,
      "grad_norm": 0.7382367253303528,
      "learning_rate": 1.7699657791113728e-06,
      "loss": 0.665,
      "step": 25400
    },
    {
      "epoch": 0.731890739353426,
      "grad_norm": 0.7351583242416382,
      "learning_rate": 1.7696096149154684e-06,
      "loss": 0.7035,
      "step": 25401
    },
    {
      "epoch": 0.7319195528150753,
      "grad_norm": 0.7291179299354553,
      "learning_rate": 1.7692534788530375e-06,
      "loss": 0.6958,
      "step": 25402
    },
    {
      "epoch": 0.7319483662767244,
      "grad_norm": 0.6968433856964111,
      "learning_rate": 1.7688973709271772e-06,
      "loss": 0.678,
      "step": 25403
    },
    {
      "epoch": 0.7319771797383737,
      "grad_norm": 0.7299618721008301,
      "learning_rate": 1.7685412911409927e-06,
      "loss": 0.6932,
      "step": 25404
    },
    {
      "epoch": 0.732005993200023,
      "grad_norm": 0.7124055624008179,
      "learning_rate": 1.7681852394975812e-06,
      "loss": 0.7084,
      "step": 25405
    },
    {
      "epoch": 0.7320348066616723,
      "grad_norm": 0.7114500999450684,
      "learning_rate": 1.7678292160000459e-06,
      "loss": 0.7196,
      "step": 25406
    },
    {
      "epoch": 0.7320636201233216,
      "grad_norm": 0.7243598699569702,
      "learning_rate": 1.767473220651489e-06,
      "loss": 0.6807,
      "step": 25407
    },
    {
      "epoch": 0.7320924335849709,
      "grad_norm": 0.7408185005187988,
      "learning_rate": 1.7671172534550068e-06,
      "loss": 0.702,
      "step": 25408
    },
    {
      "epoch": 0.7321212470466202,
      "grad_norm": 0.7441732287406921,
      "learning_rate": 1.7667613144137036e-06,
      "loss": 0.6839,
      "step": 25409
    },
    {
      "epoch": 0.7321500605082695,
      "grad_norm": 0.7068857550621033,
      "learning_rate": 1.7664054035306754e-06,
      "loss": 0.6632,
      "step": 25410
    },
    {
      "epoch": 0.7321788739699188,
      "grad_norm": 0.723892867565155,
      "learning_rate": 1.7660495208090261e-06,
      "loss": 0.693,
      "step": 25411
    },
    {
      "epoch": 0.732207687431568,
      "grad_norm": 0.7406616806983948,
      "learning_rate": 1.7656936662518487e-06,
      "loss": 0.6816,
      "step": 25412
    },
    {
      "epoch": 0.7322365008932173,
      "grad_norm": 0.7213314771652222,
      "learning_rate": 1.76533783986225e-06,
      "loss": 0.6881,
      "step": 25413
    },
    {
      "epoch": 0.7322653143548666,
      "grad_norm": 0.7315065264701843,
      "learning_rate": 1.7649820416433232e-06,
      "loss": 0.6903,
      "step": 25414
    },
    {
      "epoch": 0.7322941278165159,
      "grad_norm": 0.710669994354248,
      "learning_rate": 1.764626271598171e-06,
      "loss": 0.6642,
      "step": 25415
    },
    {
      "epoch": 0.7323229412781651,
      "grad_norm": 0.7171844244003296,
      "learning_rate": 1.7642705297298878e-06,
      "loss": 0.6817,
      "step": 25416
    },
    {
      "epoch": 0.7323517547398144,
      "grad_norm": 0.7351605296134949,
      "learning_rate": 1.763914816041576e-06,
      "loss": 0.6883,
      "step": 25417
    },
    {
      "epoch": 0.7323805682014637,
      "grad_norm": 0.7349781394004822,
      "learning_rate": 1.7635591305363293e-06,
      "loss": 0.6878,
      "step": 25418
    },
    {
      "epoch": 0.732409381663113,
      "grad_norm": 0.722513735294342,
      "learning_rate": 1.7632034732172488e-06,
      "loss": 0.6731,
      "step": 25419
    },
    {
      "epoch": 0.7324381951247623,
      "grad_norm": 0.722570538520813,
      "learning_rate": 1.7628478440874286e-06,
      "loss": 0.6808,
      "step": 25420
    },
    {
      "epoch": 0.7324670085864116,
      "grad_norm": 0.7198770642280579,
      "learning_rate": 1.7624922431499697e-06,
      "loss": 0.6769,
      "step": 25421
    },
    {
      "epoch": 0.7324958220480609,
      "grad_norm": 0.7058025002479553,
      "learning_rate": 1.7621366704079657e-06,
      "loss": 0.675,
      "step": 25422
    },
    {
      "epoch": 0.7325246355097101,
      "grad_norm": 0.7241466045379639,
      "learning_rate": 1.761781125864514e-06,
      "loss": 0.6926,
      "step": 25423
    },
    {
      "epoch": 0.7325534489713594,
      "grad_norm": 0.7180801033973694,
      "learning_rate": 1.7614256095227117e-06,
      "loss": 0.7125,
      "step": 25424
    },
    {
      "epoch": 0.7325822624330087,
      "grad_norm": 0.7178471088409424,
      "learning_rate": 1.7610701213856567e-06,
      "loss": 0.7027,
      "step": 25425
    },
    {
      "epoch": 0.732611075894658,
      "grad_norm": 0.7486597299575806,
      "learning_rate": 1.7607146614564418e-06,
      "loss": 0.7073,
      "step": 25426
    },
    {
      "epoch": 0.7326398893563073,
      "grad_norm": 0.717728316783905,
      "learning_rate": 1.760359229738164e-06,
      "loss": 0.6816,
      "step": 25427
    },
    {
      "epoch": 0.7326687028179566,
      "grad_norm": 0.7259030342102051,
      "learning_rate": 1.7600038262339202e-06,
      "loss": 0.6842,
      "step": 25428
    },
    {
      "epoch": 0.7326975162796059,
      "grad_norm": 0.7208936214447021,
      "learning_rate": 1.759648450946803e-06,
      "loss": 0.7028,
      "step": 25429
    },
    {
      "epoch": 0.7327263297412551,
      "grad_norm": 0.732081413269043,
      "learning_rate": 1.7592931038799104e-06,
      "loss": 0.7021,
      "step": 25430
    },
    {
      "epoch": 0.7327551432029044,
      "grad_norm": 0.7115381360054016,
      "learning_rate": 1.7589377850363332e-06,
      "loss": 0.6993,
      "step": 25431
    },
    {
      "epoch": 0.7327839566645536,
      "grad_norm": 0.7195394039154053,
      "learning_rate": 1.75858249441917e-06,
      "loss": 0.7057,
      "step": 25432
    },
    {
      "epoch": 0.7328127701262029,
      "grad_norm": 0.6939002275466919,
      "learning_rate": 1.7582272320315114e-06,
      "loss": 0.6765,
      "step": 25433
    },
    {
      "epoch": 0.7328415835878522,
      "grad_norm": 0.7365391254425049,
      "learning_rate": 1.7578719978764547e-06,
      "loss": 0.6877,
      "step": 25434
    },
    {
      "epoch": 0.7328703970495015,
      "grad_norm": 0.7106760740280151,
      "learning_rate": 1.7575167919570902e-06,
      "loss": 0.6884,
      "step": 25435
    },
    {
      "epoch": 0.7328992105111508,
      "grad_norm": 0.7258074283599854,
      "learning_rate": 1.757161614276513e-06,
      "loss": 0.7087,
      "step": 25436
    },
    {
      "epoch": 0.7329280239728001,
      "grad_norm": 0.7260487079620361,
      "learning_rate": 1.7568064648378164e-06,
      "loss": 0.6997,
      "step": 25437
    },
    {
      "epoch": 0.7329568374344494,
      "grad_norm": 0.7118896842002869,
      "learning_rate": 1.7564513436440955e-06,
      "loss": 0.6606,
      "step": 25438
    },
    {
      "epoch": 0.7329856508960987,
      "grad_norm": 0.7120228409767151,
      "learning_rate": 1.7560962506984392e-06,
      "loss": 0.69,
      "step": 25439
    },
    {
      "epoch": 0.733014464357748,
      "grad_norm": 0.7122225761413574,
      "learning_rate": 1.7557411860039437e-06,
      "loss": 0.6869,
      "step": 25440
    },
    {
      "epoch": 0.7330432778193973,
      "grad_norm": 0.6996551752090454,
      "learning_rate": 1.755386149563697e-06,
      "loss": 0.6863,
      "step": 25441
    },
    {
      "epoch": 0.7330720912810466,
      "grad_norm": 0.720051646232605,
      "learning_rate": 1.7550311413807959e-06,
      "loss": 0.6814,
      "step": 25442
    },
    {
      "epoch": 0.7331009047426957,
      "grad_norm": 0.7470346689224243,
      "learning_rate": 1.754676161458328e-06,
      "loss": 0.703,
      "step": 25443
    },
    {
      "epoch": 0.733129718204345,
      "grad_norm": 0.7139832973480225,
      "learning_rate": 1.7543212097993868e-06,
      "loss": 0.7074,
      "step": 25444
    },
    {
      "epoch": 0.7331585316659943,
      "grad_norm": 0.7297103404998779,
      "learning_rate": 1.7539662864070645e-06,
      "loss": 0.6623,
      "step": 25445
    },
    {
      "epoch": 0.7331873451276436,
      "grad_norm": 0.7108128070831299,
      "learning_rate": 1.75361139128445e-06,
      "loss": 0.6965,
      "step": 25446
    },
    {
      "epoch": 0.7332161585892929,
      "grad_norm": 0.7396955490112305,
      "learning_rate": 1.7532565244346362e-06,
      "loss": 0.6898,
      "step": 25447
    },
    {
      "epoch": 0.7332449720509422,
      "grad_norm": 0.7172630429267883,
      "learning_rate": 1.7529016858607107e-06,
      "loss": 0.689,
      "step": 25448
    },
    {
      "epoch": 0.7332737855125915,
      "grad_norm": 0.7166556119918823,
      "learning_rate": 1.752546875565766e-06,
      "loss": 0.6924,
      "step": 25449
    },
    {
      "epoch": 0.7333025989742408,
      "grad_norm": 0.7188957929611206,
      "learning_rate": 1.7521920935528918e-06,
      "loss": 0.6953,
      "step": 25450
    },
    {
      "epoch": 0.7333314124358901,
      "grad_norm": 0.6903784871101379,
      "learning_rate": 1.7518373398251792e-06,
      "loss": 0.6555,
      "step": 25451
    },
    {
      "epoch": 0.7333602258975394,
      "grad_norm": 0.7075906991958618,
      "learning_rate": 1.751482614385715e-06,
      "loss": 0.6785,
      "step": 25452
    },
    {
      "epoch": 0.7333890393591886,
      "grad_norm": 0.6974655389785767,
      "learning_rate": 1.7511279172375917e-06,
      "loss": 0.6775,
      "step": 25453
    },
    {
      "epoch": 0.7334178528208379,
      "grad_norm": 0.7200148105621338,
      "learning_rate": 1.7507732483838946e-06,
      "loss": 0.6944,
      "step": 25454
    },
    {
      "epoch": 0.7334466662824872,
      "grad_norm": 0.719281017780304,
      "learning_rate": 1.7504186078277158e-06,
      "loss": 0.7091,
      "step": 25455
    },
    {
      "epoch": 0.7334754797441365,
      "grad_norm": 0.7717992067337036,
      "learning_rate": 1.7500639955721417e-06,
      "loss": 0.691,
      "step": 25456
    },
    {
      "epoch": 0.7335042932057857,
      "grad_norm": 0.7093045115470886,
      "learning_rate": 1.7497094116202622e-06,
      "loss": 0.6927,
      "step": 25457
    },
    {
      "epoch": 0.733533106667435,
      "grad_norm": 0.7452431917190552,
      "learning_rate": 1.7493548559751638e-06,
      "loss": 0.7253,
      "step": 25458
    },
    {
      "epoch": 0.7335619201290843,
      "grad_norm": 0.7456854581832886,
      "learning_rate": 1.7490003286399366e-06,
      "loss": 0.6815,
      "step": 25459
    },
    {
      "epoch": 0.7335907335907336,
      "grad_norm": 0.7045573592185974,
      "learning_rate": 1.7486458296176628e-06,
      "loss": 0.7004,
      "step": 25460
    },
    {
      "epoch": 0.7336195470523829,
      "grad_norm": 0.7167637944221497,
      "learning_rate": 1.7482913589114376e-06,
      "loss": 0.6823,
      "step": 25461
    },
    {
      "epoch": 0.7336483605140321,
      "grad_norm": 0.7114520072937012,
      "learning_rate": 1.747936916524342e-06,
      "loss": 0.6777,
      "step": 25462
    },
    {
      "epoch": 0.7336771739756814,
      "grad_norm": 0.7157968878746033,
      "learning_rate": 1.7475825024594672e-06,
      "loss": 0.6854,
      "step": 25463
    },
    {
      "epoch": 0.7337059874373307,
      "grad_norm": 0.7207950949668884,
      "learning_rate": 1.7472281167198957e-06,
      "loss": 0.6942,
      "step": 25464
    },
    {
      "epoch": 0.73373480089898,
      "grad_norm": 0.7034598588943481,
      "learning_rate": 1.746873759308716e-06,
      "loss": 0.6863,
      "step": 25465
    },
    {
      "epoch": 0.7337636143606293,
      "grad_norm": 0.7161581516265869,
      "learning_rate": 1.746519430229015e-06,
      "loss": 0.7072,
      "step": 25466
    },
    {
      "epoch": 0.7337924278222786,
      "grad_norm": 0.7183427214622498,
      "learning_rate": 1.746165129483876e-06,
      "loss": 0.6947,
      "step": 25467
    },
    {
      "epoch": 0.7338212412839279,
      "grad_norm": 0.722125232219696,
      "learning_rate": 1.7458108570763882e-06,
      "loss": 0.6998,
      "step": 25468
    },
    {
      "epoch": 0.7338500547455772,
      "grad_norm": 0.8178844451904297,
      "learning_rate": 1.7454566130096328e-06,
      "loss": 0.6826,
      "step": 25469
    },
    {
      "epoch": 0.7338788682072264,
      "grad_norm": 0.7137895822525024,
      "learning_rate": 1.7451023972866981e-06,
      "loss": 0.6724,
      "step": 25470
    },
    {
      "epoch": 0.7339076816688757,
      "grad_norm": 0.7148104906082153,
      "learning_rate": 1.7447482099106672e-06,
      "loss": 0.7053,
      "step": 25471
    },
    {
      "epoch": 0.733936495130525,
      "grad_norm": 0.7045305967330933,
      "learning_rate": 1.7443940508846247e-06,
      "loss": 0.6661,
      "step": 25472
    },
    {
      "epoch": 0.7339653085921742,
      "grad_norm": 0.7433460354804993,
      "learning_rate": 1.7440399202116554e-06,
      "loss": 0.6873,
      "step": 25473
    },
    {
      "epoch": 0.7339941220538235,
      "grad_norm": 0.7322896122932434,
      "learning_rate": 1.7436858178948457e-06,
      "loss": 0.7153,
      "step": 25474
    },
    {
      "epoch": 0.7340229355154728,
      "grad_norm": 0.7277959585189819,
      "learning_rate": 1.7433317439372754e-06,
      "loss": 0.6834,
      "step": 25475
    },
    {
      "epoch": 0.7340517489771221,
      "grad_norm": 0.7094186544418335,
      "learning_rate": 1.7429776983420322e-06,
      "loss": 0.7001,
      "step": 25476
    },
    {
      "epoch": 0.7340805624387714,
      "grad_norm": 0.7104710936546326,
      "learning_rate": 1.7426236811121955e-06,
      "loss": 0.6768,
      "step": 25477
    },
    {
      "epoch": 0.7341093759004207,
      "grad_norm": 0.7370224595069885,
      "learning_rate": 1.7422696922508526e-06,
      "loss": 0.6834,
      "step": 25478
    },
    {
      "epoch": 0.73413818936207,
      "grad_norm": 0.7071545124053955,
      "learning_rate": 1.741915731761082e-06,
      "loss": 0.6844,
      "step": 25479
    },
    {
      "epoch": 0.7341670028237193,
      "grad_norm": 0.6991296410560608,
      "learning_rate": 1.74156179964597e-06,
      "loss": 0.6941,
      "step": 25480
    },
    {
      "epoch": 0.7341958162853686,
      "grad_norm": 0.710453987121582,
      "learning_rate": 1.741207895908596e-06,
      "loss": 0.6754,
      "step": 25481
    },
    {
      "epoch": 0.7342246297470179,
      "grad_norm": 0.7108107805252075,
      "learning_rate": 1.7408540205520436e-06,
      "loss": 0.6865,
      "step": 25482
    },
    {
      "epoch": 0.7342534432086671,
      "grad_norm": 0.7085833549499512,
      "learning_rate": 1.7405001735793964e-06,
      "loss": 0.6885,
      "step": 25483
    },
    {
      "epoch": 0.7342822566703163,
      "grad_norm": 0.7273765206336975,
      "learning_rate": 1.7401463549937326e-06,
      "loss": 0.7085,
      "step": 25484
    },
    {
      "epoch": 0.7343110701319656,
      "grad_norm": 0.7201477885246277,
      "learning_rate": 1.739792564798135e-06,
      "loss": 0.6831,
      "step": 25485
    },
    {
      "epoch": 0.7343398835936149,
      "grad_norm": 0.7126592397689819,
      "learning_rate": 1.739438802995686e-06,
      "loss": 0.6746,
      "step": 25486
    },
    {
      "epoch": 0.7343686970552642,
      "grad_norm": 0.7328776717185974,
      "learning_rate": 1.7390850695894667e-06,
      "loss": 0.7063,
      "step": 25487
    },
    {
      "epoch": 0.7343975105169135,
      "grad_norm": 0.727872908115387,
      "learning_rate": 1.7387313645825554e-06,
      "loss": 0.7015,
      "step": 25488
    },
    {
      "epoch": 0.7344263239785628,
      "grad_norm": 0.7050641179084778,
      "learning_rate": 1.7383776879780357e-06,
      "loss": 0.6851,
      "step": 25489
    },
    {
      "epoch": 0.7344551374402121,
      "grad_norm": 0.7098739743232727,
      "learning_rate": 1.7380240397789838e-06,
      "loss": 0.7068,
      "step": 25490
    },
    {
      "epoch": 0.7344839509018614,
      "grad_norm": 0.7103658318519592,
      "learning_rate": 1.7376704199884837e-06,
      "loss": 0.6759,
      "step": 25491
    },
    {
      "epoch": 0.7345127643635107,
      "grad_norm": 0.7245928645133972,
      "learning_rate": 1.737316828609611e-06,
      "loss": 0.6788,
      "step": 25492
    },
    {
      "epoch": 0.7345415778251599,
      "grad_norm": 0.7093212604522705,
      "learning_rate": 1.7369632656454494e-06,
      "loss": 0.6975,
      "step": 25493
    },
    {
      "epoch": 0.7345703912868092,
      "grad_norm": 0.7579019069671631,
      "learning_rate": 1.7366097310990737e-06,
      "loss": 0.7048,
      "step": 25494
    },
    {
      "epoch": 0.7345992047484585,
      "grad_norm": 0.7397403717041016,
      "learning_rate": 1.736256224973567e-06,
      "loss": 0.6943,
      "step": 25495
    },
    {
      "epoch": 0.7346280182101078,
      "grad_norm": 0.7085822224617004,
      "learning_rate": 1.7359027472720031e-06,
      "loss": 0.709,
      "step": 25496
    },
    {
      "epoch": 0.734656831671757,
      "grad_norm": 0.7047423124313354,
      "learning_rate": 1.735549297997466e-06,
      "loss": 0.6971,
      "step": 25497
    },
    {
      "epoch": 0.7346856451334063,
      "grad_norm": 0.710152804851532,
      "learning_rate": 1.7351958771530297e-06,
      "loss": 0.6945,
      "step": 25498
    },
    {
      "epoch": 0.7347144585950556,
      "grad_norm": 0.7081186771392822,
      "learning_rate": 1.734842484741776e-06,
      "loss": 0.728,
      "step": 25499
    },
    {
      "epoch": 0.7347432720567049,
      "grad_norm": 0.7076707482337952,
      "learning_rate": 1.734489120766778e-06,
      "loss": 0.7132,
      "step": 25500
    },
    {
      "epoch": 0.7347720855183542,
      "grad_norm": 0.7262732982635498,
      "learning_rate": 1.7341357852311175e-06,
      "loss": 0.7124,
      "step": 25501
    },
    {
      "epoch": 0.7348008989800034,
      "grad_norm": 0.693296492099762,
      "learning_rate": 1.7337824781378675e-06,
      "loss": 0.6735,
      "step": 25502
    },
    {
      "epoch": 0.7348297124416527,
      "grad_norm": 0.722664475440979,
      "learning_rate": 1.7334291994901076e-06,
      "loss": 0.6748,
      "step": 25503
    },
    {
      "epoch": 0.734858525903302,
      "grad_norm": 0.6939300298690796,
      "learning_rate": 1.7330759492909155e-06,
      "loss": 0.6698,
      "step": 25504
    },
    {
      "epoch": 0.7348873393649513,
      "grad_norm": 0.7271312475204468,
      "learning_rate": 1.7327227275433646e-06,
      "loss": 0.6943,
      "step": 25505
    },
    {
      "epoch": 0.7349161528266006,
      "grad_norm": 0.7189471125602722,
      "learning_rate": 1.7323695342505343e-06,
      "loss": 0.7262,
      "step": 25506
    },
    {
      "epoch": 0.7349449662882499,
      "grad_norm": 0.7171571254730225,
      "learning_rate": 1.7320163694154968e-06,
      "loss": 0.6952,
      "step": 25507
    },
    {
      "epoch": 0.7349737797498992,
      "grad_norm": 0.7225068807601929,
      "learning_rate": 1.7316632330413301e-06,
      "loss": 0.6701,
      "step": 25508
    },
    {
      "epoch": 0.7350025932115485,
      "grad_norm": 0.7138825058937073,
      "learning_rate": 1.73131012513111e-06,
      "loss": 0.6706,
      "step": 25509
    },
    {
      "epoch": 0.7350314066731977,
      "grad_norm": 0.7172941565513611,
      "learning_rate": 1.730957045687912e-06,
      "loss": 0.7056,
      "step": 25510
    },
    {
      "epoch": 0.735060220134847,
      "grad_norm": 0.7285824418067932,
      "learning_rate": 1.7306039947148095e-06,
      "loss": 0.6985,
      "step": 25511
    },
    {
      "epoch": 0.7350890335964962,
      "grad_norm": 0.714851975440979,
      "learning_rate": 1.7302509722148791e-06,
      "loss": 0.7054,
      "step": 25512
    },
    {
      "epoch": 0.7351178470581455,
      "grad_norm": 0.70821613073349,
      "learning_rate": 1.729897978191193e-06,
      "loss": 0.6706,
      "step": 25513
    },
    {
      "epoch": 0.7351466605197948,
      "grad_norm": 0.7310963869094849,
      "learning_rate": 1.7295450126468284e-06,
      "loss": 0.6736,
      "step": 25514
    },
    {
      "epoch": 0.7351754739814441,
      "grad_norm": 0.7110371589660645,
      "learning_rate": 1.7291920755848551e-06,
      "loss": 0.6815,
      "step": 25515
    },
    {
      "epoch": 0.7352042874430934,
      "grad_norm": 0.7026920318603516,
      "learning_rate": 1.728839167008351e-06,
      "loss": 0.6944,
      "step": 25516
    },
    {
      "epoch": 0.7352331009047427,
      "grad_norm": 0.7067514061927795,
      "learning_rate": 1.7284862869203867e-06,
      "loss": 0.7086,
      "step": 25517
    },
    {
      "epoch": 0.735261914366392,
      "grad_norm": 0.7209156155586243,
      "learning_rate": 1.7281334353240376e-06,
      "loss": 0.6792,
      "step": 25518
    },
    {
      "epoch": 0.7352907278280413,
      "grad_norm": 0.7192842364311218,
      "learning_rate": 1.7277806122223728e-06,
      "loss": 0.7022,
      "step": 25519
    },
    {
      "epoch": 0.7353195412896906,
      "grad_norm": 0.7373778223991394,
      "learning_rate": 1.7274278176184706e-06,
      "loss": 0.6917,
      "step": 25520
    },
    {
      "epoch": 0.7353483547513399,
      "grad_norm": 4.114299774169922,
      "learning_rate": 1.7270750515153994e-06,
      "loss": 0.6729,
      "step": 25521
    },
    {
      "epoch": 0.7353771682129892,
      "grad_norm": 0.7138318419456482,
      "learning_rate": 1.7267223139162343e-06,
      "loss": 0.6939,
      "step": 25522
    },
    {
      "epoch": 0.7354059816746384,
      "grad_norm": 0.7314566969871521,
      "learning_rate": 1.7263696048240442e-06,
      "loss": 0.7218,
      "step": 25523
    },
    {
      "epoch": 0.7354347951362876,
      "grad_norm": 0.7286038398742676,
      "learning_rate": 1.7260169242419023e-06,
      "loss": 0.6931,
      "step": 25524
    },
    {
      "epoch": 0.7354636085979369,
      "grad_norm": 0.7273135185241699,
      "learning_rate": 1.7256642721728822e-06,
      "loss": 0.6918,
      "step": 25525
    },
    {
      "epoch": 0.7354924220595862,
      "grad_norm": 0.7320488691329956,
      "learning_rate": 1.7253116486200511e-06,
      "loss": 0.6978,
      "step": 25526
    },
    {
      "epoch": 0.7355212355212355,
      "grad_norm": 0.7459088563919067,
      "learning_rate": 1.7249590535864841e-06,
      "loss": 0.6932,
      "step": 25527
    },
    {
      "epoch": 0.7355500489828848,
      "grad_norm": 0.7158744931221008,
      "learning_rate": 1.7246064870752478e-06,
      "loss": 0.6894,
      "step": 25528
    },
    {
      "epoch": 0.7355788624445341,
      "grad_norm": 0.6947843432426453,
      "learning_rate": 1.7242539490894167e-06,
      "loss": 0.6896,
      "step": 25529
    },
    {
      "epoch": 0.7356076759061834,
      "grad_norm": 0.7589659094810486,
      "learning_rate": 1.7239014396320574e-06,
      "loss": 0.7002,
      "step": 25530
    },
    {
      "epoch": 0.7356364893678327,
      "grad_norm": 0.7190447449684143,
      "learning_rate": 1.7235489587062421e-06,
      "loss": 0.6887,
      "step": 25531
    },
    {
      "epoch": 0.735665302829482,
      "grad_norm": 1.9473689794540405,
      "learning_rate": 1.7231965063150396e-06,
      "loss": 0.688,
      "step": 25532
    },
    {
      "epoch": 0.7356941162911312,
      "grad_norm": 0.7145932912826538,
      "learning_rate": 1.7228440824615216e-06,
      "loss": 0.6985,
      "step": 25533
    },
    {
      "epoch": 0.7357229297527805,
      "grad_norm": 0.7312169671058655,
      "learning_rate": 1.722491687148754e-06,
      "loss": 0.6939,
      "step": 25534
    },
    {
      "epoch": 0.7357517432144298,
      "grad_norm": 0.7380779981613159,
      "learning_rate": 1.72213932037981e-06,
      "loss": 0.686,
      "step": 25535
    },
    {
      "epoch": 0.7357805566760791,
      "grad_norm": 0.7091341018676758,
      "learning_rate": 1.7217869821577533e-06,
      "loss": 0.7125,
      "step": 25536
    },
    {
      "epoch": 0.7358093701377283,
      "grad_norm": 0.7292698621749878,
      "learning_rate": 1.7214346724856569e-06,
      "loss": 0.7102,
      "step": 25537
    },
    {
      "epoch": 0.7358381835993776,
      "grad_norm": 0.7142860293388367,
      "learning_rate": 1.7210823913665854e-06,
      "loss": 0.7044,
      "step": 25538
    },
    {
      "epoch": 0.7358669970610269,
      "grad_norm": 0.740200936794281,
      "learning_rate": 1.7207301388036102e-06,
      "loss": 0.694,
      "step": 25539
    },
    {
      "epoch": 0.7358958105226762,
      "grad_norm": 0.7324737906455994,
      "learning_rate": 1.7203779147997956e-06,
      "loss": 0.6902,
      "step": 25540
    },
    {
      "epoch": 0.7359246239843255,
      "grad_norm": 0.710427463054657,
      "learning_rate": 1.720025719358211e-06,
      "loss": 0.6746,
      "step": 25541
    },
    {
      "epoch": 0.7359534374459747,
      "grad_norm": 0.7194926738739014,
      "learning_rate": 1.7196735524819253e-06,
      "loss": 0.6907,
      "step": 25542
    },
    {
      "epoch": 0.735982250907624,
      "grad_norm": 0.70623779296875,
      "learning_rate": 1.7193214141740021e-06,
      "loss": 0.6628,
      "step": 25543
    },
    {
      "epoch": 0.7360110643692733,
      "grad_norm": 0.7218108177185059,
      "learning_rate": 1.7189693044375105e-06,
      "loss": 0.692,
      "step": 25544
    },
    {
      "epoch": 0.7360398778309226,
      "grad_norm": 0.7278954386711121,
      "learning_rate": 1.7186172232755176e-06,
      "loss": 0.6993,
      "step": 25545
    },
    {
      "epoch": 0.7360686912925719,
      "grad_norm": 0.7325723171234131,
      "learning_rate": 1.718265170691087e-06,
      "loss": 0.6911,
      "step": 25546
    },
    {
      "epoch": 0.7360975047542212,
      "grad_norm": 0.7179457545280457,
      "learning_rate": 1.717913146687286e-06,
      "loss": 0.6943,
      "step": 25547
    },
    {
      "epoch": 0.7361263182158705,
      "grad_norm": 0.7350748777389526,
      "learning_rate": 1.7175611512671824e-06,
      "loss": 0.6871,
      "step": 25548
    },
    {
      "epoch": 0.7361551316775198,
      "grad_norm": 0.7173207998275757,
      "learning_rate": 1.7172091844338385e-06,
      "loss": 0.6845,
      "step": 25549
    },
    {
      "epoch": 0.7361839451391691,
      "grad_norm": 0.7158471941947937,
      "learning_rate": 1.7168572461903221e-06,
      "loss": 0.6963,
      "step": 25550
    },
    {
      "epoch": 0.7362127586008183,
      "grad_norm": 0.7202040553092957,
      "learning_rate": 1.716505336539696e-06,
      "loss": 0.692,
      "step": 25551
    },
    {
      "epoch": 0.7362415720624675,
      "grad_norm": 0.7034358978271484,
      "learning_rate": 1.7161534554850278e-06,
      "loss": 0.6961,
      "step": 25552
    },
    {
      "epoch": 0.7362703855241168,
      "grad_norm": 0.7231038808822632,
      "learning_rate": 1.7158016030293784e-06,
      "loss": 0.7176,
      "step": 25553
    },
    {
      "epoch": 0.7362991989857661,
      "grad_norm": 0.7188616394996643,
      "learning_rate": 1.7154497791758156e-06,
      "loss": 0.6764,
      "step": 25554
    },
    {
      "epoch": 0.7363280124474154,
      "grad_norm": 0.7542707324028015,
      "learning_rate": 1.715097983927399e-06,
      "loss": 0.7034,
      "step": 25555
    },
    {
      "epoch": 0.7363568259090647,
      "grad_norm": 0.7242296934127808,
      "learning_rate": 1.714746217287198e-06,
      "loss": 0.7131,
      "step": 25556
    },
    {
      "epoch": 0.736385639370714,
      "grad_norm": 0.7326949834823608,
      "learning_rate": 1.7143944792582722e-06,
      "loss": 0.7045,
      "step": 25557
    },
    {
      "epoch": 0.7364144528323633,
      "grad_norm": 0.7190732955932617,
      "learning_rate": 1.7140427698436879e-06,
      "loss": 0.6887,
      "step": 25558
    },
    {
      "epoch": 0.7364432662940126,
      "grad_norm": 0.7316980957984924,
      "learning_rate": 1.7136910890465042e-06,
      "loss": 0.6844,
      "step": 25559
    },
    {
      "epoch": 0.7364720797556619,
      "grad_norm": 0.7355881929397583,
      "learning_rate": 1.7133394368697882e-06,
      "loss": 0.696,
      "step": 25560
    },
    {
      "epoch": 0.7365008932173112,
      "grad_norm": 0.7421307563781738,
      "learning_rate": 1.7129878133165984e-06,
      "loss": 0.71,
      "step": 25561
    },
    {
      "epoch": 0.7365297066789604,
      "grad_norm": 0.696988046169281,
      "learning_rate": 1.7126362183899985e-06,
      "loss": 0.6689,
      "step": 25562
    },
    {
      "epoch": 0.7365585201406097,
      "grad_norm": 0.7240616083145142,
      "learning_rate": 1.712284652093053e-06,
      "loss": 0.6548,
      "step": 25563
    },
    {
      "epoch": 0.7365873336022589,
      "grad_norm": 0.7314643263816833,
      "learning_rate": 1.7119331144288199e-06,
      "loss": 0.684,
      "step": 25564
    },
    {
      "epoch": 0.7366161470639082,
      "grad_norm": 0.7158297896385193,
      "learning_rate": 1.7115816054003642e-06,
      "loss": 0.7025,
      "step": 25565
    },
    {
      "epoch": 0.7366449605255575,
      "grad_norm": 0.7138926386833191,
      "learning_rate": 1.711230125010744e-06,
      "loss": 0.7021,
      "step": 25566
    },
    {
      "epoch": 0.7366737739872068,
      "grad_norm": 0.7103434205055237,
      "learning_rate": 1.7108786732630218e-06,
      "loss": 0.6818,
      "step": 25567
    },
    {
      "epoch": 0.7367025874488561,
      "grad_norm": 0.7206413149833679,
      "learning_rate": 1.7105272501602581e-06,
      "loss": 0.658,
      "step": 25568
    },
    {
      "epoch": 0.7367314009105054,
      "grad_norm": 0.7455556392669678,
      "learning_rate": 1.7101758557055153e-06,
      "loss": 0.679,
      "step": 25569
    },
    {
      "epoch": 0.7367602143721547,
      "grad_norm": 0.717643678188324,
      "learning_rate": 1.7098244899018511e-06,
      "loss": 0.6977,
      "step": 25570
    },
    {
      "epoch": 0.736789027833804,
      "grad_norm": 0.7190994024276733,
      "learning_rate": 1.7094731527523284e-06,
      "loss": 0.6994,
      "step": 25571
    },
    {
      "epoch": 0.7368178412954532,
      "grad_norm": 0.7252311110496521,
      "learning_rate": 1.7091218442600028e-06,
      "loss": 0.6838,
      "step": 25572
    },
    {
      "epoch": 0.7368466547571025,
      "grad_norm": 0.7256740927696228,
      "learning_rate": 1.7087705644279384e-06,
      "loss": 0.6787,
      "step": 25573
    },
    {
      "epoch": 0.7368754682187518,
      "grad_norm": 0.7183298468589783,
      "learning_rate": 1.708419313259191e-06,
      "loss": 0.6867,
      "step": 25574
    },
    {
      "epoch": 0.7369042816804011,
      "grad_norm": 0.7479230165481567,
      "learning_rate": 1.708068090756822e-06,
      "loss": 0.7028,
      "step": 25575
    },
    {
      "epoch": 0.7369330951420504,
      "grad_norm": 0.7169826626777649,
      "learning_rate": 1.7077168969238877e-06,
      "loss": 0.7064,
      "step": 25576
    },
    {
      "epoch": 0.7369619086036997,
      "grad_norm": 0.7004005908966064,
      "learning_rate": 1.7073657317634501e-06,
      "loss": 0.6761,
      "step": 25577
    },
    {
      "epoch": 0.7369907220653489,
      "grad_norm": 0.7304331064224243,
      "learning_rate": 1.7070145952785638e-06,
      "loss": 0.6862,
      "step": 25578
    },
    {
      "epoch": 0.7370195355269982,
      "grad_norm": 0.7266568541526794,
      "learning_rate": 1.7066634874722887e-06,
      "loss": 0.726,
      "step": 25579
    },
    {
      "epoch": 0.7370483489886475,
      "grad_norm": 0.7077539563179016,
      "learning_rate": 1.7063124083476822e-06,
      "loss": 0.6938,
      "step": 25580
    },
    {
      "epoch": 0.7370771624502968,
      "grad_norm": 0.7152023911476135,
      "learning_rate": 1.705961357907804e-06,
      "loss": 0.6813,
      "step": 25581
    },
    {
      "epoch": 0.737105975911946,
      "grad_norm": 0.7487033009529114,
      "learning_rate": 1.705610336155708e-06,
      "loss": 0.7174,
      "step": 25582
    },
    {
      "epoch": 0.7371347893735953,
      "grad_norm": 0.72227942943573,
      "learning_rate": 1.7052593430944541e-06,
      "loss": 0.708,
      "step": 25583
    },
    {
      "epoch": 0.7371636028352446,
      "grad_norm": 0.7029707431793213,
      "learning_rate": 1.7049083787270966e-06,
      "loss": 0.6994,
      "step": 25584
    },
    {
      "epoch": 0.7371924162968939,
      "grad_norm": 0.7222028374671936,
      "learning_rate": 1.7045574430566924e-06,
      "loss": 0.6776,
      "step": 25585
    },
    {
      "epoch": 0.7372212297585432,
      "grad_norm": 0.705816388130188,
      "learning_rate": 1.704206536086301e-06,
      "loss": 0.6957,
      "step": 25586
    },
    {
      "epoch": 0.7372500432201925,
      "grad_norm": 0.7422795295715332,
      "learning_rate": 1.703855657818974e-06,
      "loss": 0.7058,
      "step": 25587
    },
    {
      "epoch": 0.7372788566818418,
      "grad_norm": 0.7269429564476013,
      "learning_rate": 1.7035048082577716e-06,
      "loss": 0.7161,
      "step": 25588
    },
    {
      "epoch": 0.7373076701434911,
      "grad_norm": 0.7274188995361328,
      "learning_rate": 1.7031539874057446e-06,
      "loss": 0.6704,
      "step": 25589
    },
    {
      "epoch": 0.7373364836051404,
      "grad_norm": 0.7057502865791321,
      "learning_rate": 1.702803195265953e-06,
      "loss": 0.6993,
      "step": 25590
    },
    {
      "epoch": 0.7373652970667895,
      "grad_norm": 0.7211685180664062,
      "learning_rate": 1.7024524318414458e-06,
      "loss": 0.7099,
      "step": 25591
    },
    {
      "epoch": 0.7373941105284388,
      "grad_norm": 0.7028034329414368,
      "learning_rate": 1.7021016971352854e-06,
      "loss": 0.6625,
      "step": 25592
    },
    {
      "epoch": 0.7374229239900881,
      "grad_norm": 0.7070269584655762,
      "learning_rate": 1.7017509911505202e-06,
      "loss": 0.6547,
      "step": 25593
    },
    {
      "epoch": 0.7374517374517374,
      "grad_norm": 0.7093708515167236,
      "learning_rate": 1.701400313890209e-06,
      "loss": 0.6788,
      "step": 25594
    },
    {
      "epoch": 0.7374805509133867,
      "grad_norm": 0.6984912157058716,
      "learning_rate": 1.7010496653574021e-06,
      "loss": 0.6786,
      "step": 25595
    },
    {
      "epoch": 0.737509364375036,
      "grad_norm": 0.7186504602432251,
      "learning_rate": 1.7006990455551559e-06,
      "loss": 0.6824,
      "step": 25596
    },
    {
      "epoch": 0.7375381778366853,
      "grad_norm": 0.720382809638977,
      "learning_rate": 1.7003484544865218e-06,
      "loss": 0.6769,
      "step": 25597
    },
    {
      "epoch": 0.7375669912983346,
      "grad_norm": 0.7306437492370605,
      "learning_rate": 1.6999978921545558e-06,
      "loss": 0.6738,
      "step": 25598
    },
    {
      "epoch": 0.7375958047599839,
      "grad_norm": 0.7430118322372437,
      "learning_rate": 1.6996473585623075e-06,
      "loss": 0.7002,
      "step": 25599
    },
    {
      "epoch": 0.7376246182216332,
      "grad_norm": 0.7188990116119385,
      "learning_rate": 1.699296853712832e-06,
      "loss": 0.6808,
      "step": 25600
    },
    {
      "epoch": 0.7376534316832825,
      "grad_norm": 0.733496904373169,
      "learning_rate": 1.6989463776091824e-06,
      "loss": 0.7185,
      "step": 25601
    },
    {
      "epoch": 0.7376822451449317,
      "grad_norm": 0.7365244626998901,
      "learning_rate": 1.698595930254409e-06,
      "loss": 0.6797,
      "step": 25602
    },
    {
      "epoch": 0.737711058606581,
      "grad_norm": 0.7161151170730591,
      "learning_rate": 1.6982455116515646e-06,
      "loss": 0.6799,
      "step": 25603
    },
    {
      "epoch": 0.7377398720682303,
      "grad_norm": 0.7166601419448853,
      "learning_rate": 1.6978951218037026e-06,
      "loss": 0.6668,
      "step": 25604
    },
    {
      "epoch": 0.7377686855298795,
      "grad_norm": 0.7777171730995178,
      "learning_rate": 1.6975447607138717e-06,
      "loss": 0.6728,
      "step": 25605
    },
    {
      "epoch": 0.7377974989915288,
      "grad_norm": 0.7360907196998596,
      "learning_rate": 1.6971944283851249e-06,
      "loss": 0.6728,
      "step": 25606
    },
    {
      "epoch": 0.7378263124531781,
      "grad_norm": 0.741083025932312,
      "learning_rate": 1.6968441248205146e-06,
      "loss": 0.7145,
      "step": 25607
    },
    {
      "epoch": 0.7378551259148274,
      "grad_norm": 0.7038920521736145,
      "learning_rate": 1.6964938500230888e-06,
      "loss": 0.6785,
      "step": 25608
    },
    {
      "epoch": 0.7378839393764767,
      "grad_norm": 0.7241658568382263,
      "learning_rate": 1.6961436039959e-06,
      "loss": 0.688,
      "step": 25609
    },
    {
      "epoch": 0.737912752838126,
      "grad_norm": 0.7257488965988159,
      "learning_rate": 1.6957933867419967e-06,
      "loss": 0.6806,
      "step": 25610
    },
    {
      "epoch": 0.7379415662997753,
      "grad_norm": 0.724004328250885,
      "learning_rate": 1.6954431982644315e-06,
      "loss": 0.6772,
      "step": 25611
    },
    {
      "epoch": 0.7379703797614245,
      "grad_norm": 0.7068431973457336,
      "learning_rate": 1.695093038566251e-06,
      "loss": 0.7078,
      "step": 25612
    },
    {
      "epoch": 0.7379991932230738,
      "grad_norm": 0.7096366286277771,
      "learning_rate": 1.6947429076505083e-06,
      "loss": 0.6882,
      "step": 25613
    },
    {
      "epoch": 0.7380280066847231,
      "grad_norm": 0.7160419821739197,
      "learning_rate": 1.694392805520249e-06,
      "loss": 0.6879,
      "step": 25614
    },
    {
      "epoch": 0.7380568201463724,
      "grad_norm": 0.7279456853866577,
      "learning_rate": 1.694042732178524e-06,
      "loss": 0.675,
      "step": 25615
    },
    {
      "epoch": 0.7380856336080217,
      "grad_norm": 0.6911188960075378,
      "learning_rate": 1.6936926876283822e-06,
      "loss": 0.6917,
      "step": 25616
    },
    {
      "epoch": 0.738114447069671,
      "grad_norm": 0.7388147711753845,
      "learning_rate": 1.6933426718728734e-06,
      "loss": 0.7159,
      "step": 25617
    },
    {
      "epoch": 0.7381432605313202,
      "grad_norm": 0.7340948581695557,
      "learning_rate": 1.692992684915043e-06,
      "loss": 0.7085,
      "step": 25618
    },
    {
      "epoch": 0.7381720739929695,
      "grad_norm": 0.7402790784835815,
      "learning_rate": 1.6926427267579421e-06,
      "loss": 0.6767,
      "step": 25619
    },
    {
      "epoch": 0.7382008874546188,
      "grad_norm": 0.7508610486984253,
      "learning_rate": 1.6922927974046155e-06,
      "loss": 0.7062,
      "step": 25620
    },
    {
      "epoch": 0.738229700916268,
      "grad_norm": 0.7238842844963074,
      "learning_rate": 1.6919428968581136e-06,
      "loss": 0.6826,
      "step": 25621
    },
    {
      "epoch": 0.7382585143779173,
      "grad_norm": 0.7247014045715332,
      "learning_rate": 1.6915930251214802e-06,
      "loss": 0.6909,
      "step": 25622
    },
    {
      "epoch": 0.7382873278395666,
      "grad_norm": 0.6863507628440857,
      "learning_rate": 1.6912431821977648e-06,
      "loss": 0.6583,
      "step": 25623
    },
    {
      "epoch": 0.7383161413012159,
      "grad_norm": 0.7052937150001526,
      "learning_rate": 1.6908933680900153e-06,
      "loss": 0.7165,
      "step": 25624
    },
    {
      "epoch": 0.7383449547628652,
      "grad_norm": 0.6985569000244141,
      "learning_rate": 1.690543582801275e-06,
      "loss": 0.692,
      "step": 25625
    },
    {
      "epoch": 0.7383737682245145,
      "grad_norm": 0.7253779768943787,
      "learning_rate": 1.6901938263345935e-06,
      "loss": 0.7087,
      "step": 25626
    },
    {
      "epoch": 0.7384025816861638,
      "grad_norm": 0.7291203141212463,
      "learning_rate": 1.689844098693012e-06,
      "loss": 0.7113,
      "step": 25627
    },
    {
      "epoch": 0.7384313951478131,
      "grad_norm": 0.733996570110321,
      "learning_rate": 1.6894943998795827e-06,
      "loss": 0.6908,
      "step": 25628
    },
    {
      "epoch": 0.7384602086094624,
      "grad_norm": 0.7446103096008301,
      "learning_rate": 1.6891447298973456e-06,
      "loss": 0.7,
      "step": 25629
    },
    {
      "epoch": 0.7384890220711117,
      "grad_norm": 0.740975022315979,
      "learning_rate": 1.688795088749351e-06,
      "loss": 0.69,
      "step": 25630
    },
    {
      "epoch": 0.738517835532761,
      "grad_norm": 0.712406575679779,
      "learning_rate": 1.6884454764386388e-06,
      "loss": 0.6813,
      "step": 25631
    },
    {
      "epoch": 0.7385466489944101,
      "grad_norm": 0.7117326259613037,
      "learning_rate": 1.6880958929682579e-06,
      "loss": 0.6697,
      "step": 25632
    },
    {
      "epoch": 0.7385754624560594,
      "grad_norm": 0.7080335021018982,
      "learning_rate": 1.6877463383412496e-06,
      "loss": 0.6881,
      "step": 25633
    },
    {
      "epoch": 0.7386042759177087,
      "grad_norm": 0.732585608959198,
      "learning_rate": 1.687396812560661e-06,
      "loss": 0.6676,
      "step": 25634
    },
    {
      "epoch": 0.738633089379358,
      "grad_norm": 0.7454723715782166,
      "learning_rate": 1.6870473156295336e-06,
      "loss": 0.7047,
      "step": 25635
    },
    {
      "epoch": 0.7386619028410073,
      "grad_norm": 0.7330636978149414,
      "learning_rate": 1.6866978475509138e-06,
      "loss": 0.6921,
      "step": 25636
    },
    {
      "epoch": 0.7386907163026566,
      "grad_norm": 0.7436822652816772,
      "learning_rate": 1.6863484083278419e-06,
      "loss": 0.6947,
      "step": 25637
    },
    {
      "epoch": 0.7387195297643059,
      "grad_norm": 0.7338560223579407,
      "learning_rate": 1.6859989979633629e-06,
      "loss": 0.688,
      "step": 25638
    },
    {
      "epoch": 0.7387483432259552,
      "grad_norm": 0.733262836933136,
      "learning_rate": 1.6856496164605202e-06,
      "loss": 0.6844,
      "step": 25639
    },
    {
      "epoch": 0.7387771566876045,
      "grad_norm": 0.7195238471031189,
      "learning_rate": 1.6853002638223576e-06,
      "loss": 0.6804,
      "step": 25640
    },
    {
      "epoch": 0.7388059701492538,
      "grad_norm": 0.7198861241340637,
      "learning_rate": 1.6849509400519143e-06,
      "loss": 0.689,
      "step": 25641
    },
    {
      "epoch": 0.738834783610903,
      "grad_norm": 0.7278659343719482,
      "learning_rate": 1.6846016451522362e-06,
      "loss": 0.6872,
      "step": 25642
    },
    {
      "epoch": 0.7388635970725523,
      "grad_norm": 0.7309671640396118,
      "learning_rate": 1.6842523791263621e-06,
      "loss": 0.6558,
      "step": 25643
    },
    {
      "epoch": 0.7388924105342016,
      "grad_norm": 0.7350232005119324,
      "learning_rate": 1.6839031419773356e-06,
      "loss": 0.6905,
      "step": 25644
    },
    {
      "epoch": 0.7389212239958508,
      "grad_norm": 0.7235758900642395,
      "learning_rate": 1.683553933708199e-06,
      "loss": 0.6942,
      "step": 25645
    },
    {
      "epoch": 0.7389500374575001,
      "grad_norm": 0.7270523905754089,
      "learning_rate": 1.683204754321991e-06,
      "loss": 0.6748,
      "step": 25646
    },
    {
      "epoch": 0.7389788509191494,
      "grad_norm": 0.7321726679801941,
      "learning_rate": 1.6828556038217554e-06,
      "loss": 0.692,
      "step": 25647
    },
    {
      "epoch": 0.7390076643807987,
      "grad_norm": 0.6965051889419556,
      "learning_rate": 1.6825064822105296e-06,
      "loss": 0.6764,
      "step": 25648
    },
    {
      "epoch": 0.739036477842448,
      "grad_norm": 0.7114836573600769,
      "learning_rate": 1.6821573894913584e-06,
      "loss": 0.6803,
      "step": 25649
    },
    {
      "epoch": 0.7390652913040973,
      "grad_norm": 0.7225264310836792,
      "learning_rate": 1.681808325667278e-06,
      "loss": 0.6935,
      "step": 25650
    },
    {
      "epoch": 0.7390941047657466,
      "grad_norm": 0.7139540314674377,
      "learning_rate": 1.6814592907413296e-06,
      "loss": 0.6835,
      "step": 25651
    },
    {
      "epoch": 0.7391229182273958,
      "grad_norm": 0.7002997994422913,
      "learning_rate": 1.681110284716554e-06,
      "loss": 0.7014,
      "step": 25652
    },
    {
      "epoch": 0.7391517316890451,
      "grad_norm": 0.7164720296859741,
      "learning_rate": 1.6807613075959917e-06,
      "loss": 0.6753,
      "step": 25653
    },
    {
      "epoch": 0.7391805451506944,
      "grad_norm": 0.7041729092597961,
      "learning_rate": 1.6804123593826788e-06,
      "loss": 0.6875,
      "step": 25654
    },
    {
      "epoch": 0.7392093586123437,
      "grad_norm": 0.7119246125221252,
      "learning_rate": 1.6800634400796573e-06,
      "loss": 0.7096,
      "step": 25655
    },
    {
      "epoch": 0.739238172073993,
      "grad_norm": 0.727965235710144,
      "learning_rate": 1.6797145496899625e-06,
      "loss": 0.7046,
      "step": 25656
    },
    {
      "epoch": 0.7392669855356423,
      "grad_norm": 0.7168486714363098,
      "learning_rate": 1.679365688216637e-06,
      "loss": 0.699,
      "step": 25657
    },
    {
      "epoch": 0.7392957989972916,
      "grad_norm": 0.7324190735816956,
      "learning_rate": 1.6790168556627156e-06,
      "loss": 0.6986,
      "step": 25658
    },
    {
      "epoch": 0.7393246124589408,
      "grad_norm": 0.7413978576660156,
      "learning_rate": 1.6786680520312372e-06,
      "loss": 0.7015,
      "step": 25659
    },
    {
      "epoch": 0.7393534259205901,
      "grad_norm": 0.694629967212677,
      "learning_rate": 1.6783192773252416e-06,
      "loss": 0.6979,
      "step": 25660
    },
    {
      "epoch": 0.7393822393822393,
      "grad_norm": 0.7594257593154907,
      "learning_rate": 1.6779705315477629e-06,
      "loss": 0.6948,
      "step": 25661
    },
    {
      "epoch": 0.7394110528438886,
      "grad_norm": 0.7336023449897766,
      "learning_rate": 1.677621814701842e-06,
      "loss": 0.697,
      "step": 25662
    },
    {
      "epoch": 0.7394398663055379,
      "grad_norm": 0.7299832701683044,
      "learning_rate": 1.6772731267905119e-06,
      "loss": 0.6864,
      "step": 25663
    },
    {
      "epoch": 0.7394686797671872,
      "grad_norm": 0.7212768793106079,
      "learning_rate": 1.6769244678168118e-06,
      "loss": 0.7021,
      "step": 25664
    },
    {
      "epoch": 0.7394974932288365,
      "grad_norm": 0.7337215542793274,
      "learning_rate": 1.6765758377837777e-06,
      "loss": 0.7049,
      "step": 25665
    },
    {
      "epoch": 0.7395263066904858,
      "grad_norm": 0.7063992619514465,
      "learning_rate": 1.676227236694447e-06,
      "loss": 0.6752,
      "step": 25666
    },
    {
      "epoch": 0.7395551201521351,
      "grad_norm": 0.7300271987915039,
      "learning_rate": 1.6758786645518532e-06,
      "loss": 0.6826,
      "step": 25667
    },
    {
      "epoch": 0.7395839336137844,
      "grad_norm": 0.7225849628448486,
      "learning_rate": 1.6755301213590352e-06,
      "loss": 0.6874,
      "step": 25668
    },
    {
      "epoch": 0.7396127470754337,
      "grad_norm": 0.7152913808822632,
      "learning_rate": 1.6751816071190246e-06,
      "loss": 0.663,
      "step": 25669
    },
    {
      "epoch": 0.739641560537083,
      "grad_norm": 0.7246075868606567,
      "learning_rate": 1.67483312183486e-06,
      "loss": 0.664,
      "step": 25670
    },
    {
      "epoch": 0.7396703739987323,
      "grad_norm": 0.7445966601371765,
      "learning_rate": 1.6744846655095737e-06,
      "loss": 0.6894,
      "step": 25671
    },
    {
      "epoch": 0.7396991874603814,
      "grad_norm": 0.7305975556373596,
      "learning_rate": 1.6741362381462033e-06,
      "loss": 0.6997,
      "step": 25672
    },
    {
      "epoch": 0.7397280009220307,
      "grad_norm": 0.7079824209213257,
      "learning_rate": 1.6737878397477802e-06,
      "loss": 0.6807,
      "step": 25673
    },
    {
      "epoch": 0.73975681438368,
      "grad_norm": 0.7530113458633423,
      "learning_rate": 1.6734394703173413e-06,
      "loss": 0.7035,
      "step": 25674
    },
    {
      "epoch": 0.7397856278453293,
      "grad_norm": 0.7468416690826416,
      "learning_rate": 1.673091129857916e-06,
      "loss": 0.6918,
      "step": 25675
    },
    {
      "epoch": 0.7398144413069786,
      "grad_norm": 0.7232419848442078,
      "learning_rate": 1.6727428183725448e-06,
      "loss": 0.6938,
      "step": 25676
    },
    {
      "epoch": 0.7398432547686279,
      "grad_norm": 0.712827205657959,
      "learning_rate": 1.6723945358642557e-06,
      "loss": 0.7069,
      "step": 25677
    },
    {
      "epoch": 0.7398720682302772,
      "grad_norm": 0.7038086652755737,
      "learning_rate": 1.6720462823360855e-06,
      "loss": 0.6807,
      "step": 25678
    },
    {
      "epoch": 0.7399008816919265,
      "grad_norm": 0.726207435131073,
      "learning_rate": 1.671698057791064e-06,
      "loss": 0.6727,
      "step": 25679
    },
    {
      "epoch": 0.7399296951535758,
      "grad_norm": 0.7378556728363037,
      "learning_rate": 1.6713498622322266e-06,
      "loss": 0.6891,
      "step": 25680
    },
    {
      "epoch": 0.739958508615225,
      "grad_norm": 0.7438854575157166,
      "learning_rate": 1.671001695662603e-06,
      "loss": 0.6819,
      "step": 25681
    },
    {
      "epoch": 0.7399873220768743,
      "grad_norm": 0.7049903869628906,
      "learning_rate": 1.6706535580852268e-06,
      "loss": 0.6686,
      "step": 25682
    },
    {
      "epoch": 0.7400161355385236,
      "grad_norm": 0.7185885310173035,
      "learning_rate": 1.6703054495031317e-06,
      "loss": 0.6721,
      "step": 25683
    },
    {
      "epoch": 0.7400449490001729,
      "grad_norm": 0.7200520634651184,
      "learning_rate": 1.669957369919346e-06,
      "loss": 0.6576,
      "step": 25684
    },
    {
      "epoch": 0.7400737624618222,
      "grad_norm": 0.7507692575454712,
      "learning_rate": 1.6696093193369045e-06,
      "loss": 0.7067,
      "step": 25685
    },
    {
      "epoch": 0.7401025759234714,
      "grad_norm": 0.7230852246284485,
      "learning_rate": 1.6692612977588352e-06,
      "loss": 0.6795,
      "step": 25686
    },
    {
      "epoch": 0.7401313893851207,
      "grad_norm": 0.7278428077697754,
      "learning_rate": 1.6689133051881702e-06,
      "loss": 0.6758,
      "step": 25687
    },
    {
      "epoch": 0.74016020284677,
      "grad_norm": 0.7183472514152527,
      "learning_rate": 1.6685653416279406e-06,
      "loss": 0.7022,
      "step": 25688
    },
    {
      "epoch": 0.7401890163084193,
      "grad_norm": 0.7291963696479797,
      "learning_rate": 1.668217407081178e-06,
      "loss": 0.6683,
      "step": 25689
    },
    {
      "epoch": 0.7402178297700686,
      "grad_norm": 0.7210656404495239,
      "learning_rate": 1.66786950155091e-06,
      "loss": 0.7272,
      "step": 25690
    },
    {
      "epoch": 0.7402466432317178,
      "grad_norm": 0.7168692946434021,
      "learning_rate": 1.6675216250401694e-06,
      "loss": 0.6834,
      "step": 25691
    },
    {
      "epoch": 0.7402754566933671,
      "grad_norm": 0.738922119140625,
      "learning_rate": 1.6671737775519819e-06,
      "loss": 0.6707,
      "step": 25692
    },
    {
      "epoch": 0.7403042701550164,
      "grad_norm": 0.7031927704811096,
      "learning_rate": 1.6668259590893816e-06,
      "loss": 0.6942,
      "step": 25693
    },
    {
      "epoch": 0.7403330836166657,
      "grad_norm": 0.7160500288009644,
      "learning_rate": 1.6664781696553928e-06,
      "loss": 0.698,
      "step": 25694
    },
    {
      "epoch": 0.740361897078315,
      "grad_norm": 0.7117692232131958,
      "learning_rate": 1.6661304092530489e-06,
      "loss": 0.6846,
      "step": 25695
    },
    {
      "epoch": 0.7403907105399643,
      "grad_norm": 0.7359802722930908,
      "learning_rate": 1.6657826778853753e-06,
      "loss": 0.6771,
      "step": 25696
    },
    {
      "epoch": 0.7404195240016136,
      "grad_norm": 0.7817747592926025,
      "learning_rate": 1.665434975555401e-06,
      "loss": 0.7071,
      "step": 25697
    },
    {
      "epoch": 0.7404483374632629,
      "grad_norm": 0.7421526312828064,
      "learning_rate": 1.6650873022661562e-06,
      "loss": 0.7015,
      "step": 25698
    },
    {
      "epoch": 0.7404771509249121,
      "grad_norm": 0.7096048593521118,
      "learning_rate": 1.664739658020666e-06,
      "loss": 0.7088,
      "step": 25699
    },
    {
      "epoch": 0.7405059643865614,
      "grad_norm": 0.7023044228553772,
      "learning_rate": 1.6643920428219595e-06,
      "loss": 0.6806,
      "step": 25700
    },
    {
      "epoch": 0.7405347778482106,
      "grad_norm": 0.7264657616615295,
      "learning_rate": 1.664044456673065e-06,
      "loss": 0.6828,
      "step": 25701
    },
    {
      "epoch": 0.7405635913098599,
      "grad_norm": 0.7220238447189331,
      "learning_rate": 1.6636968995770074e-06,
      "loss": 0.696,
      "step": 25702
    },
    {
      "epoch": 0.7405924047715092,
      "grad_norm": 0.688165009021759,
      "learning_rate": 1.6633493715368148e-06,
      "loss": 0.6402,
      "step": 25703
    },
    {
      "epoch": 0.7406212182331585,
      "grad_norm": 0.7044143080711365,
      "learning_rate": 1.6630018725555154e-06,
      "loss": 0.713,
      "step": 25704
    },
    {
      "epoch": 0.7406500316948078,
      "grad_norm": 0.7311943769454956,
      "learning_rate": 1.6626544026361319e-06,
      "loss": 0.6996,
      "step": 25705
    },
    {
      "epoch": 0.7406788451564571,
      "grad_norm": 0.744961142539978,
      "learning_rate": 1.662306961781694e-06,
      "loss": 0.6919,
      "step": 25706
    },
    {
      "epoch": 0.7407076586181064,
      "grad_norm": 0.7487145662307739,
      "learning_rate": 1.6619595499952246e-06,
      "loss": 0.7061,
      "step": 25707
    },
    {
      "epoch": 0.7407364720797557,
      "grad_norm": 0.7375028729438782,
      "learning_rate": 1.6616121672797524e-06,
      "loss": 0.6593,
      "step": 25708
    },
    {
      "epoch": 0.740765285541405,
      "grad_norm": 0.716424286365509,
      "learning_rate": 1.661264813638299e-06,
      "loss": 0.6943,
      "step": 25709
    },
    {
      "epoch": 0.7407940990030543,
      "grad_norm": 0.7080779075622559,
      "learning_rate": 1.6609174890738937e-06,
      "loss": 0.683,
      "step": 25710
    },
    {
      "epoch": 0.7408229124647036,
      "grad_norm": 0.7178975939750671,
      "learning_rate": 1.6605701935895557e-06,
      "loss": 0.7002,
      "step": 25711
    },
    {
      "epoch": 0.7408517259263528,
      "grad_norm": 0.7200661301612854,
      "learning_rate": 1.6602229271883164e-06,
      "loss": 0.6969,
      "step": 25712
    },
    {
      "epoch": 0.740880539388002,
      "grad_norm": 0.7154927849769592,
      "learning_rate": 1.6598756898731955e-06,
      "loss": 0.6996,
      "step": 25713
    },
    {
      "epoch": 0.7409093528496513,
      "grad_norm": 0.7331766486167908,
      "learning_rate": 1.6595284816472196e-06,
      "loss": 0.7002,
      "step": 25714
    },
    {
      "epoch": 0.7409381663113006,
      "grad_norm": 0.7154497504234314,
      "learning_rate": 1.65918130251341e-06,
      "loss": 0.6679,
      "step": 25715
    },
    {
      "epoch": 0.7409669797729499,
      "grad_norm": 0.7347590923309326,
      "learning_rate": 1.6588341524747936e-06,
      "loss": 0.6883,
      "step": 25716
    },
    {
      "epoch": 0.7409957932345992,
      "grad_norm": 0.6983991861343384,
      "learning_rate": 1.6584870315343899e-06,
      "loss": 0.6843,
      "step": 25717
    },
    {
      "epoch": 0.7410246066962485,
      "grad_norm": 0.7099025249481201,
      "learning_rate": 1.6581399396952251e-06,
      "loss": 0.6882,
      "step": 25718
    },
    {
      "epoch": 0.7410534201578978,
      "grad_norm": 0.7095404267311096,
      "learning_rate": 1.6577928769603196e-06,
      "loss": 0.6925,
      "step": 25719
    },
    {
      "epoch": 0.7410822336195471,
      "grad_norm": 0.7040132284164429,
      "learning_rate": 1.6574458433326967e-06,
      "loss": 0.6982,
      "step": 25720
    },
    {
      "epoch": 0.7411110470811964,
      "grad_norm": 0.7111178040504456,
      "learning_rate": 1.6570988388153814e-06,
      "loss": 0.6835,
      "step": 25721
    },
    {
      "epoch": 0.7411398605428456,
      "grad_norm": 0.7492201328277588,
      "learning_rate": 1.6567518634113916e-06,
      "loss": 0.7095,
      "step": 25722
    },
    {
      "epoch": 0.7411686740044949,
      "grad_norm": 0.7176716327667236,
      "learning_rate": 1.6564049171237512e-06,
      "loss": 0.6741,
      "step": 25723
    },
    {
      "epoch": 0.7411974874661442,
      "grad_norm": 0.6977111101150513,
      "learning_rate": 1.6560579999554814e-06,
      "loss": 0.6749,
      "step": 25724
    },
    {
      "epoch": 0.7412263009277935,
      "grad_norm": 0.7117222547531128,
      "learning_rate": 1.6557111119096052e-06,
      "loss": 0.6998,
      "step": 25725
    },
    {
      "epoch": 0.7412551143894427,
      "grad_norm": 0.7393256425857544,
      "learning_rate": 1.6553642529891407e-06,
      "loss": 0.697,
      "step": 25726
    },
    {
      "epoch": 0.741283927851092,
      "grad_norm": 0.7282257676124573,
      "learning_rate": 1.6550174231971122e-06,
      "loss": 0.7004,
      "step": 25727
    },
    {
      "epoch": 0.7413127413127413,
      "grad_norm": 0.7302136421203613,
      "learning_rate": 1.654670622536536e-06,
      "loss": 0.6812,
      "step": 25728
    },
    {
      "epoch": 0.7413415547743906,
      "grad_norm": 0.7230346202850342,
      "learning_rate": 1.6543238510104365e-06,
      "loss": 0.6678,
      "step": 25729
    },
    {
      "epoch": 0.7413703682360399,
      "grad_norm": 0.7079528570175171,
      "learning_rate": 1.65397710862183e-06,
      "loss": 0.6775,
      "step": 25730
    },
    {
      "epoch": 0.7413991816976891,
      "grad_norm": 0.7172343134880066,
      "learning_rate": 1.65363039537374e-06,
      "loss": 0.6975,
      "step": 25731
    },
    {
      "epoch": 0.7414279951593384,
      "grad_norm": 0.7074718475341797,
      "learning_rate": 1.6532837112691818e-06,
      "loss": 0.6699,
      "step": 25732
    },
    {
      "epoch": 0.7414568086209877,
      "grad_norm": 0.7251349091529846,
      "learning_rate": 1.652937056311179e-06,
      "loss": 0.6824,
      "step": 25733
    },
    {
      "epoch": 0.741485622082637,
      "grad_norm": 0.7322149276733398,
      "learning_rate": 1.6525904305027467e-06,
      "loss": 0.677,
      "step": 25734
    },
    {
      "epoch": 0.7415144355442863,
      "grad_norm": 0.7233763933181763,
      "learning_rate": 1.652243833846906e-06,
      "loss": 0.7009,
      "step": 25735
    },
    {
      "epoch": 0.7415432490059356,
      "grad_norm": 0.7248538732528687,
      "learning_rate": 1.6518972663466754e-06,
      "loss": 0.7181,
      "step": 25736
    },
    {
      "epoch": 0.7415720624675849,
      "grad_norm": 0.7447075843811035,
      "learning_rate": 1.6515507280050735e-06,
      "loss": 0.6786,
      "step": 25737
    },
    {
      "epoch": 0.7416008759292342,
      "grad_norm": 0.7136714458465576,
      "learning_rate": 1.6512042188251164e-06,
      "loss": 0.6861,
      "step": 25738
    },
    {
      "epoch": 0.7416296893908835,
      "grad_norm": 0.6992204189300537,
      "learning_rate": 1.6508577388098246e-06,
      "loss": 0.6765,
      "step": 25739
    },
    {
      "epoch": 0.7416585028525327,
      "grad_norm": 0.7147189378738403,
      "learning_rate": 1.6505112879622122e-06,
      "loss": 0.6849,
      "step": 25740
    },
    {
      "epoch": 0.7416873163141819,
      "grad_norm": 0.6989285349845886,
      "learning_rate": 1.6501648662852982e-06,
      "loss": 0.6778,
      "step": 25741
    },
    {
      "epoch": 0.7417161297758312,
      "grad_norm": 0.7039116621017456,
      "learning_rate": 1.649818473782101e-06,
      "loss": 0.6622,
      "step": 25742
    },
    {
      "epoch": 0.7417449432374805,
      "grad_norm": 0.7449871897697449,
      "learning_rate": 1.6494721104556344e-06,
      "loss": 0.6871,
      "step": 25743
    },
    {
      "epoch": 0.7417737566991298,
      "grad_norm": 0.7124353051185608,
      "learning_rate": 1.6491257763089186e-06,
      "loss": 0.7119,
      "step": 25744
    },
    {
      "epoch": 0.7418025701607791,
      "grad_norm": 0.7267957925796509,
      "learning_rate": 1.6487794713449651e-06,
      "loss": 0.6887,
      "step": 25745
    },
    {
      "epoch": 0.7418313836224284,
      "grad_norm": 0.7192544937133789,
      "learning_rate": 1.6484331955667948e-06,
      "loss": 0.7063,
      "step": 25746
    },
    {
      "epoch": 0.7418601970840777,
      "grad_norm": 0.696587860584259,
      "learning_rate": 1.6480869489774176e-06,
      "loss": 0.6713,
      "step": 25747
    },
    {
      "epoch": 0.741889010545727,
      "grad_norm": 0.7151362895965576,
      "learning_rate": 1.6477407315798554e-06,
      "loss": 0.6919,
      "step": 25748
    },
    {
      "epoch": 0.7419178240073763,
      "grad_norm": 0.7225248217582703,
      "learning_rate": 1.6473945433771189e-06,
      "loss": 0.7099,
      "step": 25749
    },
    {
      "epoch": 0.7419466374690256,
      "grad_norm": 0.7094969153404236,
      "learning_rate": 1.6470483843722263e-06,
      "loss": 0.6936,
      "step": 25750
    },
    {
      "epoch": 0.7419754509306749,
      "grad_norm": 0.7487609386444092,
      "learning_rate": 1.6467022545681888e-06,
      "loss": 0.7022,
      "step": 25751
    },
    {
      "epoch": 0.7420042643923241,
      "grad_norm": 0.7134296298027039,
      "learning_rate": 1.6463561539680239e-06,
      "loss": 0.6714,
      "step": 25752
    },
    {
      "epoch": 0.7420330778539733,
      "grad_norm": 0.7255064845085144,
      "learning_rate": 1.6460100825747433e-06,
      "loss": 0.672,
      "step": 25753
    },
    {
      "epoch": 0.7420618913156226,
      "grad_norm": 0.7020930647850037,
      "learning_rate": 1.6456640403913637e-06,
      "loss": 0.6528,
      "step": 25754
    },
    {
      "epoch": 0.7420907047772719,
      "grad_norm": 0.7208055853843689,
      "learning_rate": 1.6453180274208958e-06,
      "loss": 0.6875,
      "step": 25755
    },
    {
      "epoch": 0.7421195182389212,
      "grad_norm": 0.7399609684944153,
      "learning_rate": 1.6449720436663558e-06,
      "loss": 0.6579,
      "step": 25756
    },
    {
      "epoch": 0.7421483317005705,
      "grad_norm": 0.7179895639419556,
      "learning_rate": 1.6446260891307537e-06,
      "loss": 0.7055,
      "step": 25757
    },
    {
      "epoch": 0.7421771451622198,
      "grad_norm": 0.7183464169502258,
      "learning_rate": 1.6442801638171041e-06,
      "loss": 0.6906,
      "step": 25758
    },
    {
      "epoch": 0.7422059586238691,
      "grad_norm": 0.6795687079429626,
      "learning_rate": 1.6439342677284203e-06,
      "loss": 0.6399,
      "step": 25759
    },
    {
      "epoch": 0.7422347720855184,
      "grad_norm": 0.6952900886535645,
      "learning_rate": 1.6435884008677155e-06,
      "loss": 0.6876,
      "step": 25760
    },
    {
      "epoch": 0.7422635855471676,
      "grad_norm": 0.7215628027915955,
      "learning_rate": 1.6432425632379988e-06,
      "loss": 0.7033,
      "step": 25761
    },
    {
      "epoch": 0.7422923990088169,
      "grad_norm": 0.7207431793212891,
      "learning_rate": 1.6428967548422841e-06,
      "loss": 0.6793,
      "step": 25762
    },
    {
      "epoch": 0.7423212124704662,
      "grad_norm": 0.7131759524345398,
      "learning_rate": 1.6425509756835845e-06,
      "loss": 0.7022,
      "step": 25763
    },
    {
      "epoch": 0.7423500259321155,
      "grad_norm": 0.7406553030014038,
      "learning_rate": 1.642205225764908e-06,
      "loss": 0.6873,
      "step": 25764
    },
    {
      "epoch": 0.7423788393937648,
      "grad_norm": 0.7327691912651062,
      "learning_rate": 1.6418595050892689e-06,
      "loss": 0.6669,
      "step": 25765
    },
    {
      "epoch": 0.7424076528554141,
      "grad_norm": 0.704957902431488,
      "learning_rate": 1.6415138136596749e-06,
      "loss": 0.7192,
      "step": 25766
    },
    {
      "epoch": 0.7424364663170633,
      "grad_norm": 0.6969332098960876,
      "learning_rate": 1.64116815147914e-06,
      "loss": 0.7099,
      "step": 25767
    },
    {
      "epoch": 0.7424652797787126,
      "grad_norm": 0.7234644293785095,
      "learning_rate": 1.6408225185506716e-06,
      "loss": 0.7061,
      "step": 25768
    },
    {
      "epoch": 0.7424940932403619,
      "grad_norm": 0.7205415368080139,
      "learning_rate": 1.640476914877282e-06,
      "loss": 0.7186,
      "step": 25769
    },
    {
      "epoch": 0.7425229067020112,
      "grad_norm": 0.7142401933670044,
      "learning_rate": 1.6401313404619779e-06,
      "loss": 0.6875,
      "step": 25770
    },
    {
      "epoch": 0.7425517201636604,
      "grad_norm": 0.691020131111145,
      "learning_rate": 1.6397857953077739e-06,
      "loss": 0.665,
      "step": 25771
    },
    {
      "epoch": 0.7425805336253097,
      "grad_norm": 0.7174156308174133,
      "learning_rate": 1.639440279417675e-06,
      "loss": 0.6792,
      "step": 25772
    },
    {
      "epoch": 0.742609347086959,
      "grad_norm": 0.6999650597572327,
      "learning_rate": 1.6390947927946932e-06,
      "loss": 0.6508,
      "step": 25773
    },
    {
      "epoch": 0.7426381605486083,
      "grad_norm": 0.7020248770713806,
      "learning_rate": 1.6387493354418348e-06,
      "loss": 0.6892,
      "step": 25774
    },
    {
      "epoch": 0.7426669740102576,
      "grad_norm": 0.74052494764328,
      "learning_rate": 1.6384039073621116e-06,
      "loss": 0.6982,
      "step": 25775
    },
    {
      "epoch": 0.7426957874719069,
      "grad_norm": 0.734855592250824,
      "learning_rate": 1.6380585085585283e-06,
      "loss": 0.6992,
      "step": 25776
    },
    {
      "epoch": 0.7427246009335562,
      "grad_norm": 0.7219486832618713,
      "learning_rate": 1.6377131390340962e-06,
      "loss": 0.6788,
      "step": 25777
    },
    {
      "epoch": 0.7427534143952055,
      "grad_norm": 0.7108016610145569,
      "learning_rate": 1.6373677987918196e-06,
      "loss": 0.6887,
      "step": 25778
    },
    {
      "epoch": 0.7427822278568548,
      "grad_norm": 0.7448797225952148,
      "learning_rate": 1.6370224878347085e-06,
      "loss": 0.7307,
      "step": 25779
    },
    {
      "epoch": 0.742811041318504,
      "grad_norm": 0.7225753664970398,
      "learning_rate": 1.6366772061657716e-06,
      "loss": 0.7031,
      "step": 25780
    },
    {
      "epoch": 0.7428398547801532,
      "grad_norm": 0.7242028117179871,
      "learning_rate": 1.6363319537880118e-06,
      "loss": 0.7118,
      "step": 25781
    },
    {
      "epoch": 0.7428686682418025,
      "grad_norm": 0.7391883134841919,
      "learning_rate": 1.6359867307044386e-06,
      "loss": 0.6951,
      "step": 25782
    },
    {
      "epoch": 0.7428974817034518,
      "grad_norm": 0.7166962623596191,
      "learning_rate": 1.6356415369180583e-06,
      "loss": 0.6789,
      "step": 25783
    },
    {
      "epoch": 0.7429262951651011,
      "grad_norm": 0.72445148229599,
      "learning_rate": 1.6352963724318788e-06,
      "loss": 0.6801,
      "step": 25784
    },
    {
      "epoch": 0.7429551086267504,
      "grad_norm": 0.7025960087776184,
      "learning_rate": 1.634951237248903e-06,
      "loss": 0.6893,
      "step": 25785
    },
    {
      "epoch": 0.7429839220883997,
      "grad_norm": 0.6951087713241577,
      "learning_rate": 1.634606131372139e-06,
      "loss": 0.6748,
      "step": 25786
    },
    {
      "epoch": 0.743012735550049,
      "grad_norm": 0.7295413613319397,
      "learning_rate": 1.6342610548045902e-06,
      "loss": 0.6794,
      "step": 25787
    },
    {
      "epoch": 0.7430415490116983,
      "grad_norm": 0.7273287177085876,
      "learning_rate": 1.633916007549265e-06,
      "loss": 0.6795,
      "step": 25788
    },
    {
      "epoch": 0.7430703624733476,
      "grad_norm": 0.7173488736152649,
      "learning_rate": 1.6335709896091645e-06,
      "loss": 0.6856,
      "step": 25789
    },
    {
      "epoch": 0.7430991759349969,
      "grad_norm": 0.7450670003890991,
      "learning_rate": 1.633226000987297e-06,
      "loss": 0.706,
      "step": 25790
    },
    {
      "epoch": 0.7431279893966461,
      "grad_norm": 0.7164806723594666,
      "learning_rate": 1.6328810416866642e-06,
      "loss": 0.6928,
      "step": 25791
    },
    {
      "epoch": 0.7431568028582954,
      "grad_norm": 0.709273636341095,
      "learning_rate": 1.6325361117102734e-06,
      "loss": 0.6889,
      "step": 25792
    },
    {
      "epoch": 0.7431856163199447,
      "grad_norm": 0.7157324552536011,
      "learning_rate": 1.6321912110611249e-06,
      "loss": 0.6925,
      "step": 25793
    },
    {
      "epoch": 0.7432144297815939,
      "grad_norm": 0.7288388609886169,
      "learning_rate": 1.6318463397422241e-06,
      "loss": 0.7057,
      "step": 25794
    },
    {
      "epoch": 0.7432432432432432,
      "grad_norm": 0.7045283317565918,
      "learning_rate": 1.6315014977565751e-06,
      "loss": 0.6874,
      "step": 25795
    },
    {
      "epoch": 0.7432720567048925,
      "grad_norm": 0.6938254237174988,
      "learning_rate": 1.6311566851071825e-06,
      "loss": 0.6949,
      "step": 25796
    },
    {
      "epoch": 0.7433008701665418,
      "grad_norm": 0.6956788897514343,
      "learning_rate": 1.630811901797046e-06,
      "loss": 0.6742,
      "step": 25797
    },
    {
      "epoch": 0.7433296836281911,
      "grad_norm": 0.7149144411087036,
      "learning_rate": 1.6304671478291712e-06,
      "loss": 0.6852,
      "step": 25798
    },
    {
      "epoch": 0.7433584970898404,
      "grad_norm": 0.7160478830337524,
      "learning_rate": 1.6301224232065577e-06,
      "loss": 0.7012,
      "step": 25799
    },
    {
      "epoch": 0.7433873105514897,
      "grad_norm": 0.7267448306083679,
      "learning_rate": 1.6297777279322098e-06,
      "loss": 0.6982,
      "step": 25800
    },
    {
      "epoch": 0.743416124013139,
      "grad_norm": 0.718940258026123,
      "learning_rate": 1.6294330620091298e-06,
      "loss": 0.6751,
      "step": 25801
    },
    {
      "epoch": 0.7434449374747882,
      "grad_norm": 0.6958175897598267,
      "learning_rate": 1.6290884254403172e-06,
      "loss": 0.6802,
      "step": 25802
    },
    {
      "epoch": 0.7434737509364375,
      "grad_norm": 0.7272251844406128,
      "learning_rate": 1.6287438182287763e-06,
      "loss": 0.6857,
      "step": 25803
    },
    {
      "epoch": 0.7435025643980868,
      "grad_norm": 0.7185771465301514,
      "learning_rate": 1.6283992403775052e-06,
      "loss": 0.707,
      "step": 25804
    },
    {
      "epoch": 0.7435313778597361,
      "grad_norm": 0.7142826914787292,
      "learning_rate": 1.6280546918895078e-06,
      "loss": 0.6858,
      "step": 25805
    },
    {
      "epoch": 0.7435601913213854,
      "grad_norm": 0.7000510096549988,
      "learning_rate": 1.6277101727677803e-06,
      "loss": 0.6722,
      "step": 25806
    },
    {
      "epoch": 0.7435890047830346,
      "grad_norm": 0.8486197590827942,
      "learning_rate": 1.627365683015329e-06,
      "loss": 0.7022,
      "step": 25807
    },
    {
      "epoch": 0.7436178182446839,
      "grad_norm": 0.7015095353126526,
      "learning_rate": 1.6270212226351495e-06,
      "loss": 0.6889,
      "step": 25808
    },
    {
      "epoch": 0.7436466317063332,
      "grad_norm": 0.725236713886261,
      "learning_rate": 1.6266767916302457e-06,
      "loss": 0.7094,
      "step": 25809
    },
    {
      "epoch": 0.7436754451679825,
      "grad_norm": 0.6868351697921753,
      "learning_rate": 1.6263323900036126e-06,
      "loss": 0.6582,
      "step": 25810
    },
    {
      "epoch": 0.7437042586296317,
      "grad_norm": 0.7709231376647949,
      "learning_rate": 1.6259880177582544e-06,
      "loss": 0.6861,
      "step": 25811
    },
    {
      "epoch": 0.743733072091281,
      "grad_norm": 0.7304714322090149,
      "learning_rate": 1.6256436748971654e-06,
      "loss": 0.689,
      "step": 25812
    },
    {
      "epoch": 0.7437618855529303,
      "grad_norm": 0.7070335149765015,
      "learning_rate": 1.6252993614233486e-06,
      "loss": 0.7047,
      "step": 25813
    },
    {
      "epoch": 0.7437906990145796,
      "grad_norm": 0.7230888605117798,
      "learning_rate": 1.6249550773397999e-06,
      "loss": 0.7094,
      "step": 25814
    },
    {
      "epoch": 0.7438195124762289,
      "grad_norm": 0.7099641561508179,
      "learning_rate": 1.6246108226495194e-06,
      "loss": 0.6851,
      "step": 25815
    },
    {
      "epoch": 0.7438483259378782,
      "grad_norm": 0.7114304304122925,
      "learning_rate": 1.6242665973555032e-06,
      "loss": 0.6734,
      "step": 25816
    },
    {
      "epoch": 0.7438771393995275,
      "grad_norm": 0.7111594080924988,
      "learning_rate": 1.6239224014607509e-06,
      "loss": 0.6863,
      "step": 25817
    },
    {
      "epoch": 0.7439059528611768,
      "grad_norm": 0.7083269953727722,
      "learning_rate": 1.6235782349682595e-06,
      "loss": 0.6808,
      "step": 25818
    },
    {
      "epoch": 0.7439347663228261,
      "grad_norm": 0.7180680632591248,
      "learning_rate": 1.6232340978810279e-06,
      "loss": 0.6886,
      "step": 25819
    },
    {
      "epoch": 0.7439635797844754,
      "grad_norm": 0.725818932056427,
      "learning_rate": 1.62288999020205e-06,
      "loss": 0.6906,
      "step": 25820
    },
    {
      "epoch": 0.7439923932461245,
      "grad_norm": 0.7015944719314575,
      "learning_rate": 1.6225459119343245e-06,
      "loss": 0.6778,
      "step": 25821
    },
    {
      "epoch": 0.7440212067077738,
      "grad_norm": 0.7246769666671753,
      "learning_rate": 1.6222018630808495e-06,
      "loss": 0.6868,
      "step": 25822
    },
    {
      "epoch": 0.7440500201694231,
      "grad_norm": 0.7495290637016296,
      "learning_rate": 1.621857843644618e-06,
      "loss": 0.6896,
      "step": 25823
    },
    {
      "epoch": 0.7440788336310724,
      "grad_norm": 0.7210273742675781,
      "learning_rate": 1.6215138536286295e-06,
      "loss": 0.71,
      "step": 25824
    },
    {
      "epoch": 0.7441076470927217,
      "grad_norm": 0.714897632598877,
      "learning_rate": 1.6211698930358765e-06,
      "loss": 0.6741,
      "step": 25825
    },
    {
      "epoch": 0.744136460554371,
      "grad_norm": 0.7186917662620544,
      "learning_rate": 1.6208259618693585e-06,
      "loss": 0.7131,
      "step": 25826
    },
    {
      "epoch": 0.7441652740160203,
      "grad_norm": 0.7216407060623169,
      "learning_rate": 1.620482060132066e-06,
      "loss": 0.679,
      "step": 25827
    },
    {
      "epoch": 0.7441940874776696,
      "grad_norm": 0.7152360677719116,
      "learning_rate": 1.6201381878269977e-06,
      "loss": 0.6898,
      "step": 25828
    },
    {
      "epoch": 0.7442229009393189,
      "grad_norm": 0.7539121508598328,
      "learning_rate": 1.6197943449571463e-06,
      "loss": 0.6876,
      "step": 25829
    },
    {
      "epoch": 0.7442517144009682,
      "grad_norm": 0.6969320178031921,
      "learning_rate": 1.6194505315255066e-06,
      "loss": 0.6795,
      "step": 25830
    },
    {
      "epoch": 0.7442805278626174,
      "grad_norm": 0.7368804216384888,
      "learning_rate": 1.6191067475350742e-06,
      "loss": 0.691,
      "step": 25831
    },
    {
      "epoch": 0.7443093413242667,
      "grad_norm": 0.7012580633163452,
      "learning_rate": 1.6187629929888433e-06,
      "loss": 0.6909,
      "step": 25832
    },
    {
      "epoch": 0.744338154785916,
      "grad_norm": 0.7099876999855042,
      "learning_rate": 1.6184192678898053e-06,
      "loss": 0.6839,
      "step": 25833
    },
    {
      "epoch": 0.7443669682475652,
      "grad_norm": 0.7093387842178345,
      "learning_rate": 1.6180755722409569e-06,
      "loss": 0.7,
      "step": 25834
    },
    {
      "epoch": 0.7443957817092145,
      "grad_norm": 0.7273455262184143,
      "learning_rate": 1.6177319060452878e-06,
      "loss": 0.7149,
      "step": 25835
    },
    {
      "epoch": 0.7444245951708638,
      "grad_norm": 0.7224277257919312,
      "learning_rate": 1.6173882693057946e-06,
      "loss": 0.6911,
      "step": 25836
    },
    {
      "epoch": 0.7444534086325131,
      "grad_norm": 0.7154115438461304,
      "learning_rate": 1.6170446620254664e-06,
      "loss": 0.6998,
      "step": 25837
    },
    {
      "epoch": 0.7444822220941624,
      "grad_norm": 0.706608772277832,
      "learning_rate": 1.6167010842072978e-06,
      "loss": 0.6802,
      "step": 25838
    },
    {
      "epoch": 0.7445110355558117,
      "grad_norm": 0.6856206655502319,
      "learning_rate": 1.6163575358542822e-06,
      "loss": 0.6831,
      "step": 25839
    },
    {
      "epoch": 0.744539849017461,
      "grad_norm": 0.7098620533943176,
      "learning_rate": 1.6160140169694077e-06,
      "loss": 0.6765,
      "step": 25840
    },
    {
      "epoch": 0.7445686624791102,
      "grad_norm": 0.7265403270721436,
      "learning_rate": 1.6156705275556712e-06,
      "loss": 0.68,
      "step": 25841
    },
    {
      "epoch": 0.7445974759407595,
      "grad_norm": 0.7125473022460938,
      "learning_rate": 1.615327067616057e-06,
      "loss": 0.7041,
      "step": 25842
    },
    {
      "epoch": 0.7446262894024088,
      "grad_norm": 0.7505638003349304,
      "learning_rate": 1.6149836371535644e-06,
      "loss": 0.6996,
      "step": 25843
    },
    {
      "epoch": 0.7446551028640581,
      "grad_norm": 0.7264318466186523,
      "learning_rate": 1.6146402361711788e-06,
      "loss": 0.6834,
      "step": 25844
    },
    {
      "epoch": 0.7446839163257074,
      "grad_norm": 0.723649263381958,
      "learning_rate": 1.6142968646718943e-06,
      "loss": 0.696,
      "step": 25845
    },
    {
      "epoch": 0.7447127297873567,
      "grad_norm": 0.716627299785614,
      "learning_rate": 1.6139535226586978e-06,
      "loss": 0.6819,
      "step": 25846
    },
    {
      "epoch": 0.744741543249006,
      "grad_norm": 0.7517518401145935,
      "learning_rate": 1.6136102101345835e-06,
      "loss": 0.7108,
      "step": 25847
    },
    {
      "epoch": 0.7447703567106552,
      "grad_norm": 0.713979184627533,
      "learning_rate": 1.613266927102537e-06,
      "loss": 0.6712,
      "step": 25848
    },
    {
      "epoch": 0.7447991701723045,
      "grad_norm": 0.7193649411201477,
      "learning_rate": 1.6129236735655519e-06,
      "loss": 0.6699,
      "step": 25849
    },
    {
      "epoch": 0.7448279836339537,
      "grad_norm": 0.7007912397384644,
      "learning_rate": 1.6125804495266139e-06,
      "loss": 0.6614,
      "step": 25850
    },
    {
      "epoch": 0.744856797095603,
      "grad_norm": 0.719164252281189,
      "learning_rate": 1.6122372549887155e-06,
      "loss": 0.6767,
      "step": 25851
    },
    {
      "epoch": 0.7448856105572523,
      "grad_norm": 0.7401646971702576,
      "learning_rate": 1.611894089954843e-06,
      "loss": 0.6885,
      "step": 25852
    },
    {
      "epoch": 0.7449144240189016,
      "grad_norm": 0.7082920074462891,
      "learning_rate": 1.6115509544279872e-06,
      "loss": 0.69,
      "step": 25853
    },
    {
      "epoch": 0.7449432374805509,
      "grad_norm": 0.7446842789649963,
      "learning_rate": 1.6112078484111327e-06,
      "loss": 0.7089,
      "step": 25854
    },
    {
      "epoch": 0.7449720509422002,
      "grad_norm": 0.7343615293502808,
      "learning_rate": 1.6108647719072734e-06,
      "loss": 0.6842,
      "step": 25855
    },
    {
      "epoch": 0.7450008644038495,
      "grad_norm": 0.7027886509895325,
      "learning_rate": 1.6105217249193921e-06,
      "loss": 0.6632,
      "step": 25856
    },
    {
      "epoch": 0.7450296778654988,
      "grad_norm": 0.7408277988433838,
      "learning_rate": 1.6101787074504799e-06,
      "loss": 0.6891,
      "step": 25857
    },
    {
      "epoch": 0.7450584913271481,
      "grad_norm": 0.7455893158912659,
      "learning_rate": 1.6098357195035214e-06,
      "loss": 0.6999,
      "step": 25858
    },
    {
      "epoch": 0.7450873047887974,
      "grad_norm": 0.7180150151252747,
      "learning_rate": 1.6094927610815048e-06,
      "loss": 0.6708,
      "step": 25859
    },
    {
      "epoch": 0.7451161182504467,
      "grad_norm": 0.7141180634498596,
      "learning_rate": 1.6091498321874182e-06,
      "loss": 0.6778,
      "step": 25860
    },
    {
      "epoch": 0.7451449317120958,
      "grad_norm": 0.7043668031692505,
      "learning_rate": 1.6088069328242455e-06,
      "loss": 0.6936,
      "step": 25861
    },
    {
      "epoch": 0.7451737451737451,
      "grad_norm": 0.714175820350647,
      "learning_rate": 1.608464062994976e-06,
      "loss": 0.6916,
      "step": 25862
    },
    {
      "epoch": 0.7452025586353944,
      "grad_norm": 0.7208179235458374,
      "learning_rate": 1.6081212227025927e-06,
      "loss": 0.6885,
      "step": 25863
    },
    {
      "epoch": 0.7452313720970437,
      "grad_norm": 0.732851505279541,
      "learning_rate": 1.607778411950085e-06,
      "loss": 0.6974,
      "step": 25864
    },
    {
      "epoch": 0.745260185558693,
      "grad_norm": 0.6921393871307373,
      "learning_rate": 1.6074356307404342e-06,
      "loss": 0.6665,
      "step": 25865
    },
    {
      "epoch": 0.7452889990203423,
      "grad_norm": 0.7125602960586548,
      "learning_rate": 1.6070928790766276e-06,
      "loss": 0.6914,
      "step": 25866
    },
    {
      "epoch": 0.7453178124819916,
      "grad_norm": 0.7044711112976074,
      "learning_rate": 1.6067501569616505e-06,
      "loss": 0.6864,
      "step": 25867
    },
    {
      "epoch": 0.7453466259436409,
      "grad_norm": 0.7468464970588684,
      "learning_rate": 1.6064074643984896e-06,
      "loss": 0.6879,
      "step": 25868
    },
    {
      "epoch": 0.7453754394052902,
      "grad_norm": 0.7187341451644897,
      "learning_rate": 1.6060648013901252e-06,
      "loss": 0.6684,
      "step": 25869
    },
    {
      "epoch": 0.7454042528669395,
      "grad_norm": 0.6949605941772461,
      "learning_rate": 1.6057221679395457e-06,
      "loss": 0.675,
      "step": 25870
    },
    {
      "epoch": 0.7454330663285887,
      "grad_norm": 0.7125673294067383,
      "learning_rate": 1.605379564049731e-06,
      "loss": 0.6841,
      "step": 25871
    },
    {
      "epoch": 0.745461879790238,
      "grad_norm": 0.7300274968147278,
      "learning_rate": 1.6050369897236689e-06,
      "loss": 0.6866,
      "step": 25872
    },
    {
      "epoch": 0.7454906932518873,
      "grad_norm": 0.701996922492981,
      "learning_rate": 1.6046944449643392e-06,
      "loss": 0.659,
      "step": 25873
    },
    {
      "epoch": 0.7455195067135366,
      "grad_norm": 0.6929112076759338,
      "learning_rate": 1.6043519297747284e-06,
      "loss": 0.6491,
      "step": 25874
    },
    {
      "epoch": 0.7455483201751858,
      "grad_norm": 0.7190815806388855,
      "learning_rate": 1.6040094441578164e-06,
      "loss": 0.6795,
      "step": 25875
    },
    {
      "epoch": 0.7455771336368351,
      "grad_norm": 0.7355682849884033,
      "learning_rate": 1.6036669881165872e-06,
      "loss": 0.7065,
      "step": 25876
    },
    {
      "epoch": 0.7456059470984844,
      "grad_norm": 0.7378101944923401,
      "learning_rate": 1.6033245616540249e-06,
      "loss": 0.6976,
      "step": 25877
    },
    {
      "epoch": 0.7456347605601337,
      "grad_norm": 0.70433109998703,
      "learning_rate": 1.6029821647731086e-06,
      "loss": 0.6742,
      "step": 25878
    },
    {
      "epoch": 0.745663574021783,
      "grad_norm": 0.69673091173172,
      "learning_rate": 1.6026397974768225e-06,
      "loss": 0.6724,
      "step": 25879
    },
    {
      "epoch": 0.7456923874834323,
      "grad_norm": 0.7071992754936218,
      "learning_rate": 1.6022974597681474e-06,
      "loss": 0.6855,
      "step": 25880
    },
    {
      "epoch": 0.7457212009450815,
      "grad_norm": 0.7156060934066772,
      "learning_rate": 1.6019551516500664e-06,
      "loss": 0.6638,
      "step": 25881
    },
    {
      "epoch": 0.7457500144067308,
      "grad_norm": 0.7190232872962952,
      "learning_rate": 1.6016128731255577e-06,
      "loss": 0.7193,
      "step": 25882
    },
    {
      "epoch": 0.7457788278683801,
      "grad_norm": 0.7495397329330444,
      "learning_rate": 1.6012706241976051e-06,
      "loss": 0.7078,
      "step": 25883
    },
    {
      "epoch": 0.7458076413300294,
      "grad_norm": 0.7215536832809448,
      "learning_rate": 1.6009284048691864e-06,
      "loss": 0.6601,
      "step": 25884
    },
    {
      "epoch": 0.7458364547916787,
      "grad_norm": 0.7200550436973572,
      "learning_rate": 1.600586215143285e-06,
      "loss": 0.6651,
      "step": 25885
    },
    {
      "epoch": 0.745865268253328,
      "grad_norm": 0.7312784194946289,
      "learning_rate": 1.6002440550228781e-06,
      "loss": 0.6948,
      "step": 25886
    },
    {
      "epoch": 0.7458940817149773,
      "grad_norm": 0.71431565284729,
      "learning_rate": 1.5999019245109488e-06,
      "loss": 0.6772,
      "step": 25887
    },
    {
      "epoch": 0.7459228951766265,
      "grad_norm": 0.7266595959663391,
      "learning_rate": 1.5995598236104725e-06,
      "loss": 0.7122,
      "step": 25888
    },
    {
      "epoch": 0.7459517086382758,
      "grad_norm": 0.7325490117073059,
      "learning_rate": 1.5992177523244328e-06,
      "loss": 0.7084,
      "step": 25889
    },
    {
      "epoch": 0.745980522099925,
      "grad_norm": 0.7153282165527344,
      "learning_rate": 1.5988757106558045e-06,
      "loss": 0.693,
      "step": 25890
    },
    {
      "epoch": 0.7460093355615743,
      "grad_norm": 0.7216179966926575,
      "learning_rate": 1.598533698607571e-06,
      "loss": 0.6919,
      "step": 25891
    },
    {
      "epoch": 0.7460381490232236,
      "grad_norm": 0.7263622283935547,
      "learning_rate": 1.5981917161827082e-06,
      "loss": 0.7064,
      "step": 25892
    },
    {
      "epoch": 0.7460669624848729,
      "grad_norm": 0.7410537004470825,
      "learning_rate": 1.5978497633841961e-06,
      "loss": 0.6921,
      "step": 25893
    },
    {
      "epoch": 0.7460957759465222,
      "grad_norm": 0.7113615870475769,
      "learning_rate": 1.5975078402150096e-06,
      "loss": 0.7054,
      "step": 25894
    },
    {
      "epoch": 0.7461245894081715,
      "grad_norm": 0.7027305960655212,
      "learning_rate": 1.5971659466781308e-06,
      "loss": 0.6641,
      "step": 25895
    },
    {
      "epoch": 0.7461534028698208,
      "grad_norm": 0.7237069606781006,
      "learning_rate": 1.596824082776533e-06,
      "loss": 0.6896,
      "step": 25896
    },
    {
      "epoch": 0.7461822163314701,
      "grad_norm": 0.7129207849502563,
      "learning_rate": 1.596482248513196e-06,
      "loss": 0.6891,
      "step": 25897
    },
    {
      "epoch": 0.7462110297931194,
      "grad_norm": 0.7014603018760681,
      "learning_rate": 1.5961404438910976e-06,
      "loss": 0.6698,
      "step": 25898
    },
    {
      "epoch": 0.7462398432547687,
      "grad_norm": 0.6988769769668579,
      "learning_rate": 1.5957986689132115e-06,
      "loss": 0.699,
      "step": 25899
    },
    {
      "epoch": 0.746268656716418,
      "grad_norm": 0.6982099413871765,
      "learning_rate": 1.5954569235825185e-06,
      "loss": 0.6766,
      "step": 25900
    },
    {
      "epoch": 0.7462974701780672,
      "grad_norm": 0.7418314814567566,
      "learning_rate": 1.59511520790199e-06,
      "loss": 0.7085,
      "step": 25901
    },
    {
      "epoch": 0.7463262836397164,
      "grad_norm": 0.7400418519973755,
      "learning_rate": 1.5947735218746047e-06,
      "loss": 0.6893,
      "step": 25902
    },
    {
      "epoch": 0.7463550971013657,
      "grad_norm": 0.7040691375732422,
      "learning_rate": 1.5944318655033381e-06,
      "loss": 0.663,
      "step": 25903
    },
    {
      "epoch": 0.746383910563015,
      "grad_norm": 0.7214556336402893,
      "learning_rate": 1.5940902387911667e-06,
      "loss": 0.6854,
      "step": 25904
    },
    {
      "epoch": 0.7464127240246643,
      "grad_norm": 0.7144784331321716,
      "learning_rate": 1.5937486417410637e-06,
      "loss": 0.682,
      "step": 25905
    },
    {
      "epoch": 0.7464415374863136,
      "grad_norm": 0.7108849883079529,
      "learning_rate": 1.5934070743560066e-06,
      "loss": 0.6723,
      "step": 25906
    },
    {
      "epoch": 0.7464703509479629,
      "grad_norm": 0.7356094717979431,
      "learning_rate": 1.5930655366389663e-06,
      "loss": 0.7004,
      "step": 25907
    },
    {
      "epoch": 0.7464991644096122,
      "grad_norm": 0.7073661088943481,
      "learning_rate": 1.592724028592922e-06,
      "loss": 0.6824,
      "step": 25908
    },
    {
      "epoch": 0.7465279778712615,
      "grad_norm": 0.7170729637145996,
      "learning_rate": 1.5923825502208429e-06,
      "loss": 0.682,
      "step": 25909
    },
    {
      "epoch": 0.7465567913329108,
      "grad_norm": 0.7452166676521301,
      "learning_rate": 1.5920411015257075e-06,
      "loss": 0.6932,
      "step": 25910
    },
    {
      "epoch": 0.74658560479456,
      "grad_norm": 0.7134273648262024,
      "learning_rate": 1.5916996825104858e-06,
      "loss": 0.6969,
      "step": 25911
    },
    {
      "epoch": 0.7466144182562093,
      "grad_norm": 0.7236902713775635,
      "learning_rate": 1.591358293178154e-06,
      "loss": 0.6651,
      "step": 25912
    },
    {
      "epoch": 0.7466432317178586,
      "grad_norm": 0.6989777088165283,
      "learning_rate": 1.5910169335316828e-06,
      "loss": 0.6662,
      "step": 25913
    },
    {
      "epoch": 0.7466720451795079,
      "grad_norm": 0.7048904299736023,
      "learning_rate": 1.590675603574046e-06,
      "loss": 0.6973,
      "step": 25914
    },
    {
      "epoch": 0.7467008586411571,
      "grad_norm": 0.7163840532302856,
      "learning_rate": 1.5903343033082176e-06,
      "loss": 0.7014,
      "step": 25915
    },
    {
      "epoch": 0.7467296721028064,
      "grad_norm": 0.7022039294242859,
      "learning_rate": 1.58999303273717e-06,
      "loss": 0.7138,
      "step": 25916
    },
    {
      "epoch": 0.7467584855644557,
      "grad_norm": 0.7060738205909729,
      "learning_rate": 1.5896517918638727e-06,
      "loss": 0.6615,
      "step": 25917
    },
    {
      "epoch": 0.746787299026105,
      "grad_norm": 0.7035341858863831,
      "learning_rate": 1.589310580691299e-06,
      "loss": 0.6765,
      "step": 25918
    },
    {
      "epoch": 0.7468161124877543,
      "grad_norm": 0.7144138813018799,
      "learning_rate": 1.5889693992224226e-06,
      "loss": 0.6819,
      "step": 25919
    },
    {
      "epoch": 0.7468449259494035,
      "grad_norm": 0.7106720805168152,
      "learning_rate": 1.588628247460211e-06,
      "loss": 0.7082,
      "step": 25920
    },
    {
      "epoch": 0.7468737394110528,
      "grad_norm": 0.7120151519775391,
      "learning_rate": 1.5882871254076394e-06,
      "loss": 0.683,
      "step": 25921
    },
    {
      "epoch": 0.7469025528727021,
      "grad_norm": 0.7220979332923889,
      "learning_rate": 1.5879460330676744e-06,
      "loss": 0.6956,
      "step": 25922
    },
    {
      "epoch": 0.7469313663343514,
      "grad_norm": 0.7245699167251587,
      "learning_rate": 1.5876049704432905e-06,
      "loss": 0.6957,
      "step": 25923
    },
    {
      "epoch": 0.7469601797960007,
      "grad_norm": 0.7645201086997986,
      "learning_rate": 1.5872639375374542e-06,
      "loss": 0.6896,
      "step": 25924
    },
    {
      "epoch": 0.74698899325765,
      "grad_norm": 0.7149546146392822,
      "learning_rate": 1.5869229343531394e-06,
      "loss": 0.7033,
      "step": 25925
    },
    {
      "epoch": 0.7470178067192993,
      "grad_norm": 0.710543692111969,
      "learning_rate": 1.5865819608933108e-06,
      "loss": 0.6782,
      "step": 25926
    },
    {
      "epoch": 0.7470466201809486,
      "grad_norm": 0.7285243272781372,
      "learning_rate": 1.5862410171609443e-06,
      "loss": 0.7015,
      "step": 25927
    },
    {
      "epoch": 0.7470754336425979,
      "grad_norm": 0.7090859413146973,
      "learning_rate": 1.5859001031590044e-06,
      "loss": 0.6798,
      "step": 25928
    },
    {
      "epoch": 0.747104247104247,
      "grad_norm": 0.7422881722450256,
      "learning_rate": 1.5855592188904633e-06,
      "loss": 0.6747,
      "step": 25929
    },
    {
      "epoch": 0.7471330605658963,
      "grad_norm": 0.7071202397346497,
      "learning_rate": 1.5852183643582869e-06,
      "loss": 0.6806,
      "step": 25930
    },
    {
      "epoch": 0.7471618740275456,
      "grad_norm": 0.7126176357269287,
      "learning_rate": 1.5848775395654464e-06,
      "loss": 0.6823,
      "step": 25931
    },
    {
      "epoch": 0.7471906874891949,
      "grad_norm": 0.7202418446540833,
      "learning_rate": 1.5845367445149074e-06,
      "loss": 0.6814,
      "step": 25932
    },
    {
      "epoch": 0.7472195009508442,
      "grad_norm": 0.7376303672790527,
      "learning_rate": 1.5841959792096402e-06,
      "loss": 0.6905,
      "step": 25933
    },
    {
      "epoch": 0.7472483144124935,
      "grad_norm": 0.7206810712814331,
      "learning_rate": 1.58385524365261e-06,
      "loss": 0.6877,
      "step": 25934
    },
    {
      "epoch": 0.7472771278741428,
      "grad_norm": 0.7120212316513062,
      "learning_rate": 1.583514537846786e-06,
      "loss": 0.6878,
      "step": 25935
    },
    {
      "epoch": 0.7473059413357921,
      "grad_norm": 0.7408607602119446,
      "learning_rate": 1.5831738617951364e-06,
      "loss": 0.6836,
      "step": 25936
    },
    {
      "epoch": 0.7473347547974414,
      "grad_norm": 0.7495288848876953,
      "learning_rate": 1.582833215500626e-06,
      "loss": 0.6671,
      "step": 25937
    },
    {
      "epoch": 0.7473635682590907,
      "grad_norm": 0.7132789492607117,
      "learning_rate": 1.5824925989662216e-06,
      "loss": 0.6781,
      "step": 25938
    },
    {
      "epoch": 0.74739238172074,
      "grad_norm": 0.7366125583648682,
      "learning_rate": 1.5821520121948924e-06,
      "loss": 0.7105,
      "step": 25939
    },
    {
      "epoch": 0.7474211951823893,
      "grad_norm": 0.71140456199646,
      "learning_rate": 1.5818114551896007e-06,
      "loss": 0.6871,
      "step": 25940
    },
    {
      "epoch": 0.7474500086440385,
      "grad_norm": 0.7122724652290344,
      "learning_rate": 1.581470927953314e-06,
      "loss": 0.6747,
      "step": 25941
    },
    {
      "epoch": 0.7474788221056877,
      "grad_norm": 0.710806667804718,
      "learning_rate": 1.5811304304890006e-06,
      "loss": 0.6771,
      "step": 25942
    },
    {
      "epoch": 0.747507635567337,
      "grad_norm": 0.7077122926712036,
      "learning_rate": 1.5807899627996215e-06,
      "loss": 0.6811,
      "step": 25943
    },
    {
      "epoch": 0.7475364490289863,
      "grad_norm": 0.7015256285667419,
      "learning_rate": 1.5804495248881458e-06,
      "loss": 0.6927,
      "step": 25944
    },
    {
      "epoch": 0.7475652624906356,
      "grad_norm": 0.7213323712348938,
      "learning_rate": 1.5801091167575343e-06,
      "loss": 0.6718,
      "step": 25945
    },
    {
      "epoch": 0.7475940759522849,
      "grad_norm": 0.717910885810852,
      "learning_rate": 1.5797687384107557e-06,
      "loss": 0.7175,
      "step": 25946
    },
    {
      "epoch": 0.7476228894139342,
      "grad_norm": 0.6921218633651733,
      "learning_rate": 1.5794283898507711e-06,
      "loss": 0.6945,
      "step": 25947
    },
    {
      "epoch": 0.7476517028755835,
      "grad_norm": 0.7334056496620178,
      "learning_rate": 1.5790880710805468e-06,
      "loss": 0.7178,
      "step": 25948
    },
    {
      "epoch": 0.7476805163372328,
      "grad_norm": 0.7174080610275269,
      "learning_rate": 1.5787477821030444e-06,
      "loss": 0.708,
      "step": 25949
    },
    {
      "epoch": 0.747709329798882,
      "grad_norm": 0.7137632966041565,
      "learning_rate": 1.5784075229212292e-06,
      "loss": 0.6773,
      "step": 25950
    },
    {
      "epoch": 0.7477381432605313,
      "grad_norm": 0.6895909309387207,
      "learning_rate": 1.5780672935380637e-06,
      "loss": 0.6592,
      "step": 25951
    },
    {
      "epoch": 0.7477669567221806,
      "grad_norm": 0.7246680855751038,
      "learning_rate": 1.577727093956513e-06,
      "loss": 0.693,
      "step": 25952
    },
    {
      "epoch": 0.7477957701838299,
      "grad_norm": 0.736442506313324,
      "learning_rate": 1.577386924179537e-06,
      "loss": 0.6845,
      "step": 25953
    },
    {
      "epoch": 0.7478245836454792,
      "grad_norm": 0.6965761780738831,
      "learning_rate": 1.5770467842101012e-06,
      "loss": 0.6867,
      "step": 25954
    },
    {
      "epoch": 0.7478533971071285,
      "grad_norm": 0.7189645171165466,
      "learning_rate": 1.5767066740511643e-06,
      "loss": 0.7078,
      "step": 25955
    },
    {
      "epoch": 0.7478822105687777,
      "grad_norm": 0.740402102470398,
      "learning_rate": 1.5763665937056904e-06,
      "loss": 0.6974,
      "step": 25956
    },
    {
      "epoch": 0.747911024030427,
      "grad_norm": 0.7115825414657593,
      "learning_rate": 1.5760265431766424e-06,
      "loss": 0.6674,
      "step": 25957
    },
    {
      "epoch": 0.7479398374920763,
      "grad_norm": 0.7165493965148926,
      "learning_rate": 1.5756865224669792e-06,
      "loss": 0.6904,
      "step": 25958
    },
    {
      "epoch": 0.7479686509537256,
      "grad_norm": 0.7224233746528625,
      "learning_rate": 1.5753465315796646e-06,
      "loss": 0.6688,
      "step": 25959
    },
    {
      "epoch": 0.7479974644153748,
      "grad_norm": 0.7232125401496887,
      "learning_rate": 1.575006570517657e-06,
      "loss": 0.6774,
      "step": 25960
    },
    {
      "epoch": 0.7480262778770241,
      "grad_norm": 0.7066463828086853,
      "learning_rate": 1.5746666392839199e-06,
      "loss": 0.6827,
      "step": 25961
    },
    {
      "epoch": 0.7480550913386734,
      "grad_norm": 0.7304157614707947,
      "learning_rate": 1.5743267378814092e-06,
      "loss": 0.6801,
      "step": 25962
    },
    {
      "epoch": 0.7480839048003227,
      "grad_norm": 0.7240586280822754,
      "learning_rate": 1.5739868663130914e-06,
      "loss": 0.6968,
      "step": 25963
    },
    {
      "epoch": 0.748112718261972,
      "grad_norm": 0.739472508430481,
      "learning_rate": 1.5736470245819218e-06,
      "loss": 0.689,
      "step": 25964
    },
    {
      "epoch": 0.7481415317236213,
      "grad_norm": 0.7109459638595581,
      "learning_rate": 1.5733072126908627e-06,
      "loss": 0.6967,
      "step": 25965
    },
    {
      "epoch": 0.7481703451852706,
      "grad_norm": 0.7206213474273682,
      "learning_rate": 1.5729674306428711e-06,
      "loss": 0.6759,
      "step": 25966
    },
    {
      "epoch": 0.7481991586469199,
      "grad_norm": 2.3872203826904297,
      "learning_rate": 1.5726276784409088e-06,
      "loss": 0.7503,
      "step": 25967
    },
    {
      "epoch": 0.7482279721085692,
      "grad_norm": 0.725796103477478,
      "learning_rate": 1.5722879560879317e-06,
      "loss": 0.6716,
      "step": 25968
    },
    {
      "epoch": 0.7482567855702184,
      "grad_norm": 0.7103176116943359,
      "learning_rate": 1.5719482635869015e-06,
      "loss": 0.6698,
      "step": 25969
    },
    {
      "epoch": 0.7482855990318676,
      "grad_norm": 0.6872797012329102,
      "learning_rate": 1.5716086009407739e-06,
      "loss": 0.6752,
      "step": 25970
    },
    {
      "epoch": 0.7483144124935169,
      "grad_norm": 0.6959162354469299,
      "learning_rate": 1.5712689681525095e-06,
      "loss": 0.6569,
      "step": 25971
    },
    {
      "epoch": 0.7483432259551662,
      "grad_norm": 0.7618061304092407,
      "learning_rate": 1.5709293652250634e-06,
      "loss": 0.6846,
      "step": 25972
    },
    {
      "epoch": 0.7483720394168155,
      "grad_norm": 0.7227018475532532,
      "learning_rate": 1.570589792161395e-06,
      "loss": 0.6808,
      "step": 25973
    },
    {
      "epoch": 0.7484008528784648,
      "grad_norm": 0.704612672328949,
      "learning_rate": 1.5702502489644611e-06,
      "loss": 0.6751,
      "step": 25974
    },
    {
      "epoch": 0.7484296663401141,
      "grad_norm": 0.7497157454490662,
      "learning_rate": 1.5699107356372206e-06,
      "loss": 0.6781,
      "step": 25975
    },
    {
      "epoch": 0.7484584798017634,
      "grad_norm": 0.7175446152687073,
      "learning_rate": 1.5695712521826274e-06,
      "loss": 0.6755,
      "step": 25976
    },
    {
      "epoch": 0.7484872932634127,
      "grad_norm": 0.7417196035385132,
      "learning_rate": 1.5692317986036392e-06,
      "loss": 0.6733,
      "step": 25977
    },
    {
      "epoch": 0.748516106725062,
      "grad_norm": 0.7210637331008911,
      "learning_rate": 1.5688923749032143e-06,
      "loss": 0.7004,
      "step": 25978
    },
    {
      "epoch": 0.7485449201867113,
      "grad_norm": 0.7472119331359863,
      "learning_rate": 1.5685529810843047e-06,
      "loss": 0.6946,
      "step": 25979
    },
    {
      "epoch": 0.7485737336483606,
      "grad_norm": 0.7184452414512634,
      "learning_rate": 1.5682136171498707e-06,
      "loss": 0.6991,
      "step": 25980
    },
    {
      "epoch": 0.7486025471100098,
      "grad_norm": 0.7287710309028625,
      "learning_rate": 1.5678742831028637e-06,
      "loss": 0.6859,
      "step": 25981
    },
    {
      "epoch": 0.7486313605716591,
      "grad_norm": 0.7213144302368164,
      "learning_rate": 1.567534978946243e-06,
      "loss": 0.6943,
      "step": 25982
    },
    {
      "epoch": 0.7486601740333083,
      "grad_norm": 0.7327238321304321,
      "learning_rate": 1.5671957046829594e-06,
      "loss": 0.6914,
      "step": 25983
    },
    {
      "epoch": 0.7486889874949576,
      "grad_norm": 0.715526282787323,
      "learning_rate": 1.5668564603159708e-06,
      "loss": 0.6753,
      "step": 25984
    },
    {
      "epoch": 0.7487178009566069,
      "grad_norm": 0.7058976888656616,
      "learning_rate": 1.5665172458482297e-06,
      "loss": 0.6526,
      "step": 25985
    },
    {
      "epoch": 0.7487466144182562,
      "grad_norm": 0.7099912762641907,
      "learning_rate": 1.566178061282691e-06,
      "loss": 0.6611,
      "step": 25986
    },
    {
      "epoch": 0.7487754278799055,
      "grad_norm": 0.7097283601760864,
      "learning_rate": 1.5658389066223083e-06,
      "loss": 0.6806,
      "step": 25987
    },
    {
      "epoch": 0.7488042413415548,
      "grad_norm": 0.7076934576034546,
      "learning_rate": 1.565499781870038e-06,
      "loss": 0.6639,
      "step": 25988
    },
    {
      "epoch": 0.7488330548032041,
      "grad_norm": 0.7097505331039429,
      "learning_rate": 1.5651606870288295e-06,
      "loss": 0.6958,
      "step": 25989
    },
    {
      "epoch": 0.7488618682648533,
      "grad_norm": 0.7394378781318665,
      "learning_rate": 1.5648216221016394e-06,
      "loss": 0.6953,
      "step": 25990
    },
    {
      "epoch": 0.7488906817265026,
      "grad_norm": 0.7117148041725159,
      "learning_rate": 1.5644825870914176e-06,
      "loss": 0.6856,
      "step": 25991
    },
    {
      "epoch": 0.7489194951881519,
      "grad_norm": 0.7153586745262146,
      "learning_rate": 1.5641435820011197e-06,
      "loss": 0.6988,
      "step": 25992
    },
    {
      "epoch": 0.7489483086498012,
      "grad_norm": 0.7027495503425598,
      "learning_rate": 1.5638046068336949e-06,
      "loss": 0.695,
      "step": 25993
    },
    {
      "epoch": 0.7489771221114505,
      "grad_norm": 0.6950374841690063,
      "learning_rate": 1.5634656615920974e-06,
      "loss": 0.6978,
      "step": 25994
    },
    {
      "epoch": 0.7490059355730998,
      "grad_norm": 0.7312486171722412,
      "learning_rate": 1.5631267462792798e-06,
      "loss": 0.6855,
      "step": 25995
    },
    {
      "epoch": 0.749034749034749,
      "grad_norm": 0.7112937569618225,
      "learning_rate": 1.5627878608981917e-06,
      "loss": 0.6977,
      "step": 25996
    },
    {
      "epoch": 0.7490635624963983,
      "grad_norm": 0.7044049501419067,
      "learning_rate": 1.5624490054517865e-06,
      "loss": 0.6938,
      "step": 25997
    },
    {
      "epoch": 0.7490923759580476,
      "grad_norm": 0.7449430823326111,
      "learning_rate": 1.5621101799430128e-06,
      "loss": 0.6936,
      "step": 25998
    },
    {
      "epoch": 0.7491211894196969,
      "grad_norm": 0.7346492409706116,
      "learning_rate": 1.5617713843748222e-06,
      "loss": 0.6891,
      "step": 25999
    },
    {
      "epoch": 0.7491500028813461,
      "grad_norm": 0.7031539678573608,
      "learning_rate": 1.5614326187501666e-06,
      "loss": 0.7037,
      "step": 26000
    },
    {
      "epoch": 0.7491788163429954,
      "grad_norm": 0.7211589813232422,
      "learning_rate": 1.5610938830719969e-06,
      "loss": 0.6975,
      "step": 26001
    },
    {
      "epoch": 0.7492076298046447,
      "grad_norm": 0.7723156809806824,
      "learning_rate": 1.56075517734326e-06,
      "loss": 0.7015,
      "step": 26002
    },
    {
      "epoch": 0.749236443266294,
      "grad_norm": 0.6990417838096619,
      "learning_rate": 1.5604165015669098e-06,
      "loss": 0.6996,
      "step": 26003
    },
    {
      "epoch": 0.7492652567279433,
      "grad_norm": 0.7024536728858948,
      "learning_rate": 1.5600778557458912e-06,
      "loss": 0.6348,
      "step": 26004
    },
    {
      "epoch": 0.7492940701895926,
      "grad_norm": 0.7293808460235596,
      "learning_rate": 1.5597392398831579e-06,
      "loss": 0.6696,
      "step": 26005
    },
    {
      "epoch": 0.7493228836512419,
      "grad_norm": 0.7539089918136597,
      "learning_rate": 1.5594006539816548e-06,
      "loss": 0.6989,
      "step": 26006
    },
    {
      "epoch": 0.7493516971128912,
      "grad_norm": 0.7123863697052002,
      "learning_rate": 1.559062098044335e-06,
      "loss": 0.7078,
      "step": 26007
    },
    {
      "epoch": 0.7493805105745405,
      "grad_norm": 0.7096261978149414,
      "learning_rate": 1.5587235720741422e-06,
      "loss": 0.6953,
      "step": 26008
    },
    {
      "epoch": 0.7494093240361898,
      "grad_norm": 0.7086283564567566,
      "learning_rate": 1.5583850760740292e-06,
      "loss": 0.6978,
      "step": 26009
    },
    {
      "epoch": 0.7494381374978389,
      "grad_norm": 0.7313936352729797,
      "learning_rate": 1.5580466100469382e-06,
      "loss": 0.7023,
      "step": 26010
    },
    {
      "epoch": 0.7494669509594882,
      "grad_norm": 0.7141929864883423,
      "learning_rate": 1.5577081739958238e-06,
      "loss": 0.678,
      "step": 26011
    },
    {
      "epoch": 0.7494957644211375,
      "grad_norm": 0.7153409719467163,
      "learning_rate": 1.557369767923629e-06,
      "loss": 0.6944,
      "step": 26012
    },
    {
      "epoch": 0.7495245778827868,
      "grad_norm": 0.6900660395622253,
      "learning_rate": 1.5570313918333035e-06,
      "loss": 0.6686,
      "step": 26013
    },
    {
      "epoch": 0.7495533913444361,
      "grad_norm": 0.705167829990387,
      "learning_rate": 1.5566930457277912e-06,
      "loss": 0.6797,
      "step": 26014
    },
    {
      "epoch": 0.7495822048060854,
      "grad_norm": 0.6848199367523193,
      "learning_rate": 1.5563547296100406e-06,
      "loss": 0.6581,
      "step": 26015
    },
    {
      "epoch": 0.7496110182677347,
      "grad_norm": 0.7126948833465576,
      "learning_rate": 1.5560164434829994e-06,
      "loss": 0.7027,
      "step": 26016
    },
    {
      "epoch": 0.749639831729384,
      "grad_norm": 0.7053586840629578,
      "learning_rate": 1.5556781873496107e-06,
      "loss": 0.6899,
      "step": 26017
    },
    {
      "epoch": 0.7496686451910333,
      "grad_norm": 0.7296484708786011,
      "learning_rate": 1.5553399612128233e-06,
      "loss": 0.7014,
      "step": 26018
    },
    {
      "epoch": 0.7496974586526826,
      "grad_norm": 0.7593721151351929,
      "learning_rate": 1.5550017650755806e-06,
      "loss": 0.6883,
      "step": 26019
    },
    {
      "epoch": 0.7497262721143318,
      "grad_norm": 0.7186517715454102,
      "learning_rate": 1.5546635989408294e-06,
      "loss": 0.6957,
      "step": 26020
    },
    {
      "epoch": 0.7497550855759811,
      "grad_norm": 0.7100345492362976,
      "learning_rate": 1.554325462811513e-06,
      "loss": 0.6831,
      "step": 26021
    },
    {
      "epoch": 0.7497838990376304,
      "grad_norm": 0.7218177914619446,
      "learning_rate": 1.5539873566905779e-06,
      "loss": 0.6937,
      "step": 26022
    },
    {
      "epoch": 0.7498127124992796,
      "grad_norm": 0.7207426428794861,
      "learning_rate": 1.5536492805809673e-06,
      "loss": 0.6819,
      "step": 26023
    },
    {
      "epoch": 0.7498415259609289,
      "grad_norm": 0.7501432299613953,
      "learning_rate": 1.5533112344856282e-06,
      "loss": 0.7129,
      "step": 26024
    },
    {
      "epoch": 0.7498703394225782,
      "grad_norm": 0.7084473371505737,
      "learning_rate": 1.5529732184075015e-06,
      "loss": 0.6798,
      "step": 26025
    },
    {
      "epoch": 0.7498991528842275,
      "grad_norm": 0.7166042923927307,
      "learning_rate": 1.5526352323495335e-06,
      "loss": 0.6985,
      "step": 26026
    },
    {
      "epoch": 0.7499279663458768,
      "grad_norm": 0.7320324778556824,
      "learning_rate": 1.5522972763146653e-06,
      "loss": 0.6741,
      "step": 26027
    },
    {
      "epoch": 0.7499567798075261,
      "grad_norm": 0.7192941904067993,
      "learning_rate": 1.5519593503058433e-06,
      "loss": 0.6894,
      "step": 26028
    },
    {
      "epoch": 0.7499855932691754,
      "grad_norm": 0.721116304397583,
      "learning_rate": 1.5516214543260067e-06,
      "loss": 0.6799,
      "step": 26029
    },
    {
      "epoch": 0.7500144067308246,
      "grad_norm": 0.7069655060768127,
      "learning_rate": 1.5512835883781014e-06,
      "loss": 0.6789,
      "step": 26030
    },
    {
      "epoch": 0.7500432201924739,
      "grad_norm": 0.7395453453063965,
      "learning_rate": 1.5509457524650672e-06,
      "loss": 0.6965,
      "step": 26031
    },
    {
      "epoch": 0.7500720336541232,
      "grad_norm": 0.7032768130302429,
      "learning_rate": 1.5506079465898478e-06,
      "loss": 0.6802,
      "step": 26032
    },
    {
      "epoch": 0.7501008471157725,
      "grad_norm": 0.7164782285690308,
      "learning_rate": 1.5502701707553846e-06,
      "loss": 0.6958,
      "step": 26033
    },
    {
      "epoch": 0.7501296605774218,
      "grad_norm": 0.7434179186820984,
      "learning_rate": 1.549932424964622e-06,
      "loss": 0.6902,
      "step": 26034
    },
    {
      "epoch": 0.7501584740390711,
      "grad_norm": 0.7234659790992737,
      "learning_rate": 1.5495947092204972e-06,
      "loss": 0.6969,
      "step": 26035
    },
    {
      "epoch": 0.7501872875007204,
      "grad_norm": 0.7094215154647827,
      "learning_rate": 1.5492570235259547e-06,
      "loss": 0.6981,
      "step": 26036
    },
    {
      "epoch": 0.7502161009623696,
      "grad_norm": 0.7103114724159241,
      "learning_rate": 1.5489193678839326e-06,
      "loss": 0.7017,
      "step": 26037
    },
    {
      "epoch": 0.7502449144240189,
      "grad_norm": 0.7196153998374939,
      "learning_rate": 1.548581742297373e-06,
      "loss": 0.675,
      "step": 26038
    },
    {
      "epoch": 0.7502737278856682,
      "grad_norm": 0.7108132243156433,
      "learning_rate": 1.5482441467692183e-06,
      "loss": 0.7072,
      "step": 26039
    },
    {
      "epoch": 0.7503025413473174,
      "grad_norm": 0.7118484377861023,
      "learning_rate": 1.5479065813024046e-06,
      "loss": 0.6842,
      "step": 26040
    },
    {
      "epoch": 0.7503313548089667,
      "grad_norm": 0.7215368747711182,
      "learning_rate": 1.5475690458998754e-06,
      "loss": 0.6839,
      "step": 26041
    },
    {
      "epoch": 0.750360168270616,
      "grad_norm": 0.7427811622619629,
      "learning_rate": 1.5472315405645672e-06,
      "loss": 0.6833,
      "step": 26042
    },
    {
      "epoch": 0.7503889817322653,
      "grad_norm": 0.7304307222366333,
      "learning_rate": 1.5468940652994225e-06,
      "loss": 0.6918,
      "step": 26043
    },
    {
      "epoch": 0.7504177951939146,
      "grad_norm": 0.7194227576255798,
      "learning_rate": 1.5465566201073768e-06,
      "loss": 0.7114,
      "step": 26044
    },
    {
      "epoch": 0.7504466086555639,
      "grad_norm": 0.6923680901527405,
      "learning_rate": 1.5462192049913704e-06,
      "loss": 0.6952,
      "step": 26045
    },
    {
      "epoch": 0.7504754221172132,
      "grad_norm": 0.7315395474433899,
      "learning_rate": 1.5458818199543424e-06,
      "loss": 0.6905,
      "step": 26046
    },
    {
      "epoch": 0.7505042355788625,
      "grad_norm": 0.7108258008956909,
      "learning_rate": 1.545544464999233e-06,
      "loss": 0.679,
      "step": 26047
    },
    {
      "epoch": 0.7505330490405118,
      "grad_norm": 0.7262860536575317,
      "learning_rate": 1.545207140128976e-06,
      "loss": 0.6976,
      "step": 26048
    },
    {
      "epoch": 0.7505618625021611,
      "grad_norm": 0.7440662384033203,
      "learning_rate": 1.5448698453465134e-06,
      "loss": 0.7137,
      "step": 26049
    },
    {
      "epoch": 0.7505906759638102,
      "grad_norm": 0.7015725374221802,
      "learning_rate": 1.5445325806547785e-06,
      "loss": 0.6619,
      "step": 26050
    },
    {
      "epoch": 0.7506194894254595,
      "grad_norm": 0.7440477609634399,
      "learning_rate": 1.5441953460567123e-06,
      "loss": 0.7057,
      "step": 26051
    },
    {
      "epoch": 0.7506483028871088,
      "grad_norm": 0.7251977920532227,
      "learning_rate": 1.5438581415552483e-06,
      "loss": 0.6917,
      "step": 26052
    },
    {
      "epoch": 0.7506771163487581,
      "grad_norm": 0.7080700397491455,
      "learning_rate": 1.5435209671533253e-06,
      "loss": 0.683,
      "step": 26053
    },
    {
      "epoch": 0.7507059298104074,
      "grad_norm": 0.7121289372444153,
      "learning_rate": 1.5431838228538809e-06,
      "loss": 0.6776,
      "step": 26054
    },
    {
      "epoch": 0.7507347432720567,
      "grad_norm": 0.7286261320114136,
      "learning_rate": 1.5428467086598487e-06,
      "loss": 0.7045,
      "step": 26055
    },
    {
      "epoch": 0.750763556733706,
      "grad_norm": 0.7076845765113831,
      "learning_rate": 1.5425096245741672e-06,
      "loss": 0.6857,
      "step": 26056
    },
    {
      "epoch": 0.7507923701953553,
      "grad_norm": 0.7191146016120911,
      "learning_rate": 1.5421725705997686e-06,
      "loss": 0.6896,
      "step": 26057
    },
    {
      "epoch": 0.7508211836570046,
      "grad_norm": 0.7216758728027344,
      "learning_rate": 1.5418355467395906e-06,
      "loss": 0.6866,
      "step": 26058
    },
    {
      "epoch": 0.7508499971186539,
      "grad_norm": 0.7348895072937012,
      "learning_rate": 1.5414985529965682e-06,
      "loss": 0.6587,
      "step": 26059
    },
    {
      "epoch": 0.7508788105803031,
      "grad_norm": 0.7267506122589111,
      "learning_rate": 1.5411615893736375e-06,
      "loss": 0.7116,
      "step": 26060
    },
    {
      "epoch": 0.7509076240419524,
      "grad_norm": 0.7080441117286682,
      "learning_rate": 1.5408246558737305e-06,
      "loss": 0.6839,
      "step": 26061
    },
    {
      "epoch": 0.7509364375036017,
      "grad_norm": 0.7063183188438416,
      "learning_rate": 1.5404877524997847e-06,
      "loss": 0.6763,
      "step": 26062
    },
    {
      "epoch": 0.750965250965251,
      "grad_norm": 0.726954996585846,
      "learning_rate": 1.5401508792547304e-06,
      "loss": 0.7111,
      "step": 26063
    },
    {
      "epoch": 0.7509940644269002,
      "grad_norm": 0.7161133289337158,
      "learning_rate": 1.5398140361415048e-06,
      "loss": 0.699,
      "step": 26064
    },
    {
      "epoch": 0.7510228778885495,
      "grad_norm": 0.7526089549064636,
      "learning_rate": 1.5394772231630383e-06,
      "loss": 0.718,
      "step": 26065
    },
    {
      "epoch": 0.7510516913501988,
      "grad_norm": 0.7166733741760254,
      "learning_rate": 1.5391404403222676e-06,
      "loss": 0.68,
      "step": 26066
    },
    {
      "epoch": 0.7510805048118481,
      "grad_norm": 0.7198195457458496,
      "learning_rate": 1.538803687622123e-06,
      "loss": 0.7149,
      "step": 26067
    },
    {
      "epoch": 0.7511093182734974,
      "grad_norm": 0.711819052696228,
      "learning_rate": 1.5384669650655393e-06,
      "loss": 0.6727,
      "step": 26068
    },
    {
      "epoch": 0.7511381317351467,
      "grad_norm": 0.7391241788864136,
      "learning_rate": 1.5381302726554454e-06,
      "loss": 0.7034,
      "step": 26069
    },
    {
      "epoch": 0.7511669451967959,
      "grad_norm": 0.7137904763221741,
      "learning_rate": 1.5377936103947788e-06,
      "loss": 0.6961,
      "step": 26070
    },
    {
      "epoch": 0.7511957586584452,
      "grad_norm": 0.7183562517166138,
      "learning_rate": 1.537456978286468e-06,
      "loss": 0.6862,
      "step": 26071
    },
    {
      "epoch": 0.7512245721200945,
      "grad_norm": 0.7289283275604248,
      "learning_rate": 1.5371203763334469e-06,
      "loss": 0.6649,
      "step": 26072
    },
    {
      "epoch": 0.7512533855817438,
      "grad_norm": 0.7157189846038818,
      "learning_rate": 1.536783804538644e-06,
      "loss": 0.6773,
      "step": 26073
    },
    {
      "epoch": 0.7512821990433931,
      "grad_norm": 0.7181971073150635,
      "learning_rate": 1.5364472629049943e-06,
      "loss": 0.6933,
      "step": 26074
    },
    {
      "epoch": 0.7513110125050424,
      "grad_norm": 0.740319013595581,
      "learning_rate": 1.5361107514354246e-06,
      "loss": 0.6833,
      "step": 26075
    },
    {
      "epoch": 0.7513398259666917,
      "grad_norm": 0.7116426825523376,
      "learning_rate": 1.5357742701328675e-06,
      "loss": 0.6775,
      "step": 26076
    },
    {
      "epoch": 0.7513686394283409,
      "grad_norm": 0.7189112305641174,
      "learning_rate": 1.5354378190002556e-06,
      "loss": 0.7066,
      "step": 26077
    },
    {
      "epoch": 0.7513974528899902,
      "grad_norm": 0.7245518565177917,
      "learning_rate": 1.535101398040515e-06,
      "loss": 0.6921,
      "step": 26078
    },
    {
      "epoch": 0.7514262663516394,
      "grad_norm": 0.700588047504425,
      "learning_rate": 1.5347650072565795e-06,
      "loss": 0.6871,
      "step": 26079
    },
    {
      "epoch": 0.7514550798132887,
      "grad_norm": 0.7254099249839783,
      "learning_rate": 1.5344286466513753e-06,
      "loss": 0.7007,
      "step": 26080
    },
    {
      "epoch": 0.751483893274938,
      "grad_norm": 0.7114565968513489,
      "learning_rate": 1.5340923162278332e-06,
      "loss": 0.6829,
      "step": 26081
    },
    {
      "epoch": 0.7515127067365873,
      "grad_norm": 0.7095155715942383,
      "learning_rate": 1.533756015988882e-06,
      "loss": 0.6596,
      "step": 26082
    },
    {
      "epoch": 0.7515415201982366,
      "grad_norm": 0.7264817953109741,
      "learning_rate": 1.5334197459374528e-06,
      "loss": 0.6824,
      "step": 26083
    },
    {
      "epoch": 0.7515703336598859,
      "grad_norm": 0.6975746154785156,
      "learning_rate": 1.5330835060764708e-06,
      "loss": 0.6866,
      "step": 26084
    },
    {
      "epoch": 0.7515991471215352,
      "grad_norm": 0.7207280993461609,
      "learning_rate": 1.5327472964088675e-06,
      "loss": 0.7054,
      "step": 26085
    },
    {
      "epoch": 0.7516279605831845,
      "grad_norm": 0.7232341766357422,
      "learning_rate": 1.5324111169375672e-06,
      "loss": 0.6937,
      "step": 26086
    },
    {
      "epoch": 0.7516567740448338,
      "grad_norm": 0.7320263981819153,
      "learning_rate": 1.5320749676655018e-06,
      "loss": 0.6703,
      "step": 26087
    },
    {
      "epoch": 0.7516855875064831,
      "grad_norm": 0.7225409150123596,
      "learning_rate": 1.5317388485955948e-06,
      "loss": 0.7032,
      "step": 26088
    },
    {
      "epoch": 0.7517144009681324,
      "grad_norm": 0.7029889822006226,
      "learning_rate": 1.5314027597307773e-06,
      "loss": 0.6876,
      "step": 26089
    },
    {
      "epoch": 0.7517432144297816,
      "grad_norm": 0.7293941974639893,
      "learning_rate": 1.5310667010739727e-06,
      "loss": 0.67,
      "step": 26090
    },
    {
      "epoch": 0.7517720278914308,
      "grad_norm": 0.7173463106155396,
      "learning_rate": 1.5307306726281096e-06,
      "loss": 0.6983,
      "step": 26091
    },
    {
      "epoch": 0.7518008413530801,
      "grad_norm": 0.7334301471710205,
      "learning_rate": 1.5303946743961162e-06,
      "loss": 0.6962,
      "step": 26092
    },
    {
      "epoch": 0.7518296548147294,
      "grad_norm": 0.7021569609642029,
      "learning_rate": 1.5300587063809146e-06,
      "loss": 0.6907,
      "step": 26093
    },
    {
      "epoch": 0.7518584682763787,
      "grad_norm": 0.7458489537239075,
      "learning_rate": 1.5297227685854333e-06,
      "loss": 0.7171,
      "step": 26094
    },
    {
      "epoch": 0.751887281738028,
      "grad_norm": 0.7098435163497925,
      "learning_rate": 1.5293868610125994e-06,
      "loss": 0.6749,
      "step": 26095
    },
    {
      "epoch": 0.7519160951996773,
      "grad_norm": 0.7299424409866333,
      "learning_rate": 1.529050983665335e-06,
      "loss": 0.6766,
      "step": 26096
    },
    {
      "epoch": 0.7519449086613266,
      "grad_norm": 0.7120885848999023,
      "learning_rate": 1.528715136546567e-06,
      "loss": 0.6983,
      "step": 26097
    },
    {
      "epoch": 0.7519737221229759,
      "grad_norm": 0.717132031917572,
      "learning_rate": 1.528379319659221e-06,
      "loss": 0.7078,
      "step": 26098
    },
    {
      "epoch": 0.7520025355846252,
      "grad_norm": 0.7093655467033386,
      "learning_rate": 1.5280435330062198e-06,
      "loss": 0.6726,
      "step": 26099
    },
    {
      "epoch": 0.7520313490462744,
      "grad_norm": 0.7186593413352966,
      "learning_rate": 1.5277077765904902e-06,
      "loss": 0.6829,
      "step": 26100
    },
    {
      "epoch": 0.7520601625079237,
      "grad_norm": 0.7188811898231506,
      "learning_rate": 1.5273720504149536e-06,
      "loss": 0.6919,
      "step": 26101
    },
    {
      "epoch": 0.752088975969573,
      "grad_norm": 0.7063959836959839,
      "learning_rate": 1.5270363544825361e-06,
      "loss": 0.6957,
      "step": 26102
    },
    {
      "epoch": 0.7521177894312223,
      "grad_norm": 0.7211891412734985,
      "learning_rate": 1.5267006887961588e-06,
      "loss": 0.6869,
      "step": 26103
    },
    {
      "epoch": 0.7521466028928715,
      "grad_norm": 0.7327549457550049,
      "learning_rate": 1.5263650533587482e-06,
      "loss": 0.6823,
      "step": 26104
    },
    {
      "epoch": 0.7521754163545208,
      "grad_norm": 0.7485619187355042,
      "learning_rate": 1.5260294481732224e-06,
      "loss": 0.6919,
      "step": 26105
    },
    {
      "epoch": 0.7522042298161701,
      "grad_norm": 0.7134838104248047,
      "learning_rate": 1.5256938732425108e-06,
      "loss": 0.682,
      "step": 26106
    },
    {
      "epoch": 0.7522330432778194,
      "grad_norm": 0.7211158871650696,
      "learning_rate": 1.5253583285695306e-06,
      "loss": 0.6932,
      "step": 26107
    },
    {
      "epoch": 0.7522618567394687,
      "grad_norm": 0.7298281192779541,
      "learning_rate": 1.5250228141572082e-06,
      "loss": 0.6863,
      "step": 26108
    },
    {
      "epoch": 0.752290670201118,
      "grad_norm": 0.6998457312583923,
      "learning_rate": 1.5246873300084608e-06,
      "loss": 0.6757,
      "step": 26109
    },
    {
      "epoch": 0.7523194836627672,
      "grad_norm": 0.7396429777145386,
      "learning_rate": 1.524351876126215e-06,
      "loss": 0.6857,
      "step": 26110
    },
    {
      "epoch": 0.7523482971244165,
      "grad_norm": 0.7186757922172546,
      "learning_rate": 1.5240164525133883e-06,
      "loss": 0.6823,
      "step": 26111
    },
    {
      "epoch": 0.7523771105860658,
      "grad_norm": 0.727603018283844,
      "learning_rate": 1.5236810591729052e-06,
      "loss": 0.7039,
      "step": 26112
    },
    {
      "epoch": 0.7524059240477151,
      "grad_norm": 0.7257590889930725,
      "learning_rate": 1.5233456961076831e-06,
      "loss": 0.6793,
      "step": 26113
    },
    {
      "epoch": 0.7524347375093644,
      "grad_norm": 0.6864039301872253,
      "learning_rate": 1.523010363320645e-06,
      "loss": 0.6721,
      "step": 26114
    },
    {
      "epoch": 0.7524635509710137,
      "grad_norm": 0.7331482172012329,
      "learning_rate": 1.5226750608147123e-06,
      "loss": 0.6608,
      "step": 26115
    },
    {
      "epoch": 0.752492364432663,
      "grad_norm": 0.7008041739463806,
      "learning_rate": 1.522339788592802e-06,
      "loss": 0.6738,
      "step": 26116
    },
    {
      "epoch": 0.7525211778943123,
      "grad_norm": 0.7237740159034729,
      "learning_rate": 1.5220045466578365e-06,
      "loss": 0.6947,
      "step": 26117
    },
    {
      "epoch": 0.7525499913559615,
      "grad_norm": 0.6940428614616394,
      "learning_rate": 1.5216693350127337e-06,
      "loss": 0.6654,
      "step": 26118
    },
    {
      "epoch": 0.7525788048176107,
      "grad_norm": 0.7060658931732178,
      "learning_rate": 1.521334153660416e-06,
      "loss": 0.6742,
      "step": 26119
    },
    {
      "epoch": 0.75260761827926,
      "grad_norm": 0.7523646950721741,
      "learning_rate": 1.520999002603799e-06,
      "loss": 0.6896,
      "step": 26120
    },
    {
      "epoch": 0.7526364317409093,
      "grad_norm": 0.7017382979393005,
      "learning_rate": 1.5206638818458042e-06,
      "loss": 0.6856,
      "step": 26121
    },
    {
      "epoch": 0.7526652452025586,
      "grad_norm": 0.7262458801269531,
      "learning_rate": 1.5203287913893478e-06,
      "loss": 0.6887,
      "step": 26122
    },
    {
      "epoch": 0.7526940586642079,
      "grad_norm": 0.7153213024139404,
      "learning_rate": 1.5199937312373504e-06,
      "loss": 0.6637,
      "step": 26123
    },
    {
      "epoch": 0.7527228721258572,
      "grad_norm": 0.7037817239761353,
      "learning_rate": 1.5196587013927273e-06,
      "loss": 0.6876,
      "step": 26124
    },
    {
      "epoch": 0.7527516855875065,
      "grad_norm": 0.7258089184761047,
      "learning_rate": 1.5193237018583996e-06,
      "loss": 0.6864,
      "step": 26125
    },
    {
      "epoch": 0.7527804990491558,
      "grad_norm": 0.7071638107299805,
      "learning_rate": 1.5189887326372815e-06,
      "loss": 0.6802,
      "step": 26126
    },
    {
      "epoch": 0.7528093125108051,
      "grad_norm": 0.7015824317932129,
      "learning_rate": 1.5186537937322932e-06,
      "loss": 0.6778,
      "step": 26127
    },
    {
      "epoch": 0.7528381259724544,
      "grad_norm": 0.7061379551887512,
      "learning_rate": 1.5183188851463488e-06,
      "loss": 0.6771,
      "step": 26128
    },
    {
      "epoch": 0.7528669394341037,
      "grad_norm": 0.7249532341957092,
      "learning_rate": 1.5179840068823664e-06,
      "loss": 0.7024,
      "step": 26129
    },
    {
      "epoch": 0.752895752895753,
      "grad_norm": 0.7164601683616638,
      "learning_rate": 1.517649158943263e-06,
      "loss": 0.7004,
      "step": 26130
    },
    {
      "epoch": 0.7529245663574021,
      "grad_norm": 0.7338968515396118,
      "learning_rate": 1.5173143413319553e-06,
      "loss": 0.7187,
      "step": 26131
    },
    {
      "epoch": 0.7529533798190514,
      "grad_norm": 0.7324391603469849,
      "learning_rate": 1.516979554051357e-06,
      "loss": 0.682,
      "step": 26132
    },
    {
      "epoch": 0.7529821932807007,
      "grad_norm": 0.7133488655090332,
      "learning_rate": 1.5166447971043868e-06,
      "loss": 0.6946,
      "step": 26133
    },
    {
      "epoch": 0.75301100674235,
      "grad_norm": 0.7065935134887695,
      "learning_rate": 1.5163100704939565e-06,
      "loss": 0.6782,
      "step": 26134
    },
    {
      "epoch": 0.7530398202039993,
      "grad_norm": 0.72843337059021,
      "learning_rate": 1.5159753742229827e-06,
      "loss": 0.6725,
      "step": 26135
    },
    {
      "epoch": 0.7530686336656486,
      "grad_norm": 0.7375935316085815,
      "learning_rate": 1.5156407082943831e-06,
      "loss": 0.7103,
      "step": 26136
    },
    {
      "epoch": 0.7530974471272979,
      "grad_norm": 0.7018091678619385,
      "learning_rate": 1.5153060727110674e-06,
      "loss": 0.7008,
      "step": 26137
    },
    {
      "epoch": 0.7531262605889472,
      "grad_norm": 0.7307174801826477,
      "learning_rate": 1.5149714674759546e-06,
      "loss": 0.6842,
      "step": 26138
    },
    {
      "epoch": 0.7531550740505965,
      "grad_norm": 0.7069224119186401,
      "learning_rate": 1.5146368925919548e-06,
      "loss": 0.6795,
      "step": 26139
    },
    {
      "epoch": 0.7531838875122457,
      "grad_norm": 0.7197046279907227,
      "learning_rate": 1.5143023480619851e-06,
      "loss": 0.6917,
      "step": 26140
    },
    {
      "epoch": 0.753212700973895,
      "grad_norm": 0.7550519108772278,
      "learning_rate": 1.513967833888954e-06,
      "loss": 0.6899,
      "step": 26141
    },
    {
      "epoch": 0.7532415144355443,
      "grad_norm": 0.714373767375946,
      "learning_rate": 1.5136333500757822e-06,
      "loss": 0.6814,
      "step": 26142
    },
    {
      "epoch": 0.7532703278971936,
      "grad_norm": 0.7305840849876404,
      "learning_rate": 1.513298896625377e-06,
      "loss": 0.692,
      "step": 26143
    },
    {
      "epoch": 0.7532991413588429,
      "grad_norm": 0.7379244565963745,
      "learning_rate": 1.5129644735406545e-06,
      "loss": 0.6693,
      "step": 26144
    },
    {
      "epoch": 0.7533279548204921,
      "grad_norm": 0.7280529141426086,
      "learning_rate": 1.5126300808245237e-06,
      "loss": 0.7,
      "step": 26145
    },
    {
      "epoch": 0.7533567682821414,
      "grad_norm": 0.7090817093849182,
      "learning_rate": 1.512295718479901e-06,
      "loss": 0.6938,
      "step": 26146
    },
    {
      "epoch": 0.7533855817437907,
      "grad_norm": 0.7312054634094238,
      "learning_rate": 1.5119613865096944e-06,
      "loss": 0.682,
      "step": 26147
    },
    {
      "epoch": 0.75341439520544,
      "grad_norm": 0.7309475541114807,
      "learning_rate": 1.5116270849168186e-06,
      "loss": 0.7007,
      "step": 26148
    },
    {
      "epoch": 0.7534432086670892,
      "grad_norm": 0.733086884021759,
      "learning_rate": 1.5112928137041827e-06,
      "loss": 0.692,
      "step": 26149
    },
    {
      "epoch": 0.7534720221287385,
      "grad_norm": 0.7137314677238464,
      "learning_rate": 1.5109585728746984e-06,
      "loss": 0.6626,
      "step": 26150
    },
    {
      "epoch": 0.7535008355903878,
      "grad_norm": 0.7078974843025208,
      "learning_rate": 1.5106243624312794e-06,
      "loss": 0.6831,
      "step": 26151
    },
    {
      "epoch": 0.7535296490520371,
      "grad_norm": 0.7202683687210083,
      "learning_rate": 1.5102901823768324e-06,
      "loss": 0.6688,
      "step": 26152
    },
    {
      "epoch": 0.7535584625136864,
      "grad_norm": 0.7389015555381775,
      "learning_rate": 1.5099560327142693e-06,
      "loss": 0.6865,
      "step": 26153
    },
    {
      "epoch": 0.7535872759753357,
      "grad_norm": 0.737716019153595,
      "learning_rate": 1.5096219134465019e-06,
      "loss": 0.681,
      "step": 26154
    },
    {
      "epoch": 0.753616089436985,
      "grad_norm": 0.7257372736930847,
      "learning_rate": 1.509287824576437e-06,
      "loss": 0.7117,
      "step": 26155
    },
    {
      "epoch": 0.7536449028986343,
      "grad_norm": 0.7093417644500732,
      "learning_rate": 1.5089537661069852e-06,
      "loss": 0.6949,
      "step": 26156
    },
    {
      "epoch": 0.7536737163602836,
      "grad_norm": 0.7260428667068481,
      "learning_rate": 1.508619738041059e-06,
      "loss": 0.6903,
      "step": 26157
    },
    {
      "epoch": 0.7537025298219328,
      "grad_norm": 0.7116117477416992,
      "learning_rate": 1.5082857403815627e-06,
      "loss": 0.6827,
      "step": 26158
    },
    {
      "epoch": 0.753731343283582,
      "grad_norm": 0.7221023440361023,
      "learning_rate": 1.5079517731314086e-06,
      "loss": 0.6785,
      "step": 26159
    },
    {
      "epoch": 0.7537601567452313,
      "grad_norm": 0.7042455077171326,
      "learning_rate": 1.5076178362935023e-06,
      "loss": 0.6943,
      "step": 26160
    },
    {
      "epoch": 0.7537889702068806,
      "grad_norm": 0.7084299921989441,
      "learning_rate": 1.5072839298707547e-06,
      "loss": 0.707,
      "step": 26161
    },
    {
      "epoch": 0.7538177836685299,
      "grad_norm": 0.7394973039627075,
      "learning_rate": 1.5069500538660714e-06,
      "loss": 0.6948,
      "step": 26162
    },
    {
      "epoch": 0.7538465971301792,
      "grad_norm": 0.7206337451934814,
      "learning_rate": 1.5066162082823631e-06,
      "loss": 0.6944,
      "step": 26163
    },
    {
      "epoch": 0.7538754105918285,
      "grad_norm": 0.7022528648376465,
      "learning_rate": 1.506282393122534e-06,
      "loss": 0.6806,
      "step": 26164
    },
    {
      "epoch": 0.7539042240534778,
      "grad_norm": 0.7147716283798218,
      "learning_rate": 1.5059486083894926e-06,
      "loss": 0.6797,
      "step": 26165
    },
    {
      "epoch": 0.7539330375151271,
      "grad_norm": 0.7045276761054993,
      "learning_rate": 1.505614854086146e-06,
      "loss": 0.69,
      "step": 26166
    },
    {
      "epoch": 0.7539618509767764,
      "grad_norm": 0.7081694602966309,
      "learning_rate": 1.505281130215403e-06,
      "loss": 0.6733,
      "step": 26167
    },
    {
      "epoch": 0.7539906644384257,
      "grad_norm": 0.7244886755943298,
      "learning_rate": 1.504947436780166e-06,
      "loss": 0.67,
      "step": 26168
    },
    {
      "epoch": 0.754019477900075,
      "grad_norm": 0.7040635347366333,
      "learning_rate": 1.5046137737833444e-06,
      "loss": 0.6775,
      "step": 26169
    },
    {
      "epoch": 0.7540482913617242,
      "grad_norm": 0.70763099193573,
      "learning_rate": 1.5042801412278413e-06,
      "loss": 0.6699,
      "step": 26170
    },
    {
      "epoch": 0.7540771048233735,
      "grad_norm": 0.7204809784889221,
      "learning_rate": 1.5039465391165653e-06,
      "loss": 0.6918,
      "step": 26171
    },
    {
      "epoch": 0.7541059182850227,
      "grad_norm": 0.73750901222229,
      "learning_rate": 1.5036129674524185e-06,
      "loss": 0.6905,
      "step": 26172
    },
    {
      "epoch": 0.754134731746672,
      "grad_norm": 0.7349961400032043,
      "learning_rate": 1.5032794262383076e-06,
      "loss": 0.6745,
      "step": 26173
    },
    {
      "epoch": 0.7541635452083213,
      "grad_norm": 0.7454478740692139,
      "learning_rate": 1.5029459154771392e-06,
      "loss": 0.6824,
      "step": 26174
    },
    {
      "epoch": 0.7541923586699706,
      "grad_norm": 0.7248722910881042,
      "learning_rate": 1.5026124351718146e-06,
      "loss": 0.6786,
      "step": 26175
    },
    {
      "epoch": 0.7542211721316199,
      "grad_norm": 0.6999390125274658,
      "learning_rate": 1.5022789853252413e-06,
      "loss": 0.696,
      "step": 26176
    },
    {
      "epoch": 0.7542499855932692,
      "grad_norm": 0.7279070019721985,
      "learning_rate": 1.5019455659403181e-06,
      "loss": 0.6762,
      "step": 26177
    },
    {
      "epoch": 0.7542787990549185,
      "grad_norm": 0.7426008582115173,
      "learning_rate": 1.5016121770199554e-06,
      "loss": 0.6911,
      "step": 26178
    },
    {
      "epoch": 0.7543076125165677,
      "grad_norm": 0.7199581265449524,
      "learning_rate": 1.5012788185670523e-06,
      "loss": 0.6867,
      "step": 26179
    },
    {
      "epoch": 0.754336425978217,
      "grad_norm": 0.705008327960968,
      "learning_rate": 1.5009454905845145e-06,
      "loss": 0.6733,
      "step": 26180
    },
    {
      "epoch": 0.7543652394398663,
      "grad_norm": 0.7347878217697144,
      "learning_rate": 1.500612193075242e-06,
      "loss": 0.6852,
      "step": 26181
    },
    {
      "epoch": 0.7543940529015156,
      "grad_norm": 0.7021045684814453,
      "learning_rate": 1.5002789260421412e-06,
      "loss": 0.6939,
      "step": 26182
    },
    {
      "epoch": 0.7544228663631649,
      "grad_norm": 0.7277674078941345,
      "learning_rate": 1.4999456894881104e-06,
      "loss": 0.7062,
      "step": 26183
    },
    {
      "epoch": 0.7544516798248142,
      "grad_norm": 0.7182614803314209,
      "learning_rate": 1.4996124834160559e-06,
      "loss": 0.6592,
      "step": 26184
    },
    {
      "epoch": 0.7544804932864634,
      "grad_norm": 0.7359510660171509,
      "learning_rate": 1.499279307828876e-06,
      "loss": 0.6925,
      "step": 26185
    },
    {
      "epoch": 0.7545093067481127,
      "grad_norm": 0.7163605690002441,
      "learning_rate": 1.4989461627294755e-06,
      "loss": 0.6762,
      "step": 26186
    },
    {
      "epoch": 0.754538120209762,
      "grad_norm": 0.7054329514503479,
      "learning_rate": 1.4986130481207522e-06,
      "loss": 0.6782,
      "step": 26187
    },
    {
      "epoch": 0.7545669336714113,
      "grad_norm": 0.7088040709495544,
      "learning_rate": 1.4982799640056096e-06,
      "loss": 0.6838,
      "step": 26188
    },
    {
      "epoch": 0.7545957471330605,
      "grad_norm": 0.721401572227478,
      "learning_rate": 1.4979469103869475e-06,
      "loss": 0.6925,
      "step": 26189
    },
    {
      "epoch": 0.7546245605947098,
      "grad_norm": 0.7147906422615051,
      "learning_rate": 1.4976138872676687e-06,
      "loss": 0.674,
      "step": 26190
    },
    {
      "epoch": 0.7546533740563591,
      "grad_norm": 0.7177363038063049,
      "learning_rate": 1.4972808946506706e-06,
      "loss": 0.6924,
      "step": 26191
    },
    {
      "epoch": 0.7546821875180084,
      "grad_norm": 0.7219609618186951,
      "learning_rate": 1.496947932538856e-06,
      "loss": 0.6659,
      "step": 26192
    },
    {
      "epoch": 0.7547110009796577,
      "grad_norm": 0.7215374708175659,
      "learning_rate": 1.4966150009351216e-06,
      "loss": 0.6948,
      "step": 26193
    },
    {
      "epoch": 0.754739814441307,
      "grad_norm": 0.7195445895195007,
      "learning_rate": 1.4962820998423683e-06,
      "loss": 0.6828,
      "step": 26194
    },
    {
      "epoch": 0.7547686279029563,
      "grad_norm": 0.7138930559158325,
      "learning_rate": 1.495949229263497e-06,
      "loss": 0.6944,
      "step": 26195
    },
    {
      "epoch": 0.7547974413646056,
      "grad_norm": 0.7362907528877258,
      "learning_rate": 1.4956163892014037e-06,
      "loss": 0.6996,
      "step": 26196
    },
    {
      "epoch": 0.7548262548262549,
      "grad_norm": 0.7448341250419617,
      "learning_rate": 1.4952835796589903e-06,
      "loss": 0.6737,
      "step": 26197
    },
    {
      "epoch": 0.7548550682879042,
      "grad_norm": 0.7225752472877502,
      "learning_rate": 1.4949508006391517e-06,
      "loss": 0.6942,
      "step": 26198
    },
    {
      "epoch": 0.7548838817495533,
      "grad_norm": 0.725077211856842,
      "learning_rate": 1.4946180521447896e-06,
      "loss": 0.6781,
      "step": 26199
    },
    {
      "epoch": 0.7549126952112026,
      "grad_norm": 0.7341172695159912,
      "learning_rate": 1.4942853341787983e-06,
      "loss": 0.6957,
      "step": 26200
    },
    {
      "epoch": 0.7549415086728519,
      "grad_norm": 0.708411693572998,
      "learning_rate": 1.4939526467440773e-06,
      "loss": 0.7021,
      "step": 26201
    },
    {
      "epoch": 0.7549703221345012,
      "grad_norm": 0.7226793766021729,
      "learning_rate": 1.4936199898435239e-06,
      "loss": 0.6647,
      "step": 26202
    },
    {
      "epoch": 0.7549991355961505,
      "grad_norm": 0.7219664454460144,
      "learning_rate": 1.493287363480037e-06,
      "loss": 0.6904,
      "step": 26203
    },
    {
      "epoch": 0.7550279490577998,
      "grad_norm": 0.702745258808136,
      "learning_rate": 1.4929547676565103e-06,
      "loss": 0.697,
      "step": 26204
    },
    {
      "epoch": 0.7550567625194491,
      "grad_norm": 0.7028239965438843,
      "learning_rate": 1.4926222023758435e-06,
      "loss": 0.6509,
      "step": 26205
    },
    {
      "epoch": 0.7550855759810984,
      "grad_norm": 0.7193631529808044,
      "learning_rate": 1.4922896676409293e-06,
      "loss": 0.6972,
      "step": 26206
    },
    {
      "epoch": 0.7551143894427477,
      "grad_norm": 0.7383573651313782,
      "learning_rate": 1.4919571634546676e-06,
      "loss": 0.6823,
      "step": 26207
    },
    {
      "epoch": 0.755143202904397,
      "grad_norm": 0.7237043976783752,
      "learning_rate": 1.4916246898199505e-06,
      "loss": 0.6892,
      "step": 26208
    },
    {
      "epoch": 0.7551720163660463,
      "grad_norm": 0.7140964865684509,
      "learning_rate": 1.4912922467396767e-06,
      "loss": 0.6868,
      "step": 26209
    },
    {
      "epoch": 0.7552008298276955,
      "grad_norm": 0.708591639995575,
      "learning_rate": 1.4909598342167386e-06,
      "loss": 0.6877,
      "step": 26210
    },
    {
      "epoch": 0.7552296432893448,
      "grad_norm": 0.7244853973388672,
      "learning_rate": 1.4906274522540326e-06,
      "loss": 0.6761,
      "step": 26211
    },
    {
      "epoch": 0.755258456750994,
      "grad_norm": 0.7332562208175659,
      "learning_rate": 1.4902951008544548e-06,
      "loss": 0.6821,
      "step": 26212
    },
    {
      "epoch": 0.7552872702126433,
      "grad_norm": 0.7236791849136353,
      "learning_rate": 1.4899627800208972e-06,
      "loss": 0.686,
      "step": 26213
    },
    {
      "epoch": 0.7553160836742926,
      "grad_norm": 0.7252430319786072,
      "learning_rate": 1.4896304897562542e-06,
      "loss": 0.6867,
      "step": 26214
    },
    {
      "epoch": 0.7553448971359419,
      "grad_norm": 0.729257345199585,
      "learning_rate": 1.4892982300634212e-06,
      "loss": 0.6877,
      "step": 26215
    },
    {
      "epoch": 0.7553737105975912,
      "grad_norm": 0.7150086760520935,
      "learning_rate": 1.4889660009452921e-06,
      "loss": 0.7084,
      "step": 26216
    },
    {
      "epoch": 0.7554025240592405,
      "grad_norm": 0.7429845929145813,
      "learning_rate": 1.4886338024047585e-06,
      "loss": 0.7093,
      "step": 26217
    },
    {
      "epoch": 0.7554313375208898,
      "grad_norm": 0.7076511383056641,
      "learning_rate": 1.488301634444716e-06,
      "loss": 0.7042,
      "step": 26218
    },
    {
      "epoch": 0.755460150982539,
      "grad_norm": 0.741095244884491,
      "learning_rate": 1.4879694970680536e-06,
      "loss": 0.7038,
      "step": 26219
    },
    {
      "epoch": 0.7554889644441883,
      "grad_norm": 0.7275850772857666,
      "learning_rate": 1.4876373902776687e-06,
      "loss": 0.6884,
      "step": 26220
    },
    {
      "epoch": 0.7555177779058376,
      "grad_norm": 0.6872200965881348,
      "learning_rate": 1.4873053140764487e-06,
      "loss": 0.6586,
      "step": 26221
    },
    {
      "epoch": 0.7555465913674869,
      "grad_norm": 0.7285813689231873,
      "learning_rate": 1.4869732684672894e-06,
      "loss": 0.6881,
      "step": 26222
    },
    {
      "epoch": 0.7555754048291362,
      "grad_norm": 0.7292302846908569,
      "learning_rate": 1.48664125345308e-06,
      "loss": 0.7182,
      "step": 26223
    },
    {
      "epoch": 0.7556042182907855,
      "grad_norm": 0.7114002108573914,
      "learning_rate": 1.4863092690367148e-06,
      "loss": 0.6846,
      "step": 26224
    },
    {
      "epoch": 0.7556330317524348,
      "grad_norm": 0.715243399143219,
      "learning_rate": 1.4859773152210805e-06,
      "loss": 0.6706,
      "step": 26225
    },
    {
      "epoch": 0.755661845214084,
      "grad_norm": 0.7233961224555969,
      "learning_rate": 1.4856453920090742e-06,
      "loss": 0.7118,
      "step": 26226
    },
    {
      "epoch": 0.7556906586757333,
      "grad_norm": 0.7112762331962585,
      "learning_rate": 1.4853134994035817e-06,
      "loss": 0.6764,
      "step": 26227
    },
    {
      "epoch": 0.7557194721373826,
      "grad_norm": 0.7178820967674255,
      "learning_rate": 1.4849816374074965e-06,
      "loss": 0.7067,
      "step": 26228
    },
    {
      "epoch": 0.7557482855990318,
      "grad_norm": 0.7175015807151794,
      "learning_rate": 1.4846498060237063e-06,
      "loss": 0.7004,
      "step": 26229
    },
    {
      "epoch": 0.7557770990606811,
      "grad_norm": 0.7244692444801331,
      "learning_rate": 1.4843180052551038e-06,
      "loss": 0.6839,
      "step": 26230
    },
    {
      "epoch": 0.7558059125223304,
      "grad_norm": 0.7101678252220154,
      "learning_rate": 1.4839862351045754e-06,
      "loss": 0.6848,
      "step": 26231
    },
    {
      "epoch": 0.7558347259839797,
      "grad_norm": 0.723141610622406,
      "learning_rate": 1.483654495575012e-06,
      "loss": 0.6774,
      "step": 26232
    },
    {
      "epoch": 0.755863539445629,
      "grad_norm": 0.7539788484573364,
      "learning_rate": 1.4833227866693046e-06,
      "loss": 0.6978,
      "step": 26233
    },
    {
      "epoch": 0.7558923529072783,
      "grad_norm": 0.7303386926651001,
      "learning_rate": 1.4829911083903386e-06,
      "loss": 0.6671,
      "step": 26234
    },
    {
      "epoch": 0.7559211663689276,
      "grad_norm": 0.7341259121894836,
      "learning_rate": 1.4826594607410061e-06,
      "loss": 0.69,
      "step": 26235
    },
    {
      "epoch": 0.7559499798305769,
      "grad_norm": 0.7330235242843628,
      "learning_rate": 1.4823278437241923e-06,
      "loss": 0.6739,
      "step": 26236
    },
    {
      "epoch": 0.7559787932922262,
      "grad_norm": 0.7331784963607788,
      "learning_rate": 1.4819962573427864e-06,
      "loss": 0.6907,
      "step": 26237
    },
    {
      "epoch": 0.7560076067538755,
      "grad_norm": 0.7495669722557068,
      "learning_rate": 1.4816647015996766e-06,
      "loss": 0.6969,
      "step": 26238
    },
    {
      "epoch": 0.7560364202155246,
      "grad_norm": 0.7085393071174622,
      "learning_rate": 1.4813331764977513e-06,
      "loss": 0.6826,
      "step": 26239
    },
    {
      "epoch": 0.7560652336771739,
      "grad_norm": 0.7411060929298401,
      "learning_rate": 1.4810016820398953e-06,
      "loss": 0.6907,
      "step": 26240
    },
    {
      "epoch": 0.7560940471388232,
      "grad_norm": 0.7167351841926575,
      "learning_rate": 1.4806702182289984e-06,
      "loss": 0.7226,
      "step": 26241
    },
    {
      "epoch": 0.7561228606004725,
      "grad_norm": 0.7040646076202393,
      "learning_rate": 1.4803387850679446e-06,
      "loss": 0.6677,
      "step": 26242
    },
    {
      "epoch": 0.7561516740621218,
      "grad_norm": 0.7055210471153259,
      "learning_rate": 1.480007382559623e-06,
      "loss": 0.6875,
      "step": 26243
    },
    {
      "epoch": 0.7561804875237711,
      "grad_norm": 0.7260403633117676,
      "learning_rate": 1.4796760107069163e-06,
      "loss": 0.7129,
      "step": 26244
    },
    {
      "epoch": 0.7562093009854204,
      "grad_norm": 0.7248584628105164,
      "learning_rate": 1.479344669512715e-06,
      "loss": 0.6642,
      "step": 26245
    },
    {
      "epoch": 0.7562381144470697,
      "grad_norm": 0.7333604097366333,
      "learning_rate": 1.4790133589799e-06,
      "loss": 0.7283,
      "step": 26246
    },
    {
      "epoch": 0.756266927908719,
      "grad_norm": 0.7076057195663452,
      "learning_rate": 1.478682079111361e-06,
      "loss": 0.6665,
      "step": 26247
    },
    {
      "epoch": 0.7562957413703683,
      "grad_norm": 0.7169261574745178,
      "learning_rate": 1.478350829909979e-06,
      "loss": 0.6877,
      "step": 26248
    },
    {
      "epoch": 0.7563245548320175,
      "grad_norm": 0.6930731534957886,
      "learning_rate": 1.4780196113786417e-06,
      "loss": 0.6808,
      "step": 26249
    },
    {
      "epoch": 0.7563533682936668,
      "grad_norm": 0.693867564201355,
      "learning_rate": 1.477688423520232e-06,
      "loss": 0.6886,
      "step": 26250
    },
    {
      "epoch": 0.7563821817553161,
      "grad_norm": 0.7893425226211548,
      "learning_rate": 1.4773572663376373e-06,
      "loss": 0.6947,
      "step": 26251
    },
    {
      "epoch": 0.7564109952169654,
      "grad_norm": 0.741621196269989,
      "learning_rate": 1.4770261398337376e-06,
      "loss": 0.6812,
      "step": 26252
    },
    {
      "epoch": 0.7564398086786146,
      "grad_norm": 0.7367790937423706,
      "learning_rate": 1.4766950440114186e-06,
      "loss": 0.6799,
      "step": 26253
    },
    {
      "epoch": 0.7564686221402639,
      "grad_norm": 0.7166473269462585,
      "learning_rate": 1.4763639788735651e-06,
      "loss": 0.694,
      "step": 26254
    },
    {
      "epoch": 0.7564974356019132,
      "grad_norm": 0.7548424601554871,
      "learning_rate": 1.4760329444230576e-06,
      "loss": 0.6874,
      "step": 26255
    },
    {
      "epoch": 0.7565262490635625,
      "grad_norm": 0.7430239915847778,
      "learning_rate": 1.475701940662782e-06,
      "loss": 0.7056,
      "step": 26256
    },
    {
      "epoch": 0.7565550625252118,
      "grad_norm": 0.6966676712036133,
      "learning_rate": 1.475370967595618e-06,
      "loss": 0.6689,
      "step": 26257
    },
    {
      "epoch": 0.756583875986861,
      "grad_norm": 0.7269933819770813,
      "learning_rate": 1.4750400252244511e-06,
      "loss": 0.6934,
      "step": 26258
    },
    {
      "epoch": 0.7566126894485103,
      "grad_norm": 0.7323204874992371,
      "learning_rate": 1.4747091135521602e-06,
      "loss": 0.6993,
      "step": 26259
    },
    {
      "epoch": 0.7566415029101596,
      "grad_norm": 0.727111279964447,
      "learning_rate": 1.4743782325816309e-06,
      "loss": 0.6948,
      "step": 26260
    },
    {
      "epoch": 0.7566703163718089,
      "grad_norm": 0.7162575125694275,
      "learning_rate": 1.4740473823157393e-06,
      "loss": 0.67,
      "step": 26261
    },
    {
      "epoch": 0.7566991298334582,
      "grad_norm": 0.7400985360145569,
      "learning_rate": 1.4737165627573736e-06,
      "loss": 0.6804,
      "step": 26262
    },
    {
      "epoch": 0.7567279432951075,
      "grad_norm": 0.7450959086418152,
      "learning_rate": 1.4733857739094105e-06,
      "loss": 0.6967,
      "step": 26263
    },
    {
      "epoch": 0.7567567567567568,
      "grad_norm": 0.7179782390594482,
      "learning_rate": 1.473055015774733e-06,
      "loss": 0.6995,
      "step": 26264
    },
    {
      "epoch": 0.7567855702184061,
      "grad_norm": 0.7143447995185852,
      "learning_rate": 1.4727242883562192e-06,
      "loss": 0.7002,
      "step": 26265
    },
    {
      "epoch": 0.7568143836800553,
      "grad_norm": 0.7365439534187317,
      "learning_rate": 1.4723935916567523e-06,
      "loss": 0.6925,
      "step": 26266
    },
    {
      "epoch": 0.7568431971417046,
      "grad_norm": 0.7225685119628906,
      "learning_rate": 1.4720629256792096e-06,
      "loss": 0.7158,
      "step": 26267
    },
    {
      "epoch": 0.7568720106033539,
      "grad_norm": 0.7179908752441406,
      "learning_rate": 1.4717322904264735e-06,
      "loss": 0.6772,
      "step": 26268
    },
    {
      "epoch": 0.7569008240650031,
      "grad_norm": 0.7510362267494202,
      "learning_rate": 1.4714016859014208e-06,
      "loss": 0.7029,
      "step": 26269
    },
    {
      "epoch": 0.7569296375266524,
      "grad_norm": 0.708513617515564,
      "learning_rate": 1.471071112106932e-06,
      "loss": 0.6733,
      "step": 26270
    },
    {
      "epoch": 0.7569584509883017,
      "grad_norm": 0.711050271987915,
      "learning_rate": 1.470740569045888e-06,
      "loss": 0.6754,
      "step": 26271
    },
    {
      "epoch": 0.756987264449951,
      "grad_norm": 0.7279131412506104,
      "learning_rate": 1.4704100567211638e-06,
      "loss": 0.6924,
      "step": 26272
    },
    {
      "epoch": 0.7570160779116003,
      "grad_norm": 0.7501229047775269,
      "learning_rate": 1.47007957513564e-06,
      "loss": 0.7134,
      "step": 26273
    },
    {
      "epoch": 0.7570448913732496,
      "grad_norm": 0.7189348936080933,
      "learning_rate": 1.4697491242921942e-06,
      "loss": 0.6933,
      "step": 26274
    },
    {
      "epoch": 0.7570737048348989,
      "grad_norm": 0.7254137992858887,
      "learning_rate": 1.4694187041937064e-06,
      "loss": 0.6785,
      "step": 26275
    },
    {
      "epoch": 0.7571025182965482,
      "grad_norm": 0.7218815684318542,
      "learning_rate": 1.4690883148430506e-06,
      "loss": 0.6911,
      "step": 26276
    },
    {
      "epoch": 0.7571313317581975,
      "grad_norm": 0.702593982219696,
      "learning_rate": 1.4687579562431075e-06,
      "loss": 0.6782,
      "step": 26277
    },
    {
      "epoch": 0.7571601452198468,
      "grad_norm": 0.6953024864196777,
      "learning_rate": 1.4684276283967513e-06,
      "loss": 0.6864,
      "step": 26278
    },
    {
      "epoch": 0.757188958681496,
      "grad_norm": 0.7147569060325623,
      "learning_rate": 1.4680973313068613e-06,
      "loss": 0.6744,
      "step": 26279
    },
    {
      "epoch": 0.7572177721431452,
      "grad_norm": 0.7087500691413879,
      "learning_rate": 1.467767064976312e-06,
      "loss": 0.6898,
      "step": 26280
    },
    {
      "epoch": 0.7572465856047945,
      "grad_norm": 0.7220697999000549,
      "learning_rate": 1.4674368294079817e-06,
      "loss": 0.6807,
      "step": 26281
    },
    {
      "epoch": 0.7572753990664438,
      "grad_norm": 0.7245514392852783,
      "learning_rate": 1.4671066246047438e-06,
      "loss": 0.6735,
      "step": 26282
    },
    {
      "epoch": 0.7573042125280931,
      "grad_norm": 0.7070509791374207,
      "learning_rate": 1.4667764505694776e-06,
      "loss": 0.6718,
      "step": 26283
    },
    {
      "epoch": 0.7573330259897424,
      "grad_norm": 0.7145278453826904,
      "learning_rate": 1.4664463073050545e-06,
      "loss": 0.6623,
      "step": 26284
    },
    {
      "epoch": 0.7573618394513917,
      "grad_norm": 0.722137451171875,
      "learning_rate": 1.466116194814352e-06,
      "loss": 0.6823,
      "step": 26285
    },
    {
      "epoch": 0.757390652913041,
      "grad_norm": 0.7237882018089294,
      "learning_rate": 1.4657861131002448e-06,
      "loss": 0.6984,
      "step": 26286
    },
    {
      "epoch": 0.7574194663746903,
      "grad_norm": 0.7025514841079712,
      "learning_rate": 1.4654560621656094e-06,
      "loss": 0.6779,
      "step": 26287
    },
    {
      "epoch": 0.7574482798363396,
      "grad_norm": 0.7081069946289062,
      "learning_rate": 1.4651260420133168e-06,
      "loss": 0.6684,
      "step": 26288
    },
    {
      "epoch": 0.7574770932979888,
      "grad_norm": 0.7164566516876221,
      "learning_rate": 1.4647960526462446e-06,
      "loss": 0.6879,
      "step": 26289
    },
    {
      "epoch": 0.7575059067596381,
      "grad_norm": 0.7136147618293762,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.6884,
      "step": 26290
    },
    {
      "epoch": 0.7575347202212874,
      "grad_norm": 0.7365980744361877,
      "learning_rate": 1.4641361662792475e-06,
      "loss": 0.6766,
      "step": 26291
    },
    {
      "epoch": 0.7575635336829367,
      "grad_norm": 0.7259485125541687,
      "learning_rate": 1.463806269285073e-06,
      "loss": 0.6837,
      "step": 26292
    },
    {
      "epoch": 0.7575923471445859,
      "grad_norm": 0.7398556470870972,
      "learning_rate": 1.4634764030876091e-06,
      "loss": 0.6937,
      "step": 26293
    },
    {
      "epoch": 0.7576211606062352,
      "grad_norm": 0.7469826936721802,
      "learning_rate": 1.4631465676897326e-06,
      "loss": 0.6735,
      "step": 26294
    },
    {
      "epoch": 0.7576499740678845,
      "grad_norm": 0.6955668330192566,
      "learning_rate": 1.462816763094312e-06,
      "loss": 0.6984,
      "step": 26295
    },
    {
      "epoch": 0.7576787875295338,
      "grad_norm": 0.7335454225540161,
      "learning_rate": 1.4624869893042222e-06,
      "loss": 0.7155,
      "step": 26296
    },
    {
      "epoch": 0.7577076009911831,
      "grad_norm": 0.7374691963195801,
      "learning_rate": 1.4621572463223338e-06,
      "loss": 0.6856,
      "step": 26297
    },
    {
      "epoch": 0.7577364144528324,
      "grad_norm": 0.7052094340324402,
      "learning_rate": 1.461827534151521e-06,
      "loss": 0.6703,
      "step": 26298
    },
    {
      "epoch": 0.7577652279144816,
      "grad_norm": 0.7370524406433105,
      "learning_rate": 1.4614978527946515e-06,
      "loss": 0.7053,
      "step": 26299
    },
    {
      "epoch": 0.7577940413761309,
      "grad_norm": 0.7083638906478882,
      "learning_rate": 1.4611682022546009e-06,
      "loss": 0.6889,
      "step": 26300
    },
    {
      "epoch": 0.7578228548377802,
      "grad_norm": 0.726813554763794,
      "learning_rate": 1.4608385825342354e-06,
      "loss": 0.6645,
      "step": 26301
    },
    {
      "epoch": 0.7578516682994295,
      "grad_norm": 0.7243070006370544,
      "learning_rate": 1.4605089936364298e-06,
      "loss": 0.6635,
      "step": 26302
    },
    {
      "epoch": 0.7578804817610788,
      "grad_norm": 0.7119633555412292,
      "learning_rate": 1.4601794355640514e-06,
      "loss": 0.6668,
      "step": 26303
    },
    {
      "epoch": 0.7579092952227281,
      "grad_norm": 0.7073648571968079,
      "learning_rate": 1.459849908319973e-06,
      "loss": 0.6986,
      "step": 26304
    },
    {
      "epoch": 0.7579381086843774,
      "grad_norm": 0.7274613380432129,
      "learning_rate": 1.459520411907061e-06,
      "loss": 0.6884,
      "step": 26305
    },
    {
      "epoch": 0.7579669221460267,
      "grad_norm": 0.7065785527229309,
      "learning_rate": 1.4591909463281894e-06,
      "loss": 0.6896,
      "step": 26306
    },
    {
      "epoch": 0.7579957356076759,
      "grad_norm": 0.712543785572052,
      "learning_rate": 1.4588615115862232e-06,
      "loss": 0.6682,
      "step": 26307
    },
    {
      "epoch": 0.7580245490693251,
      "grad_norm": 0.7194118499755859,
      "learning_rate": 1.458532107684033e-06,
      "loss": 0.6722,
      "step": 26308
    },
    {
      "epoch": 0.7580533625309744,
      "grad_norm": 0.7252872586250305,
      "learning_rate": 1.4582027346244886e-06,
      "loss": 0.686,
      "step": 26309
    },
    {
      "epoch": 0.7580821759926237,
      "grad_norm": 0.7343339920043945,
      "learning_rate": 1.4578733924104593e-06,
      "loss": 0.6876,
      "step": 26310
    },
    {
      "epoch": 0.758110989454273,
      "grad_norm": 0.7182785868644714,
      "learning_rate": 1.4575440810448099e-06,
      "loss": 0.6639,
      "step": 26311
    },
    {
      "epoch": 0.7581398029159223,
      "grad_norm": 0.7262239456176758,
      "learning_rate": 1.457214800530411e-06,
      "loss": 0.6824,
      "step": 26312
    },
    {
      "epoch": 0.7581686163775716,
      "grad_norm": 0.7187032699584961,
      "learning_rate": 1.4568855508701308e-06,
      "loss": 0.7061,
      "step": 26313
    },
    {
      "epoch": 0.7581974298392209,
      "grad_norm": 0.7185451984405518,
      "learning_rate": 1.4565563320668347e-06,
      "loss": 0.6793,
      "step": 26314
    },
    {
      "epoch": 0.7582262433008702,
      "grad_norm": 0.7039725184440613,
      "learning_rate": 1.4562271441233917e-06,
      "loss": 0.6918,
      "step": 26315
    },
    {
      "epoch": 0.7582550567625195,
      "grad_norm": 0.7305907011032104,
      "learning_rate": 1.4558979870426664e-06,
      "loss": 0.686,
      "step": 26316
    },
    {
      "epoch": 0.7582838702241688,
      "grad_norm": 0.7119548916816711,
      "learning_rate": 1.4555688608275287e-06,
      "loss": 0.6796,
      "step": 26317
    },
    {
      "epoch": 0.7583126836858181,
      "grad_norm": 0.71600341796875,
      "learning_rate": 1.455239765480841e-06,
      "loss": 0.686,
      "step": 26318
    },
    {
      "epoch": 0.7583414971474673,
      "grad_norm": 0.7254437804222107,
      "learning_rate": 1.4549107010054735e-06,
      "loss": 0.6704,
      "step": 26319
    },
    {
      "epoch": 0.7583703106091165,
      "grad_norm": 0.7211987376213074,
      "learning_rate": 1.4545816674042868e-06,
      "loss": 0.7004,
      "step": 26320
    },
    {
      "epoch": 0.7583991240707658,
      "grad_norm": 0.7160379886627197,
      "learning_rate": 1.4542526646801524e-06,
      "loss": 0.69,
      "step": 26321
    },
    {
      "epoch": 0.7584279375324151,
      "grad_norm": 0.7265369296073914,
      "learning_rate": 1.4539236928359319e-06,
      "loss": 0.7053,
      "step": 26322
    },
    {
      "epoch": 0.7584567509940644,
      "grad_norm": 0.7194201350212097,
      "learning_rate": 1.4535947518744925e-06,
      "loss": 0.6711,
      "step": 26323
    },
    {
      "epoch": 0.7584855644557137,
      "grad_norm": 0.7151232361793518,
      "learning_rate": 1.4532658417986962e-06,
      "loss": 0.7173,
      "step": 26324
    },
    {
      "epoch": 0.758514377917363,
      "grad_norm": 0.7371615171432495,
      "learning_rate": 1.4529369626114103e-06,
      "loss": 0.6896,
      "step": 26325
    },
    {
      "epoch": 0.7585431913790123,
      "grad_norm": 0.7409868836402893,
      "learning_rate": 1.4526081143154963e-06,
      "loss": 0.7053,
      "step": 26326
    },
    {
      "epoch": 0.7585720048406616,
      "grad_norm": 0.7265288829803467,
      "learning_rate": 1.452279296913821e-06,
      "loss": 0.7014,
      "step": 26327
    },
    {
      "epoch": 0.7586008183023109,
      "grad_norm": 0.7062795758247375,
      "learning_rate": 1.451950510409245e-06,
      "loss": 0.6931,
      "step": 26328
    },
    {
      "epoch": 0.7586296317639601,
      "grad_norm": 0.7150965929031372,
      "learning_rate": 1.4516217548046336e-06,
      "loss": 0.6793,
      "step": 26329
    },
    {
      "epoch": 0.7586584452256094,
      "grad_norm": 0.7334873676300049,
      "learning_rate": 1.451293030102851e-06,
      "loss": 0.7183,
      "step": 26330
    },
    {
      "epoch": 0.7586872586872587,
      "grad_norm": 0.7312463521957397,
      "learning_rate": 1.4509643363067572e-06,
      "loss": 0.6728,
      "step": 26331
    },
    {
      "epoch": 0.758716072148908,
      "grad_norm": 0.7059553265571594,
      "learning_rate": 1.4506356734192161e-06,
      "loss": 0.6766,
      "step": 26332
    },
    {
      "epoch": 0.7587448856105573,
      "grad_norm": 0.7341227531433105,
      "learning_rate": 1.4503070414430903e-06,
      "loss": 0.7077,
      "step": 26333
    },
    {
      "epoch": 0.7587736990722065,
      "grad_norm": 0.7443519830703735,
      "learning_rate": 1.4499784403812434e-06,
      "loss": 0.6973,
      "step": 26334
    },
    {
      "epoch": 0.7588025125338558,
      "grad_norm": 0.7193964123725891,
      "learning_rate": 1.449649870236534e-06,
      "loss": 0.6875,
      "step": 26335
    },
    {
      "epoch": 0.7588313259955051,
      "grad_norm": 0.707537829875946,
      "learning_rate": 1.4493213310118271e-06,
      "loss": 0.6789,
      "step": 26336
    },
    {
      "epoch": 0.7588601394571544,
      "grad_norm": 0.7206332087516785,
      "learning_rate": 1.4489928227099799e-06,
      "loss": 0.6877,
      "step": 26337
    },
    {
      "epoch": 0.7588889529188036,
      "grad_norm": 0.7099558711051941,
      "learning_rate": 1.4486643453338573e-06,
      "loss": 0.6707,
      "step": 26338
    },
    {
      "epoch": 0.7589177663804529,
      "grad_norm": 0.7470532655715942,
      "learning_rate": 1.4483358988863166e-06,
      "loss": 0.7166,
      "step": 26339
    },
    {
      "epoch": 0.7589465798421022,
      "grad_norm": 0.7154109477996826,
      "learning_rate": 1.4480074833702219e-06,
      "loss": 0.6925,
      "step": 26340
    },
    {
      "epoch": 0.7589753933037515,
      "grad_norm": 0.7357820272445679,
      "learning_rate": 1.4476790987884288e-06,
      "loss": 0.6826,
      "step": 26341
    },
    {
      "epoch": 0.7590042067654008,
      "grad_norm": 0.7234442234039307,
      "learning_rate": 1.447350745143802e-06,
      "loss": 0.7149,
      "step": 26342
    },
    {
      "epoch": 0.7590330202270501,
      "grad_norm": 0.719828188419342,
      "learning_rate": 1.447022422439197e-06,
      "loss": 0.708,
      "step": 26343
    },
    {
      "epoch": 0.7590618336886994,
      "grad_norm": 0.7265447974205017,
      "learning_rate": 1.446694130677475e-06,
      "loss": 0.7077,
      "step": 26344
    },
    {
      "epoch": 0.7590906471503487,
      "grad_norm": 0.7405640482902527,
      "learning_rate": 1.4463658698614958e-06,
      "loss": 0.7117,
      "step": 26345
    },
    {
      "epoch": 0.759119460611998,
      "grad_norm": 0.7117171287536621,
      "learning_rate": 1.4460376399941184e-06,
      "loss": 0.7098,
      "step": 26346
    },
    {
      "epoch": 0.7591482740736472,
      "grad_norm": 0.7053387761116028,
      "learning_rate": 1.4457094410781992e-06,
      "loss": 0.6746,
      "step": 26347
    },
    {
      "epoch": 0.7591770875352964,
      "grad_norm": 0.7269791960716248,
      "learning_rate": 1.4453812731165989e-06,
      "loss": 0.6927,
      "step": 26348
    },
    {
      "epoch": 0.7592059009969457,
      "grad_norm": 0.7230656743049622,
      "learning_rate": 1.4450531361121734e-06,
      "loss": 0.7083,
      "step": 26349
    },
    {
      "epoch": 0.759234714458595,
      "grad_norm": 0.6972666382789612,
      "learning_rate": 1.4447250300677811e-06,
      "loss": 0.6685,
      "step": 26350
    },
    {
      "epoch": 0.7592635279202443,
      "grad_norm": 0.7320264577865601,
      "learning_rate": 1.4443969549862812e-06,
      "loss": 0.6963,
      "step": 26351
    },
    {
      "epoch": 0.7592923413818936,
      "grad_norm": 0.7328753471374512,
      "learning_rate": 1.4440689108705285e-06,
      "loss": 0.7089,
      "step": 26352
    },
    {
      "epoch": 0.7593211548435429,
      "grad_norm": 0.7057414054870605,
      "learning_rate": 1.4437408977233825e-06,
      "loss": 0.6799,
      "step": 26353
    },
    {
      "epoch": 0.7593499683051922,
      "grad_norm": 0.739238977432251,
      "learning_rate": 1.4434129155476961e-06,
      "loss": 0.6972,
      "step": 26354
    },
    {
      "epoch": 0.7593787817668415,
      "grad_norm": 0.7167029976844788,
      "learning_rate": 1.4430849643463303e-06,
      "loss": 0.687,
      "step": 26355
    },
    {
      "epoch": 0.7594075952284908,
      "grad_norm": 0.7122129201889038,
      "learning_rate": 1.442757044122135e-06,
      "loss": 0.6829,
      "step": 26356
    },
    {
      "epoch": 0.7594364086901401,
      "grad_norm": 0.7004997134208679,
      "learning_rate": 1.4424291548779734e-06,
      "loss": 0.6946,
      "step": 26357
    },
    {
      "epoch": 0.7594652221517894,
      "grad_norm": 0.7198208570480347,
      "learning_rate": 1.4421012966166959e-06,
      "loss": 0.7251,
      "step": 26358
    },
    {
      "epoch": 0.7594940356134386,
      "grad_norm": 0.719883143901825,
      "learning_rate": 1.441773469341161e-06,
      "loss": 0.6698,
      "step": 26359
    },
    {
      "epoch": 0.7595228490750879,
      "grad_norm": 0.7045531272888184,
      "learning_rate": 1.441445673054221e-06,
      "loss": 0.6781,
      "step": 26360
    },
    {
      "epoch": 0.7595516625367371,
      "grad_norm": 0.7261710166931152,
      "learning_rate": 1.4411179077587335e-06,
      "loss": 0.6835,
      "step": 26361
    },
    {
      "epoch": 0.7595804759983864,
      "grad_norm": 0.7265127301216125,
      "learning_rate": 1.4407901734575497e-06,
      "loss": 0.6899,
      "step": 26362
    },
    {
      "epoch": 0.7596092894600357,
      "grad_norm": 0.7238025665283203,
      "learning_rate": 1.4404624701535269e-06,
      "loss": 0.6908,
      "step": 26363
    },
    {
      "epoch": 0.759638102921685,
      "grad_norm": 0.7321039438247681,
      "learning_rate": 1.4401347978495168e-06,
      "loss": 0.6923,
      "step": 26364
    },
    {
      "epoch": 0.7596669163833343,
      "grad_norm": 0.7216202020645142,
      "learning_rate": 1.4398071565483751e-06,
      "loss": 0.688,
      "step": 26365
    },
    {
      "epoch": 0.7596957298449836,
      "grad_norm": 0.7323575615882874,
      "learning_rate": 1.4394795462529527e-06,
      "loss": 0.6964,
      "step": 26366
    },
    {
      "epoch": 0.7597245433066329,
      "grad_norm": 0.741414487361908,
      "learning_rate": 1.4391519669661041e-06,
      "loss": 0.6693,
      "step": 26367
    },
    {
      "epoch": 0.7597533567682822,
      "grad_norm": 0.7545050382614136,
      "learning_rate": 1.4388244186906831e-06,
      "loss": 0.711,
      "step": 26368
    },
    {
      "epoch": 0.7597821702299314,
      "grad_norm": 0.7186135053634644,
      "learning_rate": 1.4384969014295424e-06,
      "loss": 0.6902,
      "step": 26369
    },
    {
      "epoch": 0.7598109836915807,
      "grad_norm": 0.7233960628509521,
      "learning_rate": 1.438169415185532e-06,
      "loss": 0.6725,
      "step": 26370
    },
    {
      "epoch": 0.75983979715323,
      "grad_norm": 0.7280999422073364,
      "learning_rate": 1.4378419599615057e-06,
      "loss": 0.7028,
      "step": 26371
    },
    {
      "epoch": 0.7598686106148793,
      "grad_norm": 0.721177339553833,
      "learning_rate": 1.4375145357603166e-06,
      "loss": 0.6745,
      "step": 26372
    },
    {
      "epoch": 0.7598974240765286,
      "grad_norm": 0.7056264877319336,
      "learning_rate": 1.4371871425848126e-06,
      "loss": 0.6972,
      "step": 26373
    },
    {
      "epoch": 0.7599262375381778,
      "grad_norm": 0.7317838668823242,
      "learning_rate": 1.4368597804378493e-06,
      "loss": 0.7069,
      "step": 26374
    },
    {
      "epoch": 0.7599550509998271,
      "grad_norm": 0.7476316094398499,
      "learning_rate": 1.4365324493222738e-06,
      "loss": 0.6828,
      "step": 26375
    },
    {
      "epoch": 0.7599838644614764,
      "grad_norm": 0.7208220362663269,
      "learning_rate": 1.4362051492409396e-06,
      "loss": 0.676,
      "step": 26376
    },
    {
      "epoch": 0.7600126779231257,
      "grad_norm": 0.7093875408172607,
      "learning_rate": 1.435877880196695e-06,
      "loss": 0.6921,
      "step": 26377
    },
    {
      "epoch": 0.760041491384775,
      "grad_norm": 0.7424702048301697,
      "learning_rate": 1.4355506421923927e-06,
      "loss": 0.6972,
      "step": 26378
    },
    {
      "epoch": 0.7600703048464242,
      "grad_norm": 0.7148802280426025,
      "learning_rate": 1.4352234352308792e-06,
      "loss": 0.6991,
      "step": 26379
    },
    {
      "epoch": 0.7600991183080735,
      "grad_norm": 0.7147157788276672,
      "learning_rate": 1.4348962593150068e-06,
      "loss": 0.7048,
      "step": 26380
    },
    {
      "epoch": 0.7601279317697228,
      "grad_norm": 0.7198686003684998,
      "learning_rate": 1.4345691144476236e-06,
      "loss": 0.6831,
      "step": 26381
    },
    {
      "epoch": 0.7601567452313721,
      "grad_norm": 0.7366692423820496,
      "learning_rate": 1.434242000631581e-06,
      "loss": 0.7085,
      "step": 26382
    },
    {
      "epoch": 0.7601855586930214,
      "grad_norm": 0.7343320846557617,
      "learning_rate": 1.433914917869725e-06,
      "loss": 0.7114,
      "step": 26383
    },
    {
      "epoch": 0.7602143721546707,
      "grad_norm": 0.7290595173835754,
      "learning_rate": 1.433587866164906e-06,
      "loss": 0.6876,
      "step": 26384
    },
    {
      "epoch": 0.76024318561632,
      "grad_norm": 0.7257911562919617,
      "learning_rate": 1.433260845519971e-06,
      "loss": 0.688,
      "step": 26385
    },
    {
      "epoch": 0.7602719990779693,
      "grad_norm": 0.7076368927955627,
      "learning_rate": 1.4329338559377692e-06,
      "loss": 0.6999,
      "step": 26386
    },
    {
      "epoch": 0.7603008125396186,
      "grad_norm": 0.7161794304847717,
      "learning_rate": 1.432606897421147e-06,
      "loss": 0.6818,
      "step": 26387
    },
    {
      "epoch": 0.7603296260012677,
      "grad_norm": 0.7252507209777832,
      "learning_rate": 1.4322799699729523e-06,
      "loss": 0.7069,
      "step": 26388
    },
    {
      "epoch": 0.760358439462917,
      "grad_norm": 0.7173978090286255,
      "learning_rate": 1.4319530735960341e-06,
      "loss": 0.6564,
      "step": 26389
    },
    {
      "epoch": 0.7603872529245663,
      "grad_norm": 0.7028056383132935,
      "learning_rate": 1.4316262082932364e-06,
      "loss": 0.6723,
      "step": 26390
    },
    {
      "epoch": 0.7604160663862156,
      "grad_norm": 0.6992411017417908,
      "learning_rate": 1.4312993740674092e-06,
      "loss": 0.6846,
      "step": 26391
    },
    {
      "epoch": 0.7604448798478649,
      "grad_norm": 0.6964235305786133,
      "learning_rate": 1.430972570921395e-06,
      "loss": 0.6745,
      "step": 26392
    },
    {
      "epoch": 0.7604736933095142,
      "grad_norm": 0.7360996603965759,
      "learning_rate": 1.4306457988580425e-06,
      "loss": 0.6773,
      "step": 26393
    },
    {
      "epoch": 0.7605025067711635,
      "grad_norm": 0.7133355140686035,
      "learning_rate": 1.4303190578801967e-06,
      "loss": 0.6757,
      "step": 26394
    },
    {
      "epoch": 0.7605313202328128,
      "grad_norm": 0.7290951609611511,
      "learning_rate": 1.4299923479907051e-06,
      "loss": 0.6879,
      "step": 26395
    },
    {
      "epoch": 0.7605601336944621,
      "grad_norm": 0.7301414012908936,
      "learning_rate": 1.4296656691924098e-06,
      "loss": 0.7046,
      "step": 26396
    },
    {
      "epoch": 0.7605889471561114,
      "grad_norm": 0.724916398525238,
      "learning_rate": 1.429339021488159e-06,
      "loss": 0.6875,
      "step": 26397
    },
    {
      "epoch": 0.7606177606177607,
      "grad_norm": 0.7285179495811462,
      "learning_rate": 1.4290124048807947e-06,
      "loss": 0.6816,
      "step": 26398
    },
    {
      "epoch": 0.7606465740794099,
      "grad_norm": 0.71359783411026,
      "learning_rate": 1.4286858193731645e-06,
      "loss": 0.676,
      "step": 26399
    },
    {
      "epoch": 0.7606753875410592,
      "grad_norm": 0.6949415802955627,
      "learning_rate": 1.4283592649681089e-06,
      "loss": 0.6944,
      "step": 26400
    },
    {
      "epoch": 0.7607042010027084,
      "grad_norm": 0.7235470414161682,
      "learning_rate": 1.4280327416684752e-06,
      "loss": 0.7076,
      "step": 26401
    },
    {
      "epoch": 0.7607330144643577,
      "grad_norm": 0.7165905237197876,
      "learning_rate": 1.4277062494771042e-06,
      "loss": 0.6899,
      "step": 26402
    },
    {
      "epoch": 0.760761827926007,
      "grad_norm": 0.7024521827697754,
      "learning_rate": 1.4273797883968422e-06,
      "loss": 0.6961,
      "step": 26403
    },
    {
      "epoch": 0.7607906413876563,
      "grad_norm": 0.7012227177619934,
      "learning_rate": 1.4270533584305279e-06,
      "loss": 0.6737,
      "step": 26404
    },
    {
      "epoch": 0.7608194548493056,
      "grad_norm": 0.7215855121612549,
      "learning_rate": 1.4267269595810107e-06,
      "loss": 0.706,
      "step": 26405
    },
    {
      "epoch": 0.7608482683109549,
      "grad_norm": 0.7195836305618286,
      "learning_rate": 1.4264005918511277e-06,
      "loss": 0.7029,
      "step": 26406
    },
    {
      "epoch": 0.7608770817726042,
      "grad_norm": 0.7186933159828186,
      "learning_rate": 1.4260742552437251e-06,
      "loss": 0.6875,
      "step": 26407
    },
    {
      "epoch": 0.7609058952342534,
      "grad_norm": 0.7139626145362854,
      "learning_rate": 1.4257479497616412e-06,
      "loss": 0.6824,
      "step": 26408
    },
    {
      "epoch": 0.7609347086959027,
      "grad_norm": 0.7204126715660095,
      "learning_rate": 1.4254216754077198e-06,
      "loss": 0.6907,
      "step": 26409
    },
    {
      "epoch": 0.760963522157552,
      "grad_norm": 0.7177759408950806,
      "learning_rate": 1.4250954321848042e-06,
      "loss": 0.6763,
      "step": 26410
    },
    {
      "epoch": 0.7609923356192013,
      "grad_norm": 0.7372193336486816,
      "learning_rate": 1.424769220095732e-06,
      "loss": 0.6894,
      "step": 26411
    },
    {
      "epoch": 0.7610211490808506,
      "grad_norm": 0.7313472628593445,
      "learning_rate": 1.4244430391433479e-06,
      "loss": 0.6927,
      "step": 26412
    },
    {
      "epoch": 0.7610499625424999,
      "grad_norm": 0.7313352823257446,
      "learning_rate": 1.4241168893304886e-06,
      "loss": 0.7117,
      "step": 26413
    },
    {
      "epoch": 0.7610787760041492,
      "grad_norm": 0.7164430618286133,
      "learning_rate": 1.4237907706599986e-06,
      "loss": 0.6713,
      "step": 26414
    },
    {
      "epoch": 0.7611075894657984,
      "grad_norm": 0.7691025137901306,
      "learning_rate": 1.4234646831347144e-06,
      "loss": 0.694,
      "step": 26415
    },
    {
      "epoch": 0.7611364029274477,
      "grad_norm": 0.738399863243103,
      "learning_rate": 1.4231386267574771e-06,
      "loss": 0.6938,
      "step": 26416
    },
    {
      "epoch": 0.761165216389097,
      "grad_norm": 0.7282604575157166,
      "learning_rate": 1.4228126015311272e-06,
      "loss": 0.6872,
      "step": 26417
    },
    {
      "epoch": 0.7611940298507462,
      "grad_norm": 0.6973026990890503,
      "learning_rate": 1.4224866074585052e-06,
      "loss": 0.6662,
      "step": 26418
    },
    {
      "epoch": 0.7612228433123955,
      "grad_norm": 0.7285700440406799,
      "learning_rate": 1.4221606445424474e-06,
      "loss": 0.6846,
      "step": 26419
    },
    {
      "epoch": 0.7612516567740448,
      "grad_norm": 0.713190495967865,
      "learning_rate": 1.421834712785795e-06,
      "loss": 0.685,
      "step": 26420
    },
    {
      "epoch": 0.7612804702356941,
      "grad_norm": 0.6997606754302979,
      "learning_rate": 1.421508812191384e-06,
      "loss": 0.6851,
      "step": 26421
    },
    {
      "epoch": 0.7613092836973434,
      "grad_norm": 0.7222548723220825,
      "learning_rate": 1.4211829427620554e-06,
      "loss": 0.6924,
      "step": 26422
    },
    {
      "epoch": 0.7613380971589927,
      "grad_norm": 0.8090010285377502,
      "learning_rate": 1.4208571045006448e-06,
      "loss": 0.688,
      "step": 26423
    },
    {
      "epoch": 0.761366910620642,
      "grad_norm": 0.7199591398239136,
      "learning_rate": 1.4205312974099922e-06,
      "loss": 0.6862,
      "step": 26424
    },
    {
      "epoch": 0.7613957240822913,
      "grad_norm": 0.730004608631134,
      "learning_rate": 1.4202055214929323e-06,
      "loss": 0.6929,
      "step": 26425
    },
    {
      "epoch": 0.7614245375439406,
      "grad_norm": 0.7248981595039368,
      "learning_rate": 1.4198797767523036e-06,
      "loss": 0.6756,
      "step": 26426
    },
    {
      "epoch": 0.7614533510055899,
      "grad_norm": 0.741486132144928,
      "learning_rate": 1.4195540631909454e-06,
      "loss": 0.7184,
      "step": 26427
    },
    {
      "epoch": 0.761482164467239,
      "grad_norm": 0.7214481830596924,
      "learning_rate": 1.4192283808116902e-06,
      "loss": 0.6888,
      "step": 26428
    },
    {
      "epoch": 0.7615109779288883,
      "grad_norm": 0.7417476177215576,
      "learning_rate": 1.4189027296173762e-06,
      "loss": 0.6997,
      "step": 26429
    },
    {
      "epoch": 0.7615397913905376,
      "grad_norm": 0.712713897228241,
      "learning_rate": 1.4185771096108413e-06,
      "loss": 0.6779,
      "step": 26430
    },
    {
      "epoch": 0.7615686048521869,
      "grad_norm": 0.7320182919502258,
      "learning_rate": 1.4182515207949183e-06,
      "loss": 0.6582,
      "step": 26431
    },
    {
      "epoch": 0.7615974183138362,
      "grad_norm": 0.717078685760498,
      "learning_rate": 1.4179259631724441e-06,
      "loss": 0.6795,
      "step": 26432
    },
    {
      "epoch": 0.7616262317754855,
      "grad_norm": 0.7004851698875427,
      "learning_rate": 1.4176004367462554e-06,
      "loss": 0.6535,
      "step": 26433
    },
    {
      "epoch": 0.7616550452371348,
      "grad_norm": 0.701172411441803,
      "learning_rate": 1.4172749415191844e-06,
      "loss": 0.6773,
      "step": 26434
    },
    {
      "epoch": 0.7616838586987841,
      "grad_norm": 0.7287687659263611,
      "learning_rate": 1.416949477494069e-06,
      "loss": 0.6644,
      "step": 26435
    },
    {
      "epoch": 0.7617126721604334,
      "grad_norm": 0.7094519734382629,
      "learning_rate": 1.41662404467374e-06,
      "loss": 0.7016,
      "step": 26436
    },
    {
      "epoch": 0.7617414856220827,
      "grad_norm": 0.7130850553512573,
      "learning_rate": 1.4162986430610348e-06,
      "loss": 0.691,
      "step": 26437
    },
    {
      "epoch": 0.761770299083732,
      "grad_norm": 0.7344282269477844,
      "learning_rate": 1.4159732726587849e-06,
      "loss": 0.6701,
      "step": 26438
    },
    {
      "epoch": 0.7617991125453812,
      "grad_norm": 0.7122668623924255,
      "learning_rate": 1.4156479334698265e-06,
      "loss": 0.6908,
      "step": 26439
    },
    {
      "epoch": 0.7618279260070305,
      "grad_norm": 0.7105470895767212,
      "learning_rate": 1.4153226254969888e-06,
      "loss": 0.6781,
      "step": 26440
    },
    {
      "epoch": 0.7618567394686798,
      "grad_norm": 0.7169949412345886,
      "learning_rate": 1.4149973487431102e-06,
      "loss": 0.7089,
      "step": 26441
    },
    {
      "epoch": 0.761885552930329,
      "grad_norm": 0.7047091126441956,
      "learning_rate": 1.41467210321102e-06,
      "loss": 0.6846,
      "step": 26442
    },
    {
      "epoch": 0.7619143663919783,
      "grad_norm": 0.7241531014442444,
      "learning_rate": 1.4143468889035527e-06,
      "loss": 0.6602,
      "step": 26443
    },
    {
      "epoch": 0.7619431798536276,
      "grad_norm": 0.7152360081672668,
      "learning_rate": 1.4140217058235384e-06,
      "loss": 0.6754,
      "step": 26444
    },
    {
      "epoch": 0.7619719933152769,
      "grad_norm": 0.7399452924728394,
      "learning_rate": 1.4136965539738124e-06,
      "loss": 0.7174,
      "step": 26445
    },
    {
      "epoch": 0.7620008067769262,
      "grad_norm": 0.740147590637207,
      "learning_rate": 1.4133714333572024e-06,
      "loss": 0.6946,
      "step": 26446
    },
    {
      "epoch": 0.7620296202385755,
      "grad_norm": 0.7148163318634033,
      "learning_rate": 1.4130463439765419e-06,
      "loss": 0.6702,
      "step": 26447
    },
    {
      "epoch": 0.7620584337002247,
      "grad_norm": 0.7050145864486694,
      "learning_rate": 1.4127212858346634e-06,
      "loss": 0.6727,
      "step": 26448
    },
    {
      "epoch": 0.762087247161874,
      "grad_norm": 0.7257559895515442,
      "learning_rate": 1.4123962589343953e-06,
      "loss": 0.6869,
      "step": 26449
    },
    {
      "epoch": 0.7621160606235233,
      "grad_norm": 0.7044984698295593,
      "learning_rate": 1.412071263278571e-06,
      "loss": 0.6615,
      "step": 26450
    },
    {
      "epoch": 0.7621448740851726,
      "grad_norm": 0.7218589186668396,
      "learning_rate": 1.411746298870018e-06,
      "loss": 0.6712,
      "step": 26451
    },
    {
      "epoch": 0.7621736875468219,
      "grad_norm": 0.7309069037437439,
      "learning_rate": 1.4114213657115677e-06,
      "loss": 0.6915,
      "step": 26452
    },
    {
      "epoch": 0.7622025010084712,
      "grad_norm": 0.7244671583175659,
      "learning_rate": 1.4110964638060498e-06,
      "loss": 0.6883,
      "step": 26453
    },
    {
      "epoch": 0.7622313144701205,
      "grad_norm": 0.7226796746253967,
      "learning_rate": 1.4107715931562959e-06,
      "loss": 0.6615,
      "step": 26454
    },
    {
      "epoch": 0.7622601279317697,
      "grad_norm": 0.704721987247467,
      "learning_rate": 1.4104467537651318e-06,
      "loss": 0.6979,
      "step": 26455
    },
    {
      "epoch": 0.762288941393419,
      "grad_norm": 0.7380164861679077,
      "learning_rate": 1.4101219456353898e-06,
      "loss": 0.71,
      "step": 26456
    },
    {
      "epoch": 0.7623177548550683,
      "grad_norm": 0.7083105444908142,
      "learning_rate": 1.4097971687698957e-06,
      "loss": 0.6734,
      "step": 26457
    },
    {
      "epoch": 0.7623465683167175,
      "grad_norm": 0.7485983371734619,
      "learning_rate": 1.4094724231714812e-06,
      "loss": 0.6985,
      "step": 26458
    },
    {
      "epoch": 0.7623753817783668,
      "grad_norm": 0.739203155040741,
      "learning_rate": 1.4091477088429706e-06,
      "loss": 0.6716,
      "step": 26459
    },
    {
      "epoch": 0.7624041952400161,
      "grad_norm": 0.7267200350761414,
      "learning_rate": 1.408823025787196e-06,
      "loss": 0.6896,
      "step": 26460
    },
    {
      "epoch": 0.7624330087016654,
      "grad_norm": 0.7351954579353333,
      "learning_rate": 1.4084983740069808e-06,
      "loss": 0.697,
      "step": 26461
    },
    {
      "epoch": 0.7624618221633147,
      "grad_norm": 0.7178053855895996,
      "learning_rate": 1.4081737535051566e-06,
      "loss": 0.7001,
      "step": 26462
    },
    {
      "epoch": 0.762490635624964,
      "grad_norm": 0.740697979927063,
      "learning_rate": 1.4078491642845465e-06,
      "loss": 0.6684,
      "step": 26463
    },
    {
      "epoch": 0.7625194490866133,
      "grad_norm": 0.7380466461181641,
      "learning_rate": 1.4075246063479797e-06,
      "loss": 0.693,
      "step": 26464
    },
    {
      "epoch": 0.7625482625482626,
      "grad_norm": 0.7223900556564331,
      "learning_rate": 1.4072000796982821e-06,
      "loss": 0.6799,
      "step": 26465
    },
    {
      "epoch": 0.7625770760099119,
      "grad_norm": 0.7195135951042175,
      "learning_rate": 1.406875584338282e-06,
      "loss": 0.6943,
      "step": 26466
    },
    {
      "epoch": 0.7626058894715612,
      "grad_norm": 0.716604471206665,
      "learning_rate": 1.4065511202708027e-06,
      "loss": 0.6675,
      "step": 26467
    },
    {
      "epoch": 0.7626347029332105,
      "grad_norm": 0.7084617614746094,
      "learning_rate": 1.4062266874986702e-06,
      "loss": 0.6878,
      "step": 26468
    },
    {
      "epoch": 0.7626635163948596,
      "grad_norm": 0.713108479976654,
      "learning_rate": 1.4059022860247128e-06,
      "loss": 0.6763,
      "step": 26469
    },
    {
      "epoch": 0.7626923298565089,
      "grad_norm": 0.7292611002922058,
      "learning_rate": 1.4055779158517524e-06,
      "loss": 0.6868,
      "step": 26470
    },
    {
      "epoch": 0.7627211433181582,
      "grad_norm": 0.7171235680580139,
      "learning_rate": 1.4052535769826165e-06,
      "loss": 0.6841,
      "step": 26471
    },
    {
      "epoch": 0.7627499567798075,
      "grad_norm": 0.7365220785140991,
      "learning_rate": 1.4049292694201267e-06,
      "loss": 0.6996,
      "step": 26472
    },
    {
      "epoch": 0.7627787702414568,
      "grad_norm": 0.7623327374458313,
      "learning_rate": 1.4046049931671112e-06,
      "loss": 0.6987,
      "step": 26473
    },
    {
      "epoch": 0.7628075837031061,
      "grad_norm": 0.7204704284667969,
      "learning_rate": 1.4042807482263904e-06,
      "loss": 0.6839,
      "step": 26474
    },
    {
      "epoch": 0.7628363971647554,
      "grad_norm": 0.7189933061599731,
      "learning_rate": 1.403956534600791e-06,
      "loss": 0.6806,
      "step": 26475
    },
    {
      "epoch": 0.7628652106264047,
      "grad_norm": 0.7324743866920471,
      "learning_rate": 1.4036323522931334e-06,
      "loss": 0.7091,
      "step": 26476
    },
    {
      "epoch": 0.762894024088054,
      "grad_norm": 0.726728618144989,
      "learning_rate": 1.403308201306245e-06,
      "loss": 0.6889,
      "step": 26477
    },
    {
      "epoch": 0.7629228375497032,
      "grad_norm": 0.7065137624740601,
      "learning_rate": 1.4029840816429458e-06,
      "loss": 0.6887,
      "step": 26478
    },
    {
      "epoch": 0.7629516510113525,
      "grad_norm": 0.719127893447876,
      "learning_rate": 1.402659993306061e-06,
      "loss": 0.6705,
      "step": 26479
    },
    {
      "epoch": 0.7629804644730018,
      "grad_norm": 0.7309948801994324,
      "learning_rate": 1.4023359362984102e-06,
      "loss": 0.698,
      "step": 26480
    },
    {
      "epoch": 0.7630092779346511,
      "grad_norm": 0.7159312963485718,
      "learning_rate": 1.4020119106228181e-06,
      "loss": 0.6907,
      "step": 26481
    },
    {
      "epoch": 0.7630380913963003,
      "grad_norm": 0.7373626232147217,
      "learning_rate": 1.4016879162821046e-06,
      "loss": 0.6933,
      "step": 26482
    },
    {
      "epoch": 0.7630669048579496,
      "grad_norm": 0.7150416970252991,
      "learning_rate": 1.4013639532790934e-06,
      "loss": 0.6972,
      "step": 26483
    },
    {
      "epoch": 0.7630957183195989,
      "grad_norm": 0.7204016447067261,
      "learning_rate": 1.4010400216166036e-06,
      "loss": 0.6944,
      "step": 26484
    },
    {
      "epoch": 0.7631245317812482,
      "grad_norm": 0.7307686805725098,
      "learning_rate": 1.400716121297458e-06,
      "loss": 0.6924,
      "step": 26485
    },
    {
      "epoch": 0.7631533452428975,
      "grad_norm": 0.7282085418701172,
      "learning_rate": 1.4003922523244778e-06,
      "loss": 0.6788,
      "step": 26486
    },
    {
      "epoch": 0.7631821587045468,
      "grad_norm": 0.7235678434371948,
      "learning_rate": 1.4000684147004818e-06,
      "loss": 0.6846,
      "step": 26487
    },
    {
      "epoch": 0.763210972166196,
      "grad_norm": 0.7417546510696411,
      "learning_rate": 1.3997446084282911e-06,
      "loss": 0.68,
      "step": 26488
    },
    {
      "epoch": 0.7632397856278453,
      "grad_norm": 0.7613782286643982,
      "learning_rate": 1.3994208335107274e-06,
      "loss": 0.6954,
      "step": 26489
    },
    {
      "epoch": 0.7632685990894946,
      "grad_norm": 0.7195947170257568,
      "learning_rate": 1.3990970899506073e-06,
      "loss": 0.6734,
      "step": 26490
    },
    {
      "epoch": 0.7632974125511439,
      "grad_norm": 0.7199484705924988,
      "learning_rate": 1.3987733777507523e-06,
      "loss": 0.6832,
      "step": 26491
    },
    {
      "epoch": 0.7633262260127932,
      "grad_norm": 0.7065523862838745,
      "learning_rate": 1.3984496969139827e-06,
      "loss": 0.6884,
      "step": 26492
    },
    {
      "epoch": 0.7633550394744425,
      "grad_norm": 0.7236217260360718,
      "learning_rate": 1.398126047443114e-06,
      "loss": 0.6951,
      "step": 26493
    },
    {
      "epoch": 0.7633838529360918,
      "grad_norm": 0.8000146150588989,
      "learning_rate": 1.3978024293409693e-06,
      "loss": 0.7144,
      "step": 26494
    },
    {
      "epoch": 0.7634126663977411,
      "grad_norm": 0.7289760112762451,
      "learning_rate": 1.3974788426103625e-06,
      "loss": 0.6849,
      "step": 26495
    },
    {
      "epoch": 0.7634414798593903,
      "grad_norm": 0.7226312756538391,
      "learning_rate": 1.397155287254115e-06,
      "loss": 0.6986,
      "step": 26496
    },
    {
      "epoch": 0.7634702933210396,
      "grad_norm": 0.7212361097335815,
      "learning_rate": 1.3968317632750422e-06,
      "loss": 0.6821,
      "step": 26497
    },
    {
      "epoch": 0.7634991067826888,
      "grad_norm": 0.7238344550132751,
      "learning_rate": 1.3965082706759647e-06,
      "loss": 0.7028,
      "step": 26498
    },
    {
      "epoch": 0.7635279202443381,
      "grad_norm": 0.7153707146644592,
      "learning_rate": 1.3961848094596963e-06,
      "loss": 0.6892,
      "step": 26499
    },
    {
      "epoch": 0.7635567337059874,
      "grad_norm": 0.7275568246841431,
      "learning_rate": 1.3958613796290554e-06,
      "loss": 0.6879,
      "step": 26500
    },
    {
      "epoch": 0.7635855471676367,
      "grad_norm": 0.7266079187393188,
      "learning_rate": 1.3955379811868598e-06,
      "loss": 0.6912,
      "step": 26501
    },
    {
      "epoch": 0.763614360629286,
      "grad_norm": 0.7169862985610962,
      "learning_rate": 1.3952146141359258e-06,
      "loss": 0.6806,
      "step": 26502
    },
    {
      "epoch": 0.7636431740909353,
      "grad_norm": 0.7296160459518433,
      "learning_rate": 1.394891278479068e-06,
      "loss": 0.6695,
      "step": 26503
    },
    {
      "epoch": 0.7636719875525846,
      "grad_norm": 0.714573323726654,
      "learning_rate": 1.3945679742191048e-06,
      "loss": 0.6816,
      "step": 26504
    },
    {
      "epoch": 0.7637008010142339,
      "grad_norm": 0.722657322883606,
      "learning_rate": 1.3942447013588495e-06,
      "loss": 0.6925,
      "step": 26505
    },
    {
      "epoch": 0.7637296144758832,
      "grad_norm": 0.7394464612007141,
      "learning_rate": 1.3939214599011174e-06,
      "loss": 0.6866,
      "step": 26506
    },
    {
      "epoch": 0.7637584279375325,
      "grad_norm": 0.7232996225357056,
      "learning_rate": 1.3935982498487272e-06,
      "loss": 0.6962,
      "step": 26507
    },
    {
      "epoch": 0.7637872413991817,
      "grad_norm": 0.7278972268104553,
      "learning_rate": 1.393275071204489e-06,
      "loss": 0.688,
      "step": 26508
    },
    {
      "epoch": 0.7638160548608309,
      "grad_norm": 0.7122330665588379,
      "learning_rate": 1.3929519239712207e-06,
      "loss": 0.6821,
      "step": 26509
    },
    {
      "epoch": 0.7638448683224802,
      "grad_norm": 0.723507285118103,
      "learning_rate": 1.3926288081517346e-06,
      "loss": 0.6826,
      "step": 26510
    },
    {
      "epoch": 0.7638736817841295,
      "grad_norm": 0.7132101655006409,
      "learning_rate": 1.392305723748847e-06,
      "loss": 0.6791,
      "step": 26511
    },
    {
      "epoch": 0.7639024952457788,
      "grad_norm": 0.7393293380737305,
      "learning_rate": 1.3919826707653667e-06,
      "loss": 0.6873,
      "step": 26512
    },
    {
      "epoch": 0.7639313087074281,
      "grad_norm": 0.7344639301300049,
      "learning_rate": 1.391659649204114e-06,
      "loss": 0.7001,
      "step": 26513
    },
    {
      "epoch": 0.7639601221690774,
      "grad_norm": 0.7434249520301819,
      "learning_rate": 1.3913366590678967e-06,
      "loss": 0.6985,
      "step": 26514
    },
    {
      "epoch": 0.7639889356307267,
      "grad_norm": 0.7275723814964294,
      "learning_rate": 1.391013700359532e-06,
      "loss": 0.6868,
      "step": 26515
    },
    {
      "epoch": 0.764017749092376,
      "grad_norm": 0.7162958383560181,
      "learning_rate": 1.3906907730818286e-06,
      "loss": 0.6955,
      "step": 26516
    },
    {
      "epoch": 0.7640465625540253,
      "grad_norm": 0.7208296060562134,
      "learning_rate": 1.3903678772376017e-06,
      "loss": 0.706,
      "step": 26517
    },
    {
      "epoch": 0.7640753760156745,
      "grad_norm": 0.7271823287010193,
      "learning_rate": 1.390045012829661e-06,
      "loss": 0.6877,
      "step": 26518
    },
    {
      "epoch": 0.7641041894773238,
      "grad_norm": 0.7230120897293091,
      "learning_rate": 1.3897221798608213e-06,
      "loss": 0.6722,
      "step": 26519
    },
    {
      "epoch": 0.7641330029389731,
      "grad_norm": 0.7704513072967529,
      "learning_rate": 1.3893993783338911e-06,
      "loss": 0.7092,
      "step": 26520
    },
    {
      "epoch": 0.7641618164006224,
      "grad_norm": 0.741064727306366,
      "learning_rate": 1.3890766082516838e-06,
      "loss": 0.7058,
      "step": 26521
    },
    {
      "epoch": 0.7641906298622717,
      "grad_norm": 0.7194626927375793,
      "learning_rate": 1.3887538696170089e-06,
      "loss": 0.7051,
      "step": 26522
    },
    {
      "epoch": 0.7642194433239209,
      "grad_norm": 0.7104766368865967,
      "learning_rate": 1.3884311624326774e-06,
      "loss": 0.6998,
      "step": 26523
    },
    {
      "epoch": 0.7642482567855702,
      "grad_norm": 0.7319368720054626,
      "learning_rate": 1.3881084867015004e-06,
      "loss": 0.6942,
      "step": 26524
    },
    {
      "epoch": 0.7642770702472195,
      "grad_norm": 0.7122235894203186,
      "learning_rate": 1.3877858424262892e-06,
      "loss": 0.6884,
      "step": 26525
    },
    {
      "epoch": 0.7643058837088688,
      "grad_norm": 0.7117592096328735,
      "learning_rate": 1.3874632296098512e-06,
      "loss": 0.6929,
      "step": 26526
    },
    {
      "epoch": 0.764334697170518,
      "grad_norm": 0.6892096996307373,
      "learning_rate": 1.3871406482549987e-06,
      "loss": 0.6738,
      "step": 26527
    },
    {
      "epoch": 0.7643635106321673,
      "grad_norm": 0.7416712045669556,
      "learning_rate": 1.386818098364538e-06,
      "loss": 0.7025,
      "step": 26528
    },
    {
      "epoch": 0.7643923240938166,
      "grad_norm": 0.7176942825317383,
      "learning_rate": 1.38649557994128e-06,
      "loss": 0.6784,
      "step": 26529
    },
    {
      "epoch": 0.7644211375554659,
      "grad_norm": 0.7074705362319946,
      "learning_rate": 1.3861730929880347e-06,
      "loss": 0.6864,
      "step": 26530
    },
    {
      "epoch": 0.7644499510171152,
      "grad_norm": 0.7027961611747742,
      "learning_rate": 1.3858506375076075e-06,
      "loss": 0.6921,
      "step": 26531
    },
    {
      "epoch": 0.7644787644787645,
      "grad_norm": 0.7205295562744141,
      "learning_rate": 1.3855282135028103e-06,
      "loss": 0.6833,
      "step": 26532
    },
    {
      "epoch": 0.7645075779404138,
      "grad_norm": 0.7127604484558105,
      "learning_rate": 1.3852058209764473e-06,
      "loss": 0.6844,
      "step": 26533
    },
    {
      "epoch": 0.7645363914020631,
      "grad_norm": 0.7268697023391724,
      "learning_rate": 1.3848834599313304e-06,
      "loss": 0.6993,
      "step": 26534
    },
    {
      "epoch": 0.7645652048637124,
      "grad_norm": 0.7461169362068176,
      "learning_rate": 1.3845611303702627e-06,
      "loss": 0.6999,
      "step": 26535
    },
    {
      "epoch": 0.7645940183253616,
      "grad_norm": 0.7217694520950317,
      "learning_rate": 1.3842388322960537e-06,
      "loss": 0.6834,
      "step": 26536
    },
    {
      "epoch": 0.7646228317870108,
      "grad_norm": 0.7466334104537964,
      "learning_rate": 1.3839165657115094e-06,
      "loss": 0.703,
      "step": 26537
    },
    {
      "epoch": 0.7646516452486601,
      "grad_norm": 0.7347504496574402,
      "learning_rate": 1.3835943306194394e-06,
      "loss": 0.7046,
      "step": 26538
    },
    {
      "epoch": 0.7646804587103094,
      "grad_norm": 0.7210476398468018,
      "learning_rate": 1.3832721270226456e-06,
      "loss": 0.7067,
      "step": 26539
    },
    {
      "epoch": 0.7647092721719587,
      "grad_norm": 0.7573145627975464,
      "learning_rate": 1.382949954923938e-06,
      "loss": 0.6866,
      "step": 26540
    },
    {
      "epoch": 0.764738085633608,
      "grad_norm": 0.7265874743461609,
      "learning_rate": 1.3826278143261191e-06,
      "loss": 0.7256,
      "step": 26541
    },
    {
      "epoch": 0.7647668990952573,
      "grad_norm": 0.7004141807556152,
      "learning_rate": 1.3823057052319972e-06,
      "loss": 0.6673,
      "step": 26542
    },
    {
      "epoch": 0.7647957125569066,
      "grad_norm": 0.7148511409759521,
      "learning_rate": 1.3819836276443742e-06,
      "loss": 0.7018,
      "step": 26543
    },
    {
      "epoch": 0.7648245260185559,
      "grad_norm": 0.7049953937530518,
      "learning_rate": 1.3816615815660578e-06,
      "loss": 0.682,
      "step": 26544
    },
    {
      "epoch": 0.7648533394802052,
      "grad_norm": 0.7164357900619507,
      "learning_rate": 1.3813395669998535e-06,
      "loss": 0.6817,
      "step": 26545
    },
    {
      "epoch": 0.7648821529418545,
      "grad_norm": 0.7338948249816895,
      "learning_rate": 1.381017583948563e-06,
      "loss": 0.6901,
      "step": 26546
    },
    {
      "epoch": 0.7649109664035038,
      "grad_norm": 0.7144280672073364,
      "learning_rate": 1.3806956324149928e-06,
      "loss": 0.689,
      "step": 26547
    },
    {
      "epoch": 0.764939779865153,
      "grad_norm": 0.7660568952560425,
      "learning_rate": 1.380373712401944e-06,
      "loss": 0.7113,
      "step": 26548
    },
    {
      "epoch": 0.7649685933268023,
      "grad_norm": 0.7178001999855042,
      "learning_rate": 1.3800518239122218e-06,
      "loss": 0.7063,
      "step": 26549
    },
    {
      "epoch": 0.7649974067884515,
      "grad_norm": 0.7311683297157288,
      "learning_rate": 1.3797299669486296e-06,
      "loss": 0.6654,
      "step": 26550
    },
    {
      "epoch": 0.7650262202501008,
      "grad_norm": 0.739582359790802,
      "learning_rate": 1.3794081415139727e-06,
      "loss": 0.6798,
      "step": 26551
    },
    {
      "epoch": 0.7650550337117501,
      "grad_norm": 0.7196133136749268,
      "learning_rate": 1.3790863476110493e-06,
      "loss": 0.688,
      "step": 26552
    },
    {
      "epoch": 0.7650838471733994,
      "grad_norm": 0.7157800793647766,
      "learning_rate": 1.3787645852426663e-06,
      "loss": 0.6807,
      "step": 26553
    },
    {
      "epoch": 0.7651126606350487,
      "grad_norm": 0.7055637836456299,
      "learning_rate": 1.3784428544116218e-06,
      "loss": 0.6734,
      "step": 26554
    },
    {
      "epoch": 0.765141474096698,
      "grad_norm": 0.7100403308868408,
      "learning_rate": 1.3781211551207219e-06,
      "loss": 0.6718,
      "step": 26555
    },
    {
      "epoch": 0.7651702875583473,
      "grad_norm": 0.724446713924408,
      "learning_rate": 1.377799487372764e-06,
      "loss": 0.6985,
      "step": 26556
    },
    {
      "epoch": 0.7651991010199966,
      "grad_norm": 0.7206264734268188,
      "learning_rate": 1.3774778511705539e-06,
      "loss": 0.6906,
      "step": 26557
    },
    {
      "epoch": 0.7652279144816458,
      "grad_norm": 0.7012507915496826,
      "learning_rate": 1.3771562465168885e-06,
      "loss": 0.671,
      "step": 26558
    },
    {
      "epoch": 0.7652567279432951,
      "grad_norm": 0.7435001730918884,
      "learning_rate": 1.3768346734145705e-06,
      "loss": 0.6823,
      "step": 26559
    },
    {
      "epoch": 0.7652855414049444,
      "grad_norm": 0.7163371443748474,
      "learning_rate": 1.376513131866401e-06,
      "loss": 0.6893,
      "step": 26560
    },
    {
      "epoch": 0.7653143548665937,
      "grad_norm": 0.7268722653388977,
      "learning_rate": 1.3761916218751814e-06,
      "loss": 0.6804,
      "step": 26561
    },
    {
      "epoch": 0.765343168328243,
      "grad_norm": 0.7496002316474915,
      "learning_rate": 1.375870143443709e-06,
      "loss": 0.7094,
      "step": 26562
    },
    {
      "epoch": 0.7653719817898922,
      "grad_norm": 0.7219821810722351,
      "learning_rate": 1.375548696574786e-06,
      "loss": 0.6729,
      "step": 26563
    },
    {
      "epoch": 0.7654007952515415,
      "grad_norm": 0.7111203670501709,
      "learning_rate": 1.3752272812712097e-06,
      "loss": 0.6929,
      "step": 26564
    },
    {
      "epoch": 0.7654296087131908,
      "grad_norm": 0.7397687435150146,
      "learning_rate": 1.3749058975357814e-06,
      "loss": 0.7208,
      "step": 26565
    },
    {
      "epoch": 0.7654584221748401,
      "grad_norm": 0.7304925918579102,
      "learning_rate": 1.374584545371298e-06,
      "loss": 0.7059,
      "step": 26566
    },
    {
      "epoch": 0.7654872356364893,
      "grad_norm": 0.7210354804992676,
      "learning_rate": 1.3742632247805588e-06,
      "loss": 0.6663,
      "step": 26567
    },
    {
      "epoch": 0.7655160490981386,
      "grad_norm": 0.7280187606811523,
      "learning_rate": 1.3739419357663635e-06,
      "loss": 0.7079,
      "step": 26568
    },
    {
      "epoch": 0.7655448625597879,
      "grad_norm": 0.7198574542999268,
      "learning_rate": 1.373620678331508e-06,
      "loss": 0.6736,
      "step": 26569
    },
    {
      "epoch": 0.7655736760214372,
      "grad_norm": 0.7313529253005981,
      "learning_rate": 1.3732994524787934e-06,
      "loss": 0.6991,
      "step": 26570
    },
    {
      "epoch": 0.7656024894830865,
      "grad_norm": 0.7261677384376526,
      "learning_rate": 1.372978258211012e-06,
      "loss": 0.6856,
      "step": 26571
    },
    {
      "epoch": 0.7656313029447358,
      "grad_norm": 0.7128048539161682,
      "learning_rate": 1.3726570955309665e-06,
      "loss": 0.6919,
      "step": 26572
    },
    {
      "epoch": 0.7656601164063851,
      "grad_norm": 0.7168378233909607,
      "learning_rate": 1.3723359644414507e-06,
      "loss": 0.6448,
      "step": 26573
    },
    {
      "epoch": 0.7656889298680344,
      "grad_norm": 0.7236161231994629,
      "learning_rate": 1.372014864945264e-06,
      "loss": 0.699,
      "step": 26574
    },
    {
      "epoch": 0.7657177433296837,
      "grad_norm": 0.726719856262207,
      "learning_rate": 1.3716937970451994e-06,
      "loss": 0.676,
      "step": 26575
    },
    {
      "epoch": 0.7657465567913329,
      "grad_norm": 0.7288396954536438,
      "learning_rate": 1.3713727607440568e-06,
      "loss": 0.6746,
      "step": 26576
    },
    {
      "epoch": 0.7657753702529821,
      "grad_norm": 0.7371814250946045,
      "learning_rate": 1.3710517560446278e-06,
      "loss": 0.6959,
      "step": 26577
    },
    {
      "epoch": 0.7658041837146314,
      "grad_norm": 0.7256645560264587,
      "learning_rate": 1.370730782949713e-06,
      "loss": 0.6915,
      "step": 26578
    },
    {
      "epoch": 0.7658329971762807,
      "grad_norm": 0.7020153403282166,
      "learning_rate": 1.3704098414621031e-06,
      "loss": 0.6626,
      "step": 26579
    },
    {
      "epoch": 0.76586181063793,
      "grad_norm": 0.7066386938095093,
      "learning_rate": 1.3700889315845967e-06,
      "loss": 0.6845,
      "step": 26580
    },
    {
      "epoch": 0.7658906240995793,
      "grad_norm": 0.7357393503189087,
      "learning_rate": 1.3697680533199854e-06,
      "loss": 0.6706,
      "step": 26581
    },
    {
      "epoch": 0.7659194375612286,
      "grad_norm": 0.7352892756462097,
      "learning_rate": 1.3694472066710658e-06,
      "loss": 0.6733,
      "step": 26582
    },
    {
      "epoch": 0.7659482510228779,
      "grad_norm": 0.6982582807540894,
      "learning_rate": 1.3691263916406316e-06,
      "loss": 0.6757,
      "step": 26583
    },
    {
      "epoch": 0.7659770644845272,
      "grad_norm": 0.7271583080291748,
      "learning_rate": 1.3688056082314782e-06,
      "loss": 0.6858,
      "step": 26584
    },
    {
      "epoch": 0.7660058779461765,
      "grad_norm": 0.7307323217391968,
      "learning_rate": 1.368484856446397e-06,
      "loss": 0.683,
      "step": 26585
    },
    {
      "epoch": 0.7660346914078258,
      "grad_norm": 0.7065715789794922,
      "learning_rate": 1.3681641362881841e-06,
      "loss": 0.6842,
      "step": 26586
    },
    {
      "epoch": 0.766063504869475,
      "grad_norm": 0.7092729806900024,
      "learning_rate": 1.3678434477596291e-06,
      "loss": 0.6729,
      "step": 26587
    },
    {
      "epoch": 0.7660923183311243,
      "grad_norm": 0.762200653553009,
      "learning_rate": 1.3675227908635276e-06,
      "loss": 0.6919,
      "step": 26588
    },
    {
      "epoch": 0.7661211317927736,
      "grad_norm": 0.7162842750549316,
      "learning_rate": 1.3672021656026723e-06,
      "loss": 0.6974,
      "step": 26589
    },
    {
      "epoch": 0.7661499452544228,
      "grad_norm": 0.7302046418190002,
      "learning_rate": 1.3668815719798534e-06,
      "loss": 0.7053,
      "step": 26590
    },
    {
      "epoch": 0.7661787587160721,
      "grad_norm": 0.7109426856040955,
      "learning_rate": 1.3665610099978655e-06,
      "loss": 0.6746,
      "step": 26591
    },
    {
      "epoch": 0.7662075721777214,
      "grad_norm": 0.7470408082008362,
      "learning_rate": 1.366240479659498e-06,
      "loss": 0.6846,
      "step": 26592
    },
    {
      "epoch": 0.7662363856393707,
      "grad_norm": 0.7181761860847473,
      "learning_rate": 1.3659199809675443e-06,
      "loss": 0.6741,
      "step": 26593
    },
    {
      "epoch": 0.76626519910102,
      "grad_norm": 0.6985397934913635,
      "learning_rate": 1.365599513924794e-06,
      "loss": 0.6831,
      "step": 26594
    },
    {
      "epoch": 0.7662940125626693,
      "grad_norm": 0.7231857180595398,
      "learning_rate": 1.3652790785340386e-06,
      "loss": 0.7049,
      "step": 26595
    },
    {
      "epoch": 0.7663228260243186,
      "grad_norm": 0.7650360465049744,
      "learning_rate": 1.3649586747980692e-06,
      "loss": 0.7,
      "step": 26596
    },
    {
      "epoch": 0.7663516394859679,
      "grad_norm": 0.7191675901412964,
      "learning_rate": 1.3646383027196774e-06,
      "loss": 0.6959,
      "step": 26597
    },
    {
      "epoch": 0.7663804529476171,
      "grad_norm": 0.7272182106971741,
      "learning_rate": 1.3643179623016506e-06,
      "loss": 0.6865,
      "step": 26598
    },
    {
      "epoch": 0.7664092664092664,
      "grad_norm": 0.7256808280944824,
      "learning_rate": 1.3639976535467814e-06,
      "loss": 0.6904,
      "step": 26599
    },
    {
      "epoch": 0.7664380798709157,
      "grad_norm": 0.7406173944473267,
      "learning_rate": 1.3636773764578559e-06,
      "loss": 0.6832,
      "step": 26600
    },
    {
      "epoch": 0.766466893332565,
      "grad_norm": 0.7283905744552612,
      "learning_rate": 1.3633571310376675e-06,
      "loss": 0.6867,
      "step": 26601
    },
    {
      "epoch": 0.7664957067942143,
      "grad_norm": 0.7375782132148743,
      "learning_rate": 1.3630369172890018e-06,
      "loss": 0.6914,
      "step": 26602
    },
    {
      "epoch": 0.7665245202558635,
      "grad_norm": 0.705627977848053,
      "learning_rate": 1.3627167352146504e-06,
      "loss": 0.7015,
      "step": 26603
    },
    {
      "epoch": 0.7665533337175128,
      "grad_norm": 0.7274474501609802,
      "learning_rate": 1.3623965848173986e-06,
      "loss": 0.6751,
      "step": 26604
    },
    {
      "epoch": 0.7665821471791621,
      "grad_norm": 0.7294792532920837,
      "learning_rate": 1.3620764661000363e-06,
      "loss": 0.6786,
      "step": 26605
    },
    {
      "epoch": 0.7666109606408114,
      "grad_norm": 0.7268305420875549,
      "learning_rate": 1.3617563790653531e-06,
      "loss": 0.7051,
      "step": 26606
    },
    {
      "epoch": 0.7666397741024606,
      "grad_norm": 0.7214045524597168,
      "learning_rate": 1.3614363237161333e-06,
      "loss": 0.6837,
      "step": 26607
    },
    {
      "epoch": 0.7666685875641099,
      "grad_norm": 0.7182524800300598,
      "learning_rate": 1.3611163000551658e-06,
      "loss": 0.6592,
      "step": 26608
    },
    {
      "epoch": 0.7666974010257592,
      "grad_norm": 0.705445408821106,
      "learning_rate": 1.3607963080852377e-06,
      "loss": 0.6646,
      "step": 26609
    },
    {
      "epoch": 0.7667262144874085,
      "grad_norm": 0.7118207812309265,
      "learning_rate": 1.3604763478091375e-06,
      "loss": 0.683,
      "step": 26610
    },
    {
      "epoch": 0.7667550279490578,
      "grad_norm": 0.7258062958717346,
      "learning_rate": 1.3601564192296485e-06,
      "loss": 0.6868,
      "step": 26611
    },
    {
      "epoch": 0.7667838414107071,
      "grad_norm": 0.7189377546310425,
      "learning_rate": 1.3598365223495601e-06,
      "loss": 0.6972,
      "step": 26612
    },
    {
      "epoch": 0.7668126548723564,
      "grad_norm": 0.711666464805603,
      "learning_rate": 1.3595166571716555e-06,
      "loss": 0.676,
      "step": 26613
    },
    {
      "epoch": 0.7668414683340057,
      "grad_norm": 0.7562504410743713,
      "learning_rate": 1.359196823698723e-06,
      "loss": 0.6864,
      "step": 26614
    },
    {
      "epoch": 0.766870281795655,
      "grad_norm": 0.7344066500663757,
      "learning_rate": 1.358877021933545e-06,
      "loss": 0.693,
      "step": 26615
    },
    {
      "epoch": 0.7668990952573043,
      "grad_norm": 0.706292450428009,
      "learning_rate": 1.3585572518789102e-06,
      "loss": 0.6802,
      "step": 26616
    },
    {
      "epoch": 0.7669279087189534,
      "grad_norm": 0.7547844052314758,
      "learning_rate": 1.3582375135375997e-06,
      "loss": 0.7117,
      "step": 26617
    },
    {
      "epoch": 0.7669567221806027,
      "grad_norm": 0.7236608862876892,
      "learning_rate": 1.3579178069124022e-06,
      "loss": 0.6868,
      "step": 26618
    },
    {
      "epoch": 0.766985535642252,
      "grad_norm": 0.7235976457595825,
      "learning_rate": 1.357598132006096e-06,
      "loss": 0.7001,
      "step": 26619
    },
    {
      "epoch": 0.7670143491039013,
      "grad_norm": 0.7298413515090942,
      "learning_rate": 1.3572784888214729e-06,
      "loss": 0.6891,
      "step": 26620
    },
    {
      "epoch": 0.7670431625655506,
      "grad_norm": 0.7047087550163269,
      "learning_rate": 1.3569588773613107e-06,
      "loss": 0.6903,
      "step": 26621
    },
    {
      "epoch": 0.7670719760271999,
      "grad_norm": 0.7334681749343872,
      "learning_rate": 1.3566392976283966e-06,
      "loss": 0.7108,
      "step": 26622
    },
    {
      "epoch": 0.7671007894888492,
      "grad_norm": 0.7065392136573792,
      "learning_rate": 1.3563197496255103e-06,
      "loss": 0.6665,
      "step": 26623
    },
    {
      "epoch": 0.7671296029504985,
      "grad_norm": 0.7184056043624878,
      "learning_rate": 1.3560002333554389e-06,
      "loss": 0.6918,
      "step": 26624
    },
    {
      "epoch": 0.7671584164121478,
      "grad_norm": 0.7275425791740417,
      "learning_rate": 1.355680748820961e-06,
      "loss": 0.6698,
      "step": 26625
    },
    {
      "epoch": 0.7671872298737971,
      "grad_norm": 0.7413820624351501,
      "learning_rate": 1.3553612960248608e-06,
      "loss": 0.6629,
      "step": 26626
    },
    {
      "epoch": 0.7672160433354464,
      "grad_norm": 0.7258886098861694,
      "learning_rate": 1.3550418749699218e-06,
      "loss": 0.6991,
      "step": 26627
    },
    {
      "epoch": 0.7672448567970956,
      "grad_norm": 0.6958054900169373,
      "learning_rate": 1.354722485658923e-06,
      "loss": 0.6494,
      "step": 26628
    },
    {
      "epoch": 0.7672736702587449,
      "grad_norm": 0.7395658493041992,
      "learning_rate": 1.3544031280946485e-06,
      "loss": 0.716,
      "step": 26629
    },
    {
      "epoch": 0.7673024837203941,
      "grad_norm": 0.7218237519264221,
      "learning_rate": 1.3540838022798775e-06,
      "loss": 0.6806,
      "step": 26630
    },
    {
      "epoch": 0.7673312971820434,
      "grad_norm": 0.7351190447807312,
      "learning_rate": 1.3537645082173917e-06,
      "loss": 0.7024,
      "step": 26631
    },
    {
      "epoch": 0.7673601106436927,
      "grad_norm": 0.7315679788589478,
      "learning_rate": 1.3534452459099723e-06,
      "loss": 0.6761,
      "step": 26632
    },
    {
      "epoch": 0.767388924105342,
      "grad_norm": 0.7324588298797607,
      "learning_rate": 1.3531260153604016e-06,
      "loss": 0.6883,
      "step": 26633
    },
    {
      "epoch": 0.7674177375669913,
      "grad_norm": 0.7217755913734436,
      "learning_rate": 1.3528068165714554e-06,
      "loss": 0.6875,
      "step": 26634
    },
    {
      "epoch": 0.7674465510286406,
      "grad_norm": 0.7213778495788574,
      "learning_rate": 1.3524876495459178e-06,
      "loss": 0.6951,
      "step": 26635
    },
    {
      "epoch": 0.7674753644902899,
      "grad_norm": 0.7163661122322083,
      "learning_rate": 1.3521685142865653e-06,
      "loss": 0.7016,
      "step": 26636
    },
    {
      "epoch": 0.7675041779519391,
      "grad_norm": 0.7243016958236694,
      "learning_rate": 1.3518494107961798e-06,
      "loss": 0.6926,
      "step": 26637
    },
    {
      "epoch": 0.7675329914135884,
      "grad_norm": 0.723901093006134,
      "learning_rate": 1.3515303390775375e-06,
      "loss": 0.7175,
      "step": 26638
    },
    {
      "epoch": 0.7675618048752377,
      "grad_norm": 0.7358098030090332,
      "learning_rate": 1.351211299133421e-06,
      "loss": 0.708,
      "step": 26639
    },
    {
      "epoch": 0.767590618336887,
      "grad_norm": 0.6973048448562622,
      "learning_rate": 1.3508922909666044e-06,
      "loss": 0.6736,
      "step": 26640
    },
    {
      "epoch": 0.7676194317985363,
      "grad_norm": 0.7208430171012878,
      "learning_rate": 1.3505733145798684e-06,
      "loss": 0.6897,
      "step": 26641
    },
    {
      "epoch": 0.7676482452601856,
      "grad_norm": 0.7177976965904236,
      "learning_rate": 1.3502543699759918e-06,
      "loss": 0.6956,
      "step": 26642
    },
    {
      "epoch": 0.7676770587218349,
      "grad_norm": 0.7239261269569397,
      "learning_rate": 1.3499354571577505e-06,
      "loss": 0.6777,
      "step": 26643
    },
    {
      "epoch": 0.7677058721834841,
      "grad_norm": 0.7657850384712219,
      "learning_rate": 1.3496165761279223e-06,
      "loss": 0.7118,
      "step": 26644
    },
    {
      "epoch": 0.7677346856451334,
      "grad_norm": 0.7076495289802551,
      "learning_rate": 1.3492977268892859e-06,
      "loss": 0.6983,
      "step": 26645
    },
    {
      "epoch": 0.7677634991067827,
      "grad_norm": 0.7105011343955994,
      "learning_rate": 1.348978909444616e-06,
      "loss": 0.7155,
      "step": 26646
    },
    {
      "epoch": 0.767792312568432,
      "grad_norm": 0.7083531618118286,
      "learning_rate": 1.3486601237966895e-06,
      "loss": 0.6592,
      "step": 26647
    },
    {
      "epoch": 0.7678211260300812,
      "grad_norm": 0.7203927636146545,
      "learning_rate": 1.3483413699482855e-06,
      "loss": 0.6739,
      "step": 26648
    },
    {
      "epoch": 0.7678499394917305,
      "grad_norm": 0.7147826552391052,
      "learning_rate": 1.3480226479021758e-06,
      "loss": 0.6697,
      "step": 26649
    },
    {
      "epoch": 0.7678787529533798,
      "grad_norm": 0.7158384919166565,
      "learning_rate": 1.34770395766114e-06,
      "loss": 0.6967,
      "step": 26650
    },
    {
      "epoch": 0.7679075664150291,
      "grad_norm": 0.7413225173950195,
      "learning_rate": 1.34738529922795e-06,
      "loss": 0.7002,
      "step": 26651
    },
    {
      "epoch": 0.7679363798766784,
      "grad_norm": 0.6950060129165649,
      "learning_rate": 1.3470666726053844e-06,
      "loss": 0.6992,
      "step": 26652
    },
    {
      "epoch": 0.7679651933383277,
      "grad_norm": 0.7380635738372803,
      "learning_rate": 1.346748077796215e-06,
      "loss": 0.7018,
      "step": 26653
    },
    {
      "epoch": 0.767994006799977,
      "grad_norm": 0.7423444986343384,
      "learning_rate": 1.3464295148032198e-06,
      "loss": 0.6783,
      "step": 26654
    },
    {
      "epoch": 0.7680228202616263,
      "grad_norm": 0.7231791615486145,
      "learning_rate": 1.3461109836291674e-06,
      "loss": 0.6784,
      "step": 26655
    },
    {
      "epoch": 0.7680516337232756,
      "grad_norm": 0.7050277590751648,
      "learning_rate": 1.3457924842768393e-06,
      "loss": 0.6715,
      "step": 26656
    },
    {
      "epoch": 0.7680804471849247,
      "grad_norm": 0.7292289733886719,
      "learning_rate": 1.3454740167490044e-06,
      "loss": 0.6975,
      "step": 26657
    },
    {
      "epoch": 0.768109260646574,
      "grad_norm": 0.71676105260849,
      "learning_rate": 1.345155581048439e-06,
      "loss": 0.7077,
      "step": 26658
    },
    {
      "epoch": 0.7681380741082233,
      "grad_norm": 0.73398357629776,
      "learning_rate": 1.3448371771779134e-06,
      "loss": 0.6991,
      "step": 26659
    },
    {
      "epoch": 0.7681668875698726,
      "grad_norm": 0.7058345079421997,
      "learning_rate": 1.3445188051402042e-06,
      "loss": 0.6768,
      "step": 26660
    },
    {
      "epoch": 0.7681957010315219,
      "grad_norm": 0.733452320098877,
      "learning_rate": 1.3442004649380796e-06,
      "loss": 0.7022,
      "step": 26661
    },
    {
      "epoch": 0.7682245144931712,
      "grad_norm": 0.7282528877258301,
      "learning_rate": 1.3438821565743165e-06,
      "loss": 0.6948,
      "step": 26662
    },
    {
      "epoch": 0.7682533279548205,
      "grad_norm": 0.7139299511909485,
      "learning_rate": 1.3435638800516836e-06,
      "loss": 0.6839,
      "step": 26663
    },
    {
      "epoch": 0.7682821414164698,
      "grad_norm": 0.7127522230148315,
      "learning_rate": 1.3432456353729545e-06,
      "loss": 0.6994,
      "step": 26664
    },
    {
      "epoch": 0.7683109548781191,
      "grad_norm": 0.7262121438980103,
      "learning_rate": 1.3429274225409017e-06,
      "loss": 0.7103,
      "step": 26665
    },
    {
      "epoch": 0.7683397683397684,
      "grad_norm": 0.7739055752754211,
      "learning_rate": 1.3426092415582936e-06,
      "loss": 0.6953,
      "step": 26666
    },
    {
      "epoch": 0.7683685818014176,
      "grad_norm": 0.7478476762771606,
      "learning_rate": 1.3422910924279032e-06,
      "loss": 0.7192,
      "step": 26667
    },
    {
      "epoch": 0.7683973952630669,
      "grad_norm": 0.7022139430046082,
      "learning_rate": 1.3419729751525006e-06,
      "loss": 0.6944,
      "step": 26668
    },
    {
      "epoch": 0.7684262087247162,
      "grad_norm": 0.7123553156852722,
      "learning_rate": 1.3416548897348586e-06,
      "loss": 0.7125,
      "step": 26669
    },
    {
      "epoch": 0.7684550221863655,
      "grad_norm": 0.7287203669548035,
      "learning_rate": 1.3413368361777441e-06,
      "loss": 0.692,
      "step": 26670
    },
    {
      "epoch": 0.7684838356480147,
      "grad_norm": 0.7435636520385742,
      "learning_rate": 1.3410188144839298e-06,
      "loss": 0.6839,
      "step": 26671
    },
    {
      "epoch": 0.768512649109664,
      "grad_norm": 0.7118750214576721,
      "learning_rate": 1.3407008246561826e-06,
      "loss": 0.6839,
      "step": 26672
    },
    {
      "epoch": 0.7685414625713133,
      "grad_norm": 0.73443603515625,
      "learning_rate": 1.3403828666972746e-06,
      "loss": 0.6952,
      "step": 26673
    },
    {
      "epoch": 0.7685702760329626,
      "grad_norm": 0.7137125730514526,
      "learning_rate": 1.340064940609972e-06,
      "loss": 0.6675,
      "step": 26674
    },
    {
      "epoch": 0.7685990894946119,
      "grad_norm": 0.7323870658874512,
      "learning_rate": 1.3397470463970463e-06,
      "loss": 0.6868,
      "step": 26675
    },
    {
      "epoch": 0.7686279029562612,
      "grad_norm": 0.7224226593971252,
      "learning_rate": 1.3394291840612638e-06,
      "loss": 0.6783,
      "step": 26676
    },
    {
      "epoch": 0.7686567164179104,
      "grad_norm": 0.7384994626045227,
      "learning_rate": 1.3391113536053956e-06,
      "loss": 0.6982,
      "step": 26677
    },
    {
      "epoch": 0.7686855298795597,
      "grad_norm": 0.7265287637710571,
      "learning_rate": 1.3387935550322061e-06,
      "loss": 0.6808,
      "step": 26678
    },
    {
      "epoch": 0.768714343341209,
      "grad_norm": 0.7180007100105286,
      "learning_rate": 1.3384757883444655e-06,
      "loss": 0.7106,
      "step": 26679
    },
    {
      "epoch": 0.7687431568028583,
      "grad_norm": 0.7381971478462219,
      "learning_rate": 1.33815805354494e-06,
      "loss": 0.6901,
      "step": 26680
    },
    {
      "epoch": 0.7687719702645076,
      "grad_norm": 0.7110780477523804,
      "learning_rate": 1.3378403506363991e-06,
      "loss": 0.6693,
      "step": 26681
    },
    {
      "epoch": 0.7688007837261569,
      "grad_norm": 0.7353168725967407,
      "learning_rate": 1.3375226796216063e-06,
      "loss": 0.6795,
      "step": 26682
    },
    {
      "epoch": 0.7688295971878062,
      "grad_norm": 0.7471719980239868,
      "learning_rate": 1.337205040503331e-06,
      "loss": 0.6847,
      "step": 26683
    },
    {
      "epoch": 0.7688584106494554,
      "grad_norm": 0.7406469583511353,
      "learning_rate": 1.3368874332843374e-06,
      "loss": 0.6921,
      "step": 26684
    },
    {
      "epoch": 0.7688872241111047,
      "grad_norm": 0.7311483025550842,
      "learning_rate": 1.3365698579673924e-06,
      "loss": 0.6995,
      "step": 26685
    },
    {
      "epoch": 0.768916037572754,
      "grad_norm": 0.7061871290206909,
      "learning_rate": 1.336252314555263e-06,
      "loss": 0.6932,
      "step": 26686
    },
    {
      "epoch": 0.7689448510344032,
      "grad_norm": 0.7375069856643677,
      "learning_rate": 1.3359348030507124e-06,
      "loss": 0.6616,
      "step": 26687
    },
    {
      "epoch": 0.7689736644960525,
      "grad_norm": 0.7190086245536804,
      "learning_rate": 1.3356173234565084e-06,
      "loss": 0.6868,
      "step": 26688
    },
    {
      "epoch": 0.7690024779577018,
      "grad_norm": 0.7478740811347961,
      "learning_rate": 1.3352998757754127e-06,
      "loss": 0.6916,
      "step": 26689
    },
    {
      "epoch": 0.7690312914193511,
      "grad_norm": 0.7207801342010498,
      "learning_rate": 1.3349824600101934e-06,
      "loss": 0.6823,
      "step": 26690
    },
    {
      "epoch": 0.7690601048810004,
      "grad_norm": 0.7094595432281494,
      "learning_rate": 1.3346650761636104e-06,
      "loss": 0.6814,
      "step": 26691
    },
    {
      "epoch": 0.7690889183426497,
      "grad_norm": 0.7300196886062622,
      "learning_rate": 1.3343477242384334e-06,
      "loss": 0.6628,
      "step": 26692
    },
    {
      "epoch": 0.769117731804299,
      "grad_norm": 0.7410570979118347,
      "learning_rate": 1.3340304042374219e-06,
      "loss": 0.6863,
      "step": 26693
    },
    {
      "epoch": 0.7691465452659483,
      "grad_norm": 0.7426881194114685,
      "learning_rate": 1.333713116163342e-06,
      "loss": 0.687,
      "step": 26694
    },
    {
      "epoch": 0.7691753587275976,
      "grad_norm": 0.7316652536392212,
      "learning_rate": 1.3333958600189551e-06,
      "loss": 0.7039,
      "step": 26695
    },
    {
      "epoch": 0.7692041721892469,
      "grad_norm": 0.7325885891914368,
      "learning_rate": 1.3330786358070264e-06,
      "loss": 0.6992,
      "step": 26696
    },
    {
      "epoch": 0.7692329856508962,
      "grad_norm": 0.7171334028244019,
      "learning_rate": 1.3327614435303154e-06,
      "loss": 0.6983,
      "step": 26697
    },
    {
      "epoch": 0.7692617991125453,
      "grad_norm": 0.7107271552085876,
      "learning_rate": 1.332444283191588e-06,
      "loss": 0.6725,
      "step": 26698
    },
    {
      "epoch": 0.7692906125741946,
      "grad_norm": 0.7092114090919495,
      "learning_rate": 1.3321271547936031e-06,
      "loss": 0.6761,
      "step": 26699
    },
    {
      "epoch": 0.7693194260358439,
      "grad_norm": 0.7105408310890198,
      "learning_rate": 1.331810058339126e-06,
      "loss": 0.6742,
      "step": 26700
    },
    {
      "epoch": 0.7693482394974932,
      "grad_norm": 0.7393097877502441,
      "learning_rate": 1.3314929938309145e-06,
      "loss": 0.6995,
      "step": 26701
    },
    {
      "epoch": 0.7693770529591425,
      "grad_norm": 0.7319861054420471,
      "learning_rate": 1.3311759612717318e-06,
      "loss": 0.6744,
      "step": 26702
    },
    {
      "epoch": 0.7694058664207918,
      "grad_norm": 0.7069929242134094,
      "learning_rate": 1.3308589606643396e-06,
      "loss": 0.6989,
      "step": 26703
    },
    {
      "epoch": 0.7694346798824411,
      "grad_norm": 0.7289763689041138,
      "learning_rate": 1.3305419920114988e-06,
      "loss": 0.6913,
      "step": 26704
    },
    {
      "epoch": 0.7694634933440904,
      "grad_norm": 0.7313492894172668,
      "learning_rate": 1.330225055315968e-06,
      "loss": 0.7045,
      "step": 26705
    },
    {
      "epoch": 0.7694923068057397,
      "grad_norm": 0.7236871123313904,
      "learning_rate": 1.3299081505805088e-06,
      "loss": 0.6961,
      "step": 26706
    },
    {
      "epoch": 0.769521120267389,
      "grad_norm": 0.7465895414352417,
      "learning_rate": 1.3295912778078818e-06,
      "loss": 0.7043,
      "step": 26707
    },
    {
      "epoch": 0.7695499337290382,
      "grad_norm": 0.7509599924087524,
      "learning_rate": 1.3292744370008447e-06,
      "loss": 0.6607,
      "step": 26708
    },
    {
      "epoch": 0.7695787471906875,
      "grad_norm": 0.7337960004806519,
      "learning_rate": 1.328957628162159e-06,
      "loss": 0.7087,
      "step": 26709
    },
    {
      "epoch": 0.7696075606523368,
      "grad_norm": 0.7267659902572632,
      "learning_rate": 1.3286408512945815e-06,
      "loss": 0.6768,
      "step": 26710
    },
    {
      "epoch": 0.769636374113986,
      "grad_norm": 0.7152334451675415,
      "learning_rate": 1.3283241064008734e-06,
      "loss": 0.6852,
      "step": 26711
    },
    {
      "epoch": 0.7696651875756353,
      "grad_norm": 0.7429664134979248,
      "learning_rate": 1.3280073934837906e-06,
      "loss": 0.6918,
      "step": 26712
    },
    {
      "epoch": 0.7696940010372846,
      "grad_norm": 0.7270377278327942,
      "learning_rate": 1.3276907125460942e-06,
      "loss": 0.7043,
      "step": 26713
    },
    {
      "epoch": 0.7697228144989339,
      "grad_norm": 0.7214694023132324,
      "learning_rate": 1.3273740635905398e-06,
      "loss": 0.6956,
      "step": 26714
    },
    {
      "epoch": 0.7697516279605832,
      "grad_norm": 0.7314718961715698,
      "learning_rate": 1.327057446619885e-06,
      "loss": 0.7033,
      "step": 26715
    },
    {
      "epoch": 0.7697804414222325,
      "grad_norm": 0.7176473140716553,
      "learning_rate": 1.3267408616368888e-06,
      "loss": 0.6971,
      "step": 26716
    },
    {
      "epoch": 0.7698092548838817,
      "grad_norm": 0.7233665585517883,
      "learning_rate": 1.326424308644309e-06,
      "loss": 0.6825,
      "step": 26717
    },
    {
      "epoch": 0.769838068345531,
      "grad_norm": 0.7138750553131104,
      "learning_rate": 1.3261077876449003e-06,
      "loss": 0.6931,
      "step": 26718
    },
    {
      "epoch": 0.7698668818071803,
      "grad_norm": 0.7001321911811829,
      "learning_rate": 1.3257912986414213e-06,
      "loss": 0.674,
      "step": 26719
    },
    {
      "epoch": 0.7698956952688296,
      "grad_norm": 0.7452879548072815,
      "learning_rate": 1.325474841636626e-06,
      "loss": 0.6806,
      "step": 26720
    },
    {
      "epoch": 0.7699245087304789,
      "grad_norm": 0.7062552571296692,
      "learning_rate": 1.3251584166332727e-06,
      "loss": 0.6699,
      "step": 26721
    },
    {
      "epoch": 0.7699533221921282,
      "grad_norm": 0.7347452640533447,
      "learning_rate": 1.3248420236341148e-06,
      "loss": 0.7058,
      "step": 26722
    },
    {
      "epoch": 0.7699821356537775,
      "grad_norm": 0.7290639877319336,
      "learning_rate": 1.3245256626419094e-06,
      "loss": 0.6897,
      "step": 26723
    },
    {
      "epoch": 0.7700109491154268,
      "grad_norm": 0.7147582173347473,
      "learning_rate": 1.3242093336594124e-06,
      "loss": 0.6891,
      "step": 26724
    },
    {
      "epoch": 0.770039762577076,
      "grad_norm": 0.7221472859382629,
      "learning_rate": 1.3238930366893765e-06,
      "loss": 0.6815,
      "step": 26725
    },
    {
      "epoch": 0.7700685760387252,
      "grad_norm": 0.7098608016967773,
      "learning_rate": 1.3235767717345589e-06,
      "loss": 0.6956,
      "step": 26726
    },
    {
      "epoch": 0.7700973895003745,
      "grad_norm": 0.7151528000831604,
      "learning_rate": 1.3232605387977093e-06,
      "loss": 0.6675,
      "step": 26727
    },
    {
      "epoch": 0.7701262029620238,
      "grad_norm": 0.7214788794517517,
      "learning_rate": 1.3229443378815882e-06,
      "loss": 0.6825,
      "step": 26728
    },
    {
      "epoch": 0.7701550164236731,
      "grad_norm": 0.7332983613014221,
      "learning_rate": 1.3226281689889448e-06,
      "loss": 0.7006,
      "step": 26729
    },
    {
      "epoch": 0.7701838298853224,
      "grad_norm": 0.7316521406173706,
      "learning_rate": 1.3223120321225352e-06,
      "loss": 0.6857,
      "step": 26730
    },
    {
      "epoch": 0.7702126433469717,
      "grad_norm": 0.7224646806716919,
      "learning_rate": 1.3219959272851102e-06,
      "loss": 0.6888,
      "step": 26731
    },
    {
      "epoch": 0.770241456808621,
      "grad_norm": 0.7393214106559753,
      "learning_rate": 1.3216798544794257e-06,
      "loss": 0.6924,
      "step": 26732
    },
    {
      "epoch": 0.7702702702702703,
      "grad_norm": 0.7206940650939941,
      "learning_rate": 1.3213638137082308e-06,
      "loss": 0.6665,
      "step": 26733
    },
    {
      "epoch": 0.7702990837319196,
      "grad_norm": 0.73771733045578,
      "learning_rate": 1.3210478049742813e-06,
      "loss": 0.7046,
      "step": 26734
    },
    {
      "epoch": 0.7703278971935689,
      "grad_norm": 0.7088169455528259,
      "learning_rate": 1.3207318282803266e-06,
      "loss": 0.6658,
      "step": 26735
    },
    {
      "epoch": 0.7703567106552182,
      "grad_norm": 0.733518123626709,
      "learning_rate": 1.3204158836291208e-06,
      "loss": 0.6854,
      "step": 26736
    },
    {
      "epoch": 0.7703855241168674,
      "grad_norm": 0.7195317149162292,
      "learning_rate": 1.3200999710234135e-06,
      "loss": 0.7049,
      "step": 26737
    },
    {
      "epoch": 0.7704143375785166,
      "grad_norm": 0.7338746190071106,
      "learning_rate": 1.319784090465958e-06,
      "loss": 0.6868,
      "step": 26738
    },
    {
      "epoch": 0.7704431510401659,
      "grad_norm": 0.7241266369819641,
      "learning_rate": 1.3194682419595017e-06,
      "loss": 0.6781,
      "step": 26739
    },
    {
      "epoch": 0.7704719645018152,
      "grad_norm": 0.7308343648910522,
      "learning_rate": 1.3191524255068e-06,
      "loss": 0.6897,
      "step": 26740
    },
    {
      "epoch": 0.7705007779634645,
      "grad_norm": 0.7182387113571167,
      "learning_rate": 1.3188366411106e-06,
      "loss": 0.7062,
      "step": 26741
    },
    {
      "epoch": 0.7705295914251138,
      "grad_norm": 0.7204962372779846,
      "learning_rate": 1.318520888773655e-06,
      "loss": 0.6574,
      "step": 26742
    },
    {
      "epoch": 0.7705584048867631,
      "grad_norm": 0.7197543382644653,
      "learning_rate": 1.3182051684987112e-06,
      "loss": 0.6961,
      "step": 26743
    },
    {
      "epoch": 0.7705872183484124,
      "grad_norm": 0.7249854803085327,
      "learning_rate": 1.31788948028852e-06,
      "loss": 0.6779,
      "step": 26744
    },
    {
      "epoch": 0.7706160318100617,
      "grad_norm": 0.7142763137817383,
      "learning_rate": 1.317573824145832e-06,
      "loss": 0.6764,
      "step": 26745
    },
    {
      "epoch": 0.770644845271711,
      "grad_norm": 0.7091799974441528,
      "learning_rate": 1.317258200073393e-06,
      "loss": 0.6532,
      "step": 26746
    },
    {
      "epoch": 0.7706736587333602,
      "grad_norm": 0.7216693162918091,
      "learning_rate": 1.3169426080739555e-06,
      "loss": 0.677,
      "step": 26747
    },
    {
      "epoch": 0.7707024721950095,
      "grad_norm": 0.7456215620040894,
      "learning_rate": 1.3166270481502652e-06,
      "loss": 0.6882,
      "step": 26748
    },
    {
      "epoch": 0.7707312856566588,
      "grad_norm": 0.7412437200546265,
      "learning_rate": 1.3163115203050725e-06,
      "loss": 0.6954,
      "step": 26749
    },
    {
      "epoch": 0.7707600991183081,
      "grad_norm": 0.7146896719932556,
      "learning_rate": 1.3159960245411225e-06,
      "loss": 0.6691,
      "step": 26750
    },
    {
      "epoch": 0.7707889125799574,
      "grad_norm": 0.723301351070404,
      "learning_rate": 1.3156805608611644e-06,
      "loss": 0.6833,
      "step": 26751
    },
    {
      "epoch": 0.7708177260416066,
      "grad_norm": 0.7235809564590454,
      "learning_rate": 1.3153651292679453e-06,
      "loss": 0.6932,
      "step": 26752
    },
    {
      "epoch": 0.7708465395032559,
      "grad_norm": 0.7344992160797119,
      "learning_rate": 1.3150497297642145e-06,
      "loss": 0.6911,
      "step": 26753
    },
    {
      "epoch": 0.7708753529649052,
      "grad_norm": 0.701956570148468,
      "learning_rate": 1.314734362352716e-06,
      "loss": 0.6795,
      "step": 26754
    },
    {
      "epoch": 0.7709041664265545,
      "grad_norm": 0.7292231917381287,
      "learning_rate": 1.314419027036198e-06,
      "loss": 0.6796,
      "step": 26755
    },
    {
      "epoch": 0.7709329798882038,
      "grad_norm": 0.746849536895752,
      "learning_rate": 1.314103723817405e-06,
      "loss": 0.6604,
      "step": 26756
    },
    {
      "epoch": 0.770961793349853,
      "grad_norm": 0.7307058572769165,
      "learning_rate": 1.3137884526990852e-06,
      "loss": 0.6744,
      "step": 26757
    },
    {
      "epoch": 0.7709906068115023,
      "grad_norm": 0.7071429491043091,
      "learning_rate": 1.3134732136839818e-06,
      "loss": 0.6827,
      "step": 26758
    },
    {
      "epoch": 0.7710194202731516,
      "grad_norm": 0.7244205474853516,
      "learning_rate": 1.3131580067748428e-06,
      "loss": 0.6917,
      "step": 26759
    },
    {
      "epoch": 0.7710482337348009,
      "grad_norm": 0.6997487545013428,
      "learning_rate": 1.3128428319744108e-06,
      "loss": 0.6495,
      "step": 26760
    },
    {
      "epoch": 0.7710770471964502,
      "grad_norm": 0.7218059301376343,
      "learning_rate": 1.3125276892854316e-06,
      "loss": 0.6657,
      "step": 26761
    },
    {
      "epoch": 0.7711058606580995,
      "grad_norm": 0.7123197913169861,
      "learning_rate": 1.3122125787106522e-06,
      "loss": 0.6984,
      "step": 26762
    },
    {
      "epoch": 0.7711346741197488,
      "grad_norm": 0.7096965909004211,
      "learning_rate": 1.3118975002528128e-06,
      "loss": 0.6653,
      "step": 26763
    },
    {
      "epoch": 0.7711634875813981,
      "grad_norm": 0.7256000638008118,
      "learning_rate": 1.31158245391466e-06,
      "loss": 0.6745,
      "step": 26764
    },
    {
      "epoch": 0.7711923010430473,
      "grad_norm": 0.7417563796043396,
      "learning_rate": 1.311267439698936e-06,
      "loss": 0.6729,
      "step": 26765
    },
    {
      "epoch": 0.7712211145046965,
      "grad_norm": 0.6961289048194885,
      "learning_rate": 1.3109524576083876e-06,
      "loss": 0.6543,
      "step": 26766
    },
    {
      "epoch": 0.7712499279663458,
      "grad_norm": 0.71349036693573,
      "learning_rate": 1.3106375076457534e-06,
      "loss": 0.6745,
      "step": 26767
    },
    {
      "epoch": 0.7712787414279951,
      "grad_norm": 0.7262557744979858,
      "learning_rate": 1.3103225898137806e-06,
      "loss": 0.6975,
      "step": 26768
    },
    {
      "epoch": 0.7713075548896444,
      "grad_norm": 0.7379556894302368,
      "learning_rate": 1.310007704115208e-06,
      "loss": 0.6819,
      "step": 26769
    },
    {
      "epoch": 0.7713363683512937,
      "grad_norm": 0.6915363669395447,
      "learning_rate": 1.3096928505527812e-06,
      "loss": 0.6796,
      "step": 26770
    },
    {
      "epoch": 0.771365181812943,
      "grad_norm": 0.7357663512229919,
      "learning_rate": 1.3093780291292392e-06,
      "loss": 0.6827,
      "step": 26771
    },
    {
      "epoch": 0.7713939952745923,
      "grad_norm": 0.7506394386291504,
      "learning_rate": 1.3090632398473263e-06,
      "loss": 0.7066,
      "step": 26772
    },
    {
      "epoch": 0.7714228087362416,
      "grad_norm": 0.7157366871833801,
      "learning_rate": 1.3087484827097824e-06,
      "loss": 0.6868,
      "step": 26773
    },
    {
      "epoch": 0.7714516221978909,
      "grad_norm": 0.706134557723999,
      "learning_rate": 1.3084337577193501e-06,
      "loss": 0.6696,
      "step": 26774
    },
    {
      "epoch": 0.7714804356595402,
      "grad_norm": 0.7079234719276428,
      "learning_rate": 1.3081190648787668e-06,
      "loss": 0.6841,
      "step": 26775
    },
    {
      "epoch": 0.7715092491211895,
      "grad_norm": 0.7301884293556213,
      "learning_rate": 1.307804404190779e-06,
      "loss": 0.6654,
      "step": 26776
    },
    {
      "epoch": 0.7715380625828387,
      "grad_norm": 0.7341542840003967,
      "learning_rate": 1.3074897756581222e-06,
      "loss": 0.7082,
      "step": 26777
    },
    {
      "epoch": 0.771566876044488,
      "grad_norm": 0.7488166689872742,
      "learning_rate": 1.3071751792835403e-06,
      "loss": 0.6897,
      "step": 26778
    },
    {
      "epoch": 0.7715956895061372,
      "grad_norm": 0.7275780439376831,
      "learning_rate": 1.3068606150697694e-06,
      "loss": 0.6913,
      "step": 26779
    },
    {
      "epoch": 0.7716245029677865,
      "grad_norm": 0.7264128923416138,
      "learning_rate": 1.306546083019552e-06,
      "loss": 0.712,
      "step": 26780
    },
    {
      "epoch": 0.7716533164294358,
      "grad_norm": 0.7206423282623291,
      "learning_rate": 1.3062315831356248e-06,
      "loss": 0.6917,
      "step": 26781
    },
    {
      "epoch": 0.7716821298910851,
      "grad_norm": 0.7376484870910645,
      "learning_rate": 1.3059171154207278e-06,
      "loss": 0.693,
      "step": 26782
    },
    {
      "epoch": 0.7717109433527344,
      "grad_norm": 0.7360013723373413,
      "learning_rate": 1.3056026798776023e-06,
      "loss": 0.6709,
      "step": 26783
    },
    {
      "epoch": 0.7717397568143837,
      "grad_norm": 0.7131718993186951,
      "learning_rate": 1.305288276508982e-06,
      "loss": 0.686,
      "step": 26784
    },
    {
      "epoch": 0.771768570276033,
      "grad_norm": 0.7274304032325745,
      "learning_rate": 1.3049739053176097e-06,
      "loss": 0.6722,
      "step": 26785
    },
    {
      "epoch": 0.7717973837376823,
      "grad_norm": 0.7240880131721497,
      "learning_rate": 1.304659566306219e-06,
      "loss": 0.686,
      "step": 26786
    },
    {
      "epoch": 0.7718261971993315,
      "grad_norm": 0.743584156036377,
      "learning_rate": 1.3043452594775497e-06,
      "loss": 0.6958,
      "step": 26787
    },
    {
      "epoch": 0.7718550106609808,
      "grad_norm": 0.7210898399353027,
      "learning_rate": 1.3040309848343386e-06,
      "loss": 0.6963,
      "step": 26788
    },
    {
      "epoch": 0.7718838241226301,
      "grad_norm": 0.7367004156112671,
      "learning_rate": 1.3037167423793251e-06,
      "loss": 0.697,
      "step": 26789
    },
    {
      "epoch": 0.7719126375842794,
      "grad_norm": 0.7054991722106934,
      "learning_rate": 1.3034025321152415e-06,
      "loss": 0.6595,
      "step": 26790
    },
    {
      "epoch": 0.7719414510459287,
      "grad_norm": 0.7337292432785034,
      "learning_rate": 1.3030883540448285e-06,
      "loss": 0.7232,
      "step": 26791
    },
    {
      "epoch": 0.7719702645075779,
      "grad_norm": 0.7357632517814636,
      "learning_rate": 1.3027742081708195e-06,
      "loss": 0.708,
      "step": 26792
    },
    {
      "epoch": 0.7719990779692272,
      "grad_norm": 0.7255117297172546,
      "learning_rate": 1.302460094495952e-06,
      "loss": 0.7151,
      "step": 26793
    },
    {
      "epoch": 0.7720278914308765,
      "grad_norm": 0.7192819118499756,
      "learning_rate": 1.3021460130229596e-06,
      "loss": 0.6923,
      "step": 26794
    },
    {
      "epoch": 0.7720567048925258,
      "grad_norm": 0.7219495177268982,
      "learning_rate": 1.3018319637545806e-06,
      "loss": 0.6974,
      "step": 26795
    },
    {
      "epoch": 0.772085518354175,
      "grad_norm": 0.7181103825569153,
      "learning_rate": 1.3015179466935468e-06,
      "loss": 0.6835,
      "step": 26796
    },
    {
      "epoch": 0.7721143318158243,
      "grad_norm": 0.7362648844718933,
      "learning_rate": 1.301203961842596e-06,
      "loss": 0.7076,
      "step": 26797
    },
    {
      "epoch": 0.7721431452774736,
      "grad_norm": 0.763886034488678,
      "learning_rate": 1.3008900092044595e-06,
      "loss": 0.6885,
      "step": 26798
    },
    {
      "epoch": 0.7721719587391229,
      "grad_norm": 0.7180325984954834,
      "learning_rate": 1.3005760887818736e-06,
      "loss": 0.7103,
      "step": 26799
    },
    {
      "epoch": 0.7722007722007722,
      "grad_norm": 0.7131702303886414,
      "learning_rate": 1.3002622005775716e-06,
      "loss": 0.6759,
      "step": 26800
    },
    {
      "epoch": 0.7722295856624215,
      "grad_norm": 0.7218391299247742,
      "learning_rate": 1.299948344594289e-06,
      "loss": 0.6839,
      "step": 26801
    },
    {
      "epoch": 0.7722583991240708,
      "grad_norm": 0.715451717376709,
      "learning_rate": 1.2996345208347566e-06,
      "loss": 0.6948,
      "step": 26802
    },
    {
      "epoch": 0.7722872125857201,
      "grad_norm": 0.7153247594833374,
      "learning_rate": 1.299320729301708e-06,
      "loss": 0.6827,
      "step": 26803
    },
    {
      "epoch": 0.7723160260473694,
      "grad_norm": 0.7657586336135864,
      "learning_rate": 1.299006969997878e-06,
      "loss": 0.7095,
      "step": 26804
    },
    {
      "epoch": 0.7723448395090187,
      "grad_norm": 0.701811671257019,
      "learning_rate": 1.298693242925997e-06,
      "loss": 0.6635,
      "step": 26805
    },
    {
      "epoch": 0.7723736529706678,
      "grad_norm": 0.7272332906723022,
      "learning_rate": 1.2983795480887983e-06,
      "loss": 0.6968,
      "step": 26806
    },
    {
      "epoch": 0.7724024664323171,
      "grad_norm": 0.7331615090370178,
      "learning_rate": 1.2980658854890126e-06,
      "loss": 0.6877,
      "step": 26807
    },
    {
      "epoch": 0.7724312798939664,
      "grad_norm": 0.712161123752594,
      "learning_rate": 1.2977522551293737e-06,
      "loss": 0.6687,
      "step": 26808
    },
    {
      "epoch": 0.7724600933556157,
      "grad_norm": 0.7154074907302856,
      "learning_rate": 1.2974386570126102e-06,
      "loss": 0.6855,
      "step": 26809
    },
    {
      "epoch": 0.772488906817265,
      "grad_norm": 0.7113009095191956,
      "learning_rate": 1.2971250911414567e-06,
      "loss": 0.6766,
      "step": 26810
    },
    {
      "epoch": 0.7725177202789143,
      "grad_norm": 0.727229654788971,
      "learning_rate": 1.2968115575186386e-06,
      "loss": 0.6886,
      "step": 26811
    },
    {
      "epoch": 0.7725465337405636,
      "grad_norm": 0.7345848083496094,
      "learning_rate": 1.2964980561468932e-06,
      "loss": 0.6785,
      "step": 26812
    },
    {
      "epoch": 0.7725753472022129,
      "grad_norm": 0.775905191898346,
      "learning_rate": 1.2961845870289464e-06,
      "loss": 0.6933,
      "step": 26813
    },
    {
      "epoch": 0.7726041606638622,
      "grad_norm": 0.7275015115737915,
      "learning_rate": 1.2958711501675307e-06,
      "loss": 0.7095,
      "step": 26814
    },
    {
      "epoch": 0.7726329741255115,
      "grad_norm": 0.7248541116714478,
      "learning_rate": 1.2955577455653727e-06,
      "loss": 0.689,
      "step": 26815
    },
    {
      "epoch": 0.7726617875871608,
      "grad_norm": 0.7069190740585327,
      "learning_rate": 1.2952443732252058e-06,
      "loss": 0.682,
      "step": 26816
    },
    {
      "epoch": 0.77269060104881,
      "grad_norm": 0.7307320833206177,
      "learning_rate": 1.294931033149755e-06,
      "loss": 0.6881,
      "step": 26817
    },
    {
      "epoch": 0.7727194145104593,
      "grad_norm": 0.7188011407852173,
      "learning_rate": 1.2946177253417525e-06,
      "loss": 0.6814,
      "step": 26818
    },
    {
      "epoch": 0.7727482279721085,
      "grad_norm": 0.705743670463562,
      "learning_rate": 1.2943044498039249e-06,
      "loss": 0.6883,
      "step": 26819
    },
    {
      "epoch": 0.7727770414337578,
      "grad_norm": 0.7444191575050354,
      "learning_rate": 1.2939912065390004e-06,
      "loss": 0.6585,
      "step": 26820
    },
    {
      "epoch": 0.7728058548954071,
      "grad_norm": 0.7206732034683228,
      "learning_rate": 1.2936779955497097e-06,
      "loss": 0.6928,
      "step": 26821
    },
    {
      "epoch": 0.7728346683570564,
      "grad_norm": 0.7196761965751648,
      "learning_rate": 1.293364816838777e-06,
      "loss": 0.6926,
      "step": 26822
    },
    {
      "epoch": 0.7728634818187057,
      "grad_norm": 0.7145606279373169,
      "learning_rate": 1.2930516704089314e-06,
      "loss": 0.6744,
      "step": 26823
    },
    {
      "epoch": 0.772892295280355,
      "grad_norm": 0.7118765115737915,
      "learning_rate": 1.2927385562629002e-06,
      "loss": 0.6899,
      "step": 26824
    },
    {
      "epoch": 0.7729211087420043,
      "grad_norm": 0.7202528119087219,
      "learning_rate": 1.2924254744034121e-06,
      "loss": 0.6816,
      "step": 26825
    },
    {
      "epoch": 0.7729499222036535,
      "grad_norm": 0.720376193523407,
      "learning_rate": 1.2921124248331902e-06,
      "loss": 0.6564,
      "step": 26826
    },
    {
      "epoch": 0.7729787356653028,
      "grad_norm": 0.717585563659668,
      "learning_rate": 1.2917994075549638e-06,
      "loss": 0.6747,
      "step": 26827
    },
    {
      "epoch": 0.7730075491269521,
      "grad_norm": 0.7279205322265625,
      "learning_rate": 1.2914864225714568e-06,
      "loss": 0.6934,
      "step": 26828
    },
    {
      "epoch": 0.7730363625886014,
      "grad_norm": 0.7065817713737488,
      "learning_rate": 1.2911734698853972e-06,
      "loss": 0.671,
      "step": 26829
    },
    {
      "epoch": 0.7730651760502507,
      "grad_norm": 0.7046272158622742,
      "learning_rate": 1.2908605494995075e-06,
      "loss": 0.6996,
      "step": 26830
    },
    {
      "epoch": 0.7730939895119,
      "grad_norm": 0.7422188520431519,
      "learning_rate": 1.2905476614165163e-06,
      "loss": 0.6925,
      "step": 26831
    },
    {
      "epoch": 0.7731228029735493,
      "grad_norm": 0.7187691330909729,
      "learning_rate": 1.2902348056391451e-06,
      "loss": 0.6941,
      "step": 26832
    },
    {
      "epoch": 0.7731516164351985,
      "grad_norm": 0.7226585745811462,
      "learning_rate": 1.289921982170122e-06,
      "loss": 0.6676,
      "step": 26833
    },
    {
      "epoch": 0.7731804298968478,
      "grad_norm": 0.7079898715019226,
      "learning_rate": 1.2896091910121667e-06,
      "loss": 0.6725,
      "step": 26834
    },
    {
      "epoch": 0.7732092433584971,
      "grad_norm": 0.7346985340118408,
      "learning_rate": 1.2892964321680097e-06,
      "loss": 0.6798,
      "step": 26835
    },
    {
      "epoch": 0.7732380568201463,
      "grad_norm": 0.6922436952590942,
      "learning_rate": 1.2889837056403693e-06,
      "loss": 0.6961,
      "step": 26836
    },
    {
      "epoch": 0.7732668702817956,
      "grad_norm": 0.7298067212104797,
      "learning_rate": 1.2886710114319729e-06,
      "loss": 0.6905,
      "step": 26837
    },
    {
      "epoch": 0.7732956837434449,
      "grad_norm": 0.732062578201294,
      "learning_rate": 1.288358349545541e-06,
      "loss": 0.7134,
      "step": 26838
    },
    {
      "epoch": 0.7733244972050942,
      "grad_norm": 0.7213756442070007,
      "learning_rate": 1.2880457199837982e-06,
      "loss": 0.7067,
      "step": 26839
    },
    {
      "epoch": 0.7733533106667435,
      "grad_norm": 0.7204465270042419,
      "learning_rate": 1.2877331227494654e-06,
      "loss": 0.6693,
      "step": 26840
    },
    {
      "epoch": 0.7733821241283928,
      "grad_norm": 0.7123950123786926,
      "learning_rate": 1.2874205578452663e-06,
      "loss": 0.6972,
      "step": 26841
    },
    {
      "epoch": 0.7734109375900421,
      "grad_norm": 0.7051505446434021,
      "learning_rate": 1.2871080252739249e-06,
      "loss": 0.6821,
      "step": 26842
    },
    {
      "epoch": 0.7734397510516914,
      "grad_norm": 0.7191348075866699,
      "learning_rate": 1.2867955250381591e-06,
      "loss": 0.6858,
      "step": 26843
    },
    {
      "epoch": 0.7734685645133407,
      "grad_norm": 0.7219435572624207,
      "learning_rate": 1.2864830571406938e-06,
      "loss": 0.6957,
      "step": 26844
    },
    {
      "epoch": 0.77349737797499,
      "grad_norm": 0.7434481978416443,
      "learning_rate": 1.2861706215842479e-06,
      "loss": 0.6719,
      "step": 26845
    },
    {
      "epoch": 0.7735261914366391,
      "grad_norm": 0.7261319160461426,
      "learning_rate": 1.2858582183715428e-06,
      "loss": 0.7114,
      "step": 26846
    },
    {
      "epoch": 0.7735550048982884,
      "grad_norm": 0.7334449887275696,
      "learning_rate": 1.2855458475053e-06,
      "loss": 0.6926,
      "step": 26847
    },
    {
      "epoch": 0.7735838183599377,
      "grad_norm": 0.7097625732421875,
      "learning_rate": 1.2852335089882418e-06,
      "loss": 0.6745,
      "step": 26848
    },
    {
      "epoch": 0.773612631821587,
      "grad_norm": 0.71201092004776,
      "learning_rate": 1.284921202823084e-06,
      "loss": 0.6944,
      "step": 26849
    },
    {
      "epoch": 0.7736414452832363,
      "grad_norm": 0.7313361763954163,
      "learning_rate": 1.2846089290125507e-06,
      "loss": 0.673,
      "step": 26850
    },
    {
      "epoch": 0.7736702587448856,
      "grad_norm": 0.7624863386154175,
      "learning_rate": 1.2842966875593582e-06,
      "loss": 0.6894,
      "step": 26851
    },
    {
      "epoch": 0.7736990722065349,
      "grad_norm": 0.7396526336669922,
      "learning_rate": 1.2839844784662287e-06,
      "loss": 0.7088,
      "step": 26852
    },
    {
      "epoch": 0.7737278856681842,
      "grad_norm": 0.7222822904586792,
      "learning_rate": 1.2836723017358782e-06,
      "loss": 0.6759,
      "step": 26853
    },
    {
      "epoch": 0.7737566991298335,
      "grad_norm": 0.7138808965682983,
      "learning_rate": 1.2833601573710287e-06,
      "loss": 0.6566,
      "step": 26854
    },
    {
      "epoch": 0.7737855125914828,
      "grad_norm": 0.701643705368042,
      "learning_rate": 1.2830480453743949e-06,
      "loss": 0.6554,
      "step": 26855
    },
    {
      "epoch": 0.773814326053132,
      "grad_norm": 0.708276629447937,
      "learning_rate": 1.2827359657486987e-06,
      "loss": 0.7023,
      "step": 26856
    },
    {
      "epoch": 0.7738431395147813,
      "grad_norm": 0.7510649561882019,
      "learning_rate": 1.2824239184966552e-06,
      "loss": 0.7066,
      "step": 26857
    },
    {
      "epoch": 0.7738719529764306,
      "grad_norm": 0.7380115985870361,
      "learning_rate": 1.2821119036209828e-06,
      "loss": 0.6871,
      "step": 26858
    },
    {
      "epoch": 0.7739007664380799,
      "grad_norm": 0.7178215384483337,
      "learning_rate": 1.2817999211243993e-06,
      "loss": 0.6922,
      "step": 26859
    },
    {
      "epoch": 0.7739295798997291,
      "grad_norm": 0.7097997665405273,
      "learning_rate": 1.2814879710096234e-06,
      "loss": 0.681,
      "step": 26860
    },
    {
      "epoch": 0.7739583933613784,
      "grad_norm": 0.701380729675293,
      "learning_rate": 1.2811760532793682e-06,
      "loss": 0.6781,
      "step": 26861
    },
    {
      "epoch": 0.7739872068230277,
      "grad_norm": 0.7201482057571411,
      "learning_rate": 1.2808641679363526e-06,
      "loss": 0.6837,
      "step": 26862
    },
    {
      "epoch": 0.774016020284677,
      "grad_norm": 0.733973503112793,
      "learning_rate": 1.2805523149832933e-06,
      "loss": 0.6768,
      "step": 26863
    },
    {
      "epoch": 0.7740448337463263,
      "grad_norm": 0.7240371704101562,
      "learning_rate": 1.280240494422904e-06,
      "loss": 0.7217,
      "step": 26864
    },
    {
      "epoch": 0.7740736472079756,
      "grad_norm": 0.7262534499168396,
      "learning_rate": 1.279928706257903e-06,
      "loss": 0.708,
      "step": 26865
    },
    {
      "epoch": 0.7741024606696248,
      "grad_norm": 0.7071489095687866,
      "learning_rate": 1.2796169504910028e-06,
      "loss": 0.6724,
      "step": 26866
    },
    {
      "epoch": 0.7741312741312741,
      "grad_norm": 0.7249864339828491,
      "learning_rate": 1.2793052271249217e-06,
      "loss": 0.6738,
      "step": 26867
    },
    {
      "epoch": 0.7741600875929234,
      "grad_norm": 0.7116022706031799,
      "learning_rate": 1.2789935361623707e-06,
      "loss": 0.6675,
      "step": 26868
    },
    {
      "epoch": 0.7741889010545727,
      "grad_norm": 0.7233498096466064,
      "learning_rate": 1.2786818776060683e-06,
      "loss": 0.6665,
      "step": 26869
    },
    {
      "epoch": 0.774217714516222,
      "grad_norm": 0.7348502278327942,
      "learning_rate": 1.2783702514587237e-06,
      "loss": 0.6612,
      "step": 26870
    },
    {
      "epoch": 0.7742465279778713,
      "grad_norm": 0.6916890144348145,
      "learning_rate": 1.278058657723057e-06,
      "loss": 0.6826,
      "step": 26871
    },
    {
      "epoch": 0.7742753414395206,
      "grad_norm": 0.7219725251197815,
      "learning_rate": 1.2777470964017768e-06,
      "loss": 0.6904,
      "step": 26872
    },
    {
      "epoch": 0.7743041549011698,
      "grad_norm": 0.7323154807090759,
      "learning_rate": 1.2774355674976007e-06,
      "loss": 0.684,
      "step": 26873
    },
    {
      "epoch": 0.7743329683628191,
      "grad_norm": 0.7395747303962708,
      "learning_rate": 1.2771240710132376e-06,
      "loss": 0.7116,
      "step": 26874
    },
    {
      "epoch": 0.7743617818244684,
      "grad_norm": 0.719668984413147,
      "learning_rate": 1.2768126069514041e-06,
      "loss": 0.652,
      "step": 26875
    },
    {
      "epoch": 0.7743905952861176,
      "grad_norm": 0.7289793491363525,
      "learning_rate": 1.2765011753148092e-06,
      "loss": 0.6855,
      "step": 26876
    },
    {
      "epoch": 0.7744194087477669,
      "grad_norm": 0.7066945433616638,
      "learning_rate": 1.2761897761061687e-06,
      "loss": 0.6754,
      "step": 26877
    },
    {
      "epoch": 0.7744482222094162,
      "grad_norm": 0.7294703125953674,
      "learning_rate": 1.2758784093281912e-06,
      "loss": 0.7065,
      "step": 26878
    },
    {
      "epoch": 0.7744770356710655,
      "grad_norm": 0.715176522731781,
      "learning_rate": 1.2755670749835902e-06,
      "loss": 0.6609,
      "step": 26879
    },
    {
      "epoch": 0.7745058491327148,
      "grad_norm": 0.7566356062889099,
      "learning_rate": 1.2752557730750781e-06,
      "loss": 0.7031,
      "step": 26880
    },
    {
      "epoch": 0.7745346625943641,
      "grad_norm": 0.7274821996688843,
      "learning_rate": 1.2749445036053637e-06,
      "loss": 0.7013,
      "step": 26881
    },
    {
      "epoch": 0.7745634760560134,
      "grad_norm": 0.7301855087280273,
      "learning_rate": 1.2746332665771587e-06,
      "loss": 0.7036,
      "step": 26882
    },
    {
      "epoch": 0.7745922895176627,
      "grad_norm": 0.7250544428825378,
      "learning_rate": 1.2743220619931757e-06,
      "loss": 0.6879,
      "step": 26883
    },
    {
      "epoch": 0.774621102979312,
      "grad_norm": 0.7056683301925659,
      "learning_rate": 1.274010889856121e-06,
      "loss": 0.6817,
      "step": 26884
    },
    {
      "epoch": 0.7746499164409613,
      "grad_norm": 0.6985870599746704,
      "learning_rate": 1.2736997501687077e-06,
      "loss": 0.6764,
      "step": 26885
    },
    {
      "epoch": 0.7746787299026106,
      "grad_norm": 0.7168278098106384,
      "learning_rate": 1.2733886429336457e-06,
      "loss": 0.7032,
      "step": 26886
    },
    {
      "epoch": 0.7747075433642597,
      "grad_norm": 0.7471373081207275,
      "learning_rate": 1.2730775681536417e-06,
      "loss": 0.7046,
      "step": 26887
    },
    {
      "epoch": 0.774736356825909,
      "grad_norm": 0.7113376259803772,
      "learning_rate": 1.2727665258314087e-06,
      "loss": 0.6829,
      "step": 26888
    },
    {
      "epoch": 0.7747651702875583,
      "grad_norm": 0.7356946468353271,
      "learning_rate": 1.2724555159696511e-06,
      "loss": 0.6807,
      "step": 26889
    },
    {
      "epoch": 0.7747939837492076,
      "grad_norm": 0.7206911444664001,
      "learning_rate": 1.2721445385710817e-06,
      "loss": 0.7068,
      "step": 26890
    },
    {
      "epoch": 0.7748227972108569,
      "grad_norm": 0.7100422382354736,
      "learning_rate": 1.2718335936384052e-06,
      "loss": 0.6876,
      "step": 26891
    },
    {
      "epoch": 0.7748516106725062,
      "grad_norm": 0.7166352868080139,
      "learning_rate": 1.2715226811743332e-06,
      "loss": 0.693,
      "step": 26892
    },
    {
      "epoch": 0.7748804241341555,
      "grad_norm": 0.7333381772041321,
      "learning_rate": 1.2712118011815695e-06,
      "loss": 0.6666,
      "step": 26893
    },
    {
      "epoch": 0.7749092375958048,
      "grad_norm": 0.7162715196609497,
      "learning_rate": 1.2709009536628242e-06,
      "loss": 0.6742,
      "step": 26894
    },
    {
      "epoch": 0.7749380510574541,
      "grad_norm": 0.7071680426597595,
      "learning_rate": 1.2705901386208031e-06,
      "loss": 0.6925,
      "step": 26895
    },
    {
      "epoch": 0.7749668645191033,
      "grad_norm": 0.7493194341659546,
      "learning_rate": 1.2702793560582162e-06,
      "loss": 0.7041,
      "step": 26896
    },
    {
      "epoch": 0.7749956779807526,
      "grad_norm": 0.7414596676826477,
      "learning_rate": 1.2699686059777656e-06,
      "loss": 0.6899,
      "step": 26897
    },
    {
      "epoch": 0.7750244914424019,
      "grad_norm": 0.7133228182792664,
      "learning_rate": 1.2696578883821614e-06,
      "loss": 0.6823,
      "step": 26898
    },
    {
      "epoch": 0.7750533049040512,
      "grad_norm": 0.7572547197341919,
      "learning_rate": 1.2693472032741067e-06,
      "loss": 0.692,
      "step": 26899
    },
    {
      "epoch": 0.7750821183657004,
      "grad_norm": 0.7179920077323914,
      "learning_rate": 1.2690365506563085e-06,
      "loss": 0.702,
      "step": 26900
    },
    {
      "epoch": 0.7751109318273497,
      "grad_norm": 0.7194307446479797,
      "learning_rate": 1.2687259305314736e-06,
      "loss": 0.6854,
      "step": 26901
    },
    {
      "epoch": 0.775139745288999,
      "grad_norm": 0.7717147469520569,
      "learning_rate": 1.268415342902305e-06,
      "loss": 0.6761,
      "step": 26902
    },
    {
      "epoch": 0.7751685587506483,
      "grad_norm": 0.7457696795463562,
      "learning_rate": 1.268104787771509e-06,
      "loss": 0.701,
      "step": 26903
    },
    {
      "epoch": 0.7751973722122976,
      "grad_norm": 0.7207621335983276,
      "learning_rate": 1.267794265141789e-06,
      "loss": 0.6787,
      "step": 26904
    },
    {
      "epoch": 0.7752261856739469,
      "grad_norm": 0.7293683290481567,
      "learning_rate": 1.267483775015852e-06,
      "loss": 0.7016,
      "step": 26905
    },
    {
      "epoch": 0.7752549991355961,
      "grad_norm": 0.7602719664573669,
      "learning_rate": 1.2671733173963968e-06,
      "loss": 0.6667,
      "step": 26906
    },
    {
      "epoch": 0.7752838125972454,
      "grad_norm": 0.7104709148406982,
      "learning_rate": 1.2668628922861332e-06,
      "loss": 0.6939,
      "step": 26907
    },
    {
      "epoch": 0.7753126260588947,
      "grad_norm": 0.7130581140518188,
      "learning_rate": 1.266552499687761e-06,
      "loss": 0.6739,
      "step": 26908
    },
    {
      "epoch": 0.775341439520544,
      "grad_norm": 0.7156445980072021,
      "learning_rate": 1.2662421396039859e-06,
      "loss": 0.7022,
      "step": 26909
    },
    {
      "epoch": 0.7753702529821933,
      "grad_norm": 0.7044747471809387,
      "learning_rate": 1.2659318120375075e-06,
      "loss": 0.6937,
      "step": 26910
    },
    {
      "epoch": 0.7753990664438426,
      "grad_norm": 0.7473044991493225,
      "learning_rate": 1.2656215169910319e-06,
      "loss": 0.6777,
      "step": 26911
    },
    {
      "epoch": 0.7754278799054919,
      "grad_norm": 0.7284888029098511,
      "learning_rate": 1.2653112544672585e-06,
      "loss": 0.7028,
      "step": 26912
    },
    {
      "epoch": 0.7754566933671412,
      "grad_norm": 0.7316250801086426,
      "learning_rate": 1.2650010244688925e-06,
      "loss": 0.6825,
      "step": 26913
    },
    {
      "epoch": 0.7754855068287904,
      "grad_norm": 0.7013718485832214,
      "learning_rate": 1.2646908269986318e-06,
      "loss": 0.6815,
      "step": 26914
    },
    {
      "epoch": 0.7755143202904397,
      "grad_norm": 0.7085960507392883,
      "learning_rate": 1.2643806620591825e-06,
      "loss": 0.67,
      "step": 26915
    },
    {
      "epoch": 0.7755431337520889,
      "grad_norm": 0.7307189106941223,
      "learning_rate": 1.2640705296532414e-06,
      "loss": 0.7048,
      "step": 26916
    },
    {
      "epoch": 0.7755719472137382,
      "grad_norm": 0.7026695609092712,
      "learning_rate": 1.263760429783511e-06,
      "loss": 0.6724,
      "step": 26917
    },
    {
      "epoch": 0.7756007606753875,
      "grad_norm": 0.7251027822494507,
      "learning_rate": 1.2634503624526927e-06,
      "loss": 0.7037,
      "step": 26918
    },
    {
      "epoch": 0.7756295741370368,
      "grad_norm": 0.7070353031158447,
      "learning_rate": 1.2631403276634884e-06,
      "loss": 0.6873,
      "step": 26919
    },
    {
      "epoch": 0.7756583875986861,
      "grad_norm": 0.7277635931968689,
      "learning_rate": 1.2628303254185948e-06,
      "loss": 0.6911,
      "step": 26920
    },
    {
      "epoch": 0.7756872010603354,
      "grad_norm": 0.7272870540618896,
      "learning_rate": 1.2625203557207143e-06,
      "loss": 0.6978,
      "step": 26921
    },
    {
      "epoch": 0.7757160145219847,
      "grad_norm": 0.7306079268455505,
      "learning_rate": 1.2622104185725441e-06,
      "loss": 0.6886,
      "step": 26922
    },
    {
      "epoch": 0.775744827983634,
      "grad_norm": 0.7141600847244263,
      "learning_rate": 1.2619005139767848e-06,
      "loss": 0.6988,
      "step": 26923
    },
    {
      "epoch": 0.7757736414452833,
      "grad_norm": 0.7257081866264343,
      "learning_rate": 1.2615906419361367e-06,
      "loss": 0.6916,
      "step": 26924
    },
    {
      "epoch": 0.7758024549069326,
      "grad_norm": 0.7211412191390991,
      "learning_rate": 1.2612808024532957e-06,
      "loss": 0.6969,
      "step": 26925
    },
    {
      "epoch": 0.7758312683685819,
      "grad_norm": 0.7329844832420349,
      "learning_rate": 1.2609709955309628e-06,
      "loss": 0.6845,
      "step": 26926
    },
    {
      "epoch": 0.775860081830231,
      "grad_norm": 0.7210153937339783,
      "learning_rate": 1.2606612211718339e-06,
      "loss": 0.6703,
      "step": 26927
    },
    {
      "epoch": 0.7758888952918803,
      "grad_norm": 0.7030201554298401,
      "learning_rate": 1.2603514793786086e-06,
      "loss": 0.6604,
      "step": 26928
    },
    {
      "epoch": 0.7759177087535296,
      "grad_norm": 0.7136049866676331,
      "learning_rate": 1.2600417701539825e-06,
      "loss": 0.6714,
      "step": 26929
    },
    {
      "epoch": 0.7759465222151789,
      "grad_norm": 0.7318877577781677,
      "learning_rate": 1.2597320935006541e-06,
      "loss": 0.6924,
      "step": 26930
    },
    {
      "epoch": 0.7759753356768282,
      "grad_norm": 0.7130160927772522,
      "learning_rate": 1.2594224494213203e-06,
      "loss": 0.6485,
      "step": 26931
    },
    {
      "epoch": 0.7760041491384775,
      "grad_norm": 0.7208611369132996,
      "learning_rate": 1.259112837918679e-06,
      "loss": 0.6825,
      "step": 26932
    },
    {
      "epoch": 0.7760329626001268,
      "grad_norm": 0.7319652438163757,
      "learning_rate": 1.2588032589954235e-06,
      "loss": 0.7043,
      "step": 26933
    },
    {
      "epoch": 0.7760617760617761,
      "grad_norm": 0.7195197939872742,
      "learning_rate": 1.258493712654254e-06,
      "loss": 0.6823,
      "step": 26934
    },
    {
      "epoch": 0.7760905895234254,
      "grad_norm": 0.7266064882278442,
      "learning_rate": 1.258184198897862e-06,
      "loss": 0.6971,
      "step": 26935
    },
    {
      "epoch": 0.7761194029850746,
      "grad_norm": 0.7541806697845459,
      "learning_rate": 1.2578747177289474e-06,
      "loss": 0.7275,
      "step": 26936
    },
    {
      "epoch": 0.7761482164467239,
      "grad_norm": 0.7161931991577148,
      "learning_rate": 1.2575652691502015e-06,
      "loss": 0.6728,
      "step": 26937
    },
    {
      "epoch": 0.7761770299083732,
      "grad_norm": 0.734693169593811,
      "learning_rate": 1.2572558531643209e-06,
      "loss": 0.6931,
      "step": 26938
    },
    {
      "epoch": 0.7762058433700225,
      "grad_norm": 0.7397980093955994,
      "learning_rate": 1.2569464697740019e-06,
      "loss": 0.6633,
      "step": 26939
    },
    {
      "epoch": 0.7762346568316718,
      "grad_norm": 0.7056294083595276,
      "learning_rate": 1.256637118981936e-06,
      "loss": 0.6756,
      "step": 26940
    },
    {
      "epoch": 0.776263470293321,
      "grad_norm": 0.7139580249786377,
      "learning_rate": 1.2563278007908202e-06,
      "loss": 0.6829,
      "step": 26941
    },
    {
      "epoch": 0.7762922837549703,
      "grad_norm": 0.7160745859146118,
      "learning_rate": 1.256018515203346e-06,
      "loss": 0.6796,
      "step": 26942
    },
    {
      "epoch": 0.7763210972166196,
      "grad_norm": 0.7545351386070251,
      "learning_rate": 1.2557092622222073e-06,
      "loss": 0.7193,
      "step": 26943
    },
    {
      "epoch": 0.7763499106782689,
      "grad_norm": 0.7248485684394836,
      "learning_rate": 1.2554000418500983e-06,
      "loss": 0.688,
      "step": 26944
    },
    {
      "epoch": 0.7763787241399182,
      "grad_norm": 0.7162254452705383,
      "learning_rate": 1.2550908540897134e-06,
      "loss": 0.6803,
      "step": 26945
    },
    {
      "epoch": 0.7764075376015674,
      "grad_norm": 0.7208738327026367,
      "learning_rate": 1.2547816989437416e-06,
      "loss": 0.686,
      "step": 26946
    },
    {
      "epoch": 0.7764363510632167,
      "grad_norm": 0.6913149356842041,
      "learning_rate": 1.2544725764148791e-06,
      "loss": 0.6699,
      "step": 26947
    },
    {
      "epoch": 0.776465164524866,
      "grad_norm": 0.6913354992866516,
      "learning_rate": 1.254163486505815e-06,
      "loss": 0.66,
      "step": 26948
    },
    {
      "epoch": 0.7764939779865153,
      "grad_norm": 0.7225092053413391,
      "learning_rate": 1.2538544292192434e-06,
      "loss": 0.6906,
      "step": 26949
    },
    {
      "epoch": 0.7765227914481646,
      "grad_norm": 0.7312251329421997,
      "learning_rate": 1.2535454045578539e-06,
      "loss": 0.7024,
      "step": 26950
    },
    {
      "epoch": 0.7765516049098139,
      "grad_norm": 0.7490320205688477,
      "learning_rate": 1.2532364125243402e-06,
      "loss": 0.7035,
      "step": 26951
    },
    {
      "epoch": 0.7765804183714632,
      "grad_norm": 0.733391523361206,
      "learning_rate": 1.252927453121391e-06,
      "loss": 0.7003,
      "step": 26952
    },
    {
      "epoch": 0.7766092318331125,
      "grad_norm": 0.7165580987930298,
      "learning_rate": 1.2526185263516988e-06,
      "loss": 0.6817,
      "step": 26953
    },
    {
      "epoch": 0.7766380452947617,
      "grad_norm": 0.7061946392059326,
      "learning_rate": 1.2523096322179501e-06,
      "loss": 0.6892,
      "step": 26954
    },
    {
      "epoch": 0.776666858756411,
      "grad_norm": 0.7166497707366943,
      "learning_rate": 1.2520007707228415e-06,
      "loss": 0.7023,
      "step": 26955
    },
    {
      "epoch": 0.7766956722180602,
      "grad_norm": 0.7070522308349609,
      "learning_rate": 1.2516919418690582e-06,
      "loss": 0.6878,
      "step": 26956
    },
    {
      "epoch": 0.7767244856797095,
      "grad_norm": 0.7176619172096252,
      "learning_rate": 1.251383145659293e-06,
      "loss": 0.7177,
      "step": 26957
    },
    {
      "epoch": 0.7767532991413588,
      "grad_norm": 0.7325156331062317,
      "learning_rate": 1.2510743820962312e-06,
      "loss": 0.6805,
      "step": 26958
    },
    {
      "epoch": 0.7767821126030081,
      "grad_norm": 0.7093314528465271,
      "learning_rate": 1.250765651182564e-06,
      "loss": 0.6856,
      "step": 26959
    },
    {
      "epoch": 0.7768109260646574,
      "grad_norm": 0.7379356026649475,
      "learning_rate": 1.2504569529209815e-06,
      "loss": 0.6983,
      "step": 26960
    },
    {
      "epoch": 0.7768397395263067,
      "grad_norm": 0.7322301864624023,
      "learning_rate": 1.2501482873141696e-06,
      "loss": 0.6944,
      "step": 26961
    },
    {
      "epoch": 0.776868552987956,
      "grad_norm": 0.7344965934753418,
      "learning_rate": 1.2498396543648196e-06,
      "loss": 0.7051,
      "step": 26962
    },
    {
      "epoch": 0.7768973664496053,
      "grad_norm": 0.7117108106613159,
      "learning_rate": 1.2495310540756151e-06,
      "loss": 0.6707,
      "step": 26963
    },
    {
      "epoch": 0.7769261799112546,
      "grad_norm": 0.7409559488296509,
      "learning_rate": 1.2492224864492475e-06,
      "loss": 0.7012,
      "step": 26964
    },
    {
      "epoch": 0.7769549933729039,
      "grad_norm": 0.7369372248649597,
      "learning_rate": 1.2489139514884019e-06,
      "loss": 0.6907,
      "step": 26965
    },
    {
      "epoch": 0.7769838068345531,
      "grad_norm": 0.7328141331672668,
      "learning_rate": 1.2486054491957656e-06,
      "loss": 0.6983,
      "step": 26966
    },
    {
      "epoch": 0.7770126202962024,
      "grad_norm": 0.7108232378959656,
      "learning_rate": 1.2482969795740257e-06,
      "loss": 0.6654,
      "step": 26967
    },
    {
      "epoch": 0.7770414337578516,
      "grad_norm": 0.718392550945282,
      "learning_rate": 1.2479885426258704e-06,
      "loss": 0.6905,
      "step": 26968
    },
    {
      "epoch": 0.7770702472195009,
      "grad_norm": 0.7125677466392517,
      "learning_rate": 1.2476801383539828e-06,
      "loss": 0.6912,
      "step": 26969
    },
    {
      "epoch": 0.7770990606811502,
      "grad_norm": 0.712627112865448,
      "learning_rate": 1.2473717667610519e-06,
      "loss": 0.692,
      "step": 26970
    },
    {
      "epoch": 0.7771278741427995,
      "grad_norm": 0.7057797908782959,
      "learning_rate": 1.24706342784976e-06,
      "loss": 0.6755,
      "step": 26971
    },
    {
      "epoch": 0.7771566876044488,
      "grad_norm": 0.7032908797264099,
      "learning_rate": 1.2467551216227952e-06,
      "loss": 0.6842,
      "step": 26972
    },
    {
      "epoch": 0.7771855010660981,
      "grad_norm": 0.734558641910553,
      "learning_rate": 1.2464468480828406e-06,
      "loss": 0.6922,
      "step": 26973
    },
    {
      "epoch": 0.7772143145277474,
      "grad_norm": 0.7082686424255371,
      "learning_rate": 1.2461386072325825e-06,
      "loss": 0.694,
      "step": 26974
    },
    {
      "epoch": 0.7772431279893967,
      "grad_norm": 0.7231772541999817,
      "learning_rate": 1.245830399074704e-06,
      "loss": 0.6904,
      "step": 26975
    },
    {
      "epoch": 0.777271941451046,
      "grad_norm": 0.7417194843292236,
      "learning_rate": 1.2455222236118896e-06,
      "loss": 0.6972,
      "step": 26976
    },
    {
      "epoch": 0.7773007549126952,
      "grad_norm": 0.71328204870224,
      "learning_rate": 1.245214080846825e-06,
      "loss": 0.6673,
      "step": 26977
    },
    {
      "epoch": 0.7773295683743445,
      "grad_norm": 0.704810380935669,
      "learning_rate": 1.2449059707821902e-06,
      "loss": 0.6443,
      "step": 26978
    },
    {
      "epoch": 0.7773583818359938,
      "grad_norm": 0.7146711945533752,
      "learning_rate": 1.2445978934206715e-06,
      "loss": 0.6815,
      "step": 26979
    },
    {
      "epoch": 0.7773871952976431,
      "grad_norm": 0.7283973097801208,
      "learning_rate": 1.2442898487649524e-06,
      "loss": 0.6911,
      "step": 26980
    },
    {
      "epoch": 0.7774160087592923,
      "grad_norm": 0.7147841453552246,
      "learning_rate": 1.2439818368177132e-06,
      "loss": 0.6755,
      "step": 26981
    },
    {
      "epoch": 0.7774448222209416,
      "grad_norm": 0.7234803438186646,
      "learning_rate": 1.243673857581637e-06,
      "loss": 0.683,
      "step": 26982
    },
    {
      "epoch": 0.7774736356825909,
      "grad_norm": 0.7369281649589539,
      "learning_rate": 1.2433659110594083e-06,
      "loss": 0.6996,
      "step": 26983
    },
    {
      "epoch": 0.7775024491442402,
      "grad_norm": 0.7396050691604614,
      "learning_rate": 1.243057997253706e-06,
      "loss": 0.6692,
      "step": 26984
    },
    {
      "epoch": 0.7775312626058895,
      "grad_norm": 0.7164115309715271,
      "learning_rate": 1.242750116167214e-06,
      "loss": 0.6921,
      "step": 26985
    },
    {
      "epoch": 0.7775600760675387,
      "grad_norm": 0.7437095642089844,
      "learning_rate": 1.2424422678026116e-06,
      "loss": 0.6644,
      "step": 26986
    },
    {
      "epoch": 0.777588889529188,
      "grad_norm": 0.7147486209869385,
      "learning_rate": 1.2421344521625816e-06,
      "loss": 0.6782,
      "step": 26987
    },
    {
      "epoch": 0.7776177029908373,
      "grad_norm": 0.6980538964271545,
      "learning_rate": 1.2418266692498031e-06,
      "loss": 0.6592,
      "step": 26988
    },
    {
      "epoch": 0.7776465164524866,
      "grad_norm": 0.6906331181526184,
      "learning_rate": 1.241518919066959e-06,
      "loss": 0.66,
      "step": 26989
    },
    {
      "epoch": 0.7776753299141359,
      "grad_norm": 0.699765145778656,
      "learning_rate": 1.2412112016167248e-06,
      "loss": 0.6828,
      "step": 26990
    },
    {
      "epoch": 0.7777041433757852,
      "grad_norm": 0.7368233799934387,
      "learning_rate": 1.240903516901787e-06,
      "loss": 0.7139,
      "step": 26991
    },
    {
      "epoch": 0.7777329568374345,
      "grad_norm": 0.7267956733703613,
      "learning_rate": 1.24059586492482e-06,
      "loss": 0.6921,
      "step": 26992
    },
    {
      "epoch": 0.7777617702990838,
      "grad_norm": 0.7224470376968384,
      "learning_rate": 1.2402882456885067e-06,
      "loss": 0.7085,
      "step": 26993
    },
    {
      "epoch": 0.7777905837607331,
      "grad_norm": 0.7218875288963318,
      "learning_rate": 1.2399806591955228e-06,
      "loss": 0.674,
      "step": 26994
    },
    {
      "epoch": 0.7778193972223822,
      "grad_norm": 0.7414214611053467,
      "learning_rate": 1.23967310544855e-06,
      "loss": 0.6893,
      "step": 26995
    },
    {
      "epoch": 0.7778482106840315,
      "grad_norm": 0.7446178793907166,
      "learning_rate": 1.2393655844502645e-06,
      "loss": 0.695,
      "step": 26996
    },
    {
      "epoch": 0.7778770241456808,
      "grad_norm": 0.7384447455406189,
      "learning_rate": 1.2390580962033455e-06,
      "loss": 0.6838,
      "step": 26997
    },
    {
      "epoch": 0.7779058376073301,
      "grad_norm": 0.7124254107475281,
      "learning_rate": 1.2387506407104722e-06,
      "loss": 0.684,
      "step": 26998
    },
    {
      "epoch": 0.7779346510689794,
      "grad_norm": 0.727617084980011,
      "learning_rate": 1.2384432179743194e-06,
      "loss": 0.6639,
      "step": 26999
    },
    {
      "epoch": 0.7779634645306287,
      "grad_norm": 0.7050459980964661,
      "learning_rate": 1.2381358279975675e-06,
      "loss": 0.6786,
      "step": 27000
    },
    {
      "epoch": 0.777992277992278,
      "grad_norm": 0.7396389245986938,
      "learning_rate": 1.2378284707828903e-06,
      "loss": 0.718,
      "step": 27001
    },
    {
      "epoch": 0.7780210914539273,
      "grad_norm": 0.7262334823608398,
      "learning_rate": 1.2375211463329668e-06,
      "loss": 0.6873,
      "step": 27002
    },
    {
      "epoch": 0.7780499049155766,
      "grad_norm": 0.7353927493095398,
      "learning_rate": 1.2372138546504724e-06,
      "loss": 0.6906,
      "step": 27003
    },
    {
      "epoch": 0.7780787183772259,
      "grad_norm": 0.7430436611175537,
      "learning_rate": 1.236906595738086e-06,
      "loss": 0.6791,
      "step": 27004
    },
    {
      "epoch": 0.7781075318388752,
      "grad_norm": 0.7142235636711121,
      "learning_rate": 1.2365993695984797e-06,
      "loss": 0.6854,
      "step": 27005
    },
    {
      "epoch": 0.7781363453005244,
      "grad_norm": 0.732729971408844,
      "learning_rate": 1.2362921762343326e-06,
      "loss": 0.6637,
      "step": 27006
    },
    {
      "epoch": 0.7781651587621737,
      "grad_norm": 0.7211254835128784,
      "learning_rate": 1.2359850156483165e-06,
      "loss": 0.6967,
      "step": 27007
    },
    {
      "epoch": 0.7781939722238229,
      "grad_norm": 0.6956951022148132,
      "learning_rate": 1.2356778878431097e-06,
      "loss": 0.6828,
      "step": 27008
    },
    {
      "epoch": 0.7782227856854722,
      "grad_norm": 0.7014413475990295,
      "learning_rate": 1.2353707928213844e-06,
      "loss": 0.672,
      "step": 27009
    },
    {
      "epoch": 0.7782515991471215,
      "grad_norm": 0.7130609154701233,
      "learning_rate": 1.2350637305858177e-06,
      "loss": 0.6574,
      "step": 27010
    },
    {
      "epoch": 0.7782804126087708,
      "grad_norm": 0.7170745134353638,
      "learning_rate": 1.2347567011390809e-06,
      "loss": 0.6624,
      "step": 27011
    },
    {
      "epoch": 0.7783092260704201,
      "grad_norm": 0.729089081287384,
      "learning_rate": 1.2344497044838505e-06,
      "loss": 0.6709,
      "step": 27012
    },
    {
      "epoch": 0.7783380395320694,
      "grad_norm": 0.7160082459449768,
      "learning_rate": 1.234142740622798e-06,
      "loss": 0.6982,
      "step": 27013
    },
    {
      "epoch": 0.7783668529937187,
      "grad_norm": 0.7163348197937012,
      "learning_rate": 1.2338358095585972e-06,
      "loss": 0.689,
      "step": 27014
    },
    {
      "epoch": 0.778395666455368,
      "grad_norm": 0.7167034149169922,
      "learning_rate": 1.233528911293922e-06,
      "loss": 0.6638,
      "step": 27015
    },
    {
      "epoch": 0.7784244799170172,
      "grad_norm": 0.7183775305747986,
      "learning_rate": 1.2332220458314464e-06,
      "loss": 0.6612,
      "step": 27016
    },
    {
      "epoch": 0.7784532933786665,
      "grad_norm": 0.7000497579574585,
      "learning_rate": 1.2329152131738399e-06,
      "loss": 0.6731,
      "step": 27017
    },
    {
      "epoch": 0.7784821068403158,
      "grad_norm": 0.6913316249847412,
      "learning_rate": 1.2326084133237775e-06,
      "loss": 0.6527,
      "step": 27018
    },
    {
      "epoch": 0.7785109203019651,
      "grad_norm": 0.7396538853645325,
      "learning_rate": 1.2323016462839287e-06,
      "loss": 0.6862,
      "step": 27019
    },
    {
      "epoch": 0.7785397337636144,
      "grad_norm": 0.7223522663116455,
      "learning_rate": 1.2319949120569657e-06,
      "loss": 0.6687,
      "step": 27020
    },
    {
      "epoch": 0.7785685472252637,
      "grad_norm": 0.7296370267868042,
      "learning_rate": 1.2316882106455624e-06,
      "loss": 0.6747,
      "step": 27021
    },
    {
      "epoch": 0.7785973606869129,
      "grad_norm": 0.7248327732086182,
      "learning_rate": 1.2313815420523862e-06,
      "loss": 0.6907,
      "step": 27022
    },
    {
      "epoch": 0.7786261741485622,
      "grad_norm": 0.7252039313316345,
      "learning_rate": 1.2310749062801108e-06,
      "loss": 0.6962,
      "step": 27023
    },
    {
      "epoch": 0.7786549876102115,
      "grad_norm": 0.7196646332740784,
      "learning_rate": 1.2307683033314044e-06,
      "loss": 0.6823,
      "step": 27024
    },
    {
      "epoch": 0.7786838010718607,
      "grad_norm": 0.7600692510604858,
      "learning_rate": 1.2304617332089392e-06,
      "loss": 0.686,
      "step": 27025
    },
    {
      "epoch": 0.77871261453351,
      "grad_norm": 0.7215406894683838,
      "learning_rate": 1.2301551959153812e-06,
      "loss": 0.6906,
      "step": 27026
    },
    {
      "epoch": 0.7787414279951593,
      "grad_norm": 0.7053678631782532,
      "learning_rate": 1.2298486914534063e-06,
      "loss": 0.6887,
      "step": 27027
    },
    {
      "epoch": 0.7787702414568086,
      "grad_norm": 0.7408519387245178,
      "learning_rate": 1.229542219825679e-06,
      "loss": 0.694,
      "step": 27028
    },
    {
      "epoch": 0.7787990549184579,
      "grad_norm": 0.7353858351707458,
      "learning_rate": 1.229235781034871e-06,
      "loss": 0.6833,
      "step": 27029
    },
    {
      "epoch": 0.7788278683801072,
      "grad_norm": 0.7227752804756165,
      "learning_rate": 1.2289293750836489e-06,
      "loss": 0.696,
      "step": 27030
    },
    {
      "epoch": 0.7788566818417565,
      "grad_norm": 0.7209666967391968,
      "learning_rate": 1.2286230019746837e-06,
      "loss": 0.7006,
      "step": 27031
    },
    {
      "epoch": 0.7788854953034058,
      "grad_norm": 0.6969279646873474,
      "learning_rate": 1.22831666171064e-06,
      "loss": 0.6968,
      "step": 27032
    },
    {
      "epoch": 0.7789143087650551,
      "grad_norm": 0.7425254583358765,
      "learning_rate": 1.2280103542941902e-06,
      "loss": 0.6837,
      "step": 27033
    },
    {
      "epoch": 0.7789431222267044,
      "grad_norm": 0.7143228650093079,
      "learning_rate": 1.2277040797279977e-06,
      "loss": 0.6871,
      "step": 27034
    },
    {
      "epoch": 0.7789719356883535,
      "grad_norm": 0.7166190147399902,
      "learning_rate": 1.2273978380147317e-06,
      "loss": 0.6823,
      "step": 27035
    },
    {
      "epoch": 0.7790007491500028,
      "grad_norm": 0.7367145419120789,
      "learning_rate": 1.2270916291570612e-06,
      "loss": 0.6759,
      "step": 27036
    },
    {
      "epoch": 0.7790295626116521,
      "grad_norm": 0.7341307997703552,
      "learning_rate": 1.226785453157649e-06,
      "loss": 0.6876,
      "step": 27037
    },
    {
      "epoch": 0.7790583760733014,
      "grad_norm": 0.7248852849006653,
      "learning_rate": 1.226479310019164e-06,
      "loss": 0.68,
      "step": 27038
    },
    {
      "epoch": 0.7790871895349507,
      "grad_norm": 0.7229720950126648,
      "learning_rate": 1.2261731997442738e-06,
      "loss": 0.696,
      "step": 27039
    },
    {
      "epoch": 0.7791160029966,
      "grad_norm": 0.7176437377929688,
      "learning_rate": 1.2258671223356412e-06,
      "loss": 0.6917,
      "step": 27040
    },
    {
      "epoch": 0.7791448164582493,
      "grad_norm": 0.71483314037323,
      "learning_rate": 1.2255610777959326e-06,
      "loss": 0.6667,
      "step": 27041
    },
    {
      "epoch": 0.7791736299198986,
      "grad_norm": 0.7160323858261108,
      "learning_rate": 1.2252550661278156e-06,
      "loss": 0.6887,
      "step": 27042
    },
    {
      "epoch": 0.7792024433815479,
      "grad_norm": 0.7361361384391785,
      "learning_rate": 1.224949087333952e-06,
      "loss": 0.7112,
      "step": 27043
    },
    {
      "epoch": 0.7792312568431972,
      "grad_norm": 0.7224162220954895,
      "learning_rate": 1.2246431414170102e-06,
      "loss": 0.6886,
      "step": 27044
    },
    {
      "epoch": 0.7792600703048465,
      "grad_norm": 0.7044194340705872,
      "learning_rate": 1.2243372283796506e-06,
      "loss": 0.6569,
      "step": 27045
    },
    {
      "epoch": 0.7792888837664957,
      "grad_norm": 0.7048516273498535,
      "learning_rate": 1.2240313482245413e-06,
      "loss": 0.6684,
      "step": 27046
    },
    {
      "epoch": 0.779317697228145,
      "grad_norm": 0.738507866859436,
      "learning_rate": 1.2237255009543431e-06,
      "loss": 0.7067,
      "step": 27047
    },
    {
      "epoch": 0.7793465106897943,
      "grad_norm": 0.7388807535171509,
      "learning_rate": 1.2234196865717219e-06,
      "loss": 0.6871,
      "step": 27048
    },
    {
      "epoch": 0.7793753241514435,
      "grad_norm": 0.7140911221504211,
      "learning_rate": 1.223113905079339e-06,
      "loss": 0.6856,
      "step": 27049
    },
    {
      "epoch": 0.7794041376130928,
      "grad_norm": 0.745829701423645,
      "learning_rate": 1.2228081564798583e-06,
      "loss": 0.7093,
      "step": 27050
    },
    {
      "epoch": 0.7794329510747421,
      "grad_norm": 0.7043879628181458,
      "learning_rate": 1.2225024407759422e-06,
      "loss": 0.6693,
      "step": 27051
    },
    {
      "epoch": 0.7794617645363914,
      "grad_norm": 0.7452259063720703,
      "learning_rate": 1.222196757970256e-06,
      "loss": 0.7114,
      "step": 27052
    },
    {
      "epoch": 0.7794905779980407,
      "grad_norm": 0.7282812595367432,
      "learning_rate": 1.2218911080654577e-06,
      "loss": 0.6892,
      "step": 27053
    },
    {
      "epoch": 0.77951939145969,
      "grad_norm": 0.7503501772880554,
      "learning_rate": 1.2215854910642128e-06,
      "loss": 0.6919,
      "step": 27054
    },
    {
      "epoch": 0.7795482049213392,
      "grad_norm": 0.6908959150314331,
      "learning_rate": 1.2212799069691794e-06,
      "loss": 0.68,
      "step": 27055
    },
    {
      "epoch": 0.7795770183829885,
      "grad_norm": 0.7370312809944153,
      "learning_rate": 1.2209743557830227e-06,
      "loss": 0.6867,
      "step": 27056
    },
    {
      "epoch": 0.7796058318446378,
      "grad_norm": 0.7029189467430115,
      "learning_rate": 1.2206688375083997e-06,
      "loss": 0.6555,
      "step": 27057
    },
    {
      "epoch": 0.7796346453062871,
      "grad_norm": 0.7247611880302429,
      "learning_rate": 1.2203633521479735e-06,
      "loss": 0.6817,
      "step": 27058
    },
    {
      "epoch": 0.7796634587679364,
      "grad_norm": 0.7131620645523071,
      "learning_rate": 1.220057899704406e-06,
      "loss": 0.6958,
      "step": 27059
    },
    {
      "epoch": 0.7796922722295857,
      "grad_norm": 0.7478622198104858,
      "learning_rate": 1.2197524801803534e-06,
      "loss": 0.6906,
      "step": 27060
    },
    {
      "epoch": 0.779721085691235,
      "grad_norm": 0.7230139970779419,
      "learning_rate": 1.2194470935784796e-06,
      "loss": 0.6717,
      "step": 27061
    },
    {
      "epoch": 0.7797498991528842,
      "grad_norm": 0.7127116918563843,
      "learning_rate": 1.21914173990144e-06,
      "loss": 0.6819,
      "step": 27062
    },
    {
      "epoch": 0.7797787126145335,
      "grad_norm": 0.7334070801734924,
      "learning_rate": 1.2188364191518986e-06,
      "loss": 0.6876,
      "step": 27063
    },
    {
      "epoch": 0.7798075260761828,
      "grad_norm": 0.7205588221549988,
      "learning_rate": 1.2185311313325115e-06,
      "loss": 0.6661,
      "step": 27064
    },
    {
      "epoch": 0.779836339537832,
      "grad_norm": 0.7162211537361145,
      "learning_rate": 1.2182258764459392e-06,
      "loss": 0.6861,
      "step": 27065
    },
    {
      "epoch": 0.7798651529994813,
      "grad_norm": 0.7156692743301392,
      "learning_rate": 1.2179206544948375e-06,
      "loss": 0.6925,
      "step": 27066
    },
    {
      "epoch": 0.7798939664611306,
      "grad_norm": 0.7254343032836914,
      "learning_rate": 1.2176154654818678e-06,
      "loss": 0.6774,
      "step": 27067
    },
    {
      "epoch": 0.7799227799227799,
      "grad_norm": 0.6986673474311829,
      "learning_rate": 1.217310309409685e-06,
      "loss": 0.6485,
      "step": 27068
    },
    {
      "epoch": 0.7799515933844292,
      "grad_norm": 0.7186209559440613,
      "learning_rate": 1.2170051862809495e-06,
      "loss": 0.6842,
      "step": 27069
    },
    {
      "epoch": 0.7799804068460785,
      "grad_norm": 0.7372789978981018,
      "learning_rate": 1.216700096098316e-06,
      "loss": 0.6833,
      "step": 27070
    },
    {
      "epoch": 0.7800092203077278,
      "grad_norm": 0.7201875448226929,
      "learning_rate": 1.216395038864444e-06,
      "loss": 0.6762,
      "step": 27071
    },
    {
      "epoch": 0.7800380337693771,
      "grad_norm": 0.7498669624328613,
      "learning_rate": 1.2160900145819875e-06,
      "loss": 0.722,
      "step": 27072
    },
    {
      "epoch": 0.7800668472310264,
      "grad_norm": 0.730898380279541,
      "learning_rate": 1.2157850232536045e-06,
      "loss": 0.6831,
      "step": 27073
    },
    {
      "epoch": 0.7800956606926757,
      "grad_norm": 0.7102658152580261,
      "learning_rate": 1.215480064881951e-06,
      "loss": 0.6799,
      "step": 27074
    },
    {
      "epoch": 0.780124474154325,
      "grad_norm": 0.743127167224884,
      "learning_rate": 1.2151751394696848e-06,
      "loss": 0.6711,
      "step": 27075
    },
    {
      "epoch": 0.7801532876159741,
      "grad_norm": 0.724873423576355,
      "learning_rate": 1.2148702470194578e-06,
      "loss": 0.6879,
      "step": 27076
    },
    {
      "epoch": 0.7801821010776234,
      "grad_norm": 0.6980623602867126,
      "learning_rate": 1.2145653875339286e-06,
      "loss": 0.6914,
      "step": 27077
    },
    {
      "epoch": 0.7802109145392727,
      "grad_norm": 0.7145766019821167,
      "learning_rate": 1.2142605610157499e-06,
      "loss": 0.6827,
      "step": 27078
    },
    {
      "epoch": 0.780239728000922,
      "grad_norm": 0.7133605480194092,
      "learning_rate": 1.2139557674675773e-06,
      "loss": 0.6751,
      "step": 27079
    },
    {
      "epoch": 0.7802685414625713,
      "grad_norm": 0.7290591597557068,
      "learning_rate": 1.2136510068920666e-06,
      "loss": 0.6847,
      "step": 27080
    },
    {
      "epoch": 0.7802973549242206,
      "grad_norm": 0.7176723480224609,
      "learning_rate": 1.2133462792918689e-06,
      "loss": 0.6826,
      "step": 27081
    },
    {
      "epoch": 0.7803261683858699,
      "grad_norm": 0.7462021708488464,
      "learning_rate": 1.2130415846696413e-06,
      "loss": 0.6782,
      "step": 27082
    },
    {
      "epoch": 0.7803549818475192,
      "grad_norm": 0.7142312526702881,
      "learning_rate": 1.212736923028035e-06,
      "loss": 0.6819,
      "step": 27083
    },
    {
      "epoch": 0.7803837953091685,
      "grad_norm": 0.7486127614974976,
      "learning_rate": 1.2124322943697054e-06,
      "loss": 0.6791,
      "step": 27084
    },
    {
      "epoch": 0.7804126087708178,
      "grad_norm": 0.7454231977462769,
      "learning_rate": 1.2121276986973013e-06,
      "loss": 0.6789,
      "step": 27085
    },
    {
      "epoch": 0.780441422232467,
      "grad_norm": 0.7264445424079895,
      "learning_rate": 1.2118231360134814e-06,
      "loss": 0.6802,
      "step": 27086
    },
    {
      "epoch": 0.7804702356941163,
      "grad_norm": 0.74729984998703,
      "learning_rate": 1.2115186063208939e-06,
      "loss": 0.6915,
      "step": 27087
    },
    {
      "epoch": 0.7804990491557656,
      "grad_norm": 0.7236136794090271,
      "learning_rate": 1.2112141096221935e-06,
      "loss": 0.6786,
      "step": 27088
    },
    {
      "epoch": 0.7805278626174148,
      "grad_norm": 0.7286954522132874,
      "learning_rate": 1.2109096459200299e-06,
      "loss": 0.681,
      "step": 27089
    },
    {
      "epoch": 0.7805566760790641,
      "grad_norm": 0.7278928756713867,
      "learning_rate": 1.210605215217056e-06,
      "loss": 0.6854,
      "step": 27090
    },
    {
      "epoch": 0.7805854895407134,
      "grad_norm": 0.7435287833213806,
      "learning_rate": 1.210300817515922e-06,
      "loss": 0.7082,
      "step": 27091
    },
    {
      "epoch": 0.7806143030023627,
      "grad_norm": 0.740212082862854,
      "learning_rate": 1.2099964528192809e-06,
      "loss": 0.6802,
      "step": 27092
    },
    {
      "epoch": 0.780643116464012,
      "grad_norm": 0.736040472984314,
      "learning_rate": 1.2096921211297806e-06,
      "loss": 0.6817,
      "step": 27093
    },
    {
      "epoch": 0.7806719299256613,
      "grad_norm": 0.7240583896636963,
      "learning_rate": 1.2093878224500744e-06,
      "loss": 0.6768,
      "step": 27094
    },
    {
      "epoch": 0.7807007433873105,
      "grad_norm": 0.7156563997268677,
      "learning_rate": 1.2090835567828097e-06,
      "loss": 0.6814,
      "step": 27095
    },
    {
      "epoch": 0.7807295568489598,
      "grad_norm": 0.7138580679893494,
      "learning_rate": 1.2087793241306378e-06,
      "loss": 0.6807,
      "step": 27096
    },
    {
      "epoch": 0.7807583703106091,
      "grad_norm": 0.7134495377540588,
      "learning_rate": 1.208475124496208e-06,
      "loss": 0.6492,
      "step": 27097
    },
    {
      "epoch": 0.7807871837722584,
      "grad_norm": 0.7373895049095154,
      "learning_rate": 1.208170957882171e-06,
      "loss": 0.6744,
      "step": 27098
    },
    {
      "epoch": 0.7808159972339077,
      "grad_norm": 0.7209753394126892,
      "learning_rate": 1.2078668242911733e-06,
      "loss": 0.6812,
      "step": 27099
    },
    {
      "epoch": 0.780844810695557,
      "grad_norm": 0.7239120602607727,
      "learning_rate": 1.2075627237258648e-06,
      "loss": 0.6925,
      "step": 27100
    },
    {
      "epoch": 0.7808736241572063,
      "grad_norm": 0.7472798824310303,
      "learning_rate": 1.207258656188895e-06,
      "loss": 0.6716,
      "step": 27101
    },
    {
      "epoch": 0.7809024376188556,
      "grad_norm": 0.7462748885154724,
      "learning_rate": 1.2069546216829098e-06,
      "loss": 0.7007,
      "step": 27102
    },
    {
      "epoch": 0.7809312510805048,
      "grad_norm": 0.7043549418449402,
      "learning_rate": 1.20665062021056e-06,
      "loss": 0.6678,
      "step": 27103
    },
    {
      "epoch": 0.780960064542154,
      "grad_norm": 0.732655942440033,
      "learning_rate": 1.2063466517744894e-06,
      "loss": 0.7063,
      "step": 27104
    },
    {
      "epoch": 0.7809888780038033,
      "grad_norm": 0.7205150127410889,
      "learning_rate": 1.2060427163773486e-06,
      "loss": 0.6999,
      "step": 27105
    },
    {
      "epoch": 0.7810176914654526,
      "grad_norm": 0.7269423604011536,
      "learning_rate": 1.2057388140217818e-06,
      "loss": 0.6842,
      "step": 27106
    },
    {
      "epoch": 0.7810465049271019,
      "grad_norm": 0.7294389009475708,
      "learning_rate": 1.2054349447104386e-06,
      "loss": 0.6996,
      "step": 27107
    },
    {
      "epoch": 0.7810753183887512,
      "grad_norm": 0.7112240791320801,
      "learning_rate": 1.2051311084459627e-06,
      "loss": 0.6849,
      "step": 27108
    },
    {
      "epoch": 0.7811041318504005,
      "grad_norm": 0.7277625799179077,
      "learning_rate": 1.204827305231001e-06,
      "loss": 0.6986,
      "step": 27109
    },
    {
      "epoch": 0.7811329453120498,
      "grad_norm": 0.7088446617126465,
      "learning_rate": 1.2045235350681995e-06,
      "loss": 0.6813,
      "step": 27110
    },
    {
      "epoch": 0.7811617587736991,
      "grad_norm": 0.7230050563812256,
      "learning_rate": 1.2042197979602054e-06,
      "loss": 0.6897,
      "step": 27111
    },
    {
      "epoch": 0.7811905722353484,
      "grad_norm": 0.7007997632026672,
      "learning_rate": 1.2039160939096618e-06,
      "loss": 0.6657,
      "step": 27112
    },
    {
      "epoch": 0.7812193856969977,
      "grad_norm": 0.7353895902633667,
      "learning_rate": 1.2036124229192147e-06,
      "loss": 0.699,
      "step": 27113
    },
    {
      "epoch": 0.781248199158647,
      "grad_norm": 0.7181501984596252,
      "learning_rate": 1.2033087849915076e-06,
      "loss": 0.7026,
      "step": 27114
    },
    {
      "epoch": 0.7812770126202963,
      "grad_norm": 0.7258956432342529,
      "learning_rate": 1.203005180129187e-06,
      "loss": 0.6797,
      "step": 27115
    },
    {
      "epoch": 0.7813058260819454,
      "grad_norm": 0.7600595355033875,
      "learning_rate": 1.2027016083348942e-06,
      "loss": 0.6634,
      "step": 27116
    },
    {
      "epoch": 0.7813346395435947,
      "grad_norm": 0.721820592880249,
      "learning_rate": 1.2023980696112747e-06,
      "loss": 0.7073,
      "step": 27117
    },
    {
      "epoch": 0.781363453005244,
      "grad_norm": 0.7737361192703247,
      "learning_rate": 1.2020945639609732e-06,
      "loss": 0.6832,
      "step": 27118
    },
    {
      "epoch": 0.7813922664668933,
      "grad_norm": 0.7041388154029846,
      "learning_rate": 1.20179109138663e-06,
      "loss": 0.6744,
      "step": 27119
    },
    {
      "epoch": 0.7814210799285426,
      "grad_norm": 0.7392207980155945,
      "learning_rate": 1.2014876518908909e-06,
      "loss": 0.7069,
      "step": 27120
    },
    {
      "epoch": 0.7814498933901919,
      "grad_norm": 0.703752338886261,
      "learning_rate": 1.2011842454763944e-06,
      "loss": 0.6696,
      "step": 27121
    },
    {
      "epoch": 0.7814787068518412,
      "grad_norm": 0.741595447063446,
      "learning_rate": 1.2008808721457882e-06,
      "loss": 0.6835,
      "step": 27122
    },
    {
      "epoch": 0.7815075203134905,
      "grad_norm": 0.7292705774307251,
      "learning_rate": 1.200577531901711e-06,
      "loss": 0.6615,
      "step": 27123
    },
    {
      "epoch": 0.7815363337751398,
      "grad_norm": 0.7264224886894226,
      "learning_rate": 1.2002742247468068e-06,
      "loss": 0.6878,
      "step": 27124
    },
    {
      "epoch": 0.781565147236789,
      "grad_norm": 0.7379683256149292,
      "learning_rate": 1.1999709506837142e-06,
      "loss": 0.6981,
      "step": 27125
    },
    {
      "epoch": 0.7815939606984383,
      "grad_norm": 0.725624680519104,
      "learning_rate": 1.1996677097150777e-06,
      "loss": 0.6989,
      "step": 27126
    },
    {
      "epoch": 0.7816227741600876,
      "grad_norm": 0.7331750988960266,
      "learning_rate": 1.1993645018435347e-06,
      "loss": 0.6799,
      "step": 27127
    },
    {
      "epoch": 0.7816515876217369,
      "grad_norm": 0.7233912944793701,
      "learning_rate": 1.1990613270717294e-06,
      "loss": 0.6969,
      "step": 27128
    },
    {
      "epoch": 0.7816804010833862,
      "grad_norm": 0.7153019905090332,
      "learning_rate": 1.1987581854022984e-06,
      "loss": 0.6819,
      "step": 27129
    },
    {
      "epoch": 0.7817092145450354,
      "grad_norm": 0.7220032215118408,
      "learning_rate": 1.1984550768378856e-06,
      "loss": 0.7047,
      "step": 27130
    },
    {
      "epoch": 0.7817380280066847,
      "grad_norm": 0.724377453327179,
      "learning_rate": 1.1981520013811276e-06,
      "loss": 0.7064,
      "step": 27131
    },
    {
      "epoch": 0.781766841468334,
      "grad_norm": 0.7158029675483704,
      "learning_rate": 1.1978489590346653e-06,
      "loss": 0.6919,
      "step": 27132
    },
    {
      "epoch": 0.7817956549299833,
      "grad_norm": 0.7089945673942566,
      "learning_rate": 1.1975459498011372e-06,
      "loss": 0.6833,
      "step": 27133
    },
    {
      "epoch": 0.7818244683916326,
      "grad_norm": 0.7126234173774719,
      "learning_rate": 1.1972429736831847e-06,
      "loss": 0.6831,
      "step": 27134
    },
    {
      "epoch": 0.7818532818532818,
      "grad_norm": 0.7206403017044067,
      "learning_rate": 1.1969400306834434e-06,
      "loss": 0.7051,
      "step": 27135
    },
    {
      "epoch": 0.7818820953149311,
      "grad_norm": 0.7077433466911316,
      "learning_rate": 1.1966371208045535e-06,
      "loss": 0.6839,
      "step": 27136
    },
    {
      "epoch": 0.7819109087765804,
      "grad_norm": 0.7319901585578918,
      "learning_rate": 1.196334244049151e-06,
      "loss": 0.6971,
      "step": 27137
    },
    {
      "epoch": 0.7819397222382297,
      "grad_norm": 0.7277098298072815,
      "learning_rate": 1.1960314004198753e-06,
      "loss": 0.6613,
      "step": 27138
    },
    {
      "epoch": 0.781968535699879,
      "grad_norm": 0.8233756422996521,
      "learning_rate": 1.195728589919365e-06,
      "loss": 0.687,
      "step": 27139
    },
    {
      "epoch": 0.7819973491615283,
      "grad_norm": 0.723636269569397,
      "learning_rate": 1.1954258125502543e-06,
      "loss": 0.6801,
      "step": 27140
    },
    {
      "epoch": 0.7820261626231776,
      "grad_norm": 0.7148153781890869,
      "learning_rate": 1.1951230683151827e-06,
      "loss": 0.6734,
      "step": 27141
    },
    {
      "epoch": 0.7820549760848269,
      "grad_norm": 0.7312471270561218,
      "learning_rate": 1.1948203572167843e-06,
      "loss": 0.6865,
      "step": 27142
    },
    {
      "epoch": 0.7820837895464761,
      "grad_norm": 0.8077908158302307,
      "learning_rate": 1.1945176792576984e-06,
      "loss": 0.6899,
      "step": 27143
    },
    {
      "epoch": 0.7821126030081254,
      "grad_norm": 0.7263386845588684,
      "learning_rate": 1.1942150344405584e-06,
      "loss": 0.6707,
      "step": 27144
    },
    {
      "epoch": 0.7821414164697746,
      "grad_norm": 0.7578572630882263,
      "learning_rate": 1.1939124227680004e-06,
      "loss": 0.7048,
      "step": 27145
    },
    {
      "epoch": 0.7821702299314239,
      "grad_norm": 0.7243785262107849,
      "learning_rate": 1.1936098442426608e-06,
      "loss": 0.6719,
      "step": 27146
    },
    {
      "epoch": 0.7821990433930732,
      "grad_norm": 0.7349374294281006,
      "learning_rate": 1.1933072988671757e-06,
      "loss": 0.68,
      "step": 27147
    },
    {
      "epoch": 0.7822278568547225,
      "grad_norm": 0.7234340310096741,
      "learning_rate": 1.1930047866441775e-06,
      "loss": 0.702,
      "step": 27148
    },
    {
      "epoch": 0.7822566703163718,
      "grad_norm": 0.7138743996620178,
      "learning_rate": 1.1927023075763033e-06,
      "loss": 0.6906,
      "step": 27149
    },
    {
      "epoch": 0.7822854837780211,
      "grad_norm": 0.7155975103378296,
      "learning_rate": 1.1923998616661847e-06,
      "loss": 0.6955,
      "step": 27150
    },
    {
      "epoch": 0.7823142972396704,
      "grad_norm": 0.734164834022522,
      "learning_rate": 1.1920974489164588e-06,
      "loss": 0.6919,
      "step": 27151
    },
    {
      "epoch": 0.7823431107013197,
      "grad_norm": 0.7249000072479248,
      "learning_rate": 1.1917950693297558e-06,
      "loss": 0.6868,
      "step": 27152
    },
    {
      "epoch": 0.782371924162969,
      "grad_norm": 0.7375547885894775,
      "learning_rate": 1.1914927229087126e-06,
      "loss": 0.6961,
      "step": 27153
    },
    {
      "epoch": 0.7824007376246183,
      "grad_norm": 0.7332158088684082,
      "learning_rate": 1.191190409655959e-06,
      "loss": 0.6734,
      "step": 27154
    },
    {
      "epoch": 0.7824295510862675,
      "grad_norm": 0.7283644080162048,
      "learning_rate": 1.1908881295741299e-06,
      "loss": 0.7062,
      "step": 27155
    },
    {
      "epoch": 0.7824583645479168,
      "grad_norm": 0.7302291393280029,
      "learning_rate": 1.1905858826658584e-06,
      "loss": 0.6777,
      "step": 27156
    },
    {
      "epoch": 0.782487178009566,
      "grad_norm": 0.7230461239814758,
      "learning_rate": 1.190283668933775e-06,
      "loss": 0.6747,
      "step": 27157
    },
    {
      "epoch": 0.7825159914712153,
      "grad_norm": 0.749691367149353,
      "learning_rate": 1.1899814883805122e-06,
      "loss": 0.7045,
      "step": 27158
    },
    {
      "epoch": 0.7825448049328646,
      "grad_norm": 0.7544189691543579,
      "learning_rate": 1.1896793410087021e-06,
      "loss": 0.713,
      "step": 27159
    },
    {
      "epoch": 0.7825736183945139,
      "grad_norm": 0.7268218994140625,
      "learning_rate": 1.1893772268209775e-06,
      "loss": 0.7137,
      "step": 27160
    },
    {
      "epoch": 0.7826024318561632,
      "grad_norm": 0.7263338565826416,
      "learning_rate": 1.1890751458199661e-06,
      "loss": 0.7008,
      "step": 27161
    },
    {
      "epoch": 0.7826312453178125,
      "grad_norm": 0.7279549241065979,
      "learning_rate": 1.1887730980083024e-06,
      "loss": 0.6978,
      "step": 27162
    },
    {
      "epoch": 0.7826600587794618,
      "grad_norm": 0.7328199744224548,
      "learning_rate": 1.1884710833886138e-06,
      "loss": 0.6828,
      "step": 27163
    },
    {
      "epoch": 0.782688872241111,
      "grad_norm": 0.6904685497283936,
      "learning_rate": 1.1881691019635334e-06,
      "loss": 0.6669,
      "step": 27164
    },
    {
      "epoch": 0.7827176857027603,
      "grad_norm": 0.7067458033561707,
      "learning_rate": 1.1878671537356884e-06,
      "loss": 0.6944,
      "step": 27165
    },
    {
      "epoch": 0.7827464991644096,
      "grad_norm": 0.7568853497505188,
      "learning_rate": 1.1875652387077108e-06,
      "loss": 0.6811,
      "step": 27166
    },
    {
      "epoch": 0.7827753126260589,
      "grad_norm": 0.7420352697372437,
      "learning_rate": 1.187263356882228e-06,
      "loss": 0.6888,
      "step": 27167
    },
    {
      "epoch": 0.7828041260877082,
      "grad_norm": 0.7316928505897522,
      "learning_rate": 1.1869615082618708e-06,
      "loss": 0.68,
      "step": 27168
    },
    {
      "epoch": 0.7828329395493575,
      "grad_norm": 0.732038676738739,
      "learning_rate": 1.1866596928492646e-06,
      "loss": 0.683,
      "step": 27169
    },
    {
      "epoch": 0.7828617530110067,
      "grad_norm": 0.7120233774185181,
      "learning_rate": 1.1863579106470436e-06,
      "loss": 0.6786,
      "step": 27170
    },
    {
      "epoch": 0.782890566472656,
      "grad_norm": 0.7187555432319641,
      "learning_rate": 1.1860561616578315e-06,
      "loss": 0.6682,
      "step": 27171
    },
    {
      "epoch": 0.7829193799343053,
      "grad_norm": 0.7200859189033508,
      "learning_rate": 1.185754445884259e-06,
      "loss": 0.6645,
      "step": 27172
    },
    {
      "epoch": 0.7829481933959546,
      "grad_norm": 0.7616735100746155,
      "learning_rate": 1.185452763328951e-06,
      "loss": 0.6825,
      "step": 27173
    },
    {
      "epoch": 0.7829770068576039,
      "grad_norm": 0.7185251712799072,
      "learning_rate": 1.1851511139945377e-06,
      "loss": 0.6935,
      "step": 27174
    },
    {
      "epoch": 0.7830058203192531,
      "grad_norm": 0.7358905673027039,
      "learning_rate": 1.1848494978836434e-06,
      "loss": 0.7024,
      "step": 27175
    },
    {
      "epoch": 0.7830346337809024,
      "grad_norm": 0.7279144525527954,
      "learning_rate": 1.1845479149988964e-06,
      "loss": 0.675,
      "step": 27176
    },
    {
      "epoch": 0.7830634472425517,
      "grad_norm": 0.708818793296814,
      "learning_rate": 1.1842463653429242e-06,
      "loss": 0.6759,
      "step": 27177
    },
    {
      "epoch": 0.783092260704201,
      "grad_norm": 0.7289834022521973,
      "learning_rate": 1.1839448489183502e-06,
      "loss": 0.6745,
      "step": 27178
    },
    {
      "epoch": 0.7831210741658503,
      "grad_norm": 0.7203513979911804,
      "learning_rate": 1.1836433657278034e-06,
      "loss": 0.6914,
      "step": 27179
    },
    {
      "epoch": 0.7831498876274996,
      "grad_norm": 0.7160709500312805,
      "learning_rate": 1.1833419157739068e-06,
      "loss": 0.689,
      "step": 27180
    },
    {
      "epoch": 0.7831787010891489,
      "grad_norm": 0.7294477820396423,
      "learning_rate": 1.1830404990592864e-06,
      "loss": 0.6987,
      "step": 27181
    },
    {
      "epoch": 0.7832075145507982,
      "grad_norm": 0.7219416499137878,
      "learning_rate": 1.182739115586568e-06,
      "loss": 0.6683,
      "step": 27182
    },
    {
      "epoch": 0.7832363280124475,
      "grad_norm": 0.7311669588088989,
      "learning_rate": 1.1824377653583773e-06,
      "loss": 0.6695,
      "step": 27183
    },
    {
      "epoch": 0.7832651414740966,
      "grad_norm": 0.7390934228897095,
      "learning_rate": 1.1821364483773362e-06,
      "loss": 0.6952,
      "step": 27184
    },
    {
      "epoch": 0.7832939549357459,
      "grad_norm": 0.7232504487037659,
      "learning_rate": 1.1818351646460718e-06,
      "loss": 0.7094,
      "step": 27185
    },
    {
      "epoch": 0.7833227683973952,
      "grad_norm": 0.740130603313446,
      "learning_rate": 1.181533914167205e-06,
      "loss": 0.6974,
      "step": 27186
    },
    {
      "epoch": 0.7833515818590445,
      "grad_norm": 0.7137002348899841,
      "learning_rate": 1.1812326969433618e-06,
      "loss": 0.6886,
      "step": 27187
    },
    {
      "epoch": 0.7833803953206938,
      "grad_norm": 0.7139683961868286,
      "learning_rate": 1.1809315129771632e-06,
      "loss": 0.6772,
      "step": 27188
    },
    {
      "epoch": 0.7834092087823431,
      "grad_norm": 0.7166796326637268,
      "learning_rate": 1.1806303622712346e-06,
      "loss": 0.6933,
      "step": 27189
    },
    {
      "epoch": 0.7834380222439924,
      "grad_norm": 0.7117869853973389,
      "learning_rate": 1.1803292448281961e-06,
      "loss": 0.6748,
      "step": 27190
    },
    {
      "epoch": 0.7834668357056417,
      "grad_norm": 0.7146778702735901,
      "learning_rate": 1.1800281606506731e-06,
      "loss": 0.6774,
      "step": 27191
    },
    {
      "epoch": 0.783495649167291,
      "grad_norm": 0.708931028842926,
      "learning_rate": 1.179727109741285e-06,
      "loss": 0.6924,
      "step": 27192
    },
    {
      "epoch": 0.7835244626289403,
      "grad_norm": 0.6890703439712524,
      "learning_rate": 1.1794260921026552e-06,
      "loss": 0.6911,
      "step": 27193
    },
    {
      "epoch": 0.7835532760905896,
      "grad_norm": 0.7265089154243469,
      "learning_rate": 1.1791251077374044e-06,
      "loss": 0.6763,
      "step": 27194
    },
    {
      "epoch": 0.7835820895522388,
      "grad_norm": 0.7122440934181213,
      "learning_rate": 1.1788241566481562e-06,
      "loss": 0.6952,
      "step": 27195
    },
    {
      "epoch": 0.7836109030138881,
      "grad_norm": 0.6962663531303406,
      "learning_rate": 1.1785232388375284e-06,
      "loss": 0.6797,
      "step": 27196
    },
    {
      "epoch": 0.7836397164755373,
      "grad_norm": 0.7177677750587463,
      "learning_rate": 1.1782223543081427e-06,
      "loss": 0.6814,
      "step": 27197
    },
    {
      "epoch": 0.7836685299371866,
      "grad_norm": 0.7181498408317566,
      "learning_rate": 1.1779215030626217e-06,
      "loss": 0.6811,
      "step": 27198
    },
    {
      "epoch": 0.7836973433988359,
      "grad_norm": 0.7252610921859741,
      "learning_rate": 1.177620685103582e-06,
      "loss": 0.6985,
      "step": 27199
    },
    {
      "epoch": 0.7837261568604852,
      "grad_norm": 0.7119203209877014,
      "learning_rate": 1.1773199004336471e-06,
      "loss": 0.6666,
      "step": 27200
    },
    {
      "epoch": 0.7837549703221345,
      "grad_norm": 0.7322803735733032,
      "learning_rate": 1.1770191490554328e-06,
      "loss": 0.7022,
      "step": 27201
    },
    {
      "epoch": 0.7837837837837838,
      "grad_norm": 0.7315750122070312,
      "learning_rate": 1.1767184309715619e-06,
      "loss": 0.6628,
      "step": 27202
    },
    {
      "epoch": 0.7838125972454331,
      "grad_norm": 0.7241635918617249,
      "learning_rate": 1.1764177461846492e-06,
      "loss": 0.6605,
      "step": 27203
    },
    {
      "epoch": 0.7838414107070824,
      "grad_norm": 0.7085424661636353,
      "learning_rate": 1.1761170946973182e-06,
      "loss": 0.6884,
      "step": 27204
    },
    {
      "epoch": 0.7838702241687316,
      "grad_norm": 0.7113511562347412,
      "learning_rate": 1.1758164765121816e-06,
      "loss": 0.6913,
      "step": 27205
    },
    {
      "epoch": 0.7838990376303809,
      "grad_norm": 0.7173693180084229,
      "learning_rate": 1.1755158916318637e-06,
      "loss": 0.6846,
      "step": 27206
    },
    {
      "epoch": 0.7839278510920302,
      "grad_norm": 0.7449926733970642,
      "learning_rate": 1.1752153400589778e-06,
      "loss": 0.6886,
      "step": 27207
    },
    {
      "epoch": 0.7839566645536795,
      "grad_norm": 0.7278683185577393,
      "learning_rate": 1.1749148217961443e-06,
      "loss": 0.6836,
      "step": 27208
    },
    {
      "epoch": 0.7839854780153288,
      "grad_norm": 0.7049155235290527,
      "learning_rate": 1.1746143368459772e-06,
      "loss": 0.6672,
      "step": 27209
    },
    {
      "epoch": 0.7840142914769781,
      "grad_norm": 0.7234271764755249,
      "learning_rate": 1.174313885211097e-06,
      "loss": 0.6892,
      "step": 27210
    },
    {
      "epoch": 0.7840431049386273,
      "grad_norm": 0.7068580389022827,
      "learning_rate": 1.1740134668941172e-06,
      "loss": 0.6965,
      "step": 27211
    },
    {
      "epoch": 0.7840719184002766,
      "grad_norm": 0.7142557501792908,
      "learning_rate": 1.1737130818976571e-06,
      "loss": 0.6864,
      "step": 27212
    },
    {
      "epoch": 0.7841007318619259,
      "grad_norm": 0.7145581245422363,
      "learning_rate": 1.1734127302243298e-06,
      "loss": 0.6801,
      "step": 27213
    },
    {
      "epoch": 0.7841295453235752,
      "grad_norm": 0.735638439655304,
      "learning_rate": 1.1731124118767518e-06,
      "loss": 0.7101,
      "step": 27214
    },
    {
      "epoch": 0.7841583587852244,
      "grad_norm": 0.712253212928772,
      "learning_rate": 1.172812126857541e-06,
      "loss": 0.6918,
      "step": 27215
    },
    {
      "epoch": 0.7841871722468737,
      "grad_norm": 0.7196796536445618,
      "learning_rate": 1.1725118751693094e-06,
      "loss": 0.6729,
      "step": 27216
    },
    {
      "epoch": 0.784215985708523,
      "grad_norm": 0.7288697361946106,
      "learning_rate": 1.1722116568146734e-06,
      "loss": 0.6899,
      "step": 27217
    },
    {
      "epoch": 0.7842447991701723,
      "grad_norm": 0.7271252274513245,
      "learning_rate": 1.1719114717962476e-06,
      "loss": 0.6817,
      "step": 27218
    },
    {
      "epoch": 0.7842736126318216,
      "grad_norm": 0.8063001036643982,
      "learning_rate": 1.1716113201166474e-06,
      "loss": 0.674,
      "step": 27219
    },
    {
      "epoch": 0.7843024260934709,
      "grad_norm": 0.6994554400444031,
      "learning_rate": 1.1713112017784838e-06,
      "loss": 0.6707,
      "step": 27220
    },
    {
      "epoch": 0.7843312395551202,
      "grad_norm": 0.7195135951042175,
      "learning_rate": 1.171011116784374e-06,
      "loss": 0.7021,
      "step": 27221
    },
    {
      "epoch": 0.7843600530167695,
      "grad_norm": 0.7177205085754395,
      "learning_rate": 1.1707110651369285e-06,
      "loss": 0.6869,
      "step": 27222
    },
    {
      "epoch": 0.7843888664784188,
      "grad_norm": 0.7188910841941833,
      "learning_rate": 1.1704110468387637e-06,
      "loss": 0.6778,
      "step": 27223
    },
    {
      "epoch": 0.784417679940068,
      "grad_norm": 0.70868319272995,
      "learning_rate": 1.1701110618924881e-06,
      "loss": 0.6559,
      "step": 27224
    },
    {
      "epoch": 0.7844464934017172,
      "grad_norm": 0.729913592338562,
      "learning_rate": 1.1698111103007188e-06,
      "loss": 0.6673,
      "step": 27225
    },
    {
      "epoch": 0.7844753068633665,
      "grad_norm": 0.7342888116836548,
      "learning_rate": 1.169511192066064e-06,
      "loss": 0.6583,
      "step": 27226
    },
    {
      "epoch": 0.7845041203250158,
      "grad_norm": 0.7432074546813965,
      "learning_rate": 1.1692113071911393e-06,
      "loss": 0.7072,
      "step": 27227
    },
    {
      "epoch": 0.7845329337866651,
      "grad_norm": 0.7213001251220703,
      "learning_rate": 1.168911455678553e-06,
      "loss": 0.7034,
      "step": 27228
    },
    {
      "epoch": 0.7845617472483144,
      "grad_norm": 0.7237422466278076,
      "learning_rate": 1.1686116375309187e-06,
      "loss": 0.6775,
      "step": 27229
    },
    {
      "epoch": 0.7845905607099637,
      "grad_norm": 0.735933780670166,
      "learning_rate": 1.168311852750847e-06,
      "loss": 0.6788,
      "step": 27230
    },
    {
      "epoch": 0.784619374171613,
      "grad_norm": 0.715765118598938,
      "learning_rate": 1.16801210134095e-06,
      "loss": 0.6981,
      "step": 27231
    },
    {
      "epoch": 0.7846481876332623,
      "grad_norm": 0.7353941202163696,
      "learning_rate": 1.1677123833038356e-06,
      "loss": 0.6849,
      "step": 27232
    },
    {
      "epoch": 0.7846770010949116,
      "grad_norm": 0.7185367345809937,
      "learning_rate": 1.167412698642117e-06,
      "loss": 0.6585,
      "step": 27233
    },
    {
      "epoch": 0.7847058145565609,
      "grad_norm": 0.7743555903434753,
      "learning_rate": 1.1671130473584015e-06,
      "loss": 0.7061,
      "step": 27234
    },
    {
      "epoch": 0.7847346280182101,
      "grad_norm": 0.7444654703140259,
      "learning_rate": 1.1668134294552992e-06,
      "loss": 0.7328,
      "step": 27235
    },
    {
      "epoch": 0.7847634414798594,
      "grad_norm": 0.7152878046035767,
      "learning_rate": 1.166513844935422e-06,
      "loss": 0.7169,
      "step": 27236
    },
    {
      "epoch": 0.7847922549415087,
      "grad_norm": 0.7055512070655823,
      "learning_rate": 1.1662142938013754e-06,
      "loss": 0.6664,
      "step": 27237
    },
    {
      "epoch": 0.7848210684031579,
      "grad_norm": 0.7197995781898499,
      "learning_rate": 1.1659147760557716e-06,
      "loss": 0.6754,
      "step": 27238
    },
    {
      "epoch": 0.7848498818648072,
      "grad_norm": 0.7444571852684021,
      "learning_rate": 1.1656152917012158e-06,
      "loss": 0.7079,
      "step": 27239
    },
    {
      "epoch": 0.7848786953264565,
      "grad_norm": 0.7027870416641235,
      "learning_rate": 1.1653158407403192e-06,
      "loss": 0.652,
      "step": 27240
    },
    {
      "epoch": 0.7849075087881058,
      "grad_norm": 0.7308266758918762,
      "learning_rate": 1.1650164231756855e-06,
      "loss": 0.6641,
      "step": 27241
    },
    {
      "epoch": 0.7849363222497551,
      "grad_norm": 0.7374881505966187,
      "learning_rate": 1.1647170390099282e-06,
      "loss": 0.6917,
      "step": 27242
    },
    {
      "epoch": 0.7849651357114044,
      "grad_norm": 0.7286730408668518,
      "learning_rate": 1.1644176882456504e-06,
      "loss": 0.7102,
      "step": 27243
    },
    {
      "epoch": 0.7849939491730537,
      "grad_norm": 0.6998944878578186,
      "learning_rate": 1.1641183708854615e-06,
      "loss": 0.6843,
      "step": 27244
    },
    {
      "epoch": 0.7850227626347029,
      "grad_norm": 0.710534393787384,
      "learning_rate": 1.163819086931966e-06,
      "loss": 0.669,
      "step": 27245
    },
    {
      "epoch": 0.7850515760963522,
      "grad_norm": 0.7163732051849365,
      "learning_rate": 1.1635198363877727e-06,
      "loss": 0.6733,
      "step": 27246
    },
    {
      "epoch": 0.7850803895580015,
      "grad_norm": 0.7181457281112671,
      "learning_rate": 1.1632206192554845e-06,
      "loss": 0.6771,
      "step": 27247
    },
    {
      "epoch": 0.7851092030196508,
      "grad_norm": 0.7044693231582642,
      "learning_rate": 1.1629214355377116e-06,
      "loss": 0.6692,
      "step": 27248
    },
    {
      "epoch": 0.7851380164813001,
      "grad_norm": 0.725979208946228,
      "learning_rate": 1.1626222852370557e-06,
      "loss": 0.6993,
      "step": 27249
    },
    {
      "epoch": 0.7851668299429494,
      "grad_norm": 0.7405851483345032,
      "learning_rate": 1.1623231683561249e-06,
      "loss": 0.6899,
      "step": 27250
    },
    {
      "epoch": 0.7851956434045986,
      "grad_norm": 0.7203330397605896,
      "learning_rate": 1.1620240848975218e-06,
      "loss": 0.6895,
      "step": 27251
    },
    {
      "epoch": 0.7852244568662479,
      "grad_norm": 0.7215494513511658,
      "learning_rate": 1.161725034863852e-06,
      "loss": 0.6878,
      "step": 27252
    },
    {
      "epoch": 0.7852532703278972,
      "grad_norm": 0.7190787196159363,
      "learning_rate": 1.1614260182577207e-06,
      "loss": 0.6846,
      "step": 27253
    },
    {
      "epoch": 0.7852820837895464,
      "grad_norm": 0.6904353499412537,
      "learning_rate": 1.1611270350817333e-06,
      "loss": 0.6578,
      "step": 27254
    },
    {
      "epoch": 0.7853108972511957,
      "grad_norm": 0.7185393571853638,
      "learning_rate": 1.1608280853384901e-06,
      "loss": 0.685,
      "step": 27255
    },
    {
      "epoch": 0.785339710712845,
      "grad_norm": 0.7612518072128296,
      "learning_rate": 1.1605291690305965e-06,
      "loss": 0.7016,
      "step": 27256
    },
    {
      "epoch": 0.7853685241744943,
      "grad_norm": 0.738319993019104,
      "learning_rate": 1.1602302861606573e-06,
      "loss": 0.7074,
      "step": 27257
    },
    {
      "epoch": 0.7853973376361436,
      "grad_norm": 0.7313337922096252,
      "learning_rate": 1.1599314367312726e-06,
      "loss": 0.6799,
      "step": 27258
    },
    {
      "epoch": 0.7854261510977929,
      "grad_norm": 0.7026184797286987,
      "learning_rate": 1.1596326207450475e-06,
      "loss": 0.6929,
      "step": 27259
    },
    {
      "epoch": 0.7854549645594422,
      "grad_norm": 0.7088682055473328,
      "learning_rate": 1.1593338382045822e-06,
      "loss": 0.6692,
      "step": 27260
    },
    {
      "epoch": 0.7854837780210915,
      "grad_norm": 0.7236157655715942,
      "learning_rate": 1.1590350891124813e-06,
      "loss": 0.7178,
      "step": 27261
    },
    {
      "epoch": 0.7855125914827408,
      "grad_norm": 0.724556028842926,
      "learning_rate": 1.1587363734713437e-06,
      "loss": 0.6645,
      "step": 27262
    },
    {
      "epoch": 0.7855414049443901,
      "grad_norm": 0.699049174785614,
      "learning_rate": 1.1584376912837736e-06,
      "loss": 0.6807,
      "step": 27263
    },
    {
      "epoch": 0.7855702184060394,
      "grad_norm": 0.7206011414527893,
      "learning_rate": 1.1581390425523702e-06,
      "loss": 0.6708,
      "step": 27264
    },
    {
      "epoch": 0.7855990318676885,
      "grad_norm": 0.7127366065979004,
      "learning_rate": 1.157840427279735e-06,
      "loss": 0.6679,
      "step": 27265
    },
    {
      "epoch": 0.7856278453293378,
      "grad_norm": 0.747123122215271,
      "learning_rate": 1.157541845468469e-06,
      "loss": 0.6857,
      "step": 27266
    },
    {
      "epoch": 0.7856566587909871,
      "grad_norm": 0.7428869605064392,
      "learning_rate": 1.1572432971211733e-06,
      "loss": 0.677,
      "step": 27267
    },
    {
      "epoch": 0.7856854722526364,
      "grad_norm": 0.7210615277290344,
      "learning_rate": 1.1569447822404467e-06,
      "loss": 0.6806,
      "step": 27268
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 0.738272488117218,
      "learning_rate": 1.1566463008288902e-06,
      "loss": 0.6947,
      "step": 27269
    },
    {
      "epoch": 0.785743099175935,
      "grad_norm": 0.7079797983169556,
      "learning_rate": 1.1563478528891008e-06,
      "loss": 0.6693,
      "step": 27270
    },
    {
      "epoch": 0.7857719126375843,
      "grad_norm": 0.7452223300933838,
      "learning_rate": 1.1560494384236814e-06,
      "loss": 0.6654,
      "step": 27271
    },
    {
      "epoch": 0.7858007260992336,
      "grad_norm": 0.8240938782691956,
      "learning_rate": 1.1557510574352272e-06,
      "loss": 0.6829,
      "step": 27272
    },
    {
      "epoch": 0.7858295395608829,
      "grad_norm": 0.7309900522232056,
      "learning_rate": 1.1554527099263386e-06,
      "loss": 0.707,
      "step": 27273
    },
    {
      "epoch": 0.7858583530225322,
      "grad_norm": 0.73140549659729,
      "learning_rate": 1.1551543958996147e-06,
      "loss": 0.6988,
      "step": 27274
    },
    {
      "epoch": 0.7858871664841814,
      "grad_norm": 0.7252138257026672,
      "learning_rate": 1.1548561153576515e-06,
      "loss": 0.6915,
      "step": 27275
    },
    {
      "epoch": 0.7859159799458307,
      "grad_norm": 0.7218452095985413,
      "learning_rate": 1.1545578683030495e-06,
      "loss": 0.6748,
      "step": 27276
    },
    {
      "epoch": 0.78594479340748,
      "grad_norm": 0.7428619265556335,
      "learning_rate": 1.154259654738401e-06,
      "loss": 0.6885,
      "step": 27277
    },
    {
      "epoch": 0.7859736068691292,
      "grad_norm": 0.7322891354560852,
      "learning_rate": 1.1539614746663097e-06,
      "loss": 0.7024,
      "step": 27278
    },
    {
      "epoch": 0.7860024203307785,
      "grad_norm": 0.7013607025146484,
      "learning_rate": 1.1536633280893684e-06,
      "loss": 0.6691,
      "step": 27279
    },
    {
      "epoch": 0.7860312337924278,
      "grad_norm": 0.7358217239379883,
      "learning_rate": 1.1533652150101754e-06,
      "loss": 0.6829,
      "step": 27280
    },
    {
      "epoch": 0.7860600472540771,
      "grad_norm": 0.6982349157333374,
      "learning_rate": 1.153067135431325e-06,
      "loss": 0.6649,
      "step": 27281
    },
    {
      "epoch": 0.7860888607157264,
      "grad_norm": 0.7340772151947021,
      "learning_rate": 1.1527690893554154e-06,
      "loss": 0.7028,
      "step": 27282
    },
    {
      "epoch": 0.7861176741773757,
      "grad_norm": 0.7345883250236511,
      "learning_rate": 1.15247107678504e-06,
      "loss": 0.6645,
      "step": 27283
    },
    {
      "epoch": 0.786146487639025,
      "grad_norm": 0.7379629015922546,
      "learning_rate": 1.1521730977227968e-06,
      "loss": 0.6735,
      "step": 27284
    },
    {
      "epoch": 0.7861753011006742,
      "grad_norm": 0.7441035509109497,
      "learning_rate": 1.151875152171278e-06,
      "loss": 0.6992,
      "step": 27285
    },
    {
      "epoch": 0.7862041145623235,
      "grad_norm": 0.7504844069480896,
      "learning_rate": 1.1515772401330815e-06,
      "loss": 0.6919,
      "step": 27286
    },
    {
      "epoch": 0.7862329280239728,
      "grad_norm": 0.7435435056686401,
      "learning_rate": 1.1512793616107987e-06,
      "loss": 0.6987,
      "step": 27287
    },
    {
      "epoch": 0.7862617414856221,
      "grad_norm": 0.7267574071884155,
      "learning_rate": 1.1509815166070265e-06,
      "loss": 0.6828,
      "step": 27288
    },
    {
      "epoch": 0.7862905549472714,
      "grad_norm": 0.7079854011535645,
      "learning_rate": 1.1506837051243552e-06,
      "loss": 0.6855,
      "step": 27289
    },
    {
      "epoch": 0.7863193684089207,
      "grad_norm": 0.7353397011756897,
      "learning_rate": 1.150385927165384e-06,
      "loss": 0.6986,
      "step": 27290
    },
    {
      "epoch": 0.78634818187057,
      "grad_norm": 0.7374000549316406,
      "learning_rate": 1.150088182732701e-06,
      "loss": 0.6976,
      "step": 27291
    },
    {
      "epoch": 0.7863769953322192,
      "grad_norm": 0.74061518907547,
      "learning_rate": 1.1497904718289033e-06,
      "loss": 0.6803,
      "step": 27292
    },
    {
      "epoch": 0.7864058087938685,
      "grad_norm": 0.7253402471542358,
      "learning_rate": 1.1494927944565803e-06,
      "loss": 0.7079,
      "step": 27293
    },
    {
      "epoch": 0.7864346222555177,
      "grad_norm": 0.731259286403656,
      "learning_rate": 1.1491951506183264e-06,
      "loss": 0.6655,
      "step": 27294
    },
    {
      "epoch": 0.786463435717167,
      "grad_norm": 0.7172602415084839,
      "learning_rate": 1.148897540316734e-06,
      "loss": 0.6811,
      "step": 27295
    },
    {
      "epoch": 0.7864922491788163,
      "grad_norm": 0.7321011424064636,
      "learning_rate": 1.148599963554393e-06,
      "loss": 0.6567,
      "step": 27296
    },
    {
      "epoch": 0.7865210626404656,
      "grad_norm": 0.7302539348602295,
      "learning_rate": 1.1483024203338982e-06,
      "loss": 0.6989,
      "step": 27297
    },
    {
      "epoch": 0.7865498761021149,
      "grad_norm": 0.724887490272522,
      "learning_rate": 1.1480049106578378e-06,
      "loss": 0.7196,
      "step": 27298
    },
    {
      "epoch": 0.7865786895637642,
      "grad_norm": 0.7237124443054199,
      "learning_rate": 1.1477074345288047e-06,
      "loss": 0.6705,
      "step": 27299
    },
    {
      "epoch": 0.7866075030254135,
      "grad_norm": 0.7323867082595825,
      "learning_rate": 1.1474099919493881e-06,
      "loss": 0.7035,
      "step": 27300
    },
    {
      "epoch": 0.7866363164870628,
      "grad_norm": 0.7293916940689087,
      "learning_rate": 1.147112582922179e-06,
      "loss": 0.6899,
      "step": 27301
    },
    {
      "epoch": 0.7866651299487121,
      "grad_norm": 0.7272076606750488,
      "learning_rate": 1.146815207449768e-06,
      "loss": 0.6857,
      "step": 27302
    },
    {
      "epoch": 0.7866939434103614,
      "grad_norm": 0.7180055975914001,
      "learning_rate": 1.1465178655347464e-06,
      "loss": 0.6877,
      "step": 27303
    },
    {
      "epoch": 0.7867227568720107,
      "grad_norm": 0.7453228831291199,
      "learning_rate": 1.1462205571797002e-06,
      "loss": 0.725,
      "step": 27304
    },
    {
      "epoch": 0.7867515703336598,
      "grad_norm": 0.7320668697357178,
      "learning_rate": 1.1459232823872225e-06,
      "loss": 0.6959,
      "step": 27305
    },
    {
      "epoch": 0.7867803837953091,
      "grad_norm": 0.7242516875267029,
      "learning_rate": 1.1456260411598984e-06,
      "loss": 0.6906,
      "step": 27306
    },
    {
      "epoch": 0.7868091972569584,
      "grad_norm": 0.7294080853462219,
      "learning_rate": 1.1453288335003205e-06,
      "loss": 0.6723,
      "step": 27307
    },
    {
      "epoch": 0.7868380107186077,
      "grad_norm": 0.7194084525108337,
      "learning_rate": 1.1450316594110734e-06,
      "loss": 0.6845,
      "step": 27308
    },
    {
      "epoch": 0.786866824180257,
      "grad_norm": 0.7080174684524536,
      "learning_rate": 1.1447345188947483e-06,
      "loss": 0.6422,
      "step": 27309
    },
    {
      "epoch": 0.7868956376419063,
      "grad_norm": 0.7411344647407532,
      "learning_rate": 1.1444374119539303e-06,
      "loss": 0.705,
      "step": 27310
    },
    {
      "epoch": 0.7869244511035556,
      "grad_norm": 0.7660857439041138,
      "learning_rate": 1.1441403385912087e-06,
      "loss": 0.6983,
      "step": 27311
    },
    {
      "epoch": 0.7869532645652049,
      "grad_norm": 0.7523914575576782,
      "learning_rate": 1.1438432988091713e-06,
      "loss": 0.7073,
      "step": 27312
    },
    {
      "epoch": 0.7869820780268542,
      "grad_norm": 0.7652953863143921,
      "learning_rate": 1.143546292610403e-06,
      "loss": 0.7047,
      "step": 27313
    },
    {
      "epoch": 0.7870108914885035,
      "grad_norm": 0.7255423665046692,
      "learning_rate": 1.143249319997491e-06,
      "loss": 0.6771,
      "step": 27314
    },
    {
      "epoch": 0.7870397049501527,
      "grad_norm": 0.7121216058731079,
      "learning_rate": 1.1429523809730225e-06,
      "loss": 0.6699,
      "step": 27315
    },
    {
      "epoch": 0.787068518411802,
      "grad_norm": 0.7047502398490906,
      "learning_rate": 1.1426554755395842e-06,
      "loss": 0.6813,
      "step": 27316
    },
    {
      "epoch": 0.7870973318734513,
      "grad_norm": 0.7083380222320557,
      "learning_rate": 1.1423586036997597e-06,
      "loss": 0.6779,
      "step": 27317
    },
    {
      "epoch": 0.7871261453351006,
      "grad_norm": 0.7241533398628235,
      "learning_rate": 1.1420617654561367e-06,
      "loss": 0.6921,
      "step": 27318
    },
    {
      "epoch": 0.7871549587967498,
      "grad_norm": 0.7450725436210632,
      "learning_rate": 1.1417649608112974e-06,
      "loss": 0.7128,
      "step": 27319
    },
    {
      "epoch": 0.7871837722583991,
      "grad_norm": 0.7104443907737732,
      "learning_rate": 1.1414681897678304e-06,
      "loss": 0.6776,
      "step": 27320
    },
    {
      "epoch": 0.7872125857200484,
      "grad_norm": 0.7040958404541016,
      "learning_rate": 1.141171452328317e-06,
      "loss": 0.6813,
      "step": 27321
    },
    {
      "epoch": 0.7872413991816977,
      "grad_norm": 0.7398683428764343,
      "learning_rate": 1.1408747484953443e-06,
      "loss": 0.6643,
      "step": 27322
    },
    {
      "epoch": 0.787270212643347,
      "grad_norm": 0.7260438203811646,
      "learning_rate": 1.1405780782714931e-06,
      "loss": 0.6736,
      "step": 27323
    },
    {
      "epoch": 0.7872990261049962,
      "grad_norm": 0.7354762554168701,
      "learning_rate": 1.14028144165935e-06,
      "loss": 0.6929,
      "step": 27324
    },
    {
      "epoch": 0.7873278395666455,
      "grad_norm": 0.7321164608001709,
      "learning_rate": 1.1399848386614948e-06,
      "loss": 0.6655,
      "step": 27325
    },
    {
      "epoch": 0.7873566530282948,
      "grad_norm": 0.7694438099861145,
      "learning_rate": 1.1396882692805161e-06,
      "loss": 0.71,
      "step": 27326
    },
    {
      "epoch": 0.7873854664899441,
      "grad_norm": 0.7096608877182007,
      "learning_rate": 1.139391733518992e-06,
      "loss": 0.672,
      "step": 27327
    },
    {
      "epoch": 0.7874142799515934,
      "grad_norm": 0.7122038006782532,
      "learning_rate": 1.1390952313795078e-06,
      "loss": 0.6788,
      "step": 27328
    },
    {
      "epoch": 0.7874430934132427,
      "grad_norm": 0.7504995465278625,
      "learning_rate": 1.1387987628646436e-06,
      "loss": 0.7056,
      "step": 27329
    },
    {
      "epoch": 0.787471906874892,
      "grad_norm": 0.7161571383476257,
      "learning_rate": 1.1385023279769842e-06,
      "loss": 0.6834,
      "step": 27330
    },
    {
      "epoch": 0.7875007203365413,
      "grad_norm": 0.7457599639892578,
      "learning_rate": 1.1382059267191071e-06,
      "loss": 0.7152,
      "step": 27331
    },
    {
      "epoch": 0.7875295337981905,
      "grad_norm": 0.7213267087936401,
      "learning_rate": 1.1379095590935967e-06,
      "loss": 0.6696,
      "step": 27332
    },
    {
      "epoch": 0.7875583472598398,
      "grad_norm": 0.7123015522956848,
      "learning_rate": 1.1376132251030347e-06,
      "loss": 0.6687,
      "step": 27333
    },
    {
      "epoch": 0.787587160721489,
      "grad_norm": 0.7124361991882324,
      "learning_rate": 1.1373169247499988e-06,
      "loss": 0.673,
      "step": 27334
    },
    {
      "epoch": 0.7876159741831383,
      "grad_norm": 0.7202966809272766,
      "learning_rate": 1.137020658037073e-06,
      "loss": 0.6633,
      "step": 27335
    },
    {
      "epoch": 0.7876447876447876,
      "grad_norm": 0.7322429418563843,
      "learning_rate": 1.1367244249668336e-06,
      "loss": 0.6946,
      "step": 27336
    },
    {
      "epoch": 0.7876736011064369,
      "grad_norm": 0.7250992059707642,
      "learning_rate": 1.1364282255418635e-06,
      "loss": 0.6794,
      "step": 27337
    },
    {
      "epoch": 0.7877024145680862,
      "grad_norm": 0.7666305303573608,
      "learning_rate": 1.1361320597647407e-06,
      "loss": 0.7025,
      "step": 27338
    },
    {
      "epoch": 0.7877312280297355,
      "grad_norm": 0.7136435508728027,
      "learning_rate": 1.135835927638047e-06,
      "loss": 0.6834,
      "step": 27339
    },
    {
      "epoch": 0.7877600414913848,
      "grad_norm": 0.6981484293937683,
      "learning_rate": 1.1355398291643582e-06,
      "loss": 0.6402,
      "step": 27340
    },
    {
      "epoch": 0.7877888549530341,
      "grad_norm": 0.7037910223007202,
      "learning_rate": 1.1352437643462555e-06,
      "loss": 0.6904,
      "step": 27341
    },
    {
      "epoch": 0.7878176684146834,
      "grad_norm": 0.7089979648590088,
      "learning_rate": 1.134947733186315e-06,
      "loss": 0.6795,
      "step": 27342
    },
    {
      "epoch": 0.7878464818763327,
      "grad_norm": 0.7121332287788391,
      "learning_rate": 1.1346517356871178e-06,
      "loss": 0.6821,
      "step": 27343
    },
    {
      "epoch": 0.787875295337982,
      "grad_norm": 0.7492150068283081,
      "learning_rate": 1.1343557718512387e-06,
      "loss": 0.6904,
      "step": 27344
    },
    {
      "epoch": 0.7879041087996312,
      "grad_norm": 0.7098567485809326,
      "learning_rate": 1.1340598416812576e-06,
      "loss": 0.6949,
      "step": 27345
    },
    {
      "epoch": 0.7879329222612804,
      "grad_norm": 0.7305237054824829,
      "learning_rate": 1.1337639451797493e-06,
      "loss": 0.6802,
      "step": 27346
    },
    {
      "epoch": 0.7879617357229297,
      "grad_norm": 0.7224000096321106,
      "learning_rate": 1.1334680823492938e-06,
      "loss": 0.6865,
      "step": 27347
    },
    {
      "epoch": 0.787990549184579,
      "grad_norm": 0.709446132183075,
      "learning_rate": 1.1331722531924637e-06,
      "loss": 0.6756,
      "step": 27348
    },
    {
      "epoch": 0.7880193626462283,
      "grad_norm": 0.7340068817138672,
      "learning_rate": 1.1328764577118406e-06,
      "loss": 0.6503,
      "step": 27349
    },
    {
      "epoch": 0.7880481761078776,
      "grad_norm": 0.7213844060897827,
      "learning_rate": 1.132580695909996e-06,
      "loss": 0.6972,
      "step": 27350
    },
    {
      "epoch": 0.7880769895695269,
      "grad_norm": 0.7081080079078674,
      "learning_rate": 1.1322849677895092e-06,
      "loss": 0.7103,
      "step": 27351
    },
    {
      "epoch": 0.7881058030311762,
      "grad_norm": 0.7273975014686584,
      "learning_rate": 1.1319892733529526e-06,
      "loss": 0.7112,
      "step": 27352
    },
    {
      "epoch": 0.7881346164928255,
      "grad_norm": 0.7175927758216858,
      "learning_rate": 1.1316936126029026e-06,
      "loss": 0.6676,
      "step": 27353
    },
    {
      "epoch": 0.7881634299544747,
      "grad_norm": 0.711736261844635,
      "learning_rate": 1.131397985541936e-06,
      "loss": 0.6669,
      "step": 27354
    },
    {
      "epoch": 0.788192243416124,
      "grad_norm": 0.7399730086326599,
      "learning_rate": 1.1311023921726243e-06,
      "loss": 0.6859,
      "step": 27355
    },
    {
      "epoch": 0.7882210568777733,
      "grad_norm": 0.7245303392410278,
      "learning_rate": 1.1308068324975453e-06,
      "loss": 0.6822,
      "step": 27356
    },
    {
      "epoch": 0.7882498703394226,
      "grad_norm": 0.7296878099441528,
      "learning_rate": 1.1305113065192686e-06,
      "loss": 0.7066,
      "step": 27357
    },
    {
      "epoch": 0.7882786838010719,
      "grad_norm": 0.7371536493301392,
      "learning_rate": 1.1302158142403725e-06,
      "loss": 0.6821,
      "step": 27358
    },
    {
      "epoch": 0.7883074972627211,
      "grad_norm": 0.7202877402305603,
      "learning_rate": 1.1299203556634265e-06,
      "loss": 0.6753,
      "step": 27359
    },
    {
      "epoch": 0.7883363107243704,
      "grad_norm": 0.7098819613456726,
      "learning_rate": 1.1296249307910057e-06,
      "loss": 0.6695,
      "step": 27360
    },
    {
      "epoch": 0.7883651241860197,
      "grad_norm": 0.6993836164474487,
      "learning_rate": 1.1293295396256825e-06,
      "loss": 0.6847,
      "step": 27361
    },
    {
      "epoch": 0.788393937647669,
      "grad_norm": 0.7302024364471436,
      "learning_rate": 1.1290341821700313e-06,
      "loss": 0.6914,
      "step": 27362
    },
    {
      "epoch": 0.7884227511093183,
      "grad_norm": 0.7015407681465149,
      "learning_rate": 1.1287388584266213e-06,
      "loss": 0.6753,
      "step": 27363
    },
    {
      "epoch": 0.7884515645709675,
      "grad_norm": 0.7337392568588257,
      "learning_rate": 1.1284435683980277e-06,
      "loss": 0.6921,
      "step": 27364
    },
    {
      "epoch": 0.7884803780326168,
      "grad_norm": 0.7430980801582336,
      "learning_rate": 1.1281483120868191e-06,
      "loss": 0.7002,
      "step": 27365
    },
    {
      "epoch": 0.7885091914942661,
      "grad_norm": 0.707592248916626,
      "learning_rate": 1.1278530894955691e-06,
      "loss": 0.6656,
      "step": 27366
    },
    {
      "epoch": 0.7885380049559154,
      "grad_norm": 0.7041339874267578,
      "learning_rate": 1.1275579006268471e-06,
      "loss": 0.6747,
      "step": 27367
    },
    {
      "epoch": 0.7885668184175647,
      "grad_norm": 0.7105223536491394,
      "learning_rate": 1.1272627454832264e-06,
      "loss": 0.6829,
      "step": 27368
    },
    {
      "epoch": 0.788595631879214,
      "grad_norm": 0.7272274494171143,
      "learning_rate": 1.126967624067274e-06,
      "loss": 0.685,
      "step": 27369
    },
    {
      "epoch": 0.7886244453408633,
      "grad_norm": 0.7525834441184998,
      "learning_rate": 1.1266725363815624e-06,
      "loss": 0.7038,
      "step": 27370
    },
    {
      "epoch": 0.7886532588025126,
      "grad_norm": 0.7157248258590698,
      "learning_rate": 1.1263774824286615e-06,
      "loss": 0.7009,
      "step": 27371
    },
    {
      "epoch": 0.7886820722641619,
      "grad_norm": 0.7225334644317627,
      "learning_rate": 1.12608246221114e-06,
      "loss": 0.6885,
      "step": 27372
    },
    {
      "epoch": 0.788710885725811,
      "grad_norm": 0.7250106930732727,
      "learning_rate": 1.1257874757315678e-06,
      "loss": 0.696,
      "step": 27373
    },
    {
      "epoch": 0.7887396991874603,
      "grad_norm": 0.764359712600708,
      "learning_rate": 1.1254925229925151e-06,
      "loss": 0.7117,
      "step": 27374
    },
    {
      "epoch": 0.7887685126491096,
      "grad_norm": 0.7292718291282654,
      "learning_rate": 1.125197603996548e-06,
      "loss": 0.7075,
      "step": 27375
    },
    {
      "epoch": 0.7887973261107589,
      "grad_norm": 0.7425129413604736,
      "learning_rate": 1.1249027187462364e-06,
      "loss": 0.678,
      "step": 27376
    },
    {
      "epoch": 0.7888261395724082,
      "grad_norm": 0.7362679839134216,
      "learning_rate": 1.1246078672441497e-06,
      "loss": 0.69,
      "step": 27377
    },
    {
      "epoch": 0.7888549530340575,
      "grad_norm": 0.7014721035957336,
      "learning_rate": 1.1243130494928533e-06,
      "loss": 0.6568,
      "step": 27378
    },
    {
      "epoch": 0.7888837664957068,
      "grad_norm": 0.7108878493309021,
      "learning_rate": 1.1240182654949172e-06,
      "loss": 0.689,
      "step": 27379
    },
    {
      "epoch": 0.7889125799573561,
      "grad_norm": 0.7115381360054016,
      "learning_rate": 1.1237235152529058e-06,
      "loss": 0.6985,
      "step": 27380
    },
    {
      "epoch": 0.7889413934190054,
      "grad_norm": 0.7175463438034058,
      "learning_rate": 1.1234287987693899e-06,
      "loss": 0.6977,
      "step": 27381
    },
    {
      "epoch": 0.7889702068806547,
      "grad_norm": 0.7303648591041565,
      "learning_rate": 1.1231341160469322e-06,
      "loss": 0.6862,
      "step": 27382
    },
    {
      "epoch": 0.788999020342304,
      "grad_norm": 0.7115087509155273,
      "learning_rate": 1.122839467088102e-06,
      "loss": 0.6692,
      "step": 27383
    },
    {
      "epoch": 0.7890278338039532,
      "grad_norm": 0.7179307341575623,
      "learning_rate": 1.1225448518954619e-06,
      "loss": 0.6686,
      "step": 27384
    },
    {
      "epoch": 0.7890566472656025,
      "grad_norm": 0.7213533520698547,
      "learning_rate": 1.1222502704715827e-06,
      "loss": 0.6875,
      "step": 27385
    },
    {
      "epoch": 0.7890854607272517,
      "grad_norm": 0.7262043356895447,
      "learning_rate": 1.121955722819026e-06,
      "loss": 0.703,
      "step": 27386
    },
    {
      "epoch": 0.789114274188901,
      "grad_norm": 0.7072544097900391,
      "learning_rate": 1.1216612089403596e-06,
      "loss": 0.6731,
      "step": 27387
    },
    {
      "epoch": 0.7891430876505503,
      "grad_norm": 0.7240755558013916,
      "learning_rate": 1.1213667288381459e-06,
      "loss": 0.7035,
      "step": 27388
    },
    {
      "epoch": 0.7891719011121996,
      "grad_norm": 0.7015955448150635,
      "learning_rate": 1.1210722825149522e-06,
      "loss": 0.6842,
      "step": 27389
    },
    {
      "epoch": 0.7892007145738489,
      "grad_norm": 0.700197160243988,
      "learning_rate": 1.1207778699733402e-06,
      "loss": 0.6472,
      "step": 27390
    },
    {
      "epoch": 0.7892295280354982,
      "grad_norm": 0.7200642228126526,
      "learning_rate": 1.1204834912158747e-06,
      "loss": 0.693,
      "step": 27391
    },
    {
      "epoch": 0.7892583414971475,
      "grad_norm": 0.7139843106269836,
      "learning_rate": 1.1201891462451215e-06,
      "loss": 0.6813,
      "step": 27392
    },
    {
      "epoch": 0.7892871549587968,
      "grad_norm": 0.7059042453765869,
      "learning_rate": 1.1198948350636406e-06,
      "loss": 0.6824,
      "step": 27393
    },
    {
      "epoch": 0.789315968420446,
      "grad_norm": 0.7174078226089478,
      "learning_rate": 1.1196005576739994e-06,
      "loss": 0.6793,
      "step": 27394
    },
    {
      "epoch": 0.7893447818820953,
      "grad_norm": 0.7554156184196472,
      "learning_rate": 1.1193063140787563e-06,
      "loss": 0.6852,
      "step": 27395
    },
    {
      "epoch": 0.7893735953437446,
      "grad_norm": 0.7146477699279785,
      "learning_rate": 1.1190121042804758e-06,
      "loss": 0.6832,
      "step": 27396
    },
    {
      "epoch": 0.7894024088053939,
      "grad_norm": 0.7236357927322388,
      "learning_rate": 1.1187179282817206e-06,
      "loss": 0.6983,
      "step": 27397
    },
    {
      "epoch": 0.7894312222670432,
      "grad_norm": 0.704593300819397,
      "learning_rate": 1.1184237860850538e-06,
      "loss": 0.7109,
      "step": 27398
    },
    {
      "epoch": 0.7894600357286925,
      "grad_norm": 0.7256060242652893,
      "learning_rate": 1.118129677693034e-06,
      "loss": 0.7081,
      "step": 27399
    },
    {
      "epoch": 0.7894888491903417,
      "grad_norm": 0.7236731052398682,
      "learning_rate": 1.117835603108226e-06,
      "loss": 0.6745,
      "step": 27400
    },
    {
      "epoch": 0.789517662651991,
      "grad_norm": 0.7056634426116943,
      "learning_rate": 1.1175415623331875e-06,
      "loss": 0.6862,
      "step": 27401
    },
    {
      "epoch": 0.7895464761136403,
      "grad_norm": 0.7177596092224121,
      "learning_rate": 1.1172475553704826e-06,
      "loss": 0.6728,
      "step": 27402
    },
    {
      "epoch": 0.7895752895752896,
      "grad_norm": 0.7276804447174072,
      "learning_rate": 1.1169535822226685e-06,
      "loss": 0.6868,
      "step": 27403
    },
    {
      "epoch": 0.7896041030369388,
      "grad_norm": 0.7259981036186218,
      "learning_rate": 1.1166596428923087e-06,
      "loss": 0.6912,
      "step": 27404
    },
    {
      "epoch": 0.7896329164985881,
      "grad_norm": 0.6983143091201782,
      "learning_rate": 1.1163657373819597e-06,
      "loss": 0.6431,
      "step": 27405
    },
    {
      "epoch": 0.7896617299602374,
      "grad_norm": 0.7281087040901184,
      "learning_rate": 1.116071865694185e-06,
      "loss": 0.6946,
      "step": 27406
    },
    {
      "epoch": 0.7896905434218867,
      "grad_norm": 0.7100383639335632,
      "learning_rate": 1.11577802783154e-06,
      "loss": 0.6695,
      "step": 27407
    },
    {
      "epoch": 0.789719356883536,
      "grad_norm": 0.7197259664535522,
      "learning_rate": 1.1154842237965858e-06,
      "loss": 0.6595,
      "step": 27408
    },
    {
      "epoch": 0.7897481703451853,
      "grad_norm": 0.7185221314430237,
      "learning_rate": 1.1151904535918807e-06,
      "loss": 0.6664,
      "step": 27409
    },
    {
      "epoch": 0.7897769838068346,
      "grad_norm": 0.7241065502166748,
      "learning_rate": 1.1148967172199848e-06,
      "loss": 0.6873,
      "step": 27410
    },
    {
      "epoch": 0.7898057972684839,
      "grad_norm": 0.7304795384407043,
      "learning_rate": 1.114603014683453e-06,
      "loss": 0.6804,
      "step": 27411
    },
    {
      "epoch": 0.7898346107301332,
      "grad_norm": 0.7252199053764343,
      "learning_rate": 1.1143093459848469e-06,
      "loss": 0.7138,
      "step": 27412
    },
    {
      "epoch": 0.7898634241917823,
      "grad_norm": 0.7270364761352539,
      "learning_rate": 1.1140157111267208e-06,
      "loss": 0.7005,
      "step": 27413
    },
    {
      "epoch": 0.7898922376534316,
      "grad_norm": 0.7503368258476257,
      "learning_rate": 1.113722110111633e-06,
      "loss": 0.703,
      "step": 27414
    },
    {
      "epoch": 0.7899210511150809,
      "grad_norm": 0.697132408618927,
      "learning_rate": 1.113428542942142e-06,
      "loss": 0.6742,
      "step": 27415
    },
    {
      "epoch": 0.7899498645767302,
      "grad_norm": 0.7627835869789124,
      "learning_rate": 1.1131350096208021e-06,
      "loss": 0.7223,
      "step": 27416
    },
    {
      "epoch": 0.7899786780383795,
      "grad_norm": 0.720299482345581,
      "learning_rate": 1.1128415101501721e-06,
      "loss": 0.683,
      "step": 27417
    },
    {
      "epoch": 0.7900074915000288,
      "grad_norm": 0.7110562324523926,
      "learning_rate": 1.1125480445328057e-06,
      "loss": 0.671,
      "step": 27418
    },
    {
      "epoch": 0.7900363049616781,
      "grad_norm": 0.7212346196174622,
      "learning_rate": 1.1122546127712614e-06,
      "loss": 0.7042,
      "step": 27419
    },
    {
      "epoch": 0.7900651184233274,
      "grad_norm": 0.7222623825073242,
      "learning_rate": 1.1119612148680898e-06,
      "loss": 0.6678,
      "step": 27420
    },
    {
      "epoch": 0.7900939318849767,
      "grad_norm": 0.719606876373291,
      "learning_rate": 1.1116678508258522e-06,
      "loss": 0.6772,
      "step": 27421
    },
    {
      "epoch": 0.790122745346626,
      "grad_norm": 0.7185183167457581,
      "learning_rate": 1.1113745206470995e-06,
      "loss": 0.6824,
      "step": 27422
    },
    {
      "epoch": 0.7901515588082753,
      "grad_norm": 0.73760986328125,
      "learning_rate": 1.1110812243343888e-06,
      "loss": 0.7122,
      "step": 27423
    },
    {
      "epoch": 0.7901803722699245,
      "grad_norm": 0.7126585245132446,
      "learning_rate": 1.1107879618902717e-06,
      "loss": 0.677,
      "step": 27424
    },
    {
      "epoch": 0.7902091857315738,
      "grad_norm": 0.7495620250701904,
      "learning_rate": 1.1104947333173056e-06,
      "loss": 0.7035,
      "step": 27425
    },
    {
      "epoch": 0.7902379991932231,
      "grad_norm": 0.7112349271774292,
      "learning_rate": 1.1102015386180404e-06,
      "loss": 0.7043,
      "step": 27426
    },
    {
      "epoch": 0.7902668126548723,
      "grad_norm": 0.7051070928573608,
      "learning_rate": 1.1099083777950325e-06,
      "loss": 0.6766,
      "step": 27427
    },
    {
      "epoch": 0.7902956261165216,
      "grad_norm": 0.7050784230232239,
      "learning_rate": 1.109615250850833e-06,
      "loss": 0.672,
      "step": 27428
    },
    {
      "epoch": 0.7903244395781709,
      "grad_norm": 0.8887553215026855,
      "learning_rate": 1.109322157787996e-06,
      "loss": 0.6677,
      "step": 27429
    },
    {
      "epoch": 0.7903532530398202,
      "grad_norm": 0.7299121618270874,
      "learning_rate": 1.1090290986090747e-06,
      "loss": 0.6797,
      "step": 27430
    },
    {
      "epoch": 0.7903820665014695,
      "grad_norm": 0.7174707055091858,
      "learning_rate": 1.1087360733166192e-06,
      "loss": 0.6894,
      "step": 27431
    },
    {
      "epoch": 0.7904108799631188,
      "grad_norm": 0.7041574120521545,
      "learning_rate": 1.1084430819131826e-06,
      "loss": 0.6969,
      "step": 27432
    },
    {
      "epoch": 0.790439693424768,
      "grad_norm": 0.7027512788772583,
      "learning_rate": 1.1081501244013176e-06,
      "loss": 0.679,
      "step": 27433
    },
    {
      "epoch": 0.7904685068864173,
      "grad_norm": 0.6977672576904297,
      "learning_rate": 1.1078572007835736e-06,
      "loss": 0.6684,
      "step": 27434
    },
    {
      "epoch": 0.7904973203480666,
      "grad_norm": 0.7582093477249146,
      "learning_rate": 1.1075643110625028e-06,
      "loss": 0.6916,
      "step": 27435
    },
    {
      "epoch": 0.7905261338097159,
      "grad_norm": 0.7061236500740051,
      "learning_rate": 1.1072714552406572e-06,
      "loss": 0.6687,
      "step": 27436
    },
    {
      "epoch": 0.7905549472713652,
      "grad_norm": 0.7789704203605652,
      "learning_rate": 1.106978633320584e-06,
      "loss": 0.7016,
      "step": 27437
    },
    {
      "epoch": 0.7905837607330145,
      "grad_norm": 0.7338630557060242,
      "learning_rate": 1.1066858453048373e-06,
      "loss": 0.6906,
      "step": 27438
    },
    {
      "epoch": 0.7906125741946638,
      "grad_norm": 0.7099214196205139,
      "learning_rate": 1.1063930911959635e-06,
      "loss": 0.6702,
      "step": 27439
    },
    {
      "epoch": 0.790641387656313,
      "grad_norm": 0.7109754085540771,
      "learning_rate": 1.1061003709965152e-06,
      "loss": 0.6748,
      "step": 27440
    },
    {
      "epoch": 0.7906702011179623,
      "grad_norm": 0.7179088592529297,
      "learning_rate": 1.1058076847090381e-06,
      "loss": 0.7069,
      "step": 27441
    },
    {
      "epoch": 0.7906990145796116,
      "grad_norm": 0.7033887505531311,
      "learning_rate": 1.1055150323360852e-06,
      "loss": 0.6708,
      "step": 27442
    },
    {
      "epoch": 0.7907278280412608,
      "grad_norm": 0.7098377346992493,
      "learning_rate": 1.105222413880202e-06,
      "loss": 0.6895,
      "step": 27443
    },
    {
      "epoch": 0.7907566415029101,
      "grad_norm": 0.7246399521827698,
      "learning_rate": 1.104929829343938e-06,
      "loss": 0.6704,
      "step": 27444
    },
    {
      "epoch": 0.7907854549645594,
      "grad_norm": 0.6945539712905884,
      "learning_rate": 1.104637278729842e-06,
      "loss": 0.6587,
      "step": 27445
    },
    {
      "epoch": 0.7908142684262087,
      "grad_norm": 0.7205718755722046,
      "learning_rate": 1.1043447620404625e-06,
      "loss": 0.6775,
      "step": 27446
    },
    {
      "epoch": 0.790843081887858,
      "grad_norm": 0.7398055791854858,
      "learning_rate": 1.1040522792783443e-06,
      "loss": 0.6914,
      "step": 27447
    },
    {
      "epoch": 0.7908718953495073,
      "grad_norm": 0.7442137002944946,
      "learning_rate": 1.103759830446038e-06,
      "loss": 0.6724,
      "step": 27448
    },
    {
      "epoch": 0.7909007088111566,
      "grad_norm": 0.7319899797439575,
      "learning_rate": 1.1034674155460868e-06,
      "loss": 0.6956,
      "step": 27449
    },
    {
      "epoch": 0.7909295222728059,
      "grad_norm": 0.727571427822113,
      "learning_rate": 1.10317503458104e-06,
      "loss": 0.678,
      "step": 27450
    },
    {
      "epoch": 0.7909583357344552,
      "grad_norm": 0.731056809425354,
      "learning_rate": 1.1028826875534438e-06,
      "loss": 0.683,
      "step": 27451
    },
    {
      "epoch": 0.7909871491961045,
      "grad_norm": 0.7168851494789124,
      "learning_rate": 1.1025903744658428e-06,
      "loss": 0.6749,
      "step": 27452
    },
    {
      "epoch": 0.7910159626577538,
      "grad_norm": 0.7400619983673096,
      "learning_rate": 1.1022980953207851e-06,
      "loss": 0.7021,
      "step": 27453
    },
    {
      "epoch": 0.7910447761194029,
      "grad_norm": 0.7465507984161377,
      "learning_rate": 1.102005850120813e-06,
      "loss": 0.6901,
      "step": 27454
    },
    {
      "epoch": 0.7910735895810522,
      "grad_norm": 0.7098546028137207,
      "learning_rate": 1.101713638868475e-06,
      "loss": 0.6702,
      "step": 27455
    },
    {
      "epoch": 0.7911024030427015,
      "grad_norm": 0.7071537375450134,
      "learning_rate": 1.1014214615663116e-06,
      "loss": 0.6875,
      "step": 27456
    },
    {
      "epoch": 0.7911312165043508,
      "grad_norm": 0.7171492576599121,
      "learning_rate": 1.101129318216873e-06,
      "loss": 0.6783,
      "step": 27457
    },
    {
      "epoch": 0.7911600299660001,
      "grad_norm": 0.7293123602867126,
      "learning_rate": 1.1008372088226994e-06,
      "loss": 0.7025,
      "step": 27458
    },
    {
      "epoch": 0.7911888434276494,
      "grad_norm": 0.7284005880355835,
      "learning_rate": 1.100545133386337e-06,
      "loss": 0.6657,
      "step": 27459
    },
    {
      "epoch": 0.7912176568892987,
      "grad_norm": 0.7250007390975952,
      "learning_rate": 1.1002530919103271e-06,
      "loss": 0.6936,
      "step": 27460
    },
    {
      "epoch": 0.791246470350948,
      "grad_norm": 0.7058360576629639,
      "learning_rate": 1.0999610843972158e-06,
      "loss": 0.6815,
      "step": 27461
    },
    {
      "epoch": 0.7912752838125973,
      "grad_norm": 0.7233611941337585,
      "learning_rate": 1.0996691108495433e-06,
      "loss": 0.7113,
      "step": 27462
    },
    {
      "epoch": 0.7913040972742466,
      "grad_norm": 0.7056329846382141,
      "learning_rate": 1.0993771712698553e-06,
      "loss": 0.6719,
      "step": 27463
    },
    {
      "epoch": 0.7913329107358958,
      "grad_norm": 0.7266027927398682,
      "learning_rate": 1.099085265660692e-06,
      "loss": 0.6679,
      "step": 27464
    },
    {
      "epoch": 0.7913617241975451,
      "grad_norm": 0.7380673289299011,
      "learning_rate": 1.0987933940245976e-06,
      "loss": 0.699,
      "step": 27465
    },
    {
      "epoch": 0.7913905376591944,
      "grad_norm": 0.7289379835128784,
      "learning_rate": 1.0985015563641122e-06,
      "loss": 0.7014,
      "step": 27466
    },
    {
      "epoch": 0.7914193511208436,
      "grad_norm": 0.7050953507423401,
      "learning_rate": 1.0982097526817776e-06,
      "loss": 0.6737,
      "step": 27467
    },
    {
      "epoch": 0.7914481645824929,
      "grad_norm": 0.720760703086853,
      "learning_rate": 1.0979179829801356e-06,
      "loss": 0.7019,
      "step": 27468
    },
    {
      "epoch": 0.7914769780441422,
      "grad_norm": 0.7314438223838806,
      "learning_rate": 1.0976262472617288e-06,
      "loss": 0.6896,
      "step": 27469
    },
    {
      "epoch": 0.7915057915057915,
      "grad_norm": 0.7276628613471985,
      "learning_rate": 1.0973345455290952e-06,
      "loss": 0.6799,
      "step": 27470
    },
    {
      "epoch": 0.7915346049674408,
      "grad_norm": 0.7141009569168091,
      "learning_rate": 1.0970428777847776e-06,
      "loss": 0.679,
      "step": 27471
    },
    {
      "epoch": 0.7915634184290901,
      "grad_norm": 0.7261364459991455,
      "learning_rate": 1.0967512440313137e-06,
      "loss": 0.6948,
      "step": 27472
    },
    {
      "epoch": 0.7915922318907394,
      "grad_norm": 0.741628885269165,
      "learning_rate": 1.0964596442712449e-06,
      "loss": 0.6934,
      "step": 27473
    },
    {
      "epoch": 0.7916210453523886,
      "grad_norm": 0.7360972166061401,
      "learning_rate": 1.0961680785071117e-06,
      "loss": 0.6956,
      "step": 27474
    },
    {
      "epoch": 0.7916498588140379,
      "grad_norm": 0.7589031457901001,
      "learning_rate": 1.0958765467414505e-06,
      "loss": 0.703,
      "step": 27475
    },
    {
      "epoch": 0.7916786722756872,
      "grad_norm": 0.7251346707344055,
      "learning_rate": 1.0955850489768038e-06,
      "loss": 0.6765,
      "step": 27476
    },
    {
      "epoch": 0.7917074857373365,
      "grad_norm": 0.6999935507774353,
      "learning_rate": 1.0952935852157065e-06,
      "loss": 0.6694,
      "step": 27477
    },
    {
      "epoch": 0.7917362991989858,
      "grad_norm": 0.7077770233154297,
      "learning_rate": 1.0950021554607004e-06,
      "loss": 0.707,
      "step": 27478
    },
    {
      "epoch": 0.7917651126606351,
      "grad_norm": 0.6956283450126648,
      "learning_rate": 1.094710759714321e-06,
      "loss": 0.6735,
      "step": 27479
    },
    {
      "epoch": 0.7917939261222844,
      "grad_norm": 0.7265126705169678,
      "learning_rate": 1.0944193979791061e-06,
      "loss": 0.6786,
      "step": 27480
    },
    {
      "epoch": 0.7918227395839336,
      "grad_norm": 0.695098340511322,
      "learning_rate": 1.0941280702575946e-06,
      "loss": 0.6722,
      "step": 27481
    },
    {
      "epoch": 0.7918515530455829,
      "grad_norm": 0.7197763919830322,
      "learning_rate": 1.0938367765523244e-06,
      "loss": 0.6834,
      "step": 27482
    },
    {
      "epoch": 0.7918803665072321,
      "grad_norm": 0.7322452068328857,
      "learning_rate": 1.0935455168658298e-06,
      "loss": 0.6771,
      "step": 27483
    },
    {
      "epoch": 0.7919091799688814,
      "grad_norm": 0.7217839360237122,
      "learning_rate": 1.0932542912006505e-06,
      "loss": 0.6877,
      "step": 27484
    },
    {
      "epoch": 0.7919379934305307,
      "grad_norm": 0.7121901512145996,
      "learning_rate": 1.0929630995593194e-06,
      "loss": 0.6747,
      "step": 27485
    },
    {
      "epoch": 0.79196680689218,
      "grad_norm": 0.7339041829109192,
      "learning_rate": 1.092671941944375e-06,
      "loss": 0.6903,
      "step": 27486
    },
    {
      "epoch": 0.7919956203538293,
      "grad_norm": 0.7327755689620972,
      "learning_rate": 1.0923808183583513e-06,
      "loss": 0.6936,
      "step": 27487
    },
    {
      "epoch": 0.7920244338154786,
      "grad_norm": 0.7122098207473755,
      "learning_rate": 1.092089728803784e-06,
      "loss": 0.6867,
      "step": 27488
    },
    {
      "epoch": 0.7920532472771279,
      "grad_norm": 0.7158411145210266,
      "learning_rate": 1.0917986732832104e-06,
      "loss": 0.6815,
      "step": 27489
    },
    {
      "epoch": 0.7920820607387772,
      "grad_norm": 0.7223653793334961,
      "learning_rate": 1.0915076517991618e-06,
      "loss": 0.6567,
      "step": 27490
    },
    {
      "epoch": 0.7921108742004265,
      "grad_norm": 0.721645712852478,
      "learning_rate": 1.0912166643541766e-06,
      "loss": 0.6832,
      "step": 27491
    },
    {
      "epoch": 0.7921396876620758,
      "grad_norm": 0.7241538166999817,
      "learning_rate": 1.0909257109507848e-06,
      "loss": 0.6996,
      "step": 27492
    },
    {
      "epoch": 0.792168501123725,
      "grad_norm": 0.7151208519935608,
      "learning_rate": 1.0906347915915228e-06,
      "loss": 0.6731,
      "step": 27493
    },
    {
      "epoch": 0.7921973145853742,
      "grad_norm": 0.716454267501831,
      "learning_rate": 1.090343906278924e-06,
      "loss": 0.6746,
      "step": 27494
    },
    {
      "epoch": 0.7922261280470235,
      "grad_norm": 0.7362104058265686,
      "learning_rate": 1.0900530550155225e-06,
      "loss": 0.6857,
      "step": 27495
    },
    {
      "epoch": 0.7922549415086728,
      "grad_norm": 0.7223495841026306,
      "learning_rate": 1.0897622378038491e-06,
      "loss": 0.667,
      "step": 27496
    },
    {
      "epoch": 0.7922837549703221,
      "grad_norm": 0.712446928024292,
      "learning_rate": 1.0894714546464391e-06,
      "loss": 0.676,
      "step": 27497
    },
    {
      "epoch": 0.7923125684319714,
      "grad_norm": 0.693390965461731,
      "learning_rate": 1.0891807055458226e-06,
      "loss": 0.6765,
      "step": 27498
    },
    {
      "epoch": 0.7923413818936207,
      "grad_norm": 0.7216341495513916,
      "learning_rate": 1.088889990504534e-06,
      "loss": 0.6644,
      "step": 27499
    },
    {
      "epoch": 0.79237019535527,
      "grad_norm": 0.7103592157363892,
      "learning_rate": 1.0885993095251024e-06,
      "loss": 0.6804,
      "step": 27500
    },
    {
      "epoch": 0.7923990088169193,
      "grad_norm": 0.712307870388031,
      "learning_rate": 1.0883086626100625e-06,
      "loss": 0.6892,
      "step": 27501
    },
    {
      "epoch": 0.7924278222785686,
      "grad_norm": 0.7314326167106628,
      "learning_rate": 1.0880180497619419e-06,
      "loss": 0.6828,
      "step": 27502
    },
    {
      "epoch": 0.7924566357402179,
      "grad_norm": 0.7356597781181335,
      "learning_rate": 1.0877274709832758e-06,
      "loss": 0.6938,
      "step": 27503
    },
    {
      "epoch": 0.7924854492018671,
      "grad_norm": 0.7343268394470215,
      "learning_rate": 1.0874369262765894e-06,
      "loss": 0.6651,
      "step": 27504
    },
    {
      "epoch": 0.7925142626635164,
      "grad_norm": 0.7179305553436279,
      "learning_rate": 1.087146415644419e-06,
      "loss": 0.6815,
      "step": 27505
    },
    {
      "epoch": 0.7925430761251657,
      "grad_norm": 0.7254303693771362,
      "learning_rate": 1.0868559390892903e-06,
      "loss": 0.692,
      "step": 27506
    },
    {
      "epoch": 0.792571889586815,
      "grad_norm": 0.7198156118392944,
      "learning_rate": 1.0865654966137362e-06,
      "loss": 0.6809,
      "step": 27507
    },
    {
      "epoch": 0.7926007030484642,
      "grad_norm": 0.7327004075050354,
      "learning_rate": 1.0862750882202833e-06,
      "loss": 0.6769,
      "step": 27508
    },
    {
      "epoch": 0.7926295165101135,
      "grad_norm": 0.7350609302520752,
      "learning_rate": 1.0859847139114631e-06,
      "loss": 0.6817,
      "step": 27509
    },
    {
      "epoch": 0.7926583299717628,
      "grad_norm": 0.778035581111908,
      "learning_rate": 1.0856943736898023e-06,
      "loss": 0.6926,
      "step": 27510
    },
    {
      "epoch": 0.7926871434334121,
      "grad_norm": 0.7109453678131104,
      "learning_rate": 1.085404067557831e-06,
      "loss": 0.7126,
      "step": 27511
    },
    {
      "epoch": 0.7927159568950614,
      "grad_norm": 0.741679310798645,
      "learning_rate": 1.085113795518078e-06,
      "loss": 0.6919,
      "step": 27512
    },
    {
      "epoch": 0.7927447703567106,
      "grad_norm": 0.7227869629859924,
      "learning_rate": 1.0848235575730692e-06,
      "loss": 0.6618,
      "step": 27513
    },
    {
      "epoch": 0.7927735838183599,
      "grad_norm": 0.7347277998924255,
      "learning_rate": 1.084533353725335e-06,
      "loss": 0.6957,
      "step": 27514
    },
    {
      "epoch": 0.7928023972800092,
      "grad_norm": 0.7279448509216309,
      "learning_rate": 1.0842431839773998e-06,
      "loss": 0.6819,
      "step": 27515
    },
    {
      "epoch": 0.7928312107416585,
      "grad_norm": 0.7349333763122559,
      "learning_rate": 1.0839530483317922e-06,
      "loss": 0.691,
      "step": 27516
    },
    {
      "epoch": 0.7928600242033078,
      "grad_norm": 0.7156439423561096,
      "learning_rate": 1.0836629467910387e-06,
      "loss": 0.6746,
      "step": 27517
    },
    {
      "epoch": 0.7928888376649571,
      "grad_norm": 0.7223895192146301,
      "learning_rate": 1.0833728793576676e-06,
      "loss": 0.6984,
      "step": 27518
    },
    {
      "epoch": 0.7929176511266064,
      "grad_norm": 0.7467368841171265,
      "learning_rate": 1.0830828460342018e-06,
      "loss": 0.6988,
      "step": 27519
    },
    {
      "epoch": 0.7929464645882557,
      "grad_norm": 0.7165420055389404,
      "learning_rate": 1.0827928468231707e-06,
      "loss": 0.7264,
      "step": 27520
    },
    {
      "epoch": 0.7929752780499049,
      "grad_norm": 0.7148466110229492,
      "learning_rate": 1.0825028817270967e-06,
      "loss": 0.6987,
      "step": 27521
    },
    {
      "epoch": 0.7930040915115542,
      "grad_norm": 0.716376006603241,
      "learning_rate": 1.082212950748508e-06,
      "loss": 0.6801,
      "step": 27522
    },
    {
      "epoch": 0.7930329049732034,
      "grad_norm": 0.725205659866333,
      "learning_rate": 1.0819230538899267e-06,
      "loss": 0.6762,
      "step": 27523
    },
    {
      "epoch": 0.7930617184348527,
      "grad_norm": 0.7474694848060608,
      "learning_rate": 1.0816331911538802e-06,
      "loss": 0.6999,
      "step": 27524
    },
    {
      "epoch": 0.793090531896502,
      "grad_norm": 0.7427866458892822,
      "learning_rate": 1.0813433625428904e-06,
      "loss": 0.6855,
      "step": 27525
    },
    {
      "epoch": 0.7931193453581513,
      "grad_norm": 0.7238495945930481,
      "learning_rate": 1.0810535680594825e-06,
      "loss": 0.6712,
      "step": 27526
    },
    {
      "epoch": 0.7931481588198006,
      "grad_norm": 0.732609212398529,
      "learning_rate": 1.0807638077061822e-06,
      "loss": 0.6987,
      "step": 27527
    },
    {
      "epoch": 0.7931769722814499,
      "grad_norm": 0.7327987551689148,
      "learning_rate": 1.0804740814855097e-06,
      "loss": 0.7038,
      "step": 27528
    },
    {
      "epoch": 0.7932057857430992,
      "grad_norm": 0.7185366153717041,
      "learning_rate": 1.0801843893999898e-06,
      "loss": 0.667,
      "step": 27529
    },
    {
      "epoch": 0.7932345992047485,
      "grad_norm": 0.7110937833786011,
      "learning_rate": 1.079894731452147e-06,
      "loss": 0.6944,
      "step": 27530
    },
    {
      "epoch": 0.7932634126663978,
      "grad_norm": 0.7610950469970703,
      "learning_rate": 1.0796051076445012e-06,
      "loss": 0.6799,
      "step": 27531
    },
    {
      "epoch": 0.7932922261280471,
      "grad_norm": 0.7153584361076355,
      "learning_rate": 1.0793155179795757e-06,
      "loss": 0.706,
      "step": 27532
    },
    {
      "epoch": 0.7933210395896964,
      "grad_norm": 0.7177065014839172,
      "learning_rate": 1.0790259624598941e-06,
      "loss": 0.676,
      "step": 27533
    },
    {
      "epoch": 0.7933498530513456,
      "grad_norm": 0.7244821190834045,
      "learning_rate": 1.078736441087976e-06,
      "loss": 0.6819,
      "step": 27534
    },
    {
      "epoch": 0.7933786665129948,
      "grad_norm": 0.7267768979072571,
      "learning_rate": 1.0784469538663445e-06,
      "loss": 0.6887,
      "step": 27535
    },
    {
      "epoch": 0.7934074799746441,
      "grad_norm": 0.7329465746879578,
      "learning_rate": 1.0781575007975187e-06,
      "loss": 0.6849,
      "step": 27536
    },
    {
      "epoch": 0.7934362934362934,
      "grad_norm": 0.7302733659744263,
      "learning_rate": 1.077868081884022e-06,
      "loss": 0.7011,
      "step": 27537
    },
    {
      "epoch": 0.7934651068979427,
      "grad_norm": 0.7221158742904663,
      "learning_rate": 1.0775786971283725e-06,
      "loss": 0.6753,
      "step": 27538
    },
    {
      "epoch": 0.793493920359592,
      "grad_norm": 0.7297218441963196,
      "learning_rate": 1.0772893465330924e-06,
      "loss": 0.7016,
      "step": 27539
    },
    {
      "epoch": 0.7935227338212413,
      "grad_norm": 0.7378271818161011,
      "learning_rate": 1.0770000301006984e-06,
      "loss": 0.6682,
      "step": 27540
    },
    {
      "epoch": 0.7935515472828906,
      "grad_norm": 0.715740442276001,
      "learning_rate": 1.0767107478337152e-06,
      "loss": 0.7019,
      "step": 27541
    },
    {
      "epoch": 0.7935803607445399,
      "grad_norm": 0.7218464016914368,
      "learning_rate": 1.0764214997346583e-06,
      "loss": 0.6897,
      "step": 27542
    },
    {
      "epoch": 0.7936091742061891,
      "grad_norm": 0.7102806568145752,
      "learning_rate": 1.0761322858060498e-06,
      "loss": 0.6753,
      "step": 27543
    },
    {
      "epoch": 0.7936379876678384,
      "grad_norm": 0.7064307928085327,
      "learning_rate": 1.0758431060504044e-06,
      "loss": 0.6688,
      "step": 27544
    },
    {
      "epoch": 0.7936668011294877,
      "grad_norm": 0.7200740575790405,
      "learning_rate": 1.0755539604702447e-06,
      "loss": 0.6927,
      "step": 27545
    },
    {
      "epoch": 0.793695614591137,
      "grad_norm": 0.7219811677932739,
      "learning_rate": 1.0752648490680855e-06,
      "loss": 0.6884,
      "step": 27546
    },
    {
      "epoch": 0.7937244280527863,
      "grad_norm": 0.7379978895187378,
      "learning_rate": 1.0749757718464476e-06,
      "loss": 0.6867,
      "step": 27547
    },
    {
      "epoch": 0.7937532415144355,
      "grad_norm": 0.698183536529541,
      "learning_rate": 1.0746867288078456e-06,
      "loss": 0.6811,
      "step": 27548
    },
    {
      "epoch": 0.7937820549760848,
      "grad_norm": 0.7063596248626709,
      "learning_rate": 1.0743977199547983e-06,
      "loss": 0.6794,
      "step": 27549
    },
    {
      "epoch": 0.7938108684377341,
      "grad_norm": 0.712377667427063,
      "learning_rate": 1.0741087452898247e-06,
      "loss": 0.6966,
      "step": 27550
    },
    {
      "epoch": 0.7938396818993834,
      "grad_norm": 0.7249626517295837,
      "learning_rate": 1.0738198048154374e-06,
      "loss": 0.6828,
      "step": 27551
    },
    {
      "epoch": 0.7938684953610327,
      "grad_norm": 0.7140086889266968,
      "learning_rate": 1.0735308985341553e-06,
      "loss": 0.6784,
      "step": 27552
    },
    {
      "epoch": 0.793897308822682,
      "grad_norm": 0.7414771914482117,
      "learning_rate": 1.0732420264484933e-06,
      "loss": 0.6923,
      "step": 27553
    },
    {
      "epoch": 0.7939261222843312,
      "grad_norm": 0.7143027186393738,
      "learning_rate": 1.0729531885609702e-06,
      "loss": 0.6946,
      "step": 27554
    },
    {
      "epoch": 0.7939549357459805,
      "grad_norm": 0.7351272702217102,
      "learning_rate": 1.0726643848740974e-06,
      "loss": 0.6856,
      "step": 27555
    },
    {
      "epoch": 0.7939837492076298,
      "grad_norm": 0.7046842575073242,
      "learning_rate": 1.072375615390393e-06,
      "loss": 0.6673,
      "step": 27556
    },
    {
      "epoch": 0.7940125626692791,
      "grad_norm": 0.7422891855239868,
      "learning_rate": 1.07208688011237e-06,
      "loss": 0.704,
      "step": 27557
    },
    {
      "epoch": 0.7940413761309284,
      "grad_norm": 0.7130206227302551,
      "learning_rate": 1.0717981790425453e-06,
      "loss": 0.6734,
      "step": 27558
    },
    {
      "epoch": 0.7940701895925777,
      "grad_norm": 0.7151420712471008,
      "learning_rate": 1.0715095121834297e-06,
      "loss": 0.6609,
      "step": 27559
    },
    {
      "epoch": 0.794099003054227,
      "grad_norm": 0.7168670296669006,
      "learning_rate": 1.0712208795375407e-06,
      "loss": 0.6887,
      "step": 27560
    },
    {
      "epoch": 0.7941278165158763,
      "grad_norm": 0.6915504932403564,
      "learning_rate": 1.070932281107389e-06,
      "loss": 0.6803,
      "step": 27561
    },
    {
      "epoch": 0.7941566299775255,
      "grad_norm": 0.7540974617004395,
      "learning_rate": 1.0706437168954914e-06,
      "loss": 0.6773,
      "step": 27562
    },
    {
      "epoch": 0.7941854434391747,
      "grad_norm": 0.7288994193077087,
      "learning_rate": 1.0703551869043576e-06,
      "loss": 0.6894,
      "step": 27563
    },
    {
      "epoch": 0.794214256900824,
      "grad_norm": 0.744178295135498,
      "learning_rate": 1.0700666911365016e-06,
      "loss": 0.6511,
      "step": 27564
    },
    {
      "epoch": 0.7942430703624733,
      "grad_norm": 0.7264887690544128,
      "learning_rate": 1.0697782295944365e-06,
      "loss": 0.6647,
      "step": 27565
    },
    {
      "epoch": 0.7942718838241226,
      "grad_norm": 0.7165972590446472,
      "learning_rate": 1.0694898022806755e-06,
      "loss": 0.6879,
      "step": 27566
    },
    {
      "epoch": 0.7943006972857719,
      "grad_norm": 0.7098684906959534,
      "learning_rate": 1.0692014091977282e-06,
      "loss": 0.6726,
      "step": 27567
    },
    {
      "epoch": 0.7943295107474212,
      "grad_norm": 0.7109886407852173,
      "learning_rate": 1.0689130503481083e-06,
      "loss": 0.6723,
      "step": 27568
    },
    {
      "epoch": 0.7943583242090705,
      "grad_norm": 0.7315764427185059,
      "learning_rate": 1.0686247257343247e-06,
      "loss": 0.6882,
      "step": 27569
    },
    {
      "epoch": 0.7943871376707198,
      "grad_norm": 0.7143970131874084,
      "learning_rate": 1.06833643535889e-06,
      "loss": 0.6834,
      "step": 27570
    },
    {
      "epoch": 0.7944159511323691,
      "grad_norm": 0.7162578701972961,
      "learning_rate": 1.0680481792243159e-06,
      "loss": 0.6941,
      "step": 27571
    },
    {
      "epoch": 0.7944447645940184,
      "grad_norm": 0.7125624418258667,
      "learning_rate": 1.06775995733311e-06,
      "loss": 0.6919,
      "step": 27572
    },
    {
      "epoch": 0.7944735780556677,
      "grad_norm": 0.7169429063796997,
      "learning_rate": 1.0674717696877856e-06,
      "loss": 0.6592,
      "step": 27573
    },
    {
      "epoch": 0.7945023915173169,
      "grad_norm": 0.7293077707290649,
      "learning_rate": 1.06718361629085e-06,
      "loss": 0.6922,
      "step": 27574
    },
    {
      "epoch": 0.7945312049789661,
      "grad_norm": 0.720796525478363,
      "learning_rate": 1.0668954971448149e-06,
      "loss": 0.6722,
      "step": 27575
    },
    {
      "epoch": 0.7945600184406154,
      "grad_norm": 0.7189416885375977,
      "learning_rate": 1.0666074122521857e-06,
      "loss": 0.6795,
      "step": 27576
    },
    {
      "epoch": 0.7945888319022647,
      "grad_norm": 0.7380032539367676,
      "learning_rate": 1.0663193616154765e-06,
      "loss": 0.6937,
      "step": 27577
    },
    {
      "epoch": 0.794617645363914,
      "grad_norm": 0.7266852855682373,
      "learning_rate": 1.0660313452371922e-06,
      "loss": 0.6805,
      "step": 27578
    },
    {
      "epoch": 0.7946464588255633,
      "grad_norm": 0.7436792850494385,
      "learning_rate": 1.0657433631198438e-06,
      "loss": 0.6804,
      "step": 27579
    },
    {
      "epoch": 0.7946752722872126,
      "grad_norm": 0.7186198830604553,
      "learning_rate": 1.0654554152659363e-06,
      "loss": 0.6715,
      "step": 27580
    },
    {
      "epoch": 0.7947040857488619,
      "grad_norm": 0.7355993986129761,
      "learning_rate": 1.0651675016779806e-06,
      "loss": 0.6917,
      "step": 27581
    },
    {
      "epoch": 0.7947328992105112,
      "grad_norm": 0.7085148096084595,
      "learning_rate": 1.0648796223584812e-06,
      "loss": 0.6698,
      "step": 27582
    },
    {
      "epoch": 0.7947617126721604,
      "grad_norm": 0.7403984069824219,
      "learning_rate": 1.0645917773099479e-06,
      "loss": 0.7146,
      "step": 27583
    },
    {
      "epoch": 0.7947905261338097,
      "grad_norm": 0.7081322073936462,
      "learning_rate": 1.0643039665348847e-06,
      "loss": 0.6794,
      "step": 27584
    },
    {
      "epoch": 0.794819339595459,
      "grad_norm": 0.7541942000389099,
      "learning_rate": 1.0640161900358009e-06,
      "loss": 0.695,
      "step": 27585
    },
    {
      "epoch": 0.7948481530571083,
      "grad_norm": 0.7349581122398376,
      "learning_rate": 1.063728447815201e-06,
      "loss": 0.7032,
      "step": 27586
    },
    {
      "epoch": 0.7948769665187576,
      "grad_norm": 0.7245550751686096,
      "learning_rate": 1.0634407398755908e-06,
      "loss": 0.7023,
      "step": 27587
    },
    {
      "epoch": 0.7949057799804069,
      "grad_norm": 0.7082794308662415,
      "learning_rate": 1.0631530662194767e-06,
      "loss": 0.6664,
      "step": 27588
    },
    {
      "epoch": 0.7949345934420561,
      "grad_norm": 0.7069796323776245,
      "learning_rate": 1.0628654268493654e-06,
      "loss": 0.672,
      "step": 27589
    },
    {
      "epoch": 0.7949634069037054,
      "grad_norm": 0.7316712737083435,
      "learning_rate": 1.0625778217677597e-06,
      "loss": 0.6899,
      "step": 27590
    },
    {
      "epoch": 0.7949922203653547,
      "grad_norm": 0.7226420640945435,
      "learning_rate": 1.0622902509771643e-06,
      "loss": 0.6855,
      "step": 27591
    },
    {
      "epoch": 0.795021033827004,
      "grad_norm": 0.7269400358200073,
      "learning_rate": 1.0620027144800865e-06,
      "loss": 0.7069,
      "step": 27592
    },
    {
      "epoch": 0.7950498472886532,
      "grad_norm": 0.7073847055435181,
      "learning_rate": 1.061715212279027e-06,
      "loss": 0.6723,
      "step": 27593
    },
    {
      "epoch": 0.7950786607503025,
      "grad_norm": 0.701021671295166,
      "learning_rate": 1.0614277443764926e-06,
      "loss": 0.6835,
      "step": 27594
    },
    {
      "epoch": 0.7951074742119518,
      "grad_norm": 0.7634377479553223,
      "learning_rate": 1.0611403107749841e-06,
      "loss": 0.7073,
      "step": 27595
    },
    {
      "epoch": 0.7951362876736011,
      "grad_norm": 0.7099233269691467,
      "learning_rate": 1.0608529114770073e-06,
      "loss": 0.6987,
      "step": 27596
    },
    {
      "epoch": 0.7951651011352504,
      "grad_norm": 0.7130380272865295,
      "learning_rate": 1.0605655464850623e-06,
      "loss": 0.6776,
      "step": 27597
    },
    {
      "epoch": 0.7951939145968997,
      "grad_norm": 0.6973452568054199,
      "learning_rate": 1.060278215801655e-06,
      "loss": 0.661,
      "step": 27598
    },
    {
      "epoch": 0.795222728058549,
      "grad_norm": 0.7345787286758423,
      "learning_rate": 1.0599909194292852e-06,
      "loss": 0.6771,
      "step": 27599
    },
    {
      "epoch": 0.7952515415201983,
      "grad_norm": 0.7174092531204224,
      "learning_rate": 1.0597036573704556e-06,
      "loss": 0.6809,
      "step": 27600
    },
    {
      "epoch": 0.7952803549818476,
      "grad_norm": 0.7310867309570312,
      "learning_rate": 1.0594164296276677e-06,
      "loss": 0.6913,
      "step": 27601
    },
    {
      "epoch": 0.7953091684434968,
      "grad_norm": 0.7403260469436646,
      "learning_rate": 1.0591292362034255e-06,
      "loss": 0.657,
      "step": 27602
    },
    {
      "epoch": 0.795337981905146,
      "grad_norm": 0.7392904758453369,
      "learning_rate": 1.0588420771002273e-06,
      "loss": 0.6949,
      "step": 27603
    },
    {
      "epoch": 0.7953667953667953,
      "grad_norm": 0.7192649245262146,
      "learning_rate": 1.0585549523205757e-06,
      "loss": 0.6586,
      "step": 27604
    },
    {
      "epoch": 0.7953956088284446,
      "grad_norm": 0.7312783002853394,
      "learning_rate": 1.058267861866969e-06,
      "loss": 0.6776,
      "step": 27605
    },
    {
      "epoch": 0.7954244222900939,
      "grad_norm": 0.7217091917991638,
      "learning_rate": 1.0579808057419105e-06,
      "loss": 0.6774,
      "step": 27606
    },
    {
      "epoch": 0.7954532357517432,
      "grad_norm": 0.728495717048645,
      "learning_rate": 1.0576937839478973e-06,
      "loss": 0.6958,
      "step": 27607
    },
    {
      "epoch": 0.7954820492133925,
      "grad_norm": 0.7105664610862732,
      "learning_rate": 1.05740679648743e-06,
      "loss": 0.6534,
      "step": 27608
    },
    {
      "epoch": 0.7955108626750418,
      "grad_norm": 0.7411982417106628,
      "learning_rate": 1.0571198433630103e-06,
      "loss": 0.6799,
      "step": 27609
    },
    {
      "epoch": 0.7955396761366911,
      "grad_norm": 0.727266788482666,
      "learning_rate": 1.0568329245771335e-06,
      "loss": 0.6558,
      "step": 27610
    },
    {
      "epoch": 0.7955684895983404,
      "grad_norm": 0.7366107702255249,
      "learning_rate": 1.0565460401323003e-06,
      "loss": 0.6829,
      "step": 27611
    },
    {
      "epoch": 0.7955973030599897,
      "grad_norm": 0.709147036075592,
      "learning_rate": 1.0562591900310094e-06,
      "loss": 0.678,
      "step": 27612
    },
    {
      "epoch": 0.795626116521639,
      "grad_norm": 0.715382993221283,
      "learning_rate": 1.05597237427576e-06,
      "loss": 0.6559,
      "step": 27613
    },
    {
      "epoch": 0.7956549299832882,
      "grad_norm": 0.725741982460022,
      "learning_rate": 1.0556855928690469e-06,
      "loss": 0.6823,
      "step": 27614
    },
    {
      "epoch": 0.7956837434449375,
      "grad_norm": 0.7185972929000854,
      "learning_rate": 1.0553988458133706e-06,
      "loss": 0.7048,
      "step": 27615
    },
    {
      "epoch": 0.7957125569065867,
      "grad_norm": 0.7197971343994141,
      "learning_rate": 1.0551121331112264e-06,
      "loss": 0.6786,
      "step": 27616
    },
    {
      "epoch": 0.795741370368236,
      "grad_norm": 0.7780349254608154,
      "learning_rate": 1.0548254547651133e-06,
      "loss": 0.6983,
      "step": 27617
    },
    {
      "epoch": 0.7957701838298853,
      "grad_norm": 0.7210052609443665,
      "learning_rate": 1.054538810777525e-06,
      "loss": 0.6692,
      "step": 27618
    },
    {
      "epoch": 0.7957989972915346,
      "grad_norm": 0.7239883542060852,
      "learning_rate": 1.0542522011509615e-06,
      "loss": 0.6965,
      "step": 27619
    },
    {
      "epoch": 0.7958278107531839,
      "grad_norm": 0.7141767144203186,
      "learning_rate": 1.0539656258879155e-06,
      "loss": 0.6744,
      "step": 27620
    },
    {
      "epoch": 0.7958566242148332,
      "grad_norm": 0.7426259517669678,
      "learning_rate": 1.0536790849908857e-06,
      "loss": 0.6838,
      "step": 27621
    },
    {
      "epoch": 0.7958854376764825,
      "grad_norm": 0.7070559859275818,
      "learning_rate": 1.0533925784623645e-06,
      "loss": 0.6817,
      "step": 27622
    },
    {
      "epoch": 0.7959142511381317,
      "grad_norm": 0.7205129861831665,
      "learning_rate": 1.0531061063048493e-06,
      "loss": 0.7145,
      "step": 27623
    },
    {
      "epoch": 0.795943064599781,
      "grad_norm": 0.7197294235229492,
      "learning_rate": 1.052819668520834e-06,
      "loss": 0.6916,
      "step": 27624
    },
    {
      "epoch": 0.7959718780614303,
      "grad_norm": 0.7379843592643738,
      "learning_rate": 1.052533265112815e-06,
      "loss": 0.6931,
      "step": 27625
    },
    {
      "epoch": 0.7960006915230796,
      "grad_norm": 0.7224552035331726,
      "learning_rate": 1.0522468960832844e-06,
      "loss": 0.672,
      "step": 27626
    },
    {
      "epoch": 0.7960295049847289,
      "grad_norm": 0.706791341304779,
      "learning_rate": 1.0519605614347378e-06,
      "loss": 0.6769,
      "step": 27627
    },
    {
      "epoch": 0.7960583184463782,
      "grad_norm": 0.7104900479316711,
      "learning_rate": 1.0516742611696674e-06,
      "loss": 0.6666,
      "step": 27628
    },
    {
      "epoch": 0.7960871319080274,
      "grad_norm": 0.739088773727417,
      "learning_rate": 1.051387995290567e-06,
      "loss": 0.7103,
      "step": 27629
    },
    {
      "epoch": 0.7961159453696767,
      "grad_norm": 0.7397668957710266,
      "learning_rate": 1.051101763799931e-06,
      "loss": 0.7053,
      "step": 27630
    },
    {
      "epoch": 0.796144758831326,
      "grad_norm": 0.7264424562454224,
      "learning_rate": 1.05081556670025e-06,
      "loss": 0.6987,
      "step": 27631
    },
    {
      "epoch": 0.7961735722929753,
      "grad_norm": 0.7092482447624207,
      "learning_rate": 1.0505294039940194e-06,
      "loss": 0.6602,
      "step": 27632
    },
    {
      "epoch": 0.7962023857546245,
      "grad_norm": 0.7195245027542114,
      "learning_rate": 1.0502432756837283e-06,
      "loss": 0.6838,
      "step": 27633
    },
    {
      "epoch": 0.7962311992162738,
      "grad_norm": 0.7291868329048157,
      "learning_rate": 1.0499571817718707e-06,
      "loss": 0.6959,
      "step": 27634
    },
    {
      "epoch": 0.7962600126779231,
      "grad_norm": 0.7821921110153198,
      "learning_rate": 1.0496711222609352e-06,
      "loss": 0.6438,
      "step": 27635
    },
    {
      "epoch": 0.7962888261395724,
      "grad_norm": 0.7388720512390137,
      "learning_rate": 1.049385097153418e-06,
      "loss": 0.6828,
      "step": 27636
    },
    {
      "epoch": 0.7963176396012217,
      "grad_norm": 0.7090144157409668,
      "learning_rate": 1.0490991064518059e-06,
      "loss": 0.698,
      "step": 27637
    },
    {
      "epoch": 0.796346453062871,
      "grad_norm": 0.7173722982406616,
      "learning_rate": 1.048813150158593e-06,
      "loss": 0.6834,
      "step": 27638
    },
    {
      "epoch": 0.7963752665245203,
      "grad_norm": 0.7423394918441772,
      "learning_rate": 1.048527228276266e-06,
      "loss": 0.6951,
      "step": 27639
    },
    {
      "epoch": 0.7964040799861696,
      "grad_norm": 0.7121502757072449,
      "learning_rate": 1.0482413408073183e-06,
      "loss": 0.6743,
      "step": 27640
    },
    {
      "epoch": 0.7964328934478189,
      "grad_norm": 0.7060762643814087,
      "learning_rate": 1.0479554877542364e-06,
      "loss": 0.6659,
      "step": 27641
    },
    {
      "epoch": 0.7964617069094682,
      "grad_norm": 0.7278268337249756,
      "learning_rate": 1.0476696691195137e-06,
      "loss": 0.6782,
      "step": 27642
    },
    {
      "epoch": 0.7964905203711173,
      "grad_norm": 0.7259411215782166,
      "learning_rate": 1.0473838849056357e-06,
      "loss": 0.6541,
      "step": 27643
    },
    {
      "epoch": 0.7965193338327666,
      "grad_norm": 0.7184861302375793,
      "learning_rate": 1.0470981351150938e-06,
      "loss": 0.708,
      "step": 27644
    },
    {
      "epoch": 0.7965481472944159,
      "grad_norm": 0.7343083620071411,
      "learning_rate": 1.0468124197503748e-06,
      "loss": 0.6825,
      "step": 27645
    },
    {
      "epoch": 0.7965769607560652,
      "grad_norm": 0.7149999737739563,
      "learning_rate": 1.0465267388139678e-06,
      "loss": 0.6816,
      "step": 27646
    },
    {
      "epoch": 0.7966057742177145,
      "grad_norm": 0.7261594533920288,
      "learning_rate": 1.046241092308361e-06,
      "loss": 0.6811,
      "step": 27647
    },
    {
      "epoch": 0.7966345876793638,
      "grad_norm": 0.7347391843795776,
      "learning_rate": 1.0459554802360434e-06,
      "loss": 0.6905,
      "step": 27648
    },
    {
      "epoch": 0.7966634011410131,
      "grad_norm": 0.716790497303009,
      "learning_rate": 1.0456699025994993e-06,
      "loss": 0.6904,
      "step": 27649
    },
    {
      "epoch": 0.7966922146026624,
      "grad_norm": 0.6941149234771729,
      "learning_rate": 1.0453843594012176e-06,
      "loss": 0.6875,
      "step": 27650
    },
    {
      "epoch": 0.7967210280643117,
      "grad_norm": 0.7391681671142578,
      "learning_rate": 1.045098850643686e-06,
      "loss": 0.6803,
      "step": 27651
    },
    {
      "epoch": 0.796749841525961,
      "grad_norm": 0.7188798785209656,
      "learning_rate": 1.0448133763293888e-06,
      "loss": 0.6653,
      "step": 27652
    },
    {
      "epoch": 0.7967786549876102,
      "grad_norm": 0.6908884048461914,
      "learning_rate": 1.0445279364608152e-06,
      "loss": 0.6803,
      "step": 27653
    },
    {
      "epoch": 0.7968074684492595,
      "grad_norm": 0.740551233291626,
      "learning_rate": 1.044242531040447e-06,
      "loss": 0.6843,
      "step": 27654
    },
    {
      "epoch": 0.7968362819109088,
      "grad_norm": 0.7080620527267456,
      "learning_rate": 1.0439571600707744e-06,
      "loss": 0.6623,
      "step": 27655
    },
    {
      "epoch": 0.796865095372558,
      "grad_norm": 0.6953225135803223,
      "learning_rate": 1.043671823554278e-06,
      "loss": 0.6479,
      "step": 27656
    },
    {
      "epoch": 0.7968939088342073,
      "grad_norm": 0.7184370160102844,
      "learning_rate": 1.0433865214934468e-06,
      "loss": 0.6889,
      "step": 27657
    },
    {
      "epoch": 0.7969227222958566,
      "grad_norm": 0.7147009968757629,
      "learning_rate": 1.0431012538907631e-06,
      "loss": 0.709,
      "step": 27658
    },
    {
      "epoch": 0.7969515357575059,
      "grad_norm": 0.73504239320755,
      "learning_rate": 1.0428160207487116e-06,
      "loss": 0.7112,
      "step": 27659
    },
    {
      "epoch": 0.7969803492191552,
      "grad_norm": 0.7036213278770447,
      "learning_rate": 1.0425308220697766e-06,
      "loss": 0.6756,
      "step": 27660
    },
    {
      "epoch": 0.7970091626808045,
      "grad_norm": 0.7310253381729126,
      "learning_rate": 1.0422456578564438e-06,
      "loss": 0.6819,
      "step": 27661
    },
    {
      "epoch": 0.7970379761424538,
      "grad_norm": 0.7192420363426208,
      "learning_rate": 1.0419605281111933e-06,
      "loss": 0.6679,
      "step": 27662
    },
    {
      "epoch": 0.797066789604103,
      "grad_norm": 0.7025421857833862,
      "learning_rate": 1.0416754328365119e-06,
      "loss": 0.6585,
      "step": 27663
    },
    {
      "epoch": 0.7970956030657523,
      "grad_norm": 0.6954725980758667,
      "learning_rate": 1.0413903720348788e-06,
      "loss": 0.6444,
      "step": 27664
    },
    {
      "epoch": 0.7971244165274016,
      "grad_norm": 0.712235689163208,
      "learning_rate": 1.04110534570878e-06,
      "loss": 0.6873,
      "step": 27665
    },
    {
      "epoch": 0.7971532299890509,
      "grad_norm": 0.7490169405937195,
      "learning_rate": 1.0408203538606948e-06,
      "loss": 0.7082,
      "step": 27666
    },
    {
      "epoch": 0.7971820434507002,
      "grad_norm": 0.7113593220710754,
      "learning_rate": 1.0405353964931065e-06,
      "loss": 0.6776,
      "step": 27667
    },
    {
      "epoch": 0.7972108569123495,
      "grad_norm": 0.7262240052223206,
      "learning_rate": 1.040250473608498e-06,
      "loss": 0.7093,
      "step": 27668
    },
    {
      "epoch": 0.7972396703739987,
      "grad_norm": 0.7375838756561279,
      "learning_rate": 1.0399655852093482e-06,
      "loss": 0.6648,
      "step": 27669
    },
    {
      "epoch": 0.797268483835648,
      "grad_norm": 0.7231490612030029,
      "learning_rate": 1.0396807312981417e-06,
      "loss": 0.6868,
      "step": 27670
    },
    {
      "epoch": 0.7972972972972973,
      "grad_norm": 0.7151505947113037,
      "learning_rate": 1.0393959118773538e-06,
      "loss": 0.6957,
      "step": 27671
    },
    {
      "epoch": 0.7973261107589465,
      "grad_norm": 0.751983106136322,
      "learning_rate": 1.0391111269494713e-06,
      "loss": 0.7014,
      "step": 27672
    },
    {
      "epoch": 0.7973549242205958,
      "grad_norm": 0.7128461003303528,
      "learning_rate": 1.03882637651697e-06,
      "loss": 0.6702,
      "step": 27673
    },
    {
      "epoch": 0.7973837376822451,
      "grad_norm": 0.7239425182342529,
      "learning_rate": 1.0385416605823322e-06,
      "loss": 0.6855,
      "step": 27674
    },
    {
      "epoch": 0.7974125511438944,
      "grad_norm": 0.7738265991210938,
      "learning_rate": 1.0382569791480357e-06,
      "loss": 0.6886,
      "step": 27675
    },
    {
      "epoch": 0.7974413646055437,
      "grad_norm": 0.7212873101234436,
      "learning_rate": 1.0379723322165618e-06,
      "loss": 0.6802,
      "step": 27676
    },
    {
      "epoch": 0.797470178067193,
      "grad_norm": 0.7130926847457886,
      "learning_rate": 1.0376877197903873e-06,
      "loss": 0.6707,
      "step": 27677
    },
    {
      "epoch": 0.7974989915288423,
      "grad_norm": 0.7209067344665527,
      "learning_rate": 1.0374031418719925e-06,
      "loss": 0.6761,
      "step": 27678
    },
    {
      "epoch": 0.7975278049904916,
      "grad_norm": 0.7126108407974243,
      "learning_rate": 1.0371185984638544e-06,
      "loss": 0.6934,
      "step": 27679
    },
    {
      "epoch": 0.7975566184521409,
      "grad_norm": 0.7257248163223267,
      "learning_rate": 1.0368340895684532e-06,
      "loss": 0.6698,
      "step": 27680
    },
    {
      "epoch": 0.7975854319137902,
      "grad_norm": 0.7159350514411926,
      "learning_rate": 1.0365496151882637e-06,
      "loss": 0.6756,
      "step": 27681
    },
    {
      "epoch": 0.7976142453754395,
      "grad_norm": 0.7268869876861572,
      "learning_rate": 1.0362651753257669e-06,
      "loss": 0.6766,
      "step": 27682
    },
    {
      "epoch": 0.7976430588370886,
      "grad_norm": 0.7227224707603455,
      "learning_rate": 1.035980769983435e-06,
      "loss": 0.6695,
      "step": 27683
    },
    {
      "epoch": 0.7976718722987379,
      "grad_norm": 0.6991518139839172,
      "learning_rate": 1.0356963991637508e-06,
      "loss": 0.6772,
      "step": 27684
    },
    {
      "epoch": 0.7977006857603872,
      "grad_norm": 0.7224861979484558,
      "learning_rate": 1.035412062869187e-06,
      "loss": 0.6934,
      "step": 27685
    },
    {
      "epoch": 0.7977294992220365,
      "grad_norm": 0.7352378964424133,
      "learning_rate": 1.0351277611022221e-06,
      "loss": 0.7096,
      "step": 27686
    },
    {
      "epoch": 0.7977583126836858,
      "grad_norm": 0.7266072034835815,
      "learning_rate": 1.0348434938653296e-06,
      "loss": 0.69,
      "step": 27687
    },
    {
      "epoch": 0.7977871261453351,
      "grad_norm": 0.7385767102241516,
      "learning_rate": 1.0345592611609868e-06,
      "loss": 0.691,
      "step": 27688
    },
    {
      "epoch": 0.7978159396069844,
      "grad_norm": 0.7265457510948181,
      "learning_rate": 1.0342750629916697e-06,
      "loss": 0.6999,
      "step": 27689
    },
    {
      "epoch": 0.7978447530686337,
      "grad_norm": 0.7087198495864868,
      "learning_rate": 1.0339908993598519e-06,
      "loss": 0.6665,
      "step": 27690
    },
    {
      "epoch": 0.797873566530283,
      "grad_norm": 0.7365615367889404,
      "learning_rate": 1.0337067702680098e-06,
      "loss": 0.6814,
      "step": 27691
    },
    {
      "epoch": 0.7979023799919323,
      "grad_norm": 0.6995519995689392,
      "learning_rate": 1.0334226757186155e-06,
      "loss": 0.6581,
      "step": 27692
    },
    {
      "epoch": 0.7979311934535815,
      "grad_norm": 0.7321345806121826,
      "learning_rate": 1.033138615714146e-06,
      "loss": 0.7112,
      "step": 27693
    },
    {
      "epoch": 0.7979600069152308,
      "grad_norm": 0.6970325708389282,
      "learning_rate": 1.0328545902570725e-06,
      "loss": 0.6641,
      "step": 27694
    },
    {
      "epoch": 0.7979888203768801,
      "grad_norm": 0.7117795944213867,
      "learning_rate": 1.0325705993498698e-06,
      "loss": 0.6731,
      "step": 27695
    },
    {
      "epoch": 0.7980176338385293,
      "grad_norm": 0.7136856913566589,
      "learning_rate": 1.0322866429950108e-06,
      "loss": 0.6924,
      "step": 27696
    },
    {
      "epoch": 0.7980464473001786,
      "grad_norm": 0.7215545773506165,
      "learning_rate": 1.0320027211949707e-06,
      "loss": 0.6751,
      "step": 27697
    },
    {
      "epoch": 0.7980752607618279,
      "grad_norm": 0.7382014393806458,
      "learning_rate": 1.0317188339522188e-06,
      "loss": 0.6821,
      "step": 27698
    },
    {
      "epoch": 0.7981040742234772,
      "grad_norm": 0.7201123237609863,
      "learning_rate": 1.0314349812692304e-06,
      "loss": 0.7156,
      "step": 27699
    },
    {
      "epoch": 0.7981328876851265,
      "grad_norm": 0.707170307636261,
      "learning_rate": 1.0311511631484749e-06,
      "loss": 0.6872,
      "step": 27700
    },
    {
      "epoch": 0.7981617011467758,
      "grad_norm": 0.737184464931488,
      "learning_rate": 1.0308673795924262e-06,
      "loss": 0.7086,
      "step": 27701
    },
    {
      "epoch": 0.798190514608425,
      "grad_norm": 0.6975066065788269,
      "learning_rate": 1.0305836306035544e-06,
      "loss": 0.7078,
      "step": 27702
    },
    {
      "epoch": 0.7982193280700743,
      "grad_norm": 0.7071568369865417,
      "learning_rate": 1.0302999161843324e-06,
      "loss": 0.6701,
      "step": 27703
    },
    {
      "epoch": 0.7982481415317236,
      "grad_norm": 0.718130350112915,
      "learning_rate": 1.0300162363372279e-06,
      "loss": 0.6724,
      "step": 27704
    },
    {
      "epoch": 0.7982769549933729,
      "grad_norm": 0.717210590839386,
      "learning_rate": 1.0297325910647144e-06,
      "loss": 0.6989,
      "step": 27705
    },
    {
      "epoch": 0.7983057684550222,
      "grad_norm": 0.7343031764030457,
      "learning_rate": 1.029448980369262e-06,
      "loss": 0.6746,
      "step": 27706
    },
    {
      "epoch": 0.7983345819166715,
      "grad_norm": 0.7204026579856873,
      "learning_rate": 1.0291654042533383e-06,
      "loss": 0.6936,
      "step": 27707
    },
    {
      "epoch": 0.7983633953783208,
      "grad_norm": 0.7130686640739441,
      "learning_rate": 1.0288818627194148e-06,
      "loss": 0.6852,
      "step": 27708
    },
    {
      "epoch": 0.7983922088399701,
      "grad_norm": 0.7500728964805603,
      "learning_rate": 1.0285983557699608e-06,
      "loss": 0.7067,
      "step": 27709
    },
    {
      "epoch": 0.7984210223016193,
      "grad_norm": 0.7125537395477295,
      "learning_rate": 1.0283148834074464e-06,
      "loss": 0.6885,
      "step": 27710
    },
    {
      "epoch": 0.7984498357632686,
      "grad_norm": 0.73749840259552,
      "learning_rate": 1.0280314456343377e-06,
      "loss": 0.6878,
      "step": 27711
    },
    {
      "epoch": 0.7984786492249178,
      "grad_norm": 0.7141931056976318,
      "learning_rate": 1.0277480424531062e-06,
      "loss": 0.6903,
      "step": 27712
    },
    {
      "epoch": 0.7985074626865671,
      "grad_norm": 0.7049961686134338,
      "learning_rate": 1.027464673866217e-06,
      "loss": 0.6722,
      "step": 27713
    },
    {
      "epoch": 0.7985362761482164,
      "grad_norm": 0.7104451656341553,
      "learning_rate": 1.0271813398761404e-06,
      "loss": 0.7208,
      "step": 27714
    },
    {
      "epoch": 0.7985650896098657,
      "grad_norm": 0.7260799407958984,
      "learning_rate": 1.0268980404853418e-06,
      "loss": 0.6905,
      "step": 27715
    },
    {
      "epoch": 0.798593903071515,
      "grad_norm": 0.7425031065940857,
      "learning_rate": 1.0266147756962913e-06,
      "loss": 0.6891,
      "step": 27716
    },
    {
      "epoch": 0.7986227165331643,
      "grad_norm": 0.7358344793319702,
      "learning_rate": 1.0263315455114525e-06,
      "loss": 0.686,
      "step": 27717
    },
    {
      "epoch": 0.7986515299948136,
      "grad_norm": 0.7382036447525024,
      "learning_rate": 1.0260483499332952e-06,
      "loss": 0.6915,
      "step": 27718
    },
    {
      "epoch": 0.7986803434564629,
      "grad_norm": 0.7056440711021423,
      "learning_rate": 1.0257651889642817e-06,
      "loss": 0.6776,
      "step": 27719
    },
    {
      "epoch": 0.7987091569181122,
      "grad_norm": 0.7466558218002319,
      "learning_rate": 1.025482062606883e-06,
      "loss": 0.6808,
      "step": 27720
    },
    {
      "epoch": 0.7987379703797615,
      "grad_norm": 0.741893470287323,
      "learning_rate": 1.0251989708635613e-06,
      "loss": 0.7056,
      "step": 27721
    },
    {
      "epoch": 0.7987667838414108,
      "grad_norm": 0.7166724801063538,
      "learning_rate": 1.0249159137367841e-06,
      "loss": 0.6717,
      "step": 27722
    },
    {
      "epoch": 0.7987955973030599,
      "grad_norm": 0.7188029885292053,
      "learning_rate": 1.024632891229015e-06,
      "loss": 0.6709,
      "step": 27723
    },
    {
      "epoch": 0.7988244107647092,
      "grad_norm": 0.7546880841255188,
      "learning_rate": 1.0243499033427207e-06,
      "loss": 0.6865,
      "step": 27724
    },
    {
      "epoch": 0.7988532242263585,
      "grad_norm": 0.7073710560798645,
      "learning_rate": 1.024066950080363e-06,
      "loss": 0.6911,
      "step": 27725
    },
    {
      "epoch": 0.7988820376880078,
      "grad_norm": 0.7352192401885986,
      "learning_rate": 1.0237840314444082e-06,
      "loss": 0.7148,
      "step": 27726
    },
    {
      "epoch": 0.7989108511496571,
      "grad_norm": 0.7277381420135498,
      "learning_rate": 1.0235011474373202e-06,
      "loss": 0.668,
      "step": 27727
    },
    {
      "epoch": 0.7989396646113064,
      "grad_norm": 0.7377383708953857,
      "learning_rate": 1.0232182980615612e-06,
      "loss": 0.6797,
      "step": 27728
    },
    {
      "epoch": 0.7989684780729557,
      "grad_norm": 0.690666139125824,
      "learning_rate": 1.022935483319597e-06,
      "loss": 0.6538,
      "step": 27729
    },
    {
      "epoch": 0.798997291534605,
      "grad_norm": 0.6962644457817078,
      "learning_rate": 1.0226527032138877e-06,
      "loss": 0.6503,
      "step": 27730
    },
    {
      "epoch": 0.7990261049962543,
      "grad_norm": 0.7282092571258545,
      "learning_rate": 1.0223699577468977e-06,
      "loss": 0.6807,
      "step": 27731
    },
    {
      "epoch": 0.7990549184579036,
      "grad_norm": 0.7183127999305725,
      "learning_rate": 1.022087246921089e-06,
      "loss": 0.6873,
      "step": 27732
    },
    {
      "epoch": 0.7990837319195528,
      "grad_norm": 0.7442107200622559,
      "learning_rate": 1.0218045707389252e-06,
      "loss": 0.6886,
      "step": 27733
    },
    {
      "epoch": 0.7991125453812021,
      "grad_norm": 0.7137820720672607,
      "learning_rate": 1.0215219292028656e-06,
      "loss": 0.6909,
      "step": 27734
    },
    {
      "epoch": 0.7991413588428514,
      "grad_norm": 0.7022490501403809,
      "learning_rate": 1.0212393223153743e-06,
      "loss": 0.6815,
      "step": 27735
    },
    {
      "epoch": 0.7991701723045007,
      "grad_norm": 0.7195515036582947,
      "learning_rate": 1.0209567500789097e-06,
      "loss": 0.679,
      "step": 27736
    },
    {
      "epoch": 0.7991989857661499,
      "grad_norm": 0.7265139818191528,
      "learning_rate": 1.0206742124959358e-06,
      "loss": 0.6892,
      "step": 27737
    },
    {
      "epoch": 0.7992277992277992,
      "grad_norm": 0.725918710231781,
      "learning_rate": 1.0203917095689098e-06,
      "loss": 0.6773,
      "step": 27738
    },
    {
      "epoch": 0.7992566126894485,
      "grad_norm": 0.7170417904853821,
      "learning_rate": 1.0201092413002956e-06,
      "loss": 0.6744,
      "step": 27739
    },
    {
      "epoch": 0.7992854261510978,
      "grad_norm": 0.7503207325935364,
      "learning_rate": 1.0198268076925506e-06,
      "loss": 0.7095,
      "step": 27740
    },
    {
      "epoch": 0.7993142396127471,
      "grad_norm": 0.7460154891014099,
      "learning_rate": 1.019544408748136e-06,
      "loss": 0.7053,
      "step": 27741
    },
    {
      "epoch": 0.7993430530743963,
      "grad_norm": 0.759253740310669,
      "learning_rate": 1.0192620444695094e-06,
      "loss": 0.6863,
      "step": 27742
    },
    {
      "epoch": 0.7993718665360456,
      "grad_norm": 0.726450502872467,
      "learning_rate": 1.0189797148591313e-06,
      "loss": 0.6915,
      "step": 27743
    },
    {
      "epoch": 0.7994006799976949,
      "grad_norm": 0.7334973216056824,
      "learning_rate": 1.01869741991946e-06,
      "loss": 0.6576,
      "step": 27744
    },
    {
      "epoch": 0.7994294934593442,
      "grad_norm": 0.72157883644104,
      "learning_rate": 1.0184151596529562e-06,
      "loss": 0.6532,
      "step": 27745
    },
    {
      "epoch": 0.7994583069209935,
      "grad_norm": 0.7168774008750916,
      "learning_rate": 1.0181329340620745e-06,
      "loss": 0.6669,
      "step": 27746
    },
    {
      "epoch": 0.7994871203826428,
      "grad_norm": 0.7320044636726379,
      "learning_rate": 1.0178507431492745e-06,
      "loss": 0.6709,
      "step": 27747
    },
    {
      "epoch": 0.7995159338442921,
      "grad_norm": 0.7575556635856628,
      "learning_rate": 1.0175685869170154e-06,
      "loss": 0.6972,
      "step": 27748
    },
    {
      "epoch": 0.7995447473059414,
      "grad_norm": 0.7319959402084351,
      "learning_rate": 1.0172864653677511e-06,
      "loss": 0.7109,
      "step": 27749
    },
    {
      "epoch": 0.7995735607675906,
      "grad_norm": 0.7107036113739014,
      "learning_rate": 1.0170043785039424e-06,
      "loss": 0.7022,
      "step": 27750
    },
    {
      "epoch": 0.7996023742292399,
      "grad_norm": 0.7397536039352417,
      "learning_rate": 1.0167223263280423e-06,
      "loss": 0.6878,
      "step": 27751
    },
    {
      "epoch": 0.7996311876908891,
      "grad_norm": 0.7340419292449951,
      "learning_rate": 1.0164403088425107e-06,
      "loss": 0.6955,
      "step": 27752
    },
    {
      "epoch": 0.7996600011525384,
      "grad_norm": 0.7158066034317017,
      "learning_rate": 1.0161583260498004e-06,
      "loss": 0.6758,
      "step": 27753
    },
    {
      "epoch": 0.7996888146141877,
      "grad_norm": 0.7067080140113831,
      "learning_rate": 1.0158763779523694e-06,
      "loss": 0.6686,
      "step": 27754
    },
    {
      "epoch": 0.799717628075837,
      "grad_norm": 0.7306129336357117,
      "learning_rate": 1.0155944645526705e-06,
      "loss": 0.671,
      "step": 27755
    },
    {
      "epoch": 0.7997464415374863,
      "grad_norm": 0.7150219678878784,
      "learning_rate": 1.0153125858531632e-06,
      "loss": 0.669,
      "step": 27756
    },
    {
      "epoch": 0.7997752549991356,
      "grad_norm": 0.7166781425476074,
      "learning_rate": 1.015030741856299e-06,
      "loss": 0.6743,
      "step": 27757
    },
    {
      "epoch": 0.7998040684607849,
      "grad_norm": 0.7207764983177185,
      "learning_rate": 1.0147489325645348e-06,
      "loss": 0.7272,
      "step": 27758
    },
    {
      "epoch": 0.7998328819224342,
      "grad_norm": 0.7051752209663391,
      "learning_rate": 1.0144671579803222e-06,
      "loss": 0.6761,
      "step": 27759
    },
    {
      "epoch": 0.7998616953840835,
      "grad_norm": 0.7445788979530334,
      "learning_rate": 1.0141854181061178e-06,
      "loss": 0.7113,
      "step": 27760
    },
    {
      "epoch": 0.7998905088457328,
      "grad_norm": 0.7507584691047668,
      "learning_rate": 1.013903712944373e-06,
      "loss": 0.7113,
      "step": 27761
    },
    {
      "epoch": 0.799919322307382,
      "grad_norm": 0.710720419883728,
      "learning_rate": 1.0136220424975434e-06,
      "loss": 0.6878,
      "step": 27762
    },
    {
      "epoch": 0.7999481357690313,
      "grad_norm": 0.7240850329399109,
      "learning_rate": 1.013340406768079e-06,
      "loss": 0.7098,
      "step": 27763
    },
    {
      "epoch": 0.7999769492306805,
      "grad_norm": 0.745252788066864,
      "learning_rate": 1.013058805758435e-06,
      "loss": 0.6851,
      "step": 27764
    },
    {
      "epoch": 0.8000057626923298,
      "grad_norm": 0.7081648707389832,
      "learning_rate": 1.012777239471065e-06,
      "loss": 0.6469,
      "step": 27765
    },
    {
      "epoch": 0.8000345761539791,
      "grad_norm": 0.7135932445526123,
      "learning_rate": 1.012495707908418e-06,
      "loss": 0.6754,
      "step": 27766
    },
    {
      "epoch": 0.8000633896156284,
      "grad_norm": 0.7296233773231506,
      "learning_rate": 1.0122142110729471e-06,
      "loss": 0.6772,
      "step": 27767
    },
    {
      "epoch": 0.8000922030772777,
      "grad_norm": 0.7346250414848328,
      "learning_rate": 1.0119327489671039e-06,
      "loss": 0.676,
      "step": 27768
    },
    {
      "epoch": 0.800121016538927,
      "grad_norm": 0.7275258302688599,
      "learning_rate": 1.0116513215933416e-06,
      "loss": 0.6791,
      "step": 27769
    },
    {
      "epoch": 0.8001498300005763,
      "grad_norm": 0.6951208114624023,
      "learning_rate": 1.011369928954108e-06,
      "loss": 0.6511,
      "step": 27770
    },
    {
      "epoch": 0.8001786434622256,
      "grad_norm": 0.7308898568153381,
      "learning_rate": 1.0110885710518559e-06,
      "loss": 0.6889,
      "step": 27771
    },
    {
      "epoch": 0.8002074569238748,
      "grad_norm": 0.7374890446662903,
      "learning_rate": 1.010807247889034e-06,
      "loss": 0.6989,
      "step": 27772
    },
    {
      "epoch": 0.8002362703855241,
      "grad_norm": 0.7348030209541321,
      "learning_rate": 1.010525959468095e-06,
      "loss": 0.688,
      "step": 27773
    },
    {
      "epoch": 0.8002650838471734,
      "grad_norm": 0.7189610004425049,
      "learning_rate": 1.0102447057914849e-06,
      "loss": 0.6909,
      "step": 27774
    },
    {
      "epoch": 0.8002938973088227,
      "grad_norm": 0.7195637226104736,
      "learning_rate": 1.0099634868616565e-06,
      "loss": 0.6749,
      "step": 27775
    },
    {
      "epoch": 0.800322710770472,
      "grad_norm": 0.7153993844985962,
      "learning_rate": 1.0096823026810564e-06,
      "loss": 0.677,
      "step": 27776
    },
    {
      "epoch": 0.8003515242321212,
      "grad_norm": 0.7191781401634216,
      "learning_rate": 1.0094011532521357e-06,
      "loss": 0.6782,
      "step": 27777
    },
    {
      "epoch": 0.8003803376937705,
      "grad_norm": 0.7287774085998535,
      "learning_rate": 1.0091200385773409e-06,
      "loss": 0.6944,
      "step": 27778
    },
    {
      "epoch": 0.8004091511554198,
      "grad_norm": 0.7311477661132812,
      "learning_rate": 1.0088389586591213e-06,
      "loss": 0.6944,
      "step": 27779
    },
    {
      "epoch": 0.8004379646170691,
      "grad_norm": 0.7166070342063904,
      "learning_rate": 1.008557913499924e-06,
      "loss": 0.6832,
      "step": 27780
    },
    {
      "epoch": 0.8004667780787184,
      "grad_norm": 0.7051171660423279,
      "learning_rate": 1.0082769031021989e-06,
      "loss": 0.6688,
      "step": 27781
    },
    {
      "epoch": 0.8004955915403676,
      "grad_norm": 0.7127143740653992,
      "learning_rate": 1.0079959274683904e-06,
      "loss": 0.677,
      "step": 27782
    },
    {
      "epoch": 0.8005244050020169,
      "grad_norm": 0.7191635370254517,
      "learning_rate": 1.0077149866009483e-06,
      "loss": 0.6817,
      "step": 27783
    },
    {
      "epoch": 0.8005532184636662,
      "grad_norm": 0.7184717655181885,
      "learning_rate": 1.0074340805023164e-06,
      "loss": 0.701,
      "step": 27784
    },
    {
      "epoch": 0.8005820319253155,
      "grad_norm": 0.7222632765769958,
      "learning_rate": 1.0071532091749424e-06,
      "loss": 0.6944,
      "step": 27785
    },
    {
      "epoch": 0.8006108453869648,
      "grad_norm": 0.7146047949790955,
      "learning_rate": 1.0068723726212743e-06,
      "loss": 0.6839,
      "step": 27786
    },
    {
      "epoch": 0.8006396588486141,
      "grad_norm": 0.7292011976242065,
      "learning_rate": 1.0065915708437547e-06,
      "loss": 0.6709,
      "step": 27787
    },
    {
      "epoch": 0.8006684723102634,
      "grad_norm": 0.7082343697547913,
      "learning_rate": 1.0063108038448316e-06,
      "loss": 0.6814,
      "step": 27788
    },
    {
      "epoch": 0.8006972857719127,
      "grad_norm": 0.7015000581741333,
      "learning_rate": 1.0060300716269479e-06,
      "loss": 0.6775,
      "step": 27789
    },
    {
      "epoch": 0.800726099233562,
      "grad_norm": 0.735299825668335,
      "learning_rate": 1.005749374192551e-06,
      "loss": 0.6819,
      "step": 27790
    },
    {
      "epoch": 0.8007549126952112,
      "grad_norm": 0.7287318706512451,
      "learning_rate": 1.005468711544082e-06,
      "loss": 0.7084,
      "step": 27791
    },
    {
      "epoch": 0.8007837261568604,
      "grad_norm": 0.731971800327301,
      "learning_rate": 1.0051880836839901e-06,
      "loss": 0.7103,
      "step": 27792
    },
    {
      "epoch": 0.8008125396185097,
      "grad_norm": 0.718420147895813,
      "learning_rate": 1.0049074906147149e-06,
      "loss": 0.6675,
      "step": 27793
    },
    {
      "epoch": 0.800841353080159,
      "grad_norm": 0.7258135080337524,
      "learning_rate": 1.0046269323387037e-06,
      "loss": 0.6894,
      "step": 27794
    },
    {
      "epoch": 0.8008701665418083,
      "grad_norm": 0.7311575412750244,
      "learning_rate": 1.0043464088583966e-06,
      "loss": 0.7033,
      "step": 27795
    },
    {
      "epoch": 0.8008989800034576,
      "grad_norm": 0.7211366891860962,
      "learning_rate": 1.0040659201762386e-06,
      "loss": 0.6715,
      "step": 27796
    },
    {
      "epoch": 0.8009277934651069,
      "grad_norm": 0.7377628087997437,
      "learning_rate": 1.0037854662946716e-06,
      "loss": 0.6866,
      "step": 27797
    },
    {
      "epoch": 0.8009566069267562,
      "grad_norm": 0.6951575875282288,
      "learning_rate": 1.0035050472161395e-06,
      "loss": 0.6875,
      "step": 27798
    },
    {
      "epoch": 0.8009854203884055,
      "grad_norm": 0.7393724322319031,
      "learning_rate": 1.003224662943082e-06,
      "loss": 0.6889,
      "step": 27799
    },
    {
      "epoch": 0.8010142338500548,
      "grad_norm": 0.7387719750404358,
      "learning_rate": 1.002944313477943e-06,
      "loss": 0.7052,
      "step": 27800
    },
    {
      "epoch": 0.8010430473117041,
      "grad_norm": 0.7236261963844299,
      "learning_rate": 1.0026639988231634e-06,
      "loss": 0.6652,
      "step": 27801
    },
    {
      "epoch": 0.8010718607733534,
      "grad_norm": 0.7639651894569397,
      "learning_rate": 1.0023837189811836e-06,
      "loss": 0.7173,
      "step": 27802
    },
    {
      "epoch": 0.8011006742350026,
      "grad_norm": 0.7332925200462341,
      "learning_rate": 1.0021034739544462e-06,
      "loss": 0.6832,
      "step": 27803
    },
    {
      "epoch": 0.8011294876966518,
      "grad_norm": 0.7209461331367493,
      "learning_rate": 1.0018232637453918e-06,
      "loss": 0.6693,
      "step": 27804
    },
    {
      "epoch": 0.8011583011583011,
      "grad_norm": 0.7042028307914734,
      "learning_rate": 1.0015430883564587e-06,
      "loss": 0.6787,
      "step": 27805
    },
    {
      "epoch": 0.8011871146199504,
      "grad_norm": 0.7461622953414917,
      "learning_rate": 1.0012629477900886e-06,
      "loss": 0.698,
      "step": 27806
    },
    {
      "epoch": 0.8012159280815997,
      "grad_norm": 0.7222486138343811,
      "learning_rate": 1.0009828420487222e-06,
      "loss": 0.6801,
      "step": 27807
    },
    {
      "epoch": 0.801244741543249,
      "grad_norm": 0.7100722789764404,
      "learning_rate": 1.000702771134796e-06,
      "loss": 0.6896,
      "step": 27808
    },
    {
      "epoch": 0.8012735550048983,
      "grad_norm": 0.7158104777336121,
      "learning_rate": 1.000422735050753e-06,
      "loss": 0.6774,
      "step": 27809
    },
    {
      "epoch": 0.8013023684665476,
      "grad_norm": 0.7318658232688904,
      "learning_rate": 1.0001427337990276e-06,
      "loss": 0.6786,
      "step": 27810
    },
    {
      "epoch": 0.8013311819281969,
      "grad_norm": 0.7278659343719482,
      "learning_rate": 9.998627673820628e-07,
      "loss": 0.6724,
      "step": 27811
    },
    {
      "epoch": 0.8013599953898461,
      "grad_norm": 0.7065644860267639,
      "learning_rate": 9.995828358022925e-07,
      "loss": 0.6817,
      "step": 27812
    },
    {
      "epoch": 0.8013888088514954,
      "grad_norm": 0.7168865203857422,
      "learning_rate": 9.993029390621583e-07,
      "loss": 0.7038,
      "step": 27813
    },
    {
      "epoch": 0.8014176223131447,
      "grad_norm": 0.7206954956054688,
      "learning_rate": 9.990230771640951e-07,
      "loss": 0.6919,
      "step": 27814
    },
    {
      "epoch": 0.801446435774794,
      "grad_norm": 0.7092364430427551,
      "learning_rate": 9.987432501105414e-07,
      "loss": 0.6772,
      "step": 27815
    },
    {
      "epoch": 0.8014752492364433,
      "grad_norm": 0.7349591255187988,
      "learning_rate": 9.984634579039342e-07,
      "loss": 0.6914,
      "step": 27816
    },
    {
      "epoch": 0.8015040626980926,
      "grad_norm": 0.7041992545127869,
      "learning_rate": 9.981837005467115e-07,
      "loss": 0.6641,
      "step": 27817
    },
    {
      "epoch": 0.8015328761597418,
      "grad_norm": 0.7121021747589111,
      "learning_rate": 9.979039780413069e-07,
      "loss": 0.67,
      "step": 27818
    },
    {
      "epoch": 0.8015616896213911,
      "grad_norm": 0.7183107137680054,
      "learning_rate": 9.976242903901595e-07,
      "loss": 0.6844,
      "step": 27819
    },
    {
      "epoch": 0.8015905030830404,
      "grad_norm": 0.765303373336792,
      "learning_rate": 9.973446375957024e-07,
      "loss": 0.7094,
      "step": 27820
    },
    {
      "epoch": 0.8016193165446897,
      "grad_norm": 0.7430519461631775,
      "learning_rate": 9.970650196603738e-07,
      "loss": 0.7219,
      "step": 27821
    },
    {
      "epoch": 0.8016481300063389,
      "grad_norm": 0.7262625098228455,
      "learning_rate": 9.96785436586606e-07,
      "loss": 0.7001,
      "step": 27822
    },
    {
      "epoch": 0.8016769434679882,
      "grad_norm": 0.7013730406761169,
      "learning_rate": 9.965058883768347e-07,
      "loss": 0.6818,
      "step": 27823
    },
    {
      "epoch": 0.8017057569296375,
      "grad_norm": 0.7228525280952454,
      "learning_rate": 9.962263750334971e-07,
      "loss": 0.6677,
      "step": 27824
    },
    {
      "epoch": 0.8017345703912868,
      "grad_norm": 0.7109221816062927,
      "learning_rate": 9.95946896559024e-07,
      "loss": 0.6641,
      "step": 27825
    },
    {
      "epoch": 0.8017633838529361,
      "grad_norm": 0.7562330961227417,
      "learning_rate": 9.956674529558518e-07,
      "loss": 0.7006,
      "step": 27826
    },
    {
      "epoch": 0.8017921973145854,
      "grad_norm": 0.7111610770225525,
      "learning_rate": 9.953880442264124e-07,
      "loss": 0.6909,
      "step": 27827
    },
    {
      "epoch": 0.8018210107762347,
      "grad_norm": 0.7192673087120056,
      "learning_rate": 9.951086703731393e-07,
      "loss": 0.6712,
      "step": 27828
    },
    {
      "epoch": 0.801849824237884,
      "grad_norm": 0.7357789874076843,
      "learning_rate": 9.94829331398467e-07,
      "loss": 0.7113,
      "step": 27829
    },
    {
      "epoch": 0.8018786376995333,
      "grad_norm": 0.7307108044624329,
      "learning_rate": 9.945500273048286e-07,
      "loss": 0.6977,
      "step": 27830
    },
    {
      "epoch": 0.8019074511611824,
      "grad_norm": 0.7357045412063599,
      "learning_rate": 9.942707580946542e-07,
      "loss": 0.6778,
      "step": 27831
    },
    {
      "epoch": 0.8019362646228317,
      "grad_norm": 0.7114014625549316,
      "learning_rate": 9.939915237703784e-07,
      "loss": 0.6794,
      "step": 27832
    },
    {
      "epoch": 0.801965078084481,
      "grad_norm": 0.7440211772918701,
      "learning_rate": 9.937123243344303e-07,
      "loss": 0.6844,
      "step": 27833
    },
    {
      "epoch": 0.8019938915461303,
      "grad_norm": 0.7089316844940186,
      "learning_rate": 9.93433159789245e-07,
      "loss": 0.685,
      "step": 27834
    },
    {
      "epoch": 0.8020227050077796,
      "grad_norm": 0.7276276350021362,
      "learning_rate": 9.9315403013725e-07,
      "loss": 0.7061,
      "step": 27835
    },
    {
      "epoch": 0.8020515184694289,
      "grad_norm": 0.7156659364700317,
      "learning_rate": 9.928749353808792e-07,
      "loss": 0.6634,
      "step": 27836
    },
    {
      "epoch": 0.8020803319310782,
      "grad_norm": 0.7364010214805603,
      "learning_rate": 9.925958755225613e-07,
      "loss": 0.691,
      "step": 27837
    },
    {
      "epoch": 0.8021091453927275,
      "grad_norm": 0.7207628488540649,
      "learning_rate": 9.923168505647284e-07,
      "loss": 0.6982,
      "step": 27838
    },
    {
      "epoch": 0.8021379588543768,
      "grad_norm": 0.7259730100631714,
      "learning_rate": 9.920378605098068e-07,
      "loss": 0.683,
      "step": 27839
    },
    {
      "epoch": 0.8021667723160261,
      "grad_norm": 0.699303150177002,
      "learning_rate": 9.917589053602318e-07,
      "loss": 0.6667,
      "step": 27840
    },
    {
      "epoch": 0.8021955857776754,
      "grad_norm": 0.7349671125411987,
      "learning_rate": 9.914799851184282e-07,
      "loss": 0.7098,
      "step": 27841
    },
    {
      "epoch": 0.8022243992393246,
      "grad_norm": 0.7220246195793152,
      "learning_rate": 9.912010997868288e-07,
      "loss": 0.7064,
      "step": 27842
    },
    {
      "epoch": 0.8022532127009739,
      "grad_norm": 0.7176865935325623,
      "learning_rate": 9.909222493678588e-07,
      "loss": 0.6621,
      "step": 27843
    },
    {
      "epoch": 0.8022820261626232,
      "grad_norm": 0.7338253855705261,
      "learning_rate": 9.906434338639482e-07,
      "loss": 0.6795,
      "step": 27844
    },
    {
      "epoch": 0.8023108396242724,
      "grad_norm": 0.8176147937774658,
      "learning_rate": 9.903646532775268e-07,
      "loss": 0.6961,
      "step": 27845
    },
    {
      "epoch": 0.8023396530859217,
      "grad_norm": 0.7011461853981018,
      "learning_rate": 9.900859076110202e-07,
      "loss": 0.6852,
      "step": 27846
    },
    {
      "epoch": 0.802368466547571,
      "grad_norm": 0.7184922695159912,
      "learning_rate": 9.898071968668583e-07,
      "loss": 0.6901,
      "step": 27847
    },
    {
      "epoch": 0.8023972800092203,
      "grad_norm": 0.7283228039741516,
      "learning_rate": 9.895285210474654e-07,
      "loss": 0.6928,
      "step": 27848
    },
    {
      "epoch": 0.8024260934708696,
      "grad_norm": 0.7042943835258484,
      "learning_rate": 9.892498801552708e-07,
      "loss": 0.6591,
      "step": 27849
    },
    {
      "epoch": 0.8024549069325189,
      "grad_norm": 0.7607520818710327,
      "learning_rate": 9.889712741927e-07,
      "loss": 0.6936,
      "step": 27850
    },
    {
      "epoch": 0.8024837203941682,
      "grad_norm": 0.7234969735145569,
      "learning_rate": 9.886927031621796e-07,
      "loss": 0.6718,
      "step": 27851
    },
    {
      "epoch": 0.8025125338558174,
      "grad_norm": 0.7501067519187927,
      "learning_rate": 9.884141670661357e-07,
      "loss": 0.6901,
      "step": 27852
    },
    {
      "epoch": 0.8025413473174667,
      "grad_norm": 0.7094476819038391,
      "learning_rate": 9.881356659069957e-07,
      "loss": 0.681,
      "step": 27853
    },
    {
      "epoch": 0.802570160779116,
      "grad_norm": 0.7227343320846558,
      "learning_rate": 9.878571996871821e-07,
      "loss": 0.671,
      "step": 27854
    },
    {
      "epoch": 0.8025989742407653,
      "grad_norm": 0.711029052734375,
      "learning_rate": 9.875787684091236e-07,
      "loss": 0.7123,
      "step": 27855
    },
    {
      "epoch": 0.8026277877024146,
      "grad_norm": 0.7131268382072449,
      "learning_rate": 9.87300372075241e-07,
      "loss": 0.6969,
      "step": 27856
    },
    {
      "epoch": 0.8026566011640639,
      "grad_norm": 0.7156141996383667,
      "learning_rate": 9.87022010687963e-07,
      "loss": 0.6576,
      "step": 27857
    },
    {
      "epoch": 0.8026854146257131,
      "grad_norm": 0.7045563459396362,
      "learning_rate": 9.867436842497103e-07,
      "loss": 0.68,
      "step": 27858
    },
    {
      "epoch": 0.8027142280873624,
      "grad_norm": 0.7169761657714844,
      "learning_rate": 9.864653927629093e-07,
      "loss": 0.6843,
      "step": 27859
    },
    {
      "epoch": 0.8027430415490117,
      "grad_norm": 0.7133088707923889,
      "learning_rate": 9.861871362299812e-07,
      "loss": 0.6919,
      "step": 27860
    },
    {
      "epoch": 0.802771855010661,
      "grad_norm": 0.7441768050193787,
      "learning_rate": 9.859089146533513e-07,
      "loss": 0.6842,
      "step": 27861
    },
    {
      "epoch": 0.8028006684723102,
      "grad_norm": 0.7103492617607117,
      "learning_rate": 9.856307280354433e-07,
      "loss": 0.6654,
      "step": 27862
    },
    {
      "epoch": 0.8028294819339595,
      "grad_norm": 0.7149012088775635,
      "learning_rate": 9.85352576378677e-07,
      "loss": 0.6639,
      "step": 27863
    },
    {
      "epoch": 0.8028582953956088,
      "grad_norm": 0.7056867480278015,
      "learning_rate": 9.850744596854772e-07,
      "loss": 0.644,
      "step": 27864
    },
    {
      "epoch": 0.8028871088572581,
      "grad_norm": 0.7117286920547485,
      "learning_rate": 9.847963779582663e-07,
      "loss": 0.6499,
      "step": 27865
    },
    {
      "epoch": 0.8029159223189074,
      "grad_norm": 0.7179288864135742,
      "learning_rate": 9.845183311994637e-07,
      "loss": 0.6773,
      "step": 27866
    },
    {
      "epoch": 0.8029447357805567,
      "grad_norm": 0.7347801923751831,
      "learning_rate": 9.84240319411493e-07,
      "loss": 0.6711,
      "step": 27867
    },
    {
      "epoch": 0.802973549242206,
      "grad_norm": 0.7368819117546082,
      "learning_rate": 9.83962342596776e-07,
      "loss": 0.6718,
      "step": 27868
    },
    {
      "epoch": 0.8030023627038553,
      "grad_norm": 0.7316017746925354,
      "learning_rate": 9.836844007577313e-07,
      "loss": 0.6691,
      "step": 27869
    },
    {
      "epoch": 0.8030311761655046,
      "grad_norm": 0.7304413914680481,
      "learning_rate": 9.834064938967812e-07,
      "loss": 0.6985,
      "step": 27870
    },
    {
      "epoch": 0.8030599896271539,
      "grad_norm": 0.746728241443634,
      "learning_rate": 9.83128622016345e-07,
      "loss": 0.6924,
      "step": 27871
    },
    {
      "epoch": 0.803088803088803,
      "grad_norm": 0.7415895462036133,
      "learning_rate": 9.828507851188434e-07,
      "loss": 0.7036,
      "step": 27872
    },
    {
      "epoch": 0.8031176165504523,
      "grad_norm": 0.7153980731964111,
      "learning_rate": 9.825729832066954e-07,
      "loss": 0.6661,
      "step": 27873
    },
    {
      "epoch": 0.8031464300121016,
      "grad_norm": 0.7218804359436035,
      "learning_rate": 9.822952162823202e-07,
      "loss": 0.6752,
      "step": 27874
    },
    {
      "epoch": 0.8031752434737509,
      "grad_norm": 0.7144119739532471,
      "learning_rate": 9.820174843481377e-07,
      "loss": 0.6926,
      "step": 27875
    },
    {
      "epoch": 0.8032040569354002,
      "grad_norm": 0.7021499872207642,
      "learning_rate": 9.817397874065676e-07,
      "loss": 0.691,
      "step": 27876
    },
    {
      "epoch": 0.8032328703970495,
      "grad_norm": 0.7320951223373413,
      "learning_rate": 9.81462125460026e-07,
      "loss": 0.6906,
      "step": 27877
    },
    {
      "epoch": 0.8032616838586988,
      "grad_norm": 0.7235841751098633,
      "learning_rate": 9.811844985109336e-07,
      "loss": 0.7178,
      "step": 27878
    },
    {
      "epoch": 0.8032904973203481,
      "grad_norm": 0.7099961638450623,
      "learning_rate": 9.809069065617055e-07,
      "loss": 0.6876,
      "step": 27879
    },
    {
      "epoch": 0.8033193107819974,
      "grad_norm": 0.6995095014572144,
      "learning_rate": 9.806293496147617e-07,
      "loss": 0.6761,
      "step": 27880
    },
    {
      "epoch": 0.8033481242436467,
      "grad_norm": 0.7133795619010925,
      "learning_rate": 9.80351827672517e-07,
      "loss": 0.6814,
      "step": 27881
    },
    {
      "epoch": 0.803376937705296,
      "grad_norm": 0.7075842022895813,
      "learning_rate": 9.800743407373896e-07,
      "loss": 0.6535,
      "step": 27882
    },
    {
      "epoch": 0.8034057511669452,
      "grad_norm": 0.7133140563964844,
      "learning_rate": 9.797968888117981e-07,
      "loss": 0.6997,
      "step": 27883
    },
    {
      "epoch": 0.8034345646285945,
      "grad_norm": 0.7104277610778809,
      "learning_rate": 9.795194718981555e-07,
      "loss": 0.6666,
      "step": 27884
    },
    {
      "epoch": 0.8034633780902437,
      "grad_norm": 0.715917706489563,
      "learning_rate": 9.792420899988803e-07,
      "loss": 0.6701,
      "step": 27885
    },
    {
      "epoch": 0.803492191551893,
      "grad_norm": 0.7004788517951965,
      "learning_rate": 9.789647431163863e-07,
      "loss": 0.6637,
      "step": 27886
    },
    {
      "epoch": 0.8035210050135423,
      "grad_norm": 0.717171311378479,
      "learning_rate": 9.786874312530897e-07,
      "loss": 0.6846,
      "step": 27887
    },
    {
      "epoch": 0.8035498184751916,
      "grad_norm": 0.7303561568260193,
      "learning_rate": 9.784101544114056e-07,
      "loss": 0.7027,
      "step": 27888
    },
    {
      "epoch": 0.8035786319368409,
      "grad_norm": 0.7063262462615967,
      "learning_rate": 9.7813291259375e-07,
      "loss": 0.6644,
      "step": 27889
    },
    {
      "epoch": 0.8036074453984902,
      "grad_norm": 0.7394115328788757,
      "learning_rate": 9.778557058025357e-07,
      "loss": 0.663,
      "step": 27890
    },
    {
      "epoch": 0.8036362588601395,
      "grad_norm": 0.7084582448005676,
      "learning_rate": 9.775785340401784e-07,
      "loss": 0.6685,
      "step": 27891
    },
    {
      "epoch": 0.8036650723217887,
      "grad_norm": 0.9598954319953918,
      "learning_rate": 9.773013973090896e-07,
      "loss": 0.6707,
      "step": 27892
    },
    {
      "epoch": 0.803693885783438,
      "grad_norm": 0.7234742641448975,
      "learning_rate": 9.77024295611686e-07,
      "loss": 0.6703,
      "step": 27893
    },
    {
      "epoch": 0.8037226992450873,
      "grad_norm": 0.7405790686607361,
      "learning_rate": 9.767472289503783e-07,
      "loss": 0.7073,
      "step": 27894
    },
    {
      "epoch": 0.8037515127067366,
      "grad_norm": 0.7421794533729553,
      "learning_rate": 9.764701973275808e-07,
      "loss": 0.6901,
      "step": 27895
    },
    {
      "epoch": 0.8037803261683859,
      "grad_norm": 0.7174975872039795,
      "learning_rate": 9.761932007457053e-07,
      "loss": 0.6771,
      "step": 27896
    },
    {
      "epoch": 0.8038091396300352,
      "grad_norm": 0.7409121990203857,
      "learning_rate": 9.759162392071658e-07,
      "loss": 0.6827,
      "step": 27897
    },
    {
      "epoch": 0.8038379530916845,
      "grad_norm": 0.735284686088562,
      "learning_rate": 9.75639312714371e-07,
      "loss": 0.6673,
      "step": 27898
    },
    {
      "epoch": 0.8038667665533337,
      "grad_norm": 0.7033460140228271,
      "learning_rate": 9.753624212697372e-07,
      "loss": 0.6646,
      "step": 27899
    },
    {
      "epoch": 0.803895580014983,
      "grad_norm": 0.7160795331001282,
      "learning_rate": 9.75085564875672e-07,
      "loss": 0.653,
      "step": 27900
    },
    {
      "epoch": 0.8039243934766322,
      "grad_norm": 0.7103889584541321,
      "learning_rate": 9.748087435345894e-07,
      "loss": 0.6678,
      "step": 27901
    },
    {
      "epoch": 0.8039532069382815,
      "grad_norm": 0.7065383791923523,
      "learning_rate": 9.745319572488982e-07,
      "loss": 0.6968,
      "step": 27902
    },
    {
      "epoch": 0.8039820203999308,
      "grad_norm": 0.7418987154960632,
      "learning_rate": 9.742552060210103e-07,
      "loss": 0.6874,
      "step": 27903
    },
    {
      "epoch": 0.8040108338615801,
      "grad_norm": 0.7086259126663208,
      "learning_rate": 9.739784898533344e-07,
      "loss": 0.6887,
      "step": 27904
    },
    {
      "epoch": 0.8040396473232294,
      "grad_norm": 0.7215201258659363,
      "learning_rate": 9.737018087482813e-07,
      "loss": 0.6993,
      "step": 27905
    },
    {
      "epoch": 0.8040684607848787,
      "grad_norm": 0.7104951739311218,
      "learning_rate": 9.734251627082613e-07,
      "loss": 0.6783,
      "step": 27906
    },
    {
      "epoch": 0.804097274246528,
      "grad_norm": 0.718231737613678,
      "learning_rate": 9.731485517356825e-07,
      "loss": 0.6753,
      "step": 27907
    },
    {
      "epoch": 0.8041260877081773,
      "grad_norm": 0.7129110097885132,
      "learning_rate": 9.728719758329552e-07,
      "loss": 0.6858,
      "step": 27908
    },
    {
      "epoch": 0.8041549011698266,
      "grad_norm": 0.7311139106750488,
      "learning_rate": 9.725954350024863e-07,
      "loss": 0.6966,
      "step": 27909
    },
    {
      "epoch": 0.8041837146314759,
      "grad_norm": 0.7021878957748413,
      "learning_rate": 9.723189292466856e-07,
      "loss": 0.6905,
      "step": 27910
    },
    {
      "epoch": 0.8042125280931252,
      "grad_norm": 0.7353085875511169,
      "learning_rate": 9.720424585679606e-07,
      "loss": 0.6778,
      "step": 27911
    },
    {
      "epoch": 0.8042413415547743,
      "grad_norm": 0.7232966423034668,
      "learning_rate": 9.717660229687204e-07,
      "loss": 0.6827,
      "step": 27912
    },
    {
      "epoch": 0.8042701550164236,
      "grad_norm": 0.7397752404212952,
      "learning_rate": 9.714896224513699e-07,
      "loss": 0.6874,
      "step": 27913
    },
    {
      "epoch": 0.8042989684780729,
      "grad_norm": 0.7412394881248474,
      "learning_rate": 9.712132570183192e-07,
      "loss": 0.6741,
      "step": 27914
    },
    {
      "epoch": 0.8043277819397222,
      "grad_norm": 0.7268896698951721,
      "learning_rate": 9.709369266719726e-07,
      "loss": 0.6936,
      "step": 27915
    },
    {
      "epoch": 0.8043565954013715,
      "grad_norm": 0.7265128493309021,
      "learning_rate": 9.706606314147388e-07,
      "loss": 0.6844,
      "step": 27916
    },
    {
      "epoch": 0.8043854088630208,
      "grad_norm": 0.719785213470459,
      "learning_rate": 9.703843712490218e-07,
      "loss": 0.6993,
      "step": 27917
    },
    {
      "epoch": 0.8044142223246701,
      "grad_norm": 0.7400436997413635,
      "learning_rate": 9.701081461772306e-07,
      "loss": 0.696,
      "step": 27918
    },
    {
      "epoch": 0.8044430357863194,
      "grad_norm": 0.6905139684677124,
      "learning_rate": 9.698319562017672e-07,
      "loss": 0.6719,
      "step": 27919
    },
    {
      "epoch": 0.8044718492479687,
      "grad_norm": 0.7303112745285034,
      "learning_rate": 9.695558013250388e-07,
      "loss": 0.6959,
      "step": 27920
    },
    {
      "epoch": 0.804500662709618,
      "grad_norm": 0.7305078506469727,
      "learning_rate": 9.692796815494521e-07,
      "loss": 0.6846,
      "step": 27921
    },
    {
      "epoch": 0.8045294761712672,
      "grad_norm": 0.7300041317939758,
      "learning_rate": 9.69003596877408e-07,
      "loss": 0.6779,
      "step": 27922
    },
    {
      "epoch": 0.8045582896329165,
      "grad_norm": 0.7201499342918396,
      "learning_rate": 9.687275473113138e-07,
      "loss": 0.6761,
      "step": 27923
    },
    {
      "epoch": 0.8045871030945658,
      "grad_norm": 0.7071051597595215,
      "learning_rate": 9.684515328535738e-07,
      "loss": 0.669,
      "step": 27924
    },
    {
      "epoch": 0.8046159165562151,
      "grad_norm": 0.7338573336601257,
      "learning_rate": 9.681755535065901e-07,
      "loss": 0.6678,
      "step": 27925
    },
    {
      "epoch": 0.8046447300178643,
      "grad_norm": 0.7050053477287292,
      "learning_rate": 9.67899609272766e-07,
      "loss": 0.6684,
      "step": 27926
    },
    {
      "epoch": 0.8046735434795136,
      "grad_norm": 0.7469555735588074,
      "learning_rate": 9.67623700154508e-07,
      "loss": 0.7013,
      "step": 27927
    },
    {
      "epoch": 0.8047023569411629,
      "grad_norm": 0.7318098545074463,
      "learning_rate": 9.673478261542147e-07,
      "loss": 0.6698,
      "step": 27928
    },
    {
      "epoch": 0.8047311704028122,
      "grad_norm": 0.7206794023513794,
      "learning_rate": 9.670719872742923e-07,
      "loss": 0.6695,
      "step": 27929
    },
    {
      "epoch": 0.8047599838644615,
      "grad_norm": 0.7311433553695679,
      "learning_rate": 9.667961835171403e-07,
      "loss": 0.6708,
      "step": 27930
    },
    {
      "epoch": 0.8047887973261107,
      "grad_norm": 0.7165062427520752,
      "learning_rate": 9.665204148851625e-07,
      "loss": 0.6607,
      "step": 27931
    },
    {
      "epoch": 0.80481761078776,
      "grad_norm": 0.7063242197036743,
      "learning_rate": 9.66244681380759e-07,
      "loss": 0.6779,
      "step": 27932
    },
    {
      "epoch": 0.8048464242494093,
      "grad_norm": 0.719158947467804,
      "learning_rate": 9.659689830063335e-07,
      "loss": 0.6841,
      "step": 27933
    },
    {
      "epoch": 0.8048752377110586,
      "grad_norm": 0.7444418668746948,
      "learning_rate": 9.656933197642826e-07,
      "loss": 0.682,
      "step": 27934
    },
    {
      "epoch": 0.8049040511727079,
      "grad_norm": 0.7186392545700073,
      "learning_rate": 9.65417691657013e-07,
      "loss": 0.6941,
      "step": 27935
    },
    {
      "epoch": 0.8049328646343572,
      "grad_norm": 0.7103058695793152,
      "learning_rate": 9.651420986869203e-07,
      "loss": 0.6892,
      "step": 27936
    },
    {
      "epoch": 0.8049616780960065,
      "grad_norm": 0.7440008521080017,
      "learning_rate": 9.648665408564084e-07,
      "loss": 0.7027,
      "step": 27937
    },
    {
      "epoch": 0.8049904915576558,
      "grad_norm": 0.7206553816795349,
      "learning_rate": 9.645910181678742e-07,
      "loss": 0.6647,
      "step": 27938
    },
    {
      "epoch": 0.805019305019305,
      "grad_norm": 0.7189446091651917,
      "learning_rate": 9.643155306237189e-07,
      "loss": 0.6698,
      "step": 27939
    },
    {
      "epoch": 0.8050481184809543,
      "grad_norm": 0.7231674194335938,
      "learning_rate": 9.640400782263405e-07,
      "loss": 0.6694,
      "step": 27940
    },
    {
      "epoch": 0.8050769319426035,
      "grad_norm": 0.728681206703186,
      "learning_rate": 9.63764660978138e-07,
      "loss": 0.705,
      "step": 27941
    },
    {
      "epoch": 0.8051057454042528,
      "grad_norm": 0.7228068709373474,
      "learning_rate": 9.63489278881512e-07,
      "loss": 0.6816,
      "step": 27942
    },
    {
      "epoch": 0.8051345588659021,
      "grad_norm": 0.7296456098556519,
      "learning_rate": 9.632139319388583e-07,
      "loss": 0.6834,
      "step": 27943
    },
    {
      "epoch": 0.8051633723275514,
      "grad_norm": 0.6920475363731384,
      "learning_rate": 9.62938620152577e-07,
      "loss": 0.6517,
      "step": 27944
    },
    {
      "epoch": 0.8051921857892007,
      "grad_norm": 0.7299639582633972,
      "learning_rate": 9.626633435250633e-07,
      "loss": 0.6856,
      "step": 27945
    },
    {
      "epoch": 0.80522099925085,
      "grad_norm": 0.7157140374183655,
      "learning_rate": 9.62388102058716e-07,
      "loss": 0.6843,
      "step": 27946
    },
    {
      "epoch": 0.8052498127124993,
      "grad_norm": 0.6914533376693726,
      "learning_rate": 9.621128957559328e-07,
      "loss": 0.6763,
      "step": 27947
    },
    {
      "epoch": 0.8052786261741486,
      "grad_norm": 0.694026529788971,
      "learning_rate": 9.618377246191107e-07,
      "loss": 0.6792,
      "step": 27948
    },
    {
      "epoch": 0.8053074396357979,
      "grad_norm": 0.7128356695175171,
      "learning_rate": 9.615625886506442e-07,
      "loss": 0.6914,
      "step": 27949
    },
    {
      "epoch": 0.8053362530974472,
      "grad_norm": 0.72378009557724,
      "learning_rate": 9.612874878529315e-07,
      "loss": 0.7057,
      "step": 27950
    },
    {
      "epoch": 0.8053650665590965,
      "grad_norm": 0.7069627046585083,
      "learning_rate": 9.610124222283668e-07,
      "loss": 0.6419,
      "step": 27951
    },
    {
      "epoch": 0.8053938800207457,
      "grad_norm": 0.7551357746124268,
      "learning_rate": 9.607373917793472e-07,
      "loss": 0.7005,
      "step": 27952
    },
    {
      "epoch": 0.8054226934823949,
      "grad_norm": 0.7168772220611572,
      "learning_rate": 9.60462396508266e-07,
      "loss": 0.6607,
      "step": 27953
    },
    {
      "epoch": 0.8054515069440442,
      "grad_norm": 0.7258699536323547,
      "learning_rate": 9.601874364175206e-07,
      "loss": 0.6921,
      "step": 27954
    },
    {
      "epoch": 0.8054803204056935,
      "grad_norm": 0.7078828811645508,
      "learning_rate": 9.59912511509503e-07,
      "loss": 0.6579,
      "step": 27955
    },
    {
      "epoch": 0.8055091338673428,
      "grad_norm": 0.7177445888519287,
      "learning_rate": 9.5963762178661e-07,
      "loss": 0.7016,
      "step": 27956
    },
    {
      "epoch": 0.8055379473289921,
      "grad_norm": 0.7225925922393799,
      "learning_rate": 9.593627672512334e-07,
      "loss": 0.6708,
      "step": 27957
    },
    {
      "epoch": 0.8055667607906414,
      "grad_norm": 0.7003762722015381,
      "learning_rate": 9.59087947905768e-07,
      "loss": 0.6755,
      "step": 27958
    },
    {
      "epoch": 0.8055955742522907,
      "grad_norm": 0.7158631682395935,
      "learning_rate": 9.588131637526066e-07,
      "loss": 0.6807,
      "step": 27959
    },
    {
      "epoch": 0.80562438771394,
      "grad_norm": 0.752419650554657,
      "learning_rate": 9.585384147941447e-07,
      "loss": 0.712,
      "step": 27960
    },
    {
      "epoch": 0.8056532011755893,
      "grad_norm": 0.7253852486610413,
      "learning_rate": 9.582637010327716e-07,
      "loss": 0.6877,
      "step": 27961
    },
    {
      "epoch": 0.8056820146372385,
      "grad_norm": 0.6937529444694519,
      "learning_rate": 9.579890224708827e-07,
      "loss": 0.6787,
      "step": 27962
    },
    {
      "epoch": 0.8057108280988878,
      "grad_norm": 0.7132723331451416,
      "learning_rate": 9.577143791108678e-07,
      "loss": 0.705,
      "step": 27963
    },
    {
      "epoch": 0.8057396415605371,
      "grad_norm": 0.7308961749076843,
      "learning_rate": 9.574397709551198e-07,
      "loss": 0.6985,
      "step": 27964
    },
    {
      "epoch": 0.8057684550221864,
      "grad_norm": 0.7287998795509338,
      "learning_rate": 9.571651980060314e-07,
      "loss": 0.6855,
      "step": 27965
    },
    {
      "epoch": 0.8057972684838356,
      "grad_norm": 0.7122610807418823,
      "learning_rate": 9.568906602659917e-07,
      "loss": 0.6803,
      "step": 27966
    },
    {
      "epoch": 0.8058260819454849,
      "grad_norm": 0.7078864574432373,
      "learning_rate": 9.56616157737394e-07,
      "loss": 0.6727,
      "step": 27967
    },
    {
      "epoch": 0.8058548954071342,
      "grad_norm": 0.7208314538002014,
      "learning_rate": 9.563416904226263e-07,
      "loss": 0.6964,
      "step": 27968
    },
    {
      "epoch": 0.8058837088687835,
      "grad_norm": 0.7277323603630066,
      "learning_rate": 9.560672583240814e-07,
      "loss": 0.6734,
      "step": 27969
    },
    {
      "epoch": 0.8059125223304328,
      "grad_norm": 0.7221807837486267,
      "learning_rate": 9.557928614441458e-07,
      "loss": 0.6964,
      "step": 27970
    },
    {
      "epoch": 0.805941335792082,
      "grad_norm": 0.7278417348861694,
      "learning_rate": 9.555184997852147e-07,
      "loss": 0.689,
      "step": 27971
    },
    {
      "epoch": 0.8059701492537313,
      "grad_norm": 0.7138343453407288,
      "learning_rate": 9.552441733496726e-07,
      "loss": 0.6807,
      "step": 27972
    },
    {
      "epoch": 0.8059989627153806,
      "grad_norm": 0.7228782773017883,
      "learning_rate": 9.549698821399116e-07,
      "loss": 0.6756,
      "step": 27973
    },
    {
      "epoch": 0.8060277761770299,
      "grad_norm": 0.7444069981575012,
      "learning_rate": 9.546956261583185e-07,
      "loss": 0.6763,
      "step": 27974
    },
    {
      "epoch": 0.8060565896386792,
      "grad_norm": 0.7403919100761414,
      "learning_rate": 9.544214054072843e-07,
      "loss": 0.6767,
      "step": 27975
    },
    {
      "epoch": 0.8060854031003285,
      "grad_norm": 0.7364217042922974,
      "learning_rate": 9.541472198891931e-07,
      "loss": 0.7058,
      "step": 27976
    },
    {
      "epoch": 0.8061142165619778,
      "grad_norm": 0.717831552028656,
      "learning_rate": 9.538730696064374e-07,
      "loss": 0.6665,
      "step": 27977
    },
    {
      "epoch": 0.8061430300236271,
      "grad_norm": 0.7224264144897461,
      "learning_rate": 9.535989545614017e-07,
      "loss": 0.6865,
      "step": 27978
    },
    {
      "epoch": 0.8061718434852764,
      "grad_norm": 0.7095180153846741,
      "learning_rate": 9.533248747564733e-07,
      "loss": 0.6724,
      "step": 27979
    },
    {
      "epoch": 0.8062006569469256,
      "grad_norm": 0.7247257232666016,
      "learning_rate": 9.530508301940417e-07,
      "loss": 0.6766,
      "step": 27980
    },
    {
      "epoch": 0.8062294704085748,
      "grad_norm": 0.7159029841423035,
      "learning_rate": 9.527768208764909e-07,
      "loss": 0.6713,
      "step": 27981
    },
    {
      "epoch": 0.8062582838702241,
      "grad_norm": 0.7278997898101807,
      "learning_rate": 9.525028468062081e-07,
      "loss": 0.6806,
      "step": 27982
    },
    {
      "epoch": 0.8062870973318734,
      "grad_norm": 0.7250685691833496,
      "learning_rate": 9.522289079855801e-07,
      "loss": 0.6695,
      "step": 27983
    },
    {
      "epoch": 0.8063159107935227,
      "grad_norm": 0.7247670888900757,
      "learning_rate": 9.519550044169918e-07,
      "loss": 0.6773,
      "step": 27984
    },
    {
      "epoch": 0.806344724255172,
      "grad_norm": 0.730372428894043,
      "learning_rate": 9.516811361028283e-07,
      "loss": 0.6669,
      "step": 27985
    },
    {
      "epoch": 0.8063735377168213,
      "grad_norm": 0.6974868774414062,
      "learning_rate": 9.514073030454763e-07,
      "loss": 0.6819,
      "step": 27986
    },
    {
      "epoch": 0.8064023511784706,
      "grad_norm": 0.7399664521217346,
      "learning_rate": 9.511335052473187e-07,
      "loss": 0.6735,
      "step": 27987
    },
    {
      "epoch": 0.8064311646401199,
      "grad_norm": 0.7322019338607788,
      "learning_rate": 9.508597427107419e-07,
      "loss": 0.7147,
      "step": 27988
    },
    {
      "epoch": 0.8064599781017692,
      "grad_norm": 0.7226223349571228,
      "learning_rate": 9.50586015438128e-07,
      "loss": 0.6846,
      "step": 27989
    },
    {
      "epoch": 0.8064887915634185,
      "grad_norm": 0.7289362549781799,
      "learning_rate": 9.503123234318634e-07,
      "loss": 0.6729,
      "step": 27990
    },
    {
      "epoch": 0.8065176050250678,
      "grad_norm": 0.7146110534667969,
      "learning_rate": 9.500386666943285e-07,
      "loss": 0.6754,
      "step": 27991
    },
    {
      "epoch": 0.806546418486717,
      "grad_norm": 0.7089695930480957,
      "learning_rate": 9.4976504522791e-07,
      "loss": 0.6925,
      "step": 27992
    },
    {
      "epoch": 0.8065752319483662,
      "grad_norm": 0.7229148149490356,
      "learning_rate": 9.49491459034988e-07,
      "loss": 0.6726,
      "step": 27993
    },
    {
      "epoch": 0.8066040454100155,
      "grad_norm": 0.7224907279014587,
      "learning_rate": 9.492179081179459e-07,
      "loss": 0.7117,
      "step": 27994
    },
    {
      "epoch": 0.8066328588716648,
      "grad_norm": 0.728502094745636,
      "learning_rate": 9.489443924791669e-07,
      "loss": 0.6993,
      "step": 27995
    },
    {
      "epoch": 0.8066616723333141,
      "grad_norm": 0.7243815660476685,
      "learning_rate": 9.486709121210336e-07,
      "loss": 0.6896,
      "step": 27996
    },
    {
      "epoch": 0.8066904857949634,
      "grad_norm": 0.7492812871932983,
      "learning_rate": 9.483974670459261e-07,
      "loss": 0.7056,
      "step": 27997
    },
    {
      "epoch": 0.8067192992566127,
      "grad_norm": 0.7247233390808105,
      "learning_rate": 9.481240572562273e-07,
      "loss": 0.704,
      "step": 27998
    },
    {
      "epoch": 0.806748112718262,
      "grad_norm": 0.7155755162239075,
      "learning_rate": 9.478506827543166e-07,
      "loss": 0.6686,
      "step": 27999
    },
    {
      "epoch": 0.8067769261799113,
      "grad_norm": 0.7218701839447021,
      "learning_rate": 9.475773435425767e-07,
      "loss": 0.6982,
      "step": 28000
    },
    {
      "epoch": 0.8068057396415605,
      "grad_norm": 0.7166585326194763,
      "learning_rate": 9.473040396233862e-07,
      "loss": 0.703,
      "step": 28001
    },
    {
      "epoch": 0.8068345531032098,
      "grad_norm": 0.730754554271698,
      "learning_rate": 9.470307709991266e-07,
      "loss": 0.6769,
      "step": 28002
    },
    {
      "epoch": 0.8068633665648591,
      "grad_norm": 0.7392805814743042,
      "learning_rate": 9.467575376721782e-07,
      "loss": 0.698,
      "step": 28003
    },
    {
      "epoch": 0.8068921800265084,
      "grad_norm": 0.7143369913101196,
      "learning_rate": 9.464843396449191e-07,
      "loss": 0.6594,
      "step": 28004
    },
    {
      "epoch": 0.8069209934881577,
      "grad_norm": 0.7124889492988586,
      "learning_rate": 9.462111769197302e-07,
      "loss": 0.6734,
      "step": 28005
    },
    {
      "epoch": 0.806949806949807,
      "grad_norm": 0.7539776563644409,
      "learning_rate": 9.459380494989873e-07,
      "loss": 0.6877,
      "step": 28006
    },
    {
      "epoch": 0.8069786204114562,
      "grad_norm": 0.7271587252616882,
      "learning_rate": 9.45664957385074e-07,
      "loss": 0.6875,
      "step": 28007
    },
    {
      "epoch": 0.8070074338731055,
      "grad_norm": 0.7308655977249146,
      "learning_rate": 9.453919005803647e-07,
      "loss": 0.6916,
      "step": 28008
    },
    {
      "epoch": 0.8070362473347548,
      "grad_norm": 0.7295401692390442,
      "learning_rate": 9.451188790872401e-07,
      "loss": 0.6979,
      "step": 28009
    },
    {
      "epoch": 0.807065060796404,
      "grad_norm": 0.7295549511909485,
      "learning_rate": 9.448458929080756e-07,
      "loss": 0.6519,
      "step": 28010
    },
    {
      "epoch": 0.8070938742580533,
      "grad_norm": 0.7100515365600586,
      "learning_rate": 9.44572942045251e-07,
      "loss": 0.6738,
      "step": 28011
    },
    {
      "epoch": 0.8071226877197026,
      "grad_norm": 0.7202882766723633,
      "learning_rate": 9.443000265011404e-07,
      "loss": 0.7,
      "step": 28012
    },
    {
      "epoch": 0.8071515011813519,
      "grad_norm": 0.7205687761306763,
      "learning_rate": 9.440271462781237e-07,
      "loss": 0.6867,
      "step": 28013
    },
    {
      "epoch": 0.8071803146430012,
      "grad_norm": 0.7269625067710876,
      "learning_rate": 9.437543013785749e-07,
      "loss": 0.6734,
      "step": 28014
    },
    {
      "epoch": 0.8072091281046505,
      "grad_norm": 0.7040286064147949,
      "learning_rate": 9.434814918048729e-07,
      "loss": 0.6759,
      "step": 28015
    },
    {
      "epoch": 0.8072379415662998,
      "grad_norm": 0.7253247499465942,
      "learning_rate": 9.432087175593907e-07,
      "loss": 0.6928,
      "step": 28016
    },
    {
      "epoch": 0.8072667550279491,
      "grad_norm": 0.7064987421035767,
      "learning_rate": 9.429359786445052e-07,
      "loss": 0.6832,
      "step": 28017
    },
    {
      "epoch": 0.8072955684895984,
      "grad_norm": 0.7463707327842712,
      "learning_rate": 9.426632750625919e-07,
      "loss": 0.6658,
      "step": 28018
    },
    {
      "epoch": 0.8073243819512477,
      "grad_norm": 0.7232109904289246,
      "learning_rate": 9.423906068160265e-07,
      "loss": 0.6818,
      "step": 28019
    },
    {
      "epoch": 0.8073531954128969,
      "grad_norm": 0.7235790491104126,
      "learning_rate": 9.421179739071817e-07,
      "loss": 0.6959,
      "step": 28020
    },
    {
      "epoch": 0.8073820088745461,
      "grad_norm": 0.7460072040557861,
      "learning_rate": 9.418453763384344e-07,
      "loss": 0.6935,
      "step": 28021
    },
    {
      "epoch": 0.8074108223361954,
      "grad_norm": 0.7387884259223938,
      "learning_rate": 9.41572814112156e-07,
      "loss": 0.6854,
      "step": 28022
    },
    {
      "epoch": 0.8074396357978447,
      "grad_norm": 0.721618115901947,
      "learning_rate": 9.413002872307208e-07,
      "loss": 0.6748,
      "step": 28023
    },
    {
      "epoch": 0.807468449259494,
      "grad_norm": 0.7205725312232971,
      "learning_rate": 9.410277956965047e-07,
      "loss": 0.6955,
      "step": 28024
    },
    {
      "epoch": 0.8074972627211433,
      "grad_norm": 0.7313064932823181,
      "learning_rate": 9.407553395118779e-07,
      "loss": 0.7019,
      "step": 28025
    },
    {
      "epoch": 0.8075260761827926,
      "grad_norm": 0.7331163287162781,
      "learning_rate": 9.404829186792153e-07,
      "loss": 0.688,
      "step": 28026
    },
    {
      "epoch": 0.8075548896444419,
      "grad_norm": 0.7135646343231201,
      "learning_rate": 9.402105332008871e-07,
      "loss": 0.6865,
      "step": 28027
    },
    {
      "epoch": 0.8075837031060912,
      "grad_norm": 0.7339548468589783,
      "learning_rate": 9.399381830792686e-07,
      "loss": 0.658,
      "step": 28028
    },
    {
      "epoch": 0.8076125165677405,
      "grad_norm": 0.7061184644699097,
      "learning_rate": 9.396658683167281e-07,
      "loss": 0.667,
      "step": 28029
    },
    {
      "epoch": 0.8076413300293898,
      "grad_norm": 0.690896213054657,
      "learning_rate": 9.393935889156391e-07,
      "loss": 0.6516,
      "step": 28030
    },
    {
      "epoch": 0.807670143491039,
      "grad_norm": 0.6993281245231628,
      "learning_rate": 9.391213448783731e-07,
      "loss": 0.6768,
      "step": 28031
    },
    {
      "epoch": 0.8076989569526883,
      "grad_norm": 0.71033775806427,
      "learning_rate": 9.388491362073022e-07,
      "loss": 0.6795,
      "step": 28032
    },
    {
      "epoch": 0.8077277704143376,
      "grad_norm": 0.734448254108429,
      "learning_rate": 9.385769629047942e-07,
      "loss": 0.7117,
      "step": 28033
    },
    {
      "epoch": 0.8077565838759868,
      "grad_norm": 0.7426904439926147,
      "learning_rate": 9.383048249732218e-07,
      "loss": 0.7055,
      "step": 28034
    },
    {
      "epoch": 0.8077853973376361,
      "grad_norm": 0.740868330001831,
      "learning_rate": 9.380327224149533e-07,
      "loss": 0.6727,
      "step": 28035
    },
    {
      "epoch": 0.8078142107992854,
      "grad_norm": 0.7308614253997803,
      "learning_rate": 9.377606552323603e-07,
      "loss": 0.6893,
      "step": 28036
    },
    {
      "epoch": 0.8078430242609347,
      "grad_norm": 0.7189586758613586,
      "learning_rate": 9.374886234278102e-07,
      "loss": 0.6784,
      "step": 28037
    },
    {
      "epoch": 0.807871837722584,
      "grad_norm": 0.7139129638671875,
      "learning_rate": 9.372166270036748e-07,
      "loss": 0.6845,
      "step": 28038
    },
    {
      "epoch": 0.8079006511842333,
      "grad_norm": 0.7139006853103638,
      "learning_rate": 9.369446659623193e-07,
      "loss": 0.6809,
      "step": 28039
    },
    {
      "epoch": 0.8079294646458826,
      "grad_norm": 0.7161251902580261,
      "learning_rate": 9.366727403061149e-07,
      "loss": 0.6549,
      "step": 28040
    },
    {
      "epoch": 0.8079582781075318,
      "grad_norm": 0.7355429530143738,
      "learning_rate": 9.3640085003743e-07,
      "loss": 0.6979,
      "step": 28041
    },
    {
      "epoch": 0.8079870915691811,
      "grad_norm": 0.7224432826042175,
      "learning_rate": 9.3612899515863e-07,
      "loss": 0.6618,
      "step": 28042
    },
    {
      "epoch": 0.8080159050308304,
      "grad_norm": 0.7015830874443054,
      "learning_rate": 9.358571756720846e-07,
      "loss": 0.6565,
      "step": 28043
    },
    {
      "epoch": 0.8080447184924797,
      "grad_norm": 0.7210572361946106,
      "learning_rate": 9.355853915801599e-07,
      "loss": 0.7151,
      "step": 28044
    },
    {
      "epoch": 0.808073531954129,
      "grad_norm": 0.7237178683280945,
      "learning_rate": 9.353136428852256e-07,
      "loss": 0.6806,
      "step": 28045
    },
    {
      "epoch": 0.8081023454157783,
      "grad_norm": 0.7364790439605713,
      "learning_rate": 9.350419295896446e-07,
      "loss": 0.6854,
      "step": 28046
    },
    {
      "epoch": 0.8081311588774275,
      "grad_norm": 0.749367356300354,
      "learning_rate": 9.347702516957857e-07,
      "loss": 0.6718,
      "step": 28047
    },
    {
      "epoch": 0.8081599723390768,
      "grad_norm": 0.7186453938484192,
      "learning_rate": 9.344986092060138e-07,
      "loss": 0.6912,
      "step": 28048
    },
    {
      "epoch": 0.8081887858007261,
      "grad_norm": 0.7052575349807739,
      "learning_rate": 9.342270021226957e-07,
      "loss": 0.675,
      "step": 28049
    },
    {
      "epoch": 0.8082175992623754,
      "grad_norm": 0.7214735150337219,
      "learning_rate": 9.339554304481952e-07,
      "loss": 0.6972,
      "step": 28050
    },
    {
      "epoch": 0.8082464127240246,
      "grad_norm": 0.7251240015029907,
      "learning_rate": 9.336838941848791e-07,
      "loss": 0.6715,
      "step": 28051
    },
    {
      "epoch": 0.8082752261856739,
      "grad_norm": 0.7090298533439636,
      "learning_rate": 9.334123933351108e-07,
      "loss": 0.6792,
      "step": 28052
    },
    {
      "epoch": 0.8083040396473232,
      "grad_norm": 0.7183727622032166,
      "learning_rate": 9.33140927901256e-07,
      "loss": 0.6834,
      "step": 28053
    },
    {
      "epoch": 0.8083328531089725,
      "grad_norm": 0.7265459895133972,
      "learning_rate": 9.328694978856767e-07,
      "loss": 0.6803,
      "step": 28054
    },
    {
      "epoch": 0.8083616665706218,
      "grad_norm": 0.7303224802017212,
      "learning_rate": 9.325981032907405e-07,
      "loss": 0.6646,
      "step": 28055
    },
    {
      "epoch": 0.8083904800322711,
      "grad_norm": 0.7194771766662598,
      "learning_rate": 9.323267441188071e-07,
      "loss": 0.6659,
      "step": 28056
    },
    {
      "epoch": 0.8084192934939204,
      "grad_norm": 0.7342276573181152,
      "learning_rate": 9.320554203722438e-07,
      "loss": 0.7021,
      "step": 28057
    },
    {
      "epoch": 0.8084481069555697,
      "grad_norm": 0.744587779045105,
      "learning_rate": 9.317841320534093e-07,
      "loss": 0.6888,
      "step": 28058
    },
    {
      "epoch": 0.808476920417219,
      "grad_norm": 0.7229480743408203,
      "learning_rate": 9.315128791646699e-07,
      "loss": 0.6861,
      "step": 28059
    },
    {
      "epoch": 0.8085057338788683,
      "grad_norm": 0.7154766917228699,
      "learning_rate": 9.312416617083853e-07,
      "loss": 0.6638,
      "step": 28060
    },
    {
      "epoch": 0.8085345473405174,
      "grad_norm": 0.7133076190948486,
      "learning_rate": 9.309704796869184e-07,
      "loss": 0.676,
      "step": 28061
    },
    {
      "epoch": 0.8085633608021667,
      "grad_norm": 0.7098782062530518,
      "learning_rate": 9.306993331026321e-07,
      "loss": 0.6747,
      "step": 28062
    },
    {
      "epoch": 0.808592174263816,
      "grad_norm": 0.7278738021850586,
      "learning_rate": 9.304282219578858e-07,
      "loss": 0.6762,
      "step": 28063
    },
    {
      "epoch": 0.8086209877254653,
      "grad_norm": 0.7433375120162964,
      "learning_rate": 9.301571462550423e-07,
      "loss": 0.6824,
      "step": 28064
    },
    {
      "epoch": 0.8086498011871146,
      "grad_norm": 0.6996492147445679,
      "learning_rate": 9.298861059964615e-07,
      "loss": 0.6738,
      "step": 28065
    },
    {
      "epoch": 0.8086786146487639,
      "grad_norm": 0.7502180933952332,
      "learning_rate": 9.296151011845034e-07,
      "loss": 0.6963,
      "step": 28066
    },
    {
      "epoch": 0.8087074281104132,
      "grad_norm": 0.730068027973175,
      "learning_rate": 9.293441318215285e-07,
      "loss": 0.6938,
      "step": 28067
    },
    {
      "epoch": 0.8087362415720625,
      "grad_norm": 0.748154878616333,
      "learning_rate": 9.290731979098988e-07,
      "loss": 0.7006,
      "step": 28068
    },
    {
      "epoch": 0.8087650550337118,
      "grad_norm": 0.7292957901954651,
      "learning_rate": 9.288022994519707e-07,
      "loss": 0.6681,
      "step": 28069
    },
    {
      "epoch": 0.8087938684953611,
      "grad_norm": 0.7369060516357422,
      "learning_rate": 9.285314364501063e-07,
      "loss": 0.6955,
      "step": 28070
    },
    {
      "epoch": 0.8088226819570103,
      "grad_norm": 0.7346971035003662,
      "learning_rate": 9.282606089066621e-07,
      "loss": 0.6494,
      "step": 28071
    },
    {
      "epoch": 0.8088514954186596,
      "grad_norm": 0.7367407083511353,
      "learning_rate": 9.279898168239987e-07,
      "loss": 0.7075,
      "step": 28072
    },
    {
      "epoch": 0.8088803088803089,
      "grad_norm": 0.7356414198875427,
      "learning_rate": 9.277190602044722e-07,
      "loss": 0.694,
      "step": 28073
    },
    {
      "epoch": 0.8089091223419581,
      "grad_norm": 0.7275328636169434,
      "learning_rate": 9.27448339050443e-07,
      "loss": 0.7017,
      "step": 28074
    },
    {
      "epoch": 0.8089379358036074,
      "grad_norm": 0.7274090051651001,
      "learning_rate": 9.271776533642668e-07,
      "loss": 0.685,
      "step": 28075
    },
    {
      "epoch": 0.8089667492652567,
      "grad_norm": 0.7352001667022705,
      "learning_rate": 9.269070031483029e-07,
      "loss": 0.6919,
      "step": 28076
    },
    {
      "epoch": 0.808995562726906,
      "grad_norm": 0.7068548202514648,
      "learning_rate": 9.266363884049062e-07,
      "loss": 0.682,
      "step": 28077
    },
    {
      "epoch": 0.8090243761885553,
      "grad_norm": 0.7199859619140625,
      "learning_rate": 9.263658091364347e-07,
      "loss": 0.6987,
      "step": 28078
    },
    {
      "epoch": 0.8090531896502046,
      "grad_norm": 0.7124538421630859,
      "learning_rate": 9.260952653452449e-07,
      "loss": 0.6884,
      "step": 28079
    },
    {
      "epoch": 0.8090820031118539,
      "grad_norm": 0.7154801487922668,
      "learning_rate": 9.258247570336942e-07,
      "loss": 0.6696,
      "step": 28080
    },
    {
      "epoch": 0.8091108165735031,
      "grad_norm": 0.7334144711494446,
      "learning_rate": 9.25554284204136e-07,
      "loss": 0.675,
      "step": 28081
    },
    {
      "epoch": 0.8091396300351524,
      "grad_norm": 0.7067642211914062,
      "learning_rate": 9.252838468589265e-07,
      "loss": 0.6814,
      "step": 28082
    },
    {
      "epoch": 0.8091684434968017,
      "grad_norm": 0.7038167119026184,
      "learning_rate": 9.250134450004233e-07,
      "loss": 0.6476,
      "step": 28083
    },
    {
      "epoch": 0.809197256958451,
      "grad_norm": 0.701845645904541,
      "learning_rate": 9.24743078630978e-07,
      "loss": 0.6791,
      "step": 28084
    },
    {
      "epoch": 0.8092260704201003,
      "grad_norm": 0.7095357775688171,
      "learning_rate": 9.24472747752948e-07,
      "loss": 0.6961,
      "step": 28085
    },
    {
      "epoch": 0.8092548838817496,
      "grad_norm": 0.7132285237312317,
      "learning_rate": 9.242024523686855e-07,
      "loss": 0.6933,
      "step": 28086
    },
    {
      "epoch": 0.8092836973433989,
      "grad_norm": 0.7113072872161865,
      "learning_rate": 9.239321924805461e-07,
      "loss": 0.6752,
      "step": 28087
    },
    {
      "epoch": 0.8093125108050481,
      "grad_norm": 0.7230879068374634,
      "learning_rate": 9.236619680908815e-07,
      "loss": 0.6752,
      "step": 28088
    },
    {
      "epoch": 0.8093413242666974,
      "grad_norm": 0.7458741664886475,
      "learning_rate": 9.233917792020475e-07,
      "loss": 0.6851,
      "step": 28089
    },
    {
      "epoch": 0.8093701377283467,
      "grad_norm": 0.7190454602241516,
      "learning_rate": 9.231216258163938e-07,
      "loss": 0.6792,
      "step": 28090
    },
    {
      "epoch": 0.8093989511899959,
      "grad_norm": 0.6931467652320862,
      "learning_rate": 9.228515079362777e-07,
      "loss": 0.6748,
      "step": 28091
    },
    {
      "epoch": 0.8094277646516452,
      "grad_norm": 0.730976402759552,
      "learning_rate": 9.225814255640481e-07,
      "loss": 0.7058,
      "step": 28092
    },
    {
      "epoch": 0.8094565781132945,
      "grad_norm": 0.7207424640655518,
      "learning_rate": 9.223113787020594e-07,
      "loss": 0.6879,
      "step": 28093
    },
    {
      "epoch": 0.8094853915749438,
      "grad_norm": 0.7151322960853577,
      "learning_rate": 9.220413673526613e-07,
      "loss": 0.6736,
      "step": 28094
    },
    {
      "epoch": 0.8095142050365931,
      "grad_norm": 0.7645272612571716,
      "learning_rate": 9.217713915182074e-07,
      "loss": 0.6804,
      "step": 28095
    },
    {
      "epoch": 0.8095430184982424,
      "grad_norm": 0.738220751285553,
      "learning_rate": 9.215014512010472e-07,
      "loss": 0.6779,
      "step": 28096
    },
    {
      "epoch": 0.8095718319598917,
      "grad_norm": 0.7163179516792297,
      "learning_rate": 9.212315464035332e-07,
      "loss": 0.6885,
      "step": 28097
    },
    {
      "epoch": 0.809600645421541,
      "grad_norm": 0.7340989708900452,
      "learning_rate": 9.209616771280138e-07,
      "loss": 0.6898,
      "step": 28098
    },
    {
      "epoch": 0.8096294588831903,
      "grad_norm": 0.7404671907424927,
      "learning_rate": 9.206918433768413e-07,
      "loss": 0.6788,
      "step": 28099
    },
    {
      "epoch": 0.8096582723448396,
      "grad_norm": 0.6980739831924438,
      "learning_rate": 9.20422045152366e-07,
      "loss": 0.668,
      "step": 28100
    },
    {
      "epoch": 0.8096870858064887,
      "grad_norm": 0.7257596254348755,
      "learning_rate": 9.201522824569348e-07,
      "loss": 0.6602,
      "step": 28101
    },
    {
      "epoch": 0.809715899268138,
      "grad_norm": 0.723950982093811,
      "learning_rate": 9.198825552928998e-07,
      "loss": 0.6885,
      "step": 28102
    },
    {
      "epoch": 0.8097447127297873,
      "grad_norm": 0.7445257306098938,
      "learning_rate": 9.196128636626084e-07,
      "loss": 0.6852,
      "step": 28103
    },
    {
      "epoch": 0.8097735261914366,
      "grad_norm": 0.7074880003929138,
      "learning_rate": 9.193432075684117e-07,
      "loss": 0.6473,
      "step": 28104
    },
    {
      "epoch": 0.8098023396530859,
      "grad_norm": 0.7376971244812012,
      "learning_rate": 9.190735870126549e-07,
      "loss": 0.6922,
      "step": 28105
    },
    {
      "epoch": 0.8098311531147352,
      "grad_norm": 0.7192913889884949,
      "learning_rate": 9.188040019976891e-07,
      "loss": 0.688,
      "step": 28106
    },
    {
      "epoch": 0.8098599665763845,
      "grad_norm": 0.7164978981018066,
      "learning_rate": 9.185344525258593e-07,
      "loss": 0.6882,
      "step": 28107
    },
    {
      "epoch": 0.8098887800380338,
      "grad_norm": 0.7324068546295166,
      "learning_rate": 9.182649385995163e-07,
      "loss": 0.6847,
      "step": 28108
    },
    {
      "epoch": 0.8099175934996831,
      "grad_norm": 0.7258588075637817,
      "learning_rate": 9.179954602210039e-07,
      "loss": 0.705,
      "step": 28109
    },
    {
      "epoch": 0.8099464069613324,
      "grad_norm": 0.6896049380302429,
      "learning_rate": 9.177260173926716e-07,
      "loss": 0.6582,
      "step": 28110
    },
    {
      "epoch": 0.8099752204229816,
      "grad_norm": 0.731478214263916,
      "learning_rate": 9.174566101168636e-07,
      "loss": 0.6885,
      "step": 28111
    },
    {
      "epoch": 0.8100040338846309,
      "grad_norm": 0.7299790382385254,
      "learning_rate": 9.171872383959291e-07,
      "loss": 0.6857,
      "step": 28112
    },
    {
      "epoch": 0.8100328473462802,
      "grad_norm": 0.717576265335083,
      "learning_rate": 9.169179022322111e-07,
      "loss": 0.6952,
      "step": 28113
    },
    {
      "epoch": 0.8100616608079295,
      "grad_norm": 0.7263712286949158,
      "learning_rate": 9.166486016280562e-07,
      "loss": 0.7025,
      "step": 28114
    },
    {
      "epoch": 0.8100904742695787,
      "grad_norm": 0.7304242849349976,
      "learning_rate": 9.163793365858103e-07,
      "loss": 0.6938,
      "step": 28115
    },
    {
      "epoch": 0.810119287731228,
      "grad_norm": 0.7424890995025635,
      "learning_rate": 9.161101071078199e-07,
      "loss": 0.6953,
      "step": 28116
    },
    {
      "epoch": 0.8101481011928773,
      "grad_norm": 0.723841667175293,
      "learning_rate": 9.158409131964264e-07,
      "loss": 0.7136,
      "step": 28117
    },
    {
      "epoch": 0.8101769146545266,
      "grad_norm": 0.7050641179084778,
      "learning_rate": 9.155717548539772e-07,
      "loss": 0.6688,
      "step": 28118
    },
    {
      "epoch": 0.8102057281161759,
      "grad_norm": 0.7275577187538147,
      "learning_rate": 9.15302632082814e-07,
      "loss": 0.6909,
      "step": 28119
    },
    {
      "epoch": 0.8102345415778252,
      "grad_norm": 0.7399085760116577,
      "learning_rate": 9.150335448852815e-07,
      "loss": 0.6655,
      "step": 28120
    },
    {
      "epoch": 0.8102633550394744,
      "grad_norm": 0.7296083569526672,
      "learning_rate": 9.147644932637245e-07,
      "loss": 0.6628,
      "step": 28121
    },
    {
      "epoch": 0.8102921685011237,
      "grad_norm": 0.7185814380645752,
      "learning_rate": 9.144954772204839e-07,
      "loss": 0.6793,
      "step": 28122
    },
    {
      "epoch": 0.810320981962773,
      "grad_norm": 0.738078236579895,
      "learning_rate": 9.142264967579051e-07,
      "loss": 0.6722,
      "step": 28123
    },
    {
      "epoch": 0.8103497954244223,
      "grad_norm": 0.7014704942703247,
      "learning_rate": 9.139575518783278e-07,
      "loss": 0.6793,
      "step": 28124
    },
    {
      "epoch": 0.8103786088860716,
      "grad_norm": 0.723050057888031,
      "learning_rate": 9.136886425840968e-07,
      "loss": 0.6979,
      "step": 28125
    },
    {
      "epoch": 0.8104074223477209,
      "grad_norm": 0.7254349589347839,
      "learning_rate": 9.13419768877551e-07,
      "loss": 0.6961,
      "step": 28126
    },
    {
      "epoch": 0.8104362358093702,
      "grad_norm": 0.6977531313896179,
      "learning_rate": 9.131509307610359e-07,
      "loss": 0.6644,
      "step": 28127
    },
    {
      "epoch": 0.8104650492710194,
      "grad_norm": 0.713638961315155,
      "learning_rate": 9.1288212823689e-07,
      "loss": 0.6905,
      "step": 28128
    },
    {
      "epoch": 0.8104938627326687,
      "grad_norm": 0.7137038707733154,
      "learning_rate": 9.126133613074556e-07,
      "loss": 0.6648,
      "step": 28129
    },
    {
      "epoch": 0.810522676194318,
      "grad_norm": 0.716528058052063,
      "learning_rate": 9.123446299750721e-07,
      "loss": 0.6819,
      "step": 28130
    },
    {
      "epoch": 0.8105514896559672,
      "grad_norm": 0.7296899557113647,
      "learning_rate": 9.120759342420821e-07,
      "loss": 0.6793,
      "step": 28131
    },
    {
      "epoch": 0.8105803031176165,
      "grad_norm": 0.7132286429405212,
      "learning_rate": 9.118072741108231e-07,
      "loss": 0.6818,
      "step": 28132
    },
    {
      "epoch": 0.8106091165792658,
      "grad_norm": 0.744196891784668,
      "learning_rate": 9.11538649583637e-07,
      "loss": 0.7103,
      "step": 28133
    },
    {
      "epoch": 0.8106379300409151,
      "grad_norm": 0.7223794460296631,
      "learning_rate": 9.112700606628616e-07,
      "loss": 0.6946,
      "step": 28134
    },
    {
      "epoch": 0.8106667435025644,
      "grad_norm": 0.8561646938323975,
      "learning_rate": 9.110015073508371e-07,
      "loss": 0.6783,
      "step": 28135
    },
    {
      "epoch": 0.8106955569642137,
      "grad_norm": 0.7341058254241943,
      "learning_rate": 9.107329896499017e-07,
      "loss": 0.6838,
      "step": 28136
    },
    {
      "epoch": 0.810724370425863,
      "grad_norm": 0.7196710109710693,
      "learning_rate": 9.104645075623936e-07,
      "loss": 0.6906,
      "step": 28137
    },
    {
      "epoch": 0.8107531838875123,
      "grad_norm": 0.7336493730545044,
      "learning_rate": 9.101960610906519e-07,
      "loss": 0.6974,
      "step": 28138
    },
    {
      "epoch": 0.8107819973491616,
      "grad_norm": 0.7099345922470093,
      "learning_rate": 9.099276502370153e-07,
      "loss": 0.6718,
      "step": 28139
    },
    {
      "epoch": 0.8108108108108109,
      "grad_norm": 0.7219609618186951,
      "learning_rate": 9.096592750038186e-07,
      "loss": 0.6915,
      "step": 28140
    },
    {
      "epoch": 0.8108396242724601,
      "grad_norm": 0.7181593775749207,
      "learning_rate": 9.09390935393401e-07,
      "loss": 0.6854,
      "step": 28141
    },
    {
      "epoch": 0.8108684377341093,
      "grad_norm": 0.6965124011039734,
      "learning_rate": 9.091226314081008e-07,
      "loss": 0.6865,
      "step": 28142
    },
    {
      "epoch": 0.8108972511957586,
      "grad_norm": 0.7332257628440857,
      "learning_rate": 9.088543630502516e-07,
      "loss": 0.692,
      "step": 28143
    },
    {
      "epoch": 0.8109260646574079,
      "grad_norm": 0.709546685218811,
      "learning_rate": 9.085861303221926e-07,
      "loss": 0.6875,
      "step": 28144
    },
    {
      "epoch": 0.8109548781190572,
      "grad_norm": 0.7279673218727112,
      "learning_rate": 9.083179332262576e-07,
      "loss": 0.6721,
      "step": 28145
    },
    {
      "epoch": 0.8109836915807065,
      "grad_norm": 0.7061697840690613,
      "learning_rate": 9.080497717647841e-07,
      "loss": 0.7008,
      "step": 28146
    },
    {
      "epoch": 0.8110125050423558,
      "grad_norm": 0.7284250259399414,
      "learning_rate": 9.077816459401051e-07,
      "loss": 0.6846,
      "step": 28147
    },
    {
      "epoch": 0.8110413185040051,
      "grad_norm": 0.7083590626716614,
      "learning_rate": 9.075135557545589e-07,
      "loss": 0.6668,
      "step": 28148
    },
    {
      "epoch": 0.8110701319656544,
      "grad_norm": 0.7364417910575867,
      "learning_rate": 9.072455012104758e-07,
      "loss": 0.6762,
      "step": 28149
    },
    {
      "epoch": 0.8110989454273037,
      "grad_norm": 0.7512146830558777,
      "learning_rate": 9.069774823101962e-07,
      "loss": 0.7052,
      "step": 28150
    },
    {
      "epoch": 0.8111277588889529,
      "grad_norm": 0.6993681192398071,
      "learning_rate": 9.067094990560493e-07,
      "loss": 0.685,
      "step": 28151
    },
    {
      "epoch": 0.8111565723506022,
      "grad_norm": 0.737570583820343,
      "learning_rate": 9.064415514503727e-07,
      "loss": 0.6652,
      "step": 28152
    },
    {
      "epoch": 0.8111853858122515,
      "grad_norm": 0.7183447480201721,
      "learning_rate": 9.061736394954968e-07,
      "loss": 0.6828,
      "step": 28153
    },
    {
      "epoch": 0.8112141992739008,
      "grad_norm": 0.7329667806625366,
      "learning_rate": 9.059057631937568e-07,
      "loss": 0.6738,
      "step": 28154
    },
    {
      "epoch": 0.81124301273555,
      "grad_norm": 0.7151634097099304,
      "learning_rate": 9.056379225474843e-07,
      "loss": 0.6802,
      "step": 28155
    },
    {
      "epoch": 0.8112718261971993,
      "grad_norm": 0.7421985864639282,
      "learning_rate": 9.053701175590141e-07,
      "loss": 0.6975,
      "step": 28156
    },
    {
      "epoch": 0.8113006396588486,
      "grad_norm": 0.7587408423423767,
      "learning_rate": 9.051023482306753e-07,
      "loss": 0.6797,
      "step": 28157
    },
    {
      "epoch": 0.8113294531204979,
      "grad_norm": 0.7097722291946411,
      "learning_rate": 9.048346145648018e-07,
      "loss": 0.6873,
      "step": 28158
    },
    {
      "epoch": 0.8113582665821472,
      "grad_norm": 0.7026298642158508,
      "learning_rate": 9.045669165637261e-07,
      "loss": 0.6899,
      "step": 28159
    },
    {
      "epoch": 0.8113870800437964,
      "grad_norm": 0.7158156037330627,
      "learning_rate": 9.042992542297779e-07,
      "loss": 0.6618,
      "step": 28160
    },
    {
      "epoch": 0.8114158935054457,
      "grad_norm": 0.6973716616630554,
      "learning_rate": 9.040316275652888e-07,
      "loss": 0.6709,
      "step": 28161
    },
    {
      "epoch": 0.811444706967095,
      "grad_norm": 0.7510629296302795,
      "learning_rate": 9.037640365725897e-07,
      "loss": 0.6784,
      "step": 28162
    },
    {
      "epoch": 0.8114735204287443,
      "grad_norm": 0.7161963582038879,
      "learning_rate": 9.034964812540125e-07,
      "loss": 0.6845,
      "step": 28163
    },
    {
      "epoch": 0.8115023338903936,
      "grad_norm": 0.7208653092384338,
      "learning_rate": 9.032289616118844e-07,
      "loss": 0.6727,
      "step": 28164
    },
    {
      "epoch": 0.8115311473520429,
      "grad_norm": 0.7232275009155273,
      "learning_rate": 9.029614776485385e-07,
      "loss": 0.6682,
      "step": 28165
    },
    {
      "epoch": 0.8115599608136922,
      "grad_norm": 0.7277164459228516,
      "learning_rate": 9.026940293663006e-07,
      "loss": 0.6695,
      "step": 28166
    },
    {
      "epoch": 0.8115887742753415,
      "grad_norm": 0.7277936339378357,
      "learning_rate": 9.024266167675039e-07,
      "loss": 0.6671,
      "step": 28167
    },
    {
      "epoch": 0.8116175877369908,
      "grad_norm": 0.7366591095924377,
      "learning_rate": 9.021592398544738e-07,
      "loss": 0.6752,
      "step": 28168
    },
    {
      "epoch": 0.81164640119864,
      "grad_norm": 0.7356122136116028,
      "learning_rate": 9.018918986295411e-07,
      "loss": 0.689,
      "step": 28169
    },
    {
      "epoch": 0.8116752146602892,
      "grad_norm": 0.7235230207443237,
      "learning_rate": 9.016245930950329e-07,
      "loss": 0.6911,
      "step": 28170
    },
    {
      "epoch": 0.8117040281219385,
      "grad_norm": 0.7200557589530945,
      "learning_rate": 9.013573232532785e-07,
      "loss": 0.6809,
      "step": 28171
    },
    {
      "epoch": 0.8117328415835878,
      "grad_norm": 0.7453780770301819,
      "learning_rate": 9.010900891066032e-07,
      "loss": 0.6812,
      "step": 28172
    },
    {
      "epoch": 0.8117616550452371,
      "grad_norm": 0.7143104672431946,
      "learning_rate": 9.008228906573363e-07,
      "loss": 0.6848,
      "step": 28173
    },
    {
      "epoch": 0.8117904685068864,
      "grad_norm": 0.736196756362915,
      "learning_rate": 9.005557279078042e-07,
      "loss": 0.7056,
      "step": 28174
    },
    {
      "epoch": 0.8118192819685357,
      "grad_norm": 0.7119483351707458,
      "learning_rate": 9.002886008603346e-07,
      "loss": 0.6697,
      "step": 28175
    },
    {
      "epoch": 0.811848095430185,
      "grad_norm": 0.7081972360610962,
      "learning_rate": 9.000215095172521e-07,
      "loss": 0.684,
      "step": 28176
    },
    {
      "epoch": 0.8118769088918343,
      "grad_norm": 0.7381733655929565,
      "learning_rate": 8.99754453880885e-07,
      "loss": 0.6913,
      "step": 28177
    },
    {
      "epoch": 0.8119057223534836,
      "grad_norm": 0.7581102252006531,
      "learning_rate": 8.994874339535569e-07,
      "loss": 0.6926,
      "step": 28178
    },
    {
      "epoch": 0.8119345358151329,
      "grad_norm": 0.7249851822853088,
      "learning_rate": 8.992204497375939e-07,
      "loss": 0.6883,
      "step": 28179
    },
    {
      "epoch": 0.8119633492767822,
      "grad_norm": 0.7306979894638062,
      "learning_rate": 8.989535012353229e-07,
      "loss": 0.7175,
      "step": 28180
    },
    {
      "epoch": 0.8119921627384314,
      "grad_norm": 0.6954357028007507,
      "learning_rate": 8.986865884490659e-07,
      "loss": 0.6938,
      "step": 28181
    },
    {
      "epoch": 0.8120209762000806,
      "grad_norm": 0.7386801242828369,
      "learning_rate": 8.984197113811505e-07,
      "loss": 0.6925,
      "step": 28182
    },
    {
      "epoch": 0.8120497896617299,
      "grad_norm": 0.7179281115531921,
      "learning_rate": 8.981528700338982e-07,
      "loss": 0.6765,
      "step": 28183
    },
    {
      "epoch": 0.8120786031233792,
      "grad_norm": 0.7100579142570496,
      "learning_rate": 8.978860644096348e-07,
      "loss": 0.685,
      "step": 28184
    },
    {
      "epoch": 0.8121074165850285,
      "grad_norm": 0.7621719837188721,
      "learning_rate": 8.976192945106809e-07,
      "loss": 0.686,
      "step": 28185
    },
    {
      "epoch": 0.8121362300466778,
      "grad_norm": 0.7288024425506592,
      "learning_rate": 8.973525603393645e-07,
      "loss": 0.7015,
      "step": 28186
    },
    {
      "epoch": 0.8121650435083271,
      "grad_norm": 0.6825854778289795,
      "learning_rate": 8.97085861898005e-07,
      "loss": 0.6595,
      "step": 28187
    },
    {
      "epoch": 0.8121938569699764,
      "grad_norm": 0.7254557013511658,
      "learning_rate": 8.968191991889274e-07,
      "loss": 0.6922,
      "step": 28188
    },
    {
      "epoch": 0.8122226704316257,
      "grad_norm": 0.7366114854812622,
      "learning_rate": 8.965525722144519e-07,
      "loss": 0.6706,
      "step": 28189
    },
    {
      "epoch": 0.812251483893275,
      "grad_norm": 0.7320071458816528,
      "learning_rate": 8.962859809769025e-07,
      "loss": 0.6819,
      "step": 28190
    },
    {
      "epoch": 0.8122802973549242,
      "grad_norm": 0.7198631167411804,
      "learning_rate": 8.960194254785986e-07,
      "loss": 0.6957,
      "step": 28191
    },
    {
      "epoch": 0.8123091108165735,
      "grad_norm": 0.7601487636566162,
      "learning_rate": 8.957529057218645e-07,
      "loss": 0.6669,
      "step": 28192
    },
    {
      "epoch": 0.8123379242782228,
      "grad_norm": 0.7124453783035278,
      "learning_rate": 8.954864217090187e-07,
      "loss": 0.6586,
      "step": 28193
    },
    {
      "epoch": 0.8123667377398721,
      "grad_norm": 0.70988929271698,
      "learning_rate": 8.952199734423844e-07,
      "loss": 0.676,
      "step": 28194
    },
    {
      "epoch": 0.8123955512015214,
      "grad_norm": 0.7204540371894836,
      "learning_rate": 8.949535609242793e-07,
      "loss": 0.6654,
      "step": 28195
    },
    {
      "epoch": 0.8124243646631706,
      "grad_norm": 0.7265556454658508,
      "learning_rate": 8.946871841570254e-07,
      "loss": 0.6926,
      "step": 28196
    },
    {
      "epoch": 0.8124531781248199,
      "grad_norm": 0.7458544969558716,
      "learning_rate": 8.944208431429424e-07,
      "loss": 0.6959,
      "step": 28197
    },
    {
      "epoch": 0.8124819915864692,
      "grad_norm": 0.7396057844161987,
      "learning_rate": 8.941545378843508e-07,
      "loss": 0.6664,
      "step": 28198
    },
    {
      "epoch": 0.8125108050481185,
      "grad_norm": 0.7458338737487793,
      "learning_rate": 8.938882683835676e-07,
      "loss": 0.7016,
      "step": 28199
    },
    {
      "epoch": 0.8125396185097677,
      "grad_norm": 0.7102893590927124,
      "learning_rate": 8.93622034642913e-07,
      "loss": 0.6704,
      "step": 28200
    },
    {
      "epoch": 0.812568431971417,
      "grad_norm": 0.7041719555854797,
      "learning_rate": 8.933558366647072e-07,
      "loss": 0.6558,
      "step": 28201
    },
    {
      "epoch": 0.8125972454330663,
      "grad_norm": 0.7018919587135315,
      "learning_rate": 8.930896744512652e-07,
      "loss": 0.6681,
      "step": 28202
    },
    {
      "epoch": 0.8126260588947156,
      "grad_norm": 0.7348940968513489,
      "learning_rate": 8.92823548004908e-07,
      "loss": 0.6957,
      "step": 28203
    },
    {
      "epoch": 0.8126548723563649,
      "grad_norm": 0.7574267983436584,
      "learning_rate": 8.92557457327951e-07,
      "loss": 0.708,
      "step": 28204
    },
    {
      "epoch": 0.8126836858180142,
      "grad_norm": 0.7286965250968933,
      "learning_rate": 8.922914024227137e-07,
      "loss": 0.6853,
      "step": 28205
    },
    {
      "epoch": 0.8127124992796635,
      "grad_norm": 0.7202223539352417,
      "learning_rate": 8.92025383291511e-07,
      "loss": 0.6754,
      "step": 28206
    },
    {
      "epoch": 0.8127413127413128,
      "grad_norm": 0.7306423783302307,
      "learning_rate": 8.917593999366619e-07,
      "loss": 0.7079,
      "step": 28207
    },
    {
      "epoch": 0.8127701262029621,
      "grad_norm": 0.7156233787536621,
      "learning_rate": 8.914934523604807e-07,
      "loss": 0.706,
      "step": 28208
    },
    {
      "epoch": 0.8127989396646113,
      "grad_norm": 0.7290973663330078,
      "learning_rate": 8.912275405652843e-07,
      "loss": 0.6802,
      "step": 28209
    },
    {
      "epoch": 0.8128277531262605,
      "grad_norm": 0.7309533357620239,
      "learning_rate": 8.909616645533886e-07,
      "loss": 0.6848,
      "step": 28210
    },
    {
      "epoch": 0.8128565665879098,
      "grad_norm": 0.7131642699241638,
      "learning_rate": 8.906958243271102e-07,
      "loss": 0.6814,
      "step": 28211
    },
    {
      "epoch": 0.8128853800495591,
      "grad_norm": 0.7135278582572937,
      "learning_rate": 8.904300198887628e-07,
      "loss": 0.6942,
      "step": 28212
    },
    {
      "epoch": 0.8129141935112084,
      "grad_norm": 0.7540271282196045,
      "learning_rate": 8.901642512406628e-07,
      "loss": 0.7117,
      "step": 28213
    },
    {
      "epoch": 0.8129430069728577,
      "grad_norm": 0.7362694144248962,
      "learning_rate": 8.898985183851228e-07,
      "loss": 0.693,
      "step": 28214
    },
    {
      "epoch": 0.812971820434507,
      "grad_norm": 0.7414819002151489,
      "learning_rate": 8.896328213244592e-07,
      "loss": 0.669,
      "step": 28215
    },
    {
      "epoch": 0.8130006338961563,
      "grad_norm": 0.7073893547058105,
      "learning_rate": 8.893671600609838e-07,
      "loss": 0.668,
      "step": 28216
    },
    {
      "epoch": 0.8130294473578056,
      "grad_norm": 0.711585283279419,
      "learning_rate": 8.891015345970111e-07,
      "loss": 0.6609,
      "step": 28217
    },
    {
      "epoch": 0.8130582608194549,
      "grad_norm": 0.7270738482475281,
      "learning_rate": 8.888359449348555e-07,
      "loss": 0.697,
      "step": 28218
    },
    {
      "epoch": 0.8130870742811042,
      "grad_norm": 0.7489407658576965,
      "learning_rate": 8.885703910768284e-07,
      "loss": 0.6644,
      "step": 28219
    },
    {
      "epoch": 0.8131158877427535,
      "grad_norm": 0.7202349305152893,
      "learning_rate": 8.883048730252447e-07,
      "loss": 0.6714,
      "step": 28220
    },
    {
      "epoch": 0.8131447012044027,
      "grad_norm": 0.7232041358947754,
      "learning_rate": 8.880393907824136e-07,
      "loss": 0.6846,
      "step": 28221
    },
    {
      "epoch": 0.813173514666052,
      "grad_norm": 0.710969865322113,
      "learning_rate": 8.877739443506494e-07,
      "loss": 0.6573,
      "step": 28222
    },
    {
      "epoch": 0.8132023281277012,
      "grad_norm": 0.7294251322746277,
      "learning_rate": 8.875085337322631e-07,
      "loss": 0.6906,
      "step": 28223
    },
    {
      "epoch": 0.8132311415893505,
      "grad_norm": 0.7105085253715515,
      "learning_rate": 8.872431589295676e-07,
      "loss": 0.6777,
      "step": 28224
    },
    {
      "epoch": 0.8132599550509998,
      "grad_norm": 0.7377647757530212,
      "learning_rate": 8.86977819944872e-07,
      "loss": 0.6751,
      "step": 28225
    },
    {
      "epoch": 0.8132887685126491,
      "grad_norm": 0.7164813876152039,
      "learning_rate": 8.867125167804896e-07,
      "loss": 0.6785,
      "step": 28226
    },
    {
      "epoch": 0.8133175819742984,
      "grad_norm": 0.7199100852012634,
      "learning_rate": 8.864472494387278e-07,
      "loss": 0.6888,
      "step": 28227
    },
    {
      "epoch": 0.8133463954359477,
      "grad_norm": 0.7221864461898804,
      "learning_rate": 8.861820179218994e-07,
      "loss": 0.683,
      "step": 28228
    },
    {
      "epoch": 0.813375208897597,
      "grad_norm": 0.7194293737411499,
      "learning_rate": 8.859168222323128e-07,
      "loss": 0.691,
      "step": 28229
    },
    {
      "epoch": 0.8134040223592462,
      "grad_norm": 0.7107384204864502,
      "learning_rate": 8.856516623722789e-07,
      "loss": 0.6723,
      "step": 28230
    },
    {
      "epoch": 0.8134328358208955,
      "grad_norm": 0.7231391072273254,
      "learning_rate": 8.853865383441046e-07,
      "loss": 0.6932,
      "step": 28231
    },
    {
      "epoch": 0.8134616492825448,
      "grad_norm": 0.7332953214645386,
      "learning_rate": 8.851214501501021e-07,
      "loss": 0.6669,
      "step": 28232
    },
    {
      "epoch": 0.8134904627441941,
      "grad_norm": 0.7045085430145264,
      "learning_rate": 8.848563977925756e-07,
      "loss": 0.6711,
      "step": 28233
    },
    {
      "epoch": 0.8135192762058434,
      "grad_norm": 0.7139204740524292,
      "learning_rate": 8.845913812738393e-07,
      "loss": 0.6813,
      "step": 28234
    },
    {
      "epoch": 0.8135480896674927,
      "grad_norm": 0.7253331542015076,
      "learning_rate": 8.843264005961965e-07,
      "loss": 0.693,
      "step": 28235
    },
    {
      "epoch": 0.8135769031291419,
      "grad_norm": 0.7259519696235657,
      "learning_rate": 8.840614557619576e-07,
      "loss": 0.7016,
      "step": 28236
    },
    {
      "epoch": 0.8136057165907912,
      "grad_norm": 0.7279645204544067,
      "learning_rate": 8.837965467734283e-07,
      "loss": 0.6943,
      "step": 28237
    },
    {
      "epoch": 0.8136345300524405,
      "grad_norm": 0.719538152217865,
      "learning_rate": 8.835316736329158e-07,
      "loss": 0.6877,
      "step": 28238
    },
    {
      "epoch": 0.8136633435140898,
      "grad_norm": 0.712272584438324,
      "learning_rate": 8.832668363427294e-07,
      "loss": 0.681,
      "step": 28239
    },
    {
      "epoch": 0.813692156975739,
      "grad_norm": 0.7270583510398865,
      "learning_rate": 8.830020349051721e-07,
      "loss": 0.6764,
      "step": 28240
    },
    {
      "epoch": 0.8137209704373883,
      "grad_norm": 0.7166256308555603,
      "learning_rate": 8.827372693225528e-07,
      "loss": 0.6898,
      "step": 28241
    },
    {
      "epoch": 0.8137497838990376,
      "grad_norm": 0.7349175214767456,
      "learning_rate": 8.824725395971745e-07,
      "loss": 0.6984,
      "step": 28242
    },
    {
      "epoch": 0.8137785973606869,
      "grad_norm": 0.7078430652618408,
      "learning_rate": 8.82207845731346e-07,
      "loss": 0.6723,
      "step": 28243
    },
    {
      "epoch": 0.8138074108223362,
      "grad_norm": 0.710963249206543,
      "learning_rate": 8.819431877273694e-07,
      "loss": 0.6958,
      "step": 28244
    },
    {
      "epoch": 0.8138362242839855,
      "grad_norm": 0.739587128162384,
      "learning_rate": 8.816785655875515e-07,
      "loss": 0.6821,
      "step": 28245
    },
    {
      "epoch": 0.8138650377456348,
      "grad_norm": 0.7227058410644531,
      "learning_rate": 8.814139793141962e-07,
      "loss": 0.6816,
      "step": 28246
    },
    {
      "epoch": 0.8138938512072841,
      "grad_norm": 0.7495362162590027,
      "learning_rate": 8.81149428909609e-07,
      "loss": 0.7114,
      "step": 28247
    },
    {
      "epoch": 0.8139226646689334,
      "grad_norm": 0.7258597016334534,
      "learning_rate": 8.808849143760923e-07,
      "loss": 0.6787,
      "step": 28248
    },
    {
      "epoch": 0.8139514781305827,
      "grad_norm": 0.7164062857627869,
      "learning_rate": 8.806204357159514e-07,
      "loss": 0.6743,
      "step": 28249
    },
    {
      "epoch": 0.8139802915922318,
      "grad_norm": 0.72278892993927,
      "learning_rate": 8.80355992931487e-07,
      "loss": 0.7003,
      "step": 28250
    },
    {
      "epoch": 0.8140091050538811,
      "grad_norm": 0.7267930507659912,
      "learning_rate": 8.800915860250053e-07,
      "loss": 0.6772,
      "step": 28251
    },
    {
      "epoch": 0.8140379185155304,
      "grad_norm": 0.7212443351745605,
      "learning_rate": 8.798272149988063e-07,
      "loss": 0.691,
      "step": 28252
    },
    {
      "epoch": 0.8140667319771797,
      "grad_norm": 0.7107009291648865,
      "learning_rate": 8.795628798551948e-07,
      "loss": 0.6784,
      "step": 28253
    },
    {
      "epoch": 0.814095545438829,
      "grad_norm": 0.7089710831642151,
      "learning_rate": 8.792985805964705e-07,
      "loss": 0.6549,
      "step": 28254
    },
    {
      "epoch": 0.8141243589004783,
      "grad_norm": 0.7045729160308838,
      "learning_rate": 8.790343172249366e-07,
      "loss": 0.6792,
      "step": 28255
    },
    {
      "epoch": 0.8141531723621276,
      "grad_norm": 0.7326443195343018,
      "learning_rate": 8.787700897428952e-07,
      "loss": 0.6862,
      "step": 28256
    },
    {
      "epoch": 0.8141819858237769,
      "grad_norm": 0.7096688747406006,
      "learning_rate": 8.785058981526457e-07,
      "loss": 0.6757,
      "step": 28257
    },
    {
      "epoch": 0.8142107992854262,
      "grad_norm": 0.7061982154846191,
      "learning_rate": 8.782417424564893e-07,
      "loss": 0.6887,
      "step": 28258
    },
    {
      "epoch": 0.8142396127470755,
      "grad_norm": 0.7170839309692383,
      "learning_rate": 8.779776226567277e-07,
      "loss": 0.6911,
      "step": 28259
    },
    {
      "epoch": 0.8142684262087247,
      "grad_norm": 0.7203729152679443,
      "learning_rate": 8.777135387556612e-07,
      "loss": 0.6845,
      "step": 28260
    },
    {
      "epoch": 0.814297239670374,
      "grad_norm": 0.7364163398742676,
      "learning_rate": 8.77449490755588e-07,
      "loss": 0.665,
      "step": 28261
    },
    {
      "epoch": 0.8143260531320233,
      "grad_norm": 0.7434654831886292,
      "learning_rate": 8.771854786588096e-07,
      "loss": 0.6942,
      "step": 28262
    },
    {
      "epoch": 0.8143548665936725,
      "grad_norm": 0.7343189716339111,
      "learning_rate": 8.769215024676237e-07,
      "loss": 0.6839,
      "step": 28263
    },
    {
      "epoch": 0.8143836800553218,
      "grad_norm": 0.7190622091293335,
      "learning_rate": 8.76657562184331e-07,
      "loss": 0.6772,
      "step": 28264
    },
    {
      "epoch": 0.8144124935169711,
      "grad_norm": 0.7224199771881104,
      "learning_rate": 8.763936578112275e-07,
      "loss": 0.6981,
      "step": 28265
    },
    {
      "epoch": 0.8144413069786204,
      "grad_norm": 0.7107733488082886,
      "learning_rate": 8.761297893506149e-07,
      "loss": 0.6695,
      "step": 28266
    },
    {
      "epoch": 0.8144701204402697,
      "grad_norm": 0.7371816039085388,
      "learning_rate": 8.758659568047878e-07,
      "loss": 0.6847,
      "step": 28267
    },
    {
      "epoch": 0.814498933901919,
      "grad_norm": 0.7280049324035645,
      "learning_rate": 8.756021601760473e-07,
      "loss": 0.7107,
      "step": 28268
    },
    {
      "epoch": 0.8145277473635683,
      "grad_norm": 0.71306312084198,
      "learning_rate": 8.753383994666864e-07,
      "loss": 0.677,
      "step": 28269
    },
    {
      "epoch": 0.8145565608252175,
      "grad_norm": 0.7243507504463196,
      "learning_rate": 8.750746746790073e-07,
      "loss": 0.6602,
      "step": 28270
    },
    {
      "epoch": 0.8145853742868668,
      "grad_norm": 0.7169332504272461,
      "learning_rate": 8.748109858153031e-07,
      "loss": 0.6873,
      "step": 28271
    },
    {
      "epoch": 0.8146141877485161,
      "grad_norm": 0.7257551550865173,
      "learning_rate": 8.745473328778725e-07,
      "loss": 0.6974,
      "step": 28272
    },
    {
      "epoch": 0.8146430012101654,
      "grad_norm": 0.7236080765724182,
      "learning_rate": 8.742837158690098e-07,
      "loss": 0.6674,
      "step": 28273
    },
    {
      "epoch": 0.8146718146718147,
      "grad_norm": 0.7372819781303406,
      "learning_rate": 8.740201347910133e-07,
      "loss": 0.6835,
      "step": 28274
    },
    {
      "epoch": 0.814700628133464,
      "grad_norm": 0.7583149075508118,
      "learning_rate": 8.737565896461753e-07,
      "loss": 0.6531,
      "step": 28275
    },
    {
      "epoch": 0.8147294415951133,
      "grad_norm": 0.7353094816207886,
      "learning_rate": 8.734930804367925e-07,
      "loss": 0.6507,
      "step": 28276
    },
    {
      "epoch": 0.8147582550567625,
      "grad_norm": 0.7258081436157227,
      "learning_rate": 8.732296071651614e-07,
      "loss": 0.7107,
      "step": 28277
    },
    {
      "epoch": 0.8147870685184118,
      "grad_norm": 0.735349178314209,
      "learning_rate": 8.72966169833574e-07,
      "loss": 0.6837,
      "step": 28278
    },
    {
      "epoch": 0.814815881980061,
      "grad_norm": 0.726773738861084,
      "learning_rate": 8.727027684443268e-07,
      "loss": 0.6873,
      "step": 28279
    },
    {
      "epoch": 0.8148446954417103,
      "grad_norm": 0.6985225677490234,
      "learning_rate": 8.724394029997119e-07,
      "loss": 0.6567,
      "step": 28280
    },
    {
      "epoch": 0.8148735089033596,
      "grad_norm": 0.7176235914230347,
      "learning_rate": 8.721760735020235e-07,
      "loss": 0.6911,
      "step": 28281
    },
    {
      "epoch": 0.8149023223650089,
      "grad_norm": 0.725993275642395,
      "learning_rate": 8.719127799535548e-07,
      "loss": 0.6868,
      "step": 28282
    },
    {
      "epoch": 0.8149311358266582,
      "grad_norm": 0.7128164172172546,
      "learning_rate": 8.716495223566007e-07,
      "loss": 0.6919,
      "step": 28283
    },
    {
      "epoch": 0.8149599492883075,
      "grad_norm": 0.7200928330421448,
      "learning_rate": 8.713863007134515e-07,
      "loss": 0.6681,
      "step": 28284
    },
    {
      "epoch": 0.8149887627499568,
      "grad_norm": 0.7201747298240662,
      "learning_rate": 8.711231150264016e-07,
      "loss": 0.6966,
      "step": 28285
    },
    {
      "epoch": 0.8150175762116061,
      "grad_norm": 0.7165990471839905,
      "learning_rate": 8.708599652977406e-07,
      "loss": 0.6835,
      "step": 28286
    },
    {
      "epoch": 0.8150463896732554,
      "grad_norm": 0.7447152137756348,
      "learning_rate": 8.705968515297631e-07,
      "loss": 0.6908,
      "step": 28287
    },
    {
      "epoch": 0.8150752031349047,
      "grad_norm": 0.7378423810005188,
      "learning_rate": 8.703337737247575e-07,
      "loss": 0.6792,
      "step": 28288
    },
    {
      "epoch": 0.815104016596554,
      "grad_norm": 0.7330437898635864,
      "learning_rate": 8.700707318850182e-07,
      "loss": 0.6813,
      "step": 28289
    },
    {
      "epoch": 0.8151328300582031,
      "grad_norm": 0.7126864194869995,
      "learning_rate": 8.69807726012833e-07,
      "loss": 0.6819,
      "step": 28290
    },
    {
      "epoch": 0.8151616435198524,
      "grad_norm": 0.7318047285079956,
      "learning_rate": 8.695447561104953e-07,
      "loss": 0.6813,
      "step": 28291
    },
    {
      "epoch": 0.8151904569815017,
      "grad_norm": 0.6869613528251648,
      "learning_rate": 8.692818221802923e-07,
      "loss": 0.6629,
      "step": 28292
    },
    {
      "epoch": 0.815219270443151,
      "grad_norm": 0.7151995897293091,
      "learning_rate": 8.690189242245156e-07,
      "loss": 0.6639,
      "step": 28293
    },
    {
      "epoch": 0.8152480839048003,
      "grad_norm": 0.7279577255249023,
      "learning_rate": 8.687560622454544e-07,
      "loss": 0.6767,
      "step": 28294
    },
    {
      "epoch": 0.8152768973664496,
      "grad_norm": 0.7257125377655029,
      "learning_rate": 8.684932362453996e-07,
      "loss": 0.6945,
      "step": 28295
    },
    {
      "epoch": 0.8153057108280989,
      "grad_norm": 0.7225931286811829,
      "learning_rate": 8.682304462266372e-07,
      "loss": 0.677,
      "step": 28296
    },
    {
      "epoch": 0.8153345242897482,
      "grad_norm": 0.7079830765724182,
      "learning_rate": 8.679676921914576e-07,
      "loss": 0.6718,
      "step": 28297
    },
    {
      "epoch": 0.8153633377513975,
      "grad_norm": 0.7064787745475769,
      "learning_rate": 8.677049741421506e-07,
      "loss": 0.6795,
      "step": 28298
    },
    {
      "epoch": 0.8153921512130468,
      "grad_norm": 0.7154862284660339,
      "learning_rate": 8.674422920810005e-07,
      "loss": 0.6542,
      "step": 28299
    },
    {
      "epoch": 0.815420964674696,
      "grad_norm": 0.7424437403678894,
      "learning_rate": 8.671796460102982e-07,
      "loss": 0.7001,
      "step": 28300
    },
    {
      "epoch": 0.8154497781363453,
      "grad_norm": 0.7371419072151184,
      "learning_rate": 8.669170359323292e-07,
      "loss": 0.6858,
      "step": 28301
    },
    {
      "epoch": 0.8154785915979946,
      "grad_norm": 0.7102106809616089,
      "learning_rate": 8.666544618493822e-07,
      "loss": 0.7002,
      "step": 28302
    },
    {
      "epoch": 0.8155074050596439,
      "grad_norm": 0.7168552279472351,
      "learning_rate": 8.66391923763742e-07,
      "loss": 0.6954,
      "step": 28303
    },
    {
      "epoch": 0.8155362185212931,
      "grad_norm": 0.7239190936088562,
      "learning_rate": 8.661294216776966e-07,
      "loss": 0.7004,
      "step": 28304
    },
    {
      "epoch": 0.8155650319829424,
      "grad_norm": 0.7679945826530457,
      "learning_rate": 8.658669555935295e-07,
      "loss": 0.6911,
      "step": 28305
    },
    {
      "epoch": 0.8155938454445917,
      "grad_norm": 0.7203570604324341,
      "learning_rate": 8.656045255135314e-07,
      "loss": 0.6925,
      "step": 28306
    },
    {
      "epoch": 0.815622658906241,
      "grad_norm": 0.7786269783973694,
      "learning_rate": 8.653421314399835e-07,
      "loss": 0.7093,
      "step": 28307
    },
    {
      "epoch": 0.8156514723678903,
      "grad_norm": 0.7167369723320007,
      "learning_rate": 8.650797733751736e-07,
      "loss": 0.6781,
      "step": 28308
    },
    {
      "epoch": 0.8156802858295396,
      "grad_norm": 0.7036350965499878,
      "learning_rate": 8.648174513213847e-07,
      "loss": 0.6558,
      "step": 28309
    },
    {
      "epoch": 0.8157090992911888,
      "grad_norm": 0.7431222796440125,
      "learning_rate": 8.645551652809025e-07,
      "loss": 0.6743,
      "step": 28310
    },
    {
      "epoch": 0.8157379127528381,
      "grad_norm": 0.7192260026931763,
      "learning_rate": 8.642929152560104e-07,
      "loss": 0.6788,
      "step": 28311
    },
    {
      "epoch": 0.8157667262144874,
      "grad_norm": 0.7135221362113953,
      "learning_rate": 8.64030701248994e-07,
      "loss": 0.6953,
      "step": 28312
    },
    {
      "epoch": 0.8157955396761367,
      "grad_norm": 0.7091834545135498,
      "learning_rate": 8.63768523262134e-07,
      "loss": 0.675,
      "step": 28313
    },
    {
      "epoch": 0.815824353137786,
      "grad_norm": 0.7442996501922607,
      "learning_rate": 8.635063812977157e-07,
      "loss": 0.6937,
      "step": 28314
    },
    {
      "epoch": 0.8158531665994353,
      "grad_norm": 0.7104876637458801,
      "learning_rate": 8.632442753580234e-07,
      "loss": 0.6758,
      "step": 28315
    },
    {
      "epoch": 0.8158819800610846,
      "grad_norm": 0.7022225260734558,
      "learning_rate": 8.629822054453369e-07,
      "loss": 0.6834,
      "step": 28316
    },
    {
      "epoch": 0.8159107935227338,
      "grad_norm": 0.7135969996452332,
      "learning_rate": 8.627201715619394e-07,
      "loss": 0.6615,
      "step": 28317
    },
    {
      "epoch": 0.8159396069843831,
      "grad_norm": 0.7431056499481201,
      "learning_rate": 8.62458173710115e-07,
      "loss": 0.7113,
      "step": 28318
    },
    {
      "epoch": 0.8159684204460323,
      "grad_norm": 0.7081702947616577,
      "learning_rate": 8.621962118921428e-07,
      "loss": 0.6928,
      "step": 28319
    },
    {
      "epoch": 0.8159972339076816,
      "grad_norm": 0.730461597442627,
      "learning_rate": 8.619342861103047e-07,
      "loss": 0.6648,
      "step": 28320
    },
    {
      "epoch": 0.8160260473693309,
      "grad_norm": 0.7302228808403015,
      "learning_rate": 8.616723963668843e-07,
      "loss": 0.699,
      "step": 28321
    },
    {
      "epoch": 0.8160548608309802,
      "grad_norm": 0.7284948825836182,
      "learning_rate": 8.614105426641589e-07,
      "loss": 0.6552,
      "step": 28322
    },
    {
      "epoch": 0.8160836742926295,
      "grad_norm": 0.8267379999160767,
      "learning_rate": 8.61148725004412e-07,
      "loss": 0.6883,
      "step": 28323
    },
    {
      "epoch": 0.8161124877542788,
      "grad_norm": 0.725696861743927,
      "learning_rate": 8.608869433899214e-07,
      "loss": 0.6605,
      "step": 28324
    },
    {
      "epoch": 0.8161413012159281,
      "grad_norm": 0.722454845905304,
      "learning_rate": 8.60625197822969e-07,
      "loss": 0.6759,
      "step": 28325
    },
    {
      "epoch": 0.8161701146775774,
      "grad_norm": 0.712874710559845,
      "learning_rate": 8.603634883058315e-07,
      "loss": 0.6945,
      "step": 28326
    },
    {
      "epoch": 0.8161989281392267,
      "grad_norm": 0.7087808847427368,
      "learning_rate": 8.60101814840792e-07,
      "loss": 0.6894,
      "step": 28327
    },
    {
      "epoch": 0.816227741600876,
      "grad_norm": 0.7166204452514648,
      "learning_rate": 8.598401774301252e-07,
      "loss": 0.676,
      "step": 28328
    },
    {
      "epoch": 0.8162565550625253,
      "grad_norm": 0.7207871675491333,
      "learning_rate": 8.595785760761122e-07,
      "loss": 0.6615,
      "step": 28329
    },
    {
      "epoch": 0.8162853685241745,
      "grad_norm": 0.7305038571357727,
      "learning_rate": 8.593170107810311e-07,
      "loss": 0.688,
      "step": 28330
    },
    {
      "epoch": 0.8163141819858237,
      "grad_norm": 0.738472580909729,
      "learning_rate": 8.590554815471608e-07,
      "loss": 0.6745,
      "step": 28331
    },
    {
      "epoch": 0.816342995447473,
      "grad_norm": 0.7007125020027161,
      "learning_rate": 8.587939883767765e-07,
      "loss": 0.6734,
      "step": 28332
    },
    {
      "epoch": 0.8163718089091223,
      "grad_norm": 0.7329251766204834,
      "learning_rate": 8.585325312721582e-07,
      "loss": 0.6882,
      "step": 28333
    },
    {
      "epoch": 0.8164006223707716,
      "grad_norm": 0.7210114002227783,
      "learning_rate": 8.582711102355801e-07,
      "loss": 0.6958,
      "step": 28334
    },
    {
      "epoch": 0.8164294358324209,
      "grad_norm": 0.7279312014579773,
      "learning_rate": 8.580097252693204e-07,
      "loss": 0.6599,
      "step": 28335
    },
    {
      "epoch": 0.8164582492940702,
      "grad_norm": 0.6939286589622498,
      "learning_rate": 8.577483763756572e-07,
      "loss": 0.6542,
      "step": 28336
    },
    {
      "epoch": 0.8164870627557195,
      "grad_norm": 0.729585587978363,
      "learning_rate": 8.574870635568633e-07,
      "loss": 0.6952,
      "step": 28337
    },
    {
      "epoch": 0.8165158762173688,
      "grad_norm": 0.7178241610527039,
      "learning_rate": 8.572257868152173e-07,
      "loss": 0.6969,
      "step": 28338
    },
    {
      "epoch": 0.816544689679018,
      "grad_norm": 0.7048150897026062,
      "learning_rate": 8.569645461529924e-07,
      "loss": 0.6965,
      "step": 28339
    },
    {
      "epoch": 0.8165735031406673,
      "grad_norm": 0.7167966961860657,
      "learning_rate": 8.567033415724663e-07,
      "loss": 0.6455,
      "step": 28340
    },
    {
      "epoch": 0.8166023166023166,
      "grad_norm": 0.7456853985786438,
      "learning_rate": 8.564421730759092e-07,
      "loss": 0.7,
      "step": 28341
    },
    {
      "epoch": 0.8166311300639659,
      "grad_norm": 0.7131472826004028,
      "learning_rate": 8.561810406656018e-07,
      "loss": 0.6991,
      "step": 28342
    },
    {
      "epoch": 0.8166599435256152,
      "grad_norm": 0.7341444492340088,
      "learning_rate": 8.559199443438138e-07,
      "loss": 0.6638,
      "step": 28343
    },
    {
      "epoch": 0.8166887569872644,
      "grad_norm": 0.7217327952384949,
      "learning_rate": 8.556588841128216e-07,
      "loss": 0.6926,
      "step": 28344
    },
    {
      "epoch": 0.8167175704489137,
      "grad_norm": 0.7275899052619934,
      "learning_rate": 8.553978599748963e-07,
      "loss": 0.6678,
      "step": 28345
    },
    {
      "epoch": 0.816746383910563,
      "grad_norm": 0.7205598950386047,
      "learning_rate": 8.551368719323139e-07,
      "loss": 0.6575,
      "step": 28346
    },
    {
      "epoch": 0.8167751973722123,
      "grad_norm": 0.7527217864990234,
      "learning_rate": 8.548759199873452e-07,
      "loss": 0.6969,
      "step": 28347
    },
    {
      "epoch": 0.8168040108338616,
      "grad_norm": 0.7326048612594604,
      "learning_rate": 8.546150041422646e-07,
      "loss": 0.6913,
      "step": 28348
    },
    {
      "epoch": 0.8168328242955109,
      "grad_norm": 0.705839991569519,
      "learning_rate": 8.54354124399342e-07,
      "loss": 0.6698,
      "step": 28349
    },
    {
      "epoch": 0.8168616377571601,
      "grad_norm": 0.7306333780288696,
      "learning_rate": 8.540932807608521e-07,
      "loss": 0.6965,
      "step": 28350
    },
    {
      "epoch": 0.8168904512188094,
      "grad_norm": 0.723991334438324,
      "learning_rate": 8.538324732290642e-07,
      "loss": 0.7008,
      "step": 28351
    },
    {
      "epoch": 0.8169192646804587,
      "grad_norm": 0.7258735299110413,
      "learning_rate": 8.535717018062512e-07,
      "loss": 0.6819,
      "step": 28352
    },
    {
      "epoch": 0.816948078142108,
      "grad_norm": 0.732374906539917,
      "learning_rate": 8.533109664946831e-07,
      "loss": 0.7055,
      "step": 28353
    },
    {
      "epoch": 0.8169768916037573,
      "grad_norm": 0.708350658416748,
      "learning_rate": 8.530502672966329e-07,
      "loss": 0.6735,
      "step": 28354
    },
    {
      "epoch": 0.8170057050654066,
      "grad_norm": 0.723928689956665,
      "learning_rate": 8.527896042143679e-07,
      "loss": 0.6878,
      "step": 28355
    },
    {
      "epoch": 0.8170345185270559,
      "grad_norm": 0.7426628470420837,
      "learning_rate": 8.525289772501616e-07,
      "loss": 0.6929,
      "step": 28356
    },
    {
      "epoch": 0.8170633319887052,
      "grad_norm": 0.7431721091270447,
      "learning_rate": 8.522683864062803e-07,
      "loss": 0.7122,
      "step": 28357
    },
    {
      "epoch": 0.8170921454503544,
      "grad_norm": 0.7265186309814453,
      "learning_rate": 8.52007831684995e-07,
      "loss": 0.6746,
      "step": 28358
    },
    {
      "epoch": 0.8171209589120036,
      "grad_norm": 0.7245837450027466,
      "learning_rate": 8.517473130885767e-07,
      "loss": 0.685,
      "step": 28359
    },
    {
      "epoch": 0.8171497723736529,
      "grad_norm": 0.7231097221374512,
      "learning_rate": 8.514868306192909e-07,
      "loss": 0.6774,
      "step": 28360
    },
    {
      "epoch": 0.8171785858353022,
      "grad_norm": 0.7732059955596924,
      "learning_rate": 8.512263842794094e-07,
      "loss": 0.6871,
      "step": 28361
    },
    {
      "epoch": 0.8172073992969515,
      "grad_norm": 0.7090074419975281,
      "learning_rate": 8.509659740711973e-07,
      "loss": 0.6991,
      "step": 28362
    },
    {
      "epoch": 0.8172362127586008,
      "grad_norm": 0.7252430319786072,
      "learning_rate": 8.507055999969254e-07,
      "loss": 0.7032,
      "step": 28363
    },
    {
      "epoch": 0.8172650262202501,
      "grad_norm": 0.7482470870018005,
      "learning_rate": 8.504452620588589e-07,
      "loss": 0.6871,
      "step": 28364
    },
    {
      "epoch": 0.8172938396818994,
      "grad_norm": 0.7322761416435242,
      "learning_rate": 8.50184960259266e-07,
      "loss": 0.6759,
      "step": 28365
    },
    {
      "epoch": 0.8173226531435487,
      "grad_norm": 0.7437650561332703,
      "learning_rate": 8.499246946004136e-07,
      "loss": 0.6771,
      "step": 28366
    },
    {
      "epoch": 0.817351466605198,
      "grad_norm": 0.720405101776123,
      "learning_rate": 8.496644650845698e-07,
      "loss": 0.7032,
      "step": 28367
    },
    {
      "epoch": 0.8173802800668473,
      "grad_norm": 0.7106844186782837,
      "learning_rate": 8.494042717139989e-07,
      "loss": 0.6821,
      "step": 28368
    },
    {
      "epoch": 0.8174090935284966,
      "grad_norm": 0.7154229879379272,
      "learning_rate": 8.491441144909684e-07,
      "loss": 0.6737,
      "step": 28369
    },
    {
      "epoch": 0.8174379069901458,
      "grad_norm": 0.73077791929245,
      "learning_rate": 8.488839934177424e-07,
      "loss": 0.6549,
      "step": 28370
    },
    {
      "epoch": 0.817466720451795,
      "grad_norm": 0.720291793346405,
      "learning_rate": 8.486239084965885e-07,
      "loss": 0.6682,
      "step": 28371
    },
    {
      "epoch": 0.8174955339134443,
      "grad_norm": 0.747504472732544,
      "learning_rate": 8.483638597297695e-07,
      "loss": 0.6806,
      "step": 28372
    },
    {
      "epoch": 0.8175243473750936,
      "grad_norm": 0.7196568846702576,
      "learning_rate": 8.481038471195507e-07,
      "loss": 0.6971,
      "step": 28373
    },
    {
      "epoch": 0.8175531608367429,
      "grad_norm": 0.7055001854896545,
      "learning_rate": 8.47843870668198e-07,
      "loss": 0.6828,
      "step": 28374
    },
    {
      "epoch": 0.8175819742983922,
      "grad_norm": 0.7212250232696533,
      "learning_rate": 8.475839303779737e-07,
      "loss": 0.6687,
      "step": 28375
    },
    {
      "epoch": 0.8176107877600415,
      "grad_norm": 0.7194840312004089,
      "learning_rate": 8.473240262511439e-07,
      "loss": 0.6904,
      "step": 28376
    },
    {
      "epoch": 0.8176396012216908,
      "grad_norm": 0.724033534526825,
      "learning_rate": 8.470641582899691e-07,
      "loss": 0.7078,
      "step": 28377
    },
    {
      "epoch": 0.8176684146833401,
      "grad_norm": 0.7339673042297363,
      "learning_rate": 8.468043264967141e-07,
      "loss": 0.6917,
      "step": 28378
    },
    {
      "epoch": 0.8176972281449894,
      "grad_norm": 0.7225167751312256,
      "learning_rate": 8.46544530873642e-07,
      "loss": 0.673,
      "step": 28379
    },
    {
      "epoch": 0.8177260416066386,
      "grad_norm": 0.7368442416191101,
      "learning_rate": 8.462847714230161e-07,
      "loss": 0.6791,
      "step": 28380
    },
    {
      "epoch": 0.8177548550682879,
      "grad_norm": 0.7305612564086914,
      "learning_rate": 8.460250481470966e-07,
      "loss": 0.6718,
      "step": 28381
    },
    {
      "epoch": 0.8177836685299372,
      "grad_norm": 0.7127610445022583,
      "learning_rate": 8.457653610481475e-07,
      "loss": 0.6547,
      "step": 28382
    },
    {
      "epoch": 0.8178124819915865,
      "grad_norm": 0.7067694067955017,
      "learning_rate": 8.455057101284287e-07,
      "loss": 0.6719,
      "step": 28383
    },
    {
      "epoch": 0.8178412954532358,
      "grad_norm": 0.7326192855834961,
      "learning_rate": 8.452460953902026e-07,
      "loss": 0.6814,
      "step": 28384
    },
    {
      "epoch": 0.817870108914885,
      "grad_norm": 0.7258254289627075,
      "learning_rate": 8.44986516835729e-07,
      "loss": 0.6943,
      "step": 28385
    },
    {
      "epoch": 0.8178989223765343,
      "grad_norm": 0.7271794676780701,
      "learning_rate": 8.447269744672703e-07,
      "loss": 0.6872,
      "step": 28386
    },
    {
      "epoch": 0.8179277358381836,
      "grad_norm": 0.7200278043746948,
      "learning_rate": 8.444674682870852e-07,
      "loss": 0.6977,
      "step": 28387
    },
    {
      "epoch": 0.8179565492998329,
      "grad_norm": 0.7197072505950928,
      "learning_rate": 8.442079982974355e-07,
      "loss": 0.6543,
      "step": 28388
    },
    {
      "epoch": 0.8179853627614821,
      "grad_norm": 0.7218455076217651,
      "learning_rate": 8.439485645005768e-07,
      "loss": 0.6653,
      "step": 28389
    },
    {
      "epoch": 0.8180141762231314,
      "grad_norm": 0.7236260175704956,
      "learning_rate": 8.436891668987745e-07,
      "loss": 0.6865,
      "step": 28390
    },
    {
      "epoch": 0.8180429896847807,
      "grad_norm": 0.6974601745605469,
      "learning_rate": 8.434298054942836e-07,
      "loss": 0.6601,
      "step": 28391
    },
    {
      "epoch": 0.81807180314643,
      "grad_norm": 0.802933931350708,
      "learning_rate": 8.431704802893653e-07,
      "loss": 0.6831,
      "step": 28392
    },
    {
      "epoch": 0.8181006166080793,
      "grad_norm": 0.7496660351753235,
      "learning_rate": 8.429111912862753e-07,
      "loss": 0.6855,
      "step": 28393
    },
    {
      "epoch": 0.8181294300697286,
      "grad_norm": 0.7434720396995544,
      "learning_rate": 8.426519384872733e-07,
      "loss": 0.6999,
      "step": 28394
    },
    {
      "epoch": 0.8181582435313779,
      "grad_norm": 0.7344042658805847,
      "learning_rate": 8.423927218946188e-07,
      "loss": 0.6827,
      "step": 28395
    },
    {
      "epoch": 0.8181870569930272,
      "grad_norm": 0.7155306935310364,
      "learning_rate": 8.421335415105658e-07,
      "loss": 0.6648,
      "step": 28396
    },
    {
      "epoch": 0.8182158704546765,
      "grad_norm": 0.7109860181808472,
      "learning_rate": 8.418743973373749e-07,
      "loss": 0.667,
      "step": 28397
    },
    {
      "epoch": 0.8182446839163257,
      "grad_norm": 0.6984483003616333,
      "learning_rate": 8.416152893772999e-07,
      "loss": 0.6715,
      "step": 28398
    },
    {
      "epoch": 0.818273497377975,
      "grad_norm": 0.7433425784111023,
      "learning_rate": 8.413562176326001e-07,
      "loss": 0.668,
      "step": 28399
    },
    {
      "epoch": 0.8183023108396242,
      "grad_norm": 0.7412384152412415,
      "learning_rate": 8.410971821055286e-07,
      "loss": 0.6771,
      "step": 28400
    },
    {
      "epoch": 0.8183311243012735,
      "grad_norm": 0.7259994745254517,
      "learning_rate": 8.408381827983453e-07,
      "loss": 0.6885,
      "step": 28401
    },
    {
      "epoch": 0.8183599377629228,
      "grad_norm": 0.7250427603721619,
      "learning_rate": 8.405792197133023e-07,
      "loss": 0.6866,
      "step": 28402
    },
    {
      "epoch": 0.8183887512245721,
      "grad_norm": 0.7265950441360474,
      "learning_rate": 8.403202928526577e-07,
      "loss": 0.6639,
      "step": 28403
    },
    {
      "epoch": 0.8184175646862214,
      "grad_norm": 0.7494430541992188,
      "learning_rate": 8.400614022186643e-07,
      "loss": 0.7028,
      "step": 28404
    },
    {
      "epoch": 0.8184463781478707,
      "grad_norm": 0.7530084252357483,
      "learning_rate": 8.398025478135791e-07,
      "loss": 0.7045,
      "step": 28405
    },
    {
      "epoch": 0.81847519160952,
      "grad_norm": 0.726852536201477,
      "learning_rate": 8.39543729639653e-07,
      "loss": 0.6658,
      "step": 28406
    },
    {
      "epoch": 0.8185040050711693,
      "grad_norm": 0.7415156364440918,
      "learning_rate": 8.392849476991438e-07,
      "loss": 0.6762,
      "step": 28407
    },
    {
      "epoch": 0.8185328185328186,
      "grad_norm": 0.731232225894928,
      "learning_rate": 8.390262019943024e-07,
      "loss": 0.7157,
      "step": 28408
    },
    {
      "epoch": 0.8185616319944679,
      "grad_norm": 0.7139700055122375,
      "learning_rate": 8.387674925273843e-07,
      "loss": 0.6737,
      "step": 28409
    },
    {
      "epoch": 0.8185904454561171,
      "grad_norm": 0.737983763217926,
      "learning_rate": 8.385088193006408e-07,
      "loss": 0.7051,
      "step": 28410
    },
    {
      "epoch": 0.8186192589177664,
      "grad_norm": 0.7093951106071472,
      "learning_rate": 8.382501823163253e-07,
      "loss": 0.6777,
      "step": 28411
    },
    {
      "epoch": 0.8186480723794156,
      "grad_norm": 0.7395523190498352,
      "learning_rate": 8.379915815766903e-07,
      "loss": 0.6747,
      "step": 28412
    },
    {
      "epoch": 0.8186768858410649,
      "grad_norm": 0.7125381827354431,
      "learning_rate": 8.377330170839892e-07,
      "loss": 0.6884,
      "step": 28413
    },
    {
      "epoch": 0.8187056993027142,
      "grad_norm": 0.711098849773407,
      "learning_rate": 8.374744888404717e-07,
      "loss": 0.688,
      "step": 28414
    },
    {
      "epoch": 0.8187345127643635,
      "grad_norm": 0.740229070186615,
      "learning_rate": 8.372159968483917e-07,
      "loss": 0.6633,
      "step": 28415
    },
    {
      "epoch": 0.8187633262260128,
      "grad_norm": 0.7218957543373108,
      "learning_rate": 8.369575411099984e-07,
      "loss": 0.662,
      "step": 28416
    },
    {
      "epoch": 0.8187921396876621,
      "grad_norm": 0.7203813195228577,
      "learning_rate": 8.366991216275427e-07,
      "loss": 0.6834,
      "step": 28417
    },
    {
      "epoch": 0.8188209531493114,
      "grad_norm": 0.7449474334716797,
      "learning_rate": 8.364407384032775e-07,
      "loss": 0.6914,
      "step": 28418
    },
    {
      "epoch": 0.8188497666109606,
      "grad_norm": 0.7481963038444519,
      "learning_rate": 8.361823914394501e-07,
      "loss": 0.6866,
      "step": 28419
    },
    {
      "epoch": 0.8188785800726099,
      "grad_norm": 0.734259843826294,
      "learning_rate": 8.359240807383129e-07,
      "loss": 0.6918,
      "step": 28420
    },
    {
      "epoch": 0.8189073935342592,
      "grad_norm": 0.7251520156860352,
      "learning_rate": 8.356658063021128e-07,
      "loss": 0.6757,
      "step": 28421
    },
    {
      "epoch": 0.8189362069959085,
      "grad_norm": 0.7169475555419922,
      "learning_rate": 8.354075681331026e-07,
      "loss": 0.6711,
      "step": 28422
    },
    {
      "epoch": 0.8189650204575578,
      "grad_norm": 0.6954594850540161,
      "learning_rate": 8.35149366233528e-07,
      "loss": 0.6669,
      "step": 28423
    },
    {
      "epoch": 0.8189938339192071,
      "grad_norm": 0.6961492300033569,
      "learning_rate": 8.348912006056392e-07,
      "loss": 0.6809,
      "step": 28424
    },
    {
      "epoch": 0.8190226473808563,
      "grad_norm": 0.725389838218689,
      "learning_rate": 8.346330712516842e-07,
      "loss": 0.7102,
      "step": 28425
    },
    {
      "epoch": 0.8190514608425056,
      "grad_norm": 0.7281590700149536,
      "learning_rate": 8.343749781739125e-07,
      "loss": 0.6794,
      "step": 28426
    },
    {
      "epoch": 0.8190802743041549,
      "grad_norm": 0.6951471567153931,
      "learning_rate": 8.341169213745698e-07,
      "loss": 0.6641,
      "step": 28427
    },
    {
      "epoch": 0.8191090877658042,
      "grad_norm": 0.714672863483429,
      "learning_rate": 8.33858900855905e-07,
      "loss": 0.6794,
      "step": 28428
    },
    {
      "epoch": 0.8191379012274534,
      "grad_norm": 0.7290200591087341,
      "learning_rate": 8.336009166201636e-07,
      "loss": 0.6866,
      "step": 28429
    },
    {
      "epoch": 0.8191667146891027,
      "grad_norm": 0.7175307273864746,
      "learning_rate": 8.333429686695948e-07,
      "loss": 0.6921,
      "step": 28430
    },
    {
      "epoch": 0.819195528150752,
      "grad_norm": 0.7360631227493286,
      "learning_rate": 8.330850570064419e-07,
      "loss": 0.6968,
      "step": 28431
    },
    {
      "epoch": 0.8192243416124013,
      "grad_norm": 0.7122015357017517,
      "learning_rate": 8.32827181632953e-07,
      "loss": 0.6708,
      "step": 28432
    },
    {
      "epoch": 0.8192531550740506,
      "grad_norm": 0.7261105179786682,
      "learning_rate": 8.325693425513753e-07,
      "loss": 0.7001,
      "step": 28433
    },
    {
      "epoch": 0.8192819685356999,
      "grad_norm": 0.7033377885818481,
      "learning_rate": 8.323115397639514e-07,
      "loss": 0.6866,
      "step": 28434
    },
    {
      "epoch": 0.8193107819973492,
      "grad_norm": 0.7216832637786865,
      "learning_rate": 8.320537732729289e-07,
      "loss": 0.676,
      "step": 28435
    },
    {
      "epoch": 0.8193395954589985,
      "grad_norm": 0.711875855922699,
      "learning_rate": 8.317960430805499e-07,
      "loss": 0.6707,
      "step": 28436
    },
    {
      "epoch": 0.8193684089206478,
      "grad_norm": 0.7050720453262329,
      "learning_rate": 8.31538349189061e-07,
      "loss": 0.6781,
      "step": 28437
    },
    {
      "epoch": 0.8193972223822971,
      "grad_norm": 0.7207843065261841,
      "learning_rate": 8.312806916007066e-07,
      "loss": 0.71,
      "step": 28438
    },
    {
      "epoch": 0.8194260358439462,
      "grad_norm": 0.7265329360961914,
      "learning_rate": 8.310230703177307e-07,
      "loss": 0.6784,
      "step": 28439
    },
    {
      "epoch": 0.8194548493055955,
      "grad_norm": 0.7357749342918396,
      "learning_rate": 8.307654853423758e-07,
      "loss": 0.681,
      "step": 28440
    },
    {
      "epoch": 0.8194836627672448,
      "grad_norm": 0.7121925950050354,
      "learning_rate": 8.30507936676887e-07,
      "loss": 0.6608,
      "step": 28441
    },
    {
      "epoch": 0.8195124762288941,
      "grad_norm": 0.7375826239585876,
      "learning_rate": 8.302504243235043e-07,
      "loss": 0.6799,
      "step": 28442
    },
    {
      "epoch": 0.8195412896905434,
      "grad_norm": 0.744405210018158,
      "learning_rate": 8.299929482844737e-07,
      "loss": 0.678,
      "step": 28443
    },
    {
      "epoch": 0.8195701031521927,
      "grad_norm": 0.7237051129341125,
      "learning_rate": 8.297355085620351e-07,
      "loss": 0.7067,
      "step": 28444
    },
    {
      "epoch": 0.819598916613842,
      "grad_norm": 0.7285897731781006,
      "learning_rate": 8.294781051584328e-07,
      "loss": 0.6857,
      "step": 28445
    },
    {
      "epoch": 0.8196277300754913,
      "grad_norm": 0.7688995003700256,
      "learning_rate": 8.292207380759055e-07,
      "loss": 0.6942,
      "step": 28446
    },
    {
      "epoch": 0.8196565435371406,
      "grad_norm": 0.7118412256240845,
      "learning_rate": 8.289634073166974e-07,
      "loss": 0.6736,
      "step": 28447
    },
    {
      "epoch": 0.8196853569987899,
      "grad_norm": 0.722992479801178,
      "learning_rate": 8.287061128830465e-07,
      "loss": 0.7174,
      "step": 28448
    },
    {
      "epoch": 0.8197141704604392,
      "grad_norm": 0.7284054756164551,
      "learning_rate": 8.284488547771979e-07,
      "loss": 0.6715,
      "step": 28449
    },
    {
      "epoch": 0.8197429839220884,
      "grad_norm": 0.749897301197052,
      "learning_rate": 8.281916330013889e-07,
      "loss": 0.6797,
      "step": 28450
    },
    {
      "epoch": 0.8197717973837377,
      "grad_norm": 0.711764395236969,
      "learning_rate": 8.27934447557861e-07,
      "loss": 0.6669,
      "step": 28451
    },
    {
      "epoch": 0.8198006108453869,
      "grad_norm": 0.7303224205970764,
      "learning_rate": 8.276772984488529e-07,
      "loss": 0.6769,
      "step": 28452
    },
    {
      "epoch": 0.8198294243070362,
      "grad_norm": 0.7153822183609009,
      "learning_rate": 8.274201856766056e-07,
      "loss": 0.6754,
      "step": 28453
    },
    {
      "epoch": 0.8198582377686855,
      "grad_norm": 0.7226095795631409,
      "learning_rate": 8.271631092433563e-07,
      "loss": 0.663,
      "step": 28454
    },
    {
      "epoch": 0.8198870512303348,
      "grad_norm": 0.7266067862510681,
      "learning_rate": 8.269060691513453e-07,
      "loss": 0.6789,
      "step": 28455
    },
    {
      "epoch": 0.8199158646919841,
      "grad_norm": 0.7009678483009338,
      "learning_rate": 8.266490654028114e-07,
      "loss": 0.6736,
      "step": 28456
    },
    {
      "epoch": 0.8199446781536334,
      "grad_norm": 0.7066066861152649,
      "learning_rate": 8.263920979999918e-07,
      "loss": 0.6779,
      "step": 28457
    },
    {
      "epoch": 0.8199734916152827,
      "grad_norm": 0.7442687153816223,
      "learning_rate": 8.261351669451256e-07,
      "loss": 0.654,
      "step": 28458
    },
    {
      "epoch": 0.820002305076932,
      "grad_norm": 0.7202167510986328,
      "learning_rate": 8.25878272240449e-07,
      "loss": 0.6684,
      "step": 28459
    },
    {
      "epoch": 0.8200311185385812,
      "grad_norm": 0.7314141392707825,
      "learning_rate": 8.256214138882002e-07,
      "loss": 0.6624,
      "step": 28460
    },
    {
      "epoch": 0.8200599320002305,
      "grad_norm": 0.7273057699203491,
      "learning_rate": 8.253645918906156e-07,
      "loss": 0.693,
      "step": 28461
    },
    {
      "epoch": 0.8200887454618798,
      "grad_norm": 0.7252872586250305,
      "learning_rate": 8.251078062499335e-07,
      "loss": 0.6712,
      "step": 28462
    },
    {
      "epoch": 0.8201175589235291,
      "grad_norm": 0.7037397027015686,
      "learning_rate": 8.248510569683881e-07,
      "loss": 0.6713,
      "step": 28463
    },
    {
      "epoch": 0.8201463723851784,
      "grad_norm": 0.7300017476081848,
      "learning_rate": 8.245943440482174e-07,
      "loss": 0.6788,
      "step": 28464
    },
    {
      "epoch": 0.8201751858468277,
      "grad_norm": 0.7436732649803162,
      "learning_rate": 8.243376674916548e-07,
      "loss": 0.6739,
      "step": 28465
    },
    {
      "epoch": 0.8202039993084769,
      "grad_norm": 0.7173042893409729,
      "learning_rate": 8.240810273009381e-07,
      "loss": 0.6734,
      "step": 28466
    },
    {
      "epoch": 0.8202328127701262,
      "grad_norm": 0.7243236303329468,
      "learning_rate": 8.238244234783005e-07,
      "loss": 0.6712,
      "step": 28467
    },
    {
      "epoch": 0.8202616262317755,
      "grad_norm": 0.7232280373573303,
      "learning_rate": 8.235678560259785e-07,
      "loss": 0.6863,
      "step": 28468
    },
    {
      "epoch": 0.8202904396934247,
      "grad_norm": 0.7201792001724243,
      "learning_rate": 8.233113249462043e-07,
      "loss": 0.6806,
      "step": 28469
    },
    {
      "epoch": 0.820319253155074,
      "grad_norm": 0.7616038918495178,
      "learning_rate": 8.230548302412133e-07,
      "loss": 0.6778,
      "step": 28470
    },
    {
      "epoch": 0.8203480666167233,
      "grad_norm": 0.7059892416000366,
      "learning_rate": 8.227983719132404e-07,
      "loss": 0.6764,
      "step": 28471
    },
    {
      "epoch": 0.8203768800783726,
      "grad_norm": 0.7280625700950623,
      "learning_rate": 8.22541949964517e-07,
      "loss": 0.6769,
      "step": 28472
    },
    {
      "epoch": 0.8204056935400219,
      "grad_norm": 0.7389631867408752,
      "learning_rate": 8.222855643972777e-07,
      "loss": 0.7028,
      "step": 28473
    },
    {
      "epoch": 0.8204345070016712,
      "grad_norm": 0.746432363986969,
      "learning_rate": 8.220292152137555e-07,
      "loss": 0.6951,
      "step": 28474
    },
    {
      "epoch": 0.8204633204633205,
      "grad_norm": 0.7187097668647766,
      "learning_rate": 8.217729024161814e-07,
      "loss": 0.6765,
      "step": 28475
    },
    {
      "epoch": 0.8204921339249698,
      "grad_norm": 0.7160741686820984,
      "learning_rate": 8.215166260067886e-07,
      "loss": 0.6694,
      "step": 28476
    },
    {
      "epoch": 0.8205209473866191,
      "grad_norm": 0.7364636659622192,
      "learning_rate": 8.212603859878104e-07,
      "loss": 0.7013,
      "step": 28477
    },
    {
      "epoch": 0.8205497608482684,
      "grad_norm": 0.7156825065612793,
      "learning_rate": 8.210041823614756e-07,
      "loss": 0.6629,
      "step": 28478
    },
    {
      "epoch": 0.8205785743099175,
      "grad_norm": 0.7474782466888428,
      "learning_rate": 8.207480151300184e-07,
      "loss": 0.7082,
      "step": 28479
    },
    {
      "epoch": 0.8206073877715668,
      "grad_norm": 0.7276941537857056,
      "learning_rate": 8.204918842956666e-07,
      "loss": 0.7016,
      "step": 28480
    },
    {
      "epoch": 0.8206362012332161,
      "grad_norm": 0.7161242365837097,
      "learning_rate": 8.202357898606539e-07,
      "loss": 0.6743,
      "step": 28481
    },
    {
      "epoch": 0.8206650146948654,
      "grad_norm": 1.0506781339645386,
      "learning_rate": 8.199797318272085e-07,
      "loss": 0.6645,
      "step": 28482
    },
    {
      "epoch": 0.8206938281565147,
      "grad_norm": 0.7128627300262451,
      "learning_rate": 8.19723710197562e-07,
      "loss": 0.707,
      "step": 28483
    },
    {
      "epoch": 0.820722641618164,
      "grad_norm": 0.7386172413825989,
      "learning_rate": 8.194677249739403e-07,
      "loss": 0.7047,
      "step": 28484
    },
    {
      "epoch": 0.8207514550798133,
      "grad_norm": 0.7153728008270264,
      "learning_rate": 8.192117761585788e-07,
      "loss": 0.6762,
      "step": 28485
    },
    {
      "epoch": 0.8207802685414626,
      "grad_norm": 0.7172016501426697,
      "learning_rate": 8.189558637537021e-07,
      "loss": 0.6932,
      "step": 28486
    },
    {
      "epoch": 0.8208090820031119,
      "grad_norm": 5.8680009841918945,
      "learning_rate": 8.186999877615414e-07,
      "loss": 0.6941,
      "step": 28487
    },
    {
      "epoch": 0.8208378954647612,
      "grad_norm": 0.7261003255844116,
      "learning_rate": 8.184441481843231e-07,
      "loss": 0.6676,
      "step": 28488
    },
    {
      "epoch": 0.8208667089264104,
      "grad_norm": 0.718079149723053,
      "learning_rate": 8.181883450242773e-07,
      "loss": 0.6808,
      "step": 28489
    },
    {
      "epoch": 0.8208955223880597,
      "grad_norm": 0.7428687214851379,
      "learning_rate": 8.179325782836295e-07,
      "loss": 0.686,
      "step": 28490
    },
    {
      "epoch": 0.820924335849709,
      "grad_norm": 0.7267698049545288,
      "learning_rate": 8.176768479646091e-07,
      "loss": 0.6858,
      "step": 28491
    },
    {
      "epoch": 0.8209531493113583,
      "grad_norm": 0.7198184132575989,
      "learning_rate": 8.174211540694421e-07,
      "loss": 0.6877,
      "step": 28492
    },
    {
      "epoch": 0.8209819627730075,
      "grad_norm": 0.7163422703742981,
      "learning_rate": 8.171654966003556e-07,
      "loss": 0.6677,
      "step": 28493
    },
    {
      "epoch": 0.8210107762346568,
      "grad_norm": 0.735040545463562,
      "learning_rate": 8.169098755595773e-07,
      "loss": 0.6934,
      "step": 28494
    },
    {
      "epoch": 0.8210395896963061,
      "grad_norm": 0.7214927077293396,
      "learning_rate": 8.166542909493314e-07,
      "loss": 0.6701,
      "step": 28495
    },
    {
      "epoch": 0.8210684031579554,
      "grad_norm": 0.732069194316864,
      "learning_rate": 8.16398742771845e-07,
      "loss": 0.6874,
      "step": 28496
    },
    {
      "epoch": 0.8210972166196047,
      "grad_norm": 0.7197397947311401,
      "learning_rate": 8.161432310293433e-07,
      "loss": 0.6808,
      "step": 28497
    },
    {
      "epoch": 0.821126030081254,
      "grad_norm": 0.7294656038284302,
      "learning_rate": 8.15887755724053e-07,
      "loss": 0.6862,
      "step": 28498
    },
    {
      "epoch": 0.8211548435429032,
      "grad_norm": 0.7078737616539001,
      "learning_rate": 8.156323168581964e-07,
      "loss": 0.6706,
      "step": 28499
    },
    {
      "epoch": 0.8211836570045525,
      "grad_norm": 0.7238340377807617,
      "learning_rate": 8.153769144340007e-07,
      "loss": 0.7005,
      "step": 28500
    },
    {
      "epoch": 0.8212124704662018,
      "grad_norm": 0.7397460341453552,
      "learning_rate": 8.151215484536879e-07,
      "loss": 0.6768,
      "step": 28501
    },
    {
      "epoch": 0.8212412839278511,
      "grad_norm": 0.7139462828636169,
      "learning_rate": 8.148662189194844e-07,
      "loss": 0.6709,
      "step": 28502
    },
    {
      "epoch": 0.8212700973895004,
      "grad_norm": 0.7299792170524597,
      "learning_rate": 8.14610925833611e-07,
      "loss": 0.6947,
      "step": 28503
    },
    {
      "epoch": 0.8212989108511497,
      "grad_norm": 0.7343937158584595,
      "learning_rate": 8.14355669198294e-07,
      "loss": 0.7009,
      "step": 28504
    },
    {
      "epoch": 0.821327724312799,
      "grad_norm": 0.7366138696670532,
      "learning_rate": 8.141004490157539e-07,
      "loss": 0.7006,
      "step": 28505
    },
    {
      "epoch": 0.8213565377744482,
      "grad_norm": 0.740075945854187,
      "learning_rate": 8.138452652882156e-07,
      "loss": 0.6877,
      "step": 28506
    },
    {
      "epoch": 0.8213853512360975,
      "grad_norm": 0.7032908797264099,
      "learning_rate": 8.13590118017899e-07,
      "loss": 0.6784,
      "step": 28507
    },
    {
      "epoch": 0.8214141646977468,
      "grad_norm": 0.7211964726448059,
      "learning_rate": 8.133350072070278e-07,
      "loss": 0.6991,
      "step": 28508
    },
    {
      "epoch": 0.821442978159396,
      "grad_norm": 0.7328566908836365,
      "learning_rate": 8.130799328578238e-07,
      "loss": 0.6752,
      "step": 28509
    },
    {
      "epoch": 0.8214717916210453,
      "grad_norm": 0.7094987630844116,
      "learning_rate": 8.128248949725087e-07,
      "loss": 0.6777,
      "step": 28510
    },
    {
      "epoch": 0.8215006050826946,
      "grad_norm": 0.7245419025421143,
      "learning_rate": 8.125698935533022e-07,
      "loss": 0.6795,
      "step": 28511
    },
    {
      "epoch": 0.8215294185443439,
      "grad_norm": 0.7071787714958191,
      "learning_rate": 8.123149286024273e-07,
      "loss": 0.6933,
      "step": 28512
    },
    {
      "epoch": 0.8215582320059932,
      "grad_norm": 0.7288840413093567,
      "learning_rate": 8.120600001221024e-07,
      "loss": 0.701,
      "step": 28513
    },
    {
      "epoch": 0.8215870454676425,
      "grad_norm": 0.7188296318054199,
      "learning_rate": 8.118051081145484e-07,
      "loss": 0.6856,
      "step": 28514
    },
    {
      "epoch": 0.8216158589292918,
      "grad_norm": 0.716437041759491,
      "learning_rate": 8.115502525819862e-07,
      "loss": 0.665,
      "step": 28515
    },
    {
      "epoch": 0.8216446723909411,
      "grad_norm": 0.7199575901031494,
      "learning_rate": 8.112954335266332e-07,
      "loss": 0.6959,
      "step": 28516
    },
    {
      "epoch": 0.8216734858525904,
      "grad_norm": 0.7272732257843018,
      "learning_rate": 8.110406509507107e-07,
      "loss": 0.6844,
      "step": 28517
    },
    {
      "epoch": 0.8217022993142397,
      "grad_norm": 0.7174020409584045,
      "learning_rate": 8.107859048564359e-07,
      "loss": 0.6839,
      "step": 28518
    },
    {
      "epoch": 0.821731112775889,
      "grad_norm": 0.7074487209320068,
      "learning_rate": 8.105311952460293e-07,
      "loss": 0.6847,
      "step": 28519
    },
    {
      "epoch": 0.8217599262375381,
      "grad_norm": 0.7329456806182861,
      "learning_rate": 8.102765221217057e-07,
      "loss": 0.6609,
      "step": 28520
    },
    {
      "epoch": 0.8217887396991874,
      "grad_norm": 0.7240434885025024,
      "learning_rate": 8.100218854856878e-07,
      "loss": 0.6806,
      "step": 28521
    },
    {
      "epoch": 0.8218175531608367,
      "grad_norm": 0.715221643447876,
      "learning_rate": 8.097672853401895e-07,
      "loss": 0.6833,
      "step": 28522
    },
    {
      "epoch": 0.821846366622486,
      "grad_norm": 0.7109608054161072,
      "learning_rate": 8.09512721687431e-07,
      "loss": 0.6863,
      "step": 28523
    },
    {
      "epoch": 0.8218751800841353,
      "grad_norm": 0.7243924140930176,
      "learning_rate": 8.092581945296263e-07,
      "loss": 0.6865,
      "step": 28524
    },
    {
      "epoch": 0.8219039935457846,
      "grad_norm": 0.7108103632926941,
      "learning_rate": 8.090037038689946e-07,
      "loss": 0.6617,
      "step": 28525
    },
    {
      "epoch": 0.8219328070074339,
      "grad_norm": 0.7329779267311096,
      "learning_rate": 8.087492497077498e-07,
      "loss": 0.6946,
      "step": 28526
    },
    {
      "epoch": 0.8219616204690832,
      "grad_norm": 0.7398452162742615,
      "learning_rate": 8.084948320481112e-07,
      "loss": 0.6868,
      "step": 28527
    },
    {
      "epoch": 0.8219904339307325,
      "grad_norm": 0.7228325009346008,
      "learning_rate": 8.082404508922909e-07,
      "loss": 0.69,
      "step": 28528
    },
    {
      "epoch": 0.8220192473923817,
      "grad_norm": 0.7214890122413635,
      "learning_rate": 8.079861062425076e-07,
      "loss": 0.6827,
      "step": 28529
    },
    {
      "epoch": 0.822048060854031,
      "grad_norm": 0.7090237140655518,
      "learning_rate": 8.077317981009731e-07,
      "loss": 0.6789,
      "step": 28530
    },
    {
      "epoch": 0.8220768743156803,
      "grad_norm": 0.7416326999664307,
      "learning_rate": 8.074775264699042e-07,
      "loss": 0.658,
      "step": 28531
    },
    {
      "epoch": 0.8221056877773296,
      "grad_norm": 0.7282970547676086,
      "learning_rate": 8.072232913515149e-07,
      "loss": 0.66,
      "step": 28532
    },
    {
      "epoch": 0.8221345012389788,
      "grad_norm": 0.7092983722686768,
      "learning_rate": 8.069690927480211e-07,
      "loss": 0.6687,
      "step": 28533
    },
    {
      "epoch": 0.8221633147006281,
      "grad_norm": 0.7208132743835449,
      "learning_rate": 8.067149306616334e-07,
      "loss": 0.6731,
      "step": 28534
    },
    {
      "epoch": 0.8221921281622774,
      "grad_norm": 0.7264866828918457,
      "learning_rate": 8.064608050945671e-07,
      "loss": 0.6952,
      "step": 28535
    },
    {
      "epoch": 0.8222209416239267,
      "grad_norm": 0.7207427620887756,
      "learning_rate": 8.062067160490367e-07,
      "loss": 0.7067,
      "step": 28536
    },
    {
      "epoch": 0.822249755085576,
      "grad_norm": 0.7234732508659363,
      "learning_rate": 8.059526635272519e-07,
      "loss": 0.6709,
      "step": 28537
    },
    {
      "epoch": 0.8222785685472253,
      "grad_norm": 0.7003331184387207,
      "learning_rate": 8.056986475314282e-07,
      "loss": 0.6688,
      "step": 28538
    },
    {
      "epoch": 0.8223073820088745,
      "grad_norm": 0.7194218039512634,
      "learning_rate": 8.054446680637751e-07,
      "loss": 0.6683,
      "step": 28539
    },
    {
      "epoch": 0.8223361954705238,
      "grad_norm": 0.7481147050857544,
      "learning_rate": 8.051907251265073e-07,
      "loss": 0.7013,
      "step": 28540
    },
    {
      "epoch": 0.8223650089321731,
      "grad_norm": 0.7171043753623962,
      "learning_rate": 8.049368187218337e-07,
      "loss": 0.6851,
      "step": 28541
    },
    {
      "epoch": 0.8223938223938224,
      "grad_norm": 0.750281572341919,
      "learning_rate": 8.046829488519686e-07,
      "loss": 0.6913,
      "step": 28542
    },
    {
      "epoch": 0.8224226358554717,
      "grad_norm": 0.7278189063072205,
      "learning_rate": 8.044291155191197e-07,
      "loss": 0.6764,
      "step": 28543
    },
    {
      "epoch": 0.822451449317121,
      "grad_norm": 0.7257230877876282,
      "learning_rate": 8.041753187254992e-07,
      "loss": 0.6726,
      "step": 28544
    },
    {
      "epoch": 0.8224802627787703,
      "grad_norm": 0.751200795173645,
      "learning_rate": 8.039215584733174e-07,
      "loss": 0.7064,
      "step": 28545
    },
    {
      "epoch": 0.8225090762404196,
      "grad_norm": 0.7284293174743652,
      "learning_rate": 8.036678347647852e-07,
      "loss": 0.7248,
      "step": 28546
    },
    {
      "epoch": 0.8225378897020688,
      "grad_norm": 0.7144133448600769,
      "learning_rate": 8.034141476021102e-07,
      "loss": 0.6883,
      "step": 28547
    },
    {
      "epoch": 0.822566703163718,
      "grad_norm": 0.719139575958252,
      "learning_rate": 8.031604969875045e-07,
      "loss": 0.6476,
      "step": 28548
    },
    {
      "epoch": 0.8225955166253673,
      "grad_norm": 0.712375819683075,
      "learning_rate": 8.02906882923174e-07,
      "loss": 0.6668,
      "step": 28549
    },
    {
      "epoch": 0.8226243300870166,
      "grad_norm": 0.7213947176933289,
      "learning_rate": 8.0265330541133e-07,
      "loss": 0.6871,
      "step": 28550
    },
    {
      "epoch": 0.8226531435486659,
      "grad_norm": 0.7478143572807312,
      "learning_rate": 8.023997644541792e-07,
      "loss": 0.7311,
      "step": 28551
    },
    {
      "epoch": 0.8226819570103152,
      "grad_norm": 0.7205697894096375,
      "learning_rate": 8.021462600539304e-07,
      "loss": 0.6761,
      "step": 28552
    },
    {
      "epoch": 0.8227107704719645,
      "grad_norm": 0.7155284285545349,
      "learning_rate": 8.018927922127923e-07,
      "loss": 0.7015,
      "step": 28553
    },
    {
      "epoch": 0.8227395839336138,
      "grad_norm": 0.7339831590652466,
      "learning_rate": 8.016393609329703e-07,
      "loss": 0.6922,
      "step": 28554
    },
    {
      "epoch": 0.8227683973952631,
      "grad_norm": 0.7025407552719116,
      "learning_rate": 8.01385966216674e-07,
      "loss": 0.6641,
      "step": 28555
    },
    {
      "epoch": 0.8227972108569124,
      "grad_norm": 0.7276931405067444,
      "learning_rate": 8.011326080661058e-07,
      "loss": 0.6796,
      "step": 28556
    },
    {
      "epoch": 0.8228260243185617,
      "grad_norm": 0.7301459908485413,
      "learning_rate": 8.008792864834785e-07,
      "loss": 0.6718,
      "step": 28557
    },
    {
      "epoch": 0.822854837780211,
      "grad_norm": 0.7406924962997437,
      "learning_rate": 8.006260014709932e-07,
      "loss": 0.7084,
      "step": 28558
    },
    {
      "epoch": 0.8228836512418602,
      "grad_norm": 0.7465454936027527,
      "learning_rate": 8.003727530308592e-07,
      "loss": 0.679,
      "step": 28559
    },
    {
      "epoch": 0.8229124647035094,
      "grad_norm": 0.6964095830917358,
      "learning_rate": 8.00119541165279e-07,
      "loss": 0.6675,
      "step": 28560
    },
    {
      "epoch": 0.8229412781651587,
      "grad_norm": 0.7552551031112671,
      "learning_rate": 7.998663658764605e-07,
      "loss": 0.6924,
      "step": 28561
    },
    {
      "epoch": 0.822970091626808,
      "grad_norm": 0.726961076259613,
      "learning_rate": 7.996132271666062e-07,
      "loss": 0.6822,
      "step": 28562
    },
    {
      "epoch": 0.8229989050884573,
      "grad_norm": 0.7185803651809692,
      "learning_rate": 7.993601250379229e-07,
      "loss": 0.6707,
      "step": 28563
    },
    {
      "epoch": 0.8230277185501066,
      "grad_norm": 0.7277646660804749,
      "learning_rate": 7.991070594926132e-07,
      "loss": 0.6865,
      "step": 28564
    },
    {
      "epoch": 0.8230565320117559,
      "grad_norm": 0.7414065003395081,
      "learning_rate": 7.988540305328824e-07,
      "loss": 0.6959,
      "step": 28565
    },
    {
      "epoch": 0.8230853454734052,
      "grad_norm": 0.7107320427894592,
      "learning_rate": 7.986010381609322e-07,
      "loss": 0.6785,
      "step": 28566
    },
    {
      "epoch": 0.8231141589350545,
      "grad_norm": 0.7135798931121826,
      "learning_rate": 7.983480823789669e-07,
      "loss": 0.6643,
      "step": 28567
    },
    {
      "epoch": 0.8231429723967038,
      "grad_norm": 0.7161332964897156,
      "learning_rate": 7.980951631891897e-07,
      "loss": 0.6732,
      "step": 28568
    },
    {
      "epoch": 0.823171785858353,
      "grad_norm": 0.741288959980011,
      "learning_rate": 7.978422805938047e-07,
      "loss": 0.6914,
      "step": 28569
    },
    {
      "epoch": 0.8232005993200023,
      "grad_norm": 0.7273041605949402,
      "learning_rate": 7.975894345950114e-07,
      "loss": 0.682,
      "step": 28570
    },
    {
      "epoch": 0.8232294127816516,
      "grad_norm": 0.7414371371269226,
      "learning_rate": 7.973366251950149e-07,
      "loss": 0.7196,
      "step": 28571
    },
    {
      "epoch": 0.8232582262433009,
      "grad_norm": 0.7524053454399109,
      "learning_rate": 7.97083852396014e-07,
      "loss": 0.696,
      "step": 28572
    },
    {
      "epoch": 0.8232870397049502,
      "grad_norm": 0.7447381019592285,
      "learning_rate": 7.968311162002107e-07,
      "loss": 0.7073,
      "step": 28573
    },
    {
      "epoch": 0.8233158531665994,
      "grad_norm": 0.7338259816169739,
      "learning_rate": 7.965784166098084e-07,
      "loss": 0.6557,
      "step": 28574
    },
    {
      "epoch": 0.8233446666282487,
      "grad_norm": 0.7101179361343384,
      "learning_rate": 7.963257536270053e-07,
      "loss": 0.6924,
      "step": 28575
    },
    {
      "epoch": 0.823373480089898,
      "grad_norm": 0.7143231630325317,
      "learning_rate": 7.960731272540034e-07,
      "loss": 0.6834,
      "step": 28576
    },
    {
      "epoch": 0.8234022935515473,
      "grad_norm": 0.7061665654182434,
      "learning_rate": 7.958205374930012e-07,
      "loss": 0.6776,
      "step": 28577
    },
    {
      "epoch": 0.8234311070131966,
      "grad_norm": 0.7102747559547424,
      "learning_rate": 7.955679843462005e-07,
      "loss": 0.6892,
      "step": 28578
    },
    {
      "epoch": 0.8234599204748458,
      "grad_norm": 0.706814706325531,
      "learning_rate": 7.953154678157987e-07,
      "loss": 0.6674,
      "step": 28579
    },
    {
      "epoch": 0.8234887339364951,
      "grad_norm": 0.7809029221534729,
      "learning_rate": 7.950629879039961e-07,
      "loss": 0.6703,
      "step": 28580
    },
    {
      "epoch": 0.8235175473981444,
      "grad_norm": 0.7253649830818176,
      "learning_rate": 7.948105446129916e-07,
      "loss": 0.6851,
      "step": 28581
    },
    {
      "epoch": 0.8235463608597937,
      "grad_norm": 0.7322103977203369,
      "learning_rate": 7.945581379449846e-07,
      "loss": 0.6631,
      "step": 28582
    },
    {
      "epoch": 0.823575174321443,
      "grad_norm": 0.7263838648796082,
      "learning_rate": 7.943057679021715e-07,
      "loss": 0.6905,
      "step": 28583
    },
    {
      "epoch": 0.8236039877830923,
      "grad_norm": 0.7037688493728638,
      "learning_rate": 7.940534344867517e-07,
      "loss": 0.6887,
      "step": 28584
    },
    {
      "epoch": 0.8236328012447416,
      "grad_norm": 0.7195074558258057,
      "learning_rate": 7.938011377009214e-07,
      "loss": 0.6726,
      "step": 28585
    },
    {
      "epoch": 0.8236616147063909,
      "grad_norm": 0.7071371078491211,
      "learning_rate": 7.935488775468792e-07,
      "loss": 0.691,
      "step": 28586
    },
    {
      "epoch": 0.8236904281680401,
      "grad_norm": 0.7357611656188965,
      "learning_rate": 7.9329665402682e-07,
      "loss": 0.6993,
      "step": 28587
    },
    {
      "epoch": 0.8237192416296893,
      "grad_norm": 0.7365456223487854,
      "learning_rate": 7.930444671429438e-07,
      "loss": 0.6831,
      "step": 28588
    },
    {
      "epoch": 0.8237480550913386,
      "grad_norm": 0.7361977100372314,
      "learning_rate": 7.927923168974428e-07,
      "loss": 0.6868,
      "step": 28589
    },
    {
      "epoch": 0.8237768685529879,
      "grad_norm": 0.7109845876693726,
      "learning_rate": 7.925402032925156e-07,
      "loss": 0.6752,
      "step": 28590
    },
    {
      "epoch": 0.8238056820146372,
      "grad_norm": 0.7179432511329651,
      "learning_rate": 7.922881263303583e-07,
      "loss": 0.6812,
      "step": 28591
    },
    {
      "epoch": 0.8238344954762865,
      "grad_norm": 0.725017786026001,
      "learning_rate": 7.920360860131643e-07,
      "loss": 0.696,
      "step": 28592
    },
    {
      "epoch": 0.8238633089379358,
      "grad_norm": 0.7185972332954407,
      "learning_rate": 7.917840823431294e-07,
      "loss": 0.654,
      "step": 28593
    },
    {
      "epoch": 0.8238921223995851,
      "grad_norm": 0.7199931740760803,
      "learning_rate": 7.915321153224487e-07,
      "loss": 0.701,
      "step": 28594
    },
    {
      "epoch": 0.8239209358612344,
      "grad_norm": 0.7203496694564819,
      "learning_rate": 7.912801849533169e-07,
      "loss": 0.6793,
      "step": 28595
    },
    {
      "epoch": 0.8239497493228837,
      "grad_norm": 0.7241397500038147,
      "learning_rate": 7.91028291237927e-07,
      "loss": 0.6947,
      "step": 28596
    },
    {
      "epoch": 0.823978562784533,
      "grad_norm": 0.7386661171913147,
      "learning_rate": 7.90776434178474e-07,
      "loss": 0.6796,
      "step": 28597
    },
    {
      "epoch": 0.8240073762461823,
      "grad_norm": 0.7156903743743896,
      "learning_rate": 7.905246137771494e-07,
      "loss": 0.6641,
      "step": 28598
    },
    {
      "epoch": 0.8240361897078315,
      "grad_norm": 0.7512582540512085,
      "learning_rate": 7.902728300361489e-07,
      "loss": 0.7218,
      "step": 28599
    },
    {
      "epoch": 0.8240650031694808,
      "grad_norm": 0.7267823815345764,
      "learning_rate": 7.900210829576626e-07,
      "loss": 0.692,
      "step": 28600
    },
    {
      "epoch": 0.82409381663113,
      "grad_norm": 0.7149772644042969,
      "learning_rate": 7.897693725438854e-07,
      "loss": 0.6979,
      "step": 28601
    },
    {
      "epoch": 0.8241226300927793,
      "grad_norm": 0.7627220153808594,
      "learning_rate": 7.895176987970071e-07,
      "loss": 0.7035,
      "step": 28602
    },
    {
      "epoch": 0.8241514435544286,
      "grad_norm": 0.8213132619857788,
      "learning_rate": 7.892660617192222e-07,
      "loss": 0.6851,
      "step": 28603
    },
    {
      "epoch": 0.8241802570160779,
      "grad_norm": 0.7255995869636536,
      "learning_rate": 7.890144613127176e-07,
      "loss": 0.6699,
      "step": 28604
    },
    {
      "epoch": 0.8242090704777272,
      "grad_norm": 0.7168936729431152,
      "learning_rate": 7.887628975796907e-07,
      "loss": 0.6695,
      "step": 28605
    },
    {
      "epoch": 0.8242378839393765,
      "grad_norm": 0.7369714379310608,
      "learning_rate": 7.885113705223274e-07,
      "loss": 0.6803,
      "step": 28606
    },
    {
      "epoch": 0.8242666974010258,
      "grad_norm": 0.7160630226135254,
      "learning_rate": 7.88259880142822e-07,
      "loss": 0.687,
      "step": 28607
    },
    {
      "epoch": 0.824295510862675,
      "grad_norm": 0.7247445583343506,
      "learning_rate": 7.880084264433613e-07,
      "loss": 0.6939,
      "step": 28608
    },
    {
      "epoch": 0.8243243243243243,
      "grad_norm": 0.6980509757995605,
      "learning_rate": 7.877570094261377e-07,
      "loss": 0.6482,
      "step": 28609
    },
    {
      "epoch": 0.8243531377859736,
      "grad_norm": 0.7046775817871094,
      "learning_rate": 7.875056290933392e-07,
      "loss": 0.6626,
      "step": 28610
    },
    {
      "epoch": 0.8243819512476229,
      "grad_norm": 0.7185201644897461,
      "learning_rate": 7.872542854471554e-07,
      "loss": 0.6531,
      "step": 28611
    },
    {
      "epoch": 0.8244107647092722,
      "grad_norm": 0.7471967935562134,
      "learning_rate": 7.870029784897765e-07,
      "loss": 0.7145,
      "step": 28612
    },
    {
      "epoch": 0.8244395781709215,
      "grad_norm": 0.7136958241462708,
      "learning_rate": 7.867517082233895e-07,
      "loss": 0.6735,
      "step": 28613
    },
    {
      "epoch": 0.8244683916325707,
      "grad_norm": 0.725811779499054,
      "learning_rate": 7.865004746501842e-07,
      "loss": 0.6821,
      "step": 28614
    },
    {
      "epoch": 0.82449720509422,
      "grad_norm": 0.7176691293716431,
      "learning_rate": 7.862492777723468e-07,
      "loss": 0.6699,
      "step": 28615
    },
    {
      "epoch": 0.8245260185558693,
      "grad_norm": 0.7288048267364502,
      "learning_rate": 7.859981175920661e-07,
      "loss": 0.6718,
      "step": 28616
    },
    {
      "epoch": 0.8245548320175186,
      "grad_norm": 0.7096377611160278,
      "learning_rate": 7.857469941115292e-07,
      "loss": 0.7005,
      "step": 28617
    },
    {
      "epoch": 0.8245836454791678,
      "grad_norm": 0.7254801392555237,
      "learning_rate": 7.854959073329249e-07,
      "loss": 0.6651,
      "step": 28618
    },
    {
      "epoch": 0.8246124589408171,
      "grad_norm": 0.718632161617279,
      "learning_rate": 7.852448572584364e-07,
      "loss": 0.685,
      "step": 28619
    },
    {
      "epoch": 0.8246412724024664,
      "grad_norm": 0.7226888537406921,
      "learning_rate": 7.849938438902538e-07,
      "loss": 0.6861,
      "step": 28620
    },
    {
      "epoch": 0.8246700858641157,
      "grad_norm": 0.7258561253547668,
      "learning_rate": 7.847428672305596e-07,
      "loss": 0.6886,
      "step": 28621
    },
    {
      "epoch": 0.824698899325765,
      "grad_norm": 0.7129559516906738,
      "learning_rate": 7.844919272815426e-07,
      "loss": 0.6719,
      "step": 28622
    },
    {
      "epoch": 0.8247277127874143,
      "grad_norm": 0.7030047178268433,
      "learning_rate": 7.842410240453862e-07,
      "loss": 0.6659,
      "step": 28623
    },
    {
      "epoch": 0.8247565262490636,
      "grad_norm": 0.744478702545166,
      "learning_rate": 7.83990157524277e-07,
      "loss": 0.6765,
      "step": 28624
    },
    {
      "epoch": 0.8247853397107129,
      "grad_norm": 0.7267180681228638,
      "learning_rate": 7.837393277203981e-07,
      "loss": 0.6829,
      "step": 28625
    },
    {
      "epoch": 0.8248141531723622,
      "grad_norm": 0.725476086139679,
      "learning_rate": 7.834885346359361e-07,
      "loss": 0.6632,
      "step": 28626
    },
    {
      "epoch": 0.8248429666340115,
      "grad_norm": 0.7146382331848145,
      "learning_rate": 7.832377782730727e-07,
      "loss": 0.691,
      "step": 28627
    },
    {
      "epoch": 0.8248717800956606,
      "grad_norm": 0.7242558002471924,
      "learning_rate": 7.829870586339933e-07,
      "loss": 0.6866,
      "step": 28628
    },
    {
      "epoch": 0.8249005935573099,
      "grad_norm": 0.7184585332870483,
      "learning_rate": 7.827363757208806e-07,
      "loss": 0.6909,
      "step": 28629
    },
    {
      "epoch": 0.8249294070189592,
      "grad_norm": 0.71455979347229,
      "learning_rate": 7.824857295359201e-07,
      "loss": 0.6735,
      "step": 28630
    },
    {
      "epoch": 0.8249582204806085,
      "grad_norm": 0.7216135263442993,
      "learning_rate": 7.822351200812917e-07,
      "loss": 0.6968,
      "step": 28631
    },
    {
      "epoch": 0.8249870339422578,
      "grad_norm": 0.7102413177490234,
      "learning_rate": 7.819845473591787e-07,
      "loss": 0.6744,
      "step": 28632
    },
    {
      "epoch": 0.8250158474039071,
      "grad_norm": 0.706082820892334,
      "learning_rate": 7.817340113717653e-07,
      "loss": 0.686,
      "step": 28633
    },
    {
      "epoch": 0.8250446608655564,
      "grad_norm": 0.7185386419296265,
      "learning_rate": 7.814835121212305e-07,
      "loss": 0.6819,
      "step": 28634
    },
    {
      "epoch": 0.8250734743272057,
      "grad_norm": 0.7311177253723145,
      "learning_rate": 7.812330496097592e-07,
      "loss": 0.6498,
      "step": 28635
    },
    {
      "epoch": 0.825102287788855,
      "grad_norm": 0.7288163900375366,
      "learning_rate": 7.809826238395291e-07,
      "loss": 0.6689,
      "step": 28636
    },
    {
      "epoch": 0.8251311012505043,
      "grad_norm": 0.7109376788139343,
      "learning_rate": 7.807322348127239e-07,
      "loss": 0.6644,
      "step": 28637
    },
    {
      "epoch": 0.8251599147121536,
      "grad_norm": 0.7192198038101196,
      "learning_rate": 7.804818825315225e-07,
      "loss": 0.6871,
      "step": 28638
    },
    {
      "epoch": 0.8251887281738028,
      "grad_norm": 0.7192742228507996,
      "learning_rate": 7.80231566998107e-07,
      "loss": 0.6812,
      "step": 28639
    },
    {
      "epoch": 0.8252175416354521,
      "grad_norm": 0.7241104245185852,
      "learning_rate": 7.799812882146541e-07,
      "loss": 0.6934,
      "step": 28640
    },
    {
      "epoch": 0.8252463550971013,
      "grad_norm": 0.7113146781921387,
      "learning_rate": 7.797310461833474e-07,
      "loss": 0.6829,
      "step": 28641
    },
    {
      "epoch": 0.8252751685587506,
      "grad_norm": 0.7086080312728882,
      "learning_rate": 7.794808409063637e-07,
      "loss": 0.6947,
      "step": 28642
    },
    {
      "epoch": 0.8253039820203999,
      "grad_norm": 0.7428293228149414,
      "learning_rate": 7.792306723858845e-07,
      "loss": 0.6816,
      "step": 28643
    },
    {
      "epoch": 0.8253327954820492,
      "grad_norm": 0.7090675830841064,
      "learning_rate": 7.789805406240852e-07,
      "loss": 0.696,
      "step": 28644
    },
    {
      "epoch": 0.8253616089436985,
      "grad_norm": 0.7238500118255615,
      "learning_rate": 7.787304456231476e-07,
      "loss": 0.6838,
      "step": 28645
    },
    {
      "epoch": 0.8253904224053478,
      "grad_norm": 0.7064415216445923,
      "learning_rate": 7.784803873852464e-07,
      "loss": 0.6683,
      "step": 28646
    },
    {
      "epoch": 0.8254192358669971,
      "grad_norm": 0.7202078104019165,
      "learning_rate": 7.782303659125629e-07,
      "loss": 0.6749,
      "step": 28647
    },
    {
      "epoch": 0.8254480493286463,
      "grad_norm": 0.7447091341018677,
      "learning_rate": 7.779803812072706e-07,
      "loss": 0.7066,
      "step": 28648
    },
    {
      "epoch": 0.8254768627902956,
      "grad_norm": 0.7659872174263,
      "learning_rate": 7.777304332715496e-07,
      "loss": 0.6923,
      "step": 28649
    },
    {
      "epoch": 0.8255056762519449,
      "grad_norm": 0.7617698311805725,
      "learning_rate": 7.774805221075765e-07,
      "loss": 0.6939,
      "step": 28650
    },
    {
      "epoch": 0.8255344897135942,
      "grad_norm": 0.7229052782058716,
      "learning_rate": 7.772306477175257e-07,
      "loss": 0.6812,
      "step": 28651
    },
    {
      "epoch": 0.8255633031752435,
      "grad_norm": 0.693884551525116,
      "learning_rate": 7.769808101035753e-07,
      "loss": 0.6816,
      "step": 28652
    },
    {
      "epoch": 0.8255921166368928,
      "grad_norm": 0.7256773114204407,
      "learning_rate": 7.767310092679004e-07,
      "loss": 0.6668,
      "step": 28653
    },
    {
      "epoch": 0.8256209300985421,
      "grad_norm": 0.707258939743042,
      "learning_rate": 7.764812452126774e-07,
      "loss": 0.6564,
      "step": 28654
    },
    {
      "epoch": 0.8256497435601913,
      "grad_norm": 0.7296473383903503,
      "learning_rate": 7.762315179400798e-07,
      "loss": 0.6959,
      "step": 28655
    },
    {
      "epoch": 0.8256785570218406,
      "grad_norm": 0.734376847743988,
      "learning_rate": 7.759818274522851e-07,
      "loss": 0.6954,
      "step": 28656
    },
    {
      "epoch": 0.8257073704834899,
      "grad_norm": 0.7115340828895569,
      "learning_rate": 7.757321737514645e-07,
      "loss": 0.6797,
      "step": 28657
    },
    {
      "epoch": 0.8257361839451391,
      "grad_norm": 0.7320231199264526,
      "learning_rate": 7.754825568397955e-07,
      "loss": 0.7022,
      "step": 28658
    },
    {
      "epoch": 0.8257649974067884,
      "grad_norm": 0.7001258730888367,
      "learning_rate": 7.752329767194489e-07,
      "loss": 0.6515,
      "step": 28659
    },
    {
      "epoch": 0.8257938108684377,
      "grad_norm": 0.7095749378204346,
      "learning_rate": 7.749834333926015e-07,
      "loss": 0.6918,
      "step": 28660
    },
    {
      "epoch": 0.825822624330087,
      "grad_norm": 0.7245579957962036,
      "learning_rate": 7.747339268614234e-07,
      "loss": 0.692,
      "step": 28661
    },
    {
      "epoch": 0.8258514377917363,
      "grad_norm": 0.7492696046829224,
      "learning_rate": 7.744844571280901e-07,
      "loss": 0.6814,
      "step": 28662
    },
    {
      "epoch": 0.8258802512533856,
      "grad_norm": 0.726288378238678,
      "learning_rate": 7.74235024194771e-07,
      "loss": 0.6835,
      "step": 28663
    },
    {
      "epoch": 0.8259090647150349,
      "grad_norm": 0.7176610827445984,
      "learning_rate": 7.739856280636432e-07,
      "loss": 0.6874,
      "step": 28664
    },
    {
      "epoch": 0.8259378781766842,
      "grad_norm": 0.7084599733352661,
      "learning_rate": 7.737362687368749e-07,
      "loss": 0.6961,
      "step": 28665
    },
    {
      "epoch": 0.8259666916383335,
      "grad_norm": 0.7200964093208313,
      "learning_rate": 7.7348694621664e-07,
      "loss": 0.7126,
      "step": 28666
    },
    {
      "epoch": 0.8259955050999828,
      "grad_norm": 0.7151501178741455,
      "learning_rate": 7.732376605051079e-07,
      "loss": 0.6659,
      "step": 28667
    },
    {
      "epoch": 0.8260243185616319,
      "grad_norm": 0.720195472240448,
      "learning_rate": 7.729884116044517e-07,
      "loss": 0.6727,
      "step": 28668
    },
    {
      "epoch": 0.8260531320232812,
      "grad_norm": 0.7196551561355591,
      "learning_rate": 7.727391995168393e-07,
      "loss": 0.6766,
      "step": 28669
    },
    {
      "epoch": 0.8260819454849305,
      "grad_norm": 0.7134928703308105,
      "learning_rate": 7.724900242444433e-07,
      "loss": 0.6774,
      "step": 28670
    },
    {
      "epoch": 0.8261107589465798,
      "grad_norm": 0.7229552268981934,
      "learning_rate": 7.722408857894342e-07,
      "loss": 0.6939,
      "step": 28671
    },
    {
      "epoch": 0.8261395724082291,
      "grad_norm": 0.742222249507904,
      "learning_rate": 7.719917841539798e-07,
      "loss": 0.6903,
      "step": 28672
    },
    {
      "epoch": 0.8261683858698784,
      "grad_norm": 0.6968215703964233,
      "learning_rate": 7.71742719340251e-07,
      "loss": 0.687,
      "step": 28673
    },
    {
      "epoch": 0.8261971993315277,
      "grad_norm": 0.7417076826095581,
      "learning_rate": 7.714936913504162e-07,
      "loss": 0.6897,
      "step": 28674
    },
    {
      "epoch": 0.826226012793177,
      "grad_norm": 0.7297963500022888,
      "learning_rate": 7.712447001866436e-07,
      "loss": 0.6941,
      "step": 28675
    },
    {
      "epoch": 0.8262548262548263,
      "grad_norm": 0.7354097366333008,
      "learning_rate": 7.709957458511031e-07,
      "loss": 0.6867,
      "step": 28676
    },
    {
      "epoch": 0.8262836397164756,
      "grad_norm": 0.7068871855735779,
      "learning_rate": 7.707468283459629e-07,
      "loss": 0.6862,
      "step": 28677
    },
    {
      "epoch": 0.8263124531781249,
      "grad_norm": 0.7318645119667053,
      "learning_rate": 7.704979476733887e-07,
      "loss": 0.6957,
      "step": 28678
    },
    {
      "epoch": 0.8263412666397741,
      "grad_norm": 0.7181503176689148,
      "learning_rate": 7.702491038355514e-07,
      "loss": 0.6711,
      "step": 28679
    },
    {
      "epoch": 0.8263700801014234,
      "grad_norm": 0.7176823019981384,
      "learning_rate": 7.700002968346143e-07,
      "loss": 0.6924,
      "step": 28680
    },
    {
      "epoch": 0.8263988935630727,
      "grad_norm": 0.7450169920921326,
      "learning_rate": 7.697515266727479e-07,
      "loss": 0.6822,
      "step": 28681
    },
    {
      "epoch": 0.8264277070247219,
      "grad_norm": 0.7142382860183716,
      "learning_rate": 7.695027933521154e-07,
      "loss": 0.6871,
      "step": 28682
    },
    {
      "epoch": 0.8264565204863712,
      "grad_norm": 0.7404525279998779,
      "learning_rate": 7.692540968748857e-07,
      "loss": 0.6756,
      "step": 28683
    },
    {
      "epoch": 0.8264853339480205,
      "grad_norm": 0.7129675149917603,
      "learning_rate": 7.690054372432232e-07,
      "loss": 0.6798,
      "step": 28684
    },
    {
      "epoch": 0.8265141474096698,
      "grad_norm": 0.7094569206237793,
      "learning_rate": 7.687568144592944e-07,
      "loss": 0.6709,
      "step": 28685
    },
    {
      "epoch": 0.8265429608713191,
      "grad_norm": 0.7229377627372742,
      "learning_rate": 7.685082285252632e-07,
      "loss": 0.6465,
      "step": 28686
    },
    {
      "epoch": 0.8265717743329684,
      "grad_norm": 0.6945552229881287,
      "learning_rate": 7.682596794432951e-07,
      "loss": 0.6649,
      "step": 28687
    },
    {
      "epoch": 0.8266005877946176,
      "grad_norm": 0.7159227132797241,
      "learning_rate": 7.680111672155549e-07,
      "loss": 0.6814,
      "step": 28688
    },
    {
      "epoch": 0.8266294012562669,
      "grad_norm": 0.7257715463638306,
      "learning_rate": 7.677626918442083e-07,
      "loss": 0.6937,
      "step": 28689
    },
    {
      "epoch": 0.8266582147179162,
      "grad_norm": 0.7138848900794983,
      "learning_rate": 7.675142533314173e-07,
      "loss": 0.6618,
      "step": 28690
    },
    {
      "epoch": 0.8266870281795655,
      "grad_norm": 0.7390158772468567,
      "learning_rate": 7.672658516793458e-07,
      "loss": 0.7028,
      "step": 28691
    },
    {
      "epoch": 0.8267158416412148,
      "grad_norm": 0.7440508008003235,
      "learning_rate": 7.670174868901586e-07,
      "loss": 0.6962,
      "step": 28692
    },
    {
      "epoch": 0.8267446551028641,
      "grad_norm": 0.7380315661430359,
      "learning_rate": 7.667691589660165e-07,
      "loss": 0.6807,
      "step": 28693
    },
    {
      "epoch": 0.8267734685645134,
      "grad_norm": 0.7292938828468323,
      "learning_rate": 7.665208679090847e-07,
      "loss": 0.6729,
      "step": 28694
    },
    {
      "epoch": 0.8268022820261626,
      "grad_norm": 0.7092307806015015,
      "learning_rate": 7.662726137215226e-07,
      "loss": 0.6969,
      "step": 28695
    },
    {
      "epoch": 0.8268310954878119,
      "grad_norm": 0.7121194005012512,
      "learning_rate": 7.660243964054959e-07,
      "loss": 0.6639,
      "step": 28696
    },
    {
      "epoch": 0.8268599089494612,
      "grad_norm": 0.7402951717376709,
      "learning_rate": 7.657762159631627e-07,
      "loss": 0.6917,
      "step": 28697
    },
    {
      "epoch": 0.8268887224111104,
      "grad_norm": 0.7250449657440186,
      "learning_rate": 7.655280723966868e-07,
      "loss": 0.6934,
      "step": 28698
    },
    {
      "epoch": 0.8269175358727597,
      "grad_norm": 0.7204811573028564,
      "learning_rate": 7.65279965708226e-07,
      "loss": 0.6721,
      "step": 28699
    },
    {
      "epoch": 0.826946349334409,
      "grad_norm": 0.7282282710075378,
      "learning_rate": 7.650318958999464e-07,
      "loss": 0.668,
      "step": 28700
    },
    {
      "epoch": 0.8269751627960583,
      "grad_norm": 0.7073618173599243,
      "learning_rate": 7.647838629740045e-07,
      "loss": 0.6631,
      "step": 28701
    },
    {
      "epoch": 0.8270039762577076,
      "grad_norm": 0.7510277032852173,
      "learning_rate": 7.645358669325626e-07,
      "loss": 0.6819,
      "step": 28702
    },
    {
      "epoch": 0.8270327897193569,
      "grad_norm": 0.7210621237754822,
      "learning_rate": 7.642879077777776e-07,
      "loss": 0.6786,
      "step": 28703
    },
    {
      "epoch": 0.8270616031810062,
      "grad_norm": 0.7338146567344666,
      "learning_rate": 7.640399855118125e-07,
      "loss": 0.6895,
      "step": 28704
    },
    {
      "epoch": 0.8270904166426555,
      "grad_norm": 0.7174311876296997,
      "learning_rate": 7.637921001368232e-07,
      "loss": 0.6537,
      "step": 28705
    },
    {
      "epoch": 0.8271192301043048,
      "grad_norm": 0.6964939832687378,
      "learning_rate": 7.635442516549718e-07,
      "loss": 0.6845,
      "step": 28706
    },
    {
      "epoch": 0.8271480435659541,
      "grad_norm": 0.7223250865936279,
      "learning_rate": 7.632964400684134e-07,
      "loss": 0.6637,
      "step": 28707
    },
    {
      "epoch": 0.8271768570276034,
      "grad_norm": 0.7337855696678162,
      "learning_rate": 7.630486653793079e-07,
      "loss": 0.6886,
      "step": 28708
    },
    {
      "epoch": 0.8272056704892525,
      "grad_norm": 0.7249306440353394,
      "learning_rate": 7.62800927589814e-07,
      "loss": 0.669,
      "step": 28709
    },
    {
      "epoch": 0.8272344839509018,
      "grad_norm": 0.7102347612380981,
      "learning_rate": 7.62553226702088e-07,
      "loss": 0.6631,
      "step": 28710
    },
    {
      "epoch": 0.8272632974125511,
      "grad_norm": 0.7388527989387512,
      "learning_rate": 7.623055627182868e-07,
      "loss": 0.7021,
      "step": 28711
    },
    {
      "epoch": 0.8272921108742004,
      "grad_norm": 0.706970751285553,
      "learning_rate": 7.620579356405683e-07,
      "loss": 0.6803,
      "step": 28712
    },
    {
      "epoch": 0.8273209243358497,
      "grad_norm": 0.7325886487960815,
      "learning_rate": 7.618103454710901e-07,
      "loss": 0.6769,
      "step": 28713
    },
    {
      "epoch": 0.827349737797499,
      "grad_norm": 0.7277024984359741,
      "learning_rate": 7.615627922120055e-07,
      "loss": 0.695,
      "step": 28714
    },
    {
      "epoch": 0.8273785512591483,
      "grad_norm": 0.733110249042511,
      "learning_rate": 7.613152758654741e-07,
      "loss": 0.7076,
      "step": 28715
    },
    {
      "epoch": 0.8274073647207976,
      "grad_norm": 0.7120025157928467,
      "learning_rate": 7.610677964336477e-07,
      "loss": 0.6788,
      "step": 28716
    },
    {
      "epoch": 0.8274361781824469,
      "grad_norm": 0.719848096370697,
      "learning_rate": 7.608203539186842e-07,
      "loss": 0.6761,
      "step": 28717
    },
    {
      "epoch": 0.8274649916440961,
      "grad_norm": 0.7148953676223755,
      "learning_rate": 7.605729483227375e-07,
      "loss": 0.6808,
      "step": 28718
    },
    {
      "epoch": 0.8274938051057454,
      "grad_norm": 0.7421728372573853,
      "learning_rate": 7.60325579647963e-07,
      "loss": 0.7033,
      "step": 28719
    },
    {
      "epoch": 0.8275226185673947,
      "grad_norm": 0.7829330563545227,
      "learning_rate": 7.600782478965141e-07,
      "loss": 0.7027,
      "step": 28720
    },
    {
      "epoch": 0.827551432029044,
      "grad_norm": 0.7301638126373291,
      "learning_rate": 7.598309530705462e-07,
      "loss": 0.687,
      "step": 28721
    },
    {
      "epoch": 0.8275802454906932,
      "grad_norm": 0.723947286605835,
      "learning_rate": 7.595836951722107e-07,
      "loss": 0.6953,
      "step": 28722
    },
    {
      "epoch": 0.8276090589523425,
      "grad_norm": 0.7280598282814026,
      "learning_rate": 7.593364742036624e-07,
      "loss": 0.6771,
      "step": 28723
    },
    {
      "epoch": 0.8276378724139918,
      "grad_norm": 0.7337408661842346,
      "learning_rate": 7.590892901670543e-07,
      "loss": 0.6902,
      "step": 28724
    },
    {
      "epoch": 0.8276666858756411,
      "grad_norm": 0.7331892251968384,
      "learning_rate": 7.588421430645404e-07,
      "loss": 0.6945,
      "step": 28725
    },
    {
      "epoch": 0.8276954993372904,
      "grad_norm": 0.730637788772583,
      "learning_rate": 7.585950328982705e-07,
      "loss": 0.687,
      "step": 28726
    },
    {
      "epoch": 0.8277243127989397,
      "grad_norm": 0.701768696308136,
      "learning_rate": 7.583479596703991e-07,
      "loss": 0.6854,
      "step": 28727
    },
    {
      "epoch": 0.827753126260589,
      "grad_norm": 0.7266085743904114,
      "learning_rate": 7.581009233830761e-07,
      "loss": 0.6633,
      "step": 28728
    },
    {
      "epoch": 0.8277819397222382,
      "grad_norm": 0.7377723455429077,
      "learning_rate": 7.578539240384536e-07,
      "loss": 0.7093,
      "step": 28729
    },
    {
      "epoch": 0.8278107531838875,
      "grad_norm": 0.7178870439529419,
      "learning_rate": 7.576069616386838e-07,
      "loss": 0.6755,
      "step": 28730
    },
    {
      "epoch": 0.8278395666455368,
      "grad_norm": 0.7148408889770508,
      "learning_rate": 7.573600361859152e-07,
      "loss": 0.6758,
      "step": 28731
    },
    {
      "epoch": 0.8278683801071861,
      "grad_norm": 0.7177572250366211,
      "learning_rate": 7.571131476823013e-07,
      "loss": 0.6818,
      "step": 28732
    },
    {
      "epoch": 0.8278971935688354,
      "grad_norm": 0.7069646120071411,
      "learning_rate": 7.568662961299889e-07,
      "loss": 0.7011,
      "step": 28733
    },
    {
      "epoch": 0.8279260070304847,
      "grad_norm": 0.7388742566108704,
      "learning_rate": 7.566194815311312e-07,
      "loss": 0.6662,
      "step": 28734
    },
    {
      "epoch": 0.827954820492134,
      "grad_norm": 0.7198246717453003,
      "learning_rate": 7.563727038878732e-07,
      "loss": 0.6845,
      "step": 28735
    },
    {
      "epoch": 0.8279836339537832,
      "grad_norm": 0.6988127827644348,
      "learning_rate": 7.561259632023693e-07,
      "loss": 0.6764,
      "step": 28736
    },
    {
      "epoch": 0.8280124474154325,
      "grad_norm": 0.7129178643226624,
      "learning_rate": 7.558792594767644e-07,
      "loss": 0.666,
      "step": 28737
    },
    {
      "epoch": 0.8280412608770817,
      "grad_norm": 0.7291674613952637,
      "learning_rate": 7.556325927132102e-07,
      "loss": 0.7016,
      "step": 28738
    },
    {
      "epoch": 0.828070074338731,
      "grad_norm": 0.727371096611023,
      "learning_rate": 7.553859629138521e-07,
      "loss": 0.6852,
      "step": 28739
    },
    {
      "epoch": 0.8280988878003803,
      "grad_norm": 0.727760910987854,
      "learning_rate": 7.551393700808407e-07,
      "loss": 0.692,
      "step": 28740
    },
    {
      "epoch": 0.8281277012620296,
      "grad_norm": 0.7119631767272949,
      "learning_rate": 7.548928142163203e-07,
      "loss": 0.6769,
      "step": 28741
    },
    {
      "epoch": 0.8281565147236789,
      "grad_norm": 0.7205119132995605,
      "learning_rate": 7.546462953224415e-07,
      "loss": 0.656,
      "step": 28742
    },
    {
      "epoch": 0.8281853281853282,
      "grad_norm": 0.7119359970092773,
      "learning_rate": 7.54399813401348e-07,
      "loss": 0.6856,
      "step": 28743
    },
    {
      "epoch": 0.8282141416469775,
      "grad_norm": 0.7418450117111206,
      "learning_rate": 7.5415336845519e-07,
      "loss": 0.7115,
      "step": 28744
    },
    {
      "epoch": 0.8282429551086268,
      "grad_norm": 0.7133148312568665,
      "learning_rate": 7.539069604861099e-07,
      "loss": 0.6872,
      "step": 28745
    },
    {
      "epoch": 0.8282717685702761,
      "grad_norm": 0.7172723412513733,
      "learning_rate": 7.536605894962562e-07,
      "loss": 0.6751,
      "step": 28746
    },
    {
      "epoch": 0.8283005820319254,
      "grad_norm": 0.7182271480560303,
      "learning_rate": 7.534142554877738e-07,
      "loss": 0.6887,
      "step": 28747
    },
    {
      "epoch": 0.8283293954935746,
      "grad_norm": 0.7450639605522156,
      "learning_rate": 7.531679584628093e-07,
      "loss": 0.6795,
      "step": 28748
    },
    {
      "epoch": 0.8283582089552238,
      "grad_norm": 0.7298884987831116,
      "learning_rate": 7.529216984235055e-07,
      "loss": 0.6535,
      "step": 28749
    },
    {
      "epoch": 0.8283870224168731,
      "grad_norm": 0.7339590191841125,
      "learning_rate": 7.526754753720084e-07,
      "loss": 0.6801,
      "step": 28750
    },
    {
      "epoch": 0.8284158358785224,
      "grad_norm": 0.7167612314224243,
      "learning_rate": 7.524292893104629e-07,
      "loss": 0.679,
      "step": 28751
    },
    {
      "epoch": 0.8284446493401717,
      "grad_norm": 0.7568314671516418,
      "learning_rate": 7.521831402410113e-07,
      "loss": 0.6958,
      "step": 28752
    },
    {
      "epoch": 0.828473462801821,
      "grad_norm": 0.7382767796516418,
      "learning_rate": 7.519370281657995e-07,
      "loss": 0.7075,
      "step": 28753
    },
    {
      "epoch": 0.8285022762634703,
      "grad_norm": 0.7199894785881042,
      "learning_rate": 7.516909530869687e-07,
      "loss": 0.6742,
      "step": 28754
    },
    {
      "epoch": 0.8285310897251196,
      "grad_norm": 0.7266822457313538,
      "learning_rate": 7.514449150066638e-07,
      "loss": 0.6999,
      "step": 28755
    },
    {
      "epoch": 0.8285599031867689,
      "grad_norm": 0.7110693454742432,
      "learning_rate": 7.511989139270253e-07,
      "loss": 0.6726,
      "step": 28756
    },
    {
      "epoch": 0.8285887166484182,
      "grad_norm": 0.7298365235328674,
      "learning_rate": 7.509529498501988e-07,
      "loss": 0.6694,
      "step": 28757
    },
    {
      "epoch": 0.8286175301100674,
      "grad_norm": 0.7351731657981873,
      "learning_rate": 7.50707022778323e-07,
      "loss": 0.697,
      "step": 28758
    },
    {
      "epoch": 0.8286463435717167,
      "grad_norm": 0.7280498743057251,
      "learning_rate": 7.504611327135414e-07,
      "loss": 0.7181,
      "step": 28759
    },
    {
      "epoch": 0.828675157033366,
      "grad_norm": 0.7436148524284363,
      "learning_rate": 7.502152796579954e-07,
      "loss": 0.6802,
      "step": 28760
    },
    {
      "epoch": 0.8287039704950153,
      "grad_norm": 0.7124040722846985,
      "learning_rate": 7.499694636138272e-07,
      "loss": 0.6602,
      "step": 28761
    },
    {
      "epoch": 0.8287327839566645,
      "grad_norm": 0.728674590587616,
      "learning_rate": 7.497236845831751e-07,
      "loss": 0.6751,
      "step": 28762
    },
    {
      "epoch": 0.8287615974183138,
      "grad_norm": 0.7394266128540039,
      "learning_rate": 7.494779425681825e-07,
      "loss": 0.6952,
      "step": 28763
    },
    {
      "epoch": 0.8287904108799631,
      "grad_norm": 0.7118693590164185,
      "learning_rate": 7.49232237570987e-07,
      "loss": 0.6531,
      "step": 28764
    },
    {
      "epoch": 0.8288192243416124,
      "grad_norm": 0.7256437540054321,
      "learning_rate": 7.489865695937304e-07,
      "loss": 0.6965,
      "step": 28765
    },
    {
      "epoch": 0.8288480378032617,
      "grad_norm": 0.7273404598236084,
      "learning_rate": 7.487409386385503e-07,
      "loss": 0.6636,
      "step": 28766
    },
    {
      "epoch": 0.828876851264911,
      "grad_norm": 0.7211886048316956,
      "learning_rate": 7.484953447075865e-07,
      "loss": 0.6722,
      "step": 28767
    },
    {
      "epoch": 0.8289056647265602,
      "grad_norm": 0.7426328659057617,
      "learning_rate": 7.482497878029799e-07,
      "loss": 0.6895,
      "step": 28768
    },
    {
      "epoch": 0.8289344781882095,
      "grad_norm": 0.7504976391792297,
      "learning_rate": 7.480042679268662e-07,
      "loss": 0.68,
      "step": 28769
    },
    {
      "epoch": 0.8289632916498588,
      "grad_norm": 0.7139580845832825,
      "learning_rate": 7.477587850813861e-07,
      "loss": 0.6944,
      "step": 28770
    },
    {
      "epoch": 0.8289921051115081,
      "grad_norm": 0.7314478158950806,
      "learning_rate": 7.475133392686751e-07,
      "loss": 0.6799,
      "step": 28771
    },
    {
      "epoch": 0.8290209185731574,
      "grad_norm": 0.7492403984069824,
      "learning_rate": 7.47267930490872e-07,
      "loss": 0.6665,
      "step": 28772
    },
    {
      "epoch": 0.8290497320348067,
      "grad_norm": 0.7357889413833618,
      "learning_rate": 7.470225587501139e-07,
      "loss": 0.7085,
      "step": 28773
    },
    {
      "epoch": 0.829078545496456,
      "grad_norm": 0.7294045686721802,
      "learning_rate": 7.467772240485394e-07,
      "loss": 0.6601,
      "step": 28774
    },
    {
      "epoch": 0.8291073589581053,
      "grad_norm": 0.7296888828277588,
      "learning_rate": 7.46531926388282e-07,
      "loss": 0.6983,
      "step": 28775
    },
    {
      "epoch": 0.8291361724197545,
      "grad_norm": 0.7205684185028076,
      "learning_rate": 7.462866657714813e-07,
      "loss": 0.6653,
      "step": 28776
    },
    {
      "epoch": 0.8291649858814037,
      "grad_norm": 0.7430093884468079,
      "learning_rate": 7.460414422002698e-07,
      "loss": 0.6623,
      "step": 28777
    },
    {
      "epoch": 0.829193799343053,
      "grad_norm": 0.7246084809303284,
      "learning_rate": 7.457962556767867e-07,
      "loss": 0.6882,
      "step": 28778
    },
    {
      "epoch": 0.8292226128047023,
      "grad_norm": 0.7075446844100952,
      "learning_rate": 7.455511062031639e-07,
      "loss": 0.6647,
      "step": 28779
    },
    {
      "epoch": 0.8292514262663516,
      "grad_norm": 0.7141609191894531,
      "learning_rate": 7.453059937815393e-07,
      "loss": 0.7027,
      "step": 28780
    },
    {
      "epoch": 0.8292802397280009,
      "grad_norm": 0.7344251871109009,
      "learning_rate": 7.450609184140456e-07,
      "loss": 0.6766,
      "step": 28781
    },
    {
      "epoch": 0.8293090531896502,
      "grad_norm": 0.7308998107910156,
      "learning_rate": 7.448158801028188e-07,
      "loss": 0.6863,
      "step": 28782
    },
    {
      "epoch": 0.8293378666512995,
      "grad_norm": 0.7267900109291077,
      "learning_rate": 7.445708788499895e-07,
      "loss": 0.6836,
      "step": 28783
    },
    {
      "epoch": 0.8293666801129488,
      "grad_norm": 0.7317128777503967,
      "learning_rate": 7.443259146576964e-07,
      "loss": 0.688,
      "step": 28784
    },
    {
      "epoch": 0.8293954935745981,
      "grad_norm": 0.7215599417686462,
      "learning_rate": 7.440809875280696e-07,
      "loss": 0.6961,
      "step": 28785
    },
    {
      "epoch": 0.8294243070362474,
      "grad_norm": 0.7346200942993164,
      "learning_rate": 7.438360974632442e-07,
      "loss": 0.69,
      "step": 28786
    },
    {
      "epoch": 0.8294531204978967,
      "grad_norm": 0.7314306497573853,
      "learning_rate": 7.435912444653503e-07,
      "loss": 0.6839,
      "step": 28787
    },
    {
      "epoch": 0.829481933959546,
      "grad_norm": 0.7226257920265198,
      "learning_rate": 7.433464285365222e-07,
      "loss": 0.687,
      "step": 28788
    },
    {
      "epoch": 0.8295107474211951,
      "grad_norm": 0.6985678672790527,
      "learning_rate": 7.431016496788923e-07,
      "loss": 0.6528,
      "step": 28789
    },
    {
      "epoch": 0.8295395608828444,
      "grad_norm": 0.7319417595863342,
      "learning_rate": 7.428569078945907e-07,
      "loss": 0.6865,
      "step": 28790
    },
    {
      "epoch": 0.8295683743444937,
      "grad_norm": 0.7344585061073303,
      "learning_rate": 7.426122031857514e-07,
      "loss": 0.6959,
      "step": 28791
    },
    {
      "epoch": 0.829597187806143,
      "grad_norm": 0.7233390808105469,
      "learning_rate": 7.423675355545023e-07,
      "loss": 0.673,
      "step": 28792
    },
    {
      "epoch": 0.8296260012677923,
      "grad_norm": 0.7192974090576172,
      "learning_rate": 7.421229050029777e-07,
      "loss": 0.7012,
      "step": 28793
    },
    {
      "epoch": 0.8296548147294416,
      "grad_norm": 0.7255287170410156,
      "learning_rate": 7.418783115333045e-07,
      "loss": 0.6804,
      "step": 28794
    },
    {
      "epoch": 0.8296836281910909,
      "grad_norm": 0.7188823223114014,
      "learning_rate": 7.416337551476149e-07,
      "loss": 0.6682,
      "step": 28795
    },
    {
      "epoch": 0.8297124416527402,
      "grad_norm": 0.7314239740371704,
      "learning_rate": 7.413892358480385e-07,
      "loss": 0.6913,
      "step": 28796
    },
    {
      "epoch": 0.8297412551143895,
      "grad_norm": 0.7532740235328674,
      "learning_rate": 7.411447536367056e-07,
      "loss": 0.6914,
      "step": 28797
    },
    {
      "epoch": 0.8297700685760387,
      "grad_norm": 0.7147694230079651,
      "learning_rate": 7.409003085157435e-07,
      "loss": 0.6805,
      "step": 28798
    },
    {
      "epoch": 0.829798882037688,
      "grad_norm": 0.7113422751426697,
      "learning_rate": 7.406559004872837e-07,
      "loss": 0.6489,
      "step": 28799
    },
    {
      "epoch": 0.8298276954993373,
      "grad_norm": 0.7382357716560364,
      "learning_rate": 7.40411529553452e-07,
      "loss": 0.7013,
      "step": 28800
    },
    {
      "epoch": 0.8298565089609866,
      "grad_norm": 0.7092444896697998,
      "learning_rate": 7.401671957163787e-07,
      "loss": 0.6828,
      "step": 28801
    },
    {
      "epoch": 0.8298853224226359,
      "grad_norm": 0.714627206325531,
      "learning_rate": 7.399228989781898e-07,
      "loss": 0.6855,
      "step": 28802
    },
    {
      "epoch": 0.8299141358842851,
      "grad_norm": 0.7142435908317566,
      "learning_rate": 7.396786393410155e-07,
      "loss": 0.6892,
      "step": 28803
    },
    {
      "epoch": 0.8299429493459344,
      "grad_norm": 0.6976220011711121,
      "learning_rate": 7.3943441680698e-07,
      "loss": 0.661,
      "step": 28804
    },
    {
      "epoch": 0.8299717628075837,
      "grad_norm": 0.7100330591201782,
      "learning_rate": 7.391902313782117e-07,
      "loss": 0.697,
      "step": 28805
    },
    {
      "epoch": 0.830000576269233,
      "grad_norm": 0.7187550067901611,
      "learning_rate": 7.389460830568385e-07,
      "loss": 0.665,
      "step": 28806
    },
    {
      "epoch": 0.8300293897308822,
      "grad_norm": 0.7204616069793701,
      "learning_rate": 7.387019718449839e-07,
      "loss": 0.6865,
      "step": 28807
    },
    {
      "epoch": 0.8300582031925315,
      "grad_norm": 0.7022334337234497,
      "learning_rate": 7.384578977447759e-07,
      "loss": 0.6911,
      "step": 28808
    },
    {
      "epoch": 0.8300870166541808,
      "grad_norm": 0.7199200391769409,
      "learning_rate": 7.382138607583405e-07,
      "loss": 0.6828,
      "step": 28809
    },
    {
      "epoch": 0.8301158301158301,
      "grad_norm": 0.7151631712913513,
      "learning_rate": 7.379698608878011e-07,
      "loss": 0.6753,
      "step": 28810
    },
    {
      "epoch": 0.8301446435774794,
      "grad_norm": 0.7228091359138489,
      "learning_rate": 7.377258981352836e-07,
      "loss": 0.6832,
      "step": 28811
    },
    {
      "epoch": 0.8301734570391287,
      "grad_norm": 0.7363631725311279,
      "learning_rate": 7.374819725029142e-07,
      "loss": 0.6654,
      "step": 28812
    },
    {
      "epoch": 0.830202270500778,
      "grad_norm": 0.707105815410614,
      "learning_rate": 7.372380839928145e-07,
      "loss": 0.6484,
      "step": 28813
    },
    {
      "epoch": 0.8302310839624273,
      "grad_norm": 0.7203279137611389,
      "learning_rate": 7.369942326071111e-07,
      "loss": 0.6742,
      "step": 28814
    },
    {
      "epoch": 0.8302598974240766,
      "grad_norm": 0.7226201891899109,
      "learning_rate": 7.367504183479257e-07,
      "loss": 0.6698,
      "step": 28815
    },
    {
      "epoch": 0.8302887108857258,
      "grad_norm": 0.7311261892318726,
      "learning_rate": 7.365066412173833e-07,
      "loss": 0.7028,
      "step": 28816
    },
    {
      "epoch": 0.830317524347375,
      "grad_norm": 0.7152165174484253,
      "learning_rate": 7.362629012176048e-07,
      "loss": 0.6879,
      "step": 28817
    },
    {
      "epoch": 0.8303463378090243,
      "grad_norm": 0.7256792783737183,
      "learning_rate": 7.360191983507154e-07,
      "loss": 0.6803,
      "step": 28818
    },
    {
      "epoch": 0.8303751512706736,
      "grad_norm": 0.7371349930763245,
      "learning_rate": 7.357755326188343e-07,
      "loss": 0.6749,
      "step": 28819
    },
    {
      "epoch": 0.8304039647323229,
      "grad_norm": 0.7266667485237122,
      "learning_rate": 7.355319040240877e-07,
      "loss": 0.6574,
      "step": 28820
    },
    {
      "epoch": 0.8304327781939722,
      "grad_norm": 0.7407572269439697,
      "learning_rate": 7.352883125685944e-07,
      "loss": 0.6789,
      "step": 28821
    },
    {
      "epoch": 0.8304615916556215,
      "grad_norm": 0.7221948504447937,
      "learning_rate": 7.350447582544778e-07,
      "loss": 0.6827,
      "step": 28822
    },
    {
      "epoch": 0.8304904051172708,
      "grad_norm": 0.7228736281394958,
      "learning_rate": 7.348012410838568e-07,
      "loss": 0.7122,
      "step": 28823
    },
    {
      "epoch": 0.8305192185789201,
      "grad_norm": 0.717548131942749,
      "learning_rate": 7.345577610588545e-07,
      "loss": 0.6627,
      "step": 28824
    },
    {
      "epoch": 0.8305480320405694,
      "grad_norm": 0.7054880261421204,
      "learning_rate": 7.343143181815893e-07,
      "loss": 0.6597,
      "step": 28825
    },
    {
      "epoch": 0.8305768455022187,
      "grad_norm": 0.7113574743270874,
      "learning_rate": 7.340709124541817e-07,
      "loss": 0.6937,
      "step": 28826
    },
    {
      "epoch": 0.830605658963868,
      "grad_norm": 0.7443816661834717,
      "learning_rate": 7.338275438787539e-07,
      "loss": 0.6913,
      "step": 28827
    },
    {
      "epoch": 0.8306344724255172,
      "grad_norm": 0.7414054274559021,
      "learning_rate": 7.335842124574222e-07,
      "loss": 0.6822,
      "step": 28828
    },
    {
      "epoch": 0.8306632858871665,
      "grad_norm": 0.7418216466903687,
      "learning_rate": 7.33340918192308e-07,
      "loss": 0.6969,
      "step": 28829
    },
    {
      "epoch": 0.8306920993488157,
      "grad_norm": 0.7231153249740601,
      "learning_rate": 7.330976610855284e-07,
      "loss": 0.6839,
      "step": 28830
    },
    {
      "epoch": 0.830720912810465,
      "grad_norm": 0.7207005023956299,
      "learning_rate": 7.32854441139203e-07,
      "loss": 0.6996,
      "step": 28831
    },
    {
      "epoch": 0.8307497262721143,
      "grad_norm": 0.7114543318748474,
      "learning_rate": 7.326112583554496e-07,
      "loss": 0.6888,
      "step": 28832
    },
    {
      "epoch": 0.8307785397337636,
      "grad_norm": 0.7260099649429321,
      "learning_rate": 7.323681127363874e-07,
      "loss": 0.6885,
      "step": 28833
    },
    {
      "epoch": 0.8308073531954129,
      "grad_norm": 0.6963470578193665,
      "learning_rate": 7.321250042841316e-07,
      "loss": 0.6673,
      "step": 28834
    },
    {
      "epoch": 0.8308361666570622,
      "grad_norm": 0.7279927134513855,
      "learning_rate": 7.31881933000802e-07,
      "loss": 0.6905,
      "step": 28835
    },
    {
      "epoch": 0.8308649801187115,
      "grad_norm": 0.75055330991745,
      "learning_rate": 7.316388988885137e-07,
      "loss": 0.6816,
      "step": 28836
    },
    {
      "epoch": 0.8308937935803608,
      "grad_norm": 0.7925270795822144,
      "learning_rate": 7.313959019493843e-07,
      "loss": 0.7018,
      "step": 28837
    },
    {
      "epoch": 0.83092260704201,
      "grad_norm": 0.7356323599815369,
      "learning_rate": 7.311529421855284e-07,
      "loss": 0.6932,
      "step": 28838
    },
    {
      "epoch": 0.8309514205036593,
      "grad_norm": 0.7275372743606567,
      "learning_rate": 7.309100195990643e-07,
      "loss": 0.6803,
      "step": 28839
    },
    {
      "epoch": 0.8309802339653086,
      "grad_norm": 0.72700035572052,
      "learning_rate": 7.306671341921051e-07,
      "loss": 0.6795,
      "step": 28840
    },
    {
      "epoch": 0.8310090474269579,
      "grad_norm": 0.7305644750595093,
      "learning_rate": 7.304242859667687e-07,
      "loss": 0.6883,
      "step": 28841
    },
    {
      "epoch": 0.8310378608886072,
      "grad_norm": 0.7287461757659912,
      "learning_rate": 7.301814749251679e-07,
      "loss": 0.6875,
      "step": 28842
    },
    {
      "epoch": 0.8310666743502564,
      "grad_norm": 0.7408763766288757,
      "learning_rate": 7.299387010694175e-07,
      "loss": 0.6859,
      "step": 28843
    },
    {
      "epoch": 0.8310954878119057,
      "grad_norm": 0.7149055600166321,
      "learning_rate": 7.296959644016332e-07,
      "loss": 0.6732,
      "step": 28844
    },
    {
      "epoch": 0.831124301273555,
      "grad_norm": 0.7492726445198059,
      "learning_rate": 7.294532649239289e-07,
      "loss": 0.6857,
      "step": 28845
    },
    {
      "epoch": 0.8311531147352043,
      "grad_norm": 0.7364996075630188,
      "learning_rate": 7.292106026384172e-07,
      "loss": 0.6956,
      "step": 28846
    },
    {
      "epoch": 0.8311819281968535,
      "grad_norm": 0.7130023241043091,
      "learning_rate": 7.289679775472125e-07,
      "loss": 0.6711,
      "step": 28847
    },
    {
      "epoch": 0.8312107416585028,
      "grad_norm": 0.7292895913124084,
      "learning_rate": 7.287253896524261e-07,
      "loss": 0.6567,
      "step": 28848
    },
    {
      "epoch": 0.8312395551201521,
      "grad_norm": 0.7043736577033997,
      "learning_rate": 7.284828389561721e-07,
      "loss": 0.686,
      "step": 28849
    },
    {
      "epoch": 0.8312683685818014,
      "grad_norm": 0.7423701286315918,
      "learning_rate": 7.282403254605636e-07,
      "loss": 0.6891,
      "step": 28850
    },
    {
      "epoch": 0.8312971820434507,
      "grad_norm": 0.7322592735290527,
      "learning_rate": 7.279978491677109e-07,
      "loss": 0.6891,
      "step": 28851
    },
    {
      "epoch": 0.8313259955051,
      "grad_norm": 0.7274309396743774,
      "learning_rate": 7.277554100797274e-07,
      "loss": 0.6996,
      "step": 28852
    },
    {
      "epoch": 0.8313548089667493,
      "grad_norm": 0.7267763018608093,
      "learning_rate": 7.27513008198722e-07,
      "loss": 0.6821,
      "step": 28853
    },
    {
      "epoch": 0.8313836224283986,
      "grad_norm": 0.7352450489997864,
      "learning_rate": 7.27270643526809e-07,
      "loss": 0.6689,
      "step": 28854
    },
    {
      "epoch": 0.8314124358900479,
      "grad_norm": 0.7519264817237854,
      "learning_rate": 7.270283160660951e-07,
      "loss": 0.728,
      "step": 28855
    },
    {
      "epoch": 0.8314412493516972,
      "grad_norm": 0.7391136884689331,
      "learning_rate": 7.26786025818696e-07,
      "loss": 0.6888,
      "step": 28856
    },
    {
      "epoch": 0.8314700628133463,
      "grad_norm": 0.7081487774848938,
      "learning_rate": 7.265437727867169e-07,
      "loss": 0.6874,
      "step": 28857
    },
    {
      "epoch": 0.8314988762749956,
      "grad_norm": 0.7211889624595642,
      "learning_rate": 7.263015569722715e-07,
      "loss": 0.6865,
      "step": 28858
    },
    {
      "epoch": 0.8315276897366449,
      "grad_norm": 0.7186380624771118,
      "learning_rate": 7.260593783774661e-07,
      "loss": 0.6644,
      "step": 28859
    },
    {
      "epoch": 0.8315565031982942,
      "grad_norm": 0.7355714440345764,
      "learning_rate": 7.258172370044119e-07,
      "loss": 0.6803,
      "step": 28860
    },
    {
      "epoch": 0.8315853166599435,
      "grad_norm": 0.7038397789001465,
      "learning_rate": 7.255751328552163e-07,
      "loss": 0.7036,
      "step": 28861
    },
    {
      "epoch": 0.8316141301215928,
      "grad_norm": 0.7288101315498352,
      "learning_rate": 7.25333065931989e-07,
      "loss": 0.6968,
      "step": 28862
    },
    {
      "epoch": 0.8316429435832421,
      "grad_norm": 0.7306252717971802,
      "learning_rate": 7.250910362368368e-07,
      "loss": 0.6898,
      "step": 28863
    },
    {
      "epoch": 0.8316717570448914,
      "grad_norm": 0.7256668210029602,
      "learning_rate": 7.248490437718686e-07,
      "loss": 0.6979,
      "step": 28864
    },
    {
      "epoch": 0.8317005705065407,
      "grad_norm": 0.7091861367225647,
      "learning_rate": 7.246070885391926e-07,
      "loss": 0.6864,
      "step": 28865
    },
    {
      "epoch": 0.83172938396819,
      "grad_norm": 0.7129870057106018,
      "learning_rate": 7.243651705409133e-07,
      "loss": 0.6686,
      "step": 28866
    },
    {
      "epoch": 0.8317581974298393,
      "grad_norm": 0.7061542868614197,
      "learning_rate": 7.2412328977914e-07,
      "loss": 0.6916,
      "step": 28867
    },
    {
      "epoch": 0.8317870108914885,
      "grad_norm": 0.7149056792259216,
      "learning_rate": 7.238814462559796e-07,
      "loss": 0.6755,
      "step": 28868
    },
    {
      "epoch": 0.8318158243531378,
      "grad_norm": 0.7030513882637024,
      "learning_rate": 7.236396399735357e-07,
      "loss": 0.6692,
      "step": 28869
    },
    {
      "epoch": 0.831844637814787,
      "grad_norm": 0.7246667742729187,
      "learning_rate": 7.233978709339162e-07,
      "loss": 0.6942,
      "step": 28870
    },
    {
      "epoch": 0.8318734512764363,
      "grad_norm": 0.7021811604499817,
      "learning_rate": 7.231561391392272e-07,
      "loss": 0.673,
      "step": 28871
    },
    {
      "epoch": 0.8319022647380856,
      "grad_norm": 0.7232382297515869,
      "learning_rate": 7.229144445915715e-07,
      "loss": 0.6968,
      "step": 28872
    },
    {
      "epoch": 0.8319310781997349,
      "grad_norm": 0.7180808186531067,
      "learning_rate": 7.226727872930573e-07,
      "loss": 0.6922,
      "step": 28873
    },
    {
      "epoch": 0.8319598916613842,
      "grad_norm": 0.7269206047058105,
      "learning_rate": 7.224311672457857e-07,
      "loss": 0.646,
      "step": 28874
    },
    {
      "epoch": 0.8319887051230335,
      "grad_norm": 0.7386382222175598,
      "learning_rate": 7.22189584451864e-07,
      "loss": 0.6912,
      "step": 28875
    },
    {
      "epoch": 0.8320175185846828,
      "grad_norm": 0.7361152768135071,
      "learning_rate": 7.219480389133931e-07,
      "loss": 0.6529,
      "step": 28876
    },
    {
      "epoch": 0.832046332046332,
      "grad_norm": 0.7242401242256165,
      "learning_rate": 7.2170653063248e-07,
      "loss": 0.6811,
      "step": 28877
    },
    {
      "epoch": 0.8320751455079813,
      "grad_norm": 0.7260377407073975,
      "learning_rate": 7.214650596112249e-07,
      "loss": 0.6727,
      "step": 28878
    },
    {
      "epoch": 0.8321039589696306,
      "grad_norm": 0.7358270883560181,
      "learning_rate": 7.212236258517324e-07,
      "loss": 0.678,
      "step": 28879
    },
    {
      "epoch": 0.8321327724312799,
      "grad_norm": 0.7325437664985657,
      "learning_rate": 7.209822293561047e-07,
      "loss": 0.6814,
      "step": 28880
    },
    {
      "epoch": 0.8321615858929292,
      "grad_norm": 0.7361072301864624,
      "learning_rate": 7.207408701264457e-07,
      "loss": 0.6682,
      "step": 28881
    },
    {
      "epoch": 0.8321903993545785,
      "grad_norm": 0.7388448715209961,
      "learning_rate": 7.204995481648547e-07,
      "loss": 0.6655,
      "step": 28882
    },
    {
      "epoch": 0.8322192128162278,
      "grad_norm": 0.72359699010849,
      "learning_rate": 7.202582634734356e-07,
      "loss": 0.6563,
      "step": 28883
    },
    {
      "epoch": 0.832248026277877,
      "grad_norm": 0.7133237719535828,
      "learning_rate": 7.20017016054288e-07,
      "loss": 0.6802,
      "step": 28884
    },
    {
      "epoch": 0.8322768397395263,
      "grad_norm": 0.711708664894104,
      "learning_rate": 7.197758059095139e-07,
      "loss": 0.6875,
      "step": 28885
    },
    {
      "epoch": 0.8323056532011756,
      "grad_norm": 0.7271198034286499,
      "learning_rate": 7.195346330412145e-07,
      "loss": 0.7003,
      "step": 28886
    },
    {
      "epoch": 0.8323344666628248,
      "grad_norm": 0.731256902217865,
      "learning_rate": 7.192934974514886e-07,
      "loss": 0.698,
      "step": 28887
    },
    {
      "epoch": 0.8323632801244741,
      "grad_norm": 0.7298883199691772,
      "learning_rate": 7.190523991424386e-07,
      "loss": 0.702,
      "step": 28888
    },
    {
      "epoch": 0.8323920935861234,
      "grad_norm": 0.7317258715629578,
      "learning_rate": 7.188113381161615e-07,
      "loss": 0.7044,
      "step": 28889
    },
    {
      "epoch": 0.8324209070477727,
      "grad_norm": 0.7308321595191956,
      "learning_rate": 7.185703143747597e-07,
      "loss": 0.6936,
      "step": 28890
    },
    {
      "epoch": 0.832449720509422,
      "grad_norm": 0.7511773109436035,
      "learning_rate": 7.183293279203279e-07,
      "loss": 0.6776,
      "step": 28891
    },
    {
      "epoch": 0.8324785339710713,
      "grad_norm": 0.7049068808555603,
      "learning_rate": 7.180883787549697e-07,
      "loss": 0.6845,
      "step": 28892
    },
    {
      "epoch": 0.8325073474327206,
      "grad_norm": 0.7242879867553711,
      "learning_rate": 7.1784746688078e-07,
      "loss": 0.6779,
      "step": 28893
    },
    {
      "epoch": 0.8325361608943699,
      "grad_norm": 0.710314929485321,
      "learning_rate": 7.176065922998598e-07,
      "loss": 0.6548,
      "step": 28894
    },
    {
      "epoch": 0.8325649743560192,
      "grad_norm": 0.7209511399269104,
      "learning_rate": 7.173657550143042e-07,
      "loss": 0.7033,
      "step": 28895
    },
    {
      "epoch": 0.8325937878176685,
      "grad_norm": 0.7240989804267883,
      "learning_rate": 7.171249550262132e-07,
      "loss": 0.6815,
      "step": 28896
    },
    {
      "epoch": 0.8326226012793176,
      "grad_norm": 0.7302938103675842,
      "learning_rate": 7.168841923376807e-07,
      "loss": 0.6979,
      "step": 28897
    },
    {
      "epoch": 0.8326514147409669,
      "grad_norm": 0.7261094450950623,
      "learning_rate": 7.166434669508072e-07,
      "loss": 0.6916,
      "step": 28898
    },
    {
      "epoch": 0.8326802282026162,
      "grad_norm": 0.742642879486084,
      "learning_rate": 7.164027788676858e-07,
      "loss": 0.6642,
      "step": 28899
    },
    {
      "epoch": 0.8327090416642655,
      "grad_norm": 0.7087109088897705,
      "learning_rate": 7.161621280904152e-07,
      "loss": 0.6539,
      "step": 28900
    },
    {
      "epoch": 0.8327378551259148,
      "grad_norm": 0.713428795337677,
      "learning_rate": 7.15921514621089e-07,
      "loss": 0.6874,
      "step": 28901
    },
    {
      "epoch": 0.8327666685875641,
      "grad_norm": 0.7233628630638123,
      "learning_rate": 7.156809384618047e-07,
      "loss": 0.6691,
      "step": 28902
    },
    {
      "epoch": 0.8327954820492134,
      "grad_norm": 0.7209185361862183,
      "learning_rate": 7.154403996146558e-07,
      "loss": 0.6755,
      "step": 28903
    },
    {
      "epoch": 0.8328242955108627,
      "grad_norm": 0.8170904517173767,
      "learning_rate": 7.151998980817393e-07,
      "loss": 0.676,
      "step": 28904
    },
    {
      "epoch": 0.832853108972512,
      "grad_norm": 0.7315351963043213,
      "learning_rate": 7.149594338651478e-07,
      "loss": 0.6983,
      "step": 28905
    },
    {
      "epoch": 0.8328819224341613,
      "grad_norm": 0.7456140518188477,
      "learning_rate": 7.147190069669769e-07,
      "loss": 0.6654,
      "step": 28906
    },
    {
      "epoch": 0.8329107358958106,
      "grad_norm": 0.7283255457878113,
      "learning_rate": 7.144786173893187e-07,
      "loss": 0.697,
      "step": 28907
    },
    {
      "epoch": 0.8329395493574598,
      "grad_norm": 0.7301376461982727,
      "learning_rate": 7.142382651342677e-07,
      "loss": 0.6978,
      "step": 28908
    },
    {
      "epoch": 0.8329683628191091,
      "grad_norm": 0.7250642776489258,
      "learning_rate": 7.139979502039185e-07,
      "loss": 0.6699,
      "step": 28909
    },
    {
      "epoch": 0.8329971762807584,
      "grad_norm": 0.7194370031356812,
      "learning_rate": 7.137576726003614e-07,
      "loss": 0.6823,
      "step": 28910
    },
    {
      "epoch": 0.8330259897424076,
      "grad_norm": 0.71969074010849,
      "learning_rate": 7.135174323256916e-07,
      "loss": 0.6874,
      "step": 28911
    },
    {
      "epoch": 0.8330548032040569,
      "grad_norm": 0.7108751535415649,
      "learning_rate": 7.132772293819984e-07,
      "loss": 0.6865,
      "step": 28912
    },
    {
      "epoch": 0.8330836166657062,
      "grad_norm": 0.7042509317398071,
      "learning_rate": 7.130370637713768e-07,
      "loss": 0.6867,
      "step": 28913
    },
    {
      "epoch": 0.8331124301273555,
      "grad_norm": 0.7254416346549988,
      "learning_rate": 7.127969354959164e-07,
      "loss": 0.6668,
      "step": 28914
    },
    {
      "epoch": 0.8331412435890048,
      "grad_norm": 0.7275904417037964,
      "learning_rate": 7.125568445577085e-07,
      "loss": 0.6872,
      "step": 28915
    },
    {
      "epoch": 0.8331700570506541,
      "grad_norm": 0.7257629632949829,
      "learning_rate": 7.123167909588446e-07,
      "loss": 0.7004,
      "step": 28916
    },
    {
      "epoch": 0.8331988705123033,
      "grad_norm": 0.7398699522018433,
      "learning_rate": 7.120767747014168e-07,
      "loss": 0.6772,
      "step": 28917
    },
    {
      "epoch": 0.8332276839739526,
      "grad_norm": 0.7204175591468811,
      "learning_rate": 7.118367957875128e-07,
      "loss": 0.6517,
      "step": 28918
    },
    {
      "epoch": 0.8332564974356019,
      "grad_norm": 0.7199852466583252,
      "learning_rate": 7.115968542192247e-07,
      "loss": 0.705,
      "step": 28919
    },
    {
      "epoch": 0.8332853108972512,
      "grad_norm": 0.7410869002342224,
      "learning_rate": 7.113569499986401e-07,
      "loss": 0.7176,
      "step": 28920
    },
    {
      "epoch": 0.8333141243589005,
      "grad_norm": 0.7377700209617615,
      "learning_rate": 7.111170831278502e-07,
      "loss": 0.6844,
      "step": 28921
    },
    {
      "epoch": 0.8333429378205498,
      "grad_norm": 0.7196758985519409,
      "learning_rate": 7.10877253608942e-07,
      "loss": 0.6913,
      "step": 28922
    },
    {
      "epoch": 0.8333717512821991,
      "grad_norm": 0.7370380759239197,
      "learning_rate": 7.10637461444006e-07,
      "loss": 0.6516,
      "step": 28923
    },
    {
      "epoch": 0.8334005647438483,
      "grad_norm": 0.7307515740394592,
      "learning_rate": 7.103977066351309e-07,
      "loss": 0.6699,
      "step": 28924
    },
    {
      "epoch": 0.8334293782054976,
      "grad_norm": 0.7255179286003113,
      "learning_rate": 7.101579891844024e-07,
      "loss": 0.7012,
      "step": 28925
    },
    {
      "epoch": 0.8334581916671469,
      "grad_norm": 0.7376047968864441,
      "learning_rate": 7.099183090939094e-07,
      "loss": 0.665,
      "step": 28926
    },
    {
      "epoch": 0.8334870051287961,
      "grad_norm": 0.7385253310203552,
      "learning_rate": 7.096786663657406e-07,
      "loss": 0.684,
      "step": 28927
    },
    {
      "epoch": 0.8335158185904454,
      "grad_norm": 0.719115674495697,
      "learning_rate": 7.094390610019807e-07,
      "loss": 0.661,
      "step": 28928
    },
    {
      "epoch": 0.8335446320520947,
      "grad_norm": 0.7457781434059143,
      "learning_rate": 7.091994930047181e-07,
      "loss": 0.6779,
      "step": 28929
    },
    {
      "epoch": 0.833573445513744,
      "grad_norm": 0.7234681248664856,
      "learning_rate": 7.08959962376039e-07,
      "loss": 0.685,
      "step": 28930
    },
    {
      "epoch": 0.8336022589753933,
      "grad_norm": 0.7087912559509277,
      "learning_rate": 7.087204691180289e-07,
      "loss": 0.6558,
      "step": 28931
    },
    {
      "epoch": 0.8336310724370426,
      "grad_norm": 0.7379653453826904,
      "learning_rate": 7.084810132327741e-07,
      "loss": 0.7149,
      "step": 28932
    },
    {
      "epoch": 0.8336598858986919,
      "grad_norm": 0.7472109794616699,
      "learning_rate": 7.082415947223592e-07,
      "loss": 0.6744,
      "step": 28933
    },
    {
      "epoch": 0.8336886993603412,
      "grad_norm": 0.7454600930213928,
      "learning_rate": 7.080022135888709e-07,
      "loss": 0.6959,
      "step": 28934
    },
    {
      "epoch": 0.8337175128219905,
      "grad_norm": 0.712541937828064,
      "learning_rate": 7.077628698343913e-07,
      "loss": 0.6813,
      "step": 28935
    },
    {
      "epoch": 0.8337463262836398,
      "grad_norm": 0.7403962016105652,
      "learning_rate": 7.075235634610084e-07,
      "loss": 0.6928,
      "step": 28936
    },
    {
      "epoch": 0.833775139745289,
      "grad_norm": 0.7502115368843079,
      "learning_rate": 7.072842944708025e-07,
      "loss": 0.6859,
      "step": 28937
    },
    {
      "epoch": 0.8338039532069382,
      "grad_norm": 0.7310363054275513,
      "learning_rate": 7.070450628658593e-07,
      "loss": 0.6813,
      "step": 28938
    },
    {
      "epoch": 0.8338327666685875,
      "grad_norm": 0.7119927406311035,
      "learning_rate": 7.068058686482626e-07,
      "loss": 0.6834,
      "step": 28939
    },
    {
      "epoch": 0.8338615801302368,
      "grad_norm": 0.7067298293113708,
      "learning_rate": 7.065667118200958e-07,
      "loss": 0.6682,
      "step": 28940
    },
    {
      "epoch": 0.8338903935918861,
      "grad_norm": 0.7647702097892761,
      "learning_rate": 7.063275923834406e-07,
      "loss": 0.702,
      "step": 28941
    },
    {
      "epoch": 0.8339192070535354,
      "grad_norm": 0.7382287979125977,
      "learning_rate": 7.060885103403803e-07,
      "loss": 0.6932,
      "step": 28942
    },
    {
      "epoch": 0.8339480205151847,
      "grad_norm": 0.7158991694450378,
      "learning_rate": 7.058494656929965e-07,
      "loss": 0.7067,
      "step": 28943
    },
    {
      "epoch": 0.833976833976834,
      "grad_norm": 0.7334153652191162,
      "learning_rate": 7.056104584433715e-07,
      "loss": 0.6963,
      "step": 28944
    },
    {
      "epoch": 0.8340056474384833,
      "grad_norm": 0.7313697934150696,
      "learning_rate": 7.053714885935858e-07,
      "loss": 0.6832,
      "step": 28945
    },
    {
      "epoch": 0.8340344609001326,
      "grad_norm": 0.7216899991035461,
      "learning_rate": 7.051325561457217e-07,
      "loss": 0.6658,
      "step": 28946
    },
    {
      "epoch": 0.8340632743617818,
      "grad_norm": 0.7507964372634888,
      "learning_rate": 7.048936611018609e-07,
      "loss": 0.6772,
      "step": 28947
    },
    {
      "epoch": 0.8340920878234311,
      "grad_norm": 0.7384761571884155,
      "learning_rate": 7.04654803464081e-07,
      "loss": 0.6959,
      "step": 28948
    },
    {
      "epoch": 0.8341209012850804,
      "grad_norm": 0.7299233675003052,
      "learning_rate": 7.044159832344654e-07,
      "loss": 0.6958,
      "step": 28949
    },
    {
      "epoch": 0.8341497147467297,
      "grad_norm": 0.7098830342292786,
      "learning_rate": 7.041772004150905e-07,
      "loss": 0.6644,
      "step": 28950
    },
    {
      "epoch": 0.8341785282083789,
      "grad_norm": 0.7242186665534973,
      "learning_rate": 7.039384550080402e-07,
      "loss": 0.6875,
      "step": 28951
    },
    {
      "epoch": 0.8342073416700282,
      "grad_norm": 0.7628321647644043,
      "learning_rate": 7.036997470153906e-07,
      "loss": 0.6724,
      "step": 28952
    },
    {
      "epoch": 0.8342361551316775,
      "grad_norm": 0.7151764631271362,
      "learning_rate": 7.034610764392219e-07,
      "loss": 0.6796,
      "step": 28953
    },
    {
      "epoch": 0.8342649685933268,
      "grad_norm": 0.7438299655914307,
      "learning_rate": 7.032224432816114e-07,
      "loss": 0.7044,
      "step": 28954
    },
    {
      "epoch": 0.8342937820549761,
      "grad_norm": 0.7150423526763916,
      "learning_rate": 7.029838475446394e-07,
      "loss": 0.6677,
      "step": 28955
    },
    {
      "epoch": 0.8343225955166254,
      "grad_norm": 0.7245270609855652,
      "learning_rate": 7.027452892303816e-07,
      "loss": 0.6773,
      "step": 28956
    },
    {
      "epoch": 0.8343514089782746,
      "grad_norm": 0.724241316318512,
      "learning_rate": 7.025067683409176e-07,
      "loss": 0.6824,
      "step": 28957
    },
    {
      "epoch": 0.8343802224399239,
      "grad_norm": 0.7274297475814819,
      "learning_rate": 7.02268284878323e-07,
      "loss": 0.6951,
      "step": 28958
    },
    {
      "epoch": 0.8344090359015732,
      "grad_norm": 0.7180221080780029,
      "learning_rate": 7.020298388446761e-07,
      "loss": 0.6879,
      "step": 28959
    },
    {
      "epoch": 0.8344378493632225,
      "grad_norm": 0.8611966967582703,
      "learning_rate": 7.01791430242052e-07,
      "loss": 0.6979,
      "step": 28960
    },
    {
      "epoch": 0.8344666628248718,
      "grad_norm": 0.7518422603607178,
      "learning_rate": 7.015530590725278e-07,
      "loss": 0.6811,
      "step": 28961
    },
    {
      "epoch": 0.8344954762865211,
      "grad_norm": 0.7165344953536987,
      "learning_rate": 7.013147253381797e-07,
      "loss": 0.6658,
      "step": 28962
    },
    {
      "epoch": 0.8345242897481704,
      "grad_norm": 0.7332323789596558,
      "learning_rate": 7.010764290410843e-07,
      "loss": 0.6577,
      "step": 28963
    },
    {
      "epoch": 0.8345531032098197,
      "grad_norm": 0.7219669818878174,
      "learning_rate": 7.00838170183315e-07,
      "loss": 0.6704,
      "step": 28964
    },
    {
      "epoch": 0.8345819166714689,
      "grad_norm": 0.7306843996047974,
      "learning_rate": 7.005999487669485e-07,
      "loss": 0.687,
      "step": 28965
    },
    {
      "epoch": 0.8346107301331182,
      "grad_norm": 0.7152778506278992,
      "learning_rate": 7.003617647940575e-07,
      "loss": 0.6726,
      "step": 28966
    },
    {
      "epoch": 0.8346395435947674,
      "grad_norm": 0.7222121357917786,
      "learning_rate": 7.001236182667176e-07,
      "loss": 0.673,
      "step": 28967
    },
    {
      "epoch": 0.8346683570564167,
      "grad_norm": 0.7274730205535889,
      "learning_rate": 6.998855091870038e-07,
      "loss": 0.6955,
      "step": 28968
    },
    {
      "epoch": 0.834697170518066,
      "grad_norm": 0.7215209007263184,
      "learning_rate": 6.996474375569872e-07,
      "loss": 0.7004,
      "step": 28969
    },
    {
      "epoch": 0.8347259839797153,
      "grad_norm": 0.731975793838501,
      "learning_rate": 6.99409403378744e-07,
      "loss": 0.6845,
      "step": 28970
    },
    {
      "epoch": 0.8347547974413646,
      "grad_norm": 0.7386131286621094,
      "learning_rate": 6.991714066543448e-07,
      "loss": 0.7103,
      "step": 28971
    },
    {
      "epoch": 0.8347836109030139,
      "grad_norm": 0.7189493775367737,
      "learning_rate": 6.989334473858645e-07,
      "loss": 0.6796,
      "step": 28972
    },
    {
      "epoch": 0.8348124243646632,
      "grad_norm": 0.7173219919204712,
      "learning_rate": 6.986955255753735e-07,
      "loss": 0.6686,
      "step": 28973
    },
    {
      "epoch": 0.8348412378263125,
      "grad_norm": 0.716673731803894,
      "learning_rate": 6.984576412249445e-07,
      "loss": 0.6688,
      "step": 28974
    },
    {
      "epoch": 0.8348700512879618,
      "grad_norm": 0.6930920481681824,
      "learning_rate": 6.982197943366492e-07,
      "loss": 0.679,
      "step": 28975
    },
    {
      "epoch": 0.8348988647496111,
      "grad_norm": 0.7363691926002502,
      "learning_rate": 6.979819849125608e-07,
      "loss": 0.6814,
      "step": 28976
    },
    {
      "epoch": 0.8349276782112603,
      "grad_norm": 0.712665319442749,
      "learning_rate": 6.977442129547479e-07,
      "loss": 0.6933,
      "step": 28977
    },
    {
      "epoch": 0.8349564916729095,
      "grad_norm": 0.7331464290618896,
      "learning_rate": 6.975064784652829e-07,
      "loss": 0.6841,
      "step": 28978
    },
    {
      "epoch": 0.8349853051345588,
      "grad_norm": 0.7179310917854309,
      "learning_rate": 6.972687814462348e-07,
      "loss": 0.6763,
      "step": 28979
    },
    {
      "epoch": 0.8350141185962081,
      "grad_norm": 0.7223648428916931,
      "learning_rate": 6.970311218996751e-07,
      "loss": 0.7009,
      "step": 28980
    },
    {
      "epoch": 0.8350429320578574,
      "grad_norm": 0.721780002117157,
      "learning_rate": 6.967934998276721e-07,
      "loss": 0.6783,
      "step": 28981
    },
    {
      "epoch": 0.8350717455195067,
      "grad_norm": 0.7103506326675415,
      "learning_rate": 6.965559152322971e-07,
      "loss": 0.6875,
      "step": 28982
    },
    {
      "epoch": 0.835100558981156,
      "grad_norm": 0.7235816717147827,
      "learning_rate": 6.963183681156172e-07,
      "loss": 0.6725,
      "step": 28983
    },
    {
      "epoch": 0.8351293724428053,
      "grad_norm": 0.7350837588310242,
      "learning_rate": 6.960808584797024e-07,
      "loss": 0.7031,
      "step": 28984
    },
    {
      "epoch": 0.8351581859044546,
      "grad_norm": 0.7396684288978577,
      "learning_rate": 6.958433863266217e-07,
      "loss": 0.6689,
      "step": 28985
    },
    {
      "epoch": 0.8351869993661039,
      "grad_norm": 0.7197608947753906,
      "learning_rate": 6.956059516584419e-07,
      "loss": 0.6847,
      "step": 28986
    },
    {
      "epoch": 0.8352158128277531,
      "grad_norm": 0.716587483882904,
      "learning_rate": 6.95368554477231e-07,
      "loss": 0.6616,
      "step": 28987
    },
    {
      "epoch": 0.8352446262894024,
      "grad_norm": 0.7140398621559143,
      "learning_rate": 6.95131194785057e-07,
      "loss": 0.6636,
      "step": 28988
    },
    {
      "epoch": 0.8352734397510517,
      "grad_norm": 0.7342939972877502,
      "learning_rate": 6.948938725839883e-07,
      "loss": 0.6792,
      "step": 28989
    },
    {
      "epoch": 0.835302253212701,
      "grad_norm": 0.7203136086463928,
      "learning_rate": 6.946565878760891e-07,
      "loss": 0.6672,
      "step": 28990
    },
    {
      "epoch": 0.8353310666743503,
      "grad_norm": 0.7377681136131287,
      "learning_rate": 6.944193406634286e-07,
      "loss": 0.7024,
      "step": 28991
    },
    {
      "epoch": 0.8353598801359995,
      "grad_norm": 0.7386910915374756,
      "learning_rate": 6.9418213094807e-07,
      "loss": 0.6742,
      "step": 28992
    },
    {
      "epoch": 0.8353886935976488,
      "grad_norm": 0.7213529348373413,
      "learning_rate": 6.939449587320824e-07,
      "loss": 0.7034,
      "step": 28993
    },
    {
      "epoch": 0.8354175070592981,
      "grad_norm": 0.7318434119224548,
      "learning_rate": 6.937078240175287e-07,
      "loss": 0.6819,
      "step": 28994
    },
    {
      "epoch": 0.8354463205209474,
      "grad_norm": 0.7266700863838196,
      "learning_rate": 6.934707268064761e-07,
      "loss": 0.6705,
      "step": 28995
    },
    {
      "epoch": 0.8354751339825967,
      "grad_norm": 0.7219359278678894,
      "learning_rate": 6.932336671009876e-07,
      "loss": 0.6932,
      "step": 28996
    },
    {
      "epoch": 0.8355039474442459,
      "grad_norm": 0.7048945426940918,
      "learning_rate": 6.929966449031295e-07,
      "loss": 0.6785,
      "step": 28997
    },
    {
      "epoch": 0.8355327609058952,
      "grad_norm": 0.7192205190658569,
      "learning_rate": 6.927596602149627e-07,
      "loss": 0.6692,
      "step": 28998
    },
    {
      "epoch": 0.8355615743675445,
      "grad_norm": 0.7207168340682983,
      "learning_rate": 6.925227130385565e-07,
      "loss": 0.6867,
      "step": 28999
    },
    {
      "epoch": 0.8355903878291938,
      "grad_norm": 0.7333238124847412,
      "learning_rate": 6.9228580337597e-07,
      "loss": 0.6922,
      "step": 29000
    },
    {
      "epoch": 0.8356192012908431,
      "grad_norm": 0.8503965735435486,
      "learning_rate": 6.920489312292694e-07,
      "loss": 0.6594,
      "step": 29001
    },
    {
      "epoch": 0.8356480147524924,
      "grad_norm": 0.7449647784233093,
      "learning_rate": 6.918120966005148e-07,
      "loss": 0.7281,
      "step": 29002
    },
    {
      "epoch": 0.8356768282141417,
      "grad_norm": 0.7167662978172302,
      "learning_rate": 6.915752994917713e-07,
      "loss": 0.6647,
      "step": 29003
    },
    {
      "epoch": 0.835705641675791,
      "grad_norm": 0.7373087406158447,
      "learning_rate": 6.913385399050993e-07,
      "loss": 0.69,
      "step": 29004
    },
    {
      "epoch": 0.8357344551374402,
      "grad_norm": 0.7459709644317627,
      "learning_rate": 6.911018178425616e-07,
      "loss": 0.6786,
      "step": 29005
    },
    {
      "epoch": 0.8357632685990894,
      "grad_norm": 0.720786452293396,
      "learning_rate": 6.908651333062205e-07,
      "loss": 0.715,
      "step": 29006
    },
    {
      "epoch": 0.8357920820607387,
      "grad_norm": 0.7169796228408813,
      "learning_rate": 6.906284862981355e-07,
      "loss": 0.7029,
      "step": 29007
    },
    {
      "epoch": 0.835820895522388,
      "grad_norm": 0.7251313328742981,
      "learning_rate": 6.903918768203699e-07,
      "loss": 0.6857,
      "step": 29008
    },
    {
      "epoch": 0.8358497089840373,
      "grad_norm": 0.6905713081359863,
      "learning_rate": 6.901553048749815e-07,
      "loss": 0.661,
      "step": 29009
    },
    {
      "epoch": 0.8358785224456866,
      "grad_norm": 0.7387323379516602,
      "learning_rate": 6.899187704640326e-07,
      "loss": 0.7029,
      "step": 29010
    },
    {
      "epoch": 0.8359073359073359,
      "grad_norm": 0.722809910774231,
      "learning_rate": 6.896822735895825e-07,
      "loss": 0.681,
      "step": 29011
    },
    {
      "epoch": 0.8359361493689852,
      "grad_norm": 0.7751491069793701,
      "learning_rate": 6.894458142536925e-07,
      "loss": 0.6879,
      "step": 29012
    },
    {
      "epoch": 0.8359649628306345,
      "grad_norm": 1.1023979187011719,
      "learning_rate": 6.892093924584187e-07,
      "loss": 0.6807,
      "step": 29013
    },
    {
      "epoch": 0.8359937762922838,
      "grad_norm": 0.7201986312866211,
      "learning_rate": 6.889730082058232e-07,
      "loss": 0.6844,
      "step": 29014
    },
    {
      "epoch": 0.8360225897539331,
      "grad_norm": 0.7059544324874878,
      "learning_rate": 6.887366614979629e-07,
      "loss": 0.6864,
      "step": 29015
    },
    {
      "epoch": 0.8360514032155824,
      "grad_norm": 0.7297528386116028,
      "learning_rate": 6.885003523368971e-07,
      "loss": 0.6789,
      "step": 29016
    },
    {
      "epoch": 0.8360802166772316,
      "grad_norm": 0.7277292609214783,
      "learning_rate": 6.882640807246821e-07,
      "loss": 0.6807,
      "step": 29017
    },
    {
      "epoch": 0.8361090301388809,
      "grad_norm": 0.7138065695762634,
      "learning_rate": 6.880278466633783e-07,
      "loss": 0.6886,
      "step": 29018
    },
    {
      "epoch": 0.8361378436005301,
      "grad_norm": 0.7267730832099915,
      "learning_rate": 6.877916501550403e-07,
      "loss": 0.6728,
      "step": 29019
    },
    {
      "epoch": 0.8361666570621794,
      "grad_norm": 0.7248713374137878,
      "learning_rate": 6.875554912017274e-07,
      "loss": 0.692,
      "step": 29020
    },
    {
      "epoch": 0.8361954705238287,
      "grad_norm": 0.7535969614982605,
      "learning_rate": 6.873193698054942e-07,
      "loss": 0.6933,
      "step": 29021
    },
    {
      "epoch": 0.836224283985478,
      "grad_norm": 0.7188230156898499,
      "learning_rate": 6.870832859683979e-07,
      "loss": 0.6981,
      "step": 29022
    },
    {
      "epoch": 0.8362530974471273,
      "grad_norm": 0.7161887884140015,
      "learning_rate": 6.868472396924953e-07,
      "loss": 0.6786,
      "step": 29023
    },
    {
      "epoch": 0.8362819109087766,
      "grad_norm": 0.7294310331344604,
      "learning_rate": 6.866112309798428e-07,
      "loss": 0.669,
      "step": 29024
    },
    {
      "epoch": 0.8363107243704259,
      "grad_norm": 0.7292380332946777,
      "learning_rate": 6.863752598324936e-07,
      "loss": 0.6777,
      "step": 29025
    },
    {
      "epoch": 0.8363395378320752,
      "grad_norm": 0.7400527000427246,
      "learning_rate": 6.861393262525035e-07,
      "loss": 0.6826,
      "step": 29026
    },
    {
      "epoch": 0.8363683512937244,
      "grad_norm": 0.7576344609260559,
      "learning_rate": 6.859034302419287e-07,
      "loss": 0.6956,
      "step": 29027
    },
    {
      "epoch": 0.8363971647553737,
      "grad_norm": 0.7374427318572998,
      "learning_rate": 6.856675718028216e-07,
      "loss": 0.7183,
      "step": 29028
    },
    {
      "epoch": 0.836425978217023,
      "grad_norm": 0.7297567129135132,
      "learning_rate": 6.854317509372383e-07,
      "loss": 0.6866,
      "step": 29029
    },
    {
      "epoch": 0.8364547916786723,
      "grad_norm": 0.7340295910835266,
      "learning_rate": 6.851959676472303e-07,
      "loss": 0.6992,
      "step": 29030
    },
    {
      "epoch": 0.8364836051403216,
      "grad_norm": 0.7269029021263123,
      "learning_rate": 6.849602219348533e-07,
      "loss": 0.691,
      "step": 29031
    },
    {
      "epoch": 0.8365124186019708,
      "grad_norm": 0.7316607236862183,
      "learning_rate": 6.847245138021586e-07,
      "loss": 0.6852,
      "step": 29032
    },
    {
      "epoch": 0.8365412320636201,
      "grad_norm": 0.7378653883934021,
      "learning_rate": 6.844888432512004e-07,
      "loss": 0.664,
      "step": 29033
    },
    {
      "epoch": 0.8365700455252694,
      "grad_norm": 0.7242190837860107,
      "learning_rate": 6.842532102840277e-07,
      "loss": 0.6806,
      "step": 29034
    },
    {
      "epoch": 0.8365988589869187,
      "grad_norm": 0.7197503447532654,
      "learning_rate": 6.840176149026983e-07,
      "loss": 0.6723,
      "step": 29035
    },
    {
      "epoch": 0.836627672448568,
      "grad_norm": 0.7237809300422668,
      "learning_rate": 6.837820571092596e-07,
      "loss": 0.6815,
      "step": 29036
    },
    {
      "epoch": 0.8366564859102172,
      "grad_norm": 0.7060258984565735,
      "learning_rate": 6.835465369057659e-07,
      "loss": 0.669,
      "step": 29037
    },
    {
      "epoch": 0.8366852993718665,
      "grad_norm": 0.7068501710891724,
      "learning_rate": 6.833110542942662e-07,
      "loss": 0.6517,
      "step": 29038
    },
    {
      "epoch": 0.8367141128335158,
      "grad_norm": 0.7298252582550049,
      "learning_rate": 6.830756092768126e-07,
      "loss": 0.6873,
      "step": 29039
    },
    {
      "epoch": 0.8367429262951651,
      "grad_norm": 0.7427700161933899,
      "learning_rate": 6.828402018554542e-07,
      "loss": 0.6982,
      "step": 29040
    },
    {
      "epoch": 0.8367717397568144,
      "grad_norm": 0.7093572020530701,
      "learning_rate": 6.826048320322431e-07,
      "loss": 0.7003,
      "step": 29041
    },
    {
      "epoch": 0.8368005532184637,
      "grad_norm": 0.7230256199836731,
      "learning_rate": 6.823694998092272e-07,
      "loss": 0.6833,
      "step": 29042
    },
    {
      "epoch": 0.836829366680113,
      "grad_norm": 0.7241486310958862,
      "learning_rate": 6.821342051884567e-07,
      "loss": 0.6799,
      "step": 29043
    },
    {
      "epoch": 0.8368581801417623,
      "grad_norm": 0.7206100821495056,
      "learning_rate": 6.818989481719823e-07,
      "loss": 0.6985,
      "step": 29044
    },
    {
      "epoch": 0.8368869936034116,
      "grad_norm": 0.7353971600532532,
      "learning_rate": 6.816637287618504e-07,
      "loss": 0.6698,
      "step": 29045
    },
    {
      "epoch": 0.8369158070650607,
      "grad_norm": 0.7057508826255798,
      "learning_rate": 6.814285469601106e-07,
      "loss": 0.7166,
      "step": 29046
    },
    {
      "epoch": 0.83694462052671,
      "grad_norm": 0.7320938110351562,
      "learning_rate": 6.811934027688111e-07,
      "loss": 0.6791,
      "step": 29047
    },
    {
      "epoch": 0.8369734339883593,
      "grad_norm": 0.7174010872840881,
      "learning_rate": 6.809582961900008e-07,
      "loss": 0.6477,
      "step": 29048
    },
    {
      "epoch": 0.8370022474500086,
      "grad_norm": 0.7142430543899536,
      "learning_rate": 6.807232272257253e-07,
      "loss": 0.6643,
      "step": 29049
    },
    {
      "epoch": 0.8370310609116579,
      "grad_norm": 0.7166874408721924,
      "learning_rate": 6.804881958780341e-07,
      "loss": 0.6745,
      "step": 29050
    },
    {
      "epoch": 0.8370598743733072,
      "grad_norm": 0.7404472231864929,
      "learning_rate": 6.802532021489716e-07,
      "loss": 0.654,
      "step": 29051
    },
    {
      "epoch": 0.8370886878349565,
      "grad_norm": 0.7364152073860168,
      "learning_rate": 6.800182460405863e-07,
      "loss": 0.6807,
      "step": 29052
    },
    {
      "epoch": 0.8371175012966058,
      "grad_norm": 0.7215190529823303,
      "learning_rate": 6.797833275549232e-07,
      "loss": 0.684,
      "step": 29053
    },
    {
      "epoch": 0.8371463147582551,
      "grad_norm": 0.71881502866745,
      "learning_rate": 6.795484466940295e-07,
      "loss": 0.6656,
      "step": 29054
    },
    {
      "epoch": 0.8371751282199044,
      "grad_norm": 0.7253089547157288,
      "learning_rate": 6.793136034599484e-07,
      "loss": 0.6859,
      "step": 29055
    },
    {
      "epoch": 0.8372039416815537,
      "grad_norm": 0.7304123044013977,
      "learning_rate": 6.790787978547286e-07,
      "loss": 0.7059,
      "step": 29056
    },
    {
      "epoch": 0.837232755143203,
      "grad_norm": 0.6948004961013794,
      "learning_rate": 6.788440298804117e-07,
      "loss": 0.6668,
      "step": 29057
    },
    {
      "epoch": 0.8372615686048522,
      "grad_norm": 0.6983707547187805,
      "learning_rate": 6.786092995390436e-07,
      "loss": 0.6575,
      "step": 29058
    },
    {
      "epoch": 0.8372903820665014,
      "grad_norm": 0.7174709439277649,
      "learning_rate": 6.783746068326686e-07,
      "loss": 0.6798,
      "step": 29059
    },
    {
      "epoch": 0.8373191955281507,
      "grad_norm": 0.7241666913032532,
      "learning_rate": 6.781399517633319e-07,
      "loss": 0.6619,
      "step": 29060
    },
    {
      "epoch": 0.8373480089898,
      "grad_norm": 0.7049720287322998,
      "learning_rate": 6.779053343330749e-07,
      "loss": 0.6652,
      "step": 29061
    },
    {
      "epoch": 0.8373768224514493,
      "grad_norm": 0.7032800316810608,
      "learning_rate": 6.776707545439432e-07,
      "loss": 0.6803,
      "step": 29062
    },
    {
      "epoch": 0.8374056359130986,
      "grad_norm": 0.7030571103096008,
      "learning_rate": 6.774362123979772e-07,
      "loss": 0.6986,
      "step": 29063
    },
    {
      "epoch": 0.8374344493747479,
      "grad_norm": 0.7063060402870178,
      "learning_rate": 6.772017078972209e-07,
      "loss": 0.6739,
      "step": 29064
    },
    {
      "epoch": 0.8374632628363972,
      "grad_norm": 0.7245265245437622,
      "learning_rate": 6.769672410437173e-07,
      "loss": 0.6889,
      "step": 29065
    },
    {
      "epoch": 0.8374920762980465,
      "grad_norm": 0.719963550567627,
      "learning_rate": 6.767328118395067e-07,
      "loss": 0.6768,
      "step": 29066
    },
    {
      "epoch": 0.8375208897596957,
      "grad_norm": 0.7228556275367737,
      "learning_rate": 6.764984202866332e-07,
      "loss": 0.6896,
      "step": 29067
    },
    {
      "epoch": 0.837549703221345,
      "grad_norm": 0.7308449149131775,
      "learning_rate": 6.762640663871345e-07,
      "loss": 0.6919,
      "step": 29068
    },
    {
      "epoch": 0.8375785166829943,
      "grad_norm": 0.728331983089447,
      "learning_rate": 6.760297501430557e-07,
      "loss": 0.6846,
      "step": 29069
    },
    {
      "epoch": 0.8376073301446436,
      "grad_norm": 0.7292725443840027,
      "learning_rate": 6.757954715564324e-07,
      "loss": 0.6835,
      "step": 29070
    },
    {
      "epoch": 0.8376361436062929,
      "grad_norm": 0.7377651929855347,
      "learning_rate": 6.755612306293102e-07,
      "loss": 0.6833,
      "step": 29071
    },
    {
      "epoch": 0.8376649570679422,
      "grad_norm": 0.7229037880897522,
      "learning_rate": 6.753270273637263e-07,
      "loss": 0.6893,
      "step": 29072
    },
    {
      "epoch": 0.8376937705295914,
      "grad_norm": 0.7025160193443298,
      "learning_rate": 6.750928617617214e-07,
      "loss": 0.6739,
      "step": 29073
    },
    {
      "epoch": 0.8377225839912407,
      "grad_norm": 0.7197566032409668,
      "learning_rate": 6.748587338253338e-07,
      "loss": 0.7087,
      "step": 29074
    },
    {
      "epoch": 0.83775139745289,
      "grad_norm": 0.7333875298500061,
      "learning_rate": 6.74624643556604e-07,
      "loss": 0.6872,
      "step": 29075
    },
    {
      "epoch": 0.8377802109145392,
      "grad_norm": 0.8238506317138672,
      "learning_rate": 6.743905909575682e-07,
      "loss": 0.6924,
      "step": 29076
    },
    {
      "epoch": 0.8378090243761885,
      "grad_norm": 0.7189961075782776,
      "learning_rate": 6.741565760302676e-07,
      "loss": 0.6679,
      "step": 29077
    },
    {
      "epoch": 0.8378378378378378,
      "grad_norm": 0.7065731287002563,
      "learning_rate": 6.739225987767384e-07,
      "loss": 0.6719,
      "step": 29078
    },
    {
      "epoch": 0.8378666512994871,
      "grad_norm": 0.71649169921875,
      "learning_rate": 6.736886591990193e-07,
      "loss": 0.6837,
      "step": 29079
    },
    {
      "epoch": 0.8378954647611364,
      "grad_norm": 0.712335467338562,
      "learning_rate": 6.734547572991468e-07,
      "loss": 0.6858,
      "step": 29080
    },
    {
      "epoch": 0.8379242782227857,
      "grad_norm": 0.7129342555999756,
      "learning_rate": 6.732208930791578e-07,
      "loss": 0.6639,
      "step": 29081
    },
    {
      "epoch": 0.837953091684435,
      "grad_norm": 0.7225082516670227,
      "learning_rate": 6.729870665410898e-07,
      "loss": 0.6774,
      "step": 29082
    },
    {
      "epoch": 0.8379819051460843,
      "grad_norm": 0.7283352613449097,
      "learning_rate": 6.727532776869805e-07,
      "loss": 0.6963,
      "step": 29083
    },
    {
      "epoch": 0.8380107186077336,
      "grad_norm": 0.7340595722198486,
      "learning_rate": 6.725195265188628e-07,
      "loss": 0.696,
      "step": 29084
    },
    {
      "epoch": 0.8380395320693829,
      "grad_norm": 0.7121168375015259,
      "learning_rate": 6.722858130387744e-07,
      "loss": 0.6686,
      "step": 29085
    },
    {
      "epoch": 0.838068345531032,
      "grad_norm": 0.7139633893966675,
      "learning_rate": 6.720521372487516e-07,
      "loss": 0.6791,
      "step": 29086
    },
    {
      "epoch": 0.8380971589926813,
      "grad_norm": 0.7118927836418152,
      "learning_rate": 6.71818499150827e-07,
      "loss": 0.6844,
      "step": 29087
    },
    {
      "epoch": 0.8381259724543306,
      "grad_norm": 0.7114420533180237,
      "learning_rate": 6.715848987470381e-07,
      "loss": 0.6684,
      "step": 29088
    },
    {
      "epoch": 0.8381547859159799,
      "grad_norm": 0.7231202721595764,
      "learning_rate": 6.713513360394164e-07,
      "loss": 0.6893,
      "step": 29089
    },
    {
      "epoch": 0.8381835993776292,
      "grad_norm": 0.6918326616287231,
      "learning_rate": 6.711178110299993e-07,
      "loss": 0.6787,
      "step": 29090
    },
    {
      "epoch": 0.8382124128392785,
      "grad_norm": 0.7237821221351624,
      "learning_rate": 6.708843237208168e-07,
      "loss": 0.6866,
      "step": 29091
    },
    {
      "epoch": 0.8382412263009278,
      "grad_norm": 0.7329044938087463,
      "learning_rate": 6.706508741139062e-07,
      "loss": 0.6578,
      "step": 29092
    },
    {
      "epoch": 0.8382700397625771,
      "grad_norm": 0.731486976146698,
      "learning_rate": 6.704174622112969e-07,
      "loss": 0.6712,
      "step": 29093
    },
    {
      "epoch": 0.8382988532242264,
      "grad_norm": 0.7233490943908691,
      "learning_rate": 6.70184088015024e-07,
      "loss": 0.6858,
      "step": 29094
    },
    {
      "epoch": 0.8383276666858757,
      "grad_norm": 0.7205370664596558,
      "learning_rate": 6.699507515271192e-07,
      "loss": 0.6921,
      "step": 29095
    },
    {
      "epoch": 0.838356480147525,
      "grad_norm": 0.7677876949310303,
      "learning_rate": 6.697174527496164e-07,
      "loss": 0.6827,
      "step": 29096
    },
    {
      "epoch": 0.8383852936091742,
      "grad_norm": 0.7429009675979614,
      "learning_rate": 6.69484191684544e-07,
      "loss": 0.7015,
      "step": 29097
    },
    {
      "epoch": 0.8384141070708235,
      "grad_norm": 0.7033623456954956,
      "learning_rate": 6.692509683339371e-07,
      "loss": 0.6771,
      "step": 29098
    },
    {
      "epoch": 0.8384429205324728,
      "grad_norm": 0.7279817461967468,
      "learning_rate": 6.69017782699824e-07,
      "loss": 0.6724,
      "step": 29099
    },
    {
      "epoch": 0.838471733994122,
      "grad_norm": 0.7229674458503723,
      "learning_rate": 6.687846347842375e-07,
      "loss": 0.6828,
      "step": 29100
    },
    {
      "epoch": 0.8385005474557713,
      "grad_norm": 0.7025011777877808,
      "learning_rate": 6.68551524589206e-07,
      "loss": 0.6647,
      "step": 29101
    },
    {
      "epoch": 0.8385293609174206,
      "grad_norm": 0.7265325784683228,
      "learning_rate": 6.683184521167613e-07,
      "loss": 0.675,
      "step": 29102
    },
    {
      "epoch": 0.8385581743790699,
      "grad_norm": 0.7248400449752808,
      "learning_rate": 6.680854173689333e-07,
      "loss": 0.6941,
      "step": 29103
    },
    {
      "epoch": 0.8385869878407192,
      "grad_norm": 0.7443091869354248,
      "learning_rate": 6.678524203477504e-07,
      "loss": 0.6914,
      "step": 29104
    },
    {
      "epoch": 0.8386158013023685,
      "grad_norm": 0.7297363877296448,
      "learning_rate": 6.676194610552433e-07,
      "loss": 0.6811,
      "step": 29105
    },
    {
      "epoch": 0.8386446147640177,
      "grad_norm": 0.7272025346755981,
      "learning_rate": 6.673865394934376e-07,
      "loss": 0.6605,
      "step": 29106
    },
    {
      "epoch": 0.838673428225667,
      "grad_norm": 0.7176963090896606,
      "learning_rate": 6.671536556643659e-07,
      "loss": 0.6846,
      "step": 29107
    },
    {
      "epoch": 0.8387022416873163,
      "grad_norm": 0.7107498645782471,
      "learning_rate": 6.669208095700541e-07,
      "loss": 0.6712,
      "step": 29108
    },
    {
      "epoch": 0.8387310551489656,
      "grad_norm": 0.7221946120262146,
      "learning_rate": 6.666880012125315e-07,
      "loss": 0.677,
      "step": 29109
    },
    {
      "epoch": 0.8387598686106149,
      "grad_norm": 0.713245153427124,
      "learning_rate": 6.664552305938238e-07,
      "loss": 0.6815,
      "step": 29110
    },
    {
      "epoch": 0.8387886820722642,
      "grad_norm": 0.7232970595359802,
      "learning_rate": 6.662224977159598e-07,
      "loss": 0.6707,
      "step": 29111
    },
    {
      "epoch": 0.8388174955339135,
      "grad_norm": 0.7109699249267578,
      "learning_rate": 6.659898025809647e-07,
      "loss": 0.6839,
      "step": 29112
    },
    {
      "epoch": 0.8388463089955627,
      "grad_norm": 0.723254382610321,
      "learning_rate": 6.657571451908674e-07,
      "loss": 0.673,
      "step": 29113
    },
    {
      "epoch": 0.838875122457212,
      "grad_norm": 0.7119907140731812,
      "learning_rate": 6.655245255476911e-07,
      "loss": 0.6832,
      "step": 29114
    },
    {
      "epoch": 0.8389039359188613,
      "grad_norm": 0.7487133145332336,
      "learning_rate": 6.65291943653465e-07,
      "loss": 0.6788,
      "step": 29115
    },
    {
      "epoch": 0.8389327493805105,
      "grad_norm": 0.7211758494377136,
      "learning_rate": 6.650593995102111e-07,
      "loss": 0.6826,
      "step": 29116
    },
    {
      "epoch": 0.8389615628421598,
      "grad_norm": 0.7097522616386414,
      "learning_rate": 6.648268931199581e-07,
      "loss": 0.6947,
      "step": 29117
    },
    {
      "epoch": 0.8389903763038091,
      "grad_norm": 0.7321993708610535,
      "learning_rate": 6.645944244847268e-07,
      "loss": 0.6637,
      "step": 29118
    },
    {
      "epoch": 0.8390191897654584,
      "grad_norm": 0.7065295577049255,
      "learning_rate": 6.643619936065466e-07,
      "loss": 0.6844,
      "step": 29119
    },
    {
      "epoch": 0.8390480032271077,
      "grad_norm": 0.6890040040016174,
      "learning_rate": 6.641296004874382e-07,
      "loss": 0.6622,
      "step": 29120
    },
    {
      "epoch": 0.839076816688757,
      "grad_norm": 0.7529351711273193,
      "learning_rate": 6.638972451294279e-07,
      "loss": 0.7018,
      "step": 29121
    },
    {
      "epoch": 0.8391056301504063,
      "grad_norm": 0.7016050219535828,
      "learning_rate": 6.636649275345369e-07,
      "loss": 0.6534,
      "step": 29122
    },
    {
      "epoch": 0.8391344436120556,
      "grad_norm": 0.7442736625671387,
      "learning_rate": 6.634326477047898e-07,
      "loss": 0.7109,
      "step": 29123
    },
    {
      "epoch": 0.8391632570737049,
      "grad_norm": 0.7524096965789795,
      "learning_rate": 6.632004056422104e-07,
      "loss": 0.6927,
      "step": 29124
    },
    {
      "epoch": 0.8391920705353542,
      "grad_norm": 0.7416664958000183,
      "learning_rate": 6.629682013488186e-07,
      "loss": 0.7031,
      "step": 29125
    },
    {
      "epoch": 0.8392208839970035,
      "grad_norm": 0.7271758913993835,
      "learning_rate": 6.627360348266399e-07,
      "loss": 0.6955,
      "step": 29126
    },
    {
      "epoch": 0.8392496974586526,
      "grad_norm": 0.7185454964637756,
      "learning_rate": 6.625039060776939e-07,
      "loss": 0.7051,
      "step": 29127
    },
    {
      "epoch": 0.8392785109203019,
      "grad_norm": 0.7355625033378601,
      "learning_rate": 6.622718151040036e-07,
      "loss": 0.7061,
      "step": 29128
    },
    {
      "epoch": 0.8393073243819512,
      "grad_norm": 0.7420949339866638,
      "learning_rate": 6.620397619075886e-07,
      "loss": 0.7018,
      "step": 29129
    },
    {
      "epoch": 0.8393361378436005,
      "grad_norm": 0.7127048373222351,
      "learning_rate": 6.618077464904709e-07,
      "loss": 0.6787,
      "step": 29130
    },
    {
      "epoch": 0.8393649513052498,
      "grad_norm": 0.70743727684021,
      "learning_rate": 6.615757688546714e-07,
      "loss": 0.6887,
      "step": 29131
    },
    {
      "epoch": 0.8393937647668991,
      "grad_norm": 0.7402710914611816,
      "learning_rate": 6.613438290022112e-07,
      "loss": 0.659,
      "step": 29132
    },
    {
      "epoch": 0.8394225782285484,
      "grad_norm": 0.726945161819458,
      "learning_rate": 6.611119269351074e-07,
      "loss": 0.6793,
      "step": 29133
    },
    {
      "epoch": 0.8394513916901977,
      "grad_norm": 0.7298011779785156,
      "learning_rate": 6.608800626553829e-07,
      "loss": 0.6593,
      "step": 29134
    },
    {
      "epoch": 0.839480205151847,
      "grad_norm": 0.7326124310493469,
      "learning_rate": 6.606482361650546e-07,
      "loss": 0.6725,
      "step": 29135
    },
    {
      "epoch": 0.8395090186134962,
      "grad_norm": 0.7242642045021057,
      "learning_rate": 6.604164474661434e-07,
      "loss": 0.6917,
      "step": 29136
    },
    {
      "epoch": 0.8395378320751455,
      "grad_norm": 0.7402667999267578,
      "learning_rate": 6.601846965606662e-07,
      "loss": 0.6917,
      "step": 29137
    },
    {
      "epoch": 0.8395666455367948,
      "grad_norm": 0.7245057225227356,
      "learning_rate": 6.59952983450643e-07,
      "loss": 0.6951,
      "step": 29138
    },
    {
      "epoch": 0.8395954589984441,
      "grad_norm": 0.7026066780090332,
      "learning_rate": 6.597213081380894e-07,
      "loss": 0.6719,
      "step": 29139
    },
    {
      "epoch": 0.8396242724600933,
      "grad_norm": 0.7327108383178711,
      "learning_rate": 6.594896706250248e-07,
      "loss": 0.6817,
      "step": 29140
    },
    {
      "epoch": 0.8396530859217426,
      "grad_norm": 0.7064483165740967,
      "learning_rate": 6.592580709134671e-07,
      "loss": 0.6824,
      "step": 29141
    },
    {
      "epoch": 0.8396818993833919,
      "grad_norm": 0.694242537021637,
      "learning_rate": 6.590265090054315e-07,
      "loss": 0.6409,
      "step": 29142
    },
    {
      "epoch": 0.8397107128450412,
      "grad_norm": 0.7089196443557739,
      "learning_rate": 6.587949849029357e-07,
      "loss": 0.6782,
      "step": 29143
    },
    {
      "epoch": 0.8397395263066905,
      "grad_norm": 0.7332811951637268,
      "learning_rate": 6.585634986079959e-07,
      "loss": 0.7086,
      "step": 29144
    },
    {
      "epoch": 0.8397683397683398,
      "grad_norm": 0.7403357028961182,
      "learning_rate": 6.583320501226292e-07,
      "loss": 0.6927,
      "step": 29145
    },
    {
      "epoch": 0.839797153229989,
      "grad_norm": 0.7217675447463989,
      "learning_rate": 6.581006394488493e-07,
      "loss": 0.6823,
      "step": 29146
    },
    {
      "epoch": 0.8398259666916383,
      "grad_norm": 0.7152122855186462,
      "learning_rate": 6.578692665886733e-07,
      "loss": 0.6812,
      "step": 29147
    },
    {
      "epoch": 0.8398547801532876,
      "grad_norm": 0.7300862669944763,
      "learning_rate": 6.576379315441145e-07,
      "loss": 0.6725,
      "step": 29148
    },
    {
      "epoch": 0.8398835936149369,
      "grad_norm": 0.7145205736160278,
      "learning_rate": 6.574066343171898e-07,
      "loss": 0.638,
      "step": 29149
    },
    {
      "epoch": 0.8399124070765862,
      "grad_norm": 0.7213398814201355,
      "learning_rate": 6.571753749099113e-07,
      "loss": 0.6838,
      "step": 29150
    },
    {
      "epoch": 0.8399412205382355,
      "grad_norm": 0.7224977612495422,
      "learning_rate": 6.569441533242948e-07,
      "loss": 0.6656,
      "step": 29151
    },
    {
      "epoch": 0.8399700339998848,
      "grad_norm": 0.7330282926559448,
      "learning_rate": 6.567129695623525e-07,
      "loss": 0.6995,
      "step": 29152
    },
    {
      "epoch": 0.8399988474615341,
      "grad_norm": 0.7105873227119446,
      "learning_rate": 6.564818236260994e-07,
      "loss": 0.6723,
      "step": 29153
    },
    {
      "epoch": 0.8400276609231833,
      "grad_norm": 0.7220221757888794,
      "learning_rate": 6.562507155175457e-07,
      "loss": 0.6763,
      "step": 29154
    },
    {
      "epoch": 0.8400564743848326,
      "grad_norm": 0.7068584561347961,
      "learning_rate": 6.56019645238708e-07,
      "loss": 0.6804,
      "step": 29155
    },
    {
      "epoch": 0.8400852878464818,
      "grad_norm": 0.7332095503807068,
      "learning_rate": 6.557886127915964e-07,
      "loss": 0.6811,
      "step": 29156
    },
    {
      "epoch": 0.8401141013081311,
      "grad_norm": 0.7203821539878845,
      "learning_rate": 6.555576181782236e-07,
      "loss": 0.6772,
      "step": 29157
    },
    {
      "epoch": 0.8401429147697804,
      "grad_norm": 0.6896966099739075,
      "learning_rate": 6.553266614006005e-07,
      "loss": 0.6488,
      "step": 29158
    },
    {
      "epoch": 0.8401717282314297,
      "grad_norm": 0.7115285992622375,
      "learning_rate": 6.550957424607402e-07,
      "loss": 0.6675,
      "step": 29159
    },
    {
      "epoch": 0.840200541693079,
      "grad_norm": 0.7235550284385681,
      "learning_rate": 6.548648613606512e-07,
      "loss": 0.703,
      "step": 29160
    },
    {
      "epoch": 0.8402293551547283,
      "grad_norm": 0.7236813902854919,
      "learning_rate": 6.546340181023458e-07,
      "loss": 0.6759,
      "step": 29161
    },
    {
      "epoch": 0.8402581686163776,
      "grad_norm": 0.8526968359947205,
      "learning_rate": 6.544032126878358e-07,
      "loss": 0.6843,
      "step": 29162
    },
    {
      "epoch": 0.8402869820780269,
      "grad_norm": 0.74554842710495,
      "learning_rate": 6.541724451191283e-07,
      "loss": 0.6959,
      "step": 29163
    },
    {
      "epoch": 0.8403157955396762,
      "grad_norm": 0.7287281155586243,
      "learning_rate": 6.539417153982358e-07,
      "loss": 0.6829,
      "step": 29164
    },
    {
      "epoch": 0.8403446090013255,
      "grad_norm": 0.7295098304748535,
      "learning_rate": 6.537110235271654e-07,
      "loss": 0.6794,
      "step": 29165
    },
    {
      "epoch": 0.8403734224629748,
      "grad_norm": 0.7215052843093872,
      "learning_rate": 6.534803695079272e-07,
      "loss": 0.683,
      "step": 29166
    },
    {
      "epoch": 0.8404022359246239,
      "grad_norm": 0.7295392155647278,
      "learning_rate": 6.532497533425303e-07,
      "loss": 0.6966,
      "step": 29167
    },
    {
      "epoch": 0.8404310493862732,
      "grad_norm": 0.7363985776901245,
      "learning_rate": 6.530191750329839e-07,
      "loss": 0.6818,
      "step": 29168
    },
    {
      "epoch": 0.8404598628479225,
      "grad_norm": 0.7180034518241882,
      "learning_rate": 6.527886345812939e-07,
      "loss": 0.6729,
      "step": 29169
    },
    {
      "epoch": 0.8404886763095718,
      "grad_norm": 0.72902512550354,
      "learning_rate": 6.525581319894703e-07,
      "loss": 0.6913,
      "step": 29170
    },
    {
      "epoch": 0.8405174897712211,
      "grad_norm": 0.7001736760139465,
      "learning_rate": 6.523276672595186e-07,
      "loss": 0.6625,
      "step": 29171
    },
    {
      "epoch": 0.8405463032328704,
      "grad_norm": 0.7118436694145203,
      "learning_rate": 6.520972403934478e-07,
      "loss": 0.673,
      "step": 29172
    },
    {
      "epoch": 0.8405751166945197,
      "grad_norm": 0.7168707847595215,
      "learning_rate": 6.518668513932625e-07,
      "loss": 0.6656,
      "step": 29173
    },
    {
      "epoch": 0.840603930156169,
      "grad_norm": 0.728632390499115,
      "learning_rate": 6.51636500260972e-07,
      "loss": 0.6751,
      "step": 29174
    },
    {
      "epoch": 0.8406327436178183,
      "grad_norm": 0.7198623418807983,
      "learning_rate": 6.514061869985793e-07,
      "loss": 0.6842,
      "step": 29175
    },
    {
      "epoch": 0.8406615570794675,
      "grad_norm": 0.7210376262664795,
      "learning_rate": 6.511759116080929e-07,
      "loss": 0.6788,
      "step": 29176
    },
    {
      "epoch": 0.8406903705411168,
      "grad_norm": 0.7231383323669434,
      "learning_rate": 6.50945674091516e-07,
      "loss": 0.6818,
      "step": 29177
    },
    {
      "epoch": 0.8407191840027661,
      "grad_norm": 0.7504528760910034,
      "learning_rate": 6.507154744508548e-07,
      "loss": 0.7092,
      "step": 29178
    },
    {
      "epoch": 0.8407479974644154,
      "grad_norm": 0.739283561706543,
      "learning_rate": 6.504853126881138e-07,
      "loss": 0.7039,
      "step": 29179
    },
    {
      "epoch": 0.8407768109260647,
      "grad_norm": 0.7457258105278015,
      "learning_rate": 6.502551888052993e-07,
      "loss": 0.706,
      "step": 29180
    },
    {
      "epoch": 0.8408056243877139,
      "grad_norm": 0.7282335758209229,
      "learning_rate": 6.50025102804413e-07,
      "loss": 0.6833,
      "step": 29181
    },
    {
      "epoch": 0.8408344378493632,
      "grad_norm": 0.7290641665458679,
      "learning_rate": 6.497950546874593e-07,
      "loss": 0.6652,
      "step": 29182
    },
    {
      "epoch": 0.8408632513110125,
      "grad_norm": 0.7218632102012634,
      "learning_rate": 6.495650444564433e-07,
      "loss": 0.6717,
      "step": 29183
    },
    {
      "epoch": 0.8408920647726618,
      "grad_norm": 0.7338702082633972,
      "learning_rate": 6.493350721133657e-07,
      "loss": 0.6862,
      "step": 29184
    },
    {
      "epoch": 0.840920878234311,
      "grad_norm": 0.7344332933425903,
      "learning_rate": 6.49105137660232e-07,
      "loss": 0.6815,
      "step": 29185
    },
    {
      "epoch": 0.8409496916959603,
      "grad_norm": 0.713987410068512,
      "learning_rate": 6.488752410990418e-07,
      "loss": 0.6908,
      "step": 29186
    },
    {
      "epoch": 0.8409785051576096,
      "grad_norm": 0.7254235148429871,
      "learning_rate": 6.486453824318e-07,
      "loss": 0.6749,
      "step": 29187
    },
    {
      "epoch": 0.8410073186192589,
      "grad_norm": 0.7066295146942139,
      "learning_rate": 6.484155616605059e-07,
      "loss": 0.6602,
      "step": 29188
    },
    {
      "epoch": 0.8410361320809082,
      "grad_norm": 0.7333784699440002,
      "learning_rate": 6.481857787871621e-07,
      "loss": 0.6808,
      "step": 29189
    },
    {
      "epoch": 0.8410649455425575,
      "grad_norm": 0.7201367616653442,
      "learning_rate": 6.479560338137702e-07,
      "loss": 0.6797,
      "step": 29190
    },
    {
      "epoch": 0.8410937590042068,
      "grad_norm": 0.7355621457099915,
      "learning_rate": 6.477263267423323e-07,
      "loss": 0.6745,
      "step": 29191
    },
    {
      "epoch": 0.8411225724658561,
      "grad_norm": 0.7061640620231628,
      "learning_rate": 6.474966575748459e-07,
      "loss": 0.6637,
      "step": 29192
    },
    {
      "epoch": 0.8411513859275054,
      "grad_norm": 0.70648592710495,
      "learning_rate": 6.472670263133135e-07,
      "loss": 0.6686,
      "step": 29193
    },
    {
      "epoch": 0.8411801993891546,
      "grad_norm": 0.7160478234291077,
      "learning_rate": 6.470374329597334e-07,
      "loss": 0.659,
      "step": 29194
    },
    {
      "epoch": 0.8412090128508039,
      "grad_norm": 0.7181525826454163,
      "learning_rate": 6.468078775161069e-07,
      "loss": 0.6628,
      "step": 29195
    },
    {
      "epoch": 0.8412378263124531,
      "grad_norm": 0.731835126876831,
      "learning_rate": 6.46578359984431e-07,
      "loss": 0.6658,
      "step": 29196
    },
    {
      "epoch": 0.8412666397741024,
      "grad_norm": 0.7209745049476624,
      "learning_rate": 6.463488803667073e-07,
      "loss": 0.6833,
      "step": 29197
    },
    {
      "epoch": 0.8412954532357517,
      "grad_norm": 0.7594584226608276,
      "learning_rate": 6.461194386649311e-07,
      "loss": 0.7083,
      "step": 29198
    },
    {
      "epoch": 0.841324266697401,
      "grad_norm": 0.7423915266990662,
      "learning_rate": 6.458900348811026e-07,
      "loss": 0.6696,
      "step": 29199
    },
    {
      "epoch": 0.8413530801590503,
      "grad_norm": 0.7428656220436096,
      "learning_rate": 6.456606690172207e-07,
      "loss": 0.7028,
      "step": 29200
    },
    {
      "epoch": 0.8413818936206996,
      "grad_norm": 0.7623273730278015,
      "learning_rate": 6.454313410752804e-07,
      "loss": 0.6951,
      "step": 29201
    },
    {
      "epoch": 0.8414107070823489,
      "grad_norm": 0.7139323353767395,
      "learning_rate": 6.452020510572799e-07,
      "loss": 0.6799,
      "step": 29202
    },
    {
      "epoch": 0.8414395205439982,
      "grad_norm": 0.728511393070221,
      "learning_rate": 6.449727989652166e-07,
      "loss": 0.6832,
      "step": 29203
    },
    {
      "epoch": 0.8414683340056475,
      "grad_norm": 0.7064540386199951,
      "learning_rate": 6.447435848010874e-07,
      "loss": 0.6539,
      "step": 29204
    },
    {
      "epoch": 0.8414971474672968,
      "grad_norm": 0.7272072434425354,
      "learning_rate": 6.445144085668875e-07,
      "loss": 0.6905,
      "step": 29205
    },
    {
      "epoch": 0.841525960928946,
      "grad_norm": 0.7200720906257629,
      "learning_rate": 6.442852702646141e-07,
      "loss": 0.6829,
      "step": 29206
    },
    {
      "epoch": 0.8415547743905953,
      "grad_norm": 0.7318350672721863,
      "learning_rate": 6.440561698962605e-07,
      "loss": 0.6948,
      "step": 29207
    },
    {
      "epoch": 0.8415835878522445,
      "grad_norm": 0.7256084084510803,
      "learning_rate": 6.438271074638247e-07,
      "loss": 0.6955,
      "step": 29208
    },
    {
      "epoch": 0.8416124013138938,
      "grad_norm": 0.7257792949676514,
      "learning_rate": 6.43598082969299e-07,
      "loss": 0.6846,
      "step": 29209
    },
    {
      "epoch": 0.8416412147755431,
      "grad_norm": 0.7114275097846985,
      "learning_rate": 6.433690964146799e-07,
      "loss": 0.6669,
      "step": 29210
    },
    {
      "epoch": 0.8416700282371924,
      "grad_norm": 0.7139978408813477,
      "learning_rate": 6.431401478019605e-07,
      "loss": 0.6525,
      "step": 29211
    },
    {
      "epoch": 0.8416988416988417,
      "grad_norm": 0.733109712600708,
      "learning_rate": 6.429112371331358e-07,
      "loss": 0.6951,
      "step": 29212
    },
    {
      "epoch": 0.841727655160491,
      "grad_norm": 0.7149056196212769,
      "learning_rate": 6.426823644101964e-07,
      "loss": 0.6942,
      "step": 29213
    },
    {
      "epoch": 0.8417564686221403,
      "grad_norm": 0.735369861125946,
      "learning_rate": 6.424535296351409e-07,
      "loss": 0.6746,
      "step": 29214
    },
    {
      "epoch": 0.8417852820837896,
      "grad_norm": 0.7402205467224121,
      "learning_rate": 6.422247328099574e-07,
      "loss": 0.6744,
      "step": 29215
    },
    {
      "epoch": 0.8418140955454388,
      "grad_norm": 0.7373019456863403,
      "learning_rate": 6.419959739366416e-07,
      "loss": 0.6704,
      "step": 29216
    },
    {
      "epoch": 0.8418429090070881,
      "grad_norm": 0.7320406436920166,
      "learning_rate": 6.417672530171832e-07,
      "loss": 0.7023,
      "step": 29217
    },
    {
      "epoch": 0.8418717224687374,
      "grad_norm": 0.7269023656845093,
      "learning_rate": 6.415385700535765e-07,
      "loss": 0.6681,
      "step": 29218
    },
    {
      "epoch": 0.8419005359303867,
      "grad_norm": 0.7277675271034241,
      "learning_rate": 6.413099250478111e-07,
      "loss": 0.6815,
      "step": 29219
    },
    {
      "epoch": 0.841929349392036,
      "grad_norm": 0.7333167791366577,
      "learning_rate": 6.410813180018793e-07,
      "loss": 0.7059,
      "step": 29220
    },
    {
      "epoch": 0.8419581628536852,
      "grad_norm": 0.718681275844574,
      "learning_rate": 6.408527489177724e-07,
      "loss": 0.6662,
      "step": 29221
    },
    {
      "epoch": 0.8419869763153345,
      "grad_norm": 0.7258471846580505,
      "learning_rate": 6.406242177974797e-07,
      "loss": 0.6857,
      "step": 29222
    },
    {
      "epoch": 0.8420157897769838,
      "grad_norm": 0.7191561460494995,
      "learning_rate": 6.403957246429937e-07,
      "loss": 0.6888,
      "step": 29223
    },
    {
      "epoch": 0.8420446032386331,
      "grad_norm": 0.7387733459472656,
      "learning_rate": 6.401672694563016e-07,
      "loss": 0.6728,
      "step": 29224
    },
    {
      "epoch": 0.8420734167002824,
      "grad_norm": 0.717675507068634,
      "learning_rate": 6.39938852239394e-07,
      "loss": 0.7089,
      "step": 29225
    },
    {
      "epoch": 0.8421022301619316,
      "grad_norm": 0.7241892218589783,
      "learning_rate": 6.39710472994261e-07,
      "loss": 0.6793,
      "step": 29226
    },
    {
      "epoch": 0.8421310436235809,
      "grad_norm": 0.7359380722045898,
      "learning_rate": 6.394821317228922e-07,
      "loss": 0.6904,
      "step": 29227
    },
    {
      "epoch": 0.8421598570852302,
      "grad_norm": 0.7159914970397949,
      "learning_rate": 6.392538284272737e-07,
      "loss": 0.6719,
      "step": 29228
    },
    {
      "epoch": 0.8421886705468795,
      "grad_norm": 0.735344648361206,
      "learning_rate": 6.390255631093967e-07,
      "loss": 0.6905,
      "step": 29229
    },
    {
      "epoch": 0.8422174840085288,
      "grad_norm": 0.7239437699317932,
      "learning_rate": 6.387973357712463e-07,
      "loss": 0.707,
      "step": 29230
    },
    {
      "epoch": 0.8422462974701781,
      "grad_norm": 0.7215097546577454,
      "learning_rate": 6.385691464148125e-07,
      "loss": 0.6702,
      "step": 29231
    },
    {
      "epoch": 0.8422751109318274,
      "grad_norm": 0.7190853953361511,
      "learning_rate": 6.383409950420804e-07,
      "loss": 0.6712,
      "step": 29232
    },
    {
      "epoch": 0.8423039243934767,
      "grad_norm": 0.7236627340316772,
      "learning_rate": 6.381128816550397e-07,
      "loss": 0.6946,
      "step": 29233
    },
    {
      "epoch": 0.842332737855126,
      "grad_norm": 0.7172616720199585,
      "learning_rate": 6.378848062556742e-07,
      "loss": 0.6961,
      "step": 29234
    },
    {
      "epoch": 0.8423615513167751,
      "grad_norm": 0.7459864616394043,
      "learning_rate": 6.376567688459723e-07,
      "loss": 0.7023,
      "step": 29235
    },
    {
      "epoch": 0.8423903647784244,
      "grad_norm": 0.7020838260650635,
      "learning_rate": 6.374287694279185e-07,
      "loss": 0.6712,
      "step": 29236
    },
    {
      "epoch": 0.8424191782400737,
      "grad_norm": 0.6982108354568481,
      "learning_rate": 6.372008080034992e-07,
      "loss": 0.6786,
      "step": 29237
    },
    {
      "epoch": 0.842447991701723,
      "grad_norm": 0.7120481133460999,
      "learning_rate": 6.369728845746998e-07,
      "loss": 0.6796,
      "step": 29238
    },
    {
      "epoch": 0.8424768051633723,
      "grad_norm": 0.7066956162452698,
      "learning_rate": 6.367449991435054e-07,
      "loss": 0.6757,
      "step": 29239
    },
    {
      "epoch": 0.8425056186250216,
      "grad_norm": 0.7946663498878479,
      "learning_rate": 6.365171517119001e-07,
      "loss": 0.6918,
      "step": 29240
    },
    {
      "epoch": 0.8425344320866709,
      "grad_norm": 0.7343288064002991,
      "learning_rate": 6.362893422818684e-07,
      "loss": 0.6849,
      "step": 29241
    },
    {
      "epoch": 0.8425632455483202,
      "grad_norm": 0.7156160473823547,
      "learning_rate": 6.360615708553953e-07,
      "loss": 0.6742,
      "step": 29242
    },
    {
      "epoch": 0.8425920590099695,
      "grad_norm": 0.7376039624214172,
      "learning_rate": 6.358338374344625e-07,
      "loss": 0.6741,
      "step": 29243
    },
    {
      "epoch": 0.8426208724716188,
      "grad_norm": 0.7136644721031189,
      "learning_rate": 6.356061420210558e-07,
      "loss": 0.6827,
      "step": 29244
    },
    {
      "epoch": 0.8426496859332681,
      "grad_norm": 0.756889820098877,
      "learning_rate": 6.353784846171552e-07,
      "loss": 0.6693,
      "step": 29245
    },
    {
      "epoch": 0.8426784993949173,
      "grad_norm": 0.7156919240951538,
      "learning_rate": 6.351508652247463e-07,
      "loss": 0.6864,
      "step": 29246
    },
    {
      "epoch": 0.8427073128565666,
      "grad_norm": 0.7361082434654236,
      "learning_rate": 6.349232838458091e-07,
      "loss": 0.6974,
      "step": 29247
    },
    {
      "epoch": 0.8427361263182158,
      "grad_norm": 0.7144671678543091,
      "learning_rate": 6.346957404823278e-07,
      "loss": 0.6716,
      "step": 29248
    },
    {
      "epoch": 0.8427649397798651,
      "grad_norm": 0.7223001718521118,
      "learning_rate": 6.344682351362807e-07,
      "loss": 0.6776,
      "step": 29249
    },
    {
      "epoch": 0.8427937532415144,
      "grad_norm": 0.8303882479667664,
      "learning_rate": 6.342407678096535e-07,
      "loss": 0.6768,
      "step": 29250
    },
    {
      "epoch": 0.8428225667031637,
      "grad_norm": 0.7059730291366577,
      "learning_rate": 6.340133385044239e-07,
      "loss": 0.6902,
      "step": 29251
    },
    {
      "epoch": 0.842851380164813,
      "grad_norm": 0.7199104428291321,
      "learning_rate": 6.337859472225743e-07,
      "loss": 0.6843,
      "step": 29252
    },
    {
      "epoch": 0.8428801936264623,
      "grad_norm": 0.7342395782470703,
      "learning_rate": 6.335585939660844e-07,
      "loss": 0.6804,
      "step": 29253
    },
    {
      "epoch": 0.8429090070881116,
      "grad_norm": 0.7276133894920349,
      "learning_rate": 6.333312787369344e-07,
      "loss": 0.6778,
      "step": 29254
    },
    {
      "epoch": 0.8429378205497609,
      "grad_norm": 0.722136914730072,
      "learning_rate": 6.331040015371037e-07,
      "loss": 0.6904,
      "step": 29255
    },
    {
      "epoch": 0.8429666340114101,
      "grad_norm": 0.7294650673866272,
      "learning_rate": 6.328767623685722e-07,
      "loss": 0.674,
      "step": 29256
    },
    {
      "epoch": 0.8429954474730594,
      "grad_norm": 0.7148212194442749,
      "learning_rate": 6.326495612333178e-07,
      "loss": 0.6894,
      "step": 29257
    },
    {
      "epoch": 0.8430242609347087,
      "grad_norm": 0.7204744815826416,
      "learning_rate": 6.324223981333199e-07,
      "loss": 0.6654,
      "step": 29258
    },
    {
      "epoch": 0.843053074396358,
      "grad_norm": 0.7524011135101318,
      "learning_rate": 6.321952730705582e-07,
      "loss": 0.6824,
      "step": 29259
    },
    {
      "epoch": 0.8430818878580073,
      "grad_norm": 0.7283197045326233,
      "learning_rate": 6.319681860470084e-07,
      "loss": 0.6891,
      "step": 29260
    },
    {
      "epoch": 0.8431107013196566,
      "grad_norm": 0.7231540083885193,
      "learning_rate": 6.317411370646492e-07,
      "loss": 0.6761,
      "step": 29261
    },
    {
      "epoch": 0.8431395147813058,
      "grad_norm": 0.7166106700897217,
      "learning_rate": 6.315141261254587e-07,
      "loss": 0.6966,
      "step": 29262
    },
    {
      "epoch": 0.8431683282429551,
      "grad_norm": 0.7394420504570007,
      "learning_rate": 6.31287153231413e-07,
      "loss": 0.7044,
      "step": 29263
    },
    {
      "epoch": 0.8431971417046044,
      "grad_norm": 0.7175294756889343,
      "learning_rate": 6.310602183844883e-07,
      "loss": 0.6755,
      "step": 29264
    },
    {
      "epoch": 0.8432259551662536,
      "grad_norm": 0.7271918654441833,
      "learning_rate": 6.308333215866635e-07,
      "loss": 0.6956,
      "step": 29265
    },
    {
      "epoch": 0.8432547686279029,
      "grad_norm": 0.7228686213493347,
      "learning_rate": 6.306064628399111e-07,
      "loss": 0.6914,
      "step": 29266
    },
    {
      "epoch": 0.8432835820895522,
      "grad_norm": 0.7189133167266846,
      "learning_rate": 6.303796421462105e-07,
      "loss": 0.6727,
      "step": 29267
    },
    {
      "epoch": 0.8433123955512015,
      "grad_norm": 0.7186652421951294,
      "learning_rate": 6.301528595075335e-07,
      "loss": 0.6861,
      "step": 29268
    },
    {
      "epoch": 0.8433412090128508,
      "grad_norm": 0.701718270778656,
      "learning_rate": 6.299261149258579e-07,
      "loss": 0.663,
      "step": 29269
    },
    {
      "epoch": 0.8433700224745001,
      "grad_norm": 0.7271353602409363,
      "learning_rate": 6.296994084031566e-07,
      "loss": 0.6867,
      "step": 29270
    },
    {
      "epoch": 0.8433988359361494,
      "grad_norm": 0.7169548273086548,
      "learning_rate": 6.294727399414058e-07,
      "loss": 0.7113,
      "step": 29271
    },
    {
      "epoch": 0.8434276493977987,
      "grad_norm": 0.7317166924476624,
      "learning_rate": 6.292461095425772e-07,
      "loss": 0.6997,
      "step": 29272
    },
    {
      "epoch": 0.843456462859448,
      "grad_norm": 0.7059676647186279,
      "learning_rate": 6.29019517208646e-07,
      "loss": 0.6675,
      "step": 29273
    },
    {
      "epoch": 0.8434852763210973,
      "grad_norm": 0.7224125862121582,
      "learning_rate": 6.287929629415857e-07,
      "loss": 0.6827,
      "step": 29274
    },
    {
      "epoch": 0.8435140897827464,
      "grad_norm": 0.7244434356689453,
      "learning_rate": 6.285664467433699e-07,
      "loss": 0.6694,
      "step": 29275
    },
    {
      "epoch": 0.8435429032443957,
      "grad_norm": 0.705514669418335,
      "learning_rate": 6.283399686159692e-07,
      "loss": 0.6632,
      "step": 29276
    },
    {
      "epoch": 0.843571716706045,
      "grad_norm": 0.7213442325592041,
      "learning_rate": 6.281135285613587e-07,
      "loss": 0.6942,
      "step": 29277
    },
    {
      "epoch": 0.8436005301676943,
      "grad_norm": 0.7328453660011292,
      "learning_rate": 6.278871265815079e-07,
      "loss": 0.6992,
      "step": 29278
    },
    {
      "epoch": 0.8436293436293436,
      "grad_norm": 0.7514608502388,
      "learning_rate": 6.276607626783899e-07,
      "loss": 0.6892,
      "step": 29279
    },
    {
      "epoch": 0.8436581570909929,
      "grad_norm": 0.7120373249053955,
      "learning_rate": 6.274344368539764e-07,
      "loss": 0.6637,
      "step": 29280
    },
    {
      "epoch": 0.8436869705526422,
      "grad_norm": 0.7228818535804749,
      "learning_rate": 6.272081491102378e-07,
      "loss": 0.6997,
      "step": 29281
    },
    {
      "epoch": 0.8437157840142915,
      "grad_norm": 0.7459698915481567,
      "learning_rate": 6.269818994491455e-07,
      "loss": 0.6754,
      "step": 29282
    },
    {
      "epoch": 0.8437445974759408,
      "grad_norm": 0.7325377464294434,
      "learning_rate": 6.267556878726682e-07,
      "loss": 0.698,
      "step": 29283
    },
    {
      "epoch": 0.8437734109375901,
      "grad_norm": 0.7350168824195862,
      "learning_rate": 6.265295143827787e-07,
      "loss": 0.6964,
      "step": 29284
    },
    {
      "epoch": 0.8438022243992394,
      "grad_norm": 0.7615273594856262,
      "learning_rate": 6.263033789814432e-07,
      "loss": 0.695,
      "step": 29285
    },
    {
      "epoch": 0.8438310378608886,
      "grad_norm": 0.7263681888580322,
      "learning_rate": 6.26077281670635e-07,
      "loss": 0.6837,
      "step": 29286
    },
    {
      "epoch": 0.8438598513225379,
      "grad_norm": 0.7319376468658447,
      "learning_rate": 6.258512224523206e-07,
      "loss": 0.6727,
      "step": 29287
    },
    {
      "epoch": 0.8438886647841872,
      "grad_norm": 0.7201719880104065,
      "learning_rate": 6.256252013284703e-07,
      "loss": 0.6897,
      "step": 29288
    },
    {
      "epoch": 0.8439174782458364,
      "grad_norm": 0.7071214914321899,
      "learning_rate": 6.253992183010504e-07,
      "loss": 0.6996,
      "step": 29289
    },
    {
      "epoch": 0.8439462917074857,
      "grad_norm": 0.7181738615036011,
      "learning_rate": 6.251732733720322e-07,
      "loss": 0.6808,
      "step": 29290
    },
    {
      "epoch": 0.843975105169135,
      "grad_norm": 0.7147303819656372,
      "learning_rate": 6.249473665433797e-07,
      "loss": 0.6734,
      "step": 29291
    },
    {
      "epoch": 0.8440039186307843,
      "grad_norm": 0.7074597477912903,
      "learning_rate": 6.247214978170635e-07,
      "loss": 0.6794,
      "step": 29292
    },
    {
      "epoch": 0.8440327320924336,
      "grad_norm": 0.7302065491676331,
      "learning_rate": 6.244956671950486e-07,
      "loss": 0.6856,
      "step": 29293
    },
    {
      "epoch": 0.8440615455540829,
      "grad_norm": 0.7038699984550476,
      "learning_rate": 6.242698746793036e-07,
      "loss": 0.6778,
      "step": 29294
    },
    {
      "epoch": 0.8440903590157322,
      "grad_norm": 0.7222652435302734,
      "learning_rate": 6.240441202717923e-07,
      "loss": 0.679,
      "step": 29295
    },
    {
      "epoch": 0.8441191724773814,
      "grad_norm": 0.7091370224952698,
      "learning_rate": 6.238184039744827e-07,
      "loss": 0.675,
      "step": 29296
    },
    {
      "epoch": 0.8441479859390307,
      "grad_norm": 0.7238255739212036,
      "learning_rate": 6.235927257893399e-07,
      "loss": 0.6922,
      "step": 29297
    },
    {
      "epoch": 0.84417679940068,
      "grad_norm": 0.7330099940299988,
      "learning_rate": 6.23367085718331e-07,
      "loss": 0.6886,
      "step": 29298
    },
    {
      "epoch": 0.8442056128623293,
      "grad_norm": 0.7259418368339539,
      "learning_rate": 6.231414837634187e-07,
      "loss": 0.678,
      "step": 29299
    },
    {
      "epoch": 0.8442344263239786,
      "grad_norm": 0.7293199896812439,
      "learning_rate": 6.2291591992657e-07,
      "loss": 0.6791,
      "step": 29300
    },
    {
      "epoch": 0.8442632397856279,
      "grad_norm": 0.7119577527046204,
      "learning_rate": 6.226903942097467e-07,
      "loss": 0.674,
      "step": 29301
    },
    {
      "epoch": 0.8442920532472771,
      "grad_norm": 0.7347997426986694,
      "learning_rate": 6.224649066149146e-07,
      "loss": 0.6917,
      "step": 29302
    },
    {
      "epoch": 0.8443208667089264,
      "grad_norm": 0.7160171866416931,
      "learning_rate": 6.222394571440382e-07,
      "loss": 0.6917,
      "step": 29303
    },
    {
      "epoch": 0.8443496801705757,
      "grad_norm": 0.7195186614990234,
      "learning_rate": 6.220140457990792e-07,
      "loss": 0.6767,
      "step": 29304
    },
    {
      "epoch": 0.844378493632225,
      "grad_norm": 0.7433479428291321,
      "learning_rate": 6.217886725820022e-07,
      "loss": 0.6866,
      "step": 29305
    },
    {
      "epoch": 0.8444073070938742,
      "grad_norm": 0.719377338886261,
      "learning_rate": 6.215633374947682e-07,
      "loss": 0.6554,
      "step": 29306
    },
    {
      "epoch": 0.8444361205555235,
      "grad_norm": 0.7291415929794312,
      "learning_rate": 6.213380405393421e-07,
      "loss": 0.6733,
      "step": 29307
    },
    {
      "epoch": 0.8444649340171728,
      "grad_norm": 0.7061198949813843,
      "learning_rate": 6.211127817176831e-07,
      "loss": 0.6734,
      "step": 29308
    },
    {
      "epoch": 0.8444937474788221,
      "grad_norm": 0.7210624814033508,
      "learning_rate": 6.20887561031755e-07,
      "loss": 0.7038,
      "step": 29309
    },
    {
      "epoch": 0.8445225609404714,
      "grad_norm": 0.7167291641235352,
      "learning_rate": 6.206623784835181e-07,
      "loss": 0.6932,
      "step": 29310
    },
    {
      "epoch": 0.8445513744021207,
      "grad_norm": 0.7373384237289429,
      "learning_rate": 6.204372340749359e-07,
      "loss": 0.6985,
      "step": 29311
    },
    {
      "epoch": 0.84458018786377,
      "grad_norm": 0.7032021284103394,
      "learning_rate": 6.202121278079659e-07,
      "loss": 0.6627,
      "step": 29312
    },
    {
      "epoch": 0.8446090013254193,
      "grad_norm": 0.7331556677818298,
      "learning_rate": 6.199870596845715e-07,
      "loss": 0.6988,
      "step": 29313
    },
    {
      "epoch": 0.8446378147870686,
      "grad_norm": 0.734082818031311,
      "learning_rate": 6.197620297067098e-07,
      "loss": 0.6783,
      "step": 29314
    },
    {
      "epoch": 0.8446666282487179,
      "grad_norm": 0.7088995575904846,
      "learning_rate": 6.195370378763438e-07,
      "loss": 0.6632,
      "step": 29315
    },
    {
      "epoch": 0.844695441710367,
      "grad_norm": 0.7233037352561951,
      "learning_rate": 6.193120841954297e-07,
      "loss": 0.6908,
      "step": 29316
    },
    {
      "epoch": 0.8447242551720163,
      "grad_norm": 0.7237614989280701,
      "learning_rate": 6.190871686659289e-07,
      "loss": 0.6797,
      "step": 29317
    },
    {
      "epoch": 0.8447530686336656,
      "grad_norm": 0.7388573288917542,
      "learning_rate": 6.188622912898002e-07,
      "loss": 0.6819,
      "step": 29318
    },
    {
      "epoch": 0.8447818820953149,
      "grad_norm": 0.7199153900146484,
      "learning_rate": 6.186374520690003e-07,
      "loss": 0.6882,
      "step": 29319
    },
    {
      "epoch": 0.8448106955569642,
      "grad_norm": 0.7250511646270752,
      "learning_rate": 6.184126510054894e-07,
      "loss": 0.6767,
      "step": 29320
    },
    {
      "epoch": 0.8448395090186135,
      "grad_norm": 0.7285598516464233,
      "learning_rate": 6.181878881012238e-07,
      "loss": 0.6838,
      "step": 29321
    },
    {
      "epoch": 0.8448683224802628,
      "grad_norm": 0.7287577986717224,
      "learning_rate": 6.179631633581612e-07,
      "loss": 0.7046,
      "step": 29322
    },
    {
      "epoch": 0.8448971359419121,
      "grad_norm": 0.7131011486053467,
      "learning_rate": 6.17738476778259e-07,
      "loss": 0.6661,
      "step": 29323
    },
    {
      "epoch": 0.8449259494035614,
      "grad_norm": 0.7386696934700012,
      "learning_rate": 6.175138283634752e-07,
      "loss": 0.7111,
      "step": 29324
    },
    {
      "epoch": 0.8449547628652107,
      "grad_norm": 0.7072945237159729,
      "learning_rate": 6.172892181157636e-07,
      "loss": 0.6876,
      "step": 29325
    },
    {
      "epoch": 0.8449835763268599,
      "grad_norm": 0.7459288239479065,
      "learning_rate": 6.170646460370833e-07,
      "loss": 0.6954,
      "step": 29326
    },
    {
      "epoch": 0.8450123897885092,
      "grad_norm": 0.7336947917938232,
      "learning_rate": 6.168401121293871e-07,
      "loss": 0.6583,
      "step": 29327
    },
    {
      "epoch": 0.8450412032501585,
      "grad_norm": 0.7351836562156677,
      "learning_rate": 6.166156163946336e-07,
      "loss": 0.708,
      "step": 29328
    },
    {
      "epoch": 0.8450700167118077,
      "grad_norm": 0.7234265804290771,
      "learning_rate": 6.163911588347748e-07,
      "loss": 0.6815,
      "step": 29329
    },
    {
      "epoch": 0.845098830173457,
      "grad_norm": 1.4411661624908447,
      "learning_rate": 6.161667394517684e-07,
      "loss": 0.6877,
      "step": 29330
    },
    {
      "epoch": 0.8451276436351063,
      "grad_norm": 0.7092885375022888,
      "learning_rate": 6.159423582475659e-07,
      "loss": 0.6823,
      "step": 29331
    },
    {
      "epoch": 0.8451564570967556,
      "grad_norm": 0.7229812741279602,
      "learning_rate": 6.157180152241243e-07,
      "loss": 0.7007,
      "step": 29332
    },
    {
      "epoch": 0.8451852705584049,
      "grad_norm": 0.7203019857406616,
      "learning_rate": 6.154937103833941e-07,
      "loss": 0.6974,
      "step": 29333
    },
    {
      "epoch": 0.8452140840200542,
      "grad_norm": 0.7260763645172119,
      "learning_rate": 6.152694437273332e-07,
      "loss": 0.6739,
      "step": 29334
    },
    {
      "epoch": 0.8452428974817034,
      "grad_norm": 0.73030024766922,
      "learning_rate": 6.150452152578907e-07,
      "loss": 0.6848,
      "step": 29335
    },
    {
      "epoch": 0.8452717109433527,
      "grad_norm": 0.7391850352287292,
      "learning_rate": 6.148210249770226e-07,
      "loss": 0.6754,
      "step": 29336
    },
    {
      "epoch": 0.845300524405002,
      "grad_norm": 0.7494385242462158,
      "learning_rate": 6.145968728866785e-07,
      "loss": 0.6803,
      "step": 29337
    },
    {
      "epoch": 0.8453293378666513,
      "grad_norm": 0.7687935829162598,
      "learning_rate": 6.143727589888127e-07,
      "loss": 0.6951,
      "step": 29338
    },
    {
      "epoch": 0.8453581513283006,
      "grad_norm": 0.7036562561988831,
      "learning_rate": 6.141486832853754e-07,
      "loss": 0.6743,
      "step": 29339
    },
    {
      "epoch": 0.8453869647899499,
      "grad_norm": 0.734364926815033,
      "learning_rate": 6.139246457783188e-07,
      "loss": 0.6712,
      "step": 29340
    },
    {
      "epoch": 0.8454157782515992,
      "grad_norm": 0.7253243923187256,
      "learning_rate": 6.137006464695949e-07,
      "loss": 0.7005,
      "step": 29341
    },
    {
      "epoch": 0.8454445917132485,
      "grad_norm": 0.743458092212677,
      "learning_rate": 6.134766853611524e-07,
      "loss": 0.6759,
      "step": 29342
    },
    {
      "epoch": 0.8454734051748977,
      "grad_norm": 0.7317010760307312,
      "learning_rate": 6.132527624549444e-07,
      "loss": 0.7016,
      "step": 29343
    },
    {
      "epoch": 0.845502218636547,
      "grad_norm": 0.7312187552452087,
      "learning_rate": 6.130288777529186e-07,
      "loss": 0.6771,
      "step": 29344
    },
    {
      "epoch": 0.8455310320981962,
      "grad_norm": 0.7201595306396484,
      "learning_rate": 6.12805031257026e-07,
      "loss": 0.6946,
      "step": 29345
    },
    {
      "epoch": 0.8455598455598455,
      "grad_norm": 0.7447811961174011,
      "learning_rate": 6.125812229692163e-07,
      "loss": 0.682,
      "step": 29346
    },
    {
      "epoch": 0.8455886590214948,
      "grad_norm": 0.7230570316314697,
      "learning_rate": 6.123574528914388e-07,
      "loss": 0.6633,
      "step": 29347
    },
    {
      "epoch": 0.8456174724831441,
      "grad_norm": 0.7621574997901917,
      "learning_rate": 6.121337210256411e-07,
      "loss": 0.6781,
      "step": 29348
    },
    {
      "epoch": 0.8456462859447934,
      "grad_norm": 0.7161246538162231,
      "learning_rate": 6.119100273737733e-07,
      "loss": 0.6968,
      "step": 29349
    },
    {
      "epoch": 0.8456750994064427,
      "grad_norm": 0.7254074215888977,
      "learning_rate": 6.11686371937782e-07,
      "loss": 0.6842,
      "step": 29350
    },
    {
      "epoch": 0.845703912868092,
      "grad_norm": 0.7119266390800476,
      "learning_rate": 6.114627547196167e-07,
      "loss": 0.6706,
      "step": 29351
    },
    {
      "epoch": 0.8457327263297413,
      "grad_norm": 0.711065948009491,
      "learning_rate": 6.112391757212222e-07,
      "loss": 0.6854,
      "step": 29352
    },
    {
      "epoch": 0.8457615397913906,
      "grad_norm": 0.7298967838287354,
      "learning_rate": 6.11015634944549e-07,
      "loss": 0.692,
      "step": 29353
    },
    {
      "epoch": 0.8457903532530399,
      "grad_norm": 0.7230201363563538,
      "learning_rate": 6.107921323915411e-07,
      "loss": 0.6876,
      "step": 29354
    },
    {
      "epoch": 0.8458191667146892,
      "grad_norm": 0.7256633043289185,
      "learning_rate": 6.105686680641459e-07,
      "loss": 0.6919,
      "step": 29355
    },
    {
      "epoch": 0.8458479801763383,
      "grad_norm": 0.7204330563545227,
      "learning_rate": 6.103452419643108e-07,
      "loss": 0.6706,
      "step": 29356
    },
    {
      "epoch": 0.8458767936379876,
      "grad_norm": 0.742083728313446,
      "learning_rate": 6.101218540939796e-07,
      "loss": 0.6977,
      "step": 29357
    },
    {
      "epoch": 0.8459056070996369,
      "grad_norm": 0.7039861083030701,
      "learning_rate": 6.098985044550992e-07,
      "loss": 0.661,
      "step": 29358
    },
    {
      "epoch": 0.8459344205612862,
      "grad_norm": 0.7274092435836792,
      "learning_rate": 6.096751930496147e-07,
      "loss": 0.6897,
      "step": 29359
    },
    {
      "epoch": 0.8459632340229355,
      "grad_norm": 0.7156088948249817,
      "learning_rate": 6.094519198794702e-07,
      "loss": 0.6879,
      "step": 29360
    },
    {
      "epoch": 0.8459920474845848,
      "grad_norm": 0.7385624647140503,
      "learning_rate": 6.0922868494661e-07,
      "loss": 0.7047,
      "step": 29361
    },
    {
      "epoch": 0.8460208609462341,
      "grad_norm": 0.7213502526283264,
      "learning_rate": 6.0900548825298e-07,
      "loss": 0.6804,
      "step": 29362
    },
    {
      "epoch": 0.8460496744078834,
      "grad_norm": 0.723626434803009,
      "learning_rate": 6.087823298005219e-07,
      "loss": 0.6987,
      "step": 29363
    },
    {
      "epoch": 0.8460784878695327,
      "grad_norm": 0.722857654094696,
      "learning_rate": 6.085592095911807e-07,
      "loss": 0.6928,
      "step": 29364
    },
    {
      "epoch": 0.846107301331182,
      "grad_norm": 0.7451149225234985,
      "learning_rate": 6.083361276268984e-07,
      "loss": 0.6977,
      "step": 29365
    },
    {
      "epoch": 0.8461361147928312,
      "grad_norm": 0.745026171207428,
      "learning_rate": 6.081130839096194e-07,
      "loss": 0.6934,
      "step": 29366
    },
    {
      "epoch": 0.8461649282544805,
      "grad_norm": 0.7173545360565186,
      "learning_rate": 6.078900784412839e-07,
      "loss": 0.6863,
      "step": 29367
    },
    {
      "epoch": 0.8461937417161298,
      "grad_norm": 0.713838517665863,
      "learning_rate": 6.076671112238363e-07,
      "loss": 0.6801,
      "step": 29368
    },
    {
      "epoch": 0.8462225551777791,
      "grad_norm": 0.7391849756240845,
      "learning_rate": 6.074441822592153e-07,
      "loss": 0.7141,
      "step": 29369
    },
    {
      "epoch": 0.8462513686394283,
      "grad_norm": 0.7355865240097046,
      "learning_rate": 6.07221291549367e-07,
      "loss": 0.6713,
      "step": 29370
    },
    {
      "epoch": 0.8462801821010776,
      "grad_norm": 0.7448136806488037,
      "learning_rate": 6.069984390962286e-07,
      "loss": 0.6941,
      "step": 29371
    },
    {
      "epoch": 0.8463089955627269,
      "grad_norm": 0.7159316539764404,
      "learning_rate": 6.067756249017437e-07,
      "loss": 0.6853,
      "step": 29372
    },
    {
      "epoch": 0.8463378090243762,
      "grad_norm": 0.7160570621490479,
      "learning_rate": 6.065528489678508e-07,
      "loss": 0.6988,
      "step": 29373
    },
    {
      "epoch": 0.8463666224860255,
      "grad_norm": 0.7572740316390991,
      "learning_rate": 6.063301112964908e-07,
      "loss": 0.7037,
      "step": 29374
    },
    {
      "epoch": 0.8463954359476747,
      "grad_norm": 0.7191334962844849,
      "learning_rate": 6.061074118896032e-07,
      "loss": 0.6809,
      "step": 29375
    },
    {
      "epoch": 0.846424249409324,
      "grad_norm": 0.723975658416748,
      "learning_rate": 6.058847507491277e-07,
      "loss": 0.6837,
      "step": 29376
    },
    {
      "epoch": 0.8464530628709733,
      "grad_norm": 0.738881528377533,
      "learning_rate": 6.056621278770042e-07,
      "loss": 0.67,
      "step": 29377
    },
    {
      "epoch": 0.8464818763326226,
      "grad_norm": 0.7221434712409973,
      "learning_rate": 6.054395432751703e-07,
      "loss": 0.696,
      "step": 29378
    },
    {
      "epoch": 0.8465106897942719,
      "grad_norm": 0.7146788239479065,
      "learning_rate": 6.052169969455657e-07,
      "loss": 0.6923,
      "step": 29379
    },
    {
      "epoch": 0.8465395032559212,
      "grad_norm": 0.7301772236824036,
      "learning_rate": 6.049944888901272e-07,
      "loss": 0.6878,
      "step": 29380
    },
    {
      "epoch": 0.8465683167175705,
      "grad_norm": 0.7320805788040161,
      "learning_rate": 6.047720191107925e-07,
      "loss": 0.6738,
      "step": 29381
    },
    {
      "epoch": 0.8465971301792198,
      "grad_norm": 0.7338646054267883,
      "learning_rate": 6.045495876095009e-07,
      "loss": 0.6774,
      "step": 29382
    },
    {
      "epoch": 0.846625943640869,
      "grad_norm": 0.7200069427490234,
      "learning_rate": 6.04327194388189e-07,
      "loss": 0.6809,
      "step": 29383
    },
    {
      "epoch": 0.8466547571025183,
      "grad_norm": 0.7142131924629211,
      "learning_rate": 6.04104839448792e-07,
      "loss": 0.6802,
      "step": 29384
    },
    {
      "epoch": 0.8466835705641675,
      "grad_norm": 0.7214905023574829,
      "learning_rate": 6.038825227932488e-07,
      "loss": 0.6913,
      "step": 29385
    },
    {
      "epoch": 0.8467123840258168,
      "grad_norm": 0.7156069278717041,
      "learning_rate": 6.036602444234934e-07,
      "loss": 0.6718,
      "step": 29386
    },
    {
      "epoch": 0.8467411974874661,
      "grad_norm": 0.7122294902801514,
      "learning_rate": 6.034380043414634e-07,
      "loss": 0.6543,
      "step": 29387
    },
    {
      "epoch": 0.8467700109491154,
      "grad_norm": 0.7179480791091919,
      "learning_rate": 6.032158025490925e-07,
      "loss": 0.6692,
      "step": 29388
    },
    {
      "epoch": 0.8467988244107647,
      "grad_norm": 0.7559058666229248,
      "learning_rate": 6.029936390483182e-07,
      "loss": 0.6749,
      "step": 29389
    },
    {
      "epoch": 0.846827637872414,
      "grad_norm": 0.7239348888397217,
      "learning_rate": 6.027715138410722e-07,
      "loss": 0.7059,
      "step": 29390
    },
    {
      "epoch": 0.8468564513340633,
      "grad_norm": 0.7346360683441162,
      "learning_rate": 6.02549426929292e-07,
      "loss": 0.6987,
      "step": 29391
    },
    {
      "epoch": 0.8468852647957126,
      "grad_norm": 0.7191421389579773,
      "learning_rate": 6.023273783149097e-07,
      "loss": 0.6737,
      "step": 29392
    },
    {
      "epoch": 0.8469140782573619,
      "grad_norm": 0.7114164233207703,
      "learning_rate": 6.021053679998595e-07,
      "loss": 0.6745,
      "step": 29393
    },
    {
      "epoch": 0.8469428917190112,
      "grad_norm": 0.7302844524383545,
      "learning_rate": 6.018833959860753e-07,
      "loss": 0.6915,
      "step": 29394
    },
    {
      "epoch": 0.8469717051806605,
      "grad_norm": 0.7118366360664368,
      "learning_rate": 6.016614622754913e-07,
      "loss": 0.6695,
      "step": 29395
    },
    {
      "epoch": 0.8470005186423097,
      "grad_norm": 0.7122609615325928,
      "learning_rate": 6.014395668700385e-07,
      "loss": 0.6705,
      "step": 29396
    },
    {
      "epoch": 0.8470293321039589,
      "grad_norm": 0.7206278443336487,
      "learning_rate": 6.012177097716504e-07,
      "loss": 0.6782,
      "step": 29397
    },
    {
      "epoch": 0.8470581455656082,
      "grad_norm": 0.7108685970306396,
      "learning_rate": 6.009958909822583e-07,
      "loss": 0.6946,
      "step": 29398
    },
    {
      "epoch": 0.8470869590272575,
      "grad_norm": 0.7362055778503418,
      "learning_rate": 6.007741105037951e-07,
      "loss": 0.7004,
      "step": 29399
    },
    {
      "epoch": 0.8471157724889068,
      "grad_norm": 0.7184761762619019,
      "learning_rate": 6.00552368338192e-07,
      "loss": 0.6821,
      "step": 29400
    },
    {
      "epoch": 0.8471445859505561,
      "grad_norm": 0.749254584312439,
      "learning_rate": 6.003306644873791e-07,
      "loss": 0.6803,
      "step": 29401
    },
    {
      "epoch": 0.8471733994122054,
      "grad_norm": 0.7177945375442505,
      "learning_rate": 6.001089989532893e-07,
      "loss": 0.6704,
      "step": 29402
    },
    {
      "epoch": 0.8472022128738547,
      "grad_norm": 0.7107280492782593,
      "learning_rate": 5.998873717378506e-07,
      "loss": 0.6899,
      "step": 29403
    },
    {
      "epoch": 0.847231026335504,
      "grad_norm": 0.7042818069458008,
      "learning_rate": 5.996657828429958e-07,
      "loss": 0.6856,
      "step": 29404
    },
    {
      "epoch": 0.8472598397971532,
      "grad_norm": 0.7322944402694702,
      "learning_rate": 5.994442322706506e-07,
      "loss": 0.6914,
      "step": 29405
    },
    {
      "epoch": 0.8472886532588025,
      "grad_norm": 0.7170947790145874,
      "learning_rate": 5.992227200227502e-07,
      "loss": 0.6754,
      "step": 29406
    },
    {
      "epoch": 0.8473174667204518,
      "grad_norm": 0.7200278639793396,
      "learning_rate": 5.99001246101219e-07,
      "loss": 0.681,
      "step": 29407
    },
    {
      "epoch": 0.8473462801821011,
      "grad_norm": 0.712479829788208,
      "learning_rate": 5.987798105079884e-07,
      "loss": 0.6783,
      "step": 29408
    },
    {
      "epoch": 0.8473750936437504,
      "grad_norm": 0.6949347257614136,
      "learning_rate": 5.985584132449856e-07,
      "loss": 0.6679,
      "step": 29409
    },
    {
      "epoch": 0.8474039071053996,
      "grad_norm": 0.7142931818962097,
      "learning_rate": 5.983370543141397e-07,
      "loss": 0.6736,
      "step": 29410
    },
    {
      "epoch": 0.8474327205670489,
      "grad_norm": 0.7201701402664185,
      "learning_rate": 5.981157337173777e-07,
      "loss": 0.6742,
      "step": 29411
    },
    {
      "epoch": 0.8474615340286982,
      "grad_norm": 0.7342038750648499,
      "learning_rate": 5.978944514566276e-07,
      "loss": 0.6787,
      "step": 29412
    },
    {
      "epoch": 0.8474903474903475,
      "grad_norm": 0.7205685377120972,
      "learning_rate": 5.97673207533816e-07,
      "loss": 0.6719,
      "step": 29413
    },
    {
      "epoch": 0.8475191609519968,
      "grad_norm": 0.7193386554718018,
      "learning_rate": 5.974520019508695e-07,
      "loss": 0.6919,
      "step": 29414
    },
    {
      "epoch": 0.847547974413646,
      "grad_norm": 0.7795517444610596,
      "learning_rate": 5.972308347097161e-07,
      "loss": 0.7013,
      "step": 29415
    },
    {
      "epoch": 0.8475767878752953,
      "grad_norm": 0.7456097602844238,
      "learning_rate": 5.970097058122804e-07,
      "loss": 0.6971,
      "step": 29416
    },
    {
      "epoch": 0.8476056013369446,
      "grad_norm": 0.7252076864242554,
      "learning_rate": 5.967886152604885e-07,
      "loss": 0.7041,
      "step": 29417
    },
    {
      "epoch": 0.8476344147985939,
      "grad_norm": 0.7243686318397522,
      "learning_rate": 5.965675630562672e-07,
      "loss": 0.6785,
      "step": 29418
    },
    {
      "epoch": 0.8476632282602432,
      "grad_norm": 0.7325416207313538,
      "learning_rate": 5.963465492015396e-07,
      "loss": 0.6708,
      "step": 29419
    },
    {
      "epoch": 0.8476920417218925,
      "grad_norm": 0.7550628781318665,
      "learning_rate": 5.961255736982318e-07,
      "loss": 0.6734,
      "step": 29420
    },
    {
      "epoch": 0.8477208551835418,
      "grad_norm": 0.7295154333114624,
      "learning_rate": 5.959046365482685e-07,
      "loss": 0.6776,
      "step": 29421
    },
    {
      "epoch": 0.8477496686451911,
      "grad_norm": 0.6897397041320801,
      "learning_rate": 5.956837377535729e-07,
      "loss": 0.6654,
      "step": 29422
    },
    {
      "epoch": 0.8477784821068404,
      "grad_norm": 0.7477818727493286,
      "learning_rate": 5.954628773160698e-07,
      "loss": 0.699,
      "step": 29423
    },
    {
      "epoch": 0.8478072955684895,
      "grad_norm": 0.7564745545387268,
      "learning_rate": 5.952420552376814e-07,
      "loss": 0.6672,
      "step": 29424
    },
    {
      "epoch": 0.8478361090301388,
      "grad_norm": 0.7154083251953125,
      "learning_rate": 5.950212715203324e-07,
      "loss": 0.6814,
      "step": 29425
    },
    {
      "epoch": 0.8478649224917881,
      "grad_norm": 0.7165635228157043,
      "learning_rate": 5.948005261659434e-07,
      "loss": 0.6872,
      "step": 29426
    },
    {
      "epoch": 0.8478937359534374,
      "grad_norm": 0.7315196394920349,
      "learning_rate": 5.945798191764402e-07,
      "loss": 0.6947,
      "step": 29427
    },
    {
      "epoch": 0.8479225494150867,
      "grad_norm": 0.7323501706123352,
      "learning_rate": 5.94359150553741e-07,
      "loss": 0.6871,
      "step": 29428
    },
    {
      "epoch": 0.847951362876736,
      "grad_norm": 0.7231778502464294,
      "learning_rate": 5.941385202997702e-07,
      "loss": 0.7042,
      "step": 29429
    },
    {
      "epoch": 0.8479801763383853,
      "grad_norm": 0.7345454096794128,
      "learning_rate": 5.939179284164482e-07,
      "loss": 0.6715,
      "step": 29430
    },
    {
      "epoch": 0.8480089898000346,
      "grad_norm": 0.7002178430557251,
      "learning_rate": 5.936973749056978e-07,
      "loss": 0.6424,
      "step": 29431
    },
    {
      "epoch": 0.8480378032616839,
      "grad_norm": 0.7214767336845398,
      "learning_rate": 5.934768597694373e-07,
      "loss": 0.687,
      "step": 29432
    },
    {
      "epoch": 0.8480666167233332,
      "grad_norm": 0.7198511362075806,
      "learning_rate": 5.932563830095894e-07,
      "loss": 0.6647,
      "step": 29433
    },
    {
      "epoch": 0.8480954301849825,
      "grad_norm": 0.7389722466468811,
      "learning_rate": 5.930359446280725e-07,
      "loss": 0.69,
      "step": 29434
    },
    {
      "epoch": 0.8481242436466317,
      "grad_norm": 0.713607668876648,
      "learning_rate": 5.928155446268085e-07,
      "loss": 0.6777,
      "step": 29435
    },
    {
      "epoch": 0.848153057108281,
      "grad_norm": 0.7091788649559021,
      "learning_rate": 5.925951830077142e-07,
      "loss": 0.6526,
      "step": 29436
    },
    {
      "epoch": 0.8481818705699302,
      "grad_norm": 0.728894054889679,
      "learning_rate": 5.923748597727103e-07,
      "loss": 0.6956,
      "step": 29437
    },
    {
      "epoch": 0.8482106840315795,
      "grad_norm": 0.7222148776054382,
      "learning_rate": 5.921545749237162e-07,
      "loss": 0.6841,
      "step": 29438
    },
    {
      "epoch": 0.8482394974932288,
      "grad_norm": 0.7051437497138977,
      "learning_rate": 5.919343284626483e-07,
      "loss": 0.6679,
      "step": 29439
    },
    {
      "epoch": 0.8482683109548781,
      "grad_norm": 0.7239149808883667,
      "learning_rate": 5.917141203914273e-07,
      "loss": 0.7137,
      "step": 29440
    },
    {
      "epoch": 0.8482971244165274,
      "grad_norm": 0.7278619408607483,
      "learning_rate": 5.914939507119671e-07,
      "loss": 0.6692,
      "step": 29441
    },
    {
      "epoch": 0.8483259378781767,
      "grad_norm": 0.7220496535301208,
      "learning_rate": 5.912738194261902e-07,
      "loss": 0.6632,
      "step": 29442
    },
    {
      "epoch": 0.848354751339826,
      "grad_norm": 0.7410841584205627,
      "learning_rate": 5.9105372653601e-07,
      "loss": 0.6964,
      "step": 29443
    },
    {
      "epoch": 0.8483835648014753,
      "grad_norm": 0.7472769021987915,
      "learning_rate": 5.908336720433455e-07,
      "loss": 0.7058,
      "step": 29444
    },
    {
      "epoch": 0.8484123782631245,
      "grad_norm": 0.6848234534263611,
      "learning_rate": 5.906136559501113e-07,
      "loss": 0.6608,
      "step": 29445
    },
    {
      "epoch": 0.8484411917247738,
      "grad_norm": 0.7217599749565125,
      "learning_rate": 5.903936782582253e-07,
      "loss": 0.6735,
      "step": 29446
    },
    {
      "epoch": 0.8484700051864231,
      "grad_norm": 0.7181175351142883,
      "learning_rate": 5.901737389696016e-07,
      "loss": 0.6793,
      "step": 29447
    },
    {
      "epoch": 0.8484988186480724,
      "grad_norm": 0.7052919864654541,
      "learning_rate": 5.899538380861569e-07,
      "loss": 0.6784,
      "step": 29448
    },
    {
      "epoch": 0.8485276321097217,
      "grad_norm": 0.7368022203445435,
      "learning_rate": 5.897339756098053e-07,
      "loss": 0.6751,
      "step": 29449
    },
    {
      "epoch": 0.848556445571371,
      "grad_norm": 0.7037926912307739,
      "learning_rate": 5.89514151542463e-07,
      "loss": 0.6772,
      "step": 29450
    },
    {
      "epoch": 0.8485852590330202,
      "grad_norm": 0.7395904660224915,
      "learning_rate": 5.892943658860428e-07,
      "loss": 0.6821,
      "step": 29451
    },
    {
      "epoch": 0.8486140724946695,
      "grad_norm": 0.770131528377533,
      "learning_rate": 5.89074618642459e-07,
      "loss": 0.6999,
      "step": 29452
    },
    {
      "epoch": 0.8486428859563188,
      "grad_norm": 0.733325719833374,
      "learning_rate": 5.888549098136265e-07,
      "loss": 0.6783,
      "step": 29453
    },
    {
      "epoch": 0.848671699417968,
      "grad_norm": 0.7240189909934998,
      "learning_rate": 5.886352394014589e-07,
      "loss": 0.6738,
      "step": 29454
    },
    {
      "epoch": 0.8487005128796173,
      "grad_norm": 0.7320826053619385,
      "learning_rate": 5.884156074078679e-07,
      "loss": 0.6796,
      "step": 29455
    },
    {
      "epoch": 0.8487293263412666,
      "grad_norm": 0.7250580191612244,
      "learning_rate": 5.881960138347676e-07,
      "loss": 0.6967,
      "step": 29456
    },
    {
      "epoch": 0.8487581398029159,
      "grad_norm": 0.7285422682762146,
      "learning_rate": 5.879764586840692e-07,
      "loss": 0.668,
      "step": 29457
    },
    {
      "epoch": 0.8487869532645652,
      "grad_norm": 0.7282539010047913,
      "learning_rate": 5.877569419576851e-07,
      "loss": 0.6585,
      "step": 29458
    },
    {
      "epoch": 0.8488157667262145,
      "grad_norm": 0.7353457808494568,
      "learning_rate": 5.875374636575287e-07,
      "loss": 0.6929,
      "step": 29459
    },
    {
      "epoch": 0.8488445801878638,
      "grad_norm": 0.707011342048645,
      "learning_rate": 5.873180237855086e-07,
      "loss": 0.676,
      "step": 29460
    },
    {
      "epoch": 0.8488733936495131,
      "grad_norm": 0.7119224071502686,
      "learning_rate": 5.870986223435393e-07,
      "loss": 0.6796,
      "step": 29461
    },
    {
      "epoch": 0.8489022071111624,
      "grad_norm": 0.7312584519386292,
      "learning_rate": 5.868792593335282e-07,
      "loss": 0.7013,
      "step": 29462
    },
    {
      "epoch": 0.8489310205728117,
      "grad_norm": 0.7164602875709534,
      "learning_rate": 5.866599347573881e-07,
      "loss": 0.6878,
      "step": 29463
    },
    {
      "epoch": 0.8489598340344608,
      "grad_norm": 0.7241442799568176,
      "learning_rate": 5.864406486170265e-07,
      "loss": 0.6744,
      "step": 29464
    },
    {
      "epoch": 0.8489886474961101,
      "grad_norm": 0.7132322788238525,
      "learning_rate": 5.862214009143568e-07,
      "loss": 0.6835,
      "step": 29465
    },
    {
      "epoch": 0.8490174609577594,
      "grad_norm": 0.7345489263534546,
      "learning_rate": 5.86002191651286e-07,
      "loss": 0.6813,
      "step": 29466
    },
    {
      "epoch": 0.8490462744194087,
      "grad_norm": 0.7293125987052917,
      "learning_rate": 5.857830208297238e-07,
      "loss": 0.67,
      "step": 29467
    },
    {
      "epoch": 0.849075087881058,
      "grad_norm": 0.7144190073013306,
      "learning_rate": 5.855638884515785e-07,
      "loss": 0.6757,
      "step": 29468
    },
    {
      "epoch": 0.8491039013427073,
      "grad_norm": 0.7229622602462769,
      "learning_rate": 5.853447945187601e-07,
      "loss": 0.7005,
      "step": 29469
    },
    {
      "epoch": 0.8491327148043566,
      "grad_norm": 0.7297617793083191,
      "learning_rate": 5.851257390331738e-07,
      "loss": 0.6969,
      "step": 29470
    },
    {
      "epoch": 0.8491615282660059,
      "grad_norm": 0.7384359836578369,
      "learning_rate": 5.849067219967308e-07,
      "loss": 0.6722,
      "step": 29471
    },
    {
      "epoch": 0.8491903417276552,
      "grad_norm": 0.7171801924705505,
      "learning_rate": 5.846877434113352e-07,
      "loss": 0.6653,
      "step": 29472
    },
    {
      "epoch": 0.8492191551893045,
      "grad_norm": 0.7219581604003906,
      "learning_rate": 5.84468803278897e-07,
      "loss": 0.6483,
      "step": 29473
    },
    {
      "epoch": 0.8492479686509538,
      "grad_norm": 0.7210312485694885,
      "learning_rate": 5.842499016013209e-07,
      "loss": 0.6778,
      "step": 29474
    },
    {
      "epoch": 0.849276782112603,
      "grad_norm": 0.7139390110969543,
      "learning_rate": 5.840310383805131e-07,
      "loss": 0.6788,
      "step": 29475
    },
    {
      "epoch": 0.8493055955742523,
      "grad_norm": 0.7161830067634583,
      "learning_rate": 5.838122136183816e-07,
      "loss": 0.6745,
      "step": 29476
    },
    {
      "epoch": 0.8493344090359016,
      "grad_norm": 0.7203471064567566,
      "learning_rate": 5.835934273168314e-07,
      "loss": 0.6743,
      "step": 29477
    },
    {
      "epoch": 0.8493632224975508,
      "grad_norm": 0.7398465871810913,
      "learning_rate": 5.833746794777673e-07,
      "loss": 0.7242,
      "step": 29478
    },
    {
      "epoch": 0.8493920359592001,
      "grad_norm": 0.7147142291069031,
      "learning_rate": 5.831559701030942e-07,
      "loss": 0.6571,
      "step": 29479
    },
    {
      "epoch": 0.8494208494208494,
      "grad_norm": 0.7230184674263,
      "learning_rate": 5.829372991947185e-07,
      "loss": 0.6835,
      "step": 29480
    },
    {
      "epoch": 0.8494496628824987,
      "grad_norm": 0.7355940341949463,
      "learning_rate": 5.827186667545426e-07,
      "loss": 0.666,
      "step": 29481
    },
    {
      "epoch": 0.849478476344148,
      "grad_norm": 0.7150968313217163,
      "learning_rate": 5.82500072784472e-07,
      "loss": 0.6949,
      "step": 29482
    },
    {
      "epoch": 0.8495072898057973,
      "grad_norm": 0.7221122980117798,
      "learning_rate": 5.822815172864094e-07,
      "loss": 0.6759,
      "step": 29483
    },
    {
      "epoch": 0.8495361032674466,
      "grad_norm": 0.730677604675293,
      "learning_rate": 5.820630002622595e-07,
      "loss": 0.7031,
      "step": 29484
    },
    {
      "epoch": 0.8495649167290958,
      "grad_norm": 0.7338609099388123,
      "learning_rate": 5.818445217139241e-07,
      "loss": 0.6765,
      "step": 29485
    },
    {
      "epoch": 0.8495937301907451,
      "grad_norm": 0.7156767845153809,
      "learning_rate": 5.816260816433067e-07,
      "loss": 0.6738,
      "step": 29486
    },
    {
      "epoch": 0.8496225436523944,
      "grad_norm": 0.7382395267486572,
      "learning_rate": 5.814076800523083e-07,
      "loss": 0.6757,
      "step": 29487
    },
    {
      "epoch": 0.8496513571140437,
      "grad_norm": 0.7330021858215332,
      "learning_rate": 5.811893169428323e-07,
      "loss": 0.6817,
      "step": 29488
    },
    {
      "epoch": 0.849680170575693,
      "grad_norm": 0.7271215915679932,
      "learning_rate": 5.8097099231678e-07,
      "loss": 0.6765,
      "step": 29489
    },
    {
      "epoch": 0.8497089840373423,
      "grad_norm": 0.7252494096755981,
      "learning_rate": 5.807527061760543e-07,
      "loss": 0.6771,
      "step": 29490
    },
    {
      "epoch": 0.8497377974989915,
      "grad_norm": 0.7210186719894409,
      "learning_rate": 5.805344585225536e-07,
      "loss": 0.6544,
      "step": 29491
    },
    {
      "epoch": 0.8497666109606408,
      "grad_norm": 0.7278276681900024,
      "learning_rate": 5.80316249358181e-07,
      "loss": 0.6864,
      "step": 29492
    },
    {
      "epoch": 0.8497954244222901,
      "grad_norm": 0.7165051698684692,
      "learning_rate": 5.800980786848348e-07,
      "loss": 0.6986,
      "step": 29493
    },
    {
      "epoch": 0.8498242378839393,
      "grad_norm": 0.7305518388748169,
      "learning_rate": 5.79879946504417e-07,
      "loss": 0.6642,
      "step": 29494
    },
    {
      "epoch": 0.8498530513455886,
      "grad_norm": 0.7177481651306152,
      "learning_rate": 5.796618528188253e-07,
      "loss": 0.669,
      "step": 29495
    },
    {
      "epoch": 0.8498818648072379,
      "grad_norm": 0.7390080094337463,
      "learning_rate": 5.794437976299605e-07,
      "loss": 0.6927,
      "step": 29496
    },
    {
      "epoch": 0.8499106782688872,
      "grad_norm": 0.7283539175987244,
      "learning_rate": 5.792257809397217e-07,
      "loss": 0.7022,
      "step": 29497
    },
    {
      "epoch": 0.8499394917305365,
      "grad_norm": 0.7548794150352478,
      "learning_rate": 5.790078027500068e-07,
      "loss": 0.7064,
      "step": 29498
    },
    {
      "epoch": 0.8499683051921858,
      "grad_norm": 0.7304723262786865,
      "learning_rate": 5.787898630627153e-07,
      "loss": 0.6687,
      "step": 29499
    },
    {
      "epoch": 0.8499971186538351,
      "grad_norm": 0.7149081826210022,
      "learning_rate": 5.785719618797425e-07,
      "loss": 0.6829,
      "step": 29500
    },
    {
      "epoch": 0.8500259321154844,
      "grad_norm": 0.7307959794998169,
      "learning_rate": 5.783540992029902e-07,
      "loss": 0.6714,
      "step": 29501
    },
    {
      "epoch": 0.8500547455771337,
      "grad_norm": 0.7509872913360596,
      "learning_rate": 5.781362750343522e-07,
      "loss": 0.7048,
      "step": 29502
    },
    {
      "epoch": 0.850083559038783,
      "grad_norm": 0.7280412316322327,
      "learning_rate": 5.779184893757284e-07,
      "loss": 0.6683,
      "step": 29503
    },
    {
      "epoch": 0.8501123725004323,
      "grad_norm": 0.7308732867240906,
      "learning_rate": 5.777007422290132e-07,
      "loss": 0.6859,
      "step": 29504
    },
    {
      "epoch": 0.8501411859620814,
      "grad_norm": 0.7261401414871216,
      "learning_rate": 5.774830335961046e-07,
      "loss": 0.7092,
      "step": 29505
    },
    {
      "epoch": 0.8501699994237307,
      "grad_norm": 0.735874593257904,
      "learning_rate": 5.772653634788972e-07,
      "loss": 0.6933,
      "step": 29506
    },
    {
      "epoch": 0.85019881288538,
      "grad_norm": 0.7313046455383301,
      "learning_rate": 5.770477318792877e-07,
      "loss": 0.6825,
      "step": 29507
    },
    {
      "epoch": 0.8502276263470293,
      "grad_norm": 0.7253413200378418,
      "learning_rate": 5.768301387991709e-07,
      "loss": 0.6965,
      "step": 29508
    },
    {
      "epoch": 0.8502564398086786,
      "grad_norm": 0.7265057563781738,
      "learning_rate": 5.766125842404424e-07,
      "loss": 0.6795,
      "step": 29509
    },
    {
      "epoch": 0.8502852532703279,
      "grad_norm": 0.712705135345459,
      "learning_rate": 5.763950682049963e-07,
      "loss": 0.6846,
      "step": 29510
    },
    {
      "epoch": 0.8503140667319772,
      "grad_norm": 0.7416572570800781,
      "learning_rate": 5.761775906947275e-07,
      "loss": 0.6881,
      "step": 29511
    },
    {
      "epoch": 0.8503428801936265,
      "grad_norm": 0.7077027559280396,
      "learning_rate": 5.759601517115276e-07,
      "loss": 0.6724,
      "step": 29512
    },
    {
      "epoch": 0.8503716936552758,
      "grad_norm": 0.7099609971046448,
      "learning_rate": 5.75742751257295e-07,
      "loss": 0.666,
      "step": 29513
    },
    {
      "epoch": 0.850400507116925,
      "grad_norm": 0.7224992513656616,
      "learning_rate": 5.755253893339185e-07,
      "loss": 0.6833,
      "step": 29514
    },
    {
      "epoch": 0.8504293205785743,
      "grad_norm": 0.7235820889472961,
      "learning_rate": 5.753080659432947e-07,
      "loss": 0.6636,
      "step": 29515
    },
    {
      "epoch": 0.8504581340402236,
      "grad_norm": 0.7494010329246521,
      "learning_rate": 5.750907810873136e-07,
      "loss": 0.6853,
      "step": 29516
    },
    {
      "epoch": 0.8504869475018729,
      "grad_norm": 0.7196388244628906,
      "learning_rate": 5.74873534767868e-07,
      "loss": 0.672,
      "step": 29517
    },
    {
      "epoch": 0.8505157609635221,
      "grad_norm": 0.7154604196548462,
      "learning_rate": 5.746563269868521e-07,
      "loss": 0.6939,
      "step": 29518
    },
    {
      "epoch": 0.8505445744251714,
      "grad_norm": 0.6977885365486145,
      "learning_rate": 5.744391577461544e-07,
      "loss": 0.6735,
      "step": 29519
    },
    {
      "epoch": 0.8505733878868207,
      "grad_norm": 0.7230004072189331,
      "learning_rate": 5.742220270476689e-07,
      "loss": 0.6767,
      "step": 29520
    },
    {
      "epoch": 0.85060220134847,
      "grad_norm": 0.7039861083030701,
      "learning_rate": 5.740049348932842e-07,
      "loss": 0.6619,
      "step": 29521
    },
    {
      "epoch": 0.8506310148101193,
      "grad_norm": 0.7172398567199707,
      "learning_rate": 5.73787881284893e-07,
      "loss": 0.6778,
      "step": 29522
    },
    {
      "epoch": 0.8506598282717686,
      "grad_norm": 0.7204477190971375,
      "learning_rate": 5.735708662243839e-07,
      "loss": 0.6457,
      "step": 29523
    },
    {
      "epoch": 0.8506886417334178,
      "grad_norm": 0.7069696187973022,
      "learning_rate": 5.733538897136481e-07,
      "loss": 0.6748,
      "step": 29524
    },
    {
      "epoch": 0.8507174551950671,
      "grad_norm": 0.7287244200706482,
      "learning_rate": 5.731369517545743e-07,
      "loss": 0.6973,
      "step": 29525
    },
    {
      "epoch": 0.8507462686567164,
      "grad_norm": 0.7261574268341064,
      "learning_rate": 5.729200523490541e-07,
      "loss": 0.6908,
      "step": 29526
    },
    {
      "epoch": 0.8507750821183657,
      "grad_norm": 0.7310495972633362,
      "learning_rate": 5.727031914989733e-07,
      "loss": 0.6722,
      "step": 29527
    },
    {
      "epoch": 0.850803895580015,
      "grad_norm": 0.7571013569831848,
      "learning_rate": 5.724863692062232e-07,
      "loss": 0.7062,
      "step": 29528
    },
    {
      "epoch": 0.8508327090416643,
      "grad_norm": 0.7410591244697571,
      "learning_rate": 5.722695854726895e-07,
      "loss": 0.6729,
      "step": 29529
    },
    {
      "epoch": 0.8508615225033136,
      "grad_norm": 0.725745677947998,
      "learning_rate": 5.720528403002635e-07,
      "loss": 0.6765,
      "step": 29530
    },
    {
      "epoch": 0.8508903359649629,
      "grad_norm": 0.7420704960823059,
      "learning_rate": 5.71836133690829e-07,
      "loss": 0.6803,
      "step": 29531
    },
    {
      "epoch": 0.8509191494266121,
      "grad_norm": 0.746080219745636,
      "learning_rate": 5.71619465646277e-07,
      "loss": 0.6784,
      "step": 29532
    },
    {
      "epoch": 0.8509479628882614,
      "grad_norm": 0.7086655497550964,
      "learning_rate": 5.714028361684909e-07,
      "loss": 0.6827,
      "step": 29533
    },
    {
      "epoch": 0.8509767763499106,
      "grad_norm": 0.7150079607963562,
      "learning_rate": 5.711862452593597e-07,
      "loss": 0.65,
      "step": 29534
    },
    {
      "epoch": 0.8510055898115599,
      "grad_norm": 0.7255388498306274,
      "learning_rate": 5.709696929207698e-07,
      "loss": 0.666,
      "step": 29535
    },
    {
      "epoch": 0.8510344032732092,
      "grad_norm": 0.7131117582321167,
      "learning_rate": 5.707531791546056e-07,
      "loss": 0.691,
      "step": 29536
    },
    {
      "epoch": 0.8510632167348585,
      "grad_norm": 0.7221094369888306,
      "learning_rate": 5.705367039627536e-07,
      "loss": 0.6988,
      "step": 29537
    },
    {
      "epoch": 0.8510920301965078,
      "grad_norm": 0.7074092030525208,
      "learning_rate": 5.703202673470992e-07,
      "loss": 0.6738,
      "step": 29538
    },
    {
      "epoch": 0.8511208436581571,
      "grad_norm": 0.7158239483833313,
      "learning_rate": 5.701038693095279e-07,
      "loss": 0.6682,
      "step": 29539
    },
    {
      "epoch": 0.8511496571198064,
      "grad_norm": 0.7260675430297852,
      "learning_rate": 5.69887509851923e-07,
      "loss": 0.6875,
      "step": 29540
    },
    {
      "epoch": 0.8511784705814557,
      "grad_norm": 0.7277835011482239,
      "learning_rate": 5.696711889761702e-07,
      "loss": 0.685,
      "step": 29541
    },
    {
      "epoch": 0.851207284043105,
      "grad_norm": 0.7336965799331665,
      "learning_rate": 5.694549066841515e-07,
      "loss": 0.6679,
      "step": 29542
    },
    {
      "epoch": 0.8512360975047543,
      "grad_norm": 0.7178047299385071,
      "learning_rate": 5.692386629777529e-07,
      "loss": 0.6843,
      "step": 29543
    },
    {
      "epoch": 0.8512649109664036,
      "grad_norm": 0.7338356971740723,
      "learning_rate": 5.690224578588554e-07,
      "loss": 0.7211,
      "step": 29544
    },
    {
      "epoch": 0.8512937244280527,
      "grad_norm": 0.7709207534790039,
      "learning_rate": 5.688062913293435e-07,
      "loss": 0.6887,
      "step": 29545
    },
    {
      "epoch": 0.851322537889702,
      "grad_norm": 0.7171513438224792,
      "learning_rate": 5.68590163391099e-07,
      "loss": 0.654,
      "step": 29546
    },
    {
      "epoch": 0.8513513513513513,
      "grad_norm": 0.7163903117179871,
      "learning_rate": 5.683740740460048e-07,
      "loss": 0.6705,
      "step": 29547
    },
    {
      "epoch": 0.8513801648130006,
      "grad_norm": 0.7241306304931641,
      "learning_rate": 5.681580232959405e-07,
      "loss": 0.6767,
      "step": 29548
    },
    {
      "epoch": 0.8514089782746499,
      "grad_norm": 0.7236060500144958,
      "learning_rate": 5.679420111427914e-07,
      "loss": 0.7013,
      "step": 29549
    },
    {
      "epoch": 0.8514377917362992,
      "grad_norm": 0.7081947922706604,
      "learning_rate": 5.677260375884363e-07,
      "loss": 0.6842,
      "step": 29550
    },
    {
      "epoch": 0.8514666051979485,
      "grad_norm": 0.7067747116088867,
      "learning_rate": 5.675101026347579e-07,
      "loss": 0.6681,
      "step": 29551
    },
    {
      "epoch": 0.8514954186595978,
      "grad_norm": 0.7193877696990967,
      "learning_rate": 5.67294206283634e-07,
      "loss": 0.6826,
      "step": 29552
    },
    {
      "epoch": 0.8515242321212471,
      "grad_norm": 0.7212232947349548,
      "learning_rate": 5.67078348536948e-07,
      "loss": 0.6937,
      "step": 29553
    },
    {
      "epoch": 0.8515530455828964,
      "grad_norm": 0.7389499545097351,
      "learning_rate": 5.668625293965774e-07,
      "loss": 0.6845,
      "step": 29554
    },
    {
      "epoch": 0.8515818590445456,
      "grad_norm": 0.7283833622932434,
      "learning_rate": 5.666467488644023e-07,
      "loss": 0.6729,
      "step": 29555
    },
    {
      "epoch": 0.8516106725061949,
      "grad_norm": 0.7210301160812378,
      "learning_rate": 5.664310069423034e-07,
      "loss": 0.6969,
      "step": 29556
    },
    {
      "epoch": 0.8516394859678442,
      "grad_norm": 0.7386835217475891,
      "learning_rate": 5.662153036321571e-07,
      "loss": 0.6866,
      "step": 29557
    },
    {
      "epoch": 0.8516682994294935,
      "grad_norm": 0.7284884452819824,
      "learning_rate": 5.659996389358452e-07,
      "loss": 0.6799,
      "step": 29558
    },
    {
      "epoch": 0.8516971128911427,
      "grad_norm": 0.7348463535308838,
      "learning_rate": 5.657840128552423e-07,
      "loss": 0.7007,
      "step": 29559
    },
    {
      "epoch": 0.851725926352792,
      "grad_norm": 0.7168623208999634,
      "learning_rate": 5.655684253922283e-07,
      "loss": 0.6364,
      "step": 29560
    },
    {
      "epoch": 0.8517547398144413,
      "grad_norm": 0.7169656753540039,
      "learning_rate": 5.653528765486804e-07,
      "loss": 0.6912,
      "step": 29561
    },
    {
      "epoch": 0.8517835532760906,
      "grad_norm": 0.7087724208831787,
      "learning_rate": 5.65137366326477e-07,
      "loss": 0.6593,
      "step": 29562
    },
    {
      "epoch": 0.8518123667377399,
      "grad_norm": 0.7298257946968079,
      "learning_rate": 5.649218947274932e-07,
      "loss": 0.6873,
      "step": 29563
    },
    {
      "epoch": 0.8518411801993891,
      "grad_norm": 0.7297700643539429,
      "learning_rate": 5.647064617536069e-07,
      "loss": 0.6728,
      "step": 29564
    },
    {
      "epoch": 0.8518699936610384,
      "grad_norm": 0.721831202507019,
      "learning_rate": 5.644910674066928e-07,
      "loss": 0.6834,
      "step": 29565
    },
    {
      "epoch": 0.8518988071226877,
      "grad_norm": 0.714324414730072,
      "learning_rate": 5.642757116886288e-07,
      "loss": 0.6795,
      "step": 29566
    },
    {
      "epoch": 0.851927620584337,
      "grad_norm": 0.7110297679901123,
      "learning_rate": 5.640603946012879e-07,
      "loss": 0.6665,
      "step": 29567
    },
    {
      "epoch": 0.8519564340459863,
      "grad_norm": 0.7299674153327942,
      "learning_rate": 5.638451161465475e-07,
      "loss": 0.6841,
      "step": 29568
    },
    {
      "epoch": 0.8519852475076356,
      "grad_norm": 0.7282763719558716,
      "learning_rate": 5.63629876326281e-07,
      "loss": 0.6759,
      "step": 29569
    },
    {
      "epoch": 0.8520140609692849,
      "grad_norm": 0.7163307070732117,
      "learning_rate": 5.634146751423647e-07,
      "loss": 0.6744,
      "step": 29570
    },
    {
      "epoch": 0.8520428744309342,
      "grad_norm": 0.7575705647468567,
      "learning_rate": 5.631995125966699e-07,
      "loss": 0.6943,
      "step": 29571
    },
    {
      "epoch": 0.8520716878925834,
      "grad_norm": 0.7262265682220459,
      "learning_rate": 5.62984388691073e-07,
      "loss": 0.6781,
      "step": 29572
    },
    {
      "epoch": 0.8521005013542327,
      "grad_norm": 0.7041893005371094,
      "learning_rate": 5.627693034274461e-07,
      "loss": 0.7037,
      "step": 29573
    },
    {
      "epoch": 0.8521293148158819,
      "grad_norm": 0.7605265974998474,
      "learning_rate": 5.625542568076642e-07,
      "loss": 0.6879,
      "step": 29574
    },
    {
      "epoch": 0.8521581282775312,
      "grad_norm": 0.711380124092102,
      "learning_rate": 5.623392488335977e-07,
      "loss": 0.6735,
      "step": 29575
    },
    {
      "epoch": 0.8521869417391805,
      "grad_norm": 0.7388466596603394,
      "learning_rate": 5.621242795071202e-07,
      "loss": 0.6875,
      "step": 29576
    },
    {
      "epoch": 0.8522157552008298,
      "grad_norm": 0.7176072001457214,
      "learning_rate": 5.619093488301053e-07,
      "loss": 0.6752,
      "step": 29577
    },
    {
      "epoch": 0.8522445686624791,
      "grad_norm": 0.7438915371894836,
      "learning_rate": 5.616944568044225e-07,
      "loss": 0.681,
      "step": 29578
    },
    {
      "epoch": 0.8522733821241284,
      "grad_norm": 0.7136913537979126,
      "learning_rate": 5.614796034319453e-07,
      "loss": 0.675,
      "step": 29579
    },
    {
      "epoch": 0.8523021955857777,
      "grad_norm": 0.7078502178192139,
      "learning_rate": 5.612647887145428e-07,
      "loss": 0.6773,
      "step": 29580
    },
    {
      "epoch": 0.852331009047427,
      "grad_norm": 0.7333618402481079,
      "learning_rate": 5.61050012654088e-07,
      "loss": 0.6509,
      "step": 29581
    },
    {
      "epoch": 0.8523598225090763,
      "grad_norm": 0.7306667566299438,
      "learning_rate": 5.608352752524487e-07,
      "loss": 0.6508,
      "step": 29582
    },
    {
      "epoch": 0.8523886359707256,
      "grad_norm": 0.7167965769767761,
      "learning_rate": 5.606205765114981e-07,
      "loss": 0.6695,
      "step": 29583
    },
    {
      "epoch": 0.8524174494323749,
      "grad_norm": 0.7128303647041321,
      "learning_rate": 5.604059164331027e-07,
      "loss": 0.6662,
      "step": 29584
    },
    {
      "epoch": 0.8524462628940241,
      "grad_norm": 0.7138391137123108,
      "learning_rate": 5.601912950191351e-07,
      "loss": 0.6846,
      "step": 29585
    },
    {
      "epoch": 0.8524750763556733,
      "grad_norm": 0.7244021892547607,
      "learning_rate": 5.599767122714628e-07,
      "loss": 0.6973,
      "step": 29586
    },
    {
      "epoch": 0.8525038898173226,
      "grad_norm": 0.7216370105743408,
      "learning_rate": 5.597621681919557e-07,
      "loss": 0.6559,
      "step": 29587
    },
    {
      "epoch": 0.8525327032789719,
      "grad_norm": 0.7274616956710815,
      "learning_rate": 5.595476627824809e-07,
      "loss": 0.6949,
      "step": 29588
    },
    {
      "epoch": 0.8525615167406212,
      "grad_norm": 0.703794002532959,
      "learning_rate": 5.593331960449078e-07,
      "loss": 0.6733,
      "step": 29589
    },
    {
      "epoch": 0.8525903302022705,
      "grad_norm": 0.7291758060455322,
      "learning_rate": 5.591187679811022e-07,
      "loss": 0.6773,
      "step": 29590
    },
    {
      "epoch": 0.8526191436639198,
      "grad_norm": 0.7280697822570801,
      "learning_rate": 5.589043785929343e-07,
      "loss": 0.6645,
      "step": 29591
    },
    {
      "epoch": 0.8526479571255691,
      "grad_norm": 0.7212777137756348,
      "learning_rate": 5.586900278822682e-07,
      "loss": 0.682,
      "step": 29592
    },
    {
      "epoch": 0.8526767705872184,
      "grad_norm": 0.7297698855400085,
      "learning_rate": 5.58475715850973e-07,
      "loss": 0.6828,
      "step": 29593
    },
    {
      "epoch": 0.8527055840488676,
      "grad_norm": 0.7185249924659729,
      "learning_rate": 5.582614425009153e-07,
      "loss": 0.6835,
      "step": 29594
    },
    {
      "epoch": 0.8527343975105169,
      "grad_norm": 0.7161609530448914,
      "learning_rate": 5.580472078339588e-07,
      "loss": 0.6986,
      "step": 29595
    },
    {
      "epoch": 0.8527632109721662,
      "grad_norm": 0.7238298654556274,
      "learning_rate": 5.578330118519714e-07,
      "loss": 0.7067,
      "step": 29596
    },
    {
      "epoch": 0.8527920244338155,
      "grad_norm": 0.7079625129699707,
      "learning_rate": 5.576188545568173e-07,
      "loss": 0.6707,
      "step": 29597
    },
    {
      "epoch": 0.8528208378954648,
      "grad_norm": 0.751217782497406,
      "learning_rate": 5.574047359503637e-07,
      "loss": 0.6818,
      "step": 29598
    },
    {
      "epoch": 0.852849651357114,
      "grad_norm": 0.7269324660301208,
      "learning_rate": 5.571906560344731e-07,
      "loss": 0.6761,
      "step": 29599
    },
    {
      "epoch": 0.8528784648187633,
      "grad_norm": 0.7215571403503418,
      "learning_rate": 5.569766148110112e-07,
      "loss": 0.6762,
      "step": 29600
    },
    {
      "epoch": 0.8529072782804126,
      "grad_norm": 0.7124926447868347,
      "learning_rate": 5.567626122818409e-07,
      "loss": 0.6663,
      "step": 29601
    },
    {
      "epoch": 0.8529360917420619,
      "grad_norm": 0.726347804069519,
      "learning_rate": 5.565486484488275e-07,
      "loss": 0.681,
      "step": 29602
    },
    {
      "epoch": 0.8529649052037112,
      "grad_norm": 0.7334949970245361,
      "learning_rate": 5.563347233138322e-07,
      "loss": 0.6893,
      "step": 29603
    },
    {
      "epoch": 0.8529937186653604,
      "grad_norm": 0.7287521958351135,
      "learning_rate": 5.561208368787207e-07,
      "loss": 0.6932,
      "step": 29604
    },
    {
      "epoch": 0.8530225321270097,
      "grad_norm": 0.7363523244857788,
      "learning_rate": 5.559069891453539e-07,
      "loss": 0.6802,
      "step": 29605
    },
    {
      "epoch": 0.853051345588659,
      "grad_norm": 0.7488470077514648,
      "learning_rate": 5.556931801155952e-07,
      "loss": 0.6939,
      "step": 29606
    },
    {
      "epoch": 0.8530801590503083,
      "grad_norm": 0.7445618510246277,
      "learning_rate": 5.554794097913052e-07,
      "loss": 0.6776,
      "step": 29607
    },
    {
      "epoch": 0.8531089725119576,
      "grad_norm": 0.7245949506759644,
      "learning_rate": 5.55265678174347e-07,
      "loss": 0.6731,
      "step": 29608
    },
    {
      "epoch": 0.8531377859736069,
      "grad_norm": 0.7160628437995911,
      "learning_rate": 5.550519852665815e-07,
      "loss": 0.6746,
      "step": 29609
    },
    {
      "epoch": 0.8531665994352562,
      "grad_norm": 0.7240520715713501,
      "learning_rate": 5.548383310698707e-07,
      "loss": 0.6635,
      "step": 29610
    },
    {
      "epoch": 0.8531954128969055,
      "grad_norm": 0.7240534424781799,
      "learning_rate": 5.54624715586074e-07,
      "loss": 0.6959,
      "step": 29611
    },
    {
      "epoch": 0.8532242263585548,
      "grad_norm": 0.7164846658706665,
      "learning_rate": 5.544111388170531e-07,
      "loss": 0.6776,
      "step": 29612
    },
    {
      "epoch": 0.853253039820204,
      "grad_norm": 0.7241271734237671,
      "learning_rate": 5.54197600764666e-07,
      "loss": 0.6961,
      "step": 29613
    },
    {
      "epoch": 0.8532818532818532,
      "grad_norm": 0.748986005783081,
      "learning_rate": 5.539841014307734e-07,
      "loss": 0.6709,
      "step": 29614
    },
    {
      "epoch": 0.8533106667435025,
      "grad_norm": 0.7278352379798889,
      "learning_rate": 5.537706408172361e-07,
      "loss": 0.6732,
      "step": 29615
    },
    {
      "epoch": 0.8533394802051518,
      "grad_norm": 0.733339250087738,
      "learning_rate": 5.535572189259114e-07,
      "loss": 0.6724,
      "step": 29616
    },
    {
      "epoch": 0.8533682936668011,
      "grad_norm": 0.737823486328125,
      "learning_rate": 5.53343835758659e-07,
      "loss": 0.6894,
      "step": 29617
    },
    {
      "epoch": 0.8533971071284504,
      "grad_norm": 0.730059802532196,
      "learning_rate": 5.531304913173357e-07,
      "loss": 0.6894,
      "step": 29618
    },
    {
      "epoch": 0.8534259205900997,
      "grad_norm": 0.7479413151741028,
      "learning_rate": 5.529171856038018e-07,
      "loss": 0.6903,
      "step": 29619
    },
    {
      "epoch": 0.853454734051749,
      "grad_norm": 0.7351506948471069,
      "learning_rate": 5.527039186199118e-07,
      "loss": 0.7023,
      "step": 29620
    },
    {
      "epoch": 0.8534835475133983,
      "grad_norm": 0.7270125150680542,
      "learning_rate": 5.52490690367527e-07,
      "loss": 0.6962,
      "step": 29621
    },
    {
      "epoch": 0.8535123609750476,
      "grad_norm": 0.7002527713775635,
      "learning_rate": 5.52277500848501e-07,
      "loss": 0.6848,
      "step": 29622
    },
    {
      "epoch": 0.8535411744366969,
      "grad_norm": 0.7407216429710388,
      "learning_rate": 5.520643500646932e-07,
      "loss": 0.7056,
      "step": 29623
    },
    {
      "epoch": 0.8535699878983461,
      "grad_norm": 0.7419584393501282,
      "learning_rate": 5.518512380179575e-07,
      "loss": 0.7074,
      "step": 29624
    },
    {
      "epoch": 0.8535988013599954,
      "grad_norm": 0.7119727730751038,
      "learning_rate": 5.516381647101521e-07,
      "loss": 0.6994,
      "step": 29625
    },
    {
      "epoch": 0.8536276148216446,
      "grad_norm": 0.7678387761116028,
      "learning_rate": 5.514251301431306e-07,
      "loss": 0.6601,
      "step": 29626
    },
    {
      "epoch": 0.8536564282832939,
      "grad_norm": 0.7162742018699646,
      "learning_rate": 5.512121343187504e-07,
      "loss": 0.6752,
      "step": 29627
    },
    {
      "epoch": 0.8536852417449432,
      "grad_norm": 0.7445310950279236,
      "learning_rate": 5.50999177238864e-07,
      "loss": 0.6937,
      "step": 29628
    },
    {
      "epoch": 0.8537140552065925,
      "grad_norm": 0.707103431224823,
      "learning_rate": 5.507862589053286e-07,
      "loss": 0.7022,
      "step": 29629
    },
    {
      "epoch": 0.8537428686682418,
      "grad_norm": 0.7231569886207581,
      "learning_rate": 5.505733793199963e-07,
      "loss": 0.6726,
      "step": 29630
    },
    {
      "epoch": 0.8537716821298911,
      "grad_norm": 0.7413890957832336,
      "learning_rate": 5.503605384847215e-07,
      "loss": 0.6727,
      "step": 29631
    },
    {
      "epoch": 0.8538004955915404,
      "grad_norm": 0.7103383541107178,
      "learning_rate": 5.501477364013591e-07,
      "loss": 0.6493,
      "step": 29632
    },
    {
      "epoch": 0.8538293090531897,
      "grad_norm": 0.7204602956771851,
      "learning_rate": 5.499349730717624e-07,
      "loss": 0.6672,
      "step": 29633
    },
    {
      "epoch": 0.853858122514839,
      "grad_norm": 0.7126367688179016,
      "learning_rate": 5.497222484977827e-07,
      "loss": 0.6505,
      "step": 29634
    },
    {
      "epoch": 0.8538869359764882,
      "grad_norm": 0.7362521886825562,
      "learning_rate": 5.495095626812741e-07,
      "loss": 0.6696,
      "step": 29635
    },
    {
      "epoch": 0.8539157494381375,
      "grad_norm": 0.7203823924064636,
      "learning_rate": 5.492969156240885e-07,
      "loss": 0.6759,
      "step": 29636
    },
    {
      "epoch": 0.8539445628997868,
      "grad_norm": 0.695392370223999,
      "learning_rate": 5.490843073280771e-07,
      "loss": 0.6621,
      "step": 29637
    },
    {
      "epoch": 0.8539733763614361,
      "grad_norm": 0.7007936835289001,
      "learning_rate": 5.48871737795093e-07,
      "loss": 0.6701,
      "step": 29638
    },
    {
      "epoch": 0.8540021898230854,
      "grad_norm": 0.713129460811615,
      "learning_rate": 5.486592070269858e-07,
      "loss": 0.696,
      "step": 29639
    },
    {
      "epoch": 0.8540310032847346,
      "grad_norm": 0.7313529849052429,
      "learning_rate": 5.484467150256079e-07,
      "loss": 0.6914,
      "step": 29640
    },
    {
      "epoch": 0.8540598167463839,
      "grad_norm": 0.7205318212509155,
      "learning_rate": 5.482342617928082e-07,
      "loss": 0.6803,
      "step": 29641
    },
    {
      "epoch": 0.8540886302080332,
      "grad_norm": 0.7055626511573792,
      "learning_rate": 5.480218473304388e-07,
      "loss": 0.6527,
      "step": 29642
    },
    {
      "epoch": 0.8541174436696825,
      "grad_norm": 0.713485836982727,
      "learning_rate": 5.478094716403481e-07,
      "loss": 0.6919,
      "step": 29643
    },
    {
      "epoch": 0.8541462571313317,
      "grad_norm": 0.7264829277992249,
      "learning_rate": 5.475971347243863e-07,
      "loss": 0.6859,
      "step": 29644
    },
    {
      "epoch": 0.854175070592981,
      "grad_norm": 0.7087420225143433,
      "learning_rate": 5.473848365844025e-07,
      "loss": 0.6839,
      "step": 29645
    },
    {
      "epoch": 0.8542038840546303,
      "grad_norm": 0.7199837565422058,
      "learning_rate": 5.471725772222469e-07,
      "loss": 0.6875,
      "step": 29646
    },
    {
      "epoch": 0.8542326975162796,
      "grad_norm": 0.7171457409858704,
      "learning_rate": 5.469603566397663e-07,
      "loss": 0.671,
      "step": 29647
    },
    {
      "epoch": 0.8542615109779289,
      "grad_norm": 0.7070053219795227,
      "learning_rate": 5.467481748388104e-07,
      "loss": 0.6636,
      "step": 29648
    },
    {
      "epoch": 0.8542903244395782,
      "grad_norm": 0.7275635004043579,
      "learning_rate": 5.465360318212248e-07,
      "loss": 0.6783,
      "step": 29649
    },
    {
      "epoch": 0.8543191379012275,
      "grad_norm": 0.7187708020210266,
      "learning_rate": 5.463239275888599e-07,
      "loss": 0.6674,
      "step": 29650
    },
    {
      "epoch": 0.8543479513628768,
      "grad_norm": 0.748008131980896,
      "learning_rate": 5.461118621435607e-07,
      "loss": 0.6649,
      "step": 29651
    },
    {
      "epoch": 0.8543767648245261,
      "grad_norm": 0.7485296726226807,
      "learning_rate": 5.458998354871741e-07,
      "loss": 0.7015,
      "step": 29652
    },
    {
      "epoch": 0.8544055782861752,
      "grad_norm": 0.7239049673080444,
      "learning_rate": 5.456878476215494e-07,
      "loss": 0.6833,
      "step": 29653
    },
    {
      "epoch": 0.8544343917478245,
      "grad_norm": 0.7322245240211487,
      "learning_rate": 5.454758985485298e-07,
      "loss": 0.6764,
      "step": 29654
    },
    {
      "epoch": 0.8544632052094738,
      "grad_norm": 0.7253252863883972,
      "learning_rate": 5.452639882699629e-07,
      "loss": 0.7148,
      "step": 29655
    },
    {
      "epoch": 0.8544920186711231,
      "grad_norm": 0.7044735550880432,
      "learning_rate": 5.450521167876927e-07,
      "loss": 0.6655,
      "step": 29656
    },
    {
      "epoch": 0.8545208321327724,
      "grad_norm": 0.7078360319137573,
      "learning_rate": 5.448402841035649e-07,
      "loss": 0.6546,
      "step": 29657
    },
    {
      "epoch": 0.8545496455944217,
      "grad_norm": 0.7309724688529968,
      "learning_rate": 5.446284902194249e-07,
      "loss": 0.6704,
      "step": 29658
    },
    {
      "epoch": 0.854578459056071,
      "grad_norm": 0.7128067016601562,
      "learning_rate": 5.444167351371182e-07,
      "loss": 0.6775,
      "step": 29659
    },
    {
      "epoch": 0.8546072725177203,
      "grad_norm": 0.7043278813362122,
      "learning_rate": 5.442050188584863e-07,
      "loss": 0.6672,
      "step": 29660
    },
    {
      "epoch": 0.8546360859793696,
      "grad_norm": 0.7206422686576843,
      "learning_rate": 5.439933413853759e-07,
      "loss": 0.6673,
      "step": 29661
    },
    {
      "epoch": 0.8546648994410189,
      "grad_norm": 0.7328513264656067,
      "learning_rate": 5.437817027196285e-07,
      "loss": 0.6822,
      "step": 29662
    },
    {
      "epoch": 0.8546937129026682,
      "grad_norm": 0.7533506155014038,
      "learning_rate": 5.435701028630885e-07,
      "loss": 0.6913,
      "step": 29663
    },
    {
      "epoch": 0.8547225263643174,
      "grad_norm": 0.7349432110786438,
      "learning_rate": 5.433585418175969e-07,
      "loss": 0.6791,
      "step": 29664
    },
    {
      "epoch": 0.8547513398259667,
      "grad_norm": 0.7037006616592407,
      "learning_rate": 5.431470195849986e-07,
      "loss": 0.6693,
      "step": 29665
    },
    {
      "epoch": 0.854780153287616,
      "grad_norm": 0.7491716146469116,
      "learning_rate": 5.429355361671334e-07,
      "loss": 0.6891,
      "step": 29666
    },
    {
      "epoch": 0.8548089667492652,
      "grad_norm": 0.7419015765190125,
      "learning_rate": 5.427240915658455e-07,
      "loss": 0.6878,
      "step": 29667
    },
    {
      "epoch": 0.8548377802109145,
      "grad_norm": 0.739382803440094,
      "learning_rate": 5.425126857829726e-07,
      "loss": 0.697,
      "step": 29668
    },
    {
      "epoch": 0.8548665936725638,
      "grad_norm": 0.7261124849319458,
      "learning_rate": 5.423013188203613e-07,
      "loss": 0.683,
      "step": 29669
    },
    {
      "epoch": 0.8548954071342131,
      "grad_norm": 0.7046046257019043,
      "learning_rate": 5.420899906798477e-07,
      "loss": 0.6562,
      "step": 29670
    },
    {
      "epoch": 0.8549242205958624,
      "grad_norm": 0.7363588809967041,
      "learning_rate": 5.418787013632748e-07,
      "loss": 0.7124,
      "step": 29671
    },
    {
      "epoch": 0.8549530340575117,
      "grad_norm": 0.7194614410400391,
      "learning_rate": 5.416674508724818e-07,
      "loss": 0.6892,
      "step": 29672
    },
    {
      "epoch": 0.854981847519161,
      "grad_norm": 0.7143166661262512,
      "learning_rate": 5.414562392093081e-07,
      "loss": 0.6604,
      "step": 29673
    },
    {
      "epoch": 0.8550106609808102,
      "grad_norm": 0.7153336405754089,
      "learning_rate": 5.412450663755941e-07,
      "loss": 0.6686,
      "step": 29674
    },
    {
      "epoch": 0.8550394744424595,
      "grad_norm": 0.7216978669166565,
      "learning_rate": 5.410339323731784e-07,
      "loss": 0.667,
      "step": 29675
    },
    {
      "epoch": 0.8550682879041088,
      "grad_norm": 0.7205850481987,
      "learning_rate": 5.408228372039004e-07,
      "loss": 0.6747,
      "step": 29676
    },
    {
      "epoch": 0.8550971013657581,
      "grad_norm": 0.7313755750656128,
      "learning_rate": 5.406117808695971e-07,
      "loss": 0.6887,
      "step": 29677
    },
    {
      "epoch": 0.8551259148274074,
      "grad_norm": 0.7216675281524658,
      "learning_rate": 5.404007633721081e-07,
      "loss": 0.6854,
      "step": 29678
    },
    {
      "epoch": 0.8551547282890567,
      "grad_norm": 0.717764139175415,
      "learning_rate": 5.401897847132697e-07,
      "loss": 0.6792,
      "step": 29679
    },
    {
      "epoch": 0.8551835417507059,
      "grad_norm": 0.7436772584915161,
      "learning_rate": 5.3997884489492e-07,
      "loss": 0.6957,
      "step": 29680
    },
    {
      "epoch": 0.8552123552123552,
      "grad_norm": 0.7000687718391418,
      "learning_rate": 5.397679439188964e-07,
      "loss": 0.6827,
      "step": 29681
    },
    {
      "epoch": 0.8552411686740045,
      "grad_norm": 0.7355711460113525,
      "learning_rate": 5.395570817870361e-07,
      "loss": 0.6865,
      "step": 29682
    },
    {
      "epoch": 0.8552699821356538,
      "grad_norm": 0.7063754796981812,
      "learning_rate": 5.393462585011739e-07,
      "loss": 0.6658,
      "step": 29683
    },
    {
      "epoch": 0.855298795597303,
      "grad_norm": 0.7051969170570374,
      "learning_rate": 5.391354740631477e-07,
      "loss": 0.6955,
      "step": 29684
    },
    {
      "epoch": 0.8553276090589523,
      "grad_norm": 0.7325314283370972,
      "learning_rate": 5.389247284747911e-07,
      "loss": 0.6901,
      "step": 29685
    },
    {
      "epoch": 0.8553564225206016,
      "grad_norm": 0.7240090370178223,
      "learning_rate": 5.387140217379422e-07,
      "loss": 0.6978,
      "step": 29686
    },
    {
      "epoch": 0.8553852359822509,
      "grad_norm": 0.7252761721611023,
      "learning_rate": 5.385033538544332e-07,
      "loss": 0.6916,
      "step": 29687
    },
    {
      "epoch": 0.8554140494439002,
      "grad_norm": 0.7326158881187439,
      "learning_rate": 5.382927248261017e-07,
      "loss": 0.6958,
      "step": 29688
    },
    {
      "epoch": 0.8554428629055495,
      "grad_norm": 0.7080654501914978,
      "learning_rate": 5.380821346547788e-07,
      "loss": 0.6535,
      "step": 29689
    },
    {
      "epoch": 0.8554716763671988,
      "grad_norm": 0.7427756786346436,
      "learning_rate": 5.378715833423004e-07,
      "loss": 0.7025,
      "step": 29690
    },
    {
      "epoch": 0.8555004898288481,
      "grad_norm": 0.7451484799385071,
      "learning_rate": 5.376610708905012e-07,
      "loss": 0.6742,
      "step": 29691
    },
    {
      "epoch": 0.8555293032904974,
      "grad_norm": 0.7032373547554016,
      "learning_rate": 5.374505973012118e-07,
      "loss": 0.6619,
      "step": 29692
    },
    {
      "epoch": 0.8555581167521467,
      "grad_norm": 0.7291322350502014,
      "learning_rate": 5.372401625762674e-07,
      "loss": 0.691,
      "step": 29693
    },
    {
      "epoch": 0.8555869302137958,
      "grad_norm": 0.7302272915840149,
      "learning_rate": 5.370297667175001e-07,
      "loss": 0.697,
      "step": 29694
    },
    {
      "epoch": 0.8556157436754451,
      "grad_norm": 0.7154949307441711,
      "learning_rate": 5.368194097267426e-07,
      "loss": 0.677,
      "step": 29695
    },
    {
      "epoch": 0.8556445571370944,
      "grad_norm": 0.71012943983078,
      "learning_rate": 5.366090916058258e-07,
      "loss": 0.6746,
      "step": 29696
    },
    {
      "epoch": 0.8556733705987437,
      "grad_norm": 0.7021250128746033,
      "learning_rate": 5.363988123565833e-07,
      "loss": 0.6654,
      "step": 29697
    },
    {
      "epoch": 0.855702184060393,
      "grad_norm": 0.7350288033485413,
      "learning_rate": 5.36188571980844e-07,
      "loss": 0.6789,
      "step": 29698
    },
    {
      "epoch": 0.8557309975220423,
      "grad_norm": 0.7177088856697083,
      "learning_rate": 5.359783704804411e-07,
      "loss": 0.7041,
      "step": 29699
    },
    {
      "epoch": 0.8557598109836916,
      "grad_norm": 0.7186043858528137,
      "learning_rate": 5.35768207857203e-07,
      "loss": 0.6696,
      "step": 29700
    },
    {
      "epoch": 0.8557886244453409,
      "grad_norm": 0.7196186184883118,
      "learning_rate": 5.355580841129621e-07,
      "loss": 0.6892,
      "step": 29701
    },
    {
      "epoch": 0.8558174379069902,
      "grad_norm": 0.7068287134170532,
      "learning_rate": 5.35347999249547e-07,
      "loss": 0.6646,
      "step": 29702
    },
    {
      "epoch": 0.8558462513686395,
      "grad_norm": 0.7215016484260559,
      "learning_rate": 5.351379532687883e-07,
      "loss": 0.6854,
      "step": 29703
    },
    {
      "epoch": 0.8558750648302887,
      "grad_norm": 0.7291077971458435,
      "learning_rate": 5.349279461725132e-07,
      "loss": 0.695,
      "step": 29704
    },
    {
      "epoch": 0.855903878291938,
      "grad_norm": 0.7035211324691772,
      "learning_rate": 5.347179779625539e-07,
      "loss": 0.6837,
      "step": 29705
    },
    {
      "epoch": 0.8559326917535873,
      "grad_norm": 0.6999439001083374,
      "learning_rate": 5.345080486407367e-07,
      "loss": 0.6682,
      "step": 29706
    },
    {
      "epoch": 0.8559615052152365,
      "grad_norm": 0.7448777556419373,
      "learning_rate": 5.342981582088913e-07,
      "loss": 0.6843,
      "step": 29707
    },
    {
      "epoch": 0.8559903186768858,
      "grad_norm": 0.7243582606315613,
      "learning_rate": 5.34088306668844e-07,
      "loss": 0.682,
      "step": 29708
    },
    {
      "epoch": 0.8560191321385351,
      "grad_norm": 0.7188305854797363,
      "learning_rate": 5.338784940224239e-07,
      "loss": 0.664,
      "step": 29709
    },
    {
      "epoch": 0.8560479456001844,
      "grad_norm": 0.7589099407196045,
      "learning_rate": 5.336687202714574e-07,
      "loss": 0.7036,
      "step": 29710
    },
    {
      "epoch": 0.8560767590618337,
      "grad_norm": 0.7377992868423462,
      "learning_rate": 5.334589854177708e-07,
      "loss": 0.699,
      "step": 29711
    },
    {
      "epoch": 0.856105572523483,
      "grad_norm": 0.7181414365768433,
      "learning_rate": 5.33249289463193e-07,
      "loss": 0.6753,
      "step": 29712
    },
    {
      "epoch": 0.8561343859851323,
      "grad_norm": 0.7440489530563354,
      "learning_rate": 5.330396324095477e-07,
      "loss": 0.6674,
      "step": 29713
    },
    {
      "epoch": 0.8561631994467815,
      "grad_norm": 0.7085272073745728,
      "learning_rate": 5.328300142586628e-07,
      "loss": 0.6829,
      "step": 29714
    },
    {
      "epoch": 0.8561920129084308,
      "grad_norm": 0.7260319590568542,
      "learning_rate": 5.32620435012362e-07,
      "loss": 0.6825,
      "step": 29715
    },
    {
      "epoch": 0.8562208263700801,
      "grad_norm": 0.7441990375518799,
      "learning_rate": 5.324108946724715e-07,
      "loss": 0.7162,
      "step": 29716
    },
    {
      "epoch": 0.8562496398317294,
      "grad_norm": 0.706960916519165,
      "learning_rate": 5.322013932408166e-07,
      "loss": 0.6791,
      "step": 29717
    },
    {
      "epoch": 0.8562784532933787,
      "grad_norm": 0.7004353404045105,
      "learning_rate": 5.319919307192217e-07,
      "loss": 0.677,
      "step": 29718
    },
    {
      "epoch": 0.856307266755028,
      "grad_norm": 0.7268251180648804,
      "learning_rate": 5.3178250710951e-07,
      "loss": 0.6649,
      "step": 29719
    },
    {
      "epoch": 0.8563360802166773,
      "grad_norm": 0.7384799718856812,
      "learning_rate": 5.315731224135068e-07,
      "loss": 0.6748,
      "step": 29720
    },
    {
      "epoch": 0.8563648936783265,
      "grad_norm": 0.7169135212898254,
      "learning_rate": 5.313637766330343e-07,
      "loss": 0.6667,
      "step": 29721
    },
    {
      "epoch": 0.8563937071399758,
      "grad_norm": 0.7315545678138733,
      "learning_rate": 5.311544697699173e-07,
      "loss": 0.6893,
      "step": 29722
    },
    {
      "epoch": 0.856422520601625,
      "grad_norm": 0.7896338105201721,
      "learning_rate": 5.309452018259764e-07,
      "loss": 0.6868,
      "step": 29723
    },
    {
      "epoch": 0.8564513340632743,
      "grad_norm": 0.7412247061729431,
      "learning_rate": 5.307359728030365e-07,
      "loss": 0.6944,
      "step": 29724
    },
    {
      "epoch": 0.8564801475249236,
      "grad_norm": 0.6967438459396362,
      "learning_rate": 5.305267827029176e-07,
      "loss": 0.6535,
      "step": 29725
    },
    {
      "epoch": 0.8565089609865729,
      "grad_norm": 1.2078121900558472,
      "learning_rate": 5.303176315274433e-07,
      "loss": 0.6805,
      "step": 29726
    },
    {
      "epoch": 0.8565377744482222,
      "grad_norm": 0.7049150466918945,
      "learning_rate": 5.301085192784328e-07,
      "loss": 0.6809,
      "step": 29727
    },
    {
      "epoch": 0.8565665879098715,
      "grad_norm": 0.7478529214859009,
      "learning_rate": 5.298994459577111e-07,
      "loss": 0.6864,
      "step": 29728
    },
    {
      "epoch": 0.8565954013715208,
      "grad_norm": 0.7201244831085205,
      "learning_rate": 5.296904115670954e-07,
      "loss": 0.6815,
      "step": 29729
    },
    {
      "epoch": 0.8566242148331701,
      "grad_norm": 0.7339228391647339,
      "learning_rate": 5.294814161084084e-07,
      "loss": 0.6666,
      "step": 29730
    },
    {
      "epoch": 0.8566530282948194,
      "grad_norm": 0.7323463559150696,
      "learning_rate": 5.292724595834686e-07,
      "loss": 0.6866,
      "step": 29731
    },
    {
      "epoch": 0.8566818417564687,
      "grad_norm": 0.7302373051643372,
      "learning_rate": 5.290635419940965e-07,
      "loss": 0.6681,
      "step": 29732
    },
    {
      "epoch": 0.856710655218118,
      "grad_norm": 0.7619338035583496,
      "learning_rate": 5.288546633421127e-07,
      "loss": 0.6737,
      "step": 29733
    },
    {
      "epoch": 0.8567394686797671,
      "grad_norm": 0.7248906493186951,
      "learning_rate": 5.28645823629334e-07,
      "loss": 0.6868,
      "step": 29734
    },
    {
      "epoch": 0.8567682821414164,
      "grad_norm": 0.7430935502052307,
      "learning_rate": 5.284370228575819e-07,
      "loss": 0.6695,
      "step": 29735
    },
    {
      "epoch": 0.8567970956030657,
      "grad_norm": 0.7269369959831238,
      "learning_rate": 5.282282610286721e-07,
      "loss": 0.6891,
      "step": 29736
    },
    {
      "epoch": 0.856825909064715,
      "grad_norm": 0.7262396812438965,
      "learning_rate": 5.280195381444247e-07,
      "loss": 0.695,
      "step": 29737
    },
    {
      "epoch": 0.8568547225263643,
      "grad_norm": 0.7221156358718872,
      "learning_rate": 5.278108542066563e-07,
      "loss": 0.6644,
      "step": 29738
    },
    {
      "epoch": 0.8568835359880136,
      "grad_norm": 0.7224681377410889,
      "learning_rate": 5.276022092171851e-07,
      "loss": 0.6624,
      "step": 29739
    },
    {
      "epoch": 0.8569123494496629,
      "grad_norm": 0.7245132327079773,
      "learning_rate": 5.273936031778271e-07,
      "loss": 0.6916,
      "step": 29740
    },
    {
      "epoch": 0.8569411629113122,
      "grad_norm": 0.7259500622749329,
      "learning_rate": 5.271850360904013e-07,
      "loss": 0.6737,
      "step": 29741
    },
    {
      "epoch": 0.8569699763729615,
      "grad_norm": 0.715366780757904,
      "learning_rate": 5.269765079567218e-07,
      "loss": 0.6832,
      "step": 29742
    },
    {
      "epoch": 0.8569987898346108,
      "grad_norm": 0.7306017875671387,
      "learning_rate": 5.267680187786061e-07,
      "loss": 0.6803,
      "step": 29743
    },
    {
      "epoch": 0.85702760329626,
      "grad_norm": 0.709665060043335,
      "learning_rate": 5.265595685578683e-07,
      "loss": 0.6593,
      "step": 29744
    },
    {
      "epoch": 0.8570564167579093,
      "grad_norm": 0.7142355442047119,
      "learning_rate": 5.263511572963264e-07,
      "loss": 0.6796,
      "step": 29745
    },
    {
      "epoch": 0.8570852302195586,
      "grad_norm": 0.7392949461936951,
      "learning_rate": 5.261427849957928e-07,
      "loss": 0.6787,
      "step": 29746
    },
    {
      "epoch": 0.8571140436812079,
      "grad_norm": 0.7276466488838196,
      "learning_rate": 5.259344516580839e-07,
      "loss": 0.6942,
      "step": 29747
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.7288848757743835,
      "learning_rate": 5.257261572850131e-07,
      "loss": 0.6878,
      "step": 29748
    },
    {
      "epoch": 0.8571716706045064,
      "grad_norm": 0.7221804261207581,
      "learning_rate": 5.25517901878394e-07,
      "loss": 0.6982,
      "step": 29749
    },
    {
      "epoch": 0.8572004840661557,
      "grad_norm": 0.7444857358932495,
      "learning_rate": 5.25309685440043e-07,
      "loss": 0.6981,
      "step": 29750
    },
    {
      "epoch": 0.857229297527805,
      "grad_norm": 0.757132887840271,
      "learning_rate": 5.251015079717703e-07,
      "loss": 0.6977,
      "step": 29751
    },
    {
      "epoch": 0.8572581109894543,
      "grad_norm": 0.7309980392456055,
      "learning_rate": 5.2489336947539e-07,
      "loss": 0.6683,
      "step": 29752
    },
    {
      "epoch": 0.8572869244511035,
      "grad_norm": 0.7282828092575073,
      "learning_rate": 5.24685269952716e-07,
      "loss": 0.6865,
      "step": 29753
    },
    {
      "epoch": 0.8573157379127528,
      "grad_norm": 0.7167887687683105,
      "learning_rate": 5.244772094055589e-07,
      "loss": 0.6717,
      "step": 29754
    },
    {
      "epoch": 0.8573445513744021,
      "grad_norm": 0.7156868577003479,
      "learning_rate": 5.242691878357309e-07,
      "loss": 0.6876,
      "step": 29755
    },
    {
      "epoch": 0.8573733648360514,
      "grad_norm": 0.7395557165145874,
      "learning_rate": 5.240612052450455e-07,
      "loss": 0.6894,
      "step": 29756
    },
    {
      "epoch": 0.8574021782977007,
      "grad_norm": 0.7077617049217224,
      "learning_rate": 5.23853261635312e-07,
      "loss": 0.6615,
      "step": 29757
    },
    {
      "epoch": 0.85743099175935,
      "grad_norm": 0.725356936454773,
      "learning_rate": 5.236453570083427e-07,
      "loss": 0.677,
      "step": 29758
    },
    {
      "epoch": 0.8574598052209993,
      "grad_norm": 0.7182134985923767,
      "learning_rate": 5.234374913659468e-07,
      "loss": 0.67,
      "step": 29759
    },
    {
      "epoch": 0.8574886186826486,
      "grad_norm": 0.7190275192260742,
      "learning_rate": 5.232296647099361e-07,
      "loss": 0.6607,
      "step": 29760
    },
    {
      "epoch": 0.8575174321442978,
      "grad_norm": 0.7230201959609985,
      "learning_rate": 5.230218770421192e-07,
      "loss": 0.6778,
      "step": 29761
    },
    {
      "epoch": 0.857546245605947,
      "grad_norm": 0.7021916508674622,
      "learning_rate": 5.228141283643074e-07,
      "loss": 0.678,
      "step": 29762
    },
    {
      "epoch": 0.8575750590675963,
      "grad_norm": 1.0707402229309082,
      "learning_rate": 5.226064186783069e-07,
      "loss": 0.6752,
      "step": 29763
    },
    {
      "epoch": 0.8576038725292456,
      "grad_norm": 0.7730968594551086,
      "learning_rate": 5.223987479859311e-07,
      "loss": 0.7068,
      "step": 29764
    },
    {
      "epoch": 0.8576326859908949,
      "grad_norm": 0.6987537145614624,
      "learning_rate": 5.221911162889848e-07,
      "loss": 0.6527,
      "step": 29765
    },
    {
      "epoch": 0.8576614994525442,
      "grad_norm": 0.7311108708381653,
      "learning_rate": 5.219835235892789e-07,
      "loss": 0.6815,
      "step": 29766
    },
    {
      "epoch": 0.8576903129141935,
      "grad_norm": 0.7244650721549988,
      "learning_rate": 5.217759698886188e-07,
      "loss": 0.7045,
      "step": 29767
    },
    {
      "epoch": 0.8577191263758428,
      "grad_norm": 0.7215893268585205,
      "learning_rate": 5.21568455188815e-07,
      "loss": 0.7163,
      "step": 29768
    },
    {
      "epoch": 0.8577479398374921,
      "grad_norm": 0.7176096439361572,
      "learning_rate": 5.213609794916718e-07,
      "loss": 0.6907,
      "step": 29769
    },
    {
      "epoch": 0.8577767532991414,
      "grad_norm": 0.735019326210022,
      "learning_rate": 5.211535427989972e-07,
      "loss": 0.7075,
      "step": 29770
    },
    {
      "epoch": 0.8578055667607907,
      "grad_norm": 0.7236853837966919,
      "learning_rate": 5.209461451125996e-07,
      "loss": 0.6772,
      "step": 29771
    },
    {
      "epoch": 0.85783438022244,
      "grad_norm": 0.7187258005142212,
      "learning_rate": 5.207387864342822e-07,
      "loss": 0.695,
      "step": 29772
    },
    {
      "epoch": 0.8578631936840893,
      "grad_norm": 0.7567713260650635,
      "learning_rate": 5.205314667658535e-07,
      "loss": 0.6974,
      "step": 29773
    },
    {
      "epoch": 0.8578920071457385,
      "grad_norm": 0.7120324969291687,
      "learning_rate": 5.203241861091169e-07,
      "loss": 0.6813,
      "step": 29774
    },
    {
      "epoch": 0.8579208206073877,
      "grad_norm": 0.7310377955436707,
      "learning_rate": 5.201169444658788e-07,
      "loss": 0.686,
      "step": 29775
    },
    {
      "epoch": 0.857949634069037,
      "grad_norm": 0.7025452852249146,
      "learning_rate": 5.199097418379439e-07,
      "loss": 0.672,
      "step": 29776
    },
    {
      "epoch": 0.8579784475306863,
      "grad_norm": 0.7341479659080505,
      "learning_rate": 5.197025782271176e-07,
      "loss": 0.6686,
      "step": 29777
    },
    {
      "epoch": 0.8580072609923356,
      "grad_norm": 0.7207947373390198,
      "learning_rate": 5.194954536352021e-07,
      "loss": 0.7093,
      "step": 29778
    },
    {
      "epoch": 0.8580360744539849,
      "grad_norm": 0.7189675569534302,
      "learning_rate": 5.192883680640032e-07,
      "loss": 0.6803,
      "step": 29779
    },
    {
      "epoch": 0.8580648879156342,
      "grad_norm": 0.7305344343185425,
      "learning_rate": 5.190813215153228e-07,
      "loss": 0.6665,
      "step": 29780
    },
    {
      "epoch": 0.8580937013772835,
      "grad_norm": 0.6864279508590698,
      "learning_rate": 5.188743139909652e-07,
      "loss": 0.6648,
      "step": 29781
    },
    {
      "epoch": 0.8581225148389328,
      "grad_norm": 0.7635272145271301,
      "learning_rate": 5.186673454927326e-07,
      "loss": 0.6776,
      "step": 29782
    },
    {
      "epoch": 0.858151328300582,
      "grad_norm": 0.729794442653656,
      "learning_rate": 5.184604160224282e-07,
      "loss": 0.6758,
      "step": 29783
    },
    {
      "epoch": 0.8581801417622313,
      "grad_norm": 0.7353041172027588,
      "learning_rate": 5.182535255818532e-07,
      "loss": 0.692,
      "step": 29784
    },
    {
      "epoch": 0.8582089552238806,
      "grad_norm": 0.711884617805481,
      "learning_rate": 5.180466741728108e-07,
      "loss": 0.675,
      "step": 29785
    },
    {
      "epoch": 0.8582377686855299,
      "grad_norm": 0.705569863319397,
      "learning_rate": 5.178398617971004e-07,
      "loss": 0.6753,
      "step": 29786
    },
    {
      "epoch": 0.8582665821471792,
      "grad_norm": 0.7060453295707703,
      "learning_rate": 5.17633088456524e-07,
      "loss": 0.6673,
      "step": 29787
    },
    {
      "epoch": 0.8582953956088284,
      "grad_norm": 0.724989652633667,
      "learning_rate": 5.174263541528829e-07,
      "loss": 0.6733,
      "step": 29788
    },
    {
      "epoch": 0.8583242090704777,
      "grad_norm": 0.7216662168502808,
      "learning_rate": 5.172196588879786e-07,
      "loss": 0.6927,
      "step": 29789
    },
    {
      "epoch": 0.858353022532127,
      "grad_norm": 0.7061623930931091,
      "learning_rate": 5.170130026636083e-07,
      "loss": 0.685,
      "step": 29790
    },
    {
      "epoch": 0.8583818359937763,
      "grad_norm": 0.70842045545578,
      "learning_rate": 5.168063854815752e-07,
      "loss": 0.6717,
      "step": 29791
    },
    {
      "epoch": 0.8584106494554256,
      "grad_norm": 0.7318770885467529,
      "learning_rate": 5.165998073436751e-07,
      "loss": 0.705,
      "step": 29792
    },
    {
      "epoch": 0.8584394629170748,
      "grad_norm": 0.7112048864364624,
      "learning_rate": 5.163932682517098e-07,
      "loss": 0.6596,
      "step": 29793
    },
    {
      "epoch": 0.8584682763787241,
      "grad_norm": 0.7277436852455139,
      "learning_rate": 5.161867682074772e-07,
      "loss": 0.6928,
      "step": 29794
    },
    {
      "epoch": 0.8584970898403734,
      "grad_norm": 0.7409156560897827,
      "learning_rate": 5.159803072127756e-07,
      "loss": 0.6926,
      "step": 29795
    },
    {
      "epoch": 0.8585259033020227,
      "grad_norm": 0.7422557473182678,
      "learning_rate": 5.157738852694033e-07,
      "loss": 0.6798,
      "step": 29796
    },
    {
      "epoch": 0.858554716763672,
      "grad_norm": 0.7223713994026184,
      "learning_rate": 5.155675023791573e-07,
      "loss": 0.6736,
      "step": 29797
    },
    {
      "epoch": 0.8585835302253213,
      "grad_norm": 0.7205865979194641,
      "learning_rate": 5.153611585438368e-07,
      "loss": 0.699,
      "step": 29798
    },
    {
      "epoch": 0.8586123436869706,
      "grad_norm": 0.7141391038894653,
      "learning_rate": 5.151548537652351e-07,
      "loss": 0.6855,
      "step": 29799
    },
    {
      "epoch": 0.8586411571486199,
      "grad_norm": 0.7263057231903076,
      "learning_rate": 5.149485880451538e-07,
      "loss": 0.671,
      "step": 29800
    },
    {
      "epoch": 0.8586699706102692,
      "grad_norm": 0.7123162746429443,
      "learning_rate": 5.147423613853858e-07,
      "loss": 0.6761,
      "step": 29801
    },
    {
      "epoch": 0.8586987840719184,
      "grad_norm": 0.7082406282424927,
      "learning_rate": 5.145361737877291e-07,
      "loss": 0.6664,
      "step": 29802
    },
    {
      "epoch": 0.8587275975335676,
      "grad_norm": 0.7252370119094849,
      "learning_rate": 5.143300252539779e-07,
      "loss": 0.647,
      "step": 29803
    },
    {
      "epoch": 0.8587564109952169,
      "grad_norm": 0.7075857520103455,
      "learning_rate": 5.141239157859284e-07,
      "loss": 0.6563,
      "step": 29804
    },
    {
      "epoch": 0.8587852244568662,
      "grad_norm": 0.7286818027496338,
      "learning_rate": 5.139178453853749e-07,
      "loss": 0.6597,
      "step": 29805
    },
    {
      "epoch": 0.8588140379185155,
      "grad_norm": 0.7083176970481873,
      "learning_rate": 5.137118140541137e-07,
      "loss": 0.6625,
      "step": 29806
    },
    {
      "epoch": 0.8588428513801648,
      "grad_norm": 0.719214677810669,
      "learning_rate": 5.135058217939365e-07,
      "loss": 0.6883,
      "step": 29807
    },
    {
      "epoch": 0.8588716648418141,
      "grad_norm": 0.7206954956054688,
      "learning_rate": 5.132998686066392e-07,
      "loss": 0.6708,
      "step": 29808
    },
    {
      "epoch": 0.8589004783034634,
      "grad_norm": 0.7001504302024841,
      "learning_rate": 5.130939544940155e-07,
      "loss": 0.6534,
      "step": 29809
    },
    {
      "epoch": 0.8589292917651127,
      "grad_norm": 0.7326601147651672,
      "learning_rate": 5.128880794578573e-07,
      "loss": 0.693,
      "step": 29810
    },
    {
      "epoch": 0.858958105226762,
      "grad_norm": 0.7309644222259521,
      "learning_rate": 5.12682243499959e-07,
      "loss": 0.6837,
      "step": 29811
    },
    {
      "epoch": 0.8589869186884113,
      "grad_norm": 0.7233476638793945,
      "learning_rate": 5.124764466221127e-07,
      "loss": 0.6851,
      "step": 29812
    },
    {
      "epoch": 0.8590157321500606,
      "grad_norm": 0.7142411470413208,
      "learning_rate": 5.122706888261103e-07,
      "loss": 0.6822,
      "step": 29813
    },
    {
      "epoch": 0.8590445456117098,
      "grad_norm": 0.7098224759101868,
      "learning_rate": 5.120649701137443e-07,
      "loss": 0.6671,
      "step": 29814
    },
    {
      "epoch": 0.859073359073359,
      "grad_norm": 0.7355315089225769,
      "learning_rate": 5.118592904868069e-07,
      "loss": 0.7033,
      "step": 29815
    },
    {
      "epoch": 0.8591021725350083,
      "grad_norm": 0.728496253490448,
      "learning_rate": 5.116536499470876e-07,
      "loss": 0.6669,
      "step": 29816
    },
    {
      "epoch": 0.8591309859966576,
      "grad_norm": 0.7338820695877075,
      "learning_rate": 5.114480484963791e-07,
      "loss": 0.6968,
      "step": 29817
    },
    {
      "epoch": 0.8591597994583069,
      "grad_norm": 0.7316566109657288,
      "learning_rate": 5.112424861364701e-07,
      "loss": 0.6866,
      "step": 29818
    },
    {
      "epoch": 0.8591886129199562,
      "grad_norm": 0.7226683497428894,
      "learning_rate": 5.110369628691531e-07,
      "loss": 0.6687,
      "step": 29819
    },
    {
      "epoch": 0.8592174263816055,
      "grad_norm": 0.735185980796814,
      "learning_rate": 5.10831478696216e-07,
      "loss": 0.7029,
      "step": 29820
    },
    {
      "epoch": 0.8592462398432548,
      "grad_norm": 0.7198660969734192,
      "learning_rate": 5.106260336194501e-07,
      "loss": 0.6696,
      "step": 29821
    },
    {
      "epoch": 0.8592750533049041,
      "grad_norm": 0.7251113653182983,
      "learning_rate": 5.104206276406426e-07,
      "loss": 0.6798,
      "step": 29822
    },
    {
      "epoch": 0.8593038667665533,
      "grad_norm": 0.7267308235168457,
      "learning_rate": 5.102152607615835e-07,
      "loss": 0.6907,
      "step": 29823
    },
    {
      "epoch": 0.8593326802282026,
      "grad_norm": 0.7196530699729919,
      "learning_rate": 5.100099329840618e-07,
      "loss": 0.6813,
      "step": 29824
    },
    {
      "epoch": 0.8593614936898519,
      "grad_norm": 0.7057012319564819,
      "learning_rate": 5.098046443098659e-07,
      "loss": 0.6804,
      "step": 29825
    },
    {
      "epoch": 0.8593903071515012,
      "grad_norm": 0.7529336214065552,
      "learning_rate": 5.095993947407818e-07,
      "loss": 0.6798,
      "step": 29826
    },
    {
      "epoch": 0.8594191206131505,
      "grad_norm": 0.7170563340187073,
      "learning_rate": 5.093941842785993e-07,
      "loss": 0.6742,
      "step": 29827
    },
    {
      "epoch": 0.8594479340747997,
      "grad_norm": 0.707134485244751,
      "learning_rate": 5.091890129251037e-07,
      "loss": 0.672,
      "step": 29828
    },
    {
      "epoch": 0.859476747536449,
      "grad_norm": 0.7262356281280518,
      "learning_rate": 5.089838806820835e-07,
      "loss": 0.6732,
      "step": 29829
    },
    {
      "epoch": 0.8595055609980983,
      "grad_norm": 0.7189186811447144,
      "learning_rate": 5.087787875513234e-07,
      "loss": 0.6935,
      "step": 29830
    },
    {
      "epoch": 0.8595343744597476,
      "grad_norm": 0.7206712961196899,
      "learning_rate": 5.085737335346102e-07,
      "loss": 0.6541,
      "step": 29831
    },
    {
      "epoch": 0.8595631879213969,
      "grad_norm": 0.7155765891075134,
      "learning_rate": 5.083687186337311e-07,
      "loss": 0.671,
      "step": 29832
    },
    {
      "epoch": 0.8595920013830461,
      "grad_norm": 0.7262512445449829,
      "learning_rate": 5.081637428504699e-07,
      "loss": 0.6878,
      "step": 29833
    },
    {
      "epoch": 0.8596208148446954,
      "grad_norm": 0.7027104496955872,
      "learning_rate": 5.079588061866125e-07,
      "loss": 0.6669,
      "step": 29834
    },
    {
      "epoch": 0.8596496283063447,
      "grad_norm": 0.716044545173645,
      "learning_rate": 5.07753908643942e-07,
      "loss": 0.6783,
      "step": 29835
    },
    {
      "epoch": 0.859678441767994,
      "grad_norm": 0.7323843836784363,
      "learning_rate": 5.075490502242464e-07,
      "loss": 0.6777,
      "step": 29836
    },
    {
      "epoch": 0.8597072552296433,
      "grad_norm": 0.7066076397895813,
      "learning_rate": 5.07344230929307e-07,
      "loss": 0.6774,
      "step": 29837
    },
    {
      "epoch": 0.8597360686912926,
      "grad_norm": 0.7400136590003967,
      "learning_rate": 5.07139450760909e-07,
      "loss": 0.6892,
      "step": 29838
    },
    {
      "epoch": 0.8597648821529419,
      "grad_norm": 0.7314883470535278,
      "learning_rate": 5.069347097208338e-07,
      "loss": 0.6848,
      "step": 29839
    },
    {
      "epoch": 0.8597936956145912,
      "grad_norm": 0.7161141633987427,
      "learning_rate": 5.067300078108678e-07,
      "loss": 0.6666,
      "step": 29840
    },
    {
      "epoch": 0.8598225090762405,
      "grad_norm": 0.7045028209686279,
      "learning_rate": 5.065253450327901e-07,
      "loss": 0.6776,
      "step": 29841
    },
    {
      "epoch": 0.8598513225378897,
      "grad_norm": 0.7302684783935547,
      "learning_rate": 5.063207213883858e-07,
      "loss": 0.709,
      "step": 29842
    },
    {
      "epoch": 0.8598801359995389,
      "grad_norm": 0.7257483005523682,
      "learning_rate": 5.061161368794354e-07,
      "loss": 0.7044,
      "step": 29843
    },
    {
      "epoch": 0.8599089494611882,
      "grad_norm": 0.737916111946106,
      "learning_rate": 5.059115915077217e-07,
      "loss": 0.7002,
      "step": 29844
    },
    {
      "epoch": 0.8599377629228375,
      "grad_norm": 0.6919493079185486,
      "learning_rate": 5.057070852750251e-07,
      "loss": 0.6584,
      "step": 29845
    },
    {
      "epoch": 0.8599665763844868,
      "grad_norm": 0.72389155626297,
      "learning_rate": 5.055026181831269e-07,
      "loss": 0.6983,
      "step": 29846
    },
    {
      "epoch": 0.8599953898461361,
      "grad_norm": 0.72670578956604,
      "learning_rate": 5.052981902338077e-07,
      "loss": 0.6921,
      "step": 29847
    },
    {
      "epoch": 0.8600242033077854,
      "grad_norm": 0.7336761355400085,
      "learning_rate": 5.050938014288498e-07,
      "loss": 0.7174,
      "step": 29848
    },
    {
      "epoch": 0.8600530167694347,
      "grad_norm": 0.7025620937347412,
      "learning_rate": 5.048894517700303e-07,
      "loss": 0.6864,
      "step": 29849
    },
    {
      "epoch": 0.860081830231084,
      "grad_norm": 0.7218828201293945,
      "learning_rate": 5.046851412591314e-07,
      "loss": 0.6676,
      "step": 29850
    },
    {
      "epoch": 0.8601106436927333,
      "grad_norm": 0.7253487706184387,
      "learning_rate": 5.044808698979297e-07,
      "loss": 0.6781,
      "step": 29851
    },
    {
      "epoch": 0.8601394571543826,
      "grad_norm": 0.7100808024406433,
      "learning_rate": 5.042766376882064e-07,
      "loss": 0.6804,
      "step": 29852
    },
    {
      "epoch": 0.8601682706160318,
      "grad_norm": 0.7600522041320801,
      "learning_rate": 5.040724446317402e-07,
      "loss": 0.7247,
      "step": 29853
    },
    {
      "epoch": 0.8601970840776811,
      "grad_norm": 0.7082039713859558,
      "learning_rate": 5.038682907303083e-07,
      "loss": 0.6644,
      "step": 29854
    },
    {
      "epoch": 0.8602258975393303,
      "grad_norm": 0.7298733592033386,
      "learning_rate": 5.036641759856892e-07,
      "loss": 0.6759,
      "step": 29855
    },
    {
      "epoch": 0.8602547110009796,
      "grad_norm": 0.7290384769439697,
      "learning_rate": 5.034601003996597e-07,
      "loss": 0.6814,
      "step": 29856
    },
    {
      "epoch": 0.8602835244626289,
      "grad_norm": 0.7270484566688538,
      "learning_rate": 5.032560639739991e-07,
      "loss": 0.7022,
      "step": 29857
    },
    {
      "epoch": 0.8603123379242782,
      "grad_norm": 0.7231769561767578,
      "learning_rate": 5.030520667104821e-07,
      "loss": 0.6631,
      "step": 29858
    },
    {
      "epoch": 0.8603411513859275,
      "grad_norm": 0.7381742000579834,
      "learning_rate": 5.028481086108861e-07,
      "loss": 0.6849,
      "step": 29859
    },
    {
      "epoch": 0.8603699648475768,
      "grad_norm": 0.7271268367767334,
      "learning_rate": 5.026441896769879e-07,
      "loss": 0.6789,
      "step": 29860
    },
    {
      "epoch": 0.8603987783092261,
      "grad_norm": 0.7039870023727417,
      "learning_rate": 5.024403099105634e-07,
      "loss": 0.6581,
      "step": 29861
    },
    {
      "epoch": 0.8604275917708754,
      "grad_norm": 0.7167859077453613,
      "learning_rate": 5.022364693133875e-07,
      "loss": 0.6755,
      "step": 29862
    },
    {
      "epoch": 0.8604564052325246,
      "grad_norm": 0.7343183159828186,
      "learning_rate": 5.020326678872367e-07,
      "loss": 0.693,
      "step": 29863
    },
    {
      "epoch": 0.8604852186941739,
      "grad_norm": 0.7343133687973022,
      "learning_rate": 5.018289056338843e-07,
      "loss": 0.6767,
      "step": 29864
    },
    {
      "epoch": 0.8605140321558232,
      "grad_norm": 0.7084524035453796,
      "learning_rate": 5.016251825551061e-07,
      "loss": 0.6826,
      "step": 29865
    },
    {
      "epoch": 0.8605428456174725,
      "grad_norm": 0.7132672071456909,
      "learning_rate": 5.01421498652675e-07,
      "loss": 0.6835,
      "step": 29866
    },
    {
      "epoch": 0.8605716590791218,
      "grad_norm": 0.7378749251365662,
      "learning_rate": 5.012178539283657e-07,
      "loss": 0.6978,
      "step": 29867
    },
    {
      "epoch": 0.8606004725407711,
      "grad_norm": 0.7361024022102356,
      "learning_rate": 5.010142483839531e-07,
      "loss": 0.6938,
      "step": 29868
    },
    {
      "epoch": 0.8606292860024203,
      "grad_norm": 0.7360561490058899,
      "learning_rate": 5.008106820212077e-07,
      "loss": 0.6767,
      "step": 29869
    },
    {
      "epoch": 0.8606580994640696,
      "grad_norm": 0.7043185234069824,
      "learning_rate": 5.006071548419045e-07,
      "loss": 0.6632,
      "step": 29870
    },
    {
      "epoch": 0.8606869129257189,
      "grad_norm": 0.6926538348197937,
      "learning_rate": 5.004036668478146e-07,
      "loss": 0.682,
      "step": 29871
    },
    {
      "epoch": 0.8607157263873682,
      "grad_norm": 0.7168450355529785,
      "learning_rate": 5.002002180407107e-07,
      "loss": 0.6726,
      "step": 29872
    },
    {
      "epoch": 0.8607445398490174,
      "grad_norm": 0.725523829460144,
      "learning_rate": 4.99996808422365e-07,
      "loss": 0.6911,
      "step": 29873
    },
    {
      "epoch": 0.8607733533106667,
      "grad_norm": 0.7171471118927002,
      "learning_rate": 4.99793437994549e-07,
      "loss": 0.664,
      "step": 29874
    },
    {
      "epoch": 0.860802166772316,
      "grad_norm": 0.7308340668678284,
      "learning_rate": 4.995901067590336e-07,
      "loss": 0.6859,
      "step": 29875
    },
    {
      "epoch": 0.8608309802339653,
      "grad_norm": 0.7306859493255615,
      "learning_rate": 4.993868147175901e-07,
      "loss": 0.6942,
      "step": 29876
    },
    {
      "epoch": 0.8608597936956146,
      "grad_norm": 0.6976415514945984,
      "learning_rate": 4.991835618719876e-07,
      "loss": 0.6763,
      "step": 29877
    },
    {
      "epoch": 0.8608886071572639,
      "grad_norm": 0.7087751030921936,
      "learning_rate": 4.989803482239974e-07,
      "loss": 0.683,
      "step": 29878
    },
    {
      "epoch": 0.8609174206189132,
      "grad_norm": 0.7289640307426453,
      "learning_rate": 4.987771737753888e-07,
      "loss": 0.6744,
      "step": 29879
    },
    {
      "epoch": 0.8609462340805625,
      "grad_norm": 0.7400549054145813,
      "learning_rate": 4.985740385279325e-07,
      "loss": 0.6733,
      "step": 29880
    },
    {
      "epoch": 0.8609750475422118,
      "grad_norm": 0.7401953339576721,
      "learning_rate": 4.983709424833949e-07,
      "loss": 0.6747,
      "step": 29881
    },
    {
      "epoch": 0.861003861003861,
      "grad_norm": 0.7254086136817932,
      "learning_rate": 4.981678856435479e-07,
      "loss": 0.7033,
      "step": 29882
    },
    {
      "epoch": 0.8610326744655102,
      "grad_norm": 0.7006754279136658,
      "learning_rate": 4.979648680101562e-07,
      "loss": 0.6371,
      "step": 29883
    },
    {
      "epoch": 0.8610614879271595,
      "grad_norm": 0.7099658250808716,
      "learning_rate": 4.977618895849922e-07,
      "loss": 0.6707,
      "step": 29884
    },
    {
      "epoch": 0.8610903013888088,
      "grad_norm": 0.7273566126823425,
      "learning_rate": 4.975589503698208e-07,
      "loss": 0.689,
      "step": 29885
    },
    {
      "epoch": 0.8611191148504581,
      "grad_norm": 0.7251653075218201,
      "learning_rate": 4.97356050366411e-07,
      "loss": 0.706,
      "step": 29886
    },
    {
      "epoch": 0.8611479283121074,
      "grad_norm": 0.7389711141586304,
      "learning_rate": 4.97153189576528e-07,
      "loss": 0.7018,
      "step": 29887
    },
    {
      "epoch": 0.8611767417737567,
      "grad_norm": 0.7247768640518188,
      "learning_rate": 4.969503680019405e-07,
      "loss": 0.6851,
      "step": 29888
    },
    {
      "epoch": 0.861205555235406,
      "grad_norm": 0.7524275183677673,
      "learning_rate": 4.967475856444126e-07,
      "loss": 0.7009,
      "step": 29889
    },
    {
      "epoch": 0.8612343686970553,
      "grad_norm": 1.1975181102752686,
      "learning_rate": 4.965448425057118e-07,
      "loss": 0.6659,
      "step": 29890
    },
    {
      "epoch": 0.8612631821587046,
      "grad_norm": 0.7285062670707703,
      "learning_rate": 4.96342138587605e-07,
      "loss": 0.6742,
      "step": 29891
    },
    {
      "epoch": 0.8612919956203539,
      "grad_norm": 0.7128275632858276,
      "learning_rate": 4.961394738918546e-07,
      "loss": 0.708,
      "step": 29892
    },
    {
      "epoch": 0.8613208090820031,
      "grad_norm": 0.7054543495178223,
      "learning_rate": 4.959368484202287e-07,
      "loss": 0.6727,
      "step": 29893
    },
    {
      "epoch": 0.8613496225436524,
      "grad_norm": 0.7170769572257996,
      "learning_rate": 4.957342621744887e-07,
      "loss": 0.6602,
      "step": 29894
    },
    {
      "epoch": 0.8613784360053017,
      "grad_norm": 0.7258298397064209,
      "learning_rate": 4.955317151564015e-07,
      "loss": 0.6972,
      "step": 29895
    },
    {
      "epoch": 0.8614072494669509,
      "grad_norm": 0.7419655919075012,
      "learning_rate": 4.953292073677296e-07,
      "loss": 0.7218,
      "step": 29896
    },
    {
      "epoch": 0.8614360629286002,
      "grad_norm": 0.7253895998001099,
      "learning_rate": 4.951267388102387e-07,
      "loss": 0.6751,
      "step": 29897
    },
    {
      "epoch": 0.8614648763902495,
      "grad_norm": 0.7292675971984863,
      "learning_rate": 4.949243094856892e-07,
      "loss": 0.7135,
      "step": 29898
    },
    {
      "epoch": 0.8614936898518988,
      "grad_norm": 0.7031120657920837,
      "learning_rate": 4.947219193958469e-07,
      "loss": 0.6923,
      "step": 29899
    },
    {
      "epoch": 0.8615225033135481,
      "grad_norm": 0.7247274518013,
      "learning_rate": 4.945195685424725e-07,
      "loss": 0.6803,
      "step": 29900
    },
    {
      "epoch": 0.8615513167751974,
      "grad_norm": 0.7132297158241272,
      "learning_rate": 4.943172569273291e-07,
      "loss": 0.6716,
      "step": 29901
    },
    {
      "epoch": 0.8615801302368467,
      "grad_norm": 0.7114353775978088,
      "learning_rate": 4.941149845521776e-07,
      "loss": 0.6803,
      "step": 29902
    },
    {
      "epoch": 0.8616089436984959,
      "grad_norm": 0.7140387892723083,
      "learning_rate": 4.939127514187814e-07,
      "loss": 0.6815,
      "step": 29903
    },
    {
      "epoch": 0.8616377571601452,
      "grad_norm": 0.7164041996002197,
      "learning_rate": 4.937105575288998e-07,
      "loss": 0.6712,
      "step": 29904
    },
    {
      "epoch": 0.8616665706217945,
      "grad_norm": 0.7424299120903015,
      "learning_rate": 4.93508402884294e-07,
      "loss": 0.6857,
      "step": 29905
    },
    {
      "epoch": 0.8616953840834438,
      "grad_norm": 0.7149697542190552,
      "learning_rate": 4.933062874867267e-07,
      "loss": 0.6757,
      "step": 29906
    },
    {
      "epoch": 0.8617241975450931,
      "grad_norm": 0.7267137765884399,
      "learning_rate": 4.931042113379558e-07,
      "loss": 0.6757,
      "step": 29907
    },
    {
      "epoch": 0.8617530110067424,
      "grad_norm": 0.7217210531234741,
      "learning_rate": 4.929021744397416e-07,
      "loss": 0.703,
      "step": 29908
    },
    {
      "epoch": 0.8617818244683916,
      "grad_norm": 0.7038978934288025,
      "learning_rate": 4.92700176793845e-07,
      "loss": 0.6981,
      "step": 29909
    },
    {
      "epoch": 0.8618106379300409,
      "grad_norm": 0.7532799243927002,
      "learning_rate": 4.924982184020233e-07,
      "loss": 0.687,
      "step": 29910
    },
    {
      "epoch": 0.8618394513916902,
      "grad_norm": 0.7164145708084106,
      "learning_rate": 4.922962992660363e-07,
      "loss": 0.6915,
      "step": 29911
    },
    {
      "epoch": 0.8618682648533394,
      "grad_norm": 0.7442967295646667,
      "learning_rate": 4.920944193876431e-07,
      "loss": 0.6733,
      "step": 29912
    },
    {
      "epoch": 0.8618970783149887,
      "grad_norm": 0.7327516078948975,
      "learning_rate": 4.918925787686007e-07,
      "loss": 0.6849,
      "step": 29913
    },
    {
      "epoch": 0.861925891776638,
      "grad_norm": 0.715164840221405,
      "learning_rate": 4.916907774106682e-07,
      "loss": 0.6856,
      "step": 29914
    },
    {
      "epoch": 0.8619547052382873,
      "grad_norm": 0.7546942830085754,
      "learning_rate": 4.914890153156021e-07,
      "loss": 0.6864,
      "step": 29915
    },
    {
      "epoch": 0.8619835186999366,
      "grad_norm": 0.7335284948348999,
      "learning_rate": 4.912872924851597e-07,
      "loss": 0.7068,
      "step": 29916
    },
    {
      "epoch": 0.8620123321615859,
      "grad_norm": 0.7117348313331604,
      "learning_rate": 4.91085608921098e-07,
      "loss": 0.6707,
      "step": 29917
    },
    {
      "epoch": 0.8620411456232352,
      "grad_norm": 0.7077138423919678,
      "learning_rate": 4.908839646251734e-07,
      "loss": 0.6639,
      "step": 29918
    },
    {
      "epoch": 0.8620699590848845,
      "grad_norm": 0.704423189163208,
      "learning_rate": 4.906823595991406e-07,
      "loss": 0.6565,
      "step": 29919
    },
    {
      "epoch": 0.8620987725465338,
      "grad_norm": 0.7190849781036377,
      "learning_rate": 4.904807938447587e-07,
      "loss": 0.6688,
      "step": 29920
    },
    {
      "epoch": 0.8621275860081831,
      "grad_norm": 0.7315501570701599,
      "learning_rate": 4.902792673637802e-07,
      "loss": 0.6943,
      "step": 29921
    },
    {
      "epoch": 0.8621563994698324,
      "grad_norm": 0.753614068031311,
      "learning_rate": 4.900777801579621e-07,
      "loss": 0.6741,
      "step": 29922
    },
    {
      "epoch": 0.8621852129314815,
      "grad_norm": 0.7090903520584106,
      "learning_rate": 4.898763322290573e-07,
      "loss": 0.6706,
      "step": 29923
    },
    {
      "epoch": 0.8622140263931308,
      "grad_norm": 0.7241352796554565,
      "learning_rate": 4.896749235788223e-07,
      "loss": 0.6962,
      "step": 29924
    },
    {
      "epoch": 0.8622428398547801,
      "grad_norm": 0.7320653200149536,
      "learning_rate": 4.89473554209009e-07,
      "loss": 0.69,
      "step": 29925
    },
    {
      "epoch": 0.8622716533164294,
      "grad_norm": 0.7459665536880493,
      "learning_rate": 4.892722241213732e-07,
      "loss": 0.6942,
      "step": 29926
    },
    {
      "epoch": 0.8623004667780787,
      "grad_norm": 0.7427588701248169,
      "learning_rate": 4.890709333176658e-07,
      "loss": 0.7018,
      "step": 29927
    },
    {
      "epoch": 0.862329280239728,
      "grad_norm": 0.7360875010490417,
      "learning_rate": 4.888696817996419e-07,
      "loss": 0.6959,
      "step": 29928
    },
    {
      "epoch": 0.8623580937013773,
      "grad_norm": 0.749503493309021,
      "learning_rate": 4.886684695690542e-07,
      "loss": 0.6753,
      "step": 29929
    },
    {
      "epoch": 0.8623869071630266,
      "grad_norm": 0.7455237507820129,
      "learning_rate": 4.884672966276538e-07,
      "loss": 0.6706,
      "step": 29930
    },
    {
      "epoch": 0.8624157206246759,
      "grad_norm": 0.7020900249481201,
      "learning_rate": 4.882661629771929e-07,
      "loss": 0.6674,
      "step": 29931
    },
    {
      "epoch": 0.8624445340863252,
      "grad_norm": 0.7398945093154907,
      "learning_rate": 4.880650686194238e-07,
      "loss": 0.6814,
      "step": 29932
    },
    {
      "epoch": 0.8624733475479744,
      "grad_norm": 0.7256187796592712,
      "learning_rate": 4.878640135560992e-07,
      "loss": 0.6931,
      "step": 29933
    },
    {
      "epoch": 0.8625021610096237,
      "grad_norm": 0.7492402791976929,
      "learning_rate": 4.876629977889669e-07,
      "loss": 0.6839,
      "step": 29934
    },
    {
      "epoch": 0.862530974471273,
      "grad_norm": 0.7405080795288086,
      "learning_rate": 4.874620213197806e-07,
      "loss": 0.7084,
      "step": 29935
    },
    {
      "epoch": 0.8625597879329222,
      "grad_norm": 0.7272213101387024,
      "learning_rate": 4.872610841502878e-07,
      "loss": 0.6799,
      "step": 29936
    },
    {
      "epoch": 0.8625886013945715,
      "grad_norm": 0.7054840326309204,
      "learning_rate": 4.870601862822411e-07,
      "loss": 0.6852,
      "step": 29937
    },
    {
      "epoch": 0.8626174148562208,
      "grad_norm": 0.7120703458786011,
      "learning_rate": 4.868593277173878e-07,
      "loss": 0.7058,
      "step": 29938
    },
    {
      "epoch": 0.8626462283178701,
      "grad_norm": 0.7326442003250122,
      "learning_rate": 4.866585084574793e-07,
      "loss": 0.692,
      "step": 29939
    },
    {
      "epoch": 0.8626750417795194,
      "grad_norm": 0.7154077291488647,
      "learning_rate": 4.864577285042626e-07,
      "loss": 0.6741,
      "step": 29940
    },
    {
      "epoch": 0.8627038552411687,
      "grad_norm": 0.7299014329910278,
      "learning_rate": 4.862569878594881e-07,
      "loss": 0.7108,
      "step": 29941
    },
    {
      "epoch": 0.862732668702818,
      "grad_norm": 0.7426057457923889,
      "learning_rate": 4.860562865249024e-07,
      "loss": 0.692,
      "step": 29942
    },
    {
      "epoch": 0.8627614821644672,
      "grad_norm": 0.7418368458747864,
      "learning_rate": 4.858556245022538e-07,
      "loss": 0.687,
      "step": 29943
    },
    {
      "epoch": 0.8627902956261165,
      "grad_norm": 0.7234686017036438,
      "learning_rate": 4.856550017932904e-07,
      "loss": 0.7083,
      "step": 29944
    },
    {
      "epoch": 0.8628191090877658,
      "grad_norm": 0.7244777083396912,
      "learning_rate": 4.8545441839976e-07,
      "loss": 0.6911,
      "step": 29945
    },
    {
      "epoch": 0.8628479225494151,
      "grad_norm": 0.7068914175033569,
      "learning_rate": 4.852538743234081e-07,
      "loss": 0.6719,
      "step": 29946
    },
    {
      "epoch": 0.8628767360110644,
      "grad_norm": 0.7222400307655334,
      "learning_rate": 4.850533695659826e-07,
      "loss": 0.685,
      "step": 29947
    },
    {
      "epoch": 0.8629055494727137,
      "grad_norm": 0.7123767733573914,
      "learning_rate": 4.848529041292283e-07,
      "loss": 0.6786,
      "step": 29948
    },
    {
      "epoch": 0.862934362934363,
      "grad_norm": 0.7207916378974915,
      "learning_rate": 4.846524780148915e-07,
      "loss": 0.6599,
      "step": 29949
    },
    {
      "epoch": 0.8629631763960122,
      "grad_norm": 0.732113242149353,
      "learning_rate": 4.844520912247186e-07,
      "loss": 0.6991,
      "step": 29950
    },
    {
      "epoch": 0.8629919898576615,
      "grad_norm": 0.7299013733863831,
      "learning_rate": 4.842517437604533e-07,
      "loss": 0.6827,
      "step": 29951
    },
    {
      "epoch": 0.8630208033193107,
      "grad_norm": 0.7128268480300903,
      "learning_rate": 4.840514356238424e-07,
      "loss": 0.6866,
      "step": 29952
    },
    {
      "epoch": 0.86304961678096,
      "grad_norm": 0.7225103974342346,
      "learning_rate": 4.838511668166279e-07,
      "loss": 0.6849,
      "step": 29953
    },
    {
      "epoch": 0.8630784302426093,
      "grad_norm": 0.7227070331573486,
      "learning_rate": 4.836509373405568e-07,
      "loss": 0.6694,
      "step": 29954
    },
    {
      "epoch": 0.8631072437042586,
      "grad_norm": 0.7053883671760559,
      "learning_rate": 4.834507471973687e-07,
      "loss": 0.6747,
      "step": 29955
    },
    {
      "epoch": 0.8631360571659079,
      "grad_norm": 0.723751962184906,
      "learning_rate": 4.832505963888123e-07,
      "loss": 0.6866,
      "step": 29956
    },
    {
      "epoch": 0.8631648706275572,
      "grad_norm": 0.7008308172225952,
      "learning_rate": 4.830504849166268e-07,
      "loss": 0.6801,
      "step": 29957
    },
    {
      "epoch": 0.8631936840892065,
      "grad_norm": 0.7608495354652405,
      "learning_rate": 4.828504127825568e-07,
      "loss": 0.7153,
      "step": 29958
    },
    {
      "epoch": 0.8632224975508558,
      "grad_norm": 0.7112349271774292,
      "learning_rate": 4.826503799883436e-07,
      "loss": 0.671,
      "step": 29959
    },
    {
      "epoch": 0.8632513110125051,
      "grad_norm": 0.7205828428268433,
      "learning_rate": 4.82450386535731e-07,
      "loss": 0.6625,
      "step": 29960
    },
    {
      "epoch": 0.8632801244741544,
      "grad_norm": 0.726735532283783,
      "learning_rate": 4.822504324264587e-07,
      "loss": 0.6629,
      "step": 29961
    },
    {
      "epoch": 0.8633089379358037,
      "grad_norm": 0.730644941329956,
      "learning_rate": 4.820505176622697e-07,
      "loss": 0.668,
      "step": 29962
    },
    {
      "epoch": 0.8633377513974528,
      "grad_norm": 0.7287558317184448,
      "learning_rate": 4.818506422449038e-07,
      "loss": 0.6737,
      "step": 29963
    },
    {
      "epoch": 0.8633665648591021,
      "grad_norm": 0.7147678732872009,
      "learning_rate": 4.816508061761033e-07,
      "loss": 0.6867,
      "step": 29964
    },
    {
      "epoch": 0.8633953783207514,
      "grad_norm": 0.7133921384811401,
      "learning_rate": 4.814510094576064e-07,
      "loss": 0.6445,
      "step": 29965
    },
    {
      "epoch": 0.8634241917824007,
      "grad_norm": 0.7629079222679138,
      "learning_rate": 4.812512520911544e-07,
      "loss": 0.6967,
      "step": 29966
    },
    {
      "epoch": 0.86345300524405,
      "grad_norm": 0.7119367122650146,
      "learning_rate": 4.810515340784866e-07,
      "loss": 0.6843,
      "step": 29967
    },
    {
      "epoch": 0.8634818187056993,
      "grad_norm": 0.7114050984382629,
      "learning_rate": 4.808518554213437e-07,
      "loss": 0.6788,
      "step": 29968
    },
    {
      "epoch": 0.8635106321673486,
      "grad_norm": 0.7192052006721497,
      "learning_rate": 4.806522161214627e-07,
      "loss": 0.6749,
      "step": 29969
    },
    {
      "epoch": 0.8635394456289979,
      "grad_norm": 0.7162525057792664,
      "learning_rate": 4.804526161805834e-07,
      "loss": 0.696,
      "step": 29970
    },
    {
      "epoch": 0.8635682590906472,
      "grad_norm": 0.7343030571937561,
      "learning_rate": 4.802530556004448e-07,
      "loss": 0.6786,
      "step": 29971
    },
    {
      "epoch": 0.8635970725522965,
      "grad_norm": 0.7389553785324097,
      "learning_rate": 4.800535343827834e-07,
      "loss": 0.6857,
      "step": 29972
    },
    {
      "epoch": 0.8636258860139457,
      "grad_norm": 0.7137295603752136,
      "learning_rate": 4.798540525293377e-07,
      "loss": 0.6662,
      "step": 29973
    },
    {
      "epoch": 0.863654699475595,
      "grad_norm": 0.7178401350975037,
      "learning_rate": 4.796546100418442e-07,
      "loss": 0.6755,
      "step": 29974
    },
    {
      "epoch": 0.8636835129372443,
      "grad_norm": 0.7271896600723267,
      "learning_rate": 4.79455206922041e-07,
      "loss": 0.7071,
      "step": 29975
    },
    {
      "epoch": 0.8637123263988936,
      "grad_norm": 0.7529674172401428,
      "learning_rate": 4.792558431716637e-07,
      "loss": 0.6855,
      "step": 29976
    },
    {
      "epoch": 0.8637411398605428,
      "grad_norm": 0.7351589202880859,
      "learning_rate": 4.79056518792449e-07,
      "loss": 0.6833,
      "step": 29977
    },
    {
      "epoch": 0.8637699533221921,
      "grad_norm": 0.7421127557754517,
      "learning_rate": 4.788572337861313e-07,
      "loss": 0.6826,
      "step": 29978
    },
    {
      "epoch": 0.8637987667838414,
      "grad_norm": 0.7325925230979919,
      "learning_rate": 4.786579881544495e-07,
      "loss": 0.6746,
      "step": 29979
    },
    {
      "epoch": 0.8638275802454907,
      "grad_norm": 0.7189649939537048,
      "learning_rate": 4.78458781899136e-07,
      "loss": 0.6915,
      "step": 29980
    },
    {
      "epoch": 0.86385639370714,
      "grad_norm": 0.7123776078224182,
      "learning_rate": 4.782596150219276e-07,
      "loss": 0.6658,
      "step": 29981
    },
    {
      "epoch": 0.8638852071687892,
      "grad_norm": 0.7304131984710693,
      "learning_rate": 4.780604875245571e-07,
      "loss": 0.6849,
      "step": 29982
    },
    {
      "epoch": 0.8639140206304385,
      "grad_norm": 0.7258749008178711,
      "learning_rate": 4.7786139940876e-07,
      "loss": 0.6623,
      "step": 29983
    },
    {
      "epoch": 0.8639428340920878,
      "grad_norm": 0.725330650806427,
      "learning_rate": 4.77662350676269e-07,
      "loss": 0.6712,
      "step": 29984
    },
    {
      "epoch": 0.8639716475537371,
      "grad_norm": 0.724759042263031,
      "learning_rate": 4.774633413288188e-07,
      "loss": 0.6687,
      "step": 29985
    },
    {
      "epoch": 0.8640004610153864,
      "grad_norm": 0.7684453725814819,
      "learning_rate": 4.772643713681413e-07,
      "loss": 0.6872,
      "step": 29986
    },
    {
      "epoch": 0.8640292744770357,
      "grad_norm": 0.7321337461471558,
      "learning_rate": 4.770654407959701e-07,
      "loss": 0.7061,
      "step": 29987
    },
    {
      "epoch": 0.864058087938685,
      "grad_norm": 0.7314637303352356,
      "learning_rate": 4.768665496140385e-07,
      "loss": 0.6922,
      "step": 29988
    },
    {
      "epoch": 0.8640869014003343,
      "grad_norm": 0.6931214332580566,
      "learning_rate": 4.76667697824077e-07,
      "loss": 0.6615,
      "step": 29989
    },
    {
      "epoch": 0.8641157148619835,
      "grad_norm": 0.7232334613800049,
      "learning_rate": 4.764688854278182e-07,
      "loss": 0.6727,
      "step": 29990
    },
    {
      "epoch": 0.8641445283236328,
      "grad_norm": 0.7123249769210815,
      "learning_rate": 4.7627011242699307e-07,
      "loss": 0.6688,
      "step": 29991
    },
    {
      "epoch": 0.864173341785282,
      "grad_norm": 0.734370231628418,
      "learning_rate": 4.760713788233345e-07,
      "loss": 0.6825,
      "step": 29992
    },
    {
      "epoch": 0.8642021552469313,
      "grad_norm": 0.7425662875175476,
      "learning_rate": 4.7587268461857125e-07,
      "loss": 0.6719,
      "step": 29993
    },
    {
      "epoch": 0.8642309687085806,
      "grad_norm": 0.732293963432312,
      "learning_rate": 4.7567402981443465e-07,
      "loss": 0.6997,
      "step": 29994
    },
    {
      "epoch": 0.8642597821702299,
      "grad_norm": 0.7264804840087891,
      "learning_rate": 4.7547541441265445e-07,
      "loss": 0.6721,
      "step": 29995
    },
    {
      "epoch": 0.8642885956318792,
      "grad_norm": 0.7459352016448975,
      "learning_rate": 4.752768384149609e-07,
      "loss": 0.689,
      "step": 29996
    },
    {
      "epoch": 0.8643174090935285,
      "grad_norm": 0.7274838089942932,
      "learning_rate": 4.750783018230826e-07,
      "loss": 0.6728,
      "step": 29997
    },
    {
      "epoch": 0.8643462225551778,
      "grad_norm": 0.7193573713302612,
      "learning_rate": 4.748798046387493e-07,
      "loss": 0.679,
      "step": 29998
    },
    {
      "epoch": 0.8643750360168271,
      "grad_norm": 0.7234454154968262,
      "learning_rate": 4.7468134686368903e-07,
      "loss": 0.688,
      "step": 29999
    },
    {
      "epoch": 0.8644038494784764,
      "grad_norm": 0.7200474739074707,
      "learning_rate": 4.7448292849963104e-07,
      "loss": 0.667,
      "step": 30000
    },
    {
      "epoch": 0.8644326629401257,
      "grad_norm": 0.7153880596160889,
      "learning_rate": 4.742845495483023e-07,
      "loss": 0.6707,
      "step": 30001
    },
    {
      "epoch": 0.864461476401775,
      "grad_norm": 0.7182427048683167,
      "learning_rate": 4.7408621001143076e-07,
      "loss": 0.6895,
      "step": 30002
    },
    {
      "epoch": 0.8644902898634242,
      "grad_norm": 0.7184101939201355,
      "learning_rate": 4.738879098907445e-07,
      "loss": 0.6765,
      "step": 30003
    },
    {
      "epoch": 0.8645191033250734,
      "grad_norm": 0.7131296992301941,
      "learning_rate": 4.7368964918797053e-07,
      "loss": 0.6731,
      "step": 30004
    },
    {
      "epoch": 0.8645479167867227,
      "grad_norm": 0.7159551382064819,
      "learning_rate": 4.734914279048347e-07,
      "loss": 0.6859,
      "step": 30005
    },
    {
      "epoch": 0.864576730248372,
      "grad_norm": 0.7274968028068542,
      "learning_rate": 4.7329324604306447e-07,
      "loss": 0.6848,
      "step": 30006
    },
    {
      "epoch": 0.8646055437100213,
      "grad_norm": 0.7056819796562195,
      "learning_rate": 4.7309510360438346e-07,
      "loss": 0.6647,
      "step": 30007
    },
    {
      "epoch": 0.8646343571716706,
      "grad_norm": 0.7324854135513306,
      "learning_rate": 4.728970005905198e-07,
      "loss": 0.6898,
      "step": 30008
    },
    {
      "epoch": 0.8646631706333199,
      "grad_norm": 0.7390820384025574,
      "learning_rate": 4.7269893700319813e-07,
      "loss": 0.7001,
      "step": 30009
    },
    {
      "epoch": 0.8646919840949692,
      "grad_norm": 0.728001058101654,
      "learning_rate": 4.7250091284414213e-07,
      "loss": 0.6681,
      "step": 30010
    },
    {
      "epoch": 0.8647207975566185,
      "grad_norm": 0.7231671214103699,
      "learning_rate": 4.7230292811507874e-07,
      "loss": 0.6685,
      "step": 30011
    },
    {
      "epoch": 0.8647496110182677,
      "grad_norm": 0.741252064704895,
      "learning_rate": 4.7210498281772933e-07,
      "loss": 0.7229,
      "step": 30012
    },
    {
      "epoch": 0.864778424479917,
      "grad_norm": 0.7287086248397827,
      "learning_rate": 4.719070769538203e-07,
      "loss": 0.6813,
      "step": 30013
    },
    {
      "epoch": 0.8648072379415663,
      "grad_norm": 0.7115008234977722,
      "learning_rate": 4.717092105250726e-07,
      "loss": 0.683,
      "step": 30014
    },
    {
      "epoch": 0.8648360514032156,
      "grad_norm": 0.7228734493255615,
      "learning_rate": 4.71511383533213e-07,
      "loss": 0.689,
      "step": 30015
    },
    {
      "epoch": 0.8648648648648649,
      "grad_norm": 0.7280217409133911,
      "learning_rate": 4.7131359597996087e-07,
      "loss": 0.6853,
      "step": 30016
    },
    {
      "epoch": 0.8648936783265141,
      "grad_norm": 0.7199865579605103,
      "learning_rate": 4.711158478670419e-07,
      "loss": 0.6864,
      "step": 30017
    },
    {
      "epoch": 0.8649224917881634,
      "grad_norm": 0.7473544478416443,
      "learning_rate": 4.7091813919617533e-07,
      "loss": 0.6731,
      "step": 30018
    },
    {
      "epoch": 0.8649513052498127,
      "grad_norm": 0.7223327159881592,
      "learning_rate": 4.707204699690854e-07,
      "loss": 0.6827,
      "step": 30019
    },
    {
      "epoch": 0.864980118711462,
      "grad_norm": 0.7224363684654236,
      "learning_rate": 4.7052284018749173e-07,
      "loss": 0.6827,
      "step": 30020
    },
    {
      "epoch": 0.8650089321731113,
      "grad_norm": 0.7134141325950623,
      "learning_rate": 4.703252498531169e-07,
      "loss": 0.6864,
      "step": 30021
    },
    {
      "epoch": 0.8650377456347605,
      "grad_norm": 0.7352195978164673,
      "learning_rate": 4.701276989676806e-07,
      "loss": 0.6811,
      "step": 30022
    },
    {
      "epoch": 0.8650665590964098,
      "grad_norm": 0.7259511351585388,
      "learning_rate": 4.6993018753290434e-07,
      "loss": 0.6719,
      "step": 30023
    },
    {
      "epoch": 0.8650953725580591,
      "grad_norm": 0.7256437540054321,
      "learning_rate": 4.6973271555050716e-07,
      "loss": 0.6982,
      "step": 30024
    },
    {
      "epoch": 0.8651241860197084,
      "grad_norm": 0.7080982327461243,
      "learning_rate": 4.695352830222094e-07,
      "loss": 0.6798,
      "step": 30025
    },
    {
      "epoch": 0.8651529994813577,
      "grad_norm": 0.7173497080802917,
      "learning_rate": 4.693378899497303e-07,
      "loss": 0.6736,
      "step": 30026
    },
    {
      "epoch": 0.865181812943007,
      "grad_norm": 0.7249125838279724,
      "learning_rate": 4.6914053633479005e-07,
      "loss": 0.7067,
      "step": 30027
    },
    {
      "epoch": 0.8652106264046563,
      "grad_norm": 0.7240942716598511,
      "learning_rate": 4.689432221791063e-07,
      "loss": 0.6907,
      "step": 30028
    },
    {
      "epoch": 0.8652394398663056,
      "grad_norm": 0.7224746346473694,
      "learning_rate": 4.6874594748439694e-07,
      "loss": 0.6817,
      "step": 30029
    },
    {
      "epoch": 0.8652682533279549,
      "grad_norm": 0.7156652808189392,
      "learning_rate": 4.6854871225238187e-07,
      "loss": 0.692,
      "step": 30030
    },
    {
      "epoch": 0.865297066789604,
      "grad_norm": 0.7274476289749146,
      "learning_rate": 4.6835151648477684e-07,
      "loss": 0.6994,
      "step": 30031
    },
    {
      "epoch": 0.8653258802512533,
      "grad_norm": 0.7396882176399231,
      "learning_rate": 4.6815436018330106e-07,
      "loss": 0.7147,
      "step": 30032
    },
    {
      "epoch": 0.8653546937129026,
      "grad_norm": 0.7228727340698242,
      "learning_rate": 4.6795724334966983e-07,
      "loss": 0.6923,
      "step": 30033
    },
    {
      "epoch": 0.8653835071745519,
      "grad_norm": 0.7259037494659424,
      "learning_rate": 4.6776016598560125e-07,
      "loss": 0.6932,
      "step": 30034
    },
    {
      "epoch": 0.8654123206362012,
      "grad_norm": 0.7410576343536377,
      "learning_rate": 4.6756312809281004e-07,
      "loss": 0.6582,
      "step": 30035
    },
    {
      "epoch": 0.8654411340978505,
      "grad_norm": 0.7310746312141418,
      "learning_rate": 4.6736612967301476e-07,
      "loss": 0.6946,
      "step": 30036
    },
    {
      "epoch": 0.8654699475594998,
      "grad_norm": 0.7207586765289307,
      "learning_rate": 4.67169170727928e-07,
      "loss": 0.664,
      "step": 30037
    },
    {
      "epoch": 0.8654987610211491,
      "grad_norm": 0.7146978378295898,
      "learning_rate": 4.669722512592667e-07,
      "loss": 0.6678,
      "step": 30038
    },
    {
      "epoch": 0.8655275744827984,
      "grad_norm": 0.7320386171340942,
      "learning_rate": 4.667753712687456e-07,
      "loss": 0.6858,
      "step": 30039
    },
    {
      "epoch": 0.8655563879444477,
      "grad_norm": 0.7484554052352905,
      "learning_rate": 4.6657853075808057e-07,
      "loss": 0.6893,
      "step": 30040
    },
    {
      "epoch": 0.865585201406097,
      "grad_norm": 0.705676794052124,
      "learning_rate": 4.663817297289841e-07,
      "loss": 0.6688,
      "step": 30041
    },
    {
      "epoch": 0.8656140148677463,
      "grad_norm": 0.7193440198898315,
      "learning_rate": 4.661849681831715e-07,
      "loss": 0.6748,
      "step": 30042
    },
    {
      "epoch": 0.8656428283293955,
      "grad_norm": 0.7223832607269287,
      "learning_rate": 4.659882461223547e-07,
      "loss": 0.675,
      "step": 30043
    },
    {
      "epoch": 0.8656716417910447,
      "grad_norm": 2.2557520866394043,
      "learning_rate": 4.6579156354824906e-07,
      "loss": 0.6868,
      "step": 30044
    },
    {
      "epoch": 0.865700455252694,
      "grad_norm": 0.7226327657699585,
      "learning_rate": 4.6559492046256483e-07,
      "loss": 0.6704,
      "step": 30045
    },
    {
      "epoch": 0.8657292687143433,
      "grad_norm": 0.7354984879493713,
      "learning_rate": 4.6539831686701674e-07,
      "loss": 0.6869,
      "step": 30046
    },
    {
      "epoch": 0.8657580821759926,
      "grad_norm": 0.7164591550827026,
      "learning_rate": 4.6520175276331683e-07,
      "loss": 0.6916,
      "step": 30047
    },
    {
      "epoch": 0.8657868956376419,
      "grad_norm": 0.7335914969444275,
      "learning_rate": 4.6500522815317586e-07,
      "loss": 0.6826,
      "step": 30048
    },
    {
      "epoch": 0.8658157090992912,
      "grad_norm": 0.7259799242019653,
      "learning_rate": 4.6480874303830694e-07,
      "loss": 0.6544,
      "step": 30049
    },
    {
      "epoch": 0.8658445225609405,
      "grad_norm": 0.7309336066246033,
      "learning_rate": 4.646122974204187e-07,
      "loss": 0.705,
      "step": 30050
    },
    {
      "epoch": 0.8658733360225898,
      "grad_norm": 0.7365005016326904,
      "learning_rate": 4.6441589130122535e-07,
      "loss": 0.6938,
      "step": 30051
    },
    {
      "epoch": 0.865902149484239,
      "grad_norm": 0.7340161800384521,
      "learning_rate": 4.6421952468243436e-07,
      "loss": 0.6707,
      "step": 30052
    },
    {
      "epoch": 0.8659309629458883,
      "grad_norm": 0.7122676968574524,
      "learning_rate": 4.640231975657583e-07,
      "loss": 0.6798,
      "step": 30053
    },
    {
      "epoch": 0.8659597764075376,
      "grad_norm": 0.94859379529953,
      "learning_rate": 4.6382690995290524e-07,
      "loss": 0.6852,
      "step": 30054
    },
    {
      "epoch": 0.8659885898691869,
      "grad_norm": 0.7350186705589294,
      "learning_rate": 4.6363066184558605e-07,
      "loss": 0.6744,
      "step": 30055
    },
    {
      "epoch": 0.8660174033308362,
      "grad_norm": 0.7331351637840271,
      "learning_rate": 4.634344532455076e-07,
      "loss": 0.6793,
      "step": 30056
    },
    {
      "epoch": 0.8660462167924855,
      "grad_norm": 0.7165337800979614,
      "learning_rate": 4.63238284154382e-07,
      "loss": 0.688,
      "step": 30057
    },
    {
      "epoch": 0.8660750302541347,
      "grad_norm": 0.7080308794975281,
      "learning_rate": 4.6304215457391445e-07,
      "loss": 0.6991,
      "step": 30058
    },
    {
      "epoch": 0.866103843715784,
      "grad_norm": 0.7062659859657288,
      "learning_rate": 4.6284606450581526e-07,
      "loss": 0.6855,
      "step": 30059
    },
    {
      "epoch": 0.8661326571774333,
      "grad_norm": 0.7371561527252197,
      "learning_rate": 4.626500139517903e-07,
      "loss": 0.6976,
      "step": 30060
    },
    {
      "epoch": 0.8661614706390826,
      "grad_norm": 0.7245014905929565,
      "learning_rate": 4.6245400291354936e-07,
      "loss": 0.683,
      "step": 30061
    },
    {
      "epoch": 0.8661902841007318,
      "grad_norm": 0.7108672261238098,
      "learning_rate": 4.6225803139279656e-07,
      "loss": 0.6544,
      "step": 30062
    },
    {
      "epoch": 0.8662190975623811,
      "grad_norm": 0.719865083694458,
      "learning_rate": 4.620620993912417e-07,
      "loss": 0.6709,
      "step": 30063
    },
    {
      "epoch": 0.8662479110240304,
      "grad_norm": 0.7138286232948303,
      "learning_rate": 4.6186620691058836e-07,
      "loss": 0.6871,
      "step": 30064
    },
    {
      "epoch": 0.8662767244856797,
      "grad_norm": 0.7235943675041199,
      "learning_rate": 4.6167035395254524e-07,
      "loss": 0.6687,
      "step": 30065
    },
    {
      "epoch": 0.866305537947329,
      "grad_norm": 0.7265187501907349,
      "learning_rate": 4.614745405188159e-07,
      "loss": 0.6748,
      "step": 30066
    },
    {
      "epoch": 0.8663343514089783,
      "grad_norm": 0.7229395508766174,
      "learning_rate": 4.6127876661110625e-07,
      "loss": 0.6939,
      "step": 30067
    },
    {
      "epoch": 0.8663631648706276,
      "grad_norm": 0.7135993242263794,
      "learning_rate": 4.610830322311222e-07,
      "loss": 0.6852,
      "step": 30068
    },
    {
      "epoch": 0.8663919783322769,
      "grad_norm": 0.7049998641014099,
      "learning_rate": 4.608873373805672e-07,
      "loss": 0.6554,
      "step": 30069
    },
    {
      "epoch": 0.8664207917939262,
      "grad_norm": 0.7258201241493225,
      "learning_rate": 4.6069168206114677e-07,
      "loss": 0.6941,
      "step": 30070
    },
    {
      "epoch": 0.8664496052555754,
      "grad_norm": 0.720667839050293,
      "learning_rate": 4.6049606627456276e-07,
      "loss": 0.6787,
      "step": 30071
    },
    {
      "epoch": 0.8664784187172246,
      "grad_norm": 0.712165892124176,
      "learning_rate": 4.6030049002252164e-07,
      "loss": 0.6813,
      "step": 30072
    },
    {
      "epoch": 0.8665072321788739,
      "grad_norm": 0.698353111743927,
      "learning_rate": 4.601049533067242e-07,
      "loss": 0.6879,
      "step": 30073
    },
    {
      "epoch": 0.8665360456405232,
      "grad_norm": 0.7310785055160522,
      "learning_rate": 4.599094561288742e-07,
      "loss": 0.6808,
      "step": 30074
    },
    {
      "epoch": 0.8665648591021725,
      "grad_norm": 0.7051450610160828,
      "learning_rate": 4.5971399849067457e-07,
      "loss": 0.664,
      "step": 30075
    },
    {
      "epoch": 0.8665936725638218,
      "grad_norm": 0.7223080992698669,
      "learning_rate": 4.5951858039382846e-07,
      "loss": 0.658,
      "step": 30076
    },
    {
      "epoch": 0.8666224860254711,
      "grad_norm": 0.7362619638442993,
      "learning_rate": 4.5932320184003564e-07,
      "loss": 0.6856,
      "step": 30077
    },
    {
      "epoch": 0.8666512994871204,
      "grad_norm": 0.7198723554611206,
      "learning_rate": 4.5912786283099976e-07,
      "loss": 0.6712,
      "step": 30078
    },
    {
      "epoch": 0.8666801129487697,
      "grad_norm": 0.7380257844924927,
      "learning_rate": 4.5893256336841997e-07,
      "loss": 0.699,
      "step": 30079
    },
    {
      "epoch": 0.866708926410419,
      "grad_norm": 0.7352977991104126,
      "learning_rate": 4.587373034539988e-07,
      "loss": 0.6869,
      "step": 30080
    },
    {
      "epoch": 0.8667377398720683,
      "grad_norm": 0.72127366065979,
      "learning_rate": 4.5854208308943547e-07,
      "loss": 0.6959,
      "step": 30081
    },
    {
      "epoch": 0.8667665533337175,
      "grad_norm": 0.7280701994895935,
      "learning_rate": 4.5834690227643144e-07,
      "loss": 0.6758,
      "step": 30082
    },
    {
      "epoch": 0.8667953667953668,
      "grad_norm": 0.732929527759552,
      "learning_rate": 4.581517610166852e-07,
      "loss": 0.6808,
      "step": 30083
    },
    {
      "epoch": 0.8668241802570161,
      "grad_norm": 0.7020103335380554,
      "learning_rate": 4.5795665931189723e-07,
      "loss": 0.6617,
      "step": 30084
    },
    {
      "epoch": 0.8668529937186653,
      "grad_norm": 0.7247205972671509,
      "learning_rate": 4.577615971637667e-07,
      "loss": 0.6806,
      "step": 30085
    },
    {
      "epoch": 0.8668818071803146,
      "grad_norm": 0.7956895232200623,
      "learning_rate": 4.5756657457399156e-07,
      "loss": 0.6841,
      "step": 30086
    },
    {
      "epoch": 0.8669106206419639,
      "grad_norm": 0.74836665391922,
      "learning_rate": 4.573715915442706e-07,
      "loss": 0.6992,
      "step": 30087
    },
    {
      "epoch": 0.8669394341036132,
      "grad_norm": 0.7160605788230896,
      "learning_rate": 4.5717664807630235e-07,
      "loss": 0.6843,
      "step": 30088
    },
    {
      "epoch": 0.8669682475652625,
      "grad_norm": 0.7272294759750366,
      "learning_rate": 4.56981744171785e-07,
      "loss": 0.683,
      "step": 30089
    },
    {
      "epoch": 0.8669970610269118,
      "grad_norm": 0.7752090096473694,
      "learning_rate": 4.5678687983241433e-07,
      "loss": 0.6866,
      "step": 30090
    },
    {
      "epoch": 0.867025874488561,
      "grad_norm": 0.7666521668434143,
      "learning_rate": 4.5659205505988955e-07,
      "loss": 0.676,
      "step": 30091
    },
    {
      "epoch": 0.8670546879502103,
      "grad_norm": 0.7134283781051636,
      "learning_rate": 4.5639726985590496e-07,
      "loss": 0.7074,
      "step": 30092
    },
    {
      "epoch": 0.8670835014118596,
      "grad_norm": 0.7392973899841309,
      "learning_rate": 4.562025242221596e-07,
      "loss": 0.674,
      "step": 30093
    },
    {
      "epoch": 0.8671123148735089,
      "grad_norm": 0.7302716374397278,
      "learning_rate": 4.5600781816034725e-07,
      "loss": 0.6692,
      "step": 30094
    },
    {
      "epoch": 0.8671411283351582,
      "grad_norm": 0.7243891954421997,
      "learning_rate": 4.558131516721648e-07,
      "loss": 0.6734,
      "step": 30095
    },
    {
      "epoch": 0.8671699417968075,
      "grad_norm": 0.7176621556282043,
      "learning_rate": 4.556185247593065e-07,
      "loss": 0.6724,
      "step": 30096
    },
    {
      "epoch": 0.8671987552584568,
      "grad_norm": 0.7401405572891235,
      "learning_rate": 4.554239374234692e-07,
      "loss": 0.689,
      "step": 30097
    },
    {
      "epoch": 0.867227568720106,
      "grad_norm": 0.717141330242157,
      "learning_rate": 4.5522938966634514e-07,
      "loss": 0.687,
      "step": 30098
    },
    {
      "epoch": 0.8672563821817553,
      "grad_norm": 0.7000468373298645,
      "learning_rate": 4.5503488148963104e-07,
      "loss": 0.6603,
      "step": 30099
    },
    {
      "epoch": 0.8672851956434046,
      "grad_norm": 0.7286403179168701,
      "learning_rate": 4.5484041289501903e-07,
      "loss": 0.6714,
      "step": 30100
    },
    {
      "epoch": 0.8673140091050539,
      "grad_norm": 0.7213414907455444,
      "learning_rate": 4.546459838842049e-07,
      "loss": 0.6763,
      "step": 30101
    },
    {
      "epoch": 0.8673428225667031,
      "grad_norm": 0.7272660732269287,
      "learning_rate": 4.544515944588795e-07,
      "loss": 0.6782,
      "step": 30102
    },
    {
      "epoch": 0.8673716360283524,
      "grad_norm": 0.734982967376709,
      "learning_rate": 4.542572446207377e-07,
      "loss": 0.6682,
      "step": 30103
    },
    {
      "epoch": 0.8674004494900017,
      "grad_norm": 0.7248273491859436,
      "learning_rate": 4.5406293437146977e-07,
      "loss": 0.6984,
      "step": 30104
    },
    {
      "epoch": 0.867429262951651,
      "grad_norm": 0.7362143993377686,
      "learning_rate": 4.538686637127704e-07,
      "loss": 0.6865,
      "step": 30105
    },
    {
      "epoch": 0.8674580764133003,
      "grad_norm": 0.7329164147377014,
      "learning_rate": 4.536744326463305e-07,
      "loss": 0.6535,
      "step": 30106
    },
    {
      "epoch": 0.8674868898749496,
      "grad_norm": 0.7358859777450562,
      "learning_rate": 4.5348024117384095e-07,
      "loss": 0.7031,
      "step": 30107
    },
    {
      "epoch": 0.8675157033365989,
      "grad_norm": 0.7486163377761841,
      "learning_rate": 4.532860892969948e-07,
      "loss": 0.6943,
      "step": 30108
    },
    {
      "epoch": 0.8675445167982482,
      "grad_norm": 0.7149650454521179,
      "learning_rate": 4.5309197701748076e-07,
      "loss": 0.6759,
      "step": 30109
    },
    {
      "epoch": 0.8675733302598975,
      "grad_norm": 0.7345487475395203,
      "learning_rate": 4.528979043369902e-07,
      "loss": 0.6912,
      "step": 30110
    },
    {
      "epoch": 0.8676021437215468,
      "grad_norm": 0.7118633389472961,
      "learning_rate": 4.5270387125721403e-07,
      "loss": 0.6933,
      "step": 30111
    },
    {
      "epoch": 0.8676309571831959,
      "grad_norm": 0.7060396075248718,
      "learning_rate": 4.52509877779842e-07,
      "loss": 0.669,
      "step": 30112
    },
    {
      "epoch": 0.8676597706448452,
      "grad_norm": 0.7074580192565918,
      "learning_rate": 4.523159239065622e-07,
      "loss": 0.6404,
      "step": 30113
    },
    {
      "epoch": 0.8676885841064945,
      "grad_norm": 0.7307451367378235,
      "learning_rate": 4.521220096390655e-07,
      "loss": 0.6925,
      "step": 30114
    },
    {
      "epoch": 0.8677173975681438,
      "grad_norm": 0.7004702687263489,
      "learning_rate": 4.5192813497903885e-07,
      "loss": 0.6404,
      "step": 30115
    },
    {
      "epoch": 0.8677462110297931,
      "grad_norm": 0.7152278423309326,
      "learning_rate": 4.517342999281732e-07,
      "loss": 0.6514,
      "step": 30116
    },
    {
      "epoch": 0.8677750244914424,
      "grad_norm": 0.7330604195594788,
      "learning_rate": 4.515405044881538e-07,
      "loss": 0.6899,
      "step": 30117
    },
    {
      "epoch": 0.8678038379530917,
      "grad_norm": 0.7256236672401428,
      "learning_rate": 4.5134674866067097e-07,
      "loss": 0.6949,
      "step": 30118
    },
    {
      "epoch": 0.867832651414741,
      "grad_norm": 0.7274736166000366,
      "learning_rate": 4.5115303244741006e-07,
      "loss": 0.6665,
      "step": 30119
    },
    {
      "epoch": 0.8678614648763903,
      "grad_norm": 0.7074441909790039,
      "learning_rate": 4.5095935585005966e-07,
      "loss": 0.6655,
      "step": 30120
    },
    {
      "epoch": 0.8678902783380396,
      "grad_norm": 0.7397364377975464,
      "learning_rate": 4.5076571887030516e-07,
      "loss": 0.662,
      "step": 30121
    },
    {
      "epoch": 0.8679190917996888,
      "grad_norm": 0.716992974281311,
      "learning_rate": 4.5057212150983353e-07,
      "loss": 0.6671,
      "step": 30122
    },
    {
      "epoch": 0.8679479052613381,
      "grad_norm": 0.7031657695770264,
      "learning_rate": 4.5037856377033116e-07,
      "loss": 0.6579,
      "step": 30123
    },
    {
      "epoch": 0.8679767187229874,
      "grad_norm": 0.7142385840415955,
      "learning_rate": 4.5018504565348455e-07,
      "loss": 0.6797,
      "step": 30124
    },
    {
      "epoch": 0.8680055321846366,
      "grad_norm": 0.7377951741218567,
      "learning_rate": 4.499915671609767e-07,
      "loss": 0.6949,
      "step": 30125
    },
    {
      "epoch": 0.8680343456462859,
      "grad_norm": 0.7267813086509705,
      "learning_rate": 4.4979812829449467e-07,
      "loss": 0.6978,
      "step": 30126
    },
    {
      "epoch": 0.8680631591079352,
      "grad_norm": 0.7208020687103271,
      "learning_rate": 4.496047290557226e-07,
      "loss": 0.6863,
      "step": 30127
    },
    {
      "epoch": 0.8680919725695845,
      "grad_norm": 0.7206710577011108,
      "learning_rate": 4.494113694463442e-07,
      "loss": 0.6831,
      "step": 30128
    },
    {
      "epoch": 0.8681207860312338,
      "grad_norm": 0.7482724785804749,
      "learning_rate": 4.4921804946804427e-07,
      "loss": 0.72,
      "step": 30129
    },
    {
      "epoch": 0.8681495994928831,
      "grad_norm": 0.7439186573028564,
      "learning_rate": 4.490247691225058e-07,
      "loss": 0.7106,
      "step": 30130
    },
    {
      "epoch": 0.8681784129545324,
      "grad_norm": 0.7243709564208984,
      "learning_rate": 4.4883152841141255e-07,
      "loss": 0.6885,
      "step": 30131
    },
    {
      "epoch": 0.8682072264161816,
      "grad_norm": 0.734688937664032,
      "learning_rate": 4.486383273364464e-07,
      "loss": 0.6933,
      "step": 30132
    },
    {
      "epoch": 0.8682360398778309,
      "grad_norm": 0.706895112991333,
      "learning_rate": 4.4844516589929166e-07,
      "loss": 0.6692,
      "step": 30133
    },
    {
      "epoch": 0.8682648533394802,
      "grad_norm": 0.7076665163040161,
      "learning_rate": 4.4825204410162803e-07,
      "loss": 0.6796,
      "step": 30134
    },
    {
      "epoch": 0.8682936668011295,
      "grad_norm": 0.7157620191574097,
      "learning_rate": 4.4805896194514085e-07,
      "loss": 0.6951,
      "step": 30135
    },
    {
      "epoch": 0.8683224802627788,
      "grad_norm": 0.7099992036819458,
      "learning_rate": 4.4786591943150883e-07,
      "loss": 0.6699,
      "step": 30136
    },
    {
      "epoch": 0.8683512937244281,
      "grad_norm": 0.7204059362411499,
      "learning_rate": 4.47672916562415e-07,
      "loss": 0.6865,
      "step": 30137
    },
    {
      "epoch": 0.8683801071860774,
      "grad_norm": 0.7010191679000854,
      "learning_rate": 4.474799533395385e-07,
      "loss": 0.6454,
      "step": 30138
    },
    {
      "epoch": 0.8684089206477266,
      "grad_norm": 0.7264352440834045,
      "learning_rate": 4.4728702976456153e-07,
      "loss": 0.6584,
      "step": 30139
    },
    {
      "epoch": 0.8684377341093759,
      "grad_norm": 0.7208396792411804,
      "learning_rate": 4.4709414583916313e-07,
      "loss": 0.6682,
      "step": 30140
    },
    {
      "epoch": 0.8684665475710251,
      "grad_norm": 0.7233030796051025,
      "learning_rate": 4.4690130156502367e-07,
      "loss": 0.6905,
      "step": 30141
    },
    {
      "epoch": 0.8684953610326744,
      "grad_norm": 0.7463968992233276,
      "learning_rate": 4.4670849694382177e-07,
      "loss": 0.6933,
      "step": 30142
    },
    {
      "epoch": 0.8685241744943237,
      "grad_norm": 0.7146976590156555,
      "learning_rate": 4.465157319772373e-07,
      "loss": 0.6952,
      "step": 30143
    },
    {
      "epoch": 0.868552987955973,
      "grad_norm": 0.7227776646614075,
      "learning_rate": 4.463230066669499e-07,
      "loss": 0.6898,
      "step": 30144
    },
    {
      "epoch": 0.8685818014176223,
      "grad_norm": 0.7220497131347656,
      "learning_rate": 4.4613032101463613e-07,
      "loss": 0.6778,
      "step": 30145
    },
    {
      "epoch": 0.8686106148792716,
      "grad_norm": 0.7148504853248596,
      "learning_rate": 4.459376750219757e-07,
      "loss": 0.6651,
      "step": 30146
    },
    {
      "epoch": 0.8686394283409209,
      "grad_norm": 0.708965003490448,
      "learning_rate": 4.457450686906456e-07,
      "loss": 0.6645,
      "step": 30147
    },
    {
      "epoch": 0.8686682418025702,
      "grad_norm": 0.7247734069824219,
      "learning_rate": 4.455525020223234e-07,
      "loss": 0.6855,
      "step": 30148
    },
    {
      "epoch": 0.8686970552642195,
      "grad_norm": 0.7178462743759155,
      "learning_rate": 4.4535997501868547e-07,
      "loss": 0.6551,
      "step": 30149
    },
    {
      "epoch": 0.8687258687258688,
      "grad_norm": 0.7262029051780701,
      "learning_rate": 4.451674876814105e-07,
      "loss": 0.6942,
      "step": 30150
    },
    {
      "epoch": 0.8687546821875181,
      "grad_norm": 0.6995120048522949,
      "learning_rate": 4.449750400121722e-07,
      "loss": 0.6651,
      "step": 30151
    },
    {
      "epoch": 0.8687834956491672,
      "grad_norm": 0.7232884168624878,
      "learning_rate": 4.447826320126492e-07,
      "loss": 0.6897,
      "step": 30152
    },
    {
      "epoch": 0.8688123091108165,
      "grad_norm": 0.7195199131965637,
      "learning_rate": 4.445902636845145e-07,
      "loss": 0.6839,
      "step": 30153
    },
    {
      "epoch": 0.8688411225724658,
      "grad_norm": 0.7216090559959412,
      "learning_rate": 4.4439793502944637e-07,
      "loss": 0.6391,
      "step": 30154
    },
    {
      "epoch": 0.8688699360341151,
      "grad_norm": 0.7125133275985718,
      "learning_rate": 4.4420564604911734e-07,
      "loss": 0.668,
      "step": 30155
    },
    {
      "epoch": 0.8688987494957644,
      "grad_norm": 0.7160711288452148,
      "learning_rate": 4.440133967452037e-07,
      "loss": 0.6859,
      "step": 30156
    },
    {
      "epoch": 0.8689275629574137,
      "grad_norm": 0.7320045828819275,
      "learning_rate": 4.4382118711937815e-07,
      "loss": 0.6688,
      "step": 30157
    },
    {
      "epoch": 0.868956376419063,
      "grad_norm": 0.7295023798942566,
      "learning_rate": 4.436290171733154e-07,
      "loss": 0.6753,
      "step": 30158
    },
    {
      "epoch": 0.8689851898807123,
      "grad_norm": 0.7317696213722229,
      "learning_rate": 4.434368869086897e-07,
      "loss": 0.6984,
      "step": 30159
    },
    {
      "epoch": 0.8690140033423616,
      "grad_norm": 0.7303234934806824,
      "learning_rate": 4.432447963271741e-07,
      "loss": 0.6827,
      "step": 30160
    },
    {
      "epoch": 0.8690428168040109,
      "grad_norm": 0.7212860584259033,
      "learning_rate": 4.4305274543044064e-07,
      "loss": 0.6919,
      "step": 30161
    },
    {
      "epoch": 0.8690716302656601,
      "grad_norm": 0.7355579137802124,
      "learning_rate": 4.428607342201635e-07,
      "loss": 0.6819,
      "step": 30162
    },
    {
      "epoch": 0.8691004437273094,
      "grad_norm": 0.7229055762290955,
      "learning_rate": 4.4266876269801305e-07,
      "loss": 0.68,
      "step": 30163
    },
    {
      "epoch": 0.8691292571889587,
      "grad_norm": 0.7231253981590271,
      "learning_rate": 4.4247683086566187e-07,
      "loss": 0.6757,
      "step": 30164
    },
    {
      "epoch": 0.869158070650608,
      "grad_norm": 0.7108719944953918,
      "learning_rate": 4.422849387247824e-07,
      "loss": 0.6935,
      "step": 30165
    },
    {
      "epoch": 0.8691868841122572,
      "grad_norm": 0.7187228798866272,
      "learning_rate": 4.420930862770445e-07,
      "loss": 0.7094,
      "step": 30166
    },
    {
      "epoch": 0.8692156975739065,
      "grad_norm": 0.6988017559051514,
      "learning_rate": 4.4190127352412013e-07,
      "loss": 0.667,
      "step": 30167
    },
    {
      "epoch": 0.8692445110355558,
      "grad_norm": 0.7146445512771606,
      "learning_rate": 4.417095004676786e-07,
      "loss": 0.6802,
      "step": 30168
    },
    {
      "epoch": 0.8692733244972051,
      "grad_norm": 0.7224573493003845,
      "learning_rate": 4.415177671093912e-07,
      "loss": 0.6913,
      "step": 30169
    },
    {
      "epoch": 0.8693021379588544,
      "grad_norm": 0.7062845826148987,
      "learning_rate": 4.413260734509256e-07,
      "loss": 0.6805,
      "step": 30170
    },
    {
      "epoch": 0.8693309514205037,
      "grad_norm": 0.7144404053688049,
      "learning_rate": 4.4113441949395486e-07,
      "loss": 0.6872,
      "step": 30171
    },
    {
      "epoch": 0.8693597648821529,
      "grad_norm": 0.7346684336662292,
      "learning_rate": 4.409428052401454e-07,
      "loss": 0.7005,
      "step": 30172
    },
    {
      "epoch": 0.8693885783438022,
      "grad_norm": 0.7304868102073669,
      "learning_rate": 4.407512306911671e-07,
      "loss": 0.6627,
      "step": 30173
    },
    {
      "epoch": 0.8694173918054515,
      "grad_norm": 0.7240089774131775,
      "learning_rate": 4.405596958486874e-07,
      "loss": 0.7235,
      "step": 30174
    },
    {
      "epoch": 0.8694462052671008,
      "grad_norm": 0.7221636176109314,
      "learning_rate": 4.4036820071437615e-07,
      "loss": 0.6678,
      "step": 30175
    },
    {
      "epoch": 0.8694750187287501,
      "grad_norm": 0.7185556888580322,
      "learning_rate": 4.4017674528989863e-07,
      "loss": 0.6842,
      "step": 30176
    },
    {
      "epoch": 0.8695038321903994,
      "grad_norm": 0.7147162556648254,
      "learning_rate": 4.3998532957692463e-07,
      "loss": 0.66,
      "step": 30177
    },
    {
      "epoch": 0.8695326456520487,
      "grad_norm": 0.7427768707275391,
      "learning_rate": 4.3979395357711895e-07,
      "loss": 0.6846,
      "step": 30178
    },
    {
      "epoch": 0.8695614591136979,
      "grad_norm": 0.714015007019043,
      "learning_rate": 4.3960261729215025e-07,
      "loss": 0.6872,
      "step": 30179
    },
    {
      "epoch": 0.8695902725753472,
      "grad_norm": 0.7138318419456482,
      "learning_rate": 4.394113207236833e-07,
      "loss": 0.6871,
      "step": 30180
    },
    {
      "epoch": 0.8696190860369964,
      "grad_norm": 0.7138150930404663,
      "learning_rate": 4.3922006387338457e-07,
      "loss": 0.6736,
      "step": 30181
    },
    {
      "epoch": 0.8696478994986457,
      "grad_norm": 0.7240716218948364,
      "learning_rate": 4.3902884674292045e-07,
      "loss": 0.6742,
      "step": 30182
    },
    {
      "epoch": 0.869676712960295,
      "grad_norm": 0.7075132727622986,
      "learning_rate": 4.3883766933395634e-07,
      "loss": 0.664,
      "step": 30183
    },
    {
      "epoch": 0.8697055264219443,
      "grad_norm": 0.7181166410446167,
      "learning_rate": 4.3864653164815585e-07,
      "loss": 0.6896,
      "step": 30184
    },
    {
      "epoch": 0.8697343398835936,
      "grad_norm": 0.7538742423057556,
      "learning_rate": 4.3845543368718436e-07,
      "loss": 0.7188,
      "step": 30185
    },
    {
      "epoch": 0.8697631533452429,
      "grad_norm": 0.731216549873352,
      "learning_rate": 4.382643754527072e-07,
      "loss": 0.7018,
      "step": 30186
    },
    {
      "epoch": 0.8697919668068922,
      "grad_norm": 0.7385568618774414,
      "learning_rate": 4.3807335694638573e-07,
      "loss": 0.7009,
      "step": 30187
    },
    {
      "epoch": 0.8698207802685415,
      "grad_norm": 0.7251545190811157,
      "learning_rate": 4.3788237816988654e-07,
      "loss": 0.6854,
      "step": 30188
    },
    {
      "epoch": 0.8698495937301908,
      "grad_norm": 0.7230910062789917,
      "learning_rate": 4.376914391248704e-07,
      "loss": 0.6889,
      "step": 30189
    },
    {
      "epoch": 0.8698784071918401,
      "grad_norm": 0.7355459332466125,
      "learning_rate": 4.375005398130022e-07,
      "loss": 0.7136,
      "step": 30190
    },
    {
      "epoch": 0.8699072206534894,
      "grad_norm": 0.7215323448181152,
      "learning_rate": 4.3730968023594277e-07,
      "loss": 0.7094,
      "step": 30191
    },
    {
      "epoch": 0.8699360341151386,
      "grad_norm": 0.7273064255714417,
      "learning_rate": 4.371188603953552e-07,
      "loss": 0.69,
      "step": 30192
    },
    {
      "epoch": 0.8699648475767878,
      "grad_norm": 0.7351850271224976,
      "learning_rate": 4.3692808029290103e-07,
      "loss": 0.6978,
      "step": 30193
    },
    {
      "epoch": 0.8699936610384371,
      "grad_norm": 0.7120800614356995,
      "learning_rate": 4.3673733993024103e-07,
      "loss": 0.6733,
      "step": 30194
    },
    {
      "epoch": 0.8700224745000864,
      "grad_norm": 0.7370272874832153,
      "learning_rate": 4.3654663930903787e-07,
      "loss": 0.6748,
      "step": 30195
    },
    {
      "epoch": 0.8700512879617357,
      "grad_norm": 0.6940509676933289,
      "learning_rate": 4.3635597843095236e-07,
      "loss": 0.6499,
      "step": 30196
    },
    {
      "epoch": 0.870080101423385,
      "grad_norm": 0.7217434644699097,
      "learning_rate": 4.361653572976432e-07,
      "loss": 0.6811,
      "step": 30197
    },
    {
      "epoch": 0.8701089148850343,
      "grad_norm": 0.7245029211044312,
      "learning_rate": 4.35974775910773e-07,
      "loss": 0.6778,
      "step": 30198
    },
    {
      "epoch": 0.8701377283466836,
      "grad_norm": 0.7598706483840942,
      "learning_rate": 4.357842342719987e-07,
      "loss": 0.666,
      "step": 30199
    },
    {
      "epoch": 0.8701665418083329,
      "grad_norm": 0.7145060300827026,
      "learning_rate": 4.355937323829823e-07,
      "loss": 0.6679,
      "step": 30200
    },
    {
      "epoch": 0.8701953552699822,
      "grad_norm": 0.7376024723052979,
      "learning_rate": 4.354032702453809e-07,
      "loss": 0.687,
      "step": 30201
    },
    {
      "epoch": 0.8702241687316314,
      "grad_norm": 0.7210973501205444,
      "learning_rate": 4.3521284786085414e-07,
      "loss": 0.6639,
      "step": 30202
    },
    {
      "epoch": 0.8702529821932807,
      "grad_norm": 0.7456167340278625,
      "learning_rate": 4.350224652310614e-07,
      "loss": 0.694,
      "step": 30203
    },
    {
      "epoch": 0.87028179565493,
      "grad_norm": 0.7277502417564392,
      "learning_rate": 4.348321223576585e-07,
      "loss": 0.6928,
      "step": 30204
    },
    {
      "epoch": 0.8703106091165793,
      "grad_norm": 0.7271600961685181,
      "learning_rate": 4.3464181924230517e-07,
      "loss": 0.6742,
      "step": 30205
    },
    {
      "epoch": 0.8703394225782285,
      "grad_norm": 0.7094466090202332,
      "learning_rate": 4.344515558866574e-07,
      "loss": 0.688,
      "step": 30206
    },
    {
      "epoch": 0.8703682360398778,
      "grad_norm": 0.7211031913757324,
      "learning_rate": 4.342613322923722e-07,
      "loss": 0.6896,
      "step": 30207
    },
    {
      "epoch": 0.8703970495015271,
      "grad_norm": 0.712131679058075,
      "learning_rate": 4.340711484611071e-07,
      "loss": 0.6654,
      "step": 30208
    },
    {
      "epoch": 0.8704258629631764,
      "grad_norm": 0.7286235094070435,
      "learning_rate": 4.338810043945191e-07,
      "loss": 0.6912,
      "step": 30209
    },
    {
      "epoch": 0.8704546764248257,
      "grad_norm": 0.7183901071548462,
      "learning_rate": 4.3369090009426185e-07,
      "loss": 0.6821,
      "step": 30210
    },
    {
      "epoch": 0.870483489886475,
      "grad_norm": 0.7554968595504761,
      "learning_rate": 4.3350083556199294e-07,
      "loss": 0.6919,
      "step": 30211
    },
    {
      "epoch": 0.8705123033481242,
      "grad_norm": 0.7398781180381775,
      "learning_rate": 4.333108107993667e-07,
      "loss": 0.6974,
      "step": 30212
    },
    {
      "epoch": 0.8705411168097735,
      "grad_norm": 0.8636775016784668,
      "learning_rate": 4.331208258080383e-07,
      "loss": 0.6792,
      "step": 30213
    },
    {
      "epoch": 0.8705699302714228,
      "grad_norm": 0.7293605208396912,
      "learning_rate": 4.3293088058966205e-07,
      "loss": 0.7082,
      "step": 30214
    },
    {
      "epoch": 0.8705987437330721,
      "grad_norm": 0.7236383557319641,
      "learning_rate": 4.327409751458933e-07,
      "loss": 0.6795,
      "step": 30215
    },
    {
      "epoch": 0.8706275571947214,
      "grad_norm": 0.7378845810890198,
      "learning_rate": 4.325511094783841e-07,
      "loss": 0.6875,
      "step": 30216
    },
    {
      "epoch": 0.8706563706563707,
      "grad_norm": 0.71607905626297,
      "learning_rate": 4.323612835887897e-07,
      "loss": 0.6773,
      "step": 30217
    },
    {
      "epoch": 0.87068518411802,
      "grad_norm": 0.7338182330131531,
      "learning_rate": 4.321714974787605e-07,
      "loss": 0.711,
      "step": 30218
    },
    {
      "epoch": 0.8707139975796693,
      "grad_norm": 0.7134485244750977,
      "learning_rate": 4.3198175114995356e-07,
      "loss": 0.6766,
      "step": 30219
    },
    {
      "epoch": 0.8707428110413185,
      "grad_norm": 0.7142478823661804,
      "learning_rate": 4.317920446040186e-07,
      "loss": 0.6785,
      "step": 30220
    },
    {
      "epoch": 0.8707716245029677,
      "grad_norm": 0.7403347492218018,
      "learning_rate": 4.316023778426087e-07,
      "loss": 0.6938,
      "step": 30221
    },
    {
      "epoch": 0.870800437964617,
      "grad_norm": 0.7096730470657349,
      "learning_rate": 4.3141275086737486e-07,
      "loss": 0.6616,
      "step": 30222
    },
    {
      "epoch": 0.8708292514262663,
      "grad_norm": 0.7257524132728577,
      "learning_rate": 4.312231636799685e-07,
      "loss": 0.6896,
      "step": 30223
    },
    {
      "epoch": 0.8708580648879156,
      "grad_norm": 0.7100397944450378,
      "learning_rate": 4.3103361628204276e-07,
      "loss": 0.6811,
      "step": 30224
    },
    {
      "epoch": 0.8708868783495649,
      "grad_norm": 0.7194767594337463,
      "learning_rate": 4.3084410867524576e-07,
      "loss": 0.6721,
      "step": 30225
    },
    {
      "epoch": 0.8709156918112142,
      "grad_norm": 0.7169797420501709,
      "learning_rate": 4.3065464086123055e-07,
      "loss": 0.7129,
      "step": 30226
    },
    {
      "epoch": 0.8709445052728635,
      "grad_norm": 0.7254153490066528,
      "learning_rate": 4.3046521284164424e-07,
      "loss": 0.7049,
      "step": 30227
    },
    {
      "epoch": 0.8709733187345128,
      "grad_norm": 0.7256010174751282,
      "learning_rate": 4.3027582461813876e-07,
      "loss": 0.6708,
      "step": 30228
    },
    {
      "epoch": 0.8710021321961621,
      "grad_norm": 0.726196825504303,
      "learning_rate": 4.3008647619236233e-07,
      "loss": 0.6689,
      "step": 30229
    },
    {
      "epoch": 0.8710309456578114,
      "grad_norm": 0.7233315110206604,
      "learning_rate": 4.2989716756596465e-07,
      "loss": 0.6837,
      "step": 30230
    },
    {
      "epoch": 0.8710597591194607,
      "grad_norm": 0.7334036231040955,
      "learning_rate": 4.2970789874059394e-07,
      "loss": 0.7071,
      "step": 30231
    },
    {
      "epoch": 0.8710885725811099,
      "grad_norm": 0.7181302309036255,
      "learning_rate": 4.2951866971789937e-07,
      "loss": 0.6575,
      "step": 30232
    },
    {
      "epoch": 0.8711173860427591,
      "grad_norm": 0.737040102481842,
      "learning_rate": 4.2932948049952795e-07,
      "loss": 0.6678,
      "step": 30233
    },
    {
      "epoch": 0.8711461995044084,
      "grad_norm": 0.707410454750061,
      "learning_rate": 4.291403310871284e-07,
      "loss": 0.6828,
      "step": 30234
    },
    {
      "epoch": 0.8711750129660577,
      "grad_norm": 0.7493996024131775,
      "learning_rate": 4.289512214823466e-07,
      "loss": 0.6894,
      "step": 30235
    },
    {
      "epoch": 0.871203826427707,
      "grad_norm": 0.7427471876144409,
      "learning_rate": 4.2876215168683123e-07,
      "loss": 0.6871,
      "step": 30236
    },
    {
      "epoch": 0.8712326398893563,
      "grad_norm": 0.7211115956306458,
      "learning_rate": 4.285731217022271e-07,
      "loss": 0.6849,
      "step": 30237
    },
    {
      "epoch": 0.8712614533510056,
      "grad_norm": 0.7306986451148987,
      "learning_rate": 4.283841315301823e-07,
      "loss": 0.6623,
      "step": 30238
    },
    {
      "epoch": 0.8712902668126549,
      "grad_norm": 0.7303096652030945,
      "learning_rate": 4.281951811723406e-07,
      "loss": 0.6649,
      "step": 30239
    },
    {
      "epoch": 0.8713190802743042,
      "grad_norm": 0.7214364409446716,
      "learning_rate": 4.2800627063034895e-07,
      "loss": 0.6756,
      "step": 30240
    },
    {
      "epoch": 0.8713478937359534,
      "grad_norm": 0.7317470908164978,
      "learning_rate": 4.2781739990585215e-07,
      "loss": 0.6798,
      "step": 30241
    },
    {
      "epoch": 0.8713767071976027,
      "grad_norm": 0.7286384701728821,
      "learning_rate": 4.276285690004961e-07,
      "loss": 0.683,
      "step": 30242
    },
    {
      "epoch": 0.871405520659252,
      "grad_norm": 0.7095968723297119,
      "learning_rate": 4.2743977791592394e-07,
      "loss": 0.6711,
      "step": 30243
    },
    {
      "epoch": 0.8714343341209013,
      "grad_norm": 0.7122278809547424,
      "learning_rate": 4.2725102665378104e-07,
      "loss": 0.6602,
      "step": 30244
    },
    {
      "epoch": 0.8714631475825506,
      "grad_norm": 0.7382186055183411,
      "learning_rate": 4.2706231521570996e-07,
      "loss": 0.7086,
      "step": 30245
    },
    {
      "epoch": 0.8714919610441999,
      "grad_norm": 0.7163243293762207,
      "learning_rate": 4.2687364360335495e-07,
      "loss": 0.7049,
      "step": 30246
    },
    {
      "epoch": 0.8715207745058491,
      "grad_norm": 0.7333662509918213,
      "learning_rate": 4.2668501181835964e-07,
      "loss": 0.6843,
      "step": 30247
    },
    {
      "epoch": 0.8715495879674984,
      "grad_norm": 0.7086206674575806,
      "learning_rate": 4.2649641986236556e-07,
      "loss": 0.6759,
      "step": 30248
    },
    {
      "epoch": 0.8715784014291477,
      "grad_norm": 0.717860221862793,
      "learning_rate": 4.263078677370164e-07,
      "loss": 0.6643,
      "step": 30249
    },
    {
      "epoch": 0.871607214890797,
      "grad_norm": 0.737001359462738,
      "learning_rate": 4.26119355443953e-07,
      "loss": 0.6797,
      "step": 30250
    },
    {
      "epoch": 0.8716360283524462,
      "grad_norm": 0.7219045758247375,
      "learning_rate": 4.259308829848191e-07,
      "loss": 0.6878,
      "step": 30251
    },
    {
      "epoch": 0.8716648418140955,
      "grad_norm": 0.713951051235199,
      "learning_rate": 4.257424503612534e-07,
      "loss": 0.666,
      "step": 30252
    },
    {
      "epoch": 0.8716936552757448,
      "grad_norm": 0.7219705581665039,
      "learning_rate": 4.255540575748984e-07,
      "loss": 0.6566,
      "step": 30253
    },
    {
      "epoch": 0.8717224687373941,
      "grad_norm": 0.7086055278778076,
      "learning_rate": 4.253657046273951e-07,
      "loss": 0.6761,
      "step": 30254
    },
    {
      "epoch": 0.8717512821990434,
      "grad_norm": 0.7726699113845825,
      "learning_rate": 4.251773915203844e-07,
      "loss": 0.6879,
      "step": 30255
    },
    {
      "epoch": 0.8717800956606927,
      "grad_norm": 0.7300506234169006,
      "learning_rate": 4.249891182555044e-07,
      "loss": 0.7116,
      "step": 30256
    },
    {
      "epoch": 0.871808909122342,
      "grad_norm": 0.728735625743866,
      "learning_rate": 4.248008848343965e-07,
      "loss": 0.7089,
      "step": 30257
    },
    {
      "epoch": 0.8718377225839913,
      "grad_norm": 0.7441645264625549,
      "learning_rate": 4.2461269125869896e-07,
      "loss": 0.697,
      "step": 30258
    },
    {
      "epoch": 0.8718665360456406,
      "grad_norm": 0.7294533848762512,
      "learning_rate": 4.2442453753005154e-07,
      "loss": 0.6547,
      "step": 30259
    },
    {
      "epoch": 0.8718953495072898,
      "grad_norm": 0.7191399335861206,
      "learning_rate": 4.242364236500923e-07,
      "loss": 0.7068,
      "step": 30260
    },
    {
      "epoch": 0.871924162968939,
      "grad_norm": 0.719575047492981,
      "learning_rate": 4.240483496204589e-07,
      "loss": 0.6488,
      "step": 30261
    },
    {
      "epoch": 0.8719529764305883,
      "grad_norm": 0.7488582134246826,
      "learning_rate": 4.2386031544279173e-07,
      "loss": 0.6632,
      "step": 30262
    },
    {
      "epoch": 0.8719817898922376,
      "grad_norm": 0.7029035687446594,
      "learning_rate": 4.2367232111872547e-07,
      "loss": 0.6842,
      "step": 30263
    },
    {
      "epoch": 0.8720106033538869,
      "grad_norm": 0.7148947715759277,
      "learning_rate": 4.2348436664989946e-07,
      "loss": 0.6725,
      "step": 30264
    },
    {
      "epoch": 0.8720394168155362,
      "grad_norm": 0.7182711362838745,
      "learning_rate": 4.23296452037949e-07,
      "loss": 0.6789,
      "step": 30265
    },
    {
      "epoch": 0.8720682302771855,
      "grad_norm": 0.7472562789916992,
      "learning_rate": 4.231085772845117e-07,
      "loss": 0.68,
      "step": 30266
    },
    {
      "epoch": 0.8720970437388348,
      "grad_norm": 0.7256298065185547,
      "learning_rate": 4.2292074239122294e-07,
      "loss": 0.7061,
      "step": 30267
    },
    {
      "epoch": 0.8721258572004841,
      "grad_norm": 0.7134672403335571,
      "learning_rate": 4.2273294735972025e-07,
      "loss": 0.6811,
      "step": 30268
    },
    {
      "epoch": 0.8721546706621334,
      "grad_norm": 0.6998875737190247,
      "learning_rate": 4.2254519219163735e-07,
      "loss": 0.6605,
      "step": 30269
    },
    {
      "epoch": 0.8721834841237827,
      "grad_norm": 0.7053899765014648,
      "learning_rate": 4.223574768886107e-07,
      "loss": 0.6742,
      "step": 30270
    },
    {
      "epoch": 0.872212297585432,
      "grad_norm": 0.7173348069190979,
      "learning_rate": 4.221698014522735e-07,
      "loss": 0.6968,
      "step": 30271
    },
    {
      "epoch": 0.8722411110470812,
      "grad_norm": 0.7115755081176758,
      "learning_rate": 4.219821658842621e-07,
      "loss": 0.6846,
      "step": 30272
    },
    {
      "epoch": 0.8722699245087305,
      "grad_norm": 0.7363868355751038,
      "learning_rate": 4.2179457018620916e-07,
      "loss": 0.6923,
      "step": 30273
    },
    {
      "epoch": 0.8722987379703797,
      "grad_norm": 0.7441946864128113,
      "learning_rate": 4.216070143597495e-07,
      "loss": 0.667,
      "step": 30274
    },
    {
      "epoch": 0.872327551432029,
      "grad_norm": 0.7311256527900696,
      "learning_rate": 4.214194984065151e-07,
      "loss": 0.6628,
      "step": 30275
    },
    {
      "epoch": 0.8723563648936783,
      "grad_norm": 0.7428518533706665,
      "learning_rate": 4.2123202232814077e-07,
      "loss": 0.6872,
      "step": 30276
    },
    {
      "epoch": 0.8723851783553276,
      "grad_norm": 0.7133564352989197,
      "learning_rate": 4.2104458612625685e-07,
      "loss": 0.6902,
      "step": 30277
    },
    {
      "epoch": 0.8724139918169769,
      "grad_norm": 0.7398750185966492,
      "learning_rate": 4.2085718980249933e-07,
      "loss": 0.6925,
      "step": 30278
    },
    {
      "epoch": 0.8724428052786262,
      "grad_norm": 0.7190512418746948,
      "learning_rate": 4.2066983335849685e-07,
      "loss": 0.657,
      "step": 30279
    },
    {
      "epoch": 0.8724716187402755,
      "grad_norm": 0.7227161526679993,
      "learning_rate": 4.204825167958837e-07,
      "loss": 0.6906,
      "step": 30280
    },
    {
      "epoch": 0.8725004322019247,
      "grad_norm": 0.7211740016937256,
      "learning_rate": 4.2029524011628906e-07,
      "loss": 0.6658,
      "step": 30281
    },
    {
      "epoch": 0.872529245663574,
      "grad_norm": 0.7238983511924744,
      "learning_rate": 4.201080033213456e-07,
      "loss": 0.6753,
      "step": 30282
    },
    {
      "epoch": 0.8725580591252233,
      "grad_norm": 0.7103614807128906,
      "learning_rate": 4.1992080641268253e-07,
      "loss": 0.6933,
      "step": 30283
    },
    {
      "epoch": 0.8725868725868726,
      "grad_norm": 0.6963347792625427,
      "learning_rate": 4.1973364939193074e-07,
      "loss": 0.6663,
      "step": 30284
    },
    {
      "epoch": 0.8726156860485219,
      "grad_norm": 0.7196746468544006,
      "learning_rate": 4.195465322607212e-07,
      "loss": 0.6904,
      "step": 30285
    },
    {
      "epoch": 0.8726444995101712,
      "grad_norm": 0.7107820510864258,
      "learning_rate": 4.193594550206814e-07,
      "loss": 0.6881,
      "step": 30286
    },
    {
      "epoch": 0.8726733129718204,
      "grad_norm": 0.7352875471115112,
      "learning_rate": 4.1917241767344296e-07,
      "loss": 0.6922,
      "step": 30287
    },
    {
      "epoch": 0.8727021264334697,
      "grad_norm": 0.7450560927391052,
      "learning_rate": 4.1898542022063226e-07,
      "loss": 0.6776,
      "step": 30288
    },
    {
      "epoch": 0.872730939895119,
      "grad_norm": 0.7096720933914185,
      "learning_rate": 4.187984626638797e-07,
      "loss": 0.648,
      "step": 30289
    },
    {
      "epoch": 0.8727597533567683,
      "grad_norm": 0.7255080938339233,
      "learning_rate": 4.1861154500481285e-07,
      "loss": 0.6855,
      "step": 30290
    },
    {
      "epoch": 0.8727885668184175,
      "grad_norm": 0.7400928735733032,
      "learning_rate": 4.184246672450609e-07,
      "loss": 0.7006,
      "step": 30291
    },
    {
      "epoch": 0.8728173802800668,
      "grad_norm": 0.7206777930259705,
      "learning_rate": 4.182378293862488e-07,
      "loss": 0.7083,
      "step": 30292
    },
    {
      "epoch": 0.8728461937417161,
      "grad_norm": 0.7179794311523438,
      "learning_rate": 4.180510314300068e-07,
      "loss": 0.6586,
      "step": 30293
    },
    {
      "epoch": 0.8728750072033654,
      "grad_norm": 0.7182632684707642,
      "learning_rate": 4.178642733779581e-07,
      "loss": 0.6809,
      "step": 30294
    },
    {
      "epoch": 0.8729038206650147,
      "grad_norm": 0.7270194888114929,
      "learning_rate": 4.17677555231733e-07,
      "loss": 0.6813,
      "step": 30295
    },
    {
      "epoch": 0.872932634126664,
      "grad_norm": 0.7247700691223145,
      "learning_rate": 4.174908769929542e-07,
      "loss": 0.6832,
      "step": 30296
    },
    {
      "epoch": 0.8729614475883133,
      "grad_norm": 0.708267331123352,
      "learning_rate": 4.1730423866325033e-07,
      "loss": 0.6726,
      "step": 30297
    },
    {
      "epoch": 0.8729902610499626,
      "grad_norm": 0.697874128818512,
      "learning_rate": 4.171176402442445e-07,
      "loss": 0.6668,
      "step": 30298
    },
    {
      "epoch": 0.8730190745116119,
      "grad_norm": 0.7097170352935791,
      "learning_rate": 4.169310817375627e-07,
      "loss": 0.682,
      "step": 30299
    },
    {
      "epoch": 0.8730478879732612,
      "grad_norm": 0.7075203061103821,
      "learning_rate": 4.167445631448308e-07,
      "loss": 0.6696,
      "step": 30300
    },
    {
      "epoch": 0.8730767014349103,
      "grad_norm": 0.7197156548500061,
      "learning_rate": 4.165580844676709e-07,
      "loss": 0.6982,
      "step": 30301
    },
    {
      "epoch": 0.8731055148965596,
      "grad_norm": 0.7423592805862427,
      "learning_rate": 4.1637164570770884e-07,
      "loss": 0.6942,
      "step": 30302
    },
    {
      "epoch": 0.8731343283582089,
      "grad_norm": 0.7175207138061523,
      "learning_rate": 4.1618524686656837e-07,
      "loss": 0.6887,
      "step": 30303
    },
    {
      "epoch": 0.8731631418198582,
      "grad_norm": 0.7135689854621887,
      "learning_rate": 4.1599888794587153e-07,
      "loss": 0.6869,
      "step": 30304
    },
    {
      "epoch": 0.8731919552815075,
      "grad_norm": 0.7338343858718872,
      "learning_rate": 4.15812568947242e-07,
      "loss": 0.7127,
      "step": 30305
    },
    {
      "epoch": 0.8732207687431568,
      "grad_norm": 0.7011570930480957,
      "learning_rate": 4.1562628987230344e-07,
      "loss": 0.6632,
      "step": 30306
    },
    {
      "epoch": 0.8732495822048061,
      "grad_norm": 0.7168313264846802,
      "learning_rate": 4.154400507226763e-07,
      "loss": 0.6525,
      "step": 30307
    },
    {
      "epoch": 0.8732783956664554,
      "grad_norm": 0.7267645001411438,
      "learning_rate": 4.1525385149998376e-07,
      "loss": 0.6903,
      "step": 30308
    },
    {
      "epoch": 0.8733072091281047,
      "grad_norm": 0.71845942735672,
      "learning_rate": 4.150676922058472e-07,
      "loss": 0.6608,
      "step": 30309
    },
    {
      "epoch": 0.873336022589754,
      "grad_norm": 0.7281327247619629,
      "learning_rate": 4.1488157284188813e-07,
      "loss": 0.6846,
      "step": 30310
    },
    {
      "epoch": 0.8733648360514032,
      "grad_norm": 0.7072223424911499,
      "learning_rate": 4.146954934097264e-07,
      "loss": 0.6781,
      "step": 30311
    },
    {
      "epoch": 0.8733936495130525,
      "grad_norm": 0.6961061954498291,
      "learning_rate": 4.14509453910984e-07,
      "loss": 0.6639,
      "step": 30312
    },
    {
      "epoch": 0.8734224629747018,
      "grad_norm": 0.7193545699119568,
      "learning_rate": 4.143234543472785e-07,
      "loss": 0.6808,
      "step": 30313
    },
    {
      "epoch": 0.873451276436351,
      "grad_norm": 0.7347692251205444,
      "learning_rate": 4.141374947202337e-07,
      "loss": 0.7004,
      "step": 30314
    },
    {
      "epoch": 0.8734800898980003,
      "grad_norm": 0.7186688780784607,
      "learning_rate": 4.1395157503146655e-07,
      "loss": 0.6719,
      "step": 30315
    },
    {
      "epoch": 0.8735089033596496,
      "grad_norm": 0.7193655967712402,
      "learning_rate": 4.137656952825975e-07,
      "loss": 0.6675,
      "step": 30316
    },
    {
      "epoch": 0.8735377168212989,
      "grad_norm": 0.7416533827781677,
      "learning_rate": 4.135798554752435e-07,
      "loss": 0.6866,
      "step": 30317
    },
    {
      "epoch": 0.8735665302829482,
      "grad_norm": 0.7297376990318298,
      "learning_rate": 4.1339405561102555e-07,
      "loss": 0.6998,
      "step": 30318
    },
    {
      "epoch": 0.8735953437445975,
      "grad_norm": 0.72952800989151,
      "learning_rate": 4.13208295691559e-07,
      "loss": 0.668,
      "step": 30319
    },
    {
      "epoch": 0.8736241572062468,
      "grad_norm": 0.747791588306427,
      "learning_rate": 4.1302257571846473e-07,
      "loss": 0.6902,
      "step": 30320
    },
    {
      "epoch": 0.873652970667896,
      "grad_norm": 0.7415187358856201,
      "learning_rate": 4.128368956933571e-07,
      "loss": 0.6618,
      "step": 30321
    },
    {
      "epoch": 0.8736817841295453,
      "grad_norm": 0.7086197733879089,
      "learning_rate": 4.1265125561785466e-07,
      "loss": 0.689,
      "step": 30322
    },
    {
      "epoch": 0.8737105975911946,
      "grad_norm": 0.7231952548027039,
      "learning_rate": 4.124656554935752e-07,
      "loss": 0.6747,
      "step": 30323
    },
    {
      "epoch": 0.8737394110528439,
      "grad_norm": 0.733609676361084,
      "learning_rate": 4.122800953221329e-07,
      "loss": 0.6761,
      "step": 30324
    },
    {
      "epoch": 0.8737682245144932,
      "grad_norm": 0.720969021320343,
      "learning_rate": 4.120945751051453e-07,
      "loss": 0.6652,
      "step": 30325
    },
    {
      "epoch": 0.8737970379761425,
      "grad_norm": 0.7261911630630493,
      "learning_rate": 4.1190909484422724e-07,
      "loss": 0.6734,
      "step": 30326
    },
    {
      "epoch": 0.8738258514377918,
      "grad_norm": 0.7105846405029297,
      "learning_rate": 4.117236545409958e-07,
      "loss": 0.6665,
      "step": 30327
    },
    {
      "epoch": 0.873854664899441,
      "grad_norm": 0.703131377696991,
      "learning_rate": 4.115382541970636e-07,
      "loss": 0.6521,
      "step": 30328
    },
    {
      "epoch": 0.8738834783610903,
      "grad_norm": 0.7083281874656677,
      "learning_rate": 4.1135289381404763e-07,
      "loss": 0.6454,
      "step": 30329
    },
    {
      "epoch": 0.8739122918227396,
      "grad_norm": 0.7188546657562256,
      "learning_rate": 4.1116757339355997e-07,
      "loss": 0.6741,
      "step": 30330
    },
    {
      "epoch": 0.8739411052843888,
      "grad_norm": 0.7284568548202515,
      "learning_rate": 4.109822929372165e-07,
      "loss": 0.6742,
      "step": 30331
    },
    {
      "epoch": 0.8739699187460381,
      "grad_norm": 0.7506450414657593,
      "learning_rate": 4.107970524466287e-07,
      "loss": 0.6859,
      "step": 30332
    },
    {
      "epoch": 0.8739987322076874,
      "grad_norm": 0.7299387454986572,
      "learning_rate": 4.10611851923412e-07,
      "loss": 0.7032,
      "step": 30333
    },
    {
      "epoch": 0.8740275456693367,
      "grad_norm": 0.7257240414619446,
      "learning_rate": 4.104266913691779e-07,
      "loss": 0.7079,
      "step": 30334
    },
    {
      "epoch": 0.874056359130986,
      "grad_norm": 0.7460421323776245,
      "learning_rate": 4.102415707855401e-07,
      "loss": 0.7003,
      "step": 30335
    },
    {
      "epoch": 0.8740851725926353,
      "grad_norm": 0.7112410664558411,
      "learning_rate": 4.100564901741094e-07,
      "loss": 0.6862,
      "step": 30336
    },
    {
      "epoch": 0.8741139860542846,
      "grad_norm": 0.7303874492645264,
      "learning_rate": 4.0987144953649803e-07,
      "loss": 0.6821,
      "step": 30337
    },
    {
      "epoch": 0.8741427995159339,
      "grad_norm": 0.7175160646438599,
      "learning_rate": 4.09686448874318e-07,
      "loss": 0.6744,
      "step": 30338
    },
    {
      "epoch": 0.8741716129775832,
      "grad_norm": 0.7245371341705322,
      "learning_rate": 4.095014881891812e-07,
      "loss": 0.6468,
      "step": 30339
    },
    {
      "epoch": 0.8742004264392325,
      "grad_norm": 0.7082215547561646,
      "learning_rate": 4.093165674826971e-07,
      "loss": 0.6858,
      "step": 30340
    },
    {
      "epoch": 0.8742292399008816,
      "grad_norm": 0.7123685479164124,
      "learning_rate": 4.0913168675647706e-07,
      "loss": 0.6869,
      "step": 30341
    },
    {
      "epoch": 0.8742580533625309,
      "grad_norm": 0.7296392321586609,
      "learning_rate": 4.089468460121304e-07,
      "loss": 0.673,
      "step": 30342
    },
    {
      "epoch": 0.8742868668241802,
      "grad_norm": 0.737163245677948,
      "learning_rate": 4.0876204525126753e-07,
      "loss": 0.6591,
      "step": 30343
    },
    {
      "epoch": 0.8743156802858295,
      "grad_norm": 0.7473632097244263,
      "learning_rate": 4.0857728447549816e-07,
      "loss": 0.6966,
      "step": 30344
    },
    {
      "epoch": 0.8743444937474788,
      "grad_norm": 0.7408128380775452,
      "learning_rate": 4.0839256368643e-07,
      "loss": 0.7028,
      "step": 30345
    },
    {
      "epoch": 0.8743733072091281,
      "grad_norm": 0.7101820111274719,
      "learning_rate": 4.082078828856734e-07,
      "loss": 0.6715,
      "step": 30346
    },
    {
      "epoch": 0.8744021206707774,
      "grad_norm": 0.7323710322380066,
      "learning_rate": 4.0802324207483534e-07,
      "loss": 0.684,
      "step": 30347
    },
    {
      "epoch": 0.8744309341324267,
      "grad_norm": 0.7176252007484436,
      "learning_rate": 4.078386412555257e-07,
      "loss": 0.6665,
      "step": 30348
    },
    {
      "epoch": 0.874459747594076,
      "grad_norm": 0.7211955785751343,
      "learning_rate": 4.0765408042934883e-07,
      "loss": 0.6697,
      "step": 30349
    },
    {
      "epoch": 0.8744885610557253,
      "grad_norm": 0.7214037775993347,
      "learning_rate": 4.0746955959791666e-07,
      "loss": 0.6766,
      "step": 30350
    },
    {
      "epoch": 0.8745173745173745,
      "grad_norm": 0.7371060252189636,
      "learning_rate": 4.072850787628324e-07,
      "loss": 0.6766,
      "step": 30351
    },
    {
      "epoch": 0.8745461879790238,
      "grad_norm": 0.7177982926368713,
      "learning_rate": 4.0710063792570475e-07,
      "loss": 0.6866,
      "step": 30352
    },
    {
      "epoch": 0.8745750014406731,
      "grad_norm": 0.7136693000793457,
      "learning_rate": 4.0691623708813854e-07,
      "loss": 0.6703,
      "step": 30353
    },
    {
      "epoch": 0.8746038149023224,
      "grad_norm": 0.7175105810165405,
      "learning_rate": 4.067318762517419e-07,
      "loss": 0.6963,
      "step": 30354
    },
    {
      "epoch": 0.8746326283639716,
      "grad_norm": 0.7238593101501465,
      "learning_rate": 4.065475554181175e-07,
      "loss": 0.6806,
      "step": 30355
    },
    {
      "epoch": 0.8746614418256209,
      "grad_norm": 0.7152885794639587,
      "learning_rate": 4.0636327458887346e-07,
      "loss": 0.6776,
      "step": 30356
    },
    {
      "epoch": 0.8746902552872702,
      "grad_norm": 0.7384794354438782,
      "learning_rate": 4.061790337656124e-07,
      "loss": 0.6863,
      "step": 30357
    },
    {
      "epoch": 0.8747190687489195,
      "grad_norm": 0.739123523235321,
      "learning_rate": 4.0599483294993967e-07,
      "loss": 0.6863,
      "step": 30358
    },
    {
      "epoch": 0.8747478822105688,
      "grad_norm": 0.7052496075630188,
      "learning_rate": 4.058106721434607e-07,
      "loss": 0.6834,
      "step": 30359
    },
    {
      "epoch": 0.874776695672218,
      "grad_norm": 0.706335723400116,
      "learning_rate": 4.0562655134777695e-07,
      "loss": 0.6641,
      "step": 30360
    },
    {
      "epoch": 0.8748055091338673,
      "grad_norm": 0.7254658937454224,
      "learning_rate": 4.054424705644938e-07,
      "loss": 0.6923,
      "step": 30361
    },
    {
      "epoch": 0.8748343225955166,
      "grad_norm": 0.7129892110824585,
      "learning_rate": 4.0525842979521447e-07,
      "loss": 0.6893,
      "step": 30362
    },
    {
      "epoch": 0.8748631360571659,
      "grad_norm": 0.7451180815696716,
      "learning_rate": 4.050744290415398e-07,
      "loss": 0.6894,
      "step": 30363
    },
    {
      "epoch": 0.8748919495188152,
      "grad_norm": 0.7192075848579407,
      "learning_rate": 4.048904683050742e-07,
      "loss": 0.6614,
      "step": 30364
    },
    {
      "epoch": 0.8749207629804645,
      "grad_norm": 0.7272468209266663,
      "learning_rate": 4.047065475874201e-07,
      "loss": 0.681,
      "step": 30365
    },
    {
      "epoch": 0.8749495764421138,
      "grad_norm": 0.726587176322937,
      "learning_rate": 4.045226668901775e-07,
      "loss": 0.6946,
      "step": 30366
    },
    {
      "epoch": 0.8749783899037631,
      "grad_norm": 0.7298508286476135,
      "learning_rate": 4.043388262149489e-07,
      "loss": 0.6824,
      "step": 30367
    },
    {
      "epoch": 0.8750072033654123,
      "grad_norm": 0.7204355001449585,
      "learning_rate": 4.041550255633353e-07,
      "loss": 0.6648,
      "step": 30368
    },
    {
      "epoch": 0.8750360168270616,
      "grad_norm": 0.720961332321167,
      "learning_rate": 4.0397126493693704e-07,
      "loss": 0.6745,
      "step": 30369
    },
    {
      "epoch": 0.8750648302887108,
      "grad_norm": 0.6979802846908569,
      "learning_rate": 4.0378754433735456e-07,
      "loss": 0.6751,
      "step": 30370
    },
    {
      "epoch": 0.8750936437503601,
      "grad_norm": 0.7247056365013123,
      "learning_rate": 4.0360386376618876e-07,
      "loss": 0.6553,
      "step": 30371
    },
    {
      "epoch": 0.8751224572120094,
      "grad_norm": 0.7253915071487427,
      "learning_rate": 4.034202232250378e-07,
      "loss": 0.6886,
      "step": 30372
    },
    {
      "epoch": 0.8751512706736587,
      "grad_norm": 0.7194705009460449,
      "learning_rate": 4.0323662271550157e-07,
      "loss": 0.6664,
      "step": 30373
    },
    {
      "epoch": 0.875180084135308,
      "grad_norm": 0.7277602553367615,
      "learning_rate": 4.030530622391793e-07,
      "loss": 0.6871,
      "step": 30374
    },
    {
      "epoch": 0.8752088975969573,
      "grad_norm": 0.7142103314399719,
      "learning_rate": 4.028695417976708e-07,
      "loss": 0.6864,
      "step": 30375
    },
    {
      "epoch": 0.8752377110586066,
      "grad_norm": 0.694923460483551,
      "learning_rate": 4.0268606139257205e-07,
      "loss": 0.6904,
      "step": 30376
    },
    {
      "epoch": 0.8752665245202559,
      "grad_norm": 0.7490061521530151,
      "learning_rate": 4.0250262102548287e-07,
      "loss": 0.7087,
      "step": 30377
    },
    {
      "epoch": 0.8752953379819052,
      "grad_norm": 0.71002197265625,
      "learning_rate": 4.0231922069799924e-07,
      "loss": 0.6853,
      "step": 30378
    },
    {
      "epoch": 0.8753241514435545,
      "grad_norm": 0.6980174779891968,
      "learning_rate": 4.0213586041171984e-07,
      "loss": 0.6522,
      "step": 30379
    },
    {
      "epoch": 0.8753529649052038,
      "grad_norm": 0.6984471082687378,
      "learning_rate": 4.0195254016824057e-07,
      "loss": 0.6644,
      "step": 30380
    },
    {
      "epoch": 0.875381778366853,
      "grad_norm": 0.724933385848999,
      "learning_rate": 4.0176925996915804e-07,
      "loss": 0.6839,
      "step": 30381
    },
    {
      "epoch": 0.8754105918285022,
      "grad_norm": 0.7204779386520386,
      "learning_rate": 4.015860198160693e-07,
      "loss": 0.6586,
      "step": 30382
    },
    {
      "epoch": 0.8754394052901515,
      "grad_norm": 0.7184402942657471,
      "learning_rate": 4.0140281971056905e-07,
      "loss": 0.6679,
      "step": 30383
    },
    {
      "epoch": 0.8754682187518008,
      "grad_norm": 0.740093469619751,
      "learning_rate": 4.0121965965425394e-07,
      "loss": 0.705,
      "step": 30384
    },
    {
      "epoch": 0.8754970322134501,
      "grad_norm": 0.7061135172843933,
      "learning_rate": 4.01036539648717e-07,
      "loss": 0.6675,
      "step": 30385
    },
    {
      "epoch": 0.8755258456750994,
      "grad_norm": 0.7243744730949402,
      "learning_rate": 4.0085345969555656e-07,
      "loss": 0.6586,
      "step": 30386
    },
    {
      "epoch": 0.8755546591367487,
      "grad_norm": 0.7188681960105896,
      "learning_rate": 4.006704197963635e-07,
      "loss": 0.6843,
      "step": 30387
    },
    {
      "epoch": 0.875583472598398,
      "grad_norm": 0.7116885185241699,
      "learning_rate": 4.004874199527348e-07,
      "loss": 0.6719,
      "step": 30388
    },
    {
      "epoch": 0.8756122860600473,
      "grad_norm": 0.7420713305473328,
      "learning_rate": 4.0030446016626143e-07,
      "loss": 0.69,
      "step": 30389
    },
    {
      "epoch": 0.8756410995216966,
      "grad_norm": 0.7016774415969849,
      "learning_rate": 4.0012154043853944e-07,
      "loss": 0.6616,
      "step": 30390
    },
    {
      "epoch": 0.8756699129833458,
      "grad_norm": 0.7161913514137268,
      "learning_rate": 3.9993866077115974e-07,
      "loss": 0.6893,
      "step": 30391
    },
    {
      "epoch": 0.8756987264449951,
      "grad_norm": 0.7327117919921875,
      "learning_rate": 3.997558211657171e-07,
      "loss": 0.6884,
      "step": 30392
    },
    {
      "epoch": 0.8757275399066444,
      "grad_norm": 0.7125160098075867,
      "learning_rate": 3.995730216238014e-07,
      "loss": 0.689,
      "step": 30393
    },
    {
      "epoch": 0.8757563533682937,
      "grad_norm": 0.7550694942474365,
      "learning_rate": 3.9939026214700694e-07,
      "loss": 0.6655,
      "step": 30394
    },
    {
      "epoch": 0.8757851668299429,
      "grad_norm": 0.7135712504386902,
      "learning_rate": 3.9920754273692355e-07,
      "loss": 0.6783,
      "step": 30395
    },
    {
      "epoch": 0.8758139802915922,
      "grad_norm": 0.7426183223724365,
      "learning_rate": 3.9902486339514325e-07,
      "loss": 0.6722,
      "step": 30396
    },
    {
      "epoch": 0.8758427937532415,
      "grad_norm": 0.7461870312690735,
      "learning_rate": 3.98842224123257e-07,
      "loss": 0.6975,
      "step": 30397
    },
    {
      "epoch": 0.8758716072148908,
      "grad_norm": 0.7206817269325256,
      "learning_rate": 3.986596249228564e-07,
      "loss": 0.6793,
      "step": 30398
    },
    {
      "epoch": 0.8759004206765401,
      "grad_norm": 0.7284789085388184,
      "learning_rate": 3.9847706579553e-07,
      "loss": 0.6786,
      "step": 30399
    },
    {
      "epoch": 0.8759292341381893,
      "grad_norm": 0.7057671546936035,
      "learning_rate": 3.9829454674286947e-07,
      "loss": 0.6802,
      "step": 30400
    },
    {
      "epoch": 0.8759580475998386,
      "grad_norm": 0.7434677481651306,
      "learning_rate": 3.9811206776646295e-07,
      "loss": 0.6905,
      "step": 30401
    },
    {
      "epoch": 0.8759868610614879,
      "grad_norm": 0.7266209721565247,
      "learning_rate": 3.9792962886789965e-07,
      "loss": 0.6615,
      "step": 30402
    },
    {
      "epoch": 0.8760156745231372,
      "grad_norm": 0.7399151921272278,
      "learning_rate": 3.977472300487695e-07,
      "loss": 0.6775,
      "step": 30403
    },
    {
      "epoch": 0.8760444879847865,
      "grad_norm": 0.7403619289398193,
      "learning_rate": 3.9756487131066004e-07,
      "loss": 0.7046,
      "step": 30404
    },
    {
      "epoch": 0.8760733014464358,
      "grad_norm": 0.7263587713241577,
      "learning_rate": 3.973825526551606e-07,
      "loss": 0.691,
      "step": 30405
    },
    {
      "epoch": 0.8761021149080851,
      "grad_norm": 0.7219063639640808,
      "learning_rate": 3.972002740838571e-07,
      "loss": 0.6949,
      "step": 30406
    },
    {
      "epoch": 0.8761309283697344,
      "grad_norm": 0.7145798206329346,
      "learning_rate": 3.9701803559833884e-07,
      "loss": 0.6529,
      "step": 30407
    },
    {
      "epoch": 0.8761597418313837,
      "grad_norm": 0.7436702847480774,
      "learning_rate": 3.9683583720019183e-07,
      "loss": 0.6881,
      "step": 30408
    },
    {
      "epoch": 0.8761885552930329,
      "grad_norm": 0.7034109234809875,
      "learning_rate": 3.966536788910025e-07,
      "loss": 0.6564,
      "step": 30409
    },
    {
      "epoch": 0.8762173687546821,
      "grad_norm": 0.7352377772331238,
      "learning_rate": 3.964715606723585e-07,
      "loss": 0.6938,
      "step": 30410
    },
    {
      "epoch": 0.8762461822163314,
      "grad_norm": 0.7247573733329773,
      "learning_rate": 3.9628948254584633e-07,
      "loss": 0.6642,
      "step": 30411
    },
    {
      "epoch": 0.8762749956779807,
      "grad_norm": 0.7233778238296509,
      "learning_rate": 3.961074445130492e-07,
      "loss": 0.7115,
      "step": 30412
    },
    {
      "epoch": 0.87630380913963,
      "grad_norm": 0.7464620471000671,
      "learning_rate": 3.9592544657555573e-07,
      "loss": 0.6764,
      "step": 30413
    },
    {
      "epoch": 0.8763326226012793,
      "grad_norm": 0.7304627895355225,
      "learning_rate": 3.957434887349476e-07,
      "loss": 0.6819,
      "step": 30414
    },
    {
      "epoch": 0.8763614360629286,
      "grad_norm": 0.7373536229133606,
      "learning_rate": 3.9556157099281235e-07,
      "loss": 0.675,
      "step": 30415
    },
    {
      "epoch": 0.8763902495245779,
      "grad_norm": 0.734571099281311,
      "learning_rate": 3.9537969335073256e-07,
      "loss": 0.7084,
      "step": 30416
    },
    {
      "epoch": 0.8764190629862272,
      "grad_norm": 0.7142137885093689,
      "learning_rate": 3.9519785581029315e-07,
      "loss": 0.6795,
      "step": 30417
    },
    {
      "epoch": 0.8764478764478765,
      "grad_norm": 0.7268968224525452,
      "learning_rate": 3.9501605837307613e-07,
      "loss": 0.7061,
      "step": 30418
    },
    {
      "epoch": 0.8764766899095258,
      "grad_norm": 0.7147169709205627,
      "learning_rate": 3.9483430104066635e-07,
      "loss": 0.6802,
      "step": 30419
    },
    {
      "epoch": 0.876505503371175,
      "grad_norm": 0.7047631740570068,
      "learning_rate": 3.9465258381464755e-07,
      "loss": 0.6676,
      "step": 30420
    },
    {
      "epoch": 0.8765343168328243,
      "grad_norm": 0.7527493834495544,
      "learning_rate": 3.944709066965996e-07,
      "loss": 0.689,
      "step": 30421
    },
    {
      "epoch": 0.8765631302944735,
      "grad_norm": 0.7105268836021423,
      "learning_rate": 3.9428926968810623e-07,
      "loss": 0.6758,
      "step": 30422
    },
    {
      "epoch": 0.8765919437561228,
      "grad_norm": 0.7045824527740479,
      "learning_rate": 3.9410767279074946e-07,
      "loss": 0.6773,
      "step": 30423
    },
    {
      "epoch": 0.8766207572177721,
      "grad_norm": 0.7342231869697571,
      "learning_rate": 3.9392611600611195e-07,
      "loss": 0.6508,
      "step": 30424
    },
    {
      "epoch": 0.8766495706794214,
      "grad_norm": 0.7327242493629456,
      "learning_rate": 3.937445993357719e-07,
      "loss": 0.6867,
      "step": 30425
    },
    {
      "epoch": 0.8766783841410707,
      "grad_norm": 0.7138197422027588,
      "learning_rate": 3.93563122781313e-07,
      "loss": 0.6707,
      "step": 30426
    },
    {
      "epoch": 0.87670719760272,
      "grad_norm": 0.7153539657592773,
      "learning_rate": 3.9338168634431404e-07,
      "loss": 0.6872,
      "step": 30427
    },
    {
      "epoch": 0.8767360110643693,
      "grad_norm": 0.7275574803352356,
      "learning_rate": 3.9320029002635595e-07,
      "loss": 0.6553,
      "step": 30428
    },
    {
      "epoch": 0.8767648245260186,
      "grad_norm": 0.7342010140419006,
      "learning_rate": 3.93018933829018e-07,
      "loss": 0.6904,
      "step": 30429
    },
    {
      "epoch": 0.8767936379876679,
      "grad_norm": 0.742367684841156,
      "learning_rate": 3.928376177538801e-07,
      "loss": 0.708,
      "step": 30430
    },
    {
      "epoch": 0.8768224514493171,
      "grad_norm": 0.7127272486686707,
      "learning_rate": 3.926563418025209e-07,
      "loss": 0.6793,
      "step": 30431
    },
    {
      "epoch": 0.8768512649109664,
      "grad_norm": 0.7494022846221924,
      "learning_rate": 3.9247510597651973e-07,
      "loss": 0.6983,
      "step": 30432
    },
    {
      "epoch": 0.8768800783726157,
      "grad_norm": 0.7172880172729492,
      "learning_rate": 3.9229391027745255e-07,
      "loss": 0.6661,
      "step": 30433
    },
    {
      "epoch": 0.876908891834265,
      "grad_norm": 0.7162240147590637,
      "learning_rate": 3.921127547069015e-07,
      "loss": 0.6823,
      "step": 30434
    },
    {
      "epoch": 0.8769377052959143,
      "grad_norm": 0.7255019545555115,
      "learning_rate": 3.919316392664413e-07,
      "loss": 0.6634,
      "step": 30435
    },
    {
      "epoch": 0.8769665187575635,
      "grad_norm": 0.7304233908653259,
      "learning_rate": 3.9175056395765077e-07,
      "loss": 0.7045,
      "step": 30436
    },
    {
      "epoch": 0.8769953322192128,
      "grad_norm": 0.7602471113204956,
      "learning_rate": 3.915695287821053e-07,
      "loss": 0.7078,
      "step": 30437
    },
    {
      "epoch": 0.8770241456808621,
      "grad_norm": 0.7113143801689148,
      "learning_rate": 3.913885337413836e-07,
      "loss": 0.6847,
      "step": 30438
    },
    {
      "epoch": 0.8770529591425114,
      "grad_norm": 0.7426790595054626,
      "learning_rate": 3.9120757883706006e-07,
      "loss": 0.709,
      "step": 30439
    },
    {
      "epoch": 0.8770817726041606,
      "grad_norm": 0.7295477986335754,
      "learning_rate": 3.910266640707117e-07,
      "loss": 0.6971,
      "step": 30440
    },
    {
      "epoch": 0.8771105860658099,
      "grad_norm": 0.7348253130912781,
      "learning_rate": 3.9084578944391383e-07,
      "loss": 0.692,
      "step": 30441
    },
    {
      "epoch": 0.8771393995274592,
      "grad_norm": 0.7295101881027222,
      "learning_rate": 3.9066495495824144e-07,
      "loss": 0.6746,
      "step": 30442
    },
    {
      "epoch": 0.8771682129891085,
      "grad_norm": 0.7127012014389038,
      "learning_rate": 3.904841606152704e-07,
      "loss": 0.6466,
      "step": 30443
    },
    {
      "epoch": 0.8771970264507578,
      "grad_norm": 0.7113691568374634,
      "learning_rate": 3.9030340641657336e-07,
      "loss": 0.6753,
      "step": 30444
    },
    {
      "epoch": 0.8772258399124071,
      "grad_norm": 0.7139072418212891,
      "learning_rate": 3.901226923637258e-07,
      "loss": 0.6677,
      "step": 30445
    },
    {
      "epoch": 0.8772546533740564,
      "grad_norm": 0.7176388502120972,
      "learning_rate": 3.899420184583014e-07,
      "loss": 0.6679,
      "step": 30446
    },
    {
      "epoch": 0.8772834668357057,
      "grad_norm": 0.7105616927146912,
      "learning_rate": 3.8976138470187496e-07,
      "loss": 0.6999,
      "step": 30447
    },
    {
      "epoch": 0.877312280297355,
      "grad_norm": 0.6924141645431519,
      "learning_rate": 3.89580791096017e-07,
      "loss": 0.6743,
      "step": 30448
    },
    {
      "epoch": 0.8773410937590042,
      "grad_norm": 0.7186463475227356,
      "learning_rate": 3.8940023764230227e-07,
      "loss": 0.6965,
      "step": 30449
    },
    {
      "epoch": 0.8773699072206534,
      "grad_norm": 0.7117940783500671,
      "learning_rate": 3.8921972434230183e-07,
      "loss": 0.7085,
      "step": 30450
    },
    {
      "epoch": 0.8773987206823027,
      "grad_norm": 0.7197250127792358,
      "learning_rate": 3.8903925119758937e-07,
      "loss": 0.6822,
      "step": 30451
    },
    {
      "epoch": 0.877427534143952,
      "grad_norm": 0.6922141909599304,
      "learning_rate": 3.8885881820973526e-07,
      "loss": 0.6445,
      "step": 30452
    },
    {
      "epoch": 0.8774563476056013,
      "grad_norm": 0.7311186194419861,
      "learning_rate": 3.8867842538031167e-07,
      "loss": 0.6971,
      "step": 30453
    },
    {
      "epoch": 0.8774851610672506,
      "grad_norm": 0.7501988410949707,
      "learning_rate": 3.8849807271088893e-07,
      "loss": 0.7162,
      "step": 30454
    },
    {
      "epoch": 0.8775139745288999,
      "grad_norm": 0.7085630297660828,
      "learning_rate": 3.8831776020303914e-07,
      "loss": 0.6573,
      "step": 30455
    },
    {
      "epoch": 0.8775427879905492,
      "grad_norm": 0.7233691811561584,
      "learning_rate": 3.881374878583305e-07,
      "loss": 0.6759,
      "step": 30456
    },
    {
      "epoch": 0.8775716014521985,
      "grad_norm": 0.7475682497024536,
      "learning_rate": 3.87957255678334e-07,
      "loss": 0.6758,
      "step": 30457
    },
    {
      "epoch": 0.8776004149138478,
      "grad_norm": 0.7504381537437439,
      "learning_rate": 3.8777706366462e-07,
      "loss": 0.6797,
      "step": 30458
    },
    {
      "epoch": 0.8776292283754971,
      "grad_norm": 0.7259917259216309,
      "learning_rate": 3.875969118187578e-07,
      "loss": 0.6602,
      "step": 30459
    },
    {
      "epoch": 0.8776580418371464,
      "grad_norm": 0.7517324686050415,
      "learning_rate": 3.8741680014231455e-07,
      "loss": 0.6961,
      "step": 30460
    },
    {
      "epoch": 0.8776868552987956,
      "grad_norm": 0.7226887345314026,
      "learning_rate": 3.872367286368606e-07,
      "loss": 0.6694,
      "step": 30461
    },
    {
      "epoch": 0.8777156687604449,
      "grad_norm": 0.719680666923523,
      "learning_rate": 3.8705669730396413e-07,
      "loss": 0.6894,
      "step": 30462
    },
    {
      "epoch": 0.8777444822220941,
      "grad_norm": 0.7517172694206238,
      "learning_rate": 3.8687670614519224e-07,
      "loss": 0.6969,
      "step": 30463
    },
    {
      "epoch": 0.8777732956837434,
      "grad_norm": 0.7217716574668884,
      "learning_rate": 3.8669675516211313e-07,
      "loss": 0.6797,
      "step": 30464
    },
    {
      "epoch": 0.8778021091453927,
      "grad_norm": 0.7057814598083496,
      "learning_rate": 3.8651684435629333e-07,
      "loss": 0.6864,
      "step": 30465
    },
    {
      "epoch": 0.877830922607042,
      "grad_norm": 0.7421597242355347,
      "learning_rate": 3.8633697372930043e-07,
      "loss": 0.6468,
      "step": 30466
    },
    {
      "epoch": 0.8778597360686913,
      "grad_norm": 0.7175578474998474,
      "learning_rate": 3.861571432826999e-07,
      "loss": 0.6799,
      "step": 30467
    },
    {
      "epoch": 0.8778885495303406,
      "grad_norm": 0.7195860147476196,
      "learning_rate": 3.859773530180594e-07,
      "loss": 0.6747,
      "step": 30468
    },
    {
      "epoch": 0.8779173629919899,
      "grad_norm": 0.7314512133598328,
      "learning_rate": 3.857976029369415e-07,
      "loss": 0.6695,
      "step": 30469
    },
    {
      "epoch": 0.8779461764536391,
      "grad_norm": 0.7333685159683228,
      "learning_rate": 3.856178930409166e-07,
      "loss": 0.6962,
      "step": 30470
    },
    {
      "epoch": 0.8779749899152884,
      "grad_norm": 0.7354332804679871,
      "learning_rate": 3.854382233315457e-07,
      "loss": 0.7227,
      "step": 30471
    },
    {
      "epoch": 0.8780038033769377,
      "grad_norm": 0.7295780181884766,
      "learning_rate": 3.852585938103959e-07,
      "loss": 0.6855,
      "step": 30472
    },
    {
      "epoch": 0.878032616838587,
      "grad_norm": 0.7409788370132446,
      "learning_rate": 3.850790044790298e-07,
      "loss": 0.71,
      "step": 30473
    },
    {
      "epoch": 0.8780614303002363,
      "grad_norm": 0.7224993705749512,
      "learning_rate": 3.848994553390134e-07,
      "loss": 0.7042,
      "step": 30474
    },
    {
      "epoch": 0.8780902437618856,
      "grad_norm": 0.7095020413398743,
      "learning_rate": 3.847199463919088e-07,
      "loss": 0.6902,
      "step": 30475
    },
    {
      "epoch": 0.8781190572235348,
      "grad_norm": 0.7308587431907654,
      "learning_rate": 3.845404776392803e-07,
      "loss": 0.6996,
      "step": 30476
    },
    {
      "epoch": 0.8781478706851841,
      "grad_norm": 0.7047302722930908,
      "learning_rate": 3.8436104908268934e-07,
      "loss": 0.6758,
      "step": 30477
    },
    {
      "epoch": 0.8781766841468334,
      "grad_norm": 0.7122935056686401,
      "learning_rate": 3.841816607237003e-07,
      "loss": 0.6734,
      "step": 30478
    },
    {
      "epoch": 0.8782054976084827,
      "grad_norm": 0.713172972202301,
      "learning_rate": 3.840023125638753e-07,
      "loss": 0.6703,
      "step": 30479
    },
    {
      "epoch": 0.878234311070132,
      "grad_norm": 0.7072880864143372,
      "learning_rate": 3.838230046047753e-07,
      "loss": 0.6599,
      "step": 30480
    },
    {
      "epoch": 0.8782631245317812,
      "grad_norm": 0.7210471034049988,
      "learning_rate": 3.836437368479623e-07,
      "loss": 0.6834,
      "step": 30481
    },
    {
      "epoch": 0.8782919379934305,
      "grad_norm": 0.7179501056671143,
      "learning_rate": 3.8346450929499735e-07,
      "loss": 0.6758,
      "step": 30482
    },
    {
      "epoch": 0.8783207514550798,
      "grad_norm": 0.703864574432373,
      "learning_rate": 3.8328532194744305e-07,
      "loss": 0.671,
      "step": 30483
    },
    {
      "epoch": 0.8783495649167291,
      "grad_norm": 0.7176491022109985,
      "learning_rate": 3.8310617480685763e-07,
      "loss": 0.6729,
      "step": 30484
    },
    {
      "epoch": 0.8783783783783784,
      "grad_norm": 0.7053967714309692,
      "learning_rate": 3.829270678748026e-07,
      "loss": 0.691,
      "step": 30485
    },
    {
      "epoch": 0.8784071918400277,
      "grad_norm": 0.7075200080871582,
      "learning_rate": 3.8274800115283726e-07,
      "loss": 0.65,
      "step": 30486
    },
    {
      "epoch": 0.878436005301677,
      "grad_norm": 0.7264952659606934,
      "learning_rate": 3.825689746425215e-07,
      "loss": 0.6841,
      "step": 30487
    },
    {
      "epoch": 0.8784648187633263,
      "grad_norm": 0.7129179835319519,
      "learning_rate": 3.823899883454135e-07,
      "loss": 0.6847,
      "step": 30488
    },
    {
      "epoch": 0.8784936322249756,
      "grad_norm": 0.716037929058075,
      "learning_rate": 3.8221104226307425e-07,
      "loss": 0.6489,
      "step": 30489
    },
    {
      "epoch": 0.8785224456866247,
      "grad_norm": 0.7193678021430969,
      "learning_rate": 3.8203213639705914e-07,
      "loss": 0.6901,
      "step": 30490
    },
    {
      "epoch": 0.878551259148274,
      "grad_norm": 0.7102203369140625,
      "learning_rate": 3.818532707489292e-07,
      "loss": 0.6735,
      "step": 30491
    },
    {
      "epoch": 0.8785800726099233,
      "grad_norm": 0.7152016758918762,
      "learning_rate": 3.816744453202398e-07,
      "loss": 0.6506,
      "step": 30492
    },
    {
      "epoch": 0.8786088860715726,
      "grad_norm": 0.7201405763626099,
      "learning_rate": 3.8149566011254913e-07,
      "loss": 0.6512,
      "step": 30493
    },
    {
      "epoch": 0.8786376995332219,
      "grad_norm": 0.7168666124343872,
      "learning_rate": 3.813169151274143e-07,
      "loss": 0.7054,
      "step": 30494
    },
    {
      "epoch": 0.8786665129948712,
      "grad_norm": 0.7097415328025818,
      "learning_rate": 3.8113821036639353e-07,
      "loss": 0.6729,
      "step": 30495
    },
    {
      "epoch": 0.8786953264565205,
      "grad_norm": 0.7092580795288086,
      "learning_rate": 3.809595458310406e-07,
      "loss": 0.6696,
      "step": 30496
    },
    {
      "epoch": 0.8787241399181698,
      "grad_norm": 0.7201431393623352,
      "learning_rate": 3.8078092152291357e-07,
      "loss": 0.7112,
      "step": 30497
    },
    {
      "epoch": 0.8787529533798191,
      "grad_norm": 0.7114318609237671,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.6598,
      "step": 30498
    },
    {
      "epoch": 0.8787817668414684,
      "grad_norm": 0.7004082798957825,
      "learning_rate": 3.804237935945554e-07,
      "loss": 0.6799,
      "step": 30499
    },
    {
      "epoch": 0.8788105803031176,
      "grad_norm": 0.7407037615776062,
      "learning_rate": 3.802452899774356e-07,
      "loss": 0.7018,
      "step": 30500
    },
    {
      "epoch": 0.8788393937647669,
      "grad_norm": 0.7110956311225891,
      "learning_rate": 3.800668265937607e-07,
      "loss": 0.6712,
      "step": 30501
    },
    {
      "epoch": 0.8788682072264162,
      "grad_norm": 0.7333454489707947,
      "learning_rate": 3.7988840344508614e-07,
      "loss": 0.6738,
      "step": 30502
    },
    {
      "epoch": 0.8788970206880654,
      "grad_norm": 0.7327145934104919,
      "learning_rate": 3.797100205329646e-07,
      "loss": 0.6576,
      "step": 30503
    },
    {
      "epoch": 0.8789258341497147,
      "grad_norm": 0.7570748925209045,
      "learning_rate": 3.795316778589503e-07,
      "loss": 0.6981,
      "step": 30504
    },
    {
      "epoch": 0.878954647611364,
      "grad_norm": 0.7599684596061707,
      "learning_rate": 3.7935337542459594e-07,
      "loss": 0.706,
      "step": 30505
    },
    {
      "epoch": 0.8789834610730133,
      "grad_norm": 0.7573548555374146,
      "learning_rate": 3.791751132314558e-07,
      "loss": 0.6738,
      "step": 30506
    },
    {
      "epoch": 0.8790122745346626,
      "grad_norm": 0.7204480767250061,
      "learning_rate": 3.789968912810804e-07,
      "loss": 0.6823,
      "step": 30507
    },
    {
      "epoch": 0.8790410879963119,
      "grad_norm": 0.7283498644828796,
      "learning_rate": 3.7881870957502343e-07,
      "loss": 0.6834,
      "step": 30508
    },
    {
      "epoch": 0.8790699014579612,
      "grad_norm": 0.7355622053146362,
      "learning_rate": 3.786405681148353e-07,
      "loss": 0.6881,
      "step": 30509
    },
    {
      "epoch": 0.8790987149196104,
      "grad_norm": 0.7142351865768433,
      "learning_rate": 3.7846246690206924e-07,
      "loss": 0.6654,
      "step": 30510
    },
    {
      "epoch": 0.8791275283812597,
      "grad_norm": 0.7244934439659119,
      "learning_rate": 3.78284405938274e-07,
      "loss": 0.6842,
      "step": 30511
    },
    {
      "epoch": 0.879156341842909,
      "grad_norm": 0.7289774417877197,
      "learning_rate": 3.7810638522500277e-07,
      "loss": 0.6961,
      "step": 30512
    },
    {
      "epoch": 0.8791851553045583,
      "grad_norm": 0.7087377905845642,
      "learning_rate": 3.779284047638032e-07,
      "loss": 0.6435,
      "step": 30513
    },
    {
      "epoch": 0.8792139687662076,
      "grad_norm": 0.7199524641036987,
      "learning_rate": 3.7775046455622853e-07,
      "loss": 0.6763,
      "step": 30514
    },
    {
      "epoch": 0.8792427822278569,
      "grad_norm": 0.7079601883888245,
      "learning_rate": 3.7757256460382527e-07,
      "loss": 0.6817,
      "step": 30515
    },
    {
      "epoch": 0.8792715956895062,
      "grad_norm": 0.7278611660003662,
      "learning_rate": 3.7739470490814445e-07,
      "loss": 0.679,
      "step": 30516
    },
    {
      "epoch": 0.8793004091511554,
      "grad_norm": 0.708221435546875,
      "learning_rate": 3.772168854707353e-07,
      "loss": 0.6809,
      "step": 30517
    },
    {
      "epoch": 0.8793292226128047,
      "grad_norm": 0.7329103350639343,
      "learning_rate": 3.770391062931461e-07,
      "loss": 0.6906,
      "step": 30518
    },
    {
      "epoch": 0.879358036074454,
      "grad_norm": 0.716860294342041,
      "learning_rate": 3.7686136737692446e-07,
      "loss": 0.6596,
      "step": 30519
    },
    {
      "epoch": 0.8793868495361032,
      "grad_norm": 0.7597591280937195,
      "learning_rate": 3.7668366872361916e-07,
      "loss": 0.6849,
      "step": 30520
    },
    {
      "epoch": 0.8794156629977525,
      "grad_norm": 0.7122878432273865,
      "learning_rate": 3.7650601033477787e-07,
      "loss": 0.6713,
      "step": 30521
    },
    {
      "epoch": 0.8794444764594018,
      "grad_norm": 0.7246335744857788,
      "learning_rate": 3.7632839221194704e-07,
      "loss": 0.696,
      "step": 30522
    },
    {
      "epoch": 0.8794732899210511,
      "grad_norm": 0.7400144934654236,
      "learning_rate": 3.7615081435667443e-07,
      "loss": 0.6925,
      "step": 30523
    },
    {
      "epoch": 0.8795021033827004,
      "grad_norm": 0.715798556804657,
      "learning_rate": 3.7597327677050545e-07,
      "loss": 0.6891,
      "step": 30524
    },
    {
      "epoch": 0.8795309168443497,
      "grad_norm": 0.7029371857643127,
      "learning_rate": 3.7579577945498767e-07,
      "loss": 0.6504,
      "step": 30525
    },
    {
      "epoch": 0.879559730305999,
      "grad_norm": 0.7161627411842346,
      "learning_rate": 3.7561832241166497e-07,
      "loss": 0.687,
      "step": 30526
    },
    {
      "epoch": 0.8795885437676483,
      "grad_norm": 0.7256885170936584,
      "learning_rate": 3.754409056420849e-07,
      "loss": 0.69,
      "step": 30527
    },
    {
      "epoch": 0.8796173572292976,
      "grad_norm": 0.7341511845588684,
      "learning_rate": 3.752635291477902e-07,
      "loss": 0.7013,
      "step": 30528
    },
    {
      "epoch": 0.8796461706909469,
      "grad_norm": 0.7314020991325378,
      "learning_rate": 3.75086192930329e-07,
      "loss": 0.688,
      "step": 30529
    },
    {
      "epoch": 0.879674984152596,
      "grad_norm": 0.7399877905845642,
      "learning_rate": 3.749088969912429e-07,
      "loss": 0.7002,
      "step": 30530
    },
    {
      "epoch": 0.8797037976142453,
      "grad_norm": 0.7344232201576233,
      "learning_rate": 3.747316413320773e-07,
      "loss": 0.6996,
      "step": 30531
    },
    {
      "epoch": 0.8797326110758946,
      "grad_norm": 0.7242488265037537,
      "learning_rate": 3.745544259543754e-07,
      "loss": 0.6883,
      "step": 30532
    },
    {
      "epoch": 0.8797614245375439,
      "grad_norm": 0.7441818118095398,
      "learning_rate": 3.7437725085968103e-07,
      "loss": 0.6851,
      "step": 30533
    },
    {
      "epoch": 0.8797902379991932,
      "grad_norm": 0.7265922427177429,
      "learning_rate": 3.7420011604953623e-07,
      "loss": 0.677,
      "step": 30534
    },
    {
      "epoch": 0.8798190514608425,
      "grad_norm": 0.7355857491493225,
      "learning_rate": 3.7402302152548475e-07,
      "loss": 0.6695,
      "step": 30535
    },
    {
      "epoch": 0.8798478649224918,
      "grad_norm": 0.7408021688461304,
      "learning_rate": 3.738459672890682e-07,
      "loss": 0.6906,
      "step": 30536
    },
    {
      "epoch": 0.8798766783841411,
      "grad_norm": 0.7609587907791138,
      "learning_rate": 3.736689533418286e-07,
      "loss": 0.7108,
      "step": 30537
    },
    {
      "epoch": 0.8799054918457904,
      "grad_norm": 0.7253323197364807,
      "learning_rate": 3.734919796853087e-07,
      "loss": 0.6482,
      "step": 30538
    },
    {
      "epoch": 0.8799343053074397,
      "grad_norm": 0.7199457287788391,
      "learning_rate": 3.7331504632104776e-07,
      "loss": 0.6932,
      "step": 30539
    },
    {
      "epoch": 0.879963118769089,
      "grad_norm": 0.6984308958053589,
      "learning_rate": 3.731381532505879e-07,
      "loss": 0.6713,
      "step": 30540
    },
    {
      "epoch": 0.8799919322307382,
      "grad_norm": 0.7208633422851562,
      "learning_rate": 3.7296130047546953e-07,
      "loss": 0.6864,
      "step": 30541
    },
    {
      "epoch": 0.8800207456923875,
      "grad_norm": 0.7334895133972168,
      "learning_rate": 3.7278448799723365e-07,
      "loss": 0.6917,
      "step": 30542
    },
    {
      "epoch": 0.8800495591540368,
      "grad_norm": 0.7091693878173828,
      "learning_rate": 3.7260771581741906e-07,
      "loss": 0.6875,
      "step": 30543
    },
    {
      "epoch": 0.880078372615686,
      "grad_norm": 0.7215579748153687,
      "learning_rate": 3.724309839375656e-07,
      "loss": 0.7041,
      "step": 30544
    },
    {
      "epoch": 0.8801071860773353,
      "grad_norm": 0.7366165518760681,
      "learning_rate": 3.722542923592121e-07,
      "loss": 0.6875,
      "step": 30545
    },
    {
      "epoch": 0.8801359995389846,
      "grad_norm": 0.7196812629699707,
      "learning_rate": 3.7207764108389835e-07,
      "loss": 0.7105,
      "step": 30546
    },
    {
      "epoch": 0.8801648130006339,
      "grad_norm": 0.7117120623588562,
      "learning_rate": 3.719010301131609e-07,
      "loss": 0.6613,
      "step": 30547
    },
    {
      "epoch": 0.8801936264622832,
      "grad_norm": 0.7201330661773682,
      "learning_rate": 3.717244594485403e-07,
      "loss": 0.6695,
      "step": 30548
    },
    {
      "epoch": 0.8802224399239325,
      "grad_norm": 0.7474265098571777,
      "learning_rate": 3.7154792909157243e-07,
      "loss": 0.6825,
      "step": 30549
    },
    {
      "epoch": 0.8802512533855817,
      "grad_norm": 0.7201318144798279,
      "learning_rate": 3.7137143904379616e-07,
      "loss": 0.6553,
      "step": 30550
    },
    {
      "epoch": 0.880280066847231,
      "grad_norm": 0.7427000403404236,
      "learning_rate": 3.7119498930674623e-07,
      "loss": 0.6688,
      "step": 30551
    },
    {
      "epoch": 0.8803088803088803,
      "grad_norm": 0.7236061096191406,
      "learning_rate": 3.7101857988196155e-07,
      "loss": 0.6854,
      "step": 30552
    },
    {
      "epoch": 0.8803376937705296,
      "grad_norm": 0.7074806094169617,
      "learning_rate": 3.7084221077097746e-07,
      "loss": 0.6856,
      "step": 30553
    },
    {
      "epoch": 0.8803665072321789,
      "grad_norm": 0.702887237071991,
      "learning_rate": 3.7066588197533113e-07,
      "loss": 0.6501,
      "step": 30554
    },
    {
      "epoch": 0.8803953206938282,
      "grad_norm": 0.7001146078109741,
      "learning_rate": 3.704895934965563e-07,
      "loss": 0.657,
      "step": 30555
    },
    {
      "epoch": 0.8804241341554775,
      "grad_norm": 0.7350071668624878,
      "learning_rate": 3.7031334533619066e-07,
      "loss": 0.6925,
      "step": 30556
    },
    {
      "epoch": 0.8804529476171267,
      "grad_norm": 0.7170410752296448,
      "learning_rate": 3.701371374957663e-07,
      "loss": 0.6638,
      "step": 30557
    },
    {
      "epoch": 0.880481761078776,
      "grad_norm": 0.7357214689254761,
      "learning_rate": 3.6996096997681976e-07,
      "loss": 0.6936,
      "step": 30558
    },
    {
      "epoch": 0.8805105745404253,
      "grad_norm": 0.7118806838989258,
      "learning_rate": 3.6978484278088535e-07,
      "loss": 0.6548,
      "step": 30559
    },
    {
      "epoch": 0.8805393880020745,
      "grad_norm": 0.7637447118759155,
      "learning_rate": 3.696087559094957e-07,
      "loss": 0.712,
      "step": 30560
    },
    {
      "epoch": 0.8805682014637238,
      "grad_norm": 0.7336201071739197,
      "learning_rate": 3.694327093641864e-07,
      "loss": 0.678,
      "step": 30561
    },
    {
      "epoch": 0.8805970149253731,
      "grad_norm": 1.12135910987854,
      "learning_rate": 3.6925670314648775e-07,
      "loss": 0.6959,
      "step": 30562
    },
    {
      "epoch": 0.8806258283870224,
      "grad_norm": 0.7509216070175171,
      "learning_rate": 3.6908073725793524e-07,
      "loss": 0.7019,
      "step": 30563
    },
    {
      "epoch": 0.8806546418486717,
      "grad_norm": 0.7293222546577454,
      "learning_rate": 3.6890481170005876e-07,
      "loss": 0.6733,
      "step": 30564
    },
    {
      "epoch": 0.880683455310321,
      "grad_norm": 0.7289807796478271,
      "learning_rate": 3.6872892647439373e-07,
      "loss": 0.6603,
      "step": 30565
    },
    {
      "epoch": 0.8807122687719703,
      "grad_norm": 0.7430860996246338,
      "learning_rate": 3.6855308158246894e-07,
      "loss": 0.6788,
      "step": 30566
    },
    {
      "epoch": 0.8807410822336196,
      "grad_norm": 0.715111494064331,
      "learning_rate": 3.6837727702581814e-07,
      "loss": 0.6647,
      "step": 30567
    },
    {
      "epoch": 0.8807698956952689,
      "grad_norm": 0.7337222099304199,
      "learning_rate": 3.682015128059707e-07,
      "loss": 0.6977,
      "step": 30568
    },
    {
      "epoch": 0.8807987091569182,
      "grad_norm": 0.7310346364974976,
      "learning_rate": 3.680257889244587e-07,
      "loss": 0.6824,
      "step": 30569
    },
    {
      "epoch": 0.8808275226185674,
      "grad_norm": 0.7113814949989319,
      "learning_rate": 3.6785010538281094e-07,
      "loss": 0.6641,
      "step": 30570
    },
    {
      "epoch": 0.8808563360802166,
      "grad_norm": 0.7248906493186951,
      "learning_rate": 3.67674462182559e-07,
      "loss": 0.6736,
      "step": 30571
    },
    {
      "epoch": 0.8808851495418659,
      "grad_norm": 0.685678243637085,
      "learning_rate": 3.6749885932523097e-07,
      "loss": 0.6618,
      "step": 30572
    },
    {
      "epoch": 0.8809139630035152,
      "grad_norm": 0.7255784869194031,
      "learning_rate": 3.67323296812358e-07,
      "loss": 0.6835,
      "step": 30573
    },
    {
      "epoch": 0.8809427764651645,
      "grad_norm": 0.7100962400436401,
      "learning_rate": 3.6714777464546714e-07,
      "loss": 0.6697,
      "step": 30574
    },
    {
      "epoch": 0.8809715899268138,
      "grad_norm": 0.7160577178001404,
      "learning_rate": 3.6697229282608827e-07,
      "loss": 0.6927,
      "step": 30575
    },
    {
      "epoch": 0.8810004033884631,
      "grad_norm": 0.715569257736206,
      "learning_rate": 3.667968513557496e-07,
      "loss": 0.6787,
      "step": 30576
    },
    {
      "epoch": 0.8810292168501124,
      "grad_norm": 0.7186816930770874,
      "learning_rate": 3.666214502359794e-07,
      "loss": 0.6649,
      "step": 30577
    },
    {
      "epoch": 0.8810580303117617,
      "grad_norm": 0.7326814532279968,
      "learning_rate": 3.664460894683036e-07,
      "loss": 0.6899,
      "step": 30578
    },
    {
      "epoch": 0.881086843773411,
      "grad_norm": 0.7101396322250366,
      "learning_rate": 3.662707690542511e-07,
      "loss": 0.6791,
      "step": 30579
    },
    {
      "epoch": 0.8811156572350602,
      "grad_norm": 0.7210766673088074,
      "learning_rate": 3.660954889953483e-07,
      "loss": 0.6641,
      "step": 30580
    },
    {
      "epoch": 0.8811444706967095,
      "grad_norm": 0.7126204967498779,
      "learning_rate": 3.659202492931213e-07,
      "loss": 0.6899,
      "step": 30581
    },
    {
      "epoch": 0.8811732841583588,
      "grad_norm": 0.7432328462600708,
      "learning_rate": 3.6574504994909723e-07,
      "loss": 0.7053,
      "step": 30582
    },
    {
      "epoch": 0.8812020976200081,
      "grad_norm": 0.7348480820655823,
      "learning_rate": 3.655698909648003e-07,
      "loss": 0.6754,
      "step": 30583
    },
    {
      "epoch": 0.8812309110816573,
      "grad_norm": 0.7086305618286133,
      "learning_rate": 3.653947723417578e-07,
      "loss": 0.6716,
      "step": 30584
    },
    {
      "epoch": 0.8812597245433066,
      "grad_norm": 0.7006499171257019,
      "learning_rate": 3.652196940814928e-07,
      "loss": 0.6607,
      "step": 30585
    },
    {
      "epoch": 0.8812885380049559,
      "grad_norm": 0.7132726907730103,
      "learning_rate": 3.6504465618553256e-07,
      "loss": 0.6674,
      "step": 30586
    },
    {
      "epoch": 0.8813173514666052,
      "grad_norm": 0.7239859104156494,
      "learning_rate": 3.6486965865539904e-07,
      "loss": 0.6563,
      "step": 30587
    },
    {
      "epoch": 0.8813461649282545,
      "grad_norm": 0.7254562377929688,
      "learning_rate": 3.6469470149261723e-07,
      "loss": 0.668,
      "step": 30588
    },
    {
      "epoch": 0.8813749783899038,
      "grad_norm": 0.7179539799690247,
      "learning_rate": 3.6451978469871086e-07,
      "loss": 0.6515,
      "step": 30589
    },
    {
      "epoch": 0.881403791851553,
      "grad_norm": 0.7071428298950195,
      "learning_rate": 3.643449082752043e-07,
      "loss": 0.663,
      "step": 30590
    },
    {
      "epoch": 0.8814326053132023,
      "grad_norm": 0.7507343888282776,
      "learning_rate": 3.641700722236191e-07,
      "loss": 0.7001,
      "step": 30591
    },
    {
      "epoch": 0.8814614187748516,
      "grad_norm": 0.7302632927894592,
      "learning_rate": 3.6399527654547905e-07,
      "loss": 0.6829,
      "step": 30592
    },
    {
      "epoch": 0.8814902322365009,
      "grad_norm": 0.7201176285743713,
      "learning_rate": 3.638205212423046e-07,
      "loss": 0.7153,
      "step": 30593
    },
    {
      "epoch": 0.8815190456981502,
      "grad_norm": 0.7129521369934082,
      "learning_rate": 3.636458063156206e-07,
      "loss": 0.6668,
      "step": 30594
    },
    {
      "epoch": 0.8815478591597995,
      "grad_norm": 0.7058051824569702,
      "learning_rate": 3.634711317669454e-07,
      "loss": 0.6441,
      "step": 30595
    },
    {
      "epoch": 0.8815766726214488,
      "grad_norm": 0.7241359949111938,
      "learning_rate": 3.6329649759780205e-07,
      "loss": 0.707,
      "step": 30596
    },
    {
      "epoch": 0.8816054860830981,
      "grad_norm": 0.7133126854896545,
      "learning_rate": 3.6312190380971226e-07,
      "loss": 0.67,
      "step": 30597
    },
    {
      "epoch": 0.8816342995447473,
      "grad_norm": 0.7256449460983276,
      "learning_rate": 3.629473504041947e-07,
      "loss": 0.6773,
      "step": 30598
    },
    {
      "epoch": 0.8816631130063965,
      "grad_norm": 0.7087865471839905,
      "learning_rate": 3.6277283738277105e-07,
      "loss": 0.6572,
      "step": 30599
    },
    {
      "epoch": 0.8816919264680458,
      "grad_norm": 0.6994092464447021,
      "learning_rate": 3.625983647469594e-07,
      "loss": 0.6601,
      "step": 30600
    },
    {
      "epoch": 0.8817207399296951,
      "grad_norm": 0.7386941909790039,
      "learning_rate": 3.624239324982803e-07,
      "loss": 0.6867,
      "step": 30601
    },
    {
      "epoch": 0.8817495533913444,
      "grad_norm": 0.7417094707489014,
      "learning_rate": 3.622495406382531e-07,
      "loss": 0.7016,
      "step": 30602
    },
    {
      "epoch": 0.8817783668529937,
      "grad_norm": 0.7699013352394104,
      "learning_rate": 3.6207518916839713e-07,
      "loss": 0.6808,
      "step": 30603
    },
    {
      "epoch": 0.881807180314643,
      "grad_norm": 0.7134562730789185,
      "learning_rate": 3.619008780902289e-07,
      "loss": 0.6473,
      "step": 30604
    },
    {
      "epoch": 0.8818359937762923,
      "grad_norm": 0.7496457695960999,
      "learning_rate": 3.6172660740526887e-07,
      "loss": 0.7069,
      "step": 30605
    },
    {
      "epoch": 0.8818648072379416,
      "grad_norm": 0.7206522226333618,
      "learning_rate": 3.615523771150325e-07,
      "loss": 0.7063,
      "step": 30606
    },
    {
      "epoch": 0.8818936206995909,
      "grad_norm": 0.7278789281845093,
      "learning_rate": 3.6137818722103924e-07,
      "loss": 0.6746,
      "step": 30607
    },
    {
      "epoch": 0.8819224341612402,
      "grad_norm": 0.7206349968910217,
      "learning_rate": 3.6120403772480383e-07,
      "loss": 0.6739,
      "step": 30608
    },
    {
      "epoch": 0.8819512476228895,
      "grad_norm": 0.7181640267372131,
      "learning_rate": 3.610299286278451e-07,
      "loss": 0.7009,
      "step": 30609
    },
    {
      "epoch": 0.8819800610845387,
      "grad_norm": 0.7023998498916626,
      "learning_rate": 3.6085585993167804e-07,
      "loss": 0.6539,
      "step": 30610
    },
    {
      "epoch": 0.8820088745461879,
      "grad_norm": 0.7264882326126099,
      "learning_rate": 3.606818316378191e-07,
      "loss": 0.6713,
      "step": 30611
    },
    {
      "epoch": 0.8820376880078372,
      "grad_norm": 0.7417111396789551,
      "learning_rate": 3.605078437477827e-07,
      "loss": 0.6702,
      "step": 30612
    },
    {
      "epoch": 0.8820665014694865,
      "grad_norm": 0.7430945634841919,
      "learning_rate": 3.6033389626308647e-07,
      "loss": 0.6788,
      "step": 30613
    },
    {
      "epoch": 0.8820953149311358,
      "grad_norm": 0.7080162167549133,
      "learning_rate": 3.6015998918524365e-07,
      "loss": 0.6626,
      "step": 30614
    },
    {
      "epoch": 0.8821241283927851,
      "grad_norm": 0.7396865487098694,
      "learning_rate": 3.599861225157697e-07,
      "loss": 0.7041,
      "step": 30615
    },
    {
      "epoch": 0.8821529418544344,
      "grad_norm": 0.723471462726593,
      "learning_rate": 3.598122962561773e-07,
      "loss": 0.6721,
      "step": 30616
    },
    {
      "epoch": 0.8821817553160837,
      "grad_norm": 0.7149025201797485,
      "learning_rate": 3.5963851040798136e-07,
      "loss": 0.683,
      "step": 30617
    },
    {
      "epoch": 0.882210568777733,
      "grad_norm": 0.7409001588821411,
      "learning_rate": 3.594647649726962e-07,
      "loss": 0.6789,
      "step": 30618
    },
    {
      "epoch": 0.8822393822393823,
      "grad_norm": 0.732959508895874,
      "learning_rate": 3.5929105995183344e-07,
      "loss": 0.6748,
      "step": 30619
    },
    {
      "epoch": 0.8822681957010315,
      "grad_norm": 0.7114374041557312,
      "learning_rate": 3.5911739534690735e-07,
      "loss": 0.6889,
      "step": 30620
    },
    {
      "epoch": 0.8822970091626808,
      "grad_norm": 0.7339305877685547,
      "learning_rate": 3.5894377115942847e-07,
      "loss": 0.6914,
      "step": 30621
    },
    {
      "epoch": 0.8823258226243301,
      "grad_norm": 0.7176750898361206,
      "learning_rate": 3.5877018739091107e-07,
      "loss": 0.6693,
      "step": 30622
    },
    {
      "epoch": 0.8823546360859794,
      "grad_norm": 0.7050756812095642,
      "learning_rate": 3.585966440428651e-07,
      "loss": 0.6802,
      "step": 30623
    },
    {
      "epoch": 0.8823834495476287,
      "grad_norm": 0.7670143246650696,
      "learning_rate": 3.5842314111680263e-07,
      "loss": 0.6922,
      "step": 30624
    },
    {
      "epoch": 0.8824122630092779,
      "grad_norm": 0.7478652000427246,
      "learning_rate": 3.582496786142342e-07,
      "loss": 0.6733,
      "step": 30625
    },
    {
      "epoch": 0.8824410764709272,
      "grad_norm": 0.7271844148635864,
      "learning_rate": 3.5807625653667246e-07,
      "loss": 0.6898,
      "step": 30626
    },
    {
      "epoch": 0.8824698899325765,
      "grad_norm": 0.6982645988464355,
      "learning_rate": 3.5790287488562503e-07,
      "loss": 0.6846,
      "step": 30627
    },
    {
      "epoch": 0.8824987033942258,
      "grad_norm": 0.7529781460762024,
      "learning_rate": 3.5772953366260467e-07,
      "loss": 0.7105,
      "step": 30628
    },
    {
      "epoch": 0.882527516855875,
      "grad_norm": 0.7288388609886169,
      "learning_rate": 3.5755623286911787e-07,
      "loss": 0.6994,
      "step": 30629
    },
    {
      "epoch": 0.8825563303175243,
      "grad_norm": 0.7397575974464417,
      "learning_rate": 3.5738297250667687e-07,
      "loss": 0.6997,
      "step": 30630
    },
    {
      "epoch": 0.8825851437791736,
      "grad_norm": 0.7674015760421753,
      "learning_rate": 3.5720975257678815e-07,
      "loss": 0.6932,
      "step": 30631
    },
    {
      "epoch": 0.8826139572408229,
      "grad_norm": 0.7124335765838623,
      "learning_rate": 3.570365730809622e-07,
      "loss": 0.6751,
      "step": 30632
    },
    {
      "epoch": 0.8826427707024722,
      "grad_norm": 0.7558530569076538,
      "learning_rate": 3.568634340207061e-07,
      "loss": 0.6887,
      "step": 30633
    },
    {
      "epoch": 0.8826715841641215,
      "grad_norm": 0.7225245237350464,
      "learning_rate": 3.566903353975276e-07,
      "loss": 0.6903,
      "step": 30634
    },
    {
      "epoch": 0.8827003976257708,
      "grad_norm": 0.7237273454666138,
      "learning_rate": 3.565172772129355e-07,
      "loss": 0.6692,
      "step": 30635
    },
    {
      "epoch": 0.8827292110874201,
      "grad_norm": 0.726556122303009,
      "learning_rate": 3.5634425946843523e-07,
      "loss": 0.6926,
      "step": 30636
    },
    {
      "epoch": 0.8827580245490694,
      "grad_norm": 0.7211825847625732,
      "learning_rate": 3.56171282165535e-07,
      "loss": 0.692,
      "step": 30637
    },
    {
      "epoch": 0.8827868380107186,
      "grad_norm": 0.7292521595954895,
      "learning_rate": 3.559983453057408e-07,
      "loss": 0.6773,
      "step": 30638
    },
    {
      "epoch": 0.8828156514723678,
      "grad_norm": 0.7371748089790344,
      "learning_rate": 3.5582544889055825e-07,
      "loss": 0.6905,
      "step": 30639
    },
    {
      "epoch": 0.8828444649340171,
      "grad_norm": 0.7057431936264038,
      "learning_rate": 3.556525929214938e-07,
      "loss": 0.6574,
      "step": 30640
    },
    {
      "epoch": 0.8828732783956664,
      "grad_norm": 0.746385931968689,
      "learning_rate": 3.5547977740005344e-07,
      "loss": 0.6711,
      "step": 30641
    },
    {
      "epoch": 0.8829020918573157,
      "grad_norm": 0.7197448015213013,
      "learning_rate": 3.553070023277405e-07,
      "loss": 0.6784,
      "step": 30642
    },
    {
      "epoch": 0.882930905318965,
      "grad_norm": 0.6997861266136169,
      "learning_rate": 3.551342677060615e-07,
      "loss": 0.669,
      "step": 30643
    },
    {
      "epoch": 0.8829597187806143,
      "grad_norm": 0.722402811050415,
      "learning_rate": 3.549615735365186e-07,
      "loss": 0.6917,
      "step": 30644
    },
    {
      "epoch": 0.8829885322422636,
      "grad_norm": 0.7187602519989014,
      "learning_rate": 3.5478891982061834e-07,
      "loss": 0.6872,
      "step": 30645
    },
    {
      "epoch": 0.8830173457039129,
      "grad_norm": 0.7085257768630981,
      "learning_rate": 3.546163065598618e-07,
      "loss": 0.6799,
      "step": 30646
    },
    {
      "epoch": 0.8830461591655622,
      "grad_norm": 0.7361193895339966,
      "learning_rate": 3.544437337557549e-07,
      "loss": 0.6982,
      "step": 30647
    },
    {
      "epoch": 0.8830749726272115,
      "grad_norm": 0.7090529799461365,
      "learning_rate": 3.542712014097971e-07,
      "loss": 0.679,
      "step": 30648
    },
    {
      "epoch": 0.8831037860888608,
      "grad_norm": 0.7087087631225586,
      "learning_rate": 3.540987095234949e-07,
      "loss": 0.7048,
      "step": 30649
    },
    {
      "epoch": 0.88313259955051,
      "grad_norm": 0.700027585029602,
      "learning_rate": 3.539262580983482e-07,
      "loss": 0.6829,
      "step": 30650
    },
    {
      "epoch": 0.8831614130121593,
      "grad_norm": 0.7129673361778259,
      "learning_rate": 3.5375384713585983e-07,
      "loss": 0.7045,
      "step": 30651
    },
    {
      "epoch": 0.8831902264738085,
      "grad_norm": 0.7162010073661804,
      "learning_rate": 3.5358147663753007e-07,
      "loss": 0.6828,
      "step": 30652
    },
    {
      "epoch": 0.8832190399354578,
      "grad_norm": 0.7393617033958435,
      "learning_rate": 3.5340914660486227e-07,
      "loss": 0.6953,
      "step": 30653
    },
    {
      "epoch": 0.8832478533971071,
      "grad_norm": 0.7289429903030396,
      "learning_rate": 3.532368570393546e-07,
      "loss": 0.6915,
      "step": 30654
    },
    {
      "epoch": 0.8832766668587564,
      "grad_norm": 0.7278397679328918,
      "learning_rate": 3.530646079425093e-07,
      "loss": 0.6881,
      "step": 30655
    },
    {
      "epoch": 0.8833054803204057,
      "grad_norm": 0.7192563414573669,
      "learning_rate": 3.528923993158262e-07,
      "loss": 0.6676,
      "step": 30656
    },
    {
      "epoch": 0.883334293782055,
      "grad_norm": 0.7221118211746216,
      "learning_rate": 3.527202311608047e-07,
      "loss": 0.6902,
      "step": 30657
    },
    {
      "epoch": 0.8833631072437043,
      "grad_norm": 0.7230218052864075,
      "learning_rate": 3.5254810347894463e-07,
      "loss": 0.6706,
      "step": 30658
    },
    {
      "epoch": 0.8833919207053536,
      "grad_norm": 0.7108289003372192,
      "learning_rate": 3.523760162717449e-07,
      "loss": 0.6661,
      "step": 30659
    },
    {
      "epoch": 0.8834207341670028,
      "grad_norm": 0.7172226309776306,
      "learning_rate": 3.5220396954070315e-07,
      "loss": 0.6743,
      "step": 30660
    },
    {
      "epoch": 0.8834495476286521,
      "grad_norm": 0.7092434763908386,
      "learning_rate": 3.520319632873193e-07,
      "loss": 0.6686,
      "step": 30661
    },
    {
      "epoch": 0.8834783610903014,
      "grad_norm": 0.6986984014511108,
      "learning_rate": 3.518599975130915e-07,
      "loss": 0.6659,
      "step": 30662
    },
    {
      "epoch": 0.8835071745519507,
      "grad_norm": 0.7044018507003784,
      "learning_rate": 3.5168807221951594e-07,
      "loss": 0.6571,
      "step": 30663
    },
    {
      "epoch": 0.8835359880136,
      "grad_norm": 0.7374884486198425,
      "learning_rate": 3.5151618740809137e-07,
      "loss": 0.6765,
      "step": 30664
    },
    {
      "epoch": 0.8835648014752492,
      "grad_norm": 0.7237429022789001,
      "learning_rate": 3.513443430803132e-07,
      "loss": 0.6906,
      "step": 30665
    },
    {
      "epoch": 0.8835936149368985,
      "grad_norm": 0.722755491733551,
      "learning_rate": 3.511725392376797e-07,
      "loss": 0.6946,
      "step": 30666
    },
    {
      "epoch": 0.8836224283985478,
      "grad_norm": 0.7136743068695068,
      "learning_rate": 3.5100077588168524e-07,
      "loss": 0.6702,
      "step": 30667
    },
    {
      "epoch": 0.8836512418601971,
      "grad_norm": 0.7555471062660217,
      "learning_rate": 3.508290530138275e-07,
      "loss": 0.7069,
      "step": 30668
    },
    {
      "epoch": 0.8836800553218463,
      "grad_norm": 0.713659942150116,
      "learning_rate": 3.506573706356009e-07,
      "loss": 0.6878,
      "step": 30669
    },
    {
      "epoch": 0.8837088687834956,
      "grad_norm": 0.7100513577461243,
      "learning_rate": 3.504857287485014e-07,
      "loss": 0.6772,
      "step": 30670
    },
    {
      "epoch": 0.8837376822451449,
      "grad_norm": 0.7408758401870728,
      "learning_rate": 3.5031412735402283e-07,
      "loss": 0.6819,
      "step": 30671
    },
    {
      "epoch": 0.8837664957067942,
      "grad_norm": 0.7251319885253906,
      "learning_rate": 3.501425664536601e-07,
      "loss": 0.6756,
      "step": 30672
    },
    {
      "epoch": 0.8837953091684435,
      "grad_norm": 0.742007315158844,
      "learning_rate": 3.499710460489075e-07,
      "loss": 0.6868,
      "step": 30673
    },
    {
      "epoch": 0.8838241226300928,
      "grad_norm": 0.717519998550415,
      "learning_rate": 3.4979956614125956e-07,
      "loss": 0.669,
      "step": 30674
    },
    {
      "epoch": 0.8838529360917421,
      "grad_norm": 0.7116888165473938,
      "learning_rate": 3.496281267322077e-07,
      "loss": 0.6577,
      "step": 30675
    },
    {
      "epoch": 0.8838817495533914,
      "grad_norm": 0.7060546875,
      "learning_rate": 3.4945672782324693e-07,
      "loss": 0.6655,
      "step": 30676
    },
    {
      "epoch": 0.8839105630150407,
      "grad_norm": 0.7086964249610901,
      "learning_rate": 3.492853694158693e-07,
      "loss": 0.6812,
      "step": 30677
    },
    {
      "epoch": 0.88393937647669,
      "grad_norm": 0.6996662020683289,
      "learning_rate": 3.491140515115665e-07,
      "loss": 0.6718,
      "step": 30678
    },
    {
      "epoch": 0.8839681899383391,
      "grad_norm": 0.7235449552536011,
      "learning_rate": 3.489427741118323e-07,
      "loss": 0.6821,
      "step": 30679
    },
    {
      "epoch": 0.8839970033999884,
      "grad_norm": 0.7042392492294312,
      "learning_rate": 3.48771537218156e-07,
      "loss": 0.6795,
      "step": 30680
    },
    {
      "epoch": 0.8840258168616377,
      "grad_norm": 0.7177771925926208,
      "learning_rate": 3.48600340832031e-07,
      "loss": 0.6829,
      "step": 30681
    },
    {
      "epoch": 0.884054630323287,
      "grad_norm": 0.7305740714073181,
      "learning_rate": 3.484291849549465e-07,
      "loss": 0.6753,
      "step": 30682
    },
    {
      "epoch": 0.8840834437849363,
      "grad_norm": 0.7497586011886597,
      "learning_rate": 3.4825806958839525e-07,
      "loss": 0.6918,
      "step": 30683
    },
    {
      "epoch": 0.8841122572465856,
      "grad_norm": 0.7125902771949768,
      "learning_rate": 3.480869947338639e-07,
      "loss": 0.6799,
      "step": 30684
    },
    {
      "epoch": 0.8841410707082349,
      "grad_norm": 0.7139227986335754,
      "learning_rate": 3.479159603928461e-07,
      "loss": 0.6353,
      "step": 30685
    },
    {
      "epoch": 0.8841698841698842,
      "grad_norm": 0.7204065918922424,
      "learning_rate": 3.4774496656682977e-07,
      "loss": 0.6682,
      "step": 30686
    },
    {
      "epoch": 0.8841986976315335,
      "grad_norm": 0.7265254855155945,
      "learning_rate": 3.4757401325730466e-07,
      "loss": 0.6841,
      "step": 30687
    },
    {
      "epoch": 0.8842275110931828,
      "grad_norm": 0.7329373955726624,
      "learning_rate": 3.4740310046575856e-07,
      "loss": 0.6914,
      "step": 30688
    },
    {
      "epoch": 0.884256324554832,
      "grad_norm": 0.7060727477073669,
      "learning_rate": 3.4723222819368076e-07,
      "loss": 0.6787,
      "step": 30689
    },
    {
      "epoch": 0.8842851380164813,
      "grad_norm": 0.717202365398407,
      "learning_rate": 3.4706139644255896e-07,
      "loss": 0.6977,
      "step": 30690
    },
    {
      "epoch": 0.8843139514781306,
      "grad_norm": 0.705935537815094,
      "learning_rate": 3.468906052138821e-07,
      "loss": 0.6672,
      "step": 30691
    },
    {
      "epoch": 0.8843427649397798,
      "grad_norm": 0.7070958018302917,
      "learning_rate": 3.467198545091355e-07,
      "loss": 0.6685,
      "step": 30692
    },
    {
      "epoch": 0.8843715784014291,
      "grad_norm": 1.8282628059387207,
      "learning_rate": 3.465491443298075e-07,
      "loss": 0.7138,
      "step": 30693
    },
    {
      "epoch": 0.8844003918630784,
      "grad_norm": 0.7271238565444946,
      "learning_rate": 3.4637847467738514e-07,
      "loss": 0.6658,
      "step": 30694
    },
    {
      "epoch": 0.8844292053247277,
      "grad_norm": 0.8080576062202454,
      "learning_rate": 3.4620784555335407e-07,
      "loss": 0.6874,
      "step": 30695
    },
    {
      "epoch": 0.884458018786377,
      "grad_norm": 0.734838604927063,
      "learning_rate": 3.460372569592002e-07,
      "loss": 0.6803,
      "step": 30696
    },
    {
      "epoch": 0.8844868322480263,
      "grad_norm": 0.7224105000495911,
      "learning_rate": 3.458667088964107e-07,
      "loss": 0.6689,
      "step": 30697
    },
    {
      "epoch": 0.8845156457096756,
      "grad_norm": 0.7105386257171631,
      "learning_rate": 3.4569620136646887e-07,
      "loss": 0.6796,
      "step": 30698
    },
    {
      "epoch": 0.8845444591713248,
      "grad_norm": 0.7231128215789795,
      "learning_rate": 3.455257343708601e-07,
      "loss": 0.6639,
      "step": 30699
    },
    {
      "epoch": 0.8845732726329741,
      "grad_norm": 0.7234055399894714,
      "learning_rate": 3.4535530791107053e-07,
      "loss": 0.6894,
      "step": 30700
    },
    {
      "epoch": 0.8846020860946234,
      "grad_norm": 0.7051517367362976,
      "learning_rate": 3.451849219885828e-07,
      "loss": 0.6867,
      "step": 30701
    },
    {
      "epoch": 0.8846308995562727,
      "grad_norm": 0.7223058342933655,
      "learning_rate": 3.450145766048818e-07,
      "loss": 0.7018,
      "step": 30702
    },
    {
      "epoch": 0.884659713017922,
      "grad_norm": 0.7339283227920532,
      "learning_rate": 3.4484427176144974e-07,
      "loss": 0.6888,
      "step": 30703
    },
    {
      "epoch": 0.8846885264795713,
      "grad_norm": 0.7186923027038574,
      "learning_rate": 3.4467400745977155e-07,
      "loss": 0.6802,
      "step": 30704
    },
    {
      "epoch": 0.8847173399412206,
      "grad_norm": 0.7172607183456421,
      "learning_rate": 3.445037837013282e-07,
      "loss": 0.6873,
      "step": 30705
    },
    {
      "epoch": 0.8847461534028698,
      "grad_norm": 0.7344896197319031,
      "learning_rate": 3.4433360048760357e-07,
      "loss": 0.6838,
      "step": 30706
    },
    {
      "epoch": 0.8847749668645191,
      "grad_norm": 0.7175235152244568,
      "learning_rate": 3.441634578200792e-07,
      "loss": 0.6499,
      "step": 30707
    },
    {
      "epoch": 0.8848037803261684,
      "grad_norm": 0.7243319749832153,
      "learning_rate": 3.439933557002362e-07,
      "loss": 0.6834,
      "step": 30708
    },
    {
      "epoch": 0.8848325937878176,
      "grad_norm": 0.7157882452011108,
      "learning_rate": 3.438232941295572e-07,
      "loss": 0.6853,
      "step": 30709
    },
    {
      "epoch": 0.8848614072494669,
      "grad_norm": 0.7455711960792542,
      "learning_rate": 3.436532731095238e-07,
      "loss": 0.6716,
      "step": 30710
    },
    {
      "epoch": 0.8848902207111162,
      "grad_norm": 0.7348159551620483,
      "learning_rate": 3.4348329264161484e-07,
      "loss": 0.672,
      "step": 30711
    },
    {
      "epoch": 0.8849190341727655,
      "grad_norm": 0.7211923599243164,
      "learning_rate": 3.4331335272731247e-07,
      "loss": 0.6648,
      "step": 30712
    },
    {
      "epoch": 0.8849478476344148,
      "grad_norm": 0.7287035584449768,
      "learning_rate": 3.4314345336809497e-07,
      "loss": 0.663,
      "step": 30713
    },
    {
      "epoch": 0.8849766610960641,
      "grad_norm": 0.7000629901885986,
      "learning_rate": 3.429735945654428e-07,
      "loss": 0.6731,
      "step": 30714
    },
    {
      "epoch": 0.8850054745577134,
      "grad_norm": 0.7330555319786072,
      "learning_rate": 3.428037763208358e-07,
      "loss": 0.6737,
      "step": 30715
    },
    {
      "epoch": 0.8850342880193627,
      "grad_norm": 0.7229329943656921,
      "learning_rate": 3.4263399863575184e-07,
      "loss": 0.6789,
      "step": 30716
    },
    {
      "epoch": 0.885063101481012,
      "grad_norm": 0.6948034167289734,
      "learning_rate": 3.424642615116708e-07,
      "loss": 0.6466,
      "step": 30717
    },
    {
      "epoch": 0.8850919149426613,
      "grad_norm": 0.7204497456550598,
      "learning_rate": 3.4229456495006976e-07,
      "loss": 0.6763,
      "step": 30718
    },
    {
      "epoch": 0.8851207284043104,
      "grad_norm": 0.7409271597862244,
      "learning_rate": 3.421249089524276e-07,
      "loss": 0.7032,
      "step": 30719
    },
    {
      "epoch": 0.8851495418659597,
      "grad_norm": 0.9412902593612671,
      "learning_rate": 3.4195529352021973e-07,
      "loss": 0.669,
      "step": 30720
    },
    {
      "epoch": 0.885178355327609,
      "grad_norm": 0.7439499497413635,
      "learning_rate": 3.417857186549267e-07,
      "loss": 0.6765,
      "step": 30721
    },
    {
      "epoch": 0.8852071687892583,
      "grad_norm": 0.7074027061462402,
      "learning_rate": 3.4161618435802234e-07,
      "loss": 0.6698,
      "step": 30722
    },
    {
      "epoch": 0.8852359822509076,
      "grad_norm": 0.7686452269554138,
      "learning_rate": 3.414466906309855e-07,
      "loss": 0.6996,
      "step": 30723
    },
    {
      "epoch": 0.8852647957125569,
      "grad_norm": 0.7006738185882568,
      "learning_rate": 3.4127723747528986e-07,
      "loss": 0.6631,
      "step": 30724
    },
    {
      "epoch": 0.8852936091742062,
      "grad_norm": 0.7283909916877747,
      "learning_rate": 3.411078248924138e-07,
      "loss": 0.6942,
      "step": 30725
    },
    {
      "epoch": 0.8853224226358555,
      "grad_norm": 0.739250898361206,
      "learning_rate": 3.409384528838305e-07,
      "loss": 0.6774,
      "step": 30726
    },
    {
      "epoch": 0.8853512360975048,
      "grad_norm": 0.7206910252571106,
      "learning_rate": 3.407691214510167e-07,
      "loss": 0.6994,
      "step": 30727
    },
    {
      "epoch": 0.8853800495591541,
      "grad_norm": 0.705918550491333,
      "learning_rate": 3.40599830595445e-07,
      "loss": 0.6715,
      "step": 30728
    },
    {
      "epoch": 0.8854088630208033,
      "grad_norm": 0.7248021364212036,
      "learning_rate": 3.4043058031859257e-07,
      "loss": 0.6789,
      "step": 30729
    },
    {
      "epoch": 0.8854376764824526,
      "grad_norm": 0.7296286225318909,
      "learning_rate": 3.40261370621931e-07,
      "loss": 0.6991,
      "step": 30730
    },
    {
      "epoch": 0.8854664899441019,
      "grad_norm": 0.7317928075790405,
      "learning_rate": 3.400922015069347e-07,
      "loss": 0.6886,
      "step": 30731
    },
    {
      "epoch": 0.8854953034057512,
      "grad_norm": 0.7389832139015198,
      "learning_rate": 3.3992307297507687e-07,
      "loss": 0.695,
      "step": 30732
    },
    {
      "epoch": 0.8855241168674004,
      "grad_norm": 0.7256151437759399,
      "learning_rate": 3.39753985027832e-07,
      "loss": 0.6933,
      "step": 30733
    },
    {
      "epoch": 0.8855529303290497,
      "grad_norm": 0.7415919899940491,
      "learning_rate": 3.3958493766667045e-07,
      "loss": 0.6591,
      "step": 30734
    },
    {
      "epoch": 0.885581743790699,
      "grad_norm": 0.722470760345459,
      "learning_rate": 3.3941593089306613e-07,
      "loss": 0.6691,
      "step": 30735
    },
    {
      "epoch": 0.8856105572523483,
      "grad_norm": 0.7372829914093018,
      "learning_rate": 3.392469647084895e-07,
      "loss": 0.7103,
      "step": 30736
    },
    {
      "epoch": 0.8856393707139976,
      "grad_norm": 0.7200138568878174,
      "learning_rate": 3.3907803911441215e-07,
      "loss": 0.6746,
      "step": 30737
    },
    {
      "epoch": 0.8856681841756469,
      "grad_norm": 0.7163688540458679,
      "learning_rate": 3.389091541123074e-07,
      "loss": 0.6852,
      "step": 30738
    },
    {
      "epoch": 0.8856969976372961,
      "grad_norm": 0.7117031812667847,
      "learning_rate": 3.387403097036435e-07,
      "loss": 0.692,
      "step": 30739
    },
    {
      "epoch": 0.8857258110989454,
      "grad_norm": 0.7538135647773743,
      "learning_rate": 3.385715058898925e-07,
      "loss": 0.6627,
      "step": 30740
    },
    {
      "epoch": 0.8857546245605947,
      "grad_norm": 0.72381192445755,
      "learning_rate": 3.3840274267252336e-07,
      "loss": 0.6806,
      "step": 30741
    },
    {
      "epoch": 0.885783438022244,
      "grad_norm": 0.7298758029937744,
      "learning_rate": 3.382340200530071e-07,
      "loss": 0.7106,
      "step": 30742
    },
    {
      "epoch": 0.8858122514838933,
      "grad_norm": 0.7121381759643555,
      "learning_rate": 3.380653380328119e-07,
      "loss": 0.6712,
      "step": 30743
    },
    {
      "epoch": 0.8858410649455426,
      "grad_norm": 0.7348463535308838,
      "learning_rate": 3.378966966134073e-07,
      "loss": 0.6863,
      "step": 30744
    },
    {
      "epoch": 0.8858698784071919,
      "grad_norm": 0.710120439529419,
      "learning_rate": 3.3772809579626197e-07,
      "loss": 0.6825,
      "step": 30745
    },
    {
      "epoch": 0.8858986918688411,
      "grad_norm": 0.702688992023468,
      "learning_rate": 3.375595355828454e-07,
      "loss": 0.6768,
      "step": 30746
    },
    {
      "epoch": 0.8859275053304904,
      "grad_norm": 0.7246979475021362,
      "learning_rate": 3.3739101597462354e-07,
      "loss": 0.667,
      "step": 30747
    },
    {
      "epoch": 0.8859563187921397,
      "grad_norm": 0.72368985414505,
      "learning_rate": 3.372225369730658e-07,
      "loss": 0.681,
      "step": 30748
    },
    {
      "epoch": 0.8859851322537889,
      "grad_norm": 0.699020266532898,
      "learning_rate": 3.370540985796383e-07,
      "loss": 0.6599,
      "step": 30749
    },
    {
      "epoch": 0.8860139457154382,
      "grad_norm": 0.7148356437683105,
      "learning_rate": 3.3688570079580875e-07,
      "loss": 0.7136,
      "step": 30750
    },
    {
      "epoch": 0.8860427591770875,
      "grad_norm": 0.7330763936042786,
      "learning_rate": 3.3671734362304256e-07,
      "loss": 0.6757,
      "step": 30751
    },
    {
      "epoch": 0.8860715726387368,
      "grad_norm": 0.7183628082275391,
      "learning_rate": 3.365490270628069e-07,
      "loss": 0.6734,
      "step": 30752
    },
    {
      "epoch": 0.8861003861003861,
      "grad_norm": 0.7330203652381897,
      "learning_rate": 3.3638075111656785e-07,
      "loss": 0.6903,
      "step": 30753
    },
    {
      "epoch": 0.8861291995620354,
      "grad_norm": 0.722248375415802,
      "learning_rate": 3.362125157857904e-07,
      "loss": 0.6751,
      "step": 30754
    },
    {
      "epoch": 0.8861580130236847,
      "grad_norm": 0.6927692294120789,
      "learning_rate": 3.3604432107194e-07,
      "loss": 0.6664,
      "step": 30755
    },
    {
      "epoch": 0.886186826485334,
      "grad_norm": 0.7165449857711792,
      "learning_rate": 3.3587616697648095e-07,
      "loss": 0.693,
      "step": 30756
    },
    {
      "epoch": 0.8862156399469833,
      "grad_norm": 0.7165740132331848,
      "learning_rate": 3.3570805350087775e-07,
      "loss": 0.68,
      "step": 30757
    },
    {
      "epoch": 0.8862444534086326,
      "grad_norm": 0.7209899425506592,
      "learning_rate": 3.3553998064659534e-07,
      "loss": 0.6614,
      "step": 30758
    },
    {
      "epoch": 0.8862732668702819,
      "grad_norm": 0.7427619695663452,
      "learning_rate": 3.3537194841509743e-07,
      "loss": 0.6849,
      "step": 30759
    },
    {
      "epoch": 0.886302080331931,
      "grad_norm": 0.7357274889945984,
      "learning_rate": 3.352039568078458e-07,
      "loss": 0.6484,
      "step": 30760
    },
    {
      "epoch": 0.8863308937935803,
      "grad_norm": 0.7012276649475098,
      "learning_rate": 3.350360058263058e-07,
      "loss": 0.6828,
      "step": 30761
    },
    {
      "epoch": 0.8863597072552296,
      "grad_norm": 0.738876223564148,
      "learning_rate": 3.34868095471938e-07,
      "loss": 0.6615,
      "step": 30762
    },
    {
      "epoch": 0.8863885207168789,
      "grad_norm": 0.7571622133255005,
      "learning_rate": 3.3470022574620677e-07,
      "loss": 0.6753,
      "step": 30763
    },
    {
      "epoch": 0.8864173341785282,
      "grad_norm": 0.7255525588989258,
      "learning_rate": 3.3453239665057205e-07,
      "loss": 0.7027,
      "step": 30764
    },
    {
      "epoch": 0.8864461476401775,
      "grad_norm": 0.7199150323867798,
      "learning_rate": 3.343646081864971e-07,
      "loss": 0.6971,
      "step": 30765
    },
    {
      "epoch": 0.8864749611018268,
      "grad_norm": 0.7390919327735901,
      "learning_rate": 3.341968603554413e-07,
      "loss": 0.6879,
      "step": 30766
    },
    {
      "epoch": 0.8865037745634761,
      "grad_norm": 0.7286495566368103,
      "learning_rate": 3.340291531588674e-07,
      "loss": 0.668,
      "step": 30767
    },
    {
      "epoch": 0.8865325880251254,
      "grad_norm": 0.7459454536437988,
      "learning_rate": 3.338614865982348e-07,
      "loss": 0.6831,
      "step": 30768
    },
    {
      "epoch": 0.8865614014867746,
      "grad_norm": 0.7126321196556091,
      "learning_rate": 3.3369386067500563e-07,
      "loss": 0.6846,
      "step": 30769
    },
    {
      "epoch": 0.8865902149484239,
      "grad_norm": 0.7421197295188904,
      "learning_rate": 3.3352627539063707e-07,
      "loss": 0.6987,
      "step": 30770
    },
    {
      "epoch": 0.8866190284100732,
      "grad_norm": 0.7192826867103577,
      "learning_rate": 3.3335873074659074e-07,
      "loss": 0.694,
      "step": 30771
    },
    {
      "epoch": 0.8866478418717225,
      "grad_norm": 0.7162178754806519,
      "learning_rate": 3.331912267443238e-07,
      "loss": 0.685,
      "step": 30772
    },
    {
      "epoch": 0.8866766553333717,
      "grad_norm": 0.7254626750946045,
      "learning_rate": 3.3302376338529786e-07,
      "loss": 0.6821,
      "step": 30773
    },
    {
      "epoch": 0.886705468795021,
      "grad_norm": 0.7275388240814209,
      "learning_rate": 3.328563406709678e-07,
      "loss": 0.6904,
      "step": 30774
    },
    {
      "epoch": 0.8867342822566703,
      "grad_norm": 0.7170333862304688,
      "learning_rate": 3.3268895860279425e-07,
      "loss": 0.6748,
      "step": 30775
    },
    {
      "epoch": 0.8867630957183196,
      "grad_norm": 0.7070068120956421,
      "learning_rate": 3.3252161718223486e-07,
      "loss": 0.681,
      "step": 30776
    },
    {
      "epoch": 0.8867919091799689,
      "grad_norm": 0.72735196352005,
      "learning_rate": 3.3235431641074567e-07,
      "loss": 0.6611,
      "step": 30777
    },
    {
      "epoch": 0.8868207226416182,
      "grad_norm": 0.7194053530693054,
      "learning_rate": 3.3218705628978556e-07,
      "loss": 0.6819,
      "step": 30778
    },
    {
      "epoch": 0.8868495361032674,
      "grad_norm": 0.718666136264801,
      "learning_rate": 3.320198368208083e-07,
      "loss": 0.6728,
      "step": 30779
    },
    {
      "epoch": 0.8868783495649167,
      "grad_norm": 0.7209991216659546,
      "learning_rate": 3.318526580052733e-07,
      "loss": 0.6767,
      "step": 30780
    },
    {
      "epoch": 0.886907163026566,
      "grad_norm": 0.7514073252677917,
      "learning_rate": 3.316855198446345e-07,
      "loss": 0.6744,
      "step": 30781
    },
    {
      "epoch": 0.8869359764882153,
      "grad_norm": 0.7216097116470337,
      "learning_rate": 3.3151842234034947e-07,
      "loss": 0.6926,
      "step": 30782
    },
    {
      "epoch": 0.8869647899498646,
      "grad_norm": 0.7457411289215088,
      "learning_rate": 3.31351365493871e-07,
      "loss": 0.7108,
      "step": 30783
    },
    {
      "epoch": 0.8869936034115139,
      "grad_norm": 0.733187198638916,
      "learning_rate": 3.3118434930665576e-07,
      "loss": 0.6721,
      "step": 30784
    },
    {
      "epoch": 0.8870224168731632,
      "grad_norm": 0.7042309641838074,
      "learning_rate": 3.3101737378015753e-07,
      "loss": 0.6662,
      "step": 30785
    },
    {
      "epoch": 0.8870512303348125,
      "grad_norm": 0.731867790222168,
      "learning_rate": 3.308504389158312e-07,
      "loss": 0.6928,
      "step": 30786
    },
    {
      "epoch": 0.8870800437964617,
      "grad_norm": 0.7207564115524292,
      "learning_rate": 3.306835447151291e-07,
      "loss": 0.6735,
      "step": 30787
    },
    {
      "epoch": 0.887108857258111,
      "grad_norm": 0.7245121598243713,
      "learning_rate": 3.305166911795066e-07,
      "loss": 0.6752,
      "step": 30788
    },
    {
      "epoch": 0.8871376707197602,
      "grad_norm": 0.7106683850288391,
      "learning_rate": 3.303498783104153e-07,
      "loss": 0.666,
      "step": 30789
    },
    {
      "epoch": 0.8871664841814095,
      "grad_norm": 0.7255218029022217,
      "learning_rate": 3.301831061093086e-07,
      "loss": 0.6673,
      "step": 30790
    },
    {
      "epoch": 0.8871952976430588,
      "grad_norm": 0.7199922204017639,
      "learning_rate": 3.300163745776391e-07,
      "loss": 0.6801,
      "step": 30791
    },
    {
      "epoch": 0.8872241111047081,
      "grad_norm": 0.7152511477470398,
      "learning_rate": 3.298496837168591e-07,
      "loss": 0.6699,
      "step": 30792
    },
    {
      "epoch": 0.8872529245663574,
      "grad_norm": 0.7600508332252502,
      "learning_rate": 3.2968303352841956e-07,
      "loss": 0.6914,
      "step": 30793
    },
    {
      "epoch": 0.8872817380280067,
      "grad_norm": 0.7236932516098022,
      "learning_rate": 3.295164240137727e-07,
      "loss": 0.6817,
      "step": 30794
    },
    {
      "epoch": 0.887310551489656,
      "grad_norm": 0.7346088886260986,
      "learning_rate": 3.293498551743679e-07,
      "loss": 0.6747,
      "step": 30795
    },
    {
      "epoch": 0.8873393649513053,
      "grad_norm": 0.7230314612388611,
      "learning_rate": 3.2918332701165733e-07,
      "loss": 0.6734,
      "step": 30796
    },
    {
      "epoch": 0.8873681784129546,
      "grad_norm": 0.7127465605735779,
      "learning_rate": 3.2901683952709207e-07,
      "loss": 0.6769,
      "step": 30797
    },
    {
      "epoch": 0.8873969918746039,
      "grad_norm": 0.7289774417877197,
      "learning_rate": 3.288503927221193e-07,
      "loss": 0.6928,
      "step": 30798
    },
    {
      "epoch": 0.8874258053362531,
      "grad_norm": 0.7352847456932068,
      "learning_rate": 3.2868398659819114e-07,
      "loss": 0.6845,
      "step": 30799
    },
    {
      "epoch": 0.8874546187979023,
      "grad_norm": 0.7437534332275391,
      "learning_rate": 3.2851762115675534e-07,
      "loss": 0.6897,
      "step": 30800
    },
    {
      "epoch": 0.8874834322595516,
      "grad_norm": 0.7060685157775879,
      "learning_rate": 3.283512963992619e-07,
      "loss": 0.6596,
      "step": 30801
    },
    {
      "epoch": 0.8875122457212009,
      "grad_norm": 0.6992811560630798,
      "learning_rate": 3.2818501232715794e-07,
      "loss": 0.6877,
      "step": 30802
    },
    {
      "epoch": 0.8875410591828502,
      "grad_norm": 0.7162731289863586,
      "learning_rate": 3.280187689418929e-07,
      "loss": 0.7118,
      "step": 30803
    },
    {
      "epoch": 0.8875698726444995,
      "grad_norm": 0.7275088429450989,
      "learning_rate": 3.2785256624491333e-07,
      "loss": 0.7098,
      "step": 30804
    },
    {
      "epoch": 0.8875986861061488,
      "grad_norm": 0.7314105033874512,
      "learning_rate": 3.2768640423766873e-07,
      "loss": 0.6753,
      "step": 30805
    },
    {
      "epoch": 0.8876274995677981,
      "grad_norm": 0.7176274061203003,
      "learning_rate": 3.27520282921604e-07,
      "loss": 0.6837,
      "step": 30806
    },
    {
      "epoch": 0.8876563130294474,
      "grad_norm": 0.7198130488395691,
      "learning_rate": 3.2735420229816796e-07,
      "loss": 0.6796,
      "step": 30807
    },
    {
      "epoch": 0.8876851264910967,
      "grad_norm": 0.7410292029380798,
      "learning_rate": 3.2718816236880445e-07,
      "loss": 0.6828,
      "step": 30808
    },
    {
      "epoch": 0.887713939952746,
      "grad_norm": 0.7063465118408203,
      "learning_rate": 3.270221631349624e-07,
      "loss": 0.6711,
      "step": 30809
    },
    {
      "epoch": 0.8877427534143952,
      "grad_norm": 0.7537160515785217,
      "learning_rate": 3.268562045980844e-07,
      "loss": 0.6804,
      "step": 30810
    },
    {
      "epoch": 0.8877715668760445,
      "grad_norm": 0.7149539589881897,
      "learning_rate": 3.266902867596189e-07,
      "loss": 0.6777,
      "step": 30811
    },
    {
      "epoch": 0.8878003803376938,
      "grad_norm": 0.7465508580207825,
      "learning_rate": 3.265244096210085e-07,
      "loss": 0.7027,
      "step": 30812
    },
    {
      "epoch": 0.8878291937993431,
      "grad_norm": 0.7459323406219482,
      "learning_rate": 3.2635857318369877e-07,
      "loss": 0.6995,
      "step": 30813
    },
    {
      "epoch": 0.8878580072609923,
      "grad_norm": 0.7247650623321533,
      "learning_rate": 3.2619277744913467e-07,
      "loss": 0.7064,
      "step": 30814
    },
    {
      "epoch": 0.8878868207226416,
      "grad_norm": 0.7307755947113037,
      "learning_rate": 3.260270224187584e-07,
      "loss": 0.6585,
      "step": 30815
    },
    {
      "epoch": 0.8879156341842909,
      "grad_norm": 0.7171160578727722,
      "learning_rate": 3.2586130809401485e-07,
      "loss": 0.6745,
      "step": 30816
    },
    {
      "epoch": 0.8879444476459402,
      "grad_norm": 0.7196224331855774,
      "learning_rate": 3.2569563447634677e-07,
      "loss": 0.7052,
      "step": 30817
    },
    {
      "epoch": 0.8879732611075895,
      "grad_norm": 0.720127522945404,
      "learning_rate": 3.255300015671975e-07,
      "loss": 0.679,
      "step": 30818
    },
    {
      "epoch": 0.8880020745692387,
      "grad_norm": 0.7222316265106201,
      "learning_rate": 3.253644093680086e-07,
      "loss": 0.6753,
      "step": 30819
    },
    {
      "epoch": 0.888030888030888,
      "grad_norm": 0.7002360224723816,
      "learning_rate": 3.25198857880224e-07,
      "loss": 0.6546,
      "step": 30820
    },
    {
      "epoch": 0.8880597014925373,
      "grad_norm": 0.7247027158737183,
      "learning_rate": 3.2503334710528303e-07,
      "loss": 0.678,
      "step": 30821
    },
    {
      "epoch": 0.8880885149541866,
      "grad_norm": 0.7326758503913879,
      "learning_rate": 3.2486787704462895e-07,
      "loss": 0.6746,
      "step": 30822
    },
    {
      "epoch": 0.8881173284158359,
      "grad_norm": 0.7568331956863403,
      "learning_rate": 3.247024476997018e-07,
      "loss": 0.6833,
      "step": 30823
    },
    {
      "epoch": 0.8881461418774852,
      "grad_norm": 0.7277394533157349,
      "learning_rate": 3.245370590719438e-07,
      "loss": 0.6599,
      "step": 30824
    },
    {
      "epoch": 0.8881749553391345,
      "grad_norm": 0.7240405678749084,
      "learning_rate": 3.243717111627931e-07,
      "loss": 0.6887,
      "step": 30825
    },
    {
      "epoch": 0.8882037688007838,
      "grad_norm": 0.7212233543395996,
      "learning_rate": 3.242064039736914e-07,
      "loss": 0.682,
      "step": 30826
    },
    {
      "epoch": 0.888232582262433,
      "grad_norm": 0.7002923488616943,
      "learning_rate": 3.240411375060765e-07,
      "loss": 0.6782,
      "step": 30827
    },
    {
      "epoch": 0.8882613957240822,
      "grad_norm": 0.7148131728172302,
      "learning_rate": 3.23875911761391e-07,
      "loss": 0.6665,
      "step": 30828
    },
    {
      "epoch": 0.8882902091857315,
      "grad_norm": 0.7234336137771606,
      "learning_rate": 3.237107267410705e-07,
      "loss": 0.6804,
      "step": 30829
    },
    {
      "epoch": 0.8883190226473808,
      "grad_norm": 0.7211416363716125,
      "learning_rate": 3.235455824465561e-07,
      "loss": 0.6743,
      "step": 30830
    },
    {
      "epoch": 0.8883478361090301,
      "grad_norm": 0.7131649255752563,
      "learning_rate": 3.233804788792844e-07,
      "loss": 0.6645,
      "step": 30831
    },
    {
      "epoch": 0.8883766495706794,
      "grad_norm": 0.7235488891601562,
      "learning_rate": 3.2321541604069416e-07,
      "loss": 0.6689,
      "step": 30832
    },
    {
      "epoch": 0.8884054630323287,
      "grad_norm": 0.7382998466491699,
      "learning_rate": 3.2305039393222214e-07,
      "loss": 0.7068,
      "step": 30833
    },
    {
      "epoch": 0.888434276493978,
      "grad_norm": 0.7531034350395203,
      "learning_rate": 3.2288541255530546e-07,
      "loss": 0.6858,
      "step": 30834
    },
    {
      "epoch": 0.8884630899556273,
      "grad_norm": 0.7180628776550293,
      "learning_rate": 3.2272047191138245e-07,
      "loss": 0.6655,
      "step": 30835
    },
    {
      "epoch": 0.8884919034172766,
      "grad_norm": 0.7346670627593994,
      "learning_rate": 3.22555572001888e-07,
      "loss": 0.666,
      "step": 30836
    },
    {
      "epoch": 0.8885207168789259,
      "grad_norm": 0.7369222640991211,
      "learning_rate": 3.223907128282594e-07,
      "loss": 0.6951,
      "step": 30837
    },
    {
      "epoch": 0.8885495303405752,
      "grad_norm": 0.7263765335083008,
      "learning_rate": 3.2222589439193096e-07,
      "loss": 0.67,
      "step": 30838
    },
    {
      "epoch": 0.8885783438022244,
      "grad_norm": 0.7287716865539551,
      "learning_rate": 3.2206111669433883e-07,
      "loss": 0.6673,
      "step": 30839
    },
    {
      "epoch": 0.8886071572638737,
      "grad_norm": 0.7198176980018616,
      "learning_rate": 3.2189637973691844e-07,
      "loss": 0.6923,
      "step": 30840
    },
    {
      "epoch": 0.8886359707255229,
      "grad_norm": 0.7293341755867004,
      "learning_rate": 3.2173168352110486e-07,
      "loss": 0.6742,
      "step": 30841
    },
    {
      "epoch": 0.8886647841871722,
      "grad_norm": 0.750541627407074,
      "learning_rate": 3.2156702804833075e-07,
      "loss": 0.7105,
      "step": 30842
    },
    {
      "epoch": 0.8886935976488215,
      "grad_norm": 0.7327432036399841,
      "learning_rate": 3.214024133200322e-07,
      "loss": 0.6789,
      "step": 30843
    },
    {
      "epoch": 0.8887224111104708,
      "grad_norm": 0.7218421101570129,
      "learning_rate": 3.2123783933764086e-07,
      "loss": 0.6845,
      "step": 30844
    },
    {
      "epoch": 0.8887512245721201,
      "grad_norm": 0.7362237572669983,
      "learning_rate": 3.210733061025917e-07,
      "loss": 0.6918,
      "step": 30845
    },
    {
      "epoch": 0.8887800380337694,
      "grad_norm": 0.7152573466300964,
      "learning_rate": 3.2090881361631633e-07,
      "loss": 0.6969,
      "step": 30846
    },
    {
      "epoch": 0.8888088514954187,
      "grad_norm": 0.7196869254112244,
      "learning_rate": 3.2074436188024806e-07,
      "loss": 0.6709,
      "step": 30847
    },
    {
      "epoch": 0.888837664957068,
      "grad_norm": 0.7342898845672607,
      "learning_rate": 3.205799508958185e-07,
      "loss": 0.7021,
      "step": 30848
    },
    {
      "epoch": 0.8888664784187172,
      "grad_norm": 0.7276679277420044,
      "learning_rate": 3.2041558066445935e-07,
      "loss": 0.6811,
      "step": 30849
    },
    {
      "epoch": 0.8888952918803665,
      "grad_norm": 0.7088930010795593,
      "learning_rate": 3.2025125118760383e-07,
      "loss": 0.671,
      "step": 30850
    },
    {
      "epoch": 0.8889241053420158,
      "grad_norm": 0.7352321743965149,
      "learning_rate": 3.200869624666808e-07,
      "loss": 0.6802,
      "step": 30851
    },
    {
      "epoch": 0.8889529188036651,
      "grad_norm": 0.7479607462882996,
      "learning_rate": 3.199227145031225e-07,
      "loss": 0.683,
      "step": 30852
    },
    {
      "epoch": 0.8889817322653144,
      "grad_norm": 0.7357273101806641,
      "learning_rate": 3.1975850729835945e-07,
      "loss": 0.691,
      "step": 30853
    },
    {
      "epoch": 0.8890105457269636,
      "grad_norm": 0.701274037361145,
      "learning_rate": 3.1959434085382047e-07,
      "loss": 0.6736,
      "step": 30854
    },
    {
      "epoch": 0.8890393591886129,
      "grad_norm": 0.710081160068512,
      "learning_rate": 3.194302151709361e-07,
      "loss": 0.6689,
      "step": 30855
    },
    {
      "epoch": 0.8890681726502622,
      "grad_norm": 0.7337176203727722,
      "learning_rate": 3.192661302511368e-07,
      "loss": 0.6792,
      "step": 30856
    },
    {
      "epoch": 0.8890969861119115,
      "grad_norm": 0.7667778134346008,
      "learning_rate": 3.1910208609584934e-07,
      "loss": 0.674,
      "step": 30857
    },
    {
      "epoch": 0.8891257995735607,
      "grad_norm": 0.7370890378952026,
      "learning_rate": 3.1893808270650473e-07,
      "loss": 0.6814,
      "step": 30858
    },
    {
      "epoch": 0.88915461303521,
      "grad_norm": 0.7051439881324768,
      "learning_rate": 3.18774120084529e-07,
      "loss": 0.6658,
      "step": 30859
    },
    {
      "epoch": 0.8891834264968593,
      "grad_norm": 0.7445203065872192,
      "learning_rate": 3.186101982313522e-07,
      "loss": 0.6638,
      "step": 30860
    },
    {
      "epoch": 0.8892122399585086,
      "grad_norm": 0.7456121444702148,
      "learning_rate": 3.184463171483998e-07,
      "loss": 0.7058,
      "step": 30861
    },
    {
      "epoch": 0.8892410534201579,
      "grad_norm": 0.7162070870399475,
      "learning_rate": 3.182824768371007e-07,
      "loss": 0.6803,
      "step": 30862
    },
    {
      "epoch": 0.8892698668818072,
      "grad_norm": 0.7414777278900146,
      "learning_rate": 3.1811867729888035e-07,
      "loss": 0.6885,
      "step": 30863
    },
    {
      "epoch": 0.8892986803434565,
      "grad_norm": 0.7148325443267822,
      "learning_rate": 3.1795491853516713e-07,
      "loss": 0.6612,
      "step": 30864
    },
    {
      "epoch": 0.8893274938051058,
      "grad_norm": 0.7262104749679565,
      "learning_rate": 3.1779120054738546e-07,
      "loss": 0.6924,
      "step": 30865
    },
    {
      "epoch": 0.8893563072667551,
      "grad_norm": 0.7424269914627075,
      "learning_rate": 3.1762752333696303e-07,
      "loss": 0.6834,
      "step": 30866
    },
    {
      "epoch": 0.8893851207284044,
      "grad_norm": 0.7242242097854614,
      "learning_rate": 3.1746388690532316e-07,
      "loss": 0.6737,
      "step": 30867
    },
    {
      "epoch": 0.8894139341900535,
      "grad_norm": 0.7330954670906067,
      "learning_rate": 3.1730029125389306e-07,
      "loss": 0.6723,
      "step": 30868
    },
    {
      "epoch": 0.8894427476517028,
      "grad_norm": 0.7252020239830017,
      "learning_rate": 3.1713673638409547e-07,
      "loss": 0.6873,
      "step": 30869
    },
    {
      "epoch": 0.8894715611133521,
      "grad_norm": 0.7576206922531128,
      "learning_rate": 3.1697322229735706e-07,
      "loss": 0.6991,
      "step": 30870
    },
    {
      "epoch": 0.8895003745750014,
      "grad_norm": 0.7059981226921082,
      "learning_rate": 3.168097489950989e-07,
      "loss": 0.6511,
      "step": 30871
    },
    {
      "epoch": 0.8895291880366507,
      "grad_norm": 0.7217310667037964,
      "learning_rate": 3.16646316478747e-07,
      "loss": 0.6762,
      "step": 30872
    },
    {
      "epoch": 0.8895580014983,
      "grad_norm": 0.7288281321525574,
      "learning_rate": 3.1648292474972475e-07,
      "loss": 0.67,
      "step": 30873
    },
    {
      "epoch": 0.8895868149599493,
      "grad_norm": 0.7475136518478394,
      "learning_rate": 3.163195738094532e-07,
      "loss": 0.6845,
      "step": 30874
    },
    {
      "epoch": 0.8896156284215986,
      "grad_norm": 0.7463434934616089,
      "learning_rate": 3.161562636593568e-07,
      "loss": 0.6849,
      "step": 30875
    },
    {
      "epoch": 0.8896444418832479,
      "grad_norm": 0.736311674118042,
      "learning_rate": 3.159929943008566e-07,
      "loss": 0.6926,
      "step": 30876
    },
    {
      "epoch": 0.8896732553448972,
      "grad_norm": 0.7239062190055847,
      "learning_rate": 3.1582976573537647e-07,
      "loss": 0.6831,
      "step": 30877
    },
    {
      "epoch": 0.8897020688065465,
      "grad_norm": 0.7252891063690186,
      "learning_rate": 3.156665779643359e-07,
      "loss": 0.6862,
      "step": 30878
    },
    {
      "epoch": 0.8897308822681957,
      "grad_norm": 0.7026932835578918,
      "learning_rate": 3.155034309891569e-07,
      "loss": 0.6863,
      "step": 30879
    },
    {
      "epoch": 0.889759695729845,
      "grad_norm": 0.7187500596046448,
      "learning_rate": 3.153403248112602e-07,
      "loss": 0.6726,
      "step": 30880
    },
    {
      "epoch": 0.8897885091914942,
      "grad_norm": 0.6998257040977478,
      "learning_rate": 3.151772594320668e-07,
      "loss": 0.6551,
      "step": 30881
    },
    {
      "epoch": 0.8898173226531435,
      "grad_norm": 0.7044333815574646,
      "learning_rate": 3.1501423485299555e-07,
      "loss": 0.6818,
      "step": 30882
    },
    {
      "epoch": 0.8898461361147928,
      "grad_norm": 0.7301635146141052,
      "learning_rate": 3.1485125107546754e-07,
      "loss": 0.6858,
      "step": 30883
    },
    {
      "epoch": 0.8898749495764421,
      "grad_norm": 0.7256763577461243,
      "learning_rate": 3.1468830810090113e-07,
      "loss": 0.6674,
      "step": 30884
    },
    {
      "epoch": 0.8899037630380914,
      "grad_norm": 0.7120108604431152,
      "learning_rate": 3.145254059307168e-07,
      "loss": 0.6989,
      "step": 30885
    },
    {
      "epoch": 0.8899325764997407,
      "grad_norm": 0.7291933298110962,
      "learning_rate": 3.1436254456633177e-07,
      "loss": 0.6755,
      "step": 30886
    },
    {
      "epoch": 0.88996138996139,
      "grad_norm": 0.7118679881095886,
      "learning_rate": 3.141997240091649e-07,
      "loss": 0.6754,
      "step": 30887
    },
    {
      "epoch": 0.8899902034230393,
      "grad_norm": 0.7190098166465759,
      "learning_rate": 3.140369442606345e-07,
      "loss": 0.6834,
      "step": 30888
    },
    {
      "epoch": 0.8900190168846885,
      "grad_norm": 0.7476838827133179,
      "learning_rate": 3.1387420532215887e-07,
      "loss": 0.7048,
      "step": 30889
    },
    {
      "epoch": 0.8900478303463378,
      "grad_norm": 0.7153657674789429,
      "learning_rate": 3.137115071951535e-07,
      "loss": 0.6678,
      "step": 30890
    },
    {
      "epoch": 0.8900766438079871,
      "grad_norm": 0.7134442329406738,
      "learning_rate": 3.135488498810368e-07,
      "loss": 0.6673,
      "step": 30891
    },
    {
      "epoch": 0.8901054572696364,
      "grad_norm": 0.7282447218894958,
      "learning_rate": 3.133862333812249e-07,
      "loss": 0.6815,
      "step": 30892
    },
    {
      "epoch": 0.8901342707312857,
      "grad_norm": 0.729376494884491,
      "learning_rate": 3.1322365769713313e-07,
      "loss": 0.7041,
      "step": 30893
    },
    {
      "epoch": 0.890163084192935,
      "grad_norm": 0.7397831678390503,
      "learning_rate": 3.1306112283018e-07,
      "loss": 0.6764,
      "step": 30894
    },
    {
      "epoch": 0.8901918976545842,
      "grad_norm": 0.743568480014801,
      "learning_rate": 3.128986287817776e-07,
      "loss": 0.6583,
      "step": 30895
    },
    {
      "epoch": 0.8902207111162335,
      "grad_norm": 0.7002007961273193,
      "learning_rate": 3.1273617555334426e-07,
      "loss": 0.6591,
      "step": 30896
    },
    {
      "epoch": 0.8902495245778828,
      "grad_norm": 0.7351205348968506,
      "learning_rate": 3.1257376314629216e-07,
      "loss": 0.6754,
      "step": 30897
    },
    {
      "epoch": 0.890278338039532,
      "grad_norm": 0.6947964429855347,
      "learning_rate": 3.124113915620375e-07,
      "loss": 0.6565,
      "step": 30898
    },
    {
      "epoch": 0.8903071515011813,
      "grad_norm": 0.7365518808364868,
      "learning_rate": 3.122490608019924e-07,
      "loss": 0.6962,
      "step": 30899
    },
    {
      "epoch": 0.8903359649628306,
      "grad_norm": 0.7159032225608826,
      "learning_rate": 3.1208677086757346e-07,
      "loss": 0.6859,
      "step": 30900
    },
    {
      "epoch": 0.8903647784244799,
      "grad_norm": 0.724920392036438,
      "learning_rate": 3.1192452176019184e-07,
      "loss": 0.6666,
      "step": 30901
    },
    {
      "epoch": 0.8903935918861292,
      "grad_norm": 0.747799277305603,
      "learning_rate": 3.1176231348126195e-07,
      "loss": 0.6887,
      "step": 30902
    },
    {
      "epoch": 0.8904224053477785,
      "grad_norm": 0.7345032095909119,
      "learning_rate": 3.116001460321949e-07,
      "loss": 0.7127,
      "step": 30903
    },
    {
      "epoch": 0.8904512188094278,
      "grad_norm": 0.7250088453292847,
      "learning_rate": 3.1143801941440457e-07,
      "loss": 0.6887,
      "step": 30904
    },
    {
      "epoch": 0.8904800322710771,
      "grad_norm": 0.7135794162750244,
      "learning_rate": 3.11275933629302e-07,
      "loss": 0.6772,
      "step": 30905
    },
    {
      "epoch": 0.8905088457327264,
      "grad_norm": 0.7121812105178833,
      "learning_rate": 3.111138886782994e-07,
      "loss": 0.6821,
      "step": 30906
    },
    {
      "epoch": 0.8905376591943757,
      "grad_norm": 0.734392523765564,
      "learning_rate": 3.109518845628068e-07,
      "loss": 0.679,
      "step": 30907
    },
    {
      "epoch": 0.8905664726560248,
      "grad_norm": 0.7121101021766663,
      "learning_rate": 3.107899212842369e-07,
      "loss": 0.6819,
      "step": 30908
    },
    {
      "epoch": 0.8905952861176741,
      "grad_norm": 0.7279629707336426,
      "learning_rate": 3.106279988439986e-07,
      "loss": 0.6589,
      "step": 30909
    },
    {
      "epoch": 0.8906240995793234,
      "grad_norm": 0.761985719203949,
      "learning_rate": 3.104661172435025e-07,
      "loss": 0.6933,
      "step": 30910
    },
    {
      "epoch": 0.8906529130409727,
      "grad_norm": 0.7260406613349915,
      "learning_rate": 3.103042764841591e-07,
      "loss": 0.691,
      "step": 30911
    },
    {
      "epoch": 0.890681726502622,
      "grad_norm": 0.7190552949905396,
      "learning_rate": 3.101424765673783e-07,
      "loss": 0.6679,
      "step": 30912
    },
    {
      "epoch": 0.8907105399642713,
      "grad_norm": 0.7487507462501526,
      "learning_rate": 3.099807174945668e-07,
      "loss": 0.7242,
      "step": 30913
    },
    {
      "epoch": 0.8907393534259206,
      "grad_norm": 0.8271962404251099,
      "learning_rate": 3.0981899926713574e-07,
      "loss": 0.6892,
      "step": 30914
    },
    {
      "epoch": 0.8907681668875699,
      "grad_norm": 0.7162754535675049,
      "learning_rate": 3.0965732188649346e-07,
      "loss": 0.6711,
      "step": 30915
    },
    {
      "epoch": 0.8907969803492192,
      "grad_norm": 0.7238566279411316,
      "learning_rate": 3.0949568535404593e-07,
      "loss": 0.6684,
      "step": 30916
    },
    {
      "epoch": 0.8908257938108685,
      "grad_norm": 0.7547336220741272,
      "learning_rate": 3.093340896712038e-07,
      "loss": 0.6821,
      "step": 30917
    },
    {
      "epoch": 0.8908546072725178,
      "grad_norm": 0.7176944017410278,
      "learning_rate": 3.0917253483937147e-07,
      "loss": 0.7019,
      "step": 30918
    },
    {
      "epoch": 0.890883420734167,
      "grad_norm": 0.730974018573761,
      "learning_rate": 3.090110208599578e-07,
      "loss": 0.6981,
      "step": 30919
    },
    {
      "epoch": 0.8909122341958163,
      "grad_norm": 0.7194978594779968,
      "learning_rate": 3.0884954773436895e-07,
      "loss": 0.6788,
      "step": 30920
    },
    {
      "epoch": 0.8909410476574655,
      "grad_norm": 0.7197388410568237,
      "learning_rate": 3.086881154640109e-07,
      "loss": 0.6777,
      "step": 30921
    },
    {
      "epoch": 0.8909698611191148,
      "grad_norm": 0.7338636517524719,
      "learning_rate": 3.0852672405028983e-07,
      "loss": 0.6873,
      "step": 30922
    },
    {
      "epoch": 0.8909986745807641,
      "grad_norm": 0.874696671962738,
      "learning_rate": 3.083653734946107e-07,
      "loss": 0.6926,
      "step": 30923
    },
    {
      "epoch": 0.8910274880424134,
      "grad_norm": 0.778346061706543,
      "learning_rate": 3.0820406379837963e-07,
      "loss": 0.6723,
      "step": 30924
    },
    {
      "epoch": 0.8910563015040627,
      "grad_norm": 0.71797114610672,
      "learning_rate": 3.080427949630016e-07,
      "loss": 0.6762,
      "step": 30925
    },
    {
      "epoch": 0.891085114965712,
      "grad_norm": 0.7147598266601562,
      "learning_rate": 3.078815669898799e-07,
      "loss": 0.6783,
      "step": 30926
    },
    {
      "epoch": 0.8911139284273613,
      "grad_norm": 0.7361458539962769,
      "learning_rate": 3.077203798804196e-07,
      "loss": 0.686,
      "step": 30927
    },
    {
      "epoch": 0.8911427418890105,
      "grad_norm": 0.7195905447006226,
      "learning_rate": 3.075592336360239e-07,
      "loss": 0.6716,
      "step": 30928
    },
    {
      "epoch": 0.8911715553506598,
      "grad_norm": 0.7145978212356567,
      "learning_rate": 3.0739812825809787e-07,
      "loss": 0.6644,
      "step": 30929
    },
    {
      "epoch": 0.8912003688123091,
      "grad_norm": 0.7329170107841492,
      "learning_rate": 3.072370637480415e-07,
      "loss": 0.6546,
      "step": 30930
    },
    {
      "epoch": 0.8912291822739584,
      "grad_norm": 0.6962326765060425,
      "learning_rate": 3.070760401072598e-07,
      "loss": 0.6484,
      "step": 30931
    },
    {
      "epoch": 0.8912579957356077,
      "grad_norm": 0.7362666130065918,
      "learning_rate": 3.0691505733715544e-07,
      "loss": 0.6845,
      "step": 30932
    },
    {
      "epoch": 0.891286809197257,
      "grad_norm": 0.7460411787033081,
      "learning_rate": 3.0675411543912847e-07,
      "loss": 0.6755,
      "step": 30933
    },
    {
      "epoch": 0.8913156226589063,
      "grad_norm": 0.6974814534187317,
      "learning_rate": 3.0659321441458277e-07,
      "loss": 0.6723,
      "step": 30934
    },
    {
      "epoch": 0.8913444361205555,
      "grad_norm": 0.7145587205886841,
      "learning_rate": 3.064323542649167e-07,
      "loss": 0.6854,
      "step": 30935
    },
    {
      "epoch": 0.8913732495822048,
      "grad_norm": 0.7146954536437988,
      "learning_rate": 3.062715349915346e-07,
      "loss": 0.672,
      "step": 30936
    },
    {
      "epoch": 0.891402063043854,
      "grad_norm": 0.7496393918991089,
      "learning_rate": 3.0611075659583434e-07,
      "loss": 0.7032,
      "step": 30937
    },
    {
      "epoch": 0.8914308765055033,
      "grad_norm": 0.7047626972198486,
      "learning_rate": 3.059500190792186e-07,
      "loss": 0.6485,
      "step": 30938
    },
    {
      "epoch": 0.8914596899671526,
      "grad_norm": 0.7004858255386353,
      "learning_rate": 3.057893224430847e-07,
      "loss": 0.6514,
      "step": 30939
    },
    {
      "epoch": 0.8914885034288019,
      "grad_norm": 0.7115792036056519,
      "learning_rate": 3.056286666888342e-07,
      "loss": 0.6897,
      "step": 30940
    },
    {
      "epoch": 0.8915173168904512,
      "grad_norm": 0.710146963596344,
      "learning_rate": 3.0546805181786433e-07,
      "loss": 0.688,
      "step": 30941
    },
    {
      "epoch": 0.8915461303521005,
      "grad_norm": 0.7149070501327515,
      "learning_rate": 3.053074778315757e-07,
      "loss": 0.681,
      "step": 30942
    },
    {
      "epoch": 0.8915749438137498,
      "grad_norm": 0.7487267255783081,
      "learning_rate": 3.051469447313654e-07,
      "loss": 0.6922,
      "step": 30943
    },
    {
      "epoch": 0.8916037572753991,
      "grad_norm": 0.6957665085792542,
      "learning_rate": 3.04986452518633e-07,
      "loss": 0.6596,
      "step": 30944
    },
    {
      "epoch": 0.8916325707370484,
      "grad_norm": 0.7607535123825073,
      "learning_rate": 3.04826001194774e-07,
      "loss": 0.6956,
      "step": 30945
    },
    {
      "epoch": 0.8916613841986977,
      "grad_norm": 0.7153303027153015,
      "learning_rate": 3.046655907611884e-07,
      "loss": 0.6773,
      "step": 30946
    },
    {
      "epoch": 0.891690197660347,
      "grad_norm": 0.7279678583145142,
      "learning_rate": 3.045052212192701e-07,
      "loss": 0.6844,
      "step": 30947
    },
    {
      "epoch": 0.8917190111219961,
      "grad_norm": 0.7313603758811951,
      "learning_rate": 3.04344892570419e-07,
      "loss": 0.6999,
      "step": 30948
    },
    {
      "epoch": 0.8917478245836454,
      "grad_norm": 0.7205615043640137,
      "learning_rate": 3.041846048160296e-07,
      "loss": 0.6635,
      "step": 30949
    },
    {
      "epoch": 0.8917766380452947,
      "grad_norm": 0.7135816812515259,
      "learning_rate": 3.040243579574986e-07,
      "loss": 0.6838,
      "step": 30950
    },
    {
      "epoch": 0.891805451506944,
      "grad_norm": 0.7079556584358215,
      "learning_rate": 3.0386415199622043e-07,
      "loss": 0.6767,
      "step": 30951
    },
    {
      "epoch": 0.8918342649685933,
      "grad_norm": 0.7078445553779602,
      "learning_rate": 3.0370398693359114e-07,
      "loss": 0.6767,
      "step": 30952
    },
    {
      "epoch": 0.8918630784302426,
      "grad_norm": 0.7303382158279419,
      "learning_rate": 3.0354386277100635e-07,
      "loss": 0.6562,
      "step": 30953
    },
    {
      "epoch": 0.8918918918918919,
      "grad_norm": 0.735786497592926,
      "learning_rate": 3.033837795098588e-07,
      "loss": 0.6731,
      "step": 30954
    },
    {
      "epoch": 0.8919207053535412,
      "grad_norm": 0.7455115914344788,
      "learning_rate": 3.032237371515445e-07,
      "loss": 0.6726,
      "step": 30955
    },
    {
      "epoch": 0.8919495188151905,
      "grad_norm": 0.719168484210968,
      "learning_rate": 3.0306373569745583e-07,
      "loss": 0.71,
      "step": 30956
    },
    {
      "epoch": 0.8919783322768398,
      "grad_norm": 0.7097586393356323,
      "learning_rate": 3.0290377514898714e-07,
      "loss": 0.6746,
      "step": 30957
    },
    {
      "epoch": 0.892007145738489,
      "grad_norm": 0.7375597953796387,
      "learning_rate": 3.0274385550753074e-07,
      "loss": 0.6773,
      "step": 30958
    },
    {
      "epoch": 0.8920359592001383,
      "grad_norm": 0.7466867566108704,
      "learning_rate": 3.0258397677447983e-07,
      "loss": 0.6966,
      "step": 30959
    },
    {
      "epoch": 0.8920647726617876,
      "grad_norm": 0.726342499256134,
      "learning_rate": 3.024241389512267e-07,
      "loss": 0.6757,
      "step": 30960
    },
    {
      "epoch": 0.8920935861234369,
      "grad_norm": 0.7268096804618835,
      "learning_rate": 3.022643420391641e-07,
      "loss": 0.6772,
      "step": 30961
    },
    {
      "epoch": 0.8921223995850861,
      "grad_norm": 0.735464870929718,
      "learning_rate": 3.0210458603968264e-07,
      "loss": 0.6925,
      "step": 30962
    },
    {
      "epoch": 0.8921512130467354,
      "grad_norm": 0.7376399636268616,
      "learning_rate": 3.01944870954175e-07,
      "loss": 0.6866,
      "step": 30963
    },
    {
      "epoch": 0.8921800265083847,
      "grad_norm": 0.7124887704849243,
      "learning_rate": 3.017851967840296e-07,
      "loss": 0.6798,
      "step": 30964
    },
    {
      "epoch": 0.892208839970034,
      "grad_norm": 0.7475035190582275,
      "learning_rate": 3.0162556353064034e-07,
      "loss": 0.6987,
      "step": 30965
    },
    {
      "epoch": 0.8922376534316833,
      "grad_norm": 0.7287240028381348,
      "learning_rate": 3.014659711953943e-07,
      "loss": 0.6683,
      "step": 30966
    },
    {
      "epoch": 0.8922664668933326,
      "grad_norm": 0.7350240349769592,
      "learning_rate": 3.013064197796839e-07,
      "loss": 0.6701,
      "step": 30967
    },
    {
      "epoch": 0.8922952803549818,
      "grad_norm": 0.7338509559631348,
      "learning_rate": 3.0114690928489676e-07,
      "loss": 0.7044,
      "step": 30968
    },
    {
      "epoch": 0.8923240938166311,
      "grad_norm": 0.7180945873260498,
      "learning_rate": 3.0098743971242297e-07,
      "loss": 0.6739,
      "step": 30969
    },
    {
      "epoch": 0.8923529072782804,
      "grad_norm": 0.7366533279418945,
      "learning_rate": 3.008280110636519e-07,
      "loss": 0.673,
      "step": 30970
    },
    {
      "epoch": 0.8923817207399297,
      "grad_norm": 0.7026088237762451,
      "learning_rate": 3.0066862333997085e-07,
      "loss": 0.6742,
      "step": 30971
    },
    {
      "epoch": 0.892410534201579,
      "grad_norm": 0.7290716171264648,
      "learning_rate": 3.005092765427686e-07,
      "loss": 0.7019,
      "step": 30972
    },
    {
      "epoch": 0.8924393476632283,
      "grad_norm": 0.7339767813682556,
      "learning_rate": 3.0034997067343255e-07,
      "loss": 0.6736,
      "step": 30973
    },
    {
      "epoch": 0.8924681611248776,
      "grad_norm": 0.7360999584197998,
      "learning_rate": 3.0019070573335086e-07,
      "loss": 0.7079,
      "step": 30974
    },
    {
      "epoch": 0.8924969745865268,
      "grad_norm": 0.7283414006233215,
      "learning_rate": 3.000314817239097e-07,
      "loss": 0.6817,
      "step": 30975
    },
    {
      "epoch": 0.8925257880481761,
      "grad_norm": 0.7217751741409302,
      "learning_rate": 2.998722986464964e-07,
      "loss": 0.6517,
      "step": 30976
    },
    {
      "epoch": 0.8925546015098254,
      "grad_norm": 0.728238582611084,
      "learning_rate": 2.997131565024969e-07,
      "loss": 0.6943,
      "step": 30977
    },
    {
      "epoch": 0.8925834149714746,
      "grad_norm": 0.7260767221450806,
      "learning_rate": 2.995540552932974e-07,
      "loss": 0.6779,
      "step": 30978
    },
    {
      "epoch": 0.8926122284331239,
      "grad_norm": 0.7352611422538757,
      "learning_rate": 2.993949950202829e-07,
      "loss": 0.6713,
      "step": 30979
    },
    {
      "epoch": 0.8926410418947732,
      "grad_norm": 0.7311475872993469,
      "learning_rate": 2.992359756848395e-07,
      "loss": 0.6702,
      "step": 30980
    },
    {
      "epoch": 0.8926698553564225,
      "grad_norm": 0.7096492052078247,
      "learning_rate": 2.9907699728835115e-07,
      "loss": 0.6621,
      "step": 30981
    },
    {
      "epoch": 0.8926986688180718,
      "grad_norm": 0.7225680947303772,
      "learning_rate": 2.989180598322039e-07,
      "loss": 0.695,
      "step": 30982
    },
    {
      "epoch": 0.8927274822797211,
      "grad_norm": 0.7395105361938477,
      "learning_rate": 2.9875916331777944e-07,
      "loss": 0.6959,
      "step": 30983
    },
    {
      "epoch": 0.8927562957413704,
      "grad_norm": 0.7307160496711731,
      "learning_rate": 2.9860030774646445e-07,
      "loss": 0.6936,
      "step": 30984
    },
    {
      "epoch": 0.8927851092030197,
      "grad_norm": 0.7460471391677856,
      "learning_rate": 2.984414931196405e-07,
      "loss": 0.7096,
      "step": 30985
    },
    {
      "epoch": 0.892813922664669,
      "grad_norm": 0.7049811482429504,
      "learning_rate": 2.982827194386917e-07,
      "loss": 0.6796,
      "step": 30986
    },
    {
      "epoch": 0.8928427361263183,
      "grad_norm": 0.7328605651855469,
      "learning_rate": 2.981239867050001e-07,
      "loss": 0.7049,
      "step": 30987
    },
    {
      "epoch": 0.8928715495879676,
      "grad_norm": 0.7438721656799316,
      "learning_rate": 2.979652949199491e-07,
      "loss": 0.6822,
      "step": 30988
    },
    {
      "epoch": 0.8929003630496167,
      "grad_norm": 0.7034000754356384,
      "learning_rate": 2.9780664408491923e-07,
      "loss": 0.6764,
      "step": 30989
    },
    {
      "epoch": 0.892929176511266,
      "grad_norm": 0.7204394340515137,
      "learning_rate": 2.9764803420129276e-07,
      "loss": 0.6884,
      "step": 30990
    },
    {
      "epoch": 0.8929579899729153,
      "grad_norm": 0.7391781210899353,
      "learning_rate": 2.974894652704524e-07,
      "loss": 0.6701,
      "step": 30991
    },
    {
      "epoch": 0.8929868034345646,
      "grad_norm": 0.7449357509613037,
      "learning_rate": 2.9733093729377717e-07,
      "loss": 0.6649,
      "step": 30992
    },
    {
      "epoch": 0.8930156168962139,
      "grad_norm": 0.7153799533843994,
      "learning_rate": 2.9717245027264916e-07,
      "loss": 0.6761,
      "step": 30993
    },
    {
      "epoch": 0.8930444303578632,
      "grad_norm": 0.733214795589447,
      "learning_rate": 2.970140042084474e-07,
      "loss": 0.701,
      "step": 30994
    },
    {
      "epoch": 0.8930732438195125,
      "grad_norm": 0.7352149486541748,
      "learning_rate": 2.9685559910255234e-07,
      "loss": 0.6875,
      "step": 30995
    },
    {
      "epoch": 0.8931020572811618,
      "grad_norm": 0.7334733009338379,
      "learning_rate": 2.9669723495634353e-07,
      "loss": 0.6973,
      "step": 30996
    },
    {
      "epoch": 0.8931308707428111,
      "grad_norm": 0.7061206102371216,
      "learning_rate": 2.965389117712014e-07,
      "loss": 0.6727,
      "step": 30997
    },
    {
      "epoch": 0.8931596842044603,
      "grad_norm": 0.7387806177139282,
      "learning_rate": 2.9638062954850224e-07,
      "loss": 0.6732,
      "step": 30998
    },
    {
      "epoch": 0.8931884976661096,
      "grad_norm": 0.7049663662910461,
      "learning_rate": 2.96222388289627e-07,
      "loss": 0.6726,
      "step": 30999
    },
    {
      "epoch": 0.8932173111277589,
      "grad_norm": 0.7342221736907959,
      "learning_rate": 2.9606418799595194e-07,
      "loss": 0.692,
      "step": 31000
    },
    {
      "epoch": 0.8932461245894082,
      "grad_norm": 0.7198059558868408,
      "learning_rate": 2.9590602866885644e-07,
      "loss": 0.6916,
      "step": 31001
    },
    {
      "epoch": 0.8932749380510574,
      "grad_norm": 0.7364152073860168,
      "learning_rate": 2.9574791030971604e-07,
      "loss": 0.6709,
      "step": 31002
    },
    {
      "epoch": 0.8933037515127067,
      "grad_norm": 0.7477028369903564,
      "learning_rate": 2.9558983291990974e-07,
      "loss": 0.6979,
      "step": 31003
    },
    {
      "epoch": 0.893332564974356,
      "grad_norm": 0.707904577255249,
      "learning_rate": 2.9543179650081246e-07,
      "loss": 0.6872,
      "step": 31004
    },
    {
      "epoch": 0.8933613784360053,
      "grad_norm": 0.7230920195579529,
      "learning_rate": 2.95273801053802e-07,
      "loss": 0.7115,
      "step": 31005
    },
    {
      "epoch": 0.8933901918976546,
      "grad_norm": 0.710722804069519,
      "learning_rate": 2.951158465802528e-07,
      "loss": 0.6677,
      "step": 31006
    },
    {
      "epoch": 0.8934190053593039,
      "grad_norm": 0.7454009056091309,
      "learning_rate": 2.949579330815422e-07,
      "loss": 0.6972,
      "step": 31007
    },
    {
      "epoch": 0.8934478188209531,
      "grad_norm": 0.7380065321922302,
      "learning_rate": 2.94800060559044e-07,
      "loss": 0.6852,
      "step": 31008
    },
    {
      "epoch": 0.8934766322826024,
      "grad_norm": 0.7067318558692932,
      "learning_rate": 2.9464222901413484e-07,
      "loss": 0.6569,
      "step": 31009
    },
    {
      "epoch": 0.8935054457442517,
      "grad_norm": 0.728975236415863,
      "learning_rate": 2.944844384481871e-07,
      "loss": 0.7,
      "step": 31010
    },
    {
      "epoch": 0.893534259205901,
      "grad_norm": 0.7150003910064697,
      "learning_rate": 2.9432668886257676e-07,
      "loss": 0.6698,
      "step": 31011
    },
    {
      "epoch": 0.8935630726675503,
      "grad_norm": 0.7284714579582214,
      "learning_rate": 2.9416898025867724e-07,
      "loss": 0.6832,
      "step": 31012
    },
    {
      "epoch": 0.8935918861291996,
      "grad_norm": 0.722728431224823,
      "learning_rate": 2.940113126378613e-07,
      "loss": 0.6814,
      "step": 31013
    },
    {
      "epoch": 0.8936206995908489,
      "grad_norm": 0.7472046613693237,
      "learning_rate": 2.938536860015029e-07,
      "loss": 0.6684,
      "step": 31014
    },
    {
      "epoch": 0.8936495130524982,
      "grad_norm": 0.7177254557609558,
      "learning_rate": 2.9369610035097417e-07,
      "loss": 0.6671,
      "step": 31015
    },
    {
      "epoch": 0.8936783265141474,
      "grad_norm": 0.7043174505233765,
      "learning_rate": 2.935385556876486e-07,
      "loss": 0.6565,
      "step": 31016
    },
    {
      "epoch": 0.8937071399757966,
      "grad_norm": 0.7386406660079956,
      "learning_rate": 2.9338105201289655e-07,
      "loss": 0.6836,
      "step": 31017
    },
    {
      "epoch": 0.8937359534374459,
      "grad_norm": 0.7137596607208252,
      "learning_rate": 2.9322358932809157e-07,
      "loss": 0.7023,
      "step": 31018
    },
    {
      "epoch": 0.8937647668990952,
      "grad_norm": 0.738811731338501,
      "learning_rate": 2.9306616763460194e-07,
      "loss": 0.7022,
      "step": 31019
    },
    {
      "epoch": 0.8937935803607445,
      "grad_norm": 0.7390009760856628,
      "learning_rate": 2.929087869338032e-07,
      "loss": 0.6738,
      "step": 31020
    },
    {
      "epoch": 0.8938223938223938,
      "grad_norm": 0.7269102334976196,
      "learning_rate": 2.9275144722706204e-07,
      "loss": 0.6624,
      "step": 31021
    },
    {
      "epoch": 0.8938512072840431,
      "grad_norm": 0.7357152700424194,
      "learning_rate": 2.925941485157513e-07,
      "loss": 0.6991,
      "step": 31022
    },
    {
      "epoch": 0.8938800207456924,
      "grad_norm": 0.7162861824035645,
      "learning_rate": 2.9243689080123926e-07,
      "loss": 0.6753,
      "step": 31023
    },
    {
      "epoch": 0.8939088342073417,
      "grad_norm": 0.7306215167045593,
      "learning_rate": 2.9227967408489653e-07,
      "loss": 0.6682,
      "step": 31024
    },
    {
      "epoch": 0.893937647668991,
      "grad_norm": 0.6998639106750488,
      "learning_rate": 2.9212249836809146e-07,
      "loss": 0.6644,
      "step": 31025
    },
    {
      "epoch": 0.8939664611306403,
      "grad_norm": 0.7238976359367371,
      "learning_rate": 2.919653636521935e-07,
      "loss": 0.689,
      "step": 31026
    },
    {
      "epoch": 0.8939952745922896,
      "grad_norm": 0.7228500843048096,
      "learning_rate": 2.9180826993857046e-07,
      "loss": 0.6808,
      "step": 31027
    },
    {
      "epoch": 0.8940240880539388,
      "grad_norm": 0.7117945551872253,
      "learning_rate": 2.9165121722859123e-07,
      "loss": 0.6662,
      "step": 31028
    },
    {
      "epoch": 0.894052901515588,
      "grad_norm": 0.7270484566688538,
      "learning_rate": 2.914942055236236e-07,
      "loss": 0.689,
      "step": 31029
    },
    {
      "epoch": 0.8940817149772373,
      "grad_norm": 0.716225266456604,
      "learning_rate": 2.9133723482503374e-07,
      "loss": 0.7004,
      "step": 31030
    },
    {
      "epoch": 0.8941105284388866,
      "grad_norm": 0.7106152772903442,
      "learning_rate": 2.911803051341899e-07,
      "loss": 0.6802,
      "step": 31031
    },
    {
      "epoch": 0.8941393419005359,
      "grad_norm": 0.7147844433784485,
      "learning_rate": 2.9102341645245834e-07,
      "loss": 0.6982,
      "step": 31032
    },
    {
      "epoch": 0.8941681553621852,
      "grad_norm": 0.7134337425231934,
      "learning_rate": 2.908665687812068e-07,
      "loss": 0.6532,
      "step": 31033
    },
    {
      "epoch": 0.8941969688238345,
      "grad_norm": 0.7151781320571899,
      "learning_rate": 2.907097621217986e-07,
      "loss": 0.6866,
      "step": 31034
    },
    {
      "epoch": 0.8942257822854838,
      "grad_norm": 0.7594202756881714,
      "learning_rate": 2.905529964756021e-07,
      "loss": 0.6996,
      "step": 31035
    },
    {
      "epoch": 0.8942545957471331,
      "grad_norm": 0.7265944480895996,
      "learning_rate": 2.903962718439807e-07,
      "loss": 0.6798,
      "step": 31036
    },
    {
      "epoch": 0.8942834092087824,
      "grad_norm": 0.7189310193061829,
      "learning_rate": 2.902395882282999e-07,
      "loss": 0.6894,
      "step": 31037
    },
    {
      "epoch": 0.8943122226704316,
      "grad_norm": 0.7258443236351013,
      "learning_rate": 2.900829456299237e-07,
      "loss": 0.6883,
      "step": 31038
    },
    {
      "epoch": 0.8943410361320809,
      "grad_norm": 0.7302384972572327,
      "learning_rate": 2.899263440502181e-07,
      "loss": 0.6784,
      "step": 31039
    },
    {
      "epoch": 0.8943698495937302,
      "grad_norm": 0.724723756313324,
      "learning_rate": 2.897697834905444e-07,
      "loss": 0.6888,
      "step": 31040
    },
    {
      "epoch": 0.8943986630553795,
      "grad_norm": 0.7235766053199768,
      "learning_rate": 2.8961326395226863e-07,
      "loss": 0.6917,
      "step": 31041
    },
    {
      "epoch": 0.8944274765170288,
      "grad_norm": 0.7482444047927856,
      "learning_rate": 2.894567854367508e-07,
      "loss": 0.6851,
      "step": 31042
    },
    {
      "epoch": 0.894456289978678,
      "grad_norm": 0.7312249541282654,
      "learning_rate": 2.8930034794535713e-07,
      "loss": 0.6923,
      "step": 31043
    },
    {
      "epoch": 0.8944851034403273,
      "grad_norm": 0.7037413716316223,
      "learning_rate": 2.8914395147944753e-07,
      "loss": 0.6648,
      "step": 31044
    },
    {
      "epoch": 0.8945139169019766,
      "grad_norm": 0.7204467058181763,
      "learning_rate": 2.8898759604038596e-07,
      "loss": 0.6745,
      "step": 31045
    },
    {
      "epoch": 0.8945427303636259,
      "grad_norm": 0.708939790725708,
      "learning_rate": 2.888312816295324e-07,
      "loss": 0.6943,
      "step": 31046
    },
    {
      "epoch": 0.8945715438252752,
      "grad_norm": 0.7381629347801208,
      "learning_rate": 2.886750082482498e-07,
      "loss": 0.713,
      "step": 31047
    },
    {
      "epoch": 0.8946003572869244,
      "grad_norm": 0.7158450484275818,
      "learning_rate": 2.8851877589789743e-07,
      "loss": 0.6647,
      "step": 31048
    },
    {
      "epoch": 0.8946291707485737,
      "grad_norm": 0.7170960307121277,
      "learning_rate": 2.883625845798366e-07,
      "loss": 0.6794,
      "step": 31049
    },
    {
      "epoch": 0.894657984210223,
      "grad_norm": 0.7219995260238647,
      "learning_rate": 2.8820643429542825e-07,
      "loss": 0.6716,
      "step": 31050
    },
    {
      "epoch": 0.8946867976718723,
      "grad_norm": 0.7084543108940125,
      "learning_rate": 2.880503250460315e-07,
      "loss": 0.6933,
      "step": 31051
    },
    {
      "epoch": 0.8947156111335216,
      "grad_norm": 0.7311278581619263,
      "learning_rate": 2.8789425683300677e-07,
      "loss": 0.7085,
      "step": 31052
    },
    {
      "epoch": 0.8947444245951709,
      "grad_norm": 0.7277851104736328,
      "learning_rate": 2.877382296577119e-07,
      "loss": 0.6934,
      "step": 31053
    },
    {
      "epoch": 0.8947732380568202,
      "grad_norm": 0.7213295102119446,
      "learning_rate": 2.87582243521507e-07,
      "loss": 0.6707,
      "step": 31054
    },
    {
      "epoch": 0.8948020515184695,
      "grad_norm": 0.7438281774520874,
      "learning_rate": 2.874262984257492e-07,
      "loss": 0.7066,
      "step": 31055
    },
    {
      "epoch": 0.8948308649801187,
      "grad_norm": 0.7530709505081177,
      "learning_rate": 2.872703943717986e-07,
      "loss": 0.6791,
      "step": 31056
    },
    {
      "epoch": 0.894859678441768,
      "grad_norm": 0.7122474312782288,
      "learning_rate": 2.871145313610113e-07,
      "loss": 0.6543,
      "step": 31057
    },
    {
      "epoch": 0.8948884919034172,
      "grad_norm": 0.7265203595161438,
      "learning_rate": 2.8695870939474626e-07,
      "loss": 0.6713,
      "step": 31058
    },
    {
      "epoch": 0.8949173053650665,
      "grad_norm": 0.7439653277397156,
      "learning_rate": 2.8680292847435844e-07,
      "loss": 0.6883,
      "step": 31059
    },
    {
      "epoch": 0.8949461188267158,
      "grad_norm": 0.7137603759765625,
      "learning_rate": 2.866471886012062e-07,
      "loss": 0.6972,
      "step": 31060
    },
    {
      "epoch": 0.8949749322883651,
      "grad_norm": 0.7293607592582703,
      "learning_rate": 2.8649148977664464e-07,
      "loss": 0.66,
      "step": 31061
    },
    {
      "epoch": 0.8950037457500144,
      "grad_norm": 0.727023720741272,
      "learning_rate": 2.8633583200203153e-07,
      "loss": 0.7099,
      "step": 31062
    },
    {
      "epoch": 0.8950325592116637,
      "grad_norm": 0.6989253759384155,
      "learning_rate": 2.861802152787202e-07,
      "loss": 0.6815,
      "step": 31063
    },
    {
      "epoch": 0.895061372673313,
      "grad_norm": 0.7082200050354004,
      "learning_rate": 2.860246396080679e-07,
      "loss": 0.6843,
      "step": 31064
    },
    {
      "epoch": 0.8950901861349623,
      "grad_norm": 0.7346121072769165,
      "learning_rate": 2.858691049914286e-07,
      "loss": 0.6862,
      "step": 31065
    },
    {
      "epoch": 0.8951189995966116,
      "grad_norm": 0.7203987240791321,
      "learning_rate": 2.857136114301562e-07,
      "loss": 0.6773,
      "step": 31066
    },
    {
      "epoch": 0.8951478130582609,
      "grad_norm": 0.7359403371810913,
      "learning_rate": 2.855581589256062e-07,
      "loss": 0.6882,
      "step": 31067
    },
    {
      "epoch": 0.8951766265199101,
      "grad_norm": 0.7139682769775391,
      "learning_rate": 2.8540274747913267e-07,
      "loss": 0.6693,
      "step": 31068
    },
    {
      "epoch": 0.8952054399815594,
      "grad_norm": 0.7432780861854553,
      "learning_rate": 2.8524737709208715e-07,
      "loss": 0.6737,
      "step": 31069
    },
    {
      "epoch": 0.8952342534432086,
      "grad_norm": 0.7252761125564575,
      "learning_rate": 2.8509204776582423e-07,
      "loss": 0.6731,
      "step": 31070
    },
    {
      "epoch": 0.8952630669048579,
      "grad_norm": 0.7028847932815552,
      "learning_rate": 2.849367595016972e-07,
      "loss": 0.6693,
      "step": 31071
    },
    {
      "epoch": 0.8952918803665072,
      "grad_norm": 0.6880029439926147,
      "learning_rate": 2.8478151230105665e-07,
      "loss": 0.6513,
      "step": 31072
    },
    {
      "epoch": 0.8953206938281565,
      "grad_norm": 0.7225626111030579,
      "learning_rate": 2.8462630616525657e-07,
      "loss": 0.6682,
      "step": 31073
    },
    {
      "epoch": 0.8953495072898058,
      "grad_norm": 0.7504944205284119,
      "learning_rate": 2.8447114109564687e-07,
      "loss": 0.7156,
      "step": 31074
    },
    {
      "epoch": 0.8953783207514551,
      "grad_norm": 0.7395007610321045,
      "learning_rate": 2.8431601709358046e-07,
      "loss": 0.6894,
      "step": 31075
    },
    {
      "epoch": 0.8954071342131044,
      "grad_norm": 0.7586963772773743,
      "learning_rate": 2.8416093416040737e-07,
      "loss": 0.6848,
      "step": 31076
    },
    {
      "epoch": 0.8954359476747537,
      "grad_norm": 0.7273798584938049,
      "learning_rate": 2.8400589229747865e-07,
      "loss": 0.6729,
      "step": 31077
    },
    {
      "epoch": 0.8954647611364029,
      "grad_norm": 0.7115862369537354,
      "learning_rate": 2.8385089150614273e-07,
      "loss": 0.6694,
      "step": 31078
    },
    {
      "epoch": 0.8954935745980522,
      "grad_norm": 0.7212842106819153,
      "learning_rate": 2.83695931787753e-07,
      "loss": 0.6775,
      "step": 31079
    },
    {
      "epoch": 0.8955223880597015,
      "grad_norm": 0.736475944519043,
      "learning_rate": 2.8354101314365613e-07,
      "loss": 0.6881,
      "step": 31080
    },
    {
      "epoch": 0.8955512015213508,
      "grad_norm": 0.7166276574134827,
      "learning_rate": 2.8338613557520325e-07,
      "loss": 0.678,
      "step": 31081
    },
    {
      "epoch": 0.8955800149830001,
      "grad_norm": 0.7222627401351929,
      "learning_rate": 2.8323129908374104e-07,
      "loss": 0.6546,
      "step": 31082
    },
    {
      "epoch": 0.8956088284446493,
      "grad_norm": 0.7014606595039368,
      "learning_rate": 2.830765036706207e-07,
      "loss": 0.6495,
      "step": 31083
    },
    {
      "epoch": 0.8956376419062986,
      "grad_norm": 0.7148464322090149,
      "learning_rate": 2.829217493371872e-07,
      "loss": 0.6742,
      "step": 31084
    },
    {
      "epoch": 0.8956664553679479,
      "grad_norm": 0.7204303741455078,
      "learning_rate": 2.8276703608479115e-07,
      "loss": 0.6648,
      "step": 31085
    },
    {
      "epoch": 0.8956952688295972,
      "grad_norm": 0.7054035663604736,
      "learning_rate": 2.826123639147782e-07,
      "loss": 0.6675,
      "step": 31086
    },
    {
      "epoch": 0.8957240822912464,
      "grad_norm": 0.7258641719818115,
      "learning_rate": 2.8245773282849556e-07,
      "loss": 0.6993,
      "step": 31087
    },
    {
      "epoch": 0.8957528957528957,
      "grad_norm": 0.7227033376693726,
      "learning_rate": 2.82303142827291e-07,
      "loss": 0.6684,
      "step": 31088
    },
    {
      "epoch": 0.895781709214545,
      "grad_norm": 0.7275271415710449,
      "learning_rate": 2.82148593912509e-07,
      "loss": 0.6638,
      "step": 31089
    },
    {
      "epoch": 0.8958105226761943,
      "grad_norm": 0.7000203132629395,
      "learning_rate": 2.8199408608549696e-07,
      "loss": 0.6679,
      "step": 31090
    },
    {
      "epoch": 0.8958393361378436,
      "grad_norm": 0.7408785223960876,
      "learning_rate": 2.8183961934760086e-07,
      "loss": 0.6718,
      "step": 31091
    },
    {
      "epoch": 0.8958681495994929,
      "grad_norm": 0.7243677973747253,
      "learning_rate": 2.816851937001647e-07,
      "loss": 0.6467,
      "step": 31092
    },
    {
      "epoch": 0.8958969630611422,
      "grad_norm": 0.713580846786499,
      "learning_rate": 2.81530809144534e-07,
      "loss": 0.6942,
      "step": 31093
    },
    {
      "epoch": 0.8959257765227915,
      "grad_norm": 0.712670087814331,
      "learning_rate": 2.8137646568205336e-07,
      "loss": 0.69,
      "step": 31094
    },
    {
      "epoch": 0.8959545899844408,
      "grad_norm": 0.7156190276145935,
      "learning_rate": 2.812221633140666e-07,
      "loss": 0.6947,
      "step": 31095
    },
    {
      "epoch": 0.8959834034460901,
      "grad_norm": 0.7078371644020081,
      "learning_rate": 2.8106790204191826e-07,
      "loss": 0.6801,
      "step": 31096
    },
    {
      "epoch": 0.8960122169077392,
      "grad_norm": 0.72995924949646,
      "learning_rate": 2.8091368186695056e-07,
      "loss": 0.6748,
      "step": 31097
    },
    {
      "epoch": 0.8960410303693885,
      "grad_norm": 0.7133581638336182,
      "learning_rate": 2.8075950279050856e-07,
      "loss": 0.6867,
      "step": 31098
    },
    {
      "epoch": 0.8960698438310378,
      "grad_norm": 0.7334454655647278,
      "learning_rate": 2.806053648139323e-07,
      "loss": 0.7043,
      "step": 31099
    },
    {
      "epoch": 0.8960986572926871,
      "grad_norm": 0.7366340160369873,
      "learning_rate": 2.804512679385668e-07,
      "loss": 0.6737,
      "step": 31100
    },
    {
      "epoch": 0.8961274707543364,
      "grad_norm": 0.7288605570793152,
      "learning_rate": 2.802972121657521e-07,
      "loss": 0.6902,
      "step": 31101
    },
    {
      "epoch": 0.8961562842159857,
      "grad_norm": 0.7115776538848877,
      "learning_rate": 2.801431974968305e-07,
      "loss": 0.6856,
      "step": 31102
    },
    {
      "epoch": 0.896185097677635,
      "grad_norm": 0.7234063744544983,
      "learning_rate": 2.7998922393314364e-07,
      "loss": 0.6924,
      "step": 31103
    },
    {
      "epoch": 0.8962139111392843,
      "grad_norm": 0.7263531684875488,
      "learning_rate": 2.7983529147603327e-07,
      "loss": 0.6908,
      "step": 31104
    },
    {
      "epoch": 0.8962427246009336,
      "grad_norm": 0.7210684418678284,
      "learning_rate": 2.796814001268383e-07,
      "loss": 0.6773,
      "step": 31105
    },
    {
      "epoch": 0.8962715380625829,
      "grad_norm": 0.7365482449531555,
      "learning_rate": 2.795275498869005e-07,
      "loss": 0.6986,
      "step": 31106
    },
    {
      "epoch": 0.8963003515242322,
      "grad_norm": 0.7229768633842468,
      "learning_rate": 2.7937374075755806e-07,
      "loss": 0.6853,
      "step": 31107
    },
    {
      "epoch": 0.8963291649858814,
      "grad_norm": 0.7233946919441223,
      "learning_rate": 2.7921997274015125e-07,
      "loss": 0.6577,
      "step": 31108
    },
    {
      "epoch": 0.8963579784475307,
      "grad_norm": 0.7274901866912842,
      "learning_rate": 2.7906624583602047e-07,
      "loss": 0.7017,
      "step": 31109
    },
    {
      "epoch": 0.8963867919091799,
      "grad_norm": 0.7293927073478699,
      "learning_rate": 2.7891256004650314e-07,
      "loss": 0.7039,
      "step": 31110
    },
    {
      "epoch": 0.8964156053708292,
      "grad_norm": 0.7051287889480591,
      "learning_rate": 2.787589153729381e-07,
      "loss": 0.6871,
      "step": 31111
    },
    {
      "epoch": 0.8964444188324785,
      "grad_norm": 0.7303358316421509,
      "learning_rate": 2.786053118166632e-07,
      "loss": 0.6882,
      "step": 31112
    },
    {
      "epoch": 0.8964732322941278,
      "grad_norm": 0.7371276021003723,
      "learning_rate": 2.7845174937901675e-07,
      "loss": 0.6956,
      "step": 31113
    },
    {
      "epoch": 0.8965020457557771,
      "grad_norm": 0.716562807559967,
      "learning_rate": 2.782982280613344e-07,
      "loss": 0.6549,
      "step": 31114
    },
    {
      "epoch": 0.8965308592174264,
      "grad_norm": 0.7335900664329529,
      "learning_rate": 2.7814474786495573e-07,
      "loss": 0.6879,
      "step": 31115
    },
    {
      "epoch": 0.8965596726790757,
      "grad_norm": 0.7198694944381714,
      "learning_rate": 2.779913087912156e-07,
      "loss": 0.6929,
      "step": 31116
    },
    {
      "epoch": 0.896588486140725,
      "grad_norm": 0.7222167253494263,
      "learning_rate": 2.7783791084145195e-07,
      "loss": 0.6626,
      "step": 31117
    },
    {
      "epoch": 0.8966172996023742,
      "grad_norm": 0.7188067436218262,
      "learning_rate": 2.7768455401699867e-07,
      "loss": 0.6872,
      "step": 31118
    },
    {
      "epoch": 0.8966461130640235,
      "grad_norm": 0.7171486616134644,
      "learning_rate": 2.7753123831919305e-07,
      "loss": 0.6719,
      "step": 31119
    },
    {
      "epoch": 0.8966749265256728,
      "grad_norm": 0.7389497756958008,
      "learning_rate": 2.7737796374936844e-07,
      "loss": 0.6943,
      "step": 31120
    },
    {
      "epoch": 0.8967037399873221,
      "grad_norm": 0.7220839262008667,
      "learning_rate": 2.772247303088621e-07,
      "loss": 0.6814,
      "step": 31121
    },
    {
      "epoch": 0.8967325534489714,
      "grad_norm": 0.7175712585449219,
      "learning_rate": 2.7707153799900685e-07,
      "loss": 0.6715,
      "step": 31122
    },
    {
      "epoch": 0.8967613669106207,
      "grad_norm": 0.7505812048912048,
      "learning_rate": 2.7691838682113723e-07,
      "loss": 0.6933,
      "step": 31123
    },
    {
      "epoch": 0.8967901803722699,
      "grad_norm": 0.7315046191215515,
      "learning_rate": 2.7676527677658715e-07,
      "loss": 0.6925,
      "step": 31124
    },
    {
      "epoch": 0.8968189938339192,
      "grad_norm": 0.7208439707756042,
      "learning_rate": 2.766122078666894e-07,
      "loss": 0.6711,
      "step": 31125
    },
    {
      "epoch": 0.8968478072955685,
      "grad_norm": 0.7308530807495117,
      "learning_rate": 2.764591800927785e-07,
      "loss": 0.6747,
      "step": 31126
    },
    {
      "epoch": 0.8968766207572177,
      "grad_norm": 0.7398058772087097,
      "learning_rate": 2.7630619345618614e-07,
      "loss": 0.6986,
      "step": 31127
    },
    {
      "epoch": 0.896905434218867,
      "grad_norm": 0.7677528858184814,
      "learning_rate": 2.761532479582446e-07,
      "loss": 0.6974,
      "step": 31128
    },
    {
      "epoch": 0.8969342476805163,
      "grad_norm": 0.7057787179946899,
      "learning_rate": 2.760003436002867e-07,
      "loss": 0.6758,
      "step": 31129
    },
    {
      "epoch": 0.8969630611421656,
      "grad_norm": 0.7418645620346069,
      "learning_rate": 2.758474803836431e-07,
      "loss": 0.6826,
      "step": 31130
    },
    {
      "epoch": 0.8969918746038149,
      "grad_norm": 0.7257365584373474,
      "learning_rate": 2.7569465830964535e-07,
      "loss": 0.6632,
      "step": 31131
    },
    {
      "epoch": 0.8970206880654642,
      "grad_norm": 0.731450617313385,
      "learning_rate": 2.7554187737962537e-07,
      "loss": 0.6908,
      "step": 31132
    },
    {
      "epoch": 0.8970495015271135,
      "grad_norm": 0.7436411380767822,
      "learning_rate": 2.753891375949119e-07,
      "loss": 0.6812,
      "step": 31133
    },
    {
      "epoch": 0.8970783149887628,
      "grad_norm": 0.7137928605079651,
      "learning_rate": 2.752364389568368e-07,
      "loss": 0.6649,
      "step": 31134
    },
    {
      "epoch": 0.8971071284504121,
      "grad_norm": 0.7337073087692261,
      "learning_rate": 2.75083781466729e-07,
      "loss": 0.6778,
      "step": 31135
    },
    {
      "epoch": 0.8971359419120614,
      "grad_norm": 0.740111768245697,
      "learning_rate": 2.74931165125919e-07,
      "loss": 0.6757,
      "step": 31136
    },
    {
      "epoch": 0.8971647553737105,
      "grad_norm": 0.7440890669822693,
      "learning_rate": 2.747785899357341e-07,
      "loss": 0.6805,
      "step": 31137
    },
    {
      "epoch": 0.8971935688353598,
      "grad_norm": 0.7338405251502991,
      "learning_rate": 2.7462605589750447e-07,
      "loss": 0.6874,
      "step": 31138
    },
    {
      "epoch": 0.8972223822970091,
      "grad_norm": 0.7218272089958191,
      "learning_rate": 2.744735630125583e-07,
      "loss": 0.7011,
      "step": 31139
    },
    {
      "epoch": 0.8972511957586584,
      "grad_norm": 0.7193764448165894,
      "learning_rate": 2.7432111128222417e-07,
      "loss": 0.677,
      "step": 31140
    },
    {
      "epoch": 0.8972800092203077,
      "grad_norm": 0.7247595191001892,
      "learning_rate": 2.741687007078281e-07,
      "loss": 0.663,
      "step": 31141
    },
    {
      "epoch": 0.897308822681957,
      "grad_norm": 0.7456904649734497,
      "learning_rate": 2.7401633129070017e-07,
      "loss": 0.6728,
      "step": 31142
    },
    {
      "epoch": 0.8973376361436063,
      "grad_norm": 0.9544634819030762,
      "learning_rate": 2.7386400303216434e-07,
      "loss": 0.6801,
      "step": 31143
    },
    {
      "epoch": 0.8973664496052556,
      "grad_norm": 0.7109763622283936,
      "learning_rate": 2.737117159335495e-07,
      "loss": 0.6839,
      "step": 31144
    },
    {
      "epoch": 0.8973952630669049,
      "grad_norm": 0.7075290083885193,
      "learning_rate": 2.735594699961802e-07,
      "loss": 0.6619,
      "step": 31145
    },
    {
      "epoch": 0.8974240765285542,
      "grad_norm": 0.7140193581581116,
      "learning_rate": 2.734072652213837e-07,
      "loss": 0.6805,
      "step": 31146
    },
    {
      "epoch": 0.8974528899902035,
      "grad_norm": 0.7317580580711365,
      "learning_rate": 2.732551016104851e-07,
      "loss": 0.6724,
      "step": 31147
    },
    {
      "epoch": 0.8974817034518527,
      "grad_norm": 0.7019183039665222,
      "learning_rate": 2.731029791648093e-07,
      "loss": 0.6835,
      "step": 31148
    },
    {
      "epoch": 0.897510516913502,
      "grad_norm": 0.7215520143508911,
      "learning_rate": 2.729508978856821e-07,
      "loss": 0.6791,
      "step": 31149
    },
    {
      "epoch": 0.8975393303751513,
      "grad_norm": 0.7313793301582336,
      "learning_rate": 2.727988577744262e-07,
      "loss": 0.6755,
      "step": 31150
    },
    {
      "epoch": 0.8975681438368005,
      "grad_norm": 0.7487755417823792,
      "learning_rate": 2.726468588323672e-07,
      "loss": 0.7084,
      "step": 31151
    },
    {
      "epoch": 0.8975969572984498,
      "grad_norm": 0.7167826294898987,
      "learning_rate": 2.72494901060828e-07,
      "loss": 0.6868,
      "step": 31152
    },
    {
      "epoch": 0.8976257707600991,
      "grad_norm": 0.6956612467765808,
      "learning_rate": 2.723429844611336e-07,
      "loss": 0.6537,
      "step": 31153
    },
    {
      "epoch": 0.8976545842217484,
      "grad_norm": 0.7195726037025452,
      "learning_rate": 2.721911090346052e-07,
      "loss": 0.6696,
      "step": 31154
    },
    {
      "epoch": 0.8976833976833977,
      "grad_norm": 0.7102893590927124,
      "learning_rate": 2.720392747825662e-07,
      "loss": 0.657,
      "step": 31155
    },
    {
      "epoch": 0.897712211145047,
      "grad_norm": 0.743979811668396,
      "learning_rate": 2.718874817063388e-07,
      "loss": 0.6892,
      "step": 31156
    },
    {
      "epoch": 0.8977410246066962,
      "grad_norm": 0.7204800844192505,
      "learning_rate": 2.7173572980724536e-07,
      "loss": 0.6885,
      "step": 31157
    },
    {
      "epoch": 0.8977698380683455,
      "grad_norm": 0.71832275390625,
      "learning_rate": 2.71584019086607e-07,
      "loss": 0.7005,
      "step": 31158
    },
    {
      "epoch": 0.8977986515299948,
      "grad_norm": 0.7398018836975098,
      "learning_rate": 2.714323495457455e-07,
      "loss": 0.6753,
      "step": 31159
    },
    {
      "epoch": 0.8978274649916441,
      "grad_norm": 0.7188073992729187,
      "learning_rate": 2.712807211859808e-07,
      "loss": 0.6717,
      "step": 31160
    },
    {
      "epoch": 0.8978562784532934,
      "grad_norm": 0.7133269906044006,
      "learning_rate": 2.711291340086347e-07,
      "loss": 0.6855,
      "step": 31161
    },
    {
      "epoch": 0.8978850919149427,
      "grad_norm": 0.7310001850128174,
      "learning_rate": 2.7097758801502503e-07,
      "loss": 0.6543,
      "step": 31162
    },
    {
      "epoch": 0.897913905376592,
      "grad_norm": 0.7303510308265686,
      "learning_rate": 2.7082608320647516e-07,
      "loss": 0.6585,
      "step": 31163
    },
    {
      "epoch": 0.8979427188382412,
      "grad_norm": 0.7024597525596619,
      "learning_rate": 2.7067461958430133e-07,
      "loss": 0.6678,
      "step": 31164
    },
    {
      "epoch": 0.8979715322998905,
      "grad_norm": 0.7275997400283813,
      "learning_rate": 2.705231971498251e-07,
      "loss": 0.7102,
      "step": 31165
    },
    {
      "epoch": 0.8980003457615398,
      "grad_norm": 0.7236344814300537,
      "learning_rate": 2.7037181590436334e-07,
      "loss": 0.6748,
      "step": 31166
    },
    {
      "epoch": 0.898029159223189,
      "grad_norm": 0.7350261807441711,
      "learning_rate": 2.702204758492355e-07,
      "loss": 0.6836,
      "step": 31167
    },
    {
      "epoch": 0.8980579726848383,
      "grad_norm": 0.7176305651664734,
      "learning_rate": 2.7006917698575943e-07,
      "loss": 0.6928,
      "step": 31168
    },
    {
      "epoch": 0.8980867861464876,
      "grad_norm": 0.7289866209030151,
      "learning_rate": 2.6991791931525237e-07,
      "loss": 0.6757,
      "step": 31169
    },
    {
      "epoch": 0.8981155996081369,
      "grad_norm": 0.7211480736732483,
      "learning_rate": 2.6976670283903215e-07,
      "loss": 0.6523,
      "step": 31170
    },
    {
      "epoch": 0.8981444130697862,
      "grad_norm": 0.7263332009315491,
      "learning_rate": 2.69615527558415e-07,
      "loss": 0.6559,
      "step": 31171
    },
    {
      "epoch": 0.8981732265314355,
      "grad_norm": 0.7332066297531128,
      "learning_rate": 2.694643934747182e-07,
      "loss": 0.6823,
      "step": 31172
    },
    {
      "epoch": 0.8982020399930848,
      "grad_norm": 0.7169559597969055,
      "learning_rate": 2.6931330058925787e-07,
      "loss": 0.6802,
      "step": 31173
    },
    {
      "epoch": 0.8982308534547341,
      "grad_norm": 0.727462649345398,
      "learning_rate": 2.6916224890334906e-07,
      "loss": 0.6754,
      "step": 31174
    },
    {
      "epoch": 0.8982596669163834,
      "grad_norm": 0.7087883949279785,
      "learning_rate": 2.69011238418308e-07,
      "loss": 0.6691,
      "step": 31175
    },
    {
      "epoch": 0.8982884803780327,
      "grad_norm": 0.720212996006012,
      "learning_rate": 2.6886026913545085e-07,
      "loss": 0.6517,
      "step": 31176
    },
    {
      "epoch": 0.898317293839682,
      "grad_norm": 0.7411372065544128,
      "learning_rate": 2.687093410560904e-07,
      "loss": 0.7,
      "step": 31177
    },
    {
      "epoch": 0.8983461073013311,
      "grad_norm": 0.7237768173217773,
      "learning_rate": 2.685584541815428e-07,
      "loss": 0.6884,
      "step": 31178
    },
    {
      "epoch": 0.8983749207629804,
      "grad_norm": 0.7304471135139465,
      "learning_rate": 2.6840760851312045e-07,
      "loss": 0.6853,
      "step": 31179
    },
    {
      "epoch": 0.8984037342246297,
      "grad_norm": 0.7086673974990845,
      "learning_rate": 2.6825680405213884e-07,
      "loss": 0.6725,
      "step": 31180
    },
    {
      "epoch": 0.898432547686279,
      "grad_norm": 0.725396454334259,
      "learning_rate": 2.6810604079990977e-07,
      "loss": 0.7109,
      "step": 31181
    },
    {
      "epoch": 0.8984613611479283,
      "grad_norm": 0.7154871821403503,
      "learning_rate": 2.679553187577477e-07,
      "loss": 0.6946,
      "step": 31182
    },
    {
      "epoch": 0.8984901746095776,
      "grad_norm": 0.7464247941970825,
      "learning_rate": 2.678046379269639e-07,
      "loss": 0.6796,
      "step": 31183
    },
    {
      "epoch": 0.8985189880712269,
      "grad_norm": 0.7124232649803162,
      "learning_rate": 2.676539983088711e-07,
      "loss": 0.6789,
      "step": 31184
    },
    {
      "epoch": 0.8985478015328762,
      "grad_norm": 0.7193334698677063,
      "learning_rate": 2.6750339990478215e-07,
      "loss": 0.6604,
      "step": 31185
    },
    {
      "epoch": 0.8985766149945255,
      "grad_norm": 0.7170802354812622,
      "learning_rate": 2.673528427160066e-07,
      "loss": 0.6719,
      "step": 31186
    },
    {
      "epoch": 0.8986054284561747,
      "grad_norm": 0.7711672186851501,
      "learning_rate": 2.6720232674385727e-07,
      "loss": 0.6963,
      "step": 31187
    },
    {
      "epoch": 0.898634241917824,
      "grad_norm": 0.7297846078872681,
      "learning_rate": 2.6705185198964533e-07,
      "loss": 0.6743,
      "step": 31188
    },
    {
      "epoch": 0.8986630553794733,
      "grad_norm": 0.7510559558868408,
      "learning_rate": 2.6690141845467977e-07,
      "loss": 0.6869,
      "step": 31189
    },
    {
      "epoch": 0.8986918688411226,
      "grad_norm": 0.7171012163162231,
      "learning_rate": 2.667510261402717e-07,
      "loss": 0.6586,
      "step": 31190
    },
    {
      "epoch": 0.8987206823027718,
      "grad_norm": 0.7040603756904602,
      "learning_rate": 2.6660067504773124e-07,
      "loss": 0.696,
      "step": 31191
    },
    {
      "epoch": 0.8987494957644211,
      "grad_norm": 0.7310309410095215,
      "learning_rate": 2.664503651783662e-07,
      "loss": 0.6814,
      "step": 31192
    },
    {
      "epoch": 0.8987783092260704,
      "grad_norm": 0.7533609867095947,
      "learning_rate": 2.6630009653348774e-07,
      "loss": 0.6681,
      "step": 31193
    },
    {
      "epoch": 0.8988071226877197,
      "grad_norm": 0.7454779744148254,
      "learning_rate": 2.6614986911440264e-07,
      "loss": 0.6966,
      "step": 31194
    },
    {
      "epoch": 0.898835936149369,
      "grad_norm": 0.734244167804718,
      "learning_rate": 2.6599968292242093e-07,
      "loss": 0.6762,
      "step": 31195
    },
    {
      "epoch": 0.8988647496110183,
      "grad_norm": 0.735549807548523,
      "learning_rate": 2.6584953795884884e-07,
      "loss": 0.6835,
      "step": 31196
    },
    {
      "epoch": 0.8988935630726675,
      "grad_norm": 0.7253294587135315,
      "learning_rate": 2.656994342249958e-07,
      "loss": 0.6762,
      "step": 31197
    },
    {
      "epoch": 0.8989223765343168,
      "grad_norm": 0.7186186909675598,
      "learning_rate": 2.655493717221663e-07,
      "loss": 0.6747,
      "step": 31198
    },
    {
      "epoch": 0.8989511899959661,
      "grad_norm": 0.7249698042869568,
      "learning_rate": 2.653993504516711e-07,
      "loss": 0.668,
      "step": 31199
    },
    {
      "epoch": 0.8989800034576154,
      "grad_norm": 0.7321885824203491,
      "learning_rate": 2.652493704148135e-07,
      "loss": 0.7003,
      "step": 31200
    },
    {
      "epoch": 0.8990088169192647,
      "grad_norm": 0.6981222033500671,
      "learning_rate": 2.650994316129013e-07,
      "loss": 0.6691,
      "step": 31201
    },
    {
      "epoch": 0.899037630380914,
      "grad_norm": 0.719636857509613,
      "learning_rate": 2.6494953404723967e-07,
      "loss": 0.6786,
      "step": 31202
    },
    {
      "epoch": 0.8990664438425633,
      "grad_norm": 0.7146164774894714,
      "learning_rate": 2.647996777191347e-07,
      "loss": 0.6689,
      "step": 31203
    },
    {
      "epoch": 0.8990952573042126,
      "grad_norm": 0.735823929309845,
      "learning_rate": 2.6464986262989045e-07,
      "loss": 0.6752,
      "step": 31204
    },
    {
      "epoch": 0.8991240707658618,
      "grad_norm": 0.7250683307647705,
      "learning_rate": 2.6450008878081246e-07,
      "loss": 0.6712,
      "step": 31205
    },
    {
      "epoch": 0.899152884227511,
      "grad_norm": 0.7221087217330933,
      "learning_rate": 2.643503561732058e-07,
      "loss": 0.6796,
      "step": 31206
    },
    {
      "epoch": 0.8991816976891603,
      "grad_norm": 0.7417027354240417,
      "learning_rate": 2.6420066480837227e-07,
      "loss": 0.6771,
      "step": 31207
    },
    {
      "epoch": 0.8992105111508096,
      "grad_norm": 0.7375722527503967,
      "learning_rate": 2.6405101468761797e-07,
      "loss": 0.7019,
      "step": 31208
    },
    {
      "epoch": 0.8992393246124589,
      "grad_norm": 0.703262448310852,
      "learning_rate": 2.6390140581224466e-07,
      "loss": 0.6577,
      "step": 31209
    },
    {
      "epoch": 0.8992681380741082,
      "grad_norm": 0.7409205436706543,
      "learning_rate": 2.6375183818355523e-07,
      "loss": 0.6973,
      "step": 31210
    },
    {
      "epoch": 0.8992969515357575,
      "grad_norm": 0.7276971340179443,
      "learning_rate": 2.63602311802853e-07,
      "loss": 0.6683,
      "step": 31211
    },
    {
      "epoch": 0.8993257649974068,
      "grad_norm": 0.7431744337081909,
      "learning_rate": 2.634528266714409e-07,
      "loss": 0.6558,
      "step": 31212
    },
    {
      "epoch": 0.8993545784590561,
      "grad_norm": 0.7228546738624573,
      "learning_rate": 2.6330338279061893e-07,
      "loss": 0.7093,
      "step": 31213
    },
    {
      "epoch": 0.8993833919207054,
      "grad_norm": 0.7260237336158752,
      "learning_rate": 2.6315398016169e-07,
      "loss": 0.6751,
      "step": 31214
    },
    {
      "epoch": 0.8994122053823547,
      "grad_norm": 0.7379341125488281,
      "learning_rate": 2.6300461878595476e-07,
      "loss": 0.6826,
      "step": 31215
    },
    {
      "epoch": 0.899441018844004,
      "grad_norm": 0.7127798199653625,
      "learning_rate": 2.6285529866471424e-07,
      "loss": 0.6686,
      "step": 31216
    },
    {
      "epoch": 0.8994698323056532,
      "grad_norm": 0.7234005331993103,
      "learning_rate": 2.6270601979926755e-07,
      "loss": 0.6808,
      "step": 31217
    },
    {
      "epoch": 0.8994986457673024,
      "grad_norm": 0.7213775515556335,
      "learning_rate": 2.6255678219091754e-07,
      "loss": 0.6913,
      "step": 31218
    },
    {
      "epoch": 0.8995274592289517,
      "grad_norm": 0.7146134376525879,
      "learning_rate": 2.6240758584096084e-07,
      "loss": 0.6588,
      "step": 31219
    },
    {
      "epoch": 0.899556272690601,
      "grad_norm": 0.7141595482826233,
      "learning_rate": 2.6225843075069924e-07,
      "loss": 0.6937,
      "step": 31220
    },
    {
      "epoch": 0.8995850861522503,
      "grad_norm": 0.7704012989997864,
      "learning_rate": 2.621093169214295e-07,
      "loss": 0.6936,
      "step": 31221
    },
    {
      "epoch": 0.8996138996138996,
      "grad_norm": 0.7595238089561462,
      "learning_rate": 2.619602443544522e-07,
      "loss": 0.6924,
      "step": 31222
    },
    {
      "epoch": 0.8996427130755489,
      "grad_norm": 0.7203063368797302,
      "learning_rate": 2.6181121305106417e-07,
      "loss": 0.6628,
      "step": 31223
    },
    {
      "epoch": 0.8996715265371982,
      "grad_norm": 0.7295006513595581,
      "learning_rate": 2.616622230125654e-07,
      "loss": 0.6705,
      "step": 31224
    },
    {
      "epoch": 0.8997003399988475,
      "grad_norm": 0.712965726852417,
      "learning_rate": 2.61513274240251e-07,
      "loss": 0.6512,
      "step": 31225
    },
    {
      "epoch": 0.8997291534604968,
      "grad_norm": 0.7228856086730957,
      "learning_rate": 2.6136436673541985e-07,
      "loss": 0.7006,
      "step": 31226
    },
    {
      "epoch": 0.899757966922146,
      "grad_norm": 0.733389139175415,
      "learning_rate": 2.6121550049936715e-07,
      "loss": 0.6738,
      "step": 31227
    },
    {
      "epoch": 0.8997867803837953,
      "grad_norm": 0.7290281057357788,
      "learning_rate": 2.6106667553339126e-07,
      "loss": 0.6827,
      "step": 31228
    },
    {
      "epoch": 0.8998155938454446,
      "grad_norm": 0.7233358025550842,
      "learning_rate": 2.6091789183878725e-07,
      "loss": 0.6905,
      "step": 31229
    },
    {
      "epoch": 0.8998444073070939,
      "grad_norm": 0.7308889627456665,
      "learning_rate": 2.607691494168507e-07,
      "loss": 0.686,
      "step": 31230
    },
    {
      "epoch": 0.8998732207687432,
      "grad_norm": 0.7244760990142822,
      "learning_rate": 2.6062044826887843e-07,
      "loss": 0.6861,
      "step": 31231
    },
    {
      "epoch": 0.8999020342303924,
      "grad_norm": 0.7486326694488525,
      "learning_rate": 2.6047178839616327e-07,
      "loss": 0.6954,
      "step": 31232
    },
    {
      "epoch": 0.8999308476920417,
      "grad_norm": 0.7330454587936401,
      "learning_rate": 2.603231698000019e-07,
      "loss": 0.6739,
      "step": 31233
    },
    {
      "epoch": 0.899959661153691,
      "grad_norm": 0.7327130436897278,
      "learning_rate": 2.601745924816862e-07,
      "loss": 0.7007,
      "step": 31234
    },
    {
      "epoch": 0.8999884746153403,
      "grad_norm": 0.7110689282417297,
      "learning_rate": 2.600260564425133e-07,
      "loss": 0.6596,
      "step": 31235
    },
    {
      "epoch": 0.9000172880769896,
      "grad_norm": 0.7226904630661011,
      "learning_rate": 2.5987756168377456e-07,
      "loss": 0.6767,
      "step": 31236
    },
    {
      "epoch": 0.9000461015386388,
      "grad_norm": 0.7299381494522095,
      "learning_rate": 2.5972910820676435e-07,
      "loss": 0.6736,
      "step": 31237
    },
    {
      "epoch": 0.9000749150002881,
      "grad_norm": 0.7158024311065674,
      "learning_rate": 2.5958069601277446e-07,
      "loss": 0.6892,
      "step": 31238
    },
    {
      "epoch": 0.9001037284619374,
      "grad_norm": 0.7315405011177063,
      "learning_rate": 2.594323251030989e-07,
      "loss": 0.6808,
      "step": 31239
    },
    {
      "epoch": 0.9001325419235867,
      "grad_norm": 0.7275943756103516,
      "learning_rate": 2.592839954790283e-07,
      "loss": 0.6644,
      "step": 31240
    },
    {
      "epoch": 0.900161355385236,
      "grad_norm": 0.729154646396637,
      "learning_rate": 2.5913570714185543e-07,
      "loss": 0.7088,
      "step": 31241
    },
    {
      "epoch": 0.9001901688468853,
      "grad_norm": 0.7102231383323669,
      "learning_rate": 2.5898746009287104e-07,
      "loss": 0.6708,
      "step": 31242
    },
    {
      "epoch": 0.9002189823085346,
      "grad_norm": 0.72658771276474,
      "learning_rate": 2.5883925433336623e-07,
      "loss": 0.6759,
      "step": 31243
    },
    {
      "epoch": 0.9002477957701839,
      "grad_norm": 0.7052672505378723,
      "learning_rate": 2.586910898646333e-07,
      "loss": 0.6682,
      "step": 31244
    },
    {
      "epoch": 0.9002766092318331,
      "grad_norm": 0.7234722375869751,
      "learning_rate": 2.585429666879602e-07,
      "loss": 0.6624,
      "step": 31245
    },
    {
      "epoch": 0.9003054226934823,
      "grad_norm": 0.7351647615432739,
      "learning_rate": 2.583948848046386e-07,
      "loss": 0.6428,
      "step": 31246
    },
    {
      "epoch": 0.9003342361551316,
      "grad_norm": 0.7146132588386536,
      "learning_rate": 2.5824684421595805e-07,
      "loss": 0.6686,
      "step": 31247
    },
    {
      "epoch": 0.9003630496167809,
      "grad_norm": 0.7167778611183167,
      "learning_rate": 2.580988449232069e-07,
      "loss": 0.6947,
      "step": 31248
    },
    {
      "epoch": 0.9003918630784302,
      "grad_norm": 0.7287390828132629,
      "learning_rate": 2.5795088692767476e-07,
      "loss": 0.6687,
      "step": 31249
    },
    {
      "epoch": 0.9004206765400795,
      "grad_norm": 0.7251880168914795,
      "learning_rate": 2.578029702306506e-07,
      "loss": 0.692,
      "step": 31250
    },
    {
      "epoch": 0.9004494900017288,
      "grad_norm": 0.7330154776573181,
      "learning_rate": 2.5765509483342164e-07,
      "loss": 0.6702,
      "step": 31251
    },
    {
      "epoch": 0.9004783034633781,
      "grad_norm": 0.7185541987419128,
      "learning_rate": 2.5750726073727637e-07,
      "loss": 0.6746,
      "step": 31252
    },
    {
      "epoch": 0.9005071169250274,
      "grad_norm": 0.7308835983276367,
      "learning_rate": 2.573594679435021e-07,
      "loss": 0.7039,
      "step": 31253
    },
    {
      "epoch": 0.9005359303866767,
      "grad_norm": 0.7086886167526245,
      "learning_rate": 2.572117164533861e-07,
      "loss": 0.6705,
      "step": 31254
    },
    {
      "epoch": 0.900564743848326,
      "grad_norm": 0.7097165584564209,
      "learning_rate": 2.570640062682145e-07,
      "loss": 0.6598,
      "step": 31255
    },
    {
      "epoch": 0.9005935573099753,
      "grad_norm": 0.7218631505966187,
      "learning_rate": 2.569163373892747e-07,
      "loss": 0.6784,
      "step": 31256
    },
    {
      "epoch": 0.9006223707716245,
      "grad_norm": 0.737531304359436,
      "learning_rate": 2.5676870981785173e-07,
      "loss": 0.6687,
      "step": 31257
    },
    {
      "epoch": 0.9006511842332738,
      "grad_norm": 0.7252153754234314,
      "learning_rate": 2.566211235552318e-07,
      "loss": 0.7063,
      "step": 31258
    },
    {
      "epoch": 0.900679997694923,
      "grad_norm": 0.7206835150718689,
      "learning_rate": 2.564735786027006e-07,
      "loss": 0.6586,
      "step": 31259
    },
    {
      "epoch": 0.9007088111565723,
      "grad_norm": 0.7317520380020142,
      "learning_rate": 2.5632607496154314e-07,
      "loss": 0.6782,
      "step": 31260
    },
    {
      "epoch": 0.9007376246182216,
      "grad_norm": 0.7196402549743652,
      "learning_rate": 2.561786126330429e-07,
      "loss": 0.6667,
      "step": 31261
    },
    {
      "epoch": 0.9007664380798709,
      "grad_norm": 0.717395544052124,
      "learning_rate": 2.5603119161848543e-07,
      "loss": 0.6595,
      "step": 31262
    },
    {
      "epoch": 0.9007952515415202,
      "grad_norm": 0.7283512949943542,
      "learning_rate": 2.5588381191915367e-07,
      "loss": 0.6835,
      "step": 31263
    },
    {
      "epoch": 0.9008240650031695,
      "grad_norm": 0.7246828079223633,
      "learning_rate": 2.557364735363316e-07,
      "loss": 0.6801,
      "step": 31264
    },
    {
      "epoch": 0.9008528784648188,
      "grad_norm": 0.7161308526992798,
      "learning_rate": 2.55589176471302e-07,
      "loss": 0.6796,
      "step": 31265
    },
    {
      "epoch": 0.900881691926468,
      "grad_norm": 0.720334529876709,
      "learning_rate": 2.5544192072534835e-07,
      "loss": 0.6991,
      "step": 31266
    },
    {
      "epoch": 0.9009105053881173,
      "grad_norm": 0.7133734226226807,
      "learning_rate": 2.5529470629975297e-07,
      "loss": 0.6665,
      "step": 31267
    },
    {
      "epoch": 0.9009393188497666,
      "grad_norm": 0.7117893695831299,
      "learning_rate": 2.5514753319579754e-07,
      "loss": 0.6848,
      "step": 31268
    },
    {
      "epoch": 0.9009681323114159,
      "grad_norm": 0.7378447651863098,
      "learning_rate": 2.550004014147639e-07,
      "loss": 0.6961,
      "step": 31269
    },
    {
      "epoch": 0.9009969457730652,
      "grad_norm": 0.7011449933052063,
      "learning_rate": 2.5485331095793267e-07,
      "loss": 0.6809,
      "step": 31270
    },
    {
      "epoch": 0.9010257592347145,
      "grad_norm": 0.7312465906143188,
      "learning_rate": 2.547062618265872e-07,
      "loss": 0.6916,
      "step": 31271
    },
    {
      "epoch": 0.9010545726963637,
      "grad_norm": 0.7455398440361023,
      "learning_rate": 2.545592540220054e-07,
      "loss": 0.682,
      "step": 31272
    },
    {
      "epoch": 0.901083386158013,
      "grad_norm": 0.7055689096450806,
      "learning_rate": 2.5441228754546963e-07,
      "loss": 0.6677,
      "step": 31273
    },
    {
      "epoch": 0.9011121996196623,
      "grad_norm": 0.7300613522529602,
      "learning_rate": 2.542653623982588e-07,
      "loss": 0.6807,
      "step": 31274
    },
    {
      "epoch": 0.9011410130813116,
      "grad_norm": 0.7201807498931885,
      "learning_rate": 2.5411847858165306e-07,
      "loss": 0.6789,
      "step": 31275
    },
    {
      "epoch": 0.9011698265429609,
      "grad_norm": 0.7072497010231018,
      "learning_rate": 2.539716360969302e-07,
      "loss": 0.6886,
      "step": 31276
    },
    {
      "epoch": 0.9011986400046101,
      "grad_norm": 0.7013221979141235,
      "learning_rate": 2.538248349453709e-07,
      "loss": 0.6983,
      "step": 31277
    },
    {
      "epoch": 0.9012274534662594,
      "grad_norm": 0.6882453560829163,
      "learning_rate": 2.5367807512825247e-07,
      "loss": 0.6527,
      "step": 31278
    },
    {
      "epoch": 0.9012562669279087,
      "grad_norm": 0.7356129288673401,
      "learning_rate": 2.5353135664685393e-07,
      "loss": 0.6874,
      "step": 31279
    },
    {
      "epoch": 0.901285080389558,
      "grad_norm": 0.7380784749984741,
      "learning_rate": 2.5338467950245194e-07,
      "loss": 0.6978,
      "step": 31280
    },
    {
      "epoch": 0.9013138938512073,
      "grad_norm": 0.7322987914085388,
      "learning_rate": 2.5323804369632446e-07,
      "loss": 0.6959,
      "step": 31281
    },
    {
      "epoch": 0.9013427073128566,
      "grad_norm": 0.717730700969696,
      "learning_rate": 2.530914492297487e-07,
      "loss": 0.6819,
      "step": 31282
    },
    {
      "epoch": 0.9013715207745059,
      "grad_norm": 0.7180164456367493,
      "learning_rate": 2.529448961040021e-07,
      "loss": 0.6851,
      "step": 31283
    },
    {
      "epoch": 0.9014003342361552,
      "grad_norm": 0.7382054328918457,
      "learning_rate": 2.527983843203591e-07,
      "loss": 0.6795,
      "step": 31284
    },
    {
      "epoch": 0.9014291476978045,
      "grad_norm": 0.7169815897941589,
      "learning_rate": 2.52651913880097e-07,
      "loss": 0.6775,
      "step": 31285
    },
    {
      "epoch": 0.9014579611594536,
      "grad_norm": 0.7144254446029663,
      "learning_rate": 2.52505484784491e-07,
      "loss": 0.6522,
      "step": 31286
    },
    {
      "epoch": 0.9014867746211029,
      "grad_norm": 0.7028779983520508,
      "learning_rate": 2.523590970348166e-07,
      "loss": 0.6576,
      "step": 31287
    },
    {
      "epoch": 0.9015155880827522,
      "grad_norm": 0.7161344885826111,
      "learning_rate": 2.522127506323491e-07,
      "loss": 0.697,
      "step": 31288
    },
    {
      "epoch": 0.9015444015444015,
      "grad_norm": 0.7580680251121521,
      "learning_rate": 2.520664455783617e-07,
      "loss": 0.6764,
      "step": 31289
    },
    {
      "epoch": 0.9015732150060508,
      "grad_norm": 0.7245597839355469,
      "learning_rate": 2.5192018187413013e-07,
      "loss": 0.6747,
      "step": 31290
    },
    {
      "epoch": 0.9016020284677001,
      "grad_norm": 0.7293309569358826,
      "learning_rate": 2.5177395952092674e-07,
      "loss": 0.6562,
      "step": 31291
    },
    {
      "epoch": 0.9016308419293494,
      "grad_norm": 0.7057734131813049,
      "learning_rate": 2.5162777852002605e-07,
      "loss": 0.6897,
      "step": 31292
    },
    {
      "epoch": 0.9016596553909987,
      "grad_norm": 0.7232502698898315,
      "learning_rate": 2.5148163887269983e-07,
      "loss": 0.6782,
      "step": 31293
    },
    {
      "epoch": 0.901688468852648,
      "grad_norm": 0.7099848389625549,
      "learning_rate": 2.513355405802231e-07,
      "loss": 0.6891,
      "step": 31294
    },
    {
      "epoch": 0.9017172823142973,
      "grad_norm": 0.720258891582489,
      "learning_rate": 2.511894836438661e-07,
      "loss": 0.6603,
      "step": 31295
    },
    {
      "epoch": 0.9017460957759466,
      "grad_norm": 0.7425694465637207,
      "learning_rate": 2.510434680649021e-07,
      "loss": 0.687,
      "step": 31296
    },
    {
      "epoch": 0.9017749092375958,
      "grad_norm": 0.7320901155471802,
      "learning_rate": 2.508974938446018e-07,
      "loss": 0.6963,
      "step": 31297
    },
    {
      "epoch": 0.9018037226992451,
      "grad_norm": 0.707538902759552,
      "learning_rate": 2.507515609842376e-07,
      "loss": 0.6691,
      "step": 31298
    },
    {
      "epoch": 0.9018325361608943,
      "grad_norm": 0.7144187092781067,
      "learning_rate": 2.5060566948507894e-07,
      "loss": 0.6759,
      "step": 31299
    },
    {
      "epoch": 0.9018613496225436,
      "grad_norm": 0.7420821785926819,
      "learning_rate": 2.504598193483981e-07,
      "loss": 0.6754,
      "step": 31300
    },
    {
      "epoch": 0.9018901630841929,
      "grad_norm": 0.712059497833252,
      "learning_rate": 2.5031401057546366e-07,
      "loss": 0.6819,
      "step": 31301
    },
    {
      "epoch": 0.9019189765458422,
      "grad_norm": 0.7173412442207336,
      "learning_rate": 2.501682431675473e-07,
      "loss": 0.6916,
      "step": 31302
    },
    {
      "epoch": 0.9019477900074915,
      "grad_norm": 0.7246128916740417,
      "learning_rate": 2.500225171259163e-07,
      "loss": 0.6856,
      "step": 31303
    },
    {
      "epoch": 0.9019766034691408,
      "grad_norm": 0.7271636724472046,
      "learning_rate": 2.4987683245184137e-07,
      "loss": 0.6698,
      "step": 31304
    },
    {
      "epoch": 0.9020054169307901,
      "grad_norm": 0.729909360408783,
      "learning_rate": 2.4973118914659035e-07,
      "loss": 0.6677,
      "step": 31305
    },
    {
      "epoch": 0.9020342303924394,
      "grad_norm": 0.7312185764312744,
      "learning_rate": 2.4958558721143333e-07,
      "loss": 0.6832,
      "step": 31306
    },
    {
      "epoch": 0.9020630438540886,
      "grad_norm": 0.7061076164245605,
      "learning_rate": 2.49440026647636e-07,
      "loss": 0.7085,
      "step": 31307
    },
    {
      "epoch": 0.9020918573157379,
      "grad_norm": 0.7482667565345764,
      "learning_rate": 2.4929450745646735e-07,
      "loss": 0.6566,
      "step": 31308
    },
    {
      "epoch": 0.9021206707773872,
      "grad_norm": 0.717056393623352,
      "learning_rate": 2.491490296391952e-07,
      "loss": 0.6819,
      "step": 31309
    },
    {
      "epoch": 0.9021494842390365,
      "grad_norm": 0.708184003829956,
      "learning_rate": 2.490035931970858e-07,
      "loss": 0.651,
      "step": 31310
    },
    {
      "epoch": 0.9021782977006858,
      "grad_norm": 0.7235247492790222,
      "learning_rate": 2.488581981314059e-07,
      "loss": 0.6744,
      "step": 31311
    },
    {
      "epoch": 0.9022071111623351,
      "grad_norm": 0.7210511565208435,
      "learning_rate": 2.4871284444342117e-07,
      "loss": 0.6606,
      "step": 31312
    },
    {
      "epoch": 0.9022359246239843,
      "grad_norm": 0.7249030470848083,
      "learning_rate": 2.485675321343989e-07,
      "loss": 0.6774,
      "step": 31313
    },
    {
      "epoch": 0.9022647380856336,
      "grad_norm": 0.7367075681686401,
      "learning_rate": 2.484222612056025e-07,
      "loss": 0.6755,
      "step": 31314
    },
    {
      "epoch": 0.9022935515472829,
      "grad_norm": 0.7313719987869263,
      "learning_rate": 2.482770316582994e-07,
      "loss": 0.6616,
      "step": 31315
    },
    {
      "epoch": 0.9023223650089321,
      "grad_norm": 0.7116139531135559,
      "learning_rate": 2.48131843493753e-07,
      "loss": 0.6674,
      "step": 31316
    },
    {
      "epoch": 0.9023511784705814,
      "grad_norm": 0.7170494198799133,
      "learning_rate": 2.479866967132277e-07,
      "loss": 0.6813,
      "step": 31317
    },
    {
      "epoch": 0.9023799919322307,
      "grad_norm": 0.7139401435852051,
      "learning_rate": 2.478415913179882e-07,
      "loss": 0.675,
      "step": 31318
    },
    {
      "epoch": 0.90240880539388,
      "grad_norm": 0.7126638889312744,
      "learning_rate": 2.476965273092985e-07,
      "loss": 0.684,
      "step": 31319
    },
    {
      "epoch": 0.9024376188555293,
      "grad_norm": 0.7613565325737,
      "learning_rate": 2.475515046884208e-07,
      "loss": 0.6672,
      "step": 31320
    },
    {
      "epoch": 0.9024664323171786,
      "grad_norm": 0.725124716758728,
      "learning_rate": 2.4740652345661974e-07,
      "loss": 0.661,
      "step": 31321
    },
    {
      "epoch": 0.9024952457788279,
      "grad_norm": 0.7115626931190491,
      "learning_rate": 2.4726158361515596e-07,
      "loss": 0.6488,
      "step": 31322
    },
    {
      "epoch": 0.9025240592404772,
      "grad_norm": 0.7547920942306519,
      "learning_rate": 2.47116685165294e-07,
      "loss": 0.6876,
      "step": 31323
    },
    {
      "epoch": 0.9025528727021265,
      "grad_norm": 0.7188279628753662,
      "learning_rate": 2.4697182810829333e-07,
      "loss": 0.6706,
      "step": 31324
    },
    {
      "epoch": 0.9025816861637758,
      "grad_norm": 0.7077640295028687,
      "learning_rate": 2.4682701244541694e-07,
      "loss": 0.674,
      "step": 31325
    },
    {
      "epoch": 0.902610499625425,
      "grad_norm": 0.7244452238082886,
      "learning_rate": 2.466822381779266e-07,
      "loss": 0.671,
      "step": 31326
    },
    {
      "epoch": 0.9026393130870742,
      "grad_norm": 0.7215639352798462,
      "learning_rate": 2.465375053070818e-07,
      "loss": 0.6985,
      "step": 31327
    },
    {
      "epoch": 0.9026681265487235,
      "grad_norm": 0.7117161750793457,
      "learning_rate": 2.4639281383414424e-07,
      "loss": 0.6839,
      "step": 31328
    },
    {
      "epoch": 0.9026969400103728,
      "grad_norm": 0.7466195821762085,
      "learning_rate": 2.46248163760372e-07,
      "loss": 0.6956,
      "step": 31329
    },
    {
      "epoch": 0.9027257534720221,
      "grad_norm": 0.7004942893981934,
      "learning_rate": 2.4610355508702723e-07,
      "loss": 0.6827,
      "step": 31330
    },
    {
      "epoch": 0.9027545669336714,
      "grad_norm": 0.7463592290878296,
      "learning_rate": 2.459589878153679e-07,
      "loss": 0.6937,
      "step": 31331
    },
    {
      "epoch": 0.9027833803953207,
      "grad_norm": 0.7995031476020813,
      "learning_rate": 2.458144619466546e-07,
      "loss": 0.7164,
      "step": 31332
    },
    {
      "epoch": 0.90281219385697,
      "grad_norm": 0.7150669693946838,
      "learning_rate": 2.456699774821436e-07,
      "loss": 0.7047,
      "step": 31333
    },
    {
      "epoch": 0.9028410073186193,
      "grad_norm": 0.720507025718689,
      "learning_rate": 2.4552553442309555e-07,
      "loss": 0.6689,
      "step": 31334
    },
    {
      "epoch": 0.9028698207802686,
      "grad_norm": 0.7199457287788391,
      "learning_rate": 2.453811327707667e-07,
      "loss": 0.6682,
      "step": 31335
    },
    {
      "epoch": 0.9028986342419179,
      "grad_norm": 0.747645914554596,
      "learning_rate": 2.4523677252641545e-07,
      "loss": 0.7011,
      "step": 31336
    },
    {
      "epoch": 0.9029274477035671,
      "grad_norm": 0.7187844514846802,
      "learning_rate": 2.450924536912985e-07,
      "loss": 0.7035,
      "step": 31337
    },
    {
      "epoch": 0.9029562611652164,
      "grad_norm": 0.7299436330795288,
      "learning_rate": 2.4494817626667443e-07,
      "loss": 0.6878,
      "step": 31338
    },
    {
      "epoch": 0.9029850746268657,
      "grad_norm": 0.7207320928573608,
      "learning_rate": 2.448039402537972e-07,
      "loss": 0.6908,
      "step": 31339
    },
    {
      "epoch": 0.9030138880885149,
      "grad_norm": 0.7590579986572266,
      "learning_rate": 2.4465974565392404e-07,
      "loss": 0.6768,
      "step": 31340
    },
    {
      "epoch": 0.9030427015501642,
      "grad_norm": 0.7246205806732178,
      "learning_rate": 2.4451559246831125e-07,
      "loss": 0.6796,
      "step": 31341
    },
    {
      "epoch": 0.9030715150118135,
      "grad_norm": 0.7141299247741699,
      "learning_rate": 2.443714806982145e-07,
      "loss": 0.6624,
      "step": 31342
    },
    {
      "epoch": 0.9031003284734628,
      "grad_norm": 0.7136168479919434,
      "learning_rate": 2.442274103448872e-07,
      "loss": 0.6881,
      "step": 31343
    },
    {
      "epoch": 0.9031291419351121,
      "grad_norm": 0.7137159109115601,
      "learning_rate": 2.440833814095861e-07,
      "loss": 0.6586,
      "step": 31344
    },
    {
      "epoch": 0.9031579553967614,
      "grad_norm": 0.7295387983322144,
      "learning_rate": 2.4393939389356414e-07,
      "loss": 0.6977,
      "step": 31345
    },
    {
      "epoch": 0.9031867688584106,
      "grad_norm": 0.7299234867095947,
      "learning_rate": 2.437954477980753e-07,
      "loss": 0.6739,
      "step": 31346
    },
    {
      "epoch": 0.9032155823200599,
      "grad_norm": 0.72669917345047,
      "learning_rate": 2.436515431243747e-07,
      "loss": 0.6587,
      "step": 31347
    },
    {
      "epoch": 0.9032443957817092,
      "grad_norm": 0.7202736139297485,
      "learning_rate": 2.4350767987371347e-07,
      "loss": 0.6694,
      "step": 31348
    },
    {
      "epoch": 0.9032732092433585,
      "grad_norm": 0.7406603693962097,
      "learning_rate": 2.433638580473463e-07,
      "loss": 0.6834,
      "step": 31349
    },
    {
      "epoch": 0.9033020227050078,
      "grad_norm": 0.7023654580116272,
      "learning_rate": 2.432200776465249e-07,
      "loss": 0.6711,
      "step": 31350
    },
    {
      "epoch": 0.9033308361666571,
      "grad_norm": 0.7249902486801147,
      "learning_rate": 2.430763386725016e-07,
      "loss": 0.6767,
      "step": 31351
    },
    {
      "epoch": 0.9033596496283064,
      "grad_norm": 0.7156206369400024,
      "learning_rate": 2.4293264112652816e-07,
      "loss": 0.6596,
      "step": 31352
    },
    {
      "epoch": 0.9033884630899556,
      "grad_norm": 0.7377842664718628,
      "learning_rate": 2.427889850098558e-07,
      "loss": 0.6478,
      "step": 31353
    },
    {
      "epoch": 0.9034172765516049,
      "grad_norm": 0.7092906832695007,
      "learning_rate": 2.426453703237358e-07,
      "loss": 0.6967,
      "step": 31354
    },
    {
      "epoch": 0.9034460900132542,
      "grad_norm": 0.7204397916793823,
      "learning_rate": 2.4250179706941934e-07,
      "loss": 0.7019,
      "step": 31355
    },
    {
      "epoch": 0.9034749034749034,
      "grad_norm": 0.733232855796814,
      "learning_rate": 2.4235826524815653e-07,
      "loss": 0.6678,
      "step": 31356
    },
    {
      "epoch": 0.9035037169365527,
      "grad_norm": 0.7189871072769165,
      "learning_rate": 2.422147748611975e-07,
      "loss": 0.6869,
      "step": 31357
    },
    {
      "epoch": 0.903532530398202,
      "grad_norm": 0.7322857975959778,
      "learning_rate": 2.420713259097912e-07,
      "loss": 0.677,
      "step": 31358
    },
    {
      "epoch": 0.9035613438598513,
      "grad_norm": 0.7326494455337524,
      "learning_rate": 2.4192791839518837e-07,
      "loss": 0.6625,
      "step": 31359
    },
    {
      "epoch": 0.9035901573215006,
      "grad_norm": 0.7121175527572632,
      "learning_rate": 2.417845523186363e-07,
      "loss": 0.6749,
      "step": 31360
    },
    {
      "epoch": 0.9036189707831499,
      "grad_norm": 0.7009145617485046,
      "learning_rate": 2.416412276813845e-07,
      "loss": 0.6512,
      "step": 31361
    },
    {
      "epoch": 0.9036477842447992,
      "grad_norm": 0.7453208565711975,
      "learning_rate": 2.41497944484681e-07,
      "loss": 0.6657,
      "step": 31362
    },
    {
      "epoch": 0.9036765977064485,
      "grad_norm": 0.7042986154556274,
      "learning_rate": 2.41354702729773e-07,
      "loss": 0.6612,
      "step": 31363
    },
    {
      "epoch": 0.9037054111680978,
      "grad_norm": 0.712508499622345,
      "learning_rate": 2.4121150241791014e-07,
      "loss": 0.6662,
      "step": 31364
    },
    {
      "epoch": 0.9037342246297471,
      "grad_norm": 0.7442620396614075,
      "learning_rate": 2.4106834355033636e-07,
      "loss": 0.7108,
      "step": 31365
    },
    {
      "epoch": 0.9037630380913964,
      "grad_norm": 0.7278650999069214,
      "learning_rate": 2.4092522612830074e-07,
      "loss": 0.6859,
      "step": 31366
    },
    {
      "epoch": 0.9037918515530455,
      "grad_norm": 0.7158353924751282,
      "learning_rate": 2.4078215015304883e-07,
      "loss": 0.6746,
      "step": 31367
    },
    {
      "epoch": 0.9038206650146948,
      "grad_norm": 0.7879675030708313,
      "learning_rate": 2.406391156258281e-07,
      "loss": 0.665,
      "step": 31368
    },
    {
      "epoch": 0.9038494784763441,
      "grad_norm": 0.7418292164802551,
      "learning_rate": 2.404961225478819e-07,
      "loss": 0.6853,
      "step": 31369
    },
    {
      "epoch": 0.9038782919379934,
      "grad_norm": 0.7333069443702698,
      "learning_rate": 2.4035317092045764e-07,
      "loss": 0.692,
      "step": 31370
    },
    {
      "epoch": 0.9039071053996427,
      "grad_norm": 0.713230311870575,
      "learning_rate": 2.4021026074479813e-07,
      "loss": 0.6781,
      "step": 31371
    },
    {
      "epoch": 0.903935918861292,
      "grad_norm": 0.7043179273605347,
      "learning_rate": 2.4006739202215025e-07,
      "loss": 0.6517,
      "step": 31372
    },
    {
      "epoch": 0.9039647323229413,
      "grad_norm": 0.7307421565055847,
      "learning_rate": 2.399245647537568e-07,
      "loss": 0.6623,
      "step": 31373
    },
    {
      "epoch": 0.9039935457845906,
      "grad_norm": 0.72223299741745,
      "learning_rate": 2.397817789408624e-07,
      "loss": 0.6825,
      "step": 31374
    },
    {
      "epoch": 0.9040223592462399,
      "grad_norm": 0.711513340473175,
      "learning_rate": 2.3963903458470937e-07,
      "loss": 0.669,
      "step": 31375
    },
    {
      "epoch": 0.9040511727078892,
      "grad_norm": 0.7391369938850403,
      "learning_rate": 2.3949633168654283e-07,
      "loss": 0.6947,
      "step": 31376
    },
    {
      "epoch": 0.9040799861695384,
      "grad_norm": 0.7207404375076294,
      "learning_rate": 2.3935367024760293e-07,
      "loss": 0.6829,
      "step": 31377
    },
    {
      "epoch": 0.9041087996311877,
      "grad_norm": 0.7112572193145752,
      "learning_rate": 2.392110502691353e-07,
      "loss": 0.6683,
      "step": 31378
    },
    {
      "epoch": 0.904137613092837,
      "grad_norm": 0.7287269830703735,
      "learning_rate": 2.3906847175237945e-07,
      "loss": 0.673,
      "step": 31379
    },
    {
      "epoch": 0.9041664265544862,
      "grad_norm": 0.7085961699485779,
      "learning_rate": 2.389259346985784e-07,
      "loss": 0.6846,
      "step": 31380
    },
    {
      "epoch": 0.9041952400161355,
      "grad_norm": 0.7013751268386841,
      "learning_rate": 2.387834391089727e-07,
      "loss": 0.6867,
      "step": 31381
    },
    {
      "epoch": 0.9042240534777848,
      "grad_norm": 0.7194358110427856,
      "learning_rate": 2.386409849848048e-07,
      "loss": 0.6477,
      "step": 31382
    },
    {
      "epoch": 0.9042528669394341,
      "grad_norm": 0.7038308382034302,
      "learning_rate": 2.384985723273131e-07,
      "loss": 0.6437,
      "step": 31383
    },
    {
      "epoch": 0.9042816804010834,
      "grad_norm": 0.7214747071266174,
      "learning_rate": 2.383562011377394e-07,
      "loss": 0.6755,
      "step": 31384
    },
    {
      "epoch": 0.9043104938627327,
      "grad_norm": 0.710969865322113,
      "learning_rate": 2.3821387141732376e-07,
      "loss": 0.7008,
      "step": 31385
    },
    {
      "epoch": 0.904339307324382,
      "grad_norm": 0.7621433138847351,
      "learning_rate": 2.3807158316730472e-07,
      "loss": 0.6911,
      "step": 31386
    },
    {
      "epoch": 0.9043681207860312,
      "grad_norm": 0.739287793636322,
      "learning_rate": 2.3792933638892234e-07,
      "loss": 0.6721,
      "step": 31387
    },
    {
      "epoch": 0.9043969342476805,
      "grad_norm": 0.7114126682281494,
      "learning_rate": 2.3778713108341513e-07,
      "loss": 0.6712,
      "step": 31388
    },
    {
      "epoch": 0.9044257477093298,
      "grad_norm": 0.7429977655410767,
      "learning_rate": 2.376449672520209e-07,
      "loss": 0.6698,
      "step": 31389
    },
    {
      "epoch": 0.9044545611709791,
      "grad_norm": 0.7315112948417664,
      "learning_rate": 2.3750284489597874e-07,
      "loss": 0.6741,
      "step": 31390
    },
    {
      "epoch": 0.9044833746326284,
      "grad_norm": 0.7162727117538452,
      "learning_rate": 2.3736076401652653e-07,
      "loss": 0.6726,
      "step": 31391
    },
    {
      "epoch": 0.9045121880942777,
      "grad_norm": 0.7249963879585266,
      "learning_rate": 2.3721872461490048e-07,
      "loss": 0.6831,
      "step": 31392
    },
    {
      "epoch": 0.904541001555927,
      "grad_norm": 0.7258575558662415,
      "learning_rate": 2.3707672669233907e-07,
      "loss": 0.6972,
      "step": 31393
    },
    {
      "epoch": 0.9045698150175762,
      "grad_norm": 0.7453554272651672,
      "learning_rate": 2.3693477025007738e-07,
      "loss": 0.7128,
      "step": 31394
    },
    {
      "epoch": 0.9045986284792255,
      "grad_norm": 0.7274580597877502,
      "learning_rate": 2.3679285528935336e-07,
      "loss": 0.6883,
      "step": 31395
    },
    {
      "epoch": 0.9046274419408747,
      "grad_norm": 0.7184687852859497,
      "learning_rate": 2.3665098181140156e-07,
      "loss": 0.6702,
      "step": 31396
    },
    {
      "epoch": 0.904656255402524,
      "grad_norm": 0.7161028981208801,
      "learning_rate": 2.365091498174582e-07,
      "loss": 0.693,
      "step": 31397
    },
    {
      "epoch": 0.9046850688641733,
      "grad_norm": 0.7438908219337463,
      "learning_rate": 2.3636735930875843e-07,
      "loss": 0.6864,
      "step": 31398
    },
    {
      "epoch": 0.9047138823258226,
      "grad_norm": 0.698373019695282,
      "learning_rate": 2.3622561028653735e-07,
      "loss": 0.6674,
      "step": 31399
    },
    {
      "epoch": 0.9047426957874719,
      "grad_norm": 0.7418864965438843,
      "learning_rate": 2.360839027520284e-07,
      "loss": 0.6648,
      "step": 31400
    },
    {
      "epoch": 0.9047715092491212,
      "grad_norm": 0.7042630314826965,
      "learning_rate": 2.3594223670646676e-07,
      "loss": 0.6613,
      "step": 31401
    },
    {
      "epoch": 0.9048003227107705,
      "grad_norm": 0.7228531837463379,
      "learning_rate": 2.3580061215108585e-07,
      "loss": 0.6708,
      "step": 31402
    },
    {
      "epoch": 0.9048291361724198,
      "grad_norm": 0.7230401635169983,
      "learning_rate": 2.3565902908711967e-07,
      "loss": 0.6679,
      "step": 31403
    },
    {
      "epoch": 0.9048579496340691,
      "grad_norm": 0.7123426198959351,
      "learning_rate": 2.355174875158006e-07,
      "loss": 0.6811,
      "step": 31404
    },
    {
      "epoch": 0.9048867630957184,
      "grad_norm": 0.7209210395812988,
      "learning_rate": 2.35375987438361e-07,
      "loss": 0.6711,
      "step": 31405
    },
    {
      "epoch": 0.9049155765573677,
      "grad_norm": 0.7443971633911133,
      "learning_rate": 2.3523452885603427e-07,
      "loss": 0.6864,
      "step": 31406
    },
    {
      "epoch": 0.9049443900190168,
      "grad_norm": 0.7331723570823669,
      "learning_rate": 2.3509311177005169e-07,
      "loss": 0.6879,
      "step": 31407
    },
    {
      "epoch": 0.9049732034806661,
      "grad_norm": 0.72347491979599,
      "learning_rate": 2.3495173618164502e-07,
      "loss": 0.6718,
      "step": 31408
    },
    {
      "epoch": 0.9050020169423154,
      "grad_norm": 0.7202072143554688,
      "learning_rate": 2.3481040209204498e-07,
      "loss": 0.6768,
      "step": 31409
    },
    {
      "epoch": 0.9050308304039647,
      "grad_norm": 0.7098655700683594,
      "learning_rate": 2.3466910950248334e-07,
      "loss": 0.6763,
      "step": 31410
    },
    {
      "epoch": 0.905059643865614,
      "grad_norm": 0.7105159759521484,
      "learning_rate": 2.3452785841418967e-07,
      "loss": 0.6745,
      "step": 31411
    },
    {
      "epoch": 0.9050884573272633,
      "grad_norm": 0.7195517420768738,
      "learning_rate": 2.3438664882839524e-07,
      "loss": 0.6751,
      "step": 31412
    },
    {
      "epoch": 0.9051172707889126,
      "grad_norm": 0.7288689613342285,
      "learning_rate": 2.3424548074632791e-07,
      "loss": 0.6837,
      "step": 31413
    },
    {
      "epoch": 0.9051460842505619,
      "grad_norm": 0.713737964630127,
      "learning_rate": 2.3410435416922006e-07,
      "loss": 0.6955,
      "step": 31414
    },
    {
      "epoch": 0.9051748977122112,
      "grad_norm": 0.7059574723243713,
      "learning_rate": 2.3396326909829792e-07,
      "loss": 0.659,
      "step": 31415
    },
    {
      "epoch": 0.9052037111738604,
      "grad_norm": 0.723779559135437,
      "learning_rate": 2.3382222553479274e-07,
      "loss": 0.6628,
      "step": 31416
    },
    {
      "epoch": 0.9052325246355097,
      "grad_norm": 0.7074745893478394,
      "learning_rate": 2.3368122347993017e-07,
      "loss": 0.677,
      "step": 31417
    },
    {
      "epoch": 0.905261338097159,
      "grad_norm": 0.7067629098892212,
      "learning_rate": 2.3354026293494037e-07,
      "loss": 0.656,
      "step": 31418
    },
    {
      "epoch": 0.9052901515588083,
      "grad_norm": 0.7447856068611145,
      "learning_rate": 2.3339934390105013e-07,
      "loss": 0.6887,
      "step": 31419
    },
    {
      "epoch": 0.9053189650204576,
      "grad_norm": 0.7288075685501099,
      "learning_rate": 2.3325846637948678e-07,
      "loss": 0.6752,
      "step": 31420
    },
    {
      "epoch": 0.9053477784821068,
      "grad_norm": 0.7377232909202576,
      "learning_rate": 2.3311763037147661e-07,
      "loss": 0.662,
      "step": 31421
    },
    {
      "epoch": 0.9053765919437561,
      "grad_norm": 0.7216107845306396,
      "learning_rate": 2.3297683587824692e-07,
      "loss": 0.6806,
      "step": 31422
    },
    {
      "epoch": 0.9054054054054054,
      "grad_norm": 0.7181398272514343,
      "learning_rate": 2.3283608290102456e-07,
      "loss": 0.6816,
      "step": 31423
    },
    {
      "epoch": 0.9054342188670547,
      "grad_norm": 0.7406002879142761,
      "learning_rate": 2.326953714410335e-07,
      "loss": 0.6748,
      "step": 31424
    },
    {
      "epoch": 0.905463032328704,
      "grad_norm": 0.7187787890434265,
      "learning_rate": 2.325547014995e-07,
      "loss": 0.687,
      "step": 31425
    },
    {
      "epoch": 0.9054918457903532,
      "grad_norm": 0.715950071811676,
      "learning_rate": 2.3241407307764974e-07,
      "loss": 0.6776,
      "step": 31426
    },
    {
      "epoch": 0.9055206592520025,
      "grad_norm": 0.7148599028587341,
      "learning_rate": 2.3227348617670788e-07,
      "loss": 0.6808,
      "step": 31427
    },
    {
      "epoch": 0.9055494727136518,
      "grad_norm": 0.7328236103057861,
      "learning_rate": 2.3213294079789672e-07,
      "loss": 0.6926,
      "step": 31428
    },
    {
      "epoch": 0.9055782861753011,
      "grad_norm": 0.7081111073493958,
      "learning_rate": 2.3199243694244255e-07,
      "loss": 0.6717,
      "step": 31429
    },
    {
      "epoch": 0.9056070996369504,
      "grad_norm": 0.7127276659011841,
      "learning_rate": 2.3185197461156717e-07,
      "loss": 0.6556,
      "step": 31430
    },
    {
      "epoch": 0.9056359130985997,
      "grad_norm": 0.7030760049819946,
      "learning_rate": 2.317115538064957e-07,
      "loss": 0.6711,
      "step": 31431
    },
    {
      "epoch": 0.905664726560249,
      "grad_norm": 0.8619281053543091,
      "learning_rate": 2.315711745284488e-07,
      "loss": 0.6964,
      "step": 31432
    },
    {
      "epoch": 0.9056935400218983,
      "grad_norm": 0.7335017919540405,
      "learning_rate": 2.3143083677865162e-07,
      "loss": 0.6914,
      "step": 31433
    },
    {
      "epoch": 0.9057223534835475,
      "grad_norm": 0.7378886342048645,
      "learning_rate": 2.3129054055832378e-07,
      "loss": 0.6757,
      "step": 31434
    },
    {
      "epoch": 0.9057511669451968,
      "grad_norm": 0.7163642048835754,
      "learning_rate": 2.3115028586868925e-07,
      "loss": 0.6801,
      "step": 31435
    },
    {
      "epoch": 0.905779980406846,
      "grad_norm": 0.7165055274963379,
      "learning_rate": 2.310100727109682e-07,
      "loss": 0.6965,
      "step": 31436
    },
    {
      "epoch": 0.9058087938684953,
      "grad_norm": 0.7368524670600891,
      "learning_rate": 2.3086990108638186e-07,
      "loss": 0.6959,
      "step": 31437
    },
    {
      "epoch": 0.9058376073301446,
      "grad_norm": 0.7281750440597534,
      "learning_rate": 2.3072977099615147e-07,
      "loss": 0.6784,
      "step": 31438
    },
    {
      "epoch": 0.9058664207917939,
      "grad_norm": 0.7446407079696655,
      "learning_rate": 2.3058968244149826e-07,
      "loss": 0.6884,
      "step": 31439
    },
    {
      "epoch": 0.9058952342534432,
      "grad_norm": 0.7114475965499878,
      "learning_rate": 2.304496354236402e-07,
      "loss": 0.675,
      "step": 31440
    },
    {
      "epoch": 0.9059240477150925,
      "grad_norm": 0.7271216511726379,
      "learning_rate": 2.3030962994379847e-07,
      "loss": 0.6993,
      "step": 31441
    },
    {
      "epoch": 0.9059528611767418,
      "grad_norm": 0.7353903651237488,
      "learning_rate": 2.3016966600319158e-07,
      "loss": 0.6877,
      "step": 31442
    },
    {
      "epoch": 0.9059816746383911,
      "grad_norm": 0.7046968936920166,
      "learning_rate": 2.3002974360303854e-07,
      "loss": 0.6439,
      "step": 31443
    },
    {
      "epoch": 0.9060104881000404,
      "grad_norm": 0.7371172904968262,
      "learning_rate": 2.2988986274455893e-07,
      "loss": 0.6875,
      "step": 31444
    },
    {
      "epoch": 0.9060393015616897,
      "grad_norm": 0.7146303057670593,
      "learning_rate": 2.2975002342896957e-07,
      "loss": 0.6851,
      "step": 31445
    },
    {
      "epoch": 0.906068115023339,
      "grad_norm": 0.7536616325378418,
      "learning_rate": 2.2961022565749003e-07,
      "loss": 0.6762,
      "step": 31446
    },
    {
      "epoch": 0.9060969284849882,
      "grad_norm": 0.7167180180549622,
      "learning_rate": 2.2947046943133544e-07,
      "loss": 0.6925,
      "step": 31447
    },
    {
      "epoch": 0.9061257419466374,
      "grad_norm": 0.7168353796005249,
      "learning_rate": 2.293307547517254e-07,
      "loss": 0.6912,
      "step": 31448
    },
    {
      "epoch": 0.9061545554082867,
      "grad_norm": 0.7318179607391357,
      "learning_rate": 2.2919108161987447e-07,
      "loss": 0.6858,
      "step": 31449
    },
    {
      "epoch": 0.906183368869936,
      "grad_norm": 0.7515577077865601,
      "learning_rate": 2.2905145003700112e-07,
      "loss": 0.6876,
      "step": 31450
    },
    {
      "epoch": 0.9062121823315853,
      "grad_norm": 0.7635377049446106,
      "learning_rate": 2.2891186000431997e-07,
      "loss": 0.6808,
      "step": 31451
    },
    {
      "epoch": 0.9062409957932346,
      "grad_norm": 0.7221590280532837,
      "learning_rate": 2.287723115230478e-07,
      "loss": 0.6809,
      "step": 31452
    },
    {
      "epoch": 0.9062698092548839,
      "grad_norm": 0.7215930223464966,
      "learning_rate": 2.2863280459439863e-07,
      "loss": 0.6801,
      "step": 31453
    },
    {
      "epoch": 0.9062986227165332,
      "grad_norm": 0.7067119479179382,
      "learning_rate": 2.284933392195887e-07,
      "loss": 0.6524,
      "step": 31454
    },
    {
      "epoch": 0.9063274361781825,
      "grad_norm": 0.7037613391876221,
      "learning_rate": 2.28353915399831e-07,
      "loss": 0.6684,
      "step": 31455
    },
    {
      "epoch": 0.9063562496398317,
      "grad_norm": 0.7088664770126343,
      "learning_rate": 2.2821453313634224e-07,
      "loss": 0.6923,
      "step": 31456
    },
    {
      "epoch": 0.906385063101481,
      "grad_norm": 0.735705554485321,
      "learning_rate": 2.280751924303337e-07,
      "loss": 0.6893,
      "step": 31457
    },
    {
      "epoch": 0.9064138765631303,
      "grad_norm": 0.7652257680892944,
      "learning_rate": 2.2793589328302057e-07,
      "loss": 0.6901,
      "step": 31458
    },
    {
      "epoch": 0.9064426900247796,
      "grad_norm": 0.7411969304084778,
      "learning_rate": 2.2779663569561515e-07,
      "loss": 0.6737,
      "step": 31459
    },
    {
      "epoch": 0.9064715034864289,
      "grad_norm": 0.7091256380081177,
      "learning_rate": 2.276574196693304e-07,
      "loss": 0.6728,
      "step": 31460
    },
    {
      "epoch": 0.9065003169480781,
      "grad_norm": 0.708513081073761,
      "learning_rate": 2.2751824520537923e-07,
      "loss": 0.6533,
      "step": 31461
    },
    {
      "epoch": 0.9065291304097274,
      "grad_norm": 0.7361253499984741,
      "learning_rate": 2.27379112304974e-07,
      "loss": 0.6979,
      "step": 31462
    },
    {
      "epoch": 0.9065579438713767,
      "grad_norm": 0.7225785851478577,
      "learning_rate": 2.2724002096932485e-07,
      "loss": 0.7014,
      "step": 31463
    },
    {
      "epoch": 0.906586757333026,
      "grad_norm": 0.7218043208122253,
      "learning_rate": 2.2710097119964415e-07,
      "loss": 0.6802,
      "step": 31464
    },
    {
      "epoch": 0.9066155707946753,
      "grad_norm": 0.7262313365936279,
      "learning_rate": 2.269619629971437e-07,
      "loss": 0.6907,
      "step": 31465
    },
    {
      "epoch": 0.9066443842563245,
      "grad_norm": 0.7120346426963806,
      "learning_rate": 2.268229963630325e-07,
      "loss": 0.6896,
      "step": 31466
    },
    {
      "epoch": 0.9066731977179738,
      "grad_norm": 0.7273064851760864,
      "learning_rate": 2.2668407129852243e-07,
      "loss": 0.6875,
      "step": 31467
    },
    {
      "epoch": 0.9067020111796231,
      "grad_norm": 0.7238070368766785,
      "learning_rate": 2.2654518780482136e-07,
      "loss": 0.6967,
      "step": 31468
    },
    {
      "epoch": 0.9067308246412724,
      "grad_norm": 0.7070819735527039,
      "learning_rate": 2.2640634588314114e-07,
      "loss": 0.669,
      "step": 31469
    },
    {
      "epoch": 0.9067596381029217,
      "grad_norm": 0.7265505790710449,
      "learning_rate": 2.2626754553468911e-07,
      "loss": 0.6655,
      "step": 31470
    },
    {
      "epoch": 0.906788451564571,
      "grad_norm": 0.7111327648162842,
      "learning_rate": 2.261287867606754e-07,
      "loss": 0.677,
      "step": 31471
    },
    {
      "epoch": 0.9068172650262203,
      "grad_norm": 0.7031008005142212,
      "learning_rate": 2.2599006956230686e-07,
      "loss": 0.6739,
      "step": 31472
    },
    {
      "epoch": 0.9068460784878696,
      "grad_norm": 0.7210219502449036,
      "learning_rate": 2.2585139394079303e-07,
      "loss": 0.6831,
      "step": 31473
    },
    {
      "epoch": 0.9068748919495189,
      "grad_norm": 0.7204639315605164,
      "learning_rate": 2.2571275989734075e-07,
      "loss": 0.6832,
      "step": 31474
    },
    {
      "epoch": 0.906903705411168,
      "grad_norm": 0.7427040934562683,
      "learning_rate": 2.2557416743315852e-07,
      "loss": 0.6942,
      "step": 31475
    },
    {
      "epoch": 0.9069325188728173,
      "grad_norm": 0.7403397560119629,
      "learning_rate": 2.25435616549452e-07,
      "loss": 0.6886,
      "step": 31476
    },
    {
      "epoch": 0.9069613323344666,
      "grad_norm": 0.7307195663452148,
      "learning_rate": 2.2529710724742915e-07,
      "loss": 0.6847,
      "step": 31477
    },
    {
      "epoch": 0.9069901457961159,
      "grad_norm": 0.7250280976295471,
      "learning_rate": 2.2515863952829508e-07,
      "loss": 0.6781,
      "step": 31478
    },
    {
      "epoch": 0.9070189592577652,
      "grad_norm": 0.7585778832435608,
      "learning_rate": 2.2502021339325664e-07,
      "loss": 0.6803,
      "step": 31479
    },
    {
      "epoch": 0.9070477727194145,
      "grad_norm": 0.6978474259376526,
      "learning_rate": 2.248818288435184e-07,
      "loss": 0.6568,
      "step": 31480
    },
    {
      "epoch": 0.9070765861810638,
      "grad_norm": 0.728771448135376,
      "learning_rate": 2.2474348588028606e-07,
      "loss": 0.7032,
      "step": 31481
    },
    {
      "epoch": 0.9071053996427131,
      "grad_norm": 0.7221553325653076,
      "learning_rate": 2.2460518450476477e-07,
      "loss": 0.6801,
      "step": 31482
    },
    {
      "epoch": 0.9071342131043624,
      "grad_norm": 0.7252734899520874,
      "learning_rate": 2.24466924718158e-07,
      "loss": 0.6858,
      "step": 31483
    },
    {
      "epoch": 0.9071630265660117,
      "grad_norm": 0.7126978635787964,
      "learning_rate": 2.243287065216715e-07,
      "loss": 0.6637,
      "step": 31484
    },
    {
      "epoch": 0.907191840027661,
      "grad_norm": 0.7333305478096008,
      "learning_rate": 2.2419052991650647e-07,
      "loss": 0.6744,
      "step": 31485
    },
    {
      "epoch": 0.9072206534893102,
      "grad_norm": 0.7178986072540283,
      "learning_rate": 2.240523949038692e-07,
      "loss": 0.675,
      "step": 31486
    },
    {
      "epoch": 0.9072494669509595,
      "grad_norm": 0.7038068771362305,
      "learning_rate": 2.2391430148496097e-07,
      "loss": 0.683,
      "step": 31487
    },
    {
      "epoch": 0.9072782804126087,
      "grad_norm": 0.7415990233421326,
      "learning_rate": 2.237762496609852e-07,
      "loss": 0.6949,
      "step": 31488
    },
    {
      "epoch": 0.907307093874258,
      "grad_norm": 0.7049795985221863,
      "learning_rate": 2.2363823943314377e-07,
      "loss": 0.6629,
      "step": 31489
    },
    {
      "epoch": 0.9073359073359073,
      "grad_norm": 0.7240287661552429,
      "learning_rate": 2.2350027080263848e-07,
      "loss": 0.6558,
      "step": 31490
    },
    {
      "epoch": 0.9073647207975566,
      "grad_norm": 0.6904599666595459,
      "learning_rate": 2.233623437706711e-07,
      "loss": 0.6739,
      "step": 31491
    },
    {
      "epoch": 0.9073935342592059,
      "grad_norm": 0.7380297780036926,
      "learning_rate": 2.2322445833844298e-07,
      "loss": 0.6795,
      "step": 31492
    },
    {
      "epoch": 0.9074223477208552,
      "grad_norm": 0.7047941088676453,
      "learning_rate": 2.230866145071542e-07,
      "loss": 0.6576,
      "step": 31493
    },
    {
      "epoch": 0.9074511611825045,
      "grad_norm": 0.7294546961784363,
      "learning_rate": 2.229488122780066e-07,
      "loss": 0.6724,
      "step": 31494
    },
    {
      "epoch": 0.9074799746441538,
      "grad_norm": 0.7439510226249695,
      "learning_rate": 2.228110516521992e-07,
      "loss": 0.6919,
      "step": 31495
    },
    {
      "epoch": 0.907508788105803,
      "grad_norm": 0.7256235480308533,
      "learning_rate": 2.2267333263093217e-07,
      "loss": 0.6576,
      "step": 31496
    },
    {
      "epoch": 0.9075376015674523,
      "grad_norm": 0.7260321378707886,
      "learning_rate": 2.22535655215404e-07,
      "loss": 0.6779,
      "step": 31497
    },
    {
      "epoch": 0.9075664150291016,
      "grad_norm": 0.7249727845191956,
      "learning_rate": 2.2239801940681594e-07,
      "loss": 0.6709,
      "step": 31498
    },
    {
      "epoch": 0.9075952284907509,
      "grad_norm": 0.743903636932373,
      "learning_rate": 2.2226042520636426e-07,
      "loss": 0.6694,
      "step": 31499
    },
    {
      "epoch": 0.9076240419524002,
      "grad_norm": 0.7172455787658691,
      "learning_rate": 2.2212287261524913e-07,
      "loss": 0.653,
      "step": 31500
    },
    {
      "epoch": 0.9076528554140495,
      "grad_norm": 0.7293558120727539,
      "learning_rate": 2.2198536163466678e-07,
      "loss": 0.694,
      "step": 31501
    },
    {
      "epoch": 0.9076816688756987,
      "grad_norm": 0.7471203207969666,
      "learning_rate": 2.2184789226581572e-07,
      "loss": 0.7057,
      "step": 31502
    },
    {
      "epoch": 0.907710482337348,
      "grad_norm": 0.8314016461372375,
      "learning_rate": 2.2171046450989386e-07,
      "loss": 0.6557,
      "step": 31503
    },
    {
      "epoch": 0.9077392957989973,
      "grad_norm": 0.7127790451049805,
      "learning_rate": 2.2157307836809695e-07,
      "loss": 0.6749,
      "step": 31504
    },
    {
      "epoch": 0.9077681092606465,
      "grad_norm": 0.7092307209968567,
      "learning_rate": 2.2143573384162232e-07,
      "loss": 0.6648,
      "step": 31505
    },
    {
      "epoch": 0.9077969227222958,
      "grad_norm": 0.6970167756080627,
      "learning_rate": 2.212984309316646e-07,
      "loss": 0.6765,
      "step": 31506
    },
    {
      "epoch": 0.9078257361839451,
      "grad_norm": 0.7279450297355652,
      "learning_rate": 2.2116116963942169e-07,
      "loss": 0.6904,
      "step": 31507
    },
    {
      "epoch": 0.9078545496455944,
      "grad_norm": 0.7190730571746826,
      "learning_rate": 2.2102394996608712e-07,
      "loss": 0.683,
      "step": 31508
    },
    {
      "epoch": 0.9078833631072437,
      "grad_norm": 0.7129976153373718,
      "learning_rate": 2.2088677191285657e-07,
      "loss": 0.6938,
      "step": 31509
    },
    {
      "epoch": 0.907912176568893,
      "grad_norm": 0.7263224124908447,
      "learning_rate": 2.2074963548092465e-07,
      "loss": 0.6694,
      "step": 31510
    },
    {
      "epoch": 0.9079409900305423,
      "grad_norm": 0.7424516081809998,
      "learning_rate": 2.2061254067148708e-07,
      "loss": 0.6911,
      "step": 31511
    },
    {
      "epoch": 0.9079698034921916,
      "grad_norm": 0.7360967993736267,
      "learning_rate": 2.2047548748573566e-07,
      "loss": 0.6944,
      "step": 31512
    },
    {
      "epoch": 0.9079986169538409,
      "grad_norm": 0.7113759517669678,
      "learning_rate": 2.2033847592486612e-07,
      "loss": 0.674,
      "step": 31513
    },
    {
      "epoch": 0.9080274304154902,
      "grad_norm": 0.718721866607666,
      "learning_rate": 2.2020150599006918e-07,
      "loss": 0.6969,
      "step": 31514
    },
    {
      "epoch": 0.9080562438771393,
      "grad_norm": 0.7224593162536621,
      "learning_rate": 2.2006457768253998e-07,
      "loss": 0.6548,
      "step": 31515
    },
    {
      "epoch": 0.9080850573387886,
      "grad_norm": 0.7317832112312317,
      "learning_rate": 2.199276910034698e-07,
      "loss": 0.6985,
      "step": 31516
    },
    {
      "epoch": 0.9081138708004379,
      "grad_norm": 0.7331830263137817,
      "learning_rate": 2.1979084595405154e-07,
      "loss": 0.6761,
      "step": 31517
    },
    {
      "epoch": 0.9081426842620872,
      "grad_norm": 0.7181898951530457,
      "learning_rate": 2.1965404253547595e-07,
      "loss": 0.6973,
      "step": 31518
    },
    {
      "epoch": 0.9081714977237365,
      "grad_norm": 0.7146555185317993,
      "learning_rate": 2.1951728074893486e-07,
      "loss": 0.6658,
      "step": 31519
    },
    {
      "epoch": 0.9082003111853858,
      "grad_norm": 0.7342650294303894,
      "learning_rate": 2.1938056059562009e-07,
      "loss": 0.6859,
      "step": 31520
    },
    {
      "epoch": 0.9082291246470351,
      "grad_norm": 0.7331576943397522,
      "learning_rate": 2.1924388207672175e-07,
      "loss": 0.6939,
      "step": 31521
    },
    {
      "epoch": 0.9082579381086844,
      "grad_norm": 0.7302202582359314,
      "learning_rate": 2.1910724519342952e-07,
      "loss": 0.679,
      "step": 31522
    },
    {
      "epoch": 0.9082867515703337,
      "grad_norm": 0.724600076675415,
      "learning_rate": 2.1897064994693406e-07,
      "loss": 0.6719,
      "step": 31523
    },
    {
      "epoch": 0.908315565031983,
      "grad_norm": 0.7262647747993469,
      "learning_rate": 2.1883409633842613e-07,
      "loss": 0.6758,
      "step": 31524
    },
    {
      "epoch": 0.9083443784936323,
      "grad_norm": 0.7068781852722168,
      "learning_rate": 2.1869758436909306e-07,
      "loss": 0.6812,
      "step": 31525
    },
    {
      "epoch": 0.9083731919552815,
      "grad_norm": 0.7025012969970703,
      "learning_rate": 2.185611140401245e-07,
      "loss": 0.6688,
      "step": 31526
    },
    {
      "epoch": 0.9084020054169308,
      "grad_norm": 0.7190932035446167,
      "learning_rate": 2.1842468535270888e-07,
      "loss": 0.699,
      "step": 31527
    },
    {
      "epoch": 0.9084308188785801,
      "grad_norm": 0.729071319103241,
      "learning_rate": 2.1828829830803478e-07,
      "loss": 0.7043,
      "step": 31528
    },
    {
      "epoch": 0.9084596323402293,
      "grad_norm": 0.7433184385299683,
      "learning_rate": 2.1815195290728896e-07,
      "loss": 0.7072,
      "step": 31529
    },
    {
      "epoch": 0.9084884458018786,
      "grad_norm": 0.7323890328407288,
      "learning_rate": 2.1801564915166052e-07,
      "loss": 0.6944,
      "step": 31530
    },
    {
      "epoch": 0.9085172592635279,
      "grad_norm": 0.7138328552246094,
      "learning_rate": 2.178793870423346e-07,
      "loss": 0.6648,
      "step": 31531
    },
    {
      "epoch": 0.9085460727251772,
      "grad_norm": 0.7143567800521851,
      "learning_rate": 2.1774316658049964e-07,
      "loss": 0.6776,
      "step": 31532
    },
    {
      "epoch": 0.9085748861868265,
      "grad_norm": 0.7041279673576355,
      "learning_rate": 2.176069877673398e-07,
      "loss": 0.6654,
      "step": 31533
    },
    {
      "epoch": 0.9086036996484758,
      "grad_norm": 0.7334715127944946,
      "learning_rate": 2.1747085060404404e-07,
      "loss": 0.6757,
      "step": 31534
    },
    {
      "epoch": 0.908632513110125,
      "grad_norm": 0.716002881526947,
      "learning_rate": 2.1733475509179536e-07,
      "loss": 0.6812,
      "step": 31535
    },
    {
      "epoch": 0.9086613265717743,
      "grad_norm": 0.7232450246810913,
      "learning_rate": 2.171987012317811e-07,
      "loss": 0.6767,
      "step": 31536
    },
    {
      "epoch": 0.9086901400334236,
      "grad_norm": 0.7179090976715088,
      "learning_rate": 2.1706268902518425e-07,
      "loss": 0.6646,
      "step": 31537
    },
    {
      "epoch": 0.9087189534950729,
      "grad_norm": 0.7050055861473083,
      "learning_rate": 2.1692671847319046e-07,
      "loss": 0.661,
      "step": 31538
    },
    {
      "epoch": 0.9087477669567222,
      "grad_norm": 0.7197228670120239,
      "learning_rate": 2.167907895769833e-07,
      "loss": 0.6868,
      "step": 31539
    },
    {
      "epoch": 0.9087765804183715,
      "grad_norm": 0.7037671804428101,
      "learning_rate": 2.1665490233774734e-07,
      "loss": 0.6718,
      "step": 31540
    },
    {
      "epoch": 0.9088053938800208,
      "grad_norm": 0.7102565169334412,
      "learning_rate": 2.1651905675666552e-07,
      "loss": 0.6522,
      "step": 31541
    },
    {
      "epoch": 0.90883420734167,
      "grad_norm": 0.7559085488319397,
      "learning_rate": 2.163832528349208e-07,
      "loss": 0.6798,
      "step": 31542
    },
    {
      "epoch": 0.9088630208033193,
      "grad_norm": 0.7163766026496887,
      "learning_rate": 2.1624749057369665e-07,
      "loss": 0.6844,
      "step": 31543
    },
    {
      "epoch": 0.9088918342649686,
      "grad_norm": 0.7302870750427246,
      "learning_rate": 2.1611176997417382e-07,
      "loss": 0.6878,
      "step": 31544
    },
    {
      "epoch": 0.9089206477266178,
      "grad_norm": 0.7252059578895569,
      "learning_rate": 2.159760910375358e-07,
      "loss": 0.6977,
      "step": 31545
    },
    {
      "epoch": 0.9089494611882671,
      "grad_norm": 0.7158402800559998,
      "learning_rate": 2.1584045376496388e-07,
      "loss": 0.6718,
      "step": 31546
    },
    {
      "epoch": 0.9089782746499164,
      "grad_norm": 0.7244638204574585,
      "learning_rate": 2.1570485815763932e-07,
      "loss": 0.6596,
      "step": 31547
    },
    {
      "epoch": 0.9090070881115657,
      "grad_norm": 0.7347728610038757,
      "learning_rate": 2.1556930421674227e-07,
      "loss": 0.6718,
      "step": 31548
    },
    {
      "epoch": 0.909035901573215,
      "grad_norm": 0.7307023406028748,
      "learning_rate": 2.1543379194345516e-07,
      "loss": 0.6713,
      "step": 31549
    },
    {
      "epoch": 0.9090647150348643,
      "grad_norm": 0.7060407996177673,
      "learning_rate": 2.152983213389559e-07,
      "loss": 0.6465,
      "step": 31550
    },
    {
      "epoch": 0.9090935284965136,
      "grad_norm": 0.7121175527572632,
      "learning_rate": 2.151628924044258e-07,
      "loss": 0.6885,
      "step": 31551
    },
    {
      "epoch": 0.9091223419581629,
      "grad_norm": 0.7131579518318176,
      "learning_rate": 2.150275051410433e-07,
      "loss": 0.6594,
      "step": 31552
    },
    {
      "epoch": 0.9091511554198122,
      "grad_norm": 0.7181227207183838,
      "learning_rate": 2.1489215954998867e-07,
      "loss": 0.6595,
      "step": 31553
    },
    {
      "epoch": 0.9091799688814615,
      "grad_norm": 0.7278980016708374,
      "learning_rate": 2.147568556324392e-07,
      "loss": 0.6621,
      "step": 31554
    },
    {
      "epoch": 0.9092087823431108,
      "grad_norm": 0.7257643342018127,
      "learning_rate": 2.1462159338957455e-07,
      "loss": 0.679,
      "step": 31555
    },
    {
      "epoch": 0.9092375958047599,
      "grad_norm": 0.7382318377494812,
      "learning_rate": 2.1448637282257157e-07,
      "loss": 0.6886,
      "step": 31556
    },
    {
      "epoch": 0.9092664092664092,
      "grad_norm": 0.7465295791625977,
      "learning_rate": 2.143511939326093e-07,
      "loss": 0.6776,
      "step": 31557
    },
    {
      "epoch": 0.9092952227280585,
      "grad_norm": 0.7158471345901489,
      "learning_rate": 2.1421605672086343e-07,
      "loss": 0.6755,
      "step": 31558
    },
    {
      "epoch": 0.9093240361897078,
      "grad_norm": 0.7210054397583008,
      "learning_rate": 2.1408096118851252e-07,
      "loss": 0.6749,
      "step": 31559
    },
    {
      "epoch": 0.9093528496513571,
      "grad_norm": 0.7384793162345886,
      "learning_rate": 2.139459073367317e-07,
      "loss": 0.6988,
      "step": 31560
    },
    {
      "epoch": 0.9093816631130064,
      "grad_norm": 0.7302911281585693,
      "learning_rate": 2.1381089516669783e-07,
      "loss": 0.6798,
      "step": 31561
    },
    {
      "epoch": 0.9094104765746557,
      "grad_norm": 0.7379843592643738,
      "learning_rate": 2.136759246795872e-07,
      "loss": 0.6951,
      "step": 31562
    },
    {
      "epoch": 0.909439290036305,
      "grad_norm": 0.715787947177887,
      "learning_rate": 2.1354099587657383e-07,
      "loss": 0.6911,
      "step": 31563
    },
    {
      "epoch": 0.9094681034979543,
      "grad_norm": 0.7085850238800049,
      "learning_rate": 2.134061087588346e-07,
      "loss": 0.6698,
      "step": 31564
    },
    {
      "epoch": 0.9094969169596036,
      "grad_norm": 0.7494443655014038,
      "learning_rate": 2.1327126332754243e-07,
      "loss": 0.6785,
      "step": 31565
    },
    {
      "epoch": 0.9095257304212528,
      "grad_norm": 0.730754017829895,
      "learning_rate": 2.1313645958387308e-07,
      "loss": 0.6592,
      "step": 31566
    },
    {
      "epoch": 0.9095545438829021,
      "grad_norm": 0.7054844498634338,
      "learning_rate": 2.1300169752900003e-07,
      "loss": 0.6745,
      "step": 31567
    },
    {
      "epoch": 0.9095833573445514,
      "grad_norm": 0.7071132063865662,
      "learning_rate": 2.1286697716409621e-07,
      "loss": 0.6695,
      "step": 31568
    },
    {
      "epoch": 0.9096121708062006,
      "grad_norm": 0.7360710501670837,
      "learning_rate": 2.1273229849033628e-07,
      "loss": 0.7061,
      "step": 31569
    },
    {
      "epoch": 0.9096409842678499,
      "grad_norm": 0.7371011972427368,
      "learning_rate": 2.125976615088926e-07,
      "loss": 0.6821,
      "step": 31570
    },
    {
      "epoch": 0.9096697977294992,
      "grad_norm": 0.704193115234375,
      "learning_rate": 2.124630662209376e-07,
      "loss": 0.6693,
      "step": 31571
    },
    {
      "epoch": 0.9096986111911485,
      "grad_norm": 0.7335010170936584,
      "learning_rate": 2.1232851262764366e-07,
      "loss": 0.7144,
      "step": 31572
    },
    {
      "epoch": 0.9097274246527978,
      "grad_norm": 0.7157847285270691,
      "learning_rate": 2.1219400073018203e-07,
      "loss": 0.6567,
      "step": 31573
    },
    {
      "epoch": 0.9097562381144471,
      "grad_norm": 0.7401134967803955,
      "learning_rate": 2.1205953052972516e-07,
      "loss": 0.6787,
      "step": 31574
    },
    {
      "epoch": 0.9097850515760963,
      "grad_norm": 0.7168725728988647,
      "learning_rate": 2.1192510202744322e-07,
      "loss": 0.6748,
      "step": 31575
    },
    {
      "epoch": 0.9098138650377456,
      "grad_norm": 0.7441776990890503,
      "learning_rate": 2.1179071522450745e-07,
      "loss": 0.6955,
      "step": 31576
    },
    {
      "epoch": 0.9098426784993949,
      "grad_norm": 0.7295196056365967,
      "learning_rate": 2.1165637012208806e-07,
      "loss": 0.6891,
      "step": 31577
    },
    {
      "epoch": 0.9098714919610442,
      "grad_norm": 0.7440439462661743,
      "learning_rate": 2.1152206672135468e-07,
      "loss": 0.7029,
      "step": 31578
    },
    {
      "epoch": 0.9099003054226935,
      "grad_norm": 0.711136519908905,
      "learning_rate": 2.11387805023478e-07,
      "loss": 0.6734,
      "step": 31579
    },
    {
      "epoch": 0.9099291188843428,
      "grad_norm": 0.7099027037620544,
      "learning_rate": 2.11253585029626e-07,
      "loss": 0.6547,
      "step": 31580
    },
    {
      "epoch": 0.9099579323459921,
      "grad_norm": 0.7234064936637878,
      "learning_rate": 2.1111940674096832e-07,
      "loss": 0.6853,
      "step": 31581
    },
    {
      "epoch": 0.9099867458076414,
      "grad_norm": 0.7165348529815674,
      "learning_rate": 2.1098527015867453e-07,
      "loss": 0.6966,
      "step": 31582
    },
    {
      "epoch": 0.9100155592692906,
      "grad_norm": 0.7278946042060852,
      "learning_rate": 2.1085117528391098e-07,
      "loss": 0.7003,
      "step": 31583
    },
    {
      "epoch": 0.9100443727309399,
      "grad_norm": 0.7157314419746399,
      "learning_rate": 2.1071712211784612e-07,
      "loss": 0.6697,
      "step": 31584
    },
    {
      "epoch": 0.9100731861925891,
      "grad_norm": 0.7053660750389099,
      "learning_rate": 2.1058311066164793e-07,
      "loss": 0.6695,
      "step": 31585
    },
    {
      "epoch": 0.9101019996542384,
      "grad_norm": 0.7128182649612427,
      "learning_rate": 2.1044914091648273e-07,
      "loss": 0.6618,
      "step": 31586
    },
    {
      "epoch": 0.9101308131158877,
      "grad_norm": 0.7305976152420044,
      "learning_rate": 2.103152128835184e-07,
      "loss": 0.6931,
      "step": 31587
    },
    {
      "epoch": 0.910159626577537,
      "grad_norm": 0.7080334424972534,
      "learning_rate": 2.1018132656392022e-07,
      "loss": 0.6873,
      "step": 31588
    },
    {
      "epoch": 0.9101884400391863,
      "grad_norm": 0.7260454297065735,
      "learning_rate": 2.1004748195885548e-07,
      "loss": 0.6752,
      "step": 31589
    },
    {
      "epoch": 0.9102172535008356,
      "grad_norm": 0.7179112434387207,
      "learning_rate": 2.099136790694878e-07,
      "loss": 0.6697,
      "step": 31590
    },
    {
      "epoch": 0.9102460669624849,
      "grad_norm": 0.7204611897468567,
      "learning_rate": 2.097799178969845e-07,
      "loss": 0.6732,
      "step": 31591
    },
    {
      "epoch": 0.9102748804241342,
      "grad_norm": 0.7396252155303955,
      "learning_rate": 2.0964619844250798e-07,
      "loss": 0.6893,
      "step": 31592
    },
    {
      "epoch": 0.9103036938857835,
      "grad_norm": 0.7253586649894714,
      "learning_rate": 2.0951252070722628e-07,
      "loss": 0.6899,
      "step": 31593
    },
    {
      "epoch": 0.9103325073474328,
      "grad_norm": 0.7271673083305359,
      "learning_rate": 2.0937888469230117e-07,
      "loss": 0.6655,
      "step": 31594
    },
    {
      "epoch": 0.910361320809082,
      "grad_norm": 0.7503384351730347,
      "learning_rate": 2.092452903988973e-07,
      "loss": 0.689,
      "step": 31595
    },
    {
      "epoch": 0.9103901342707312,
      "grad_norm": 0.7614181041717529,
      "learning_rate": 2.0911173782817762e-07,
      "loss": 0.689,
      "step": 31596
    },
    {
      "epoch": 0.9104189477323805,
      "grad_norm": 0.7188196778297424,
      "learning_rate": 2.089782269813062e-07,
      "loss": 0.6827,
      "step": 31597
    },
    {
      "epoch": 0.9104477611940298,
      "grad_norm": 0.7305052280426025,
      "learning_rate": 2.0884475785944437e-07,
      "loss": 0.6716,
      "step": 31598
    },
    {
      "epoch": 0.9104765746556791,
      "grad_norm": 0.7465987801551819,
      "learning_rate": 2.0871133046375557e-07,
      "loss": 0.6609,
      "step": 31599
    },
    {
      "epoch": 0.9105053881173284,
      "grad_norm": 0.7186824679374695,
      "learning_rate": 2.0857794479540172e-07,
      "loss": 0.6589,
      "step": 31600
    },
    {
      "epoch": 0.9105342015789777,
      "grad_norm": 0.7226389646530151,
      "learning_rate": 2.0844460085554408e-07,
      "loss": 0.6843,
      "step": 31601
    },
    {
      "epoch": 0.910563015040627,
      "grad_norm": 0.724929928779602,
      "learning_rate": 2.083112986453445e-07,
      "loss": 0.6887,
      "step": 31602
    },
    {
      "epoch": 0.9105918285022763,
      "grad_norm": 0.7487994432449341,
      "learning_rate": 2.0817803816596317e-07,
      "loss": 0.6942,
      "step": 31603
    },
    {
      "epoch": 0.9106206419639256,
      "grad_norm": 0.729175329208374,
      "learning_rate": 2.0804481941856081e-07,
      "loss": 0.7017,
      "step": 31604
    },
    {
      "epoch": 0.9106494554255748,
      "grad_norm": 0.7328388094902039,
      "learning_rate": 2.0791164240429818e-07,
      "loss": 0.689,
      "step": 31605
    },
    {
      "epoch": 0.9106782688872241,
      "grad_norm": 0.7054126262664795,
      "learning_rate": 2.0777850712433545e-07,
      "loss": 0.6602,
      "step": 31606
    },
    {
      "epoch": 0.9107070823488734,
      "grad_norm": 0.7293880581855774,
      "learning_rate": 2.076454135798306e-07,
      "loss": 0.6853,
      "step": 31607
    },
    {
      "epoch": 0.9107358958105227,
      "grad_norm": 0.7395138144493103,
      "learning_rate": 2.0751236177194435e-07,
      "loss": 0.6746,
      "step": 31608
    },
    {
      "epoch": 0.910764709272172,
      "grad_norm": 0.7245134711265564,
      "learning_rate": 2.073793517018341e-07,
      "loss": 0.683,
      "step": 31609
    },
    {
      "epoch": 0.9107935227338212,
      "grad_norm": 0.7269086837768555,
      "learning_rate": 2.072463833706595e-07,
      "loss": 0.6909,
      "step": 31610
    },
    {
      "epoch": 0.9108223361954705,
      "grad_norm": 0.7201257348060608,
      "learning_rate": 2.0711345677957683e-07,
      "loss": 0.6854,
      "step": 31611
    },
    {
      "epoch": 0.9108511496571198,
      "grad_norm": 0.7366417050361633,
      "learning_rate": 2.0698057192974574e-07,
      "loss": 0.6715,
      "step": 31612
    },
    {
      "epoch": 0.9108799631187691,
      "grad_norm": 0.7096394300460815,
      "learning_rate": 2.0684772882232195e-07,
      "loss": 0.6771,
      "step": 31613
    },
    {
      "epoch": 0.9109087765804184,
      "grad_norm": 0.7162228226661682,
      "learning_rate": 2.06714927458464e-07,
      "loss": 0.6784,
      "step": 31614
    },
    {
      "epoch": 0.9109375900420676,
      "grad_norm": 0.7346343994140625,
      "learning_rate": 2.065821678393265e-07,
      "loss": 0.6996,
      "step": 31615
    },
    {
      "epoch": 0.9109664035037169,
      "grad_norm": 0.7186387181282043,
      "learning_rate": 2.0644944996606687e-07,
      "loss": 0.6552,
      "step": 31616
    },
    {
      "epoch": 0.9109952169653662,
      "grad_norm": 0.7175353765487671,
      "learning_rate": 2.063167738398403e-07,
      "loss": 0.6614,
      "step": 31617
    },
    {
      "epoch": 0.9110240304270155,
      "grad_norm": 0.7170310616493225,
      "learning_rate": 2.061841394618036e-07,
      "loss": 0.6895,
      "step": 31618
    },
    {
      "epoch": 0.9110528438886648,
      "grad_norm": 0.7100851535797119,
      "learning_rate": 2.060515468331109e-07,
      "loss": 0.6789,
      "step": 31619
    },
    {
      "epoch": 0.9110816573503141,
      "grad_norm": 0.7444106340408325,
      "learning_rate": 2.0591899595491682e-07,
      "loss": 0.6829,
      "step": 31620
    },
    {
      "epoch": 0.9111104708119634,
      "grad_norm": 0.7311579585075378,
      "learning_rate": 2.05786486828376e-07,
      "loss": 0.6989,
      "step": 31621
    },
    {
      "epoch": 0.9111392842736127,
      "grad_norm": 0.7175740003585815,
      "learning_rate": 2.0565401945464193e-07,
      "loss": 0.6839,
      "step": 31622
    },
    {
      "epoch": 0.9111680977352619,
      "grad_norm": 0.7181791663169861,
      "learning_rate": 2.0552159383486925e-07,
      "loss": 0.6735,
      "step": 31623
    },
    {
      "epoch": 0.9111969111969112,
      "grad_norm": 0.7240517139434814,
      "learning_rate": 2.053892099702104e-07,
      "loss": 0.6607,
      "step": 31624
    },
    {
      "epoch": 0.9112257246585604,
      "grad_norm": 0.7282246351242065,
      "learning_rate": 2.0525686786181943e-07,
      "loss": 0.6973,
      "step": 31625
    },
    {
      "epoch": 0.9112545381202097,
      "grad_norm": 0.7403459548950195,
      "learning_rate": 2.0512456751084763e-07,
      "loss": 0.712,
      "step": 31626
    },
    {
      "epoch": 0.911283351581859,
      "grad_norm": 0.7257274389266968,
      "learning_rate": 2.04992308918448e-07,
      "loss": 0.7016,
      "step": 31627
    },
    {
      "epoch": 0.9113121650435083,
      "grad_norm": 0.7304509878158569,
      "learning_rate": 2.0486009208577128e-07,
      "loss": 0.6669,
      "step": 31628
    },
    {
      "epoch": 0.9113409785051576,
      "grad_norm": 0.7272776961326599,
      "learning_rate": 2.0472791701397043e-07,
      "loss": 0.6871,
      "step": 31629
    },
    {
      "epoch": 0.9113697919668069,
      "grad_norm": 0.734175443649292,
      "learning_rate": 2.0459578370419563e-07,
      "loss": 0.6886,
      "step": 31630
    },
    {
      "epoch": 0.9113986054284562,
      "grad_norm": 0.7311010360717773,
      "learning_rate": 2.0446369215759876e-07,
      "loss": 0.691,
      "step": 31631
    },
    {
      "epoch": 0.9114274188901055,
      "grad_norm": 0.7297621965408325,
      "learning_rate": 2.0433164237532832e-07,
      "loss": 0.7043,
      "step": 31632
    },
    {
      "epoch": 0.9114562323517548,
      "grad_norm": 0.7204145789146423,
      "learning_rate": 2.041996343585362e-07,
      "loss": 0.6806,
      "step": 31633
    },
    {
      "epoch": 0.9114850458134041,
      "grad_norm": 0.7291021943092346,
      "learning_rate": 2.040676681083703e-07,
      "loss": 0.6732,
      "step": 31634
    },
    {
      "epoch": 0.9115138592750534,
      "grad_norm": 0.7434563040733337,
      "learning_rate": 2.0393574362598202e-07,
      "loss": 0.6882,
      "step": 31635
    },
    {
      "epoch": 0.9115426727367026,
      "grad_norm": 0.7175540328025818,
      "learning_rate": 2.0380386091251813e-07,
      "loss": 0.6818,
      "step": 31636
    },
    {
      "epoch": 0.9115714861983518,
      "grad_norm": 0.7312337756156921,
      "learning_rate": 2.0367201996912779e-07,
      "loss": 0.6973,
      "step": 31637
    },
    {
      "epoch": 0.9116002996600011,
      "grad_norm": 0.7161271572113037,
      "learning_rate": 2.035402207969611e-07,
      "loss": 0.6698,
      "step": 31638
    },
    {
      "epoch": 0.9116291131216504,
      "grad_norm": 0.725724995136261,
      "learning_rate": 2.034084633971628e-07,
      "loss": 0.6657,
      "step": 31639
    },
    {
      "epoch": 0.9116579265832997,
      "grad_norm": 0.7133716940879822,
      "learning_rate": 2.0327674777088247e-07,
      "loss": 0.6814,
      "step": 31640
    },
    {
      "epoch": 0.911686740044949,
      "grad_norm": 0.7373013496398926,
      "learning_rate": 2.0314507391926753e-07,
      "loss": 0.6801,
      "step": 31641
    },
    {
      "epoch": 0.9117155535065983,
      "grad_norm": 0.8301945924758911,
      "learning_rate": 2.030134418434626e-07,
      "loss": 0.6651,
      "step": 31642
    },
    {
      "epoch": 0.9117443669682476,
      "grad_norm": 0.7172823548316956,
      "learning_rate": 2.0288185154461626e-07,
      "loss": 0.6707,
      "step": 31643
    },
    {
      "epoch": 0.9117731804298969,
      "grad_norm": 0.7253229022026062,
      "learning_rate": 2.0275030302387365e-07,
      "loss": 0.6744,
      "step": 31644
    },
    {
      "epoch": 0.9118019938915461,
      "grad_norm": 0.7115705013275146,
      "learning_rate": 2.0261879628238e-07,
      "loss": 0.6763,
      "step": 31645
    },
    {
      "epoch": 0.9118308073531954,
      "grad_norm": 0.7368462681770325,
      "learning_rate": 2.0248733132128107e-07,
      "loss": 0.6771,
      "step": 31646
    },
    {
      "epoch": 0.9118596208148447,
      "grad_norm": 0.728611946105957,
      "learning_rate": 2.02355908141722e-07,
      "loss": 0.6802,
      "step": 31647
    },
    {
      "epoch": 0.911888434276494,
      "grad_norm": 0.7306041717529297,
      "learning_rate": 2.0222452674484695e-07,
      "loss": 0.7058,
      "step": 31648
    },
    {
      "epoch": 0.9119172477381433,
      "grad_norm": 0.707865297794342,
      "learning_rate": 2.0209318713179994e-07,
      "loss": 0.6498,
      "step": 31649
    },
    {
      "epoch": 0.9119460611997925,
      "grad_norm": 0.717138946056366,
      "learning_rate": 2.0196188930372562e-07,
      "loss": 0.668,
      "step": 31650
    },
    {
      "epoch": 0.9119748746614418,
      "grad_norm": 0.717552900314331,
      "learning_rate": 2.0183063326176645e-07,
      "loss": 0.6779,
      "step": 31651
    },
    {
      "epoch": 0.9120036881230911,
      "grad_norm": 0.7434765696525574,
      "learning_rate": 2.0169941900706647e-07,
      "loss": 0.662,
      "step": 31652
    },
    {
      "epoch": 0.9120325015847404,
      "grad_norm": 0.7176098823547363,
      "learning_rate": 2.0156824654076755e-07,
      "loss": 0.6813,
      "step": 31653
    },
    {
      "epoch": 0.9120613150463897,
      "grad_norm": 0.711901068687439,
      "learning_rate": 2.0143711586401326e-07,
      "loss": 0.6873,
      "step": 31654
    },
    {
      "epoch": 0.9120901285080389,
      "grad_norm": 0.7221172451972961,
      "learning_rate": 2.0130602697794377e-07,
      "loss": 0.6942,
      "step": 31655
    },
    {
      "epoch": 0.9121189419696882,
      "grad_norm": 0.7212774157524109,
      "learning_rate": 2.0117497988370315e-07,
      "loss": 0.681,
      "step": 31656
    },
    {
      "epoch": 0.9121477554313375,
      "grad_norm": 0.7262812852859497,
      "learning_rate": 2.0104397458243052e-07,
      "loss": 0.6876,
      "step": 31657
    },
    {
      "epoch": 0.9121765688929868,
      "grad_norm": 0.7376290559768677,
      "learning_rate": 2.0091301107526773e-07,
      "loss": 0.6693,
      "step": 31658
    },
    {
      "epoch": 0.9122053823546361,
      "grad_norm": 0.7266740202903748,
      "learning_rate": 2.0078208936335552e-07,
      "loss": 0.6796,
      "step": 31659
    },
    {
      "epoch": 0.9122341958162854,
      "grad_norm": 0.7195622324943542,
      "learning_rate": 2.0065120944783355e-07,
      "loss": 0.6545,
      "step": 31660
    },
    {
      "epoch": 0.9122630092779347,
      "grad_norm": 0.7189425230026245,
      "learning_rate": 2.005203713298426e-07,
      "loss": 0.6921,
      "step": 31661
    },
    {
      "epoch": 0.912291822739584,
      "grad_norm": 0.7465268969535828,
      "learning_rate": 2.0038957501052113e-07,
      "loss": 0.6645,
      "step": 31662
    },
    {
      "epoch": 0.9123206362012333,
      "grad_norm": 0.7219449281692505,
      "learning_rate": 2.0025882049100886e-07,
      "loss": 0.6833,
      "step": 31663
    },
    {
      "epoch": 0.9123494496628825,
      "grad_norm": 0.7311064600944519,
      "learning_rate": 2.0012810777244317e-07,
      "loss": 0.6768,
      "step": 31664
    },
    {
      "epoch": 0.9123782631245317,
      "grad_norm": 0.7338333129882812,
      "learning_rate": 1.9999743685596484e-07,
      "loss": 0.6853,
      "step": 31665
    },
    {
      "epoch": 0.912407076586181,
      "grad_norm": 0.7852910757064819,
      "learning_rate": 1.998668077427096e-07,
      "loss": 0.6923,
      "step": 31666
    },
    {
      "epoch": 0.9124358900478303,
      "grad_norm": 0.7182508707046509,
      "learning_rate": 1.9973622043381712e-07,
      "loss": 0.6845,
      "step": 31667
    },
    {
      "epoch": 0.9124647035094796,
      "grad_norm": 0.7188873887062073,
      "learning_rate": 1.996056749304226e-07,
      "loss": 0.7011,
      "step": 31668
    },
    {
      "epoch": 0.9124935169711289,
      "grad_norm": 0.7303317189216614,
      "learning_rate": 1.9947517123366512e-07,
      "loss": 0.6758,
      "step": 31669
    },
    {
      "epoch": 0.9125223304327782,
      "grad_norm": 0.7315539121627808,
      "learning_rate": 1.993447093446793e-07,
      "loss": 0.6953,
      "step": 31670
    },
    {
      "epoch": 0.9125511438944275,
      "grad_norm": 0.7691141366958618,
      "learning_rate": 1.9921428926460263e-07,
      "loss": 0.7005,
      "step": 31671
    },
    {
      "epoch": 0.9125799573560768,
      "grad_norm": 0.7108851671218872,
      "learning_rate": 1.9908391099457026e-07,
      "loss": 0.6722,
      "step": 31672
    },
    {
      "epoch": 0.9126087708177261,
      "grad_norm": 0.7134304642677307,
      "learning_rate": 1.9895357453571796e-07,
      "loss": 0.6524,
      "step": 31673
    },
    {
      "epoch": 0.9126375842793754,
      "grad_norm": 0.7071855068206787,
      "learning_rate": 1.9882327988918037e-07,
      "loss": 0.6541,
      "step": 31674
    },
    {
      "epoch": 0.9126663977410246,
      "grad_norm": 0.7210395932197571,
      "learning_rate": 1.9869302705609273e-07,
      "loss": 0.6873,
      "step": 31675
    },
    {
      "epoch": 0.9126952112026739,
      "grad_norm": 0.7409003376960754,
      "learning_rate": 1.985628160375891e-07,
      "loss": 0.6795,
      "step": 31676
    },
    {
      "epoch": 0.9127240246643231,
      "grad_norm": 0.7100420594215393,
      "learning_rate": 1.9843264683480468e-07,
      "loss": 0.6712,
      "step": 31677
    },
    {
      "epoch": 0.9127528381259724,
      "grad_norm": 0.7001190185546875,
      "learning_rate": 1.983025194488708e-07,
      "loss": 0.6453,
      "step": 31678
    },
    {
      "epoch": 0.9127816515876217,
      "grad_norm": 0.7187187075614929,
      "learning_rate": 1.9817243388092323e-07,
      "loss": 0.7111,
      "step": 31679
    },
    {
      "epoch": 0.912810465049271,
      "grad_norm": 0.7150541543960571,
      "learning_rate": 1.980423901320927e-07,
      "loss": 0.6624,
      "step": 31680
    },
    {
      "epoch": 0.9128392785109203,
      "grad_norm": 0.7110422849655151,
      "learning_rate": 1.9791238820351278e-07,
      "loss": 0.6806,
      "step": 31681
    },
    {
      "epoch": 0.9128680919725696,
      "grad_norm": 0.7489230632781982,
      "learning_rate": 1.9778242809631642e-07,
      "loss": 0.6955,
      "step": 31682
    },
    {
      "epoch": 0.9128969054342189,
      "grad_norm": 0.7583437561988831,
      "learning_rate": 1.9765250981163387e-07,
      "loss": 0.6921,
      "step": 31683
    },
    {
      "epoch": 0.9129257188958682,
      "grad_norm": 0.7502217888832092,
      "learning_rate": 1.9752263335059806e-07,
      "loss": 0.6842,
      "step": 31684
    },
    {
      "epoch": 0.9129545323575174,
      "grad_norm": 0.7192662358283997,
      "learning_rate": 1.973927987143387e-07,
      "loss": 0.6838,
      "step": 31685
    },
    {
      "epoch": 0.9129833458191667,
      "grad_norm": 0.7258001565933228,
      "learning_rate": 1.9726300590398816e-07,
      "loss": 0.704,
      "step": 31686
    },
    {
      "epoch": 0.913012159280816,
      "grad_norm": 0.7526232004165649,
      "learning_rate": 1.9713325492067504e-07,
      "loss": 0.6959,
      "step": 31687
    },
    {
      "epoch": 0.9130409727424653,
      "grad_norm": 0.7277880907058716,
      "learning_rate": 1.9700354576553006e-07,
      "loss": 0.6866,
      "step": 31688
    },
    {
      "epoch": 0.9130697862041146,
      "grad_norm": 0.7152006030082703,
      "learning_rate": 1.9687387843968287e-07,
      "loss": 0.6658,
      "step": 31689
    },
    {
      "epoch": 0.9130985996657639,
      "grad_norm": 0.726241946220398,
      "learning_rate": 1.967442529442637e-07,
      "loss": 0.6907,
      "step": 31690
    },
    {
      "epoch": 0.9131274131274131,
      "grad_norm": 0.7121580839157104,
      "learning_rate": 1.9661466928039997e-07,
      "loss": 0.693,
      "step": 31691
    },
    {
      "epoch": 0.9131562265890624,
      "grad_norm": 0.7353387475013733,
      "learning_rate": 1.9648512744922132e-07,
      "loss": 0.7169,
      "step": 31692
    },
    {
      "epoch": 0.9131850400507117,
      "grad_norm": 0.7166880369186401,
      "learning_rate": 1.9635562745185466e-07,
      "loss": 0.6889,
      "step": 31693
    },
    {
      "epoch": 0.913213853512361,
      "grad_norm": 0.7491193413734436,
      "learning_rate": 1.962261692894296e-07,
      "loss": 0.6634,
      "step": 31694
    },
    {
      "epoch": 0.9132426669740102,
      "grad_norm": 0.7331994771957397,
      "learning_rate": 1.9609675296307195e-07,
      "loss": 0.6692,
      "step": 31695
    },
    {
      "epoch": 0.9132714804356595,
      "grad_norm": 0.719622015953064,
      "learning_rate": 1.9596737847390913e-07,
      "loss": 0.6745,
      "step": 31696
    },
    {
      "epoch": 0.9133002938973088,
      "grad_norm": 0.7330936193466187,
      "learning_rate": 1.958380458230691e-07,
      "loss": 0.6889,
      "step": 31697
    },
    {
      "epoch": 0.9133291073589581,
      "grad_norm": 0.6968420147895813,
      "learning_rate": 1.9570875501167651e-07,
      "loss": 0.6531,
      "step": 31698
    },
    {
      "epoch": 0.9133579208206074,
      "grad_norm": 0.7173160910606384,
      "learning_rate": 1.9557950604085885e-07,
      "loss": 0.6871,
      "step": 31699
    },
    {
      "epoch": 0.9133867342822567,
      "grad_norm": 0.7224859595298767,
      "learning_rate": 1.9545029891174017e-07,
      "loss": 0.6548,
      "step": 31700
    },
    {
      "epoch": 0.913415547743906,
      "grad_norm": 0.7264130115509033,
      "learning_rate": 1.953211336254468e-07,
      "loss": 0.6891,
      "step": 31701
    },
    {
      "epoch": 0.9134443612055553,
      "grad_norm": 0.7343130111694336,
      "learning_rate": 1.951920101831034e-07,
      "loss": 0.6903,
      "step": 31702
    },
    {
      "epoch": 0.9134731746672046,
      "grad_norm": 0.7457412481307983,
      "learning_rate": 1.950629285858352e-07,
      "loss": 0.7078,
      "step": 31703
    },
    {
      "epoch": 0.9135019881288537,
      "grad_norm": 0.7207112312316895,
      "learning_rate": 1.949338888347646e-07,
      "loss": 0.6725,
      "step": 31704
    },
    {
      "epoch": 0.913530801590503,
      "grad_norm": 0.7303913235664368,
      "learning_rate": 1.9480489093101796e-07,
      "loss": 0.6807,
      "step": 31705
    },
    {
      "epoch": 0.9135596150521523,
      "grad_norm": 0.7238057255744934,
      "learning_rate": 1.94675934875716e-07,
      "loss": 0.6876,
      "step": 31706
    },
    {
      "epoch": 0.9135884285138016,
      "grad_norm": 0.7406853437423706,
      "learning_rate": 1.94547020669984e-07,
      "loss": 0.6764,
      "step": 31707
    },
    {
      "epoch": 0.9136172419754509,
      "grad_norm": 0.7166715264320374,
      "learning_rate": 1.944181483149432e-07,
      "loss": 0.6799,
      "step": 31708
    },
    {
      "epoch": 0.9136460554371002,
      "grad_norm": 0.7397344708442688,
      "learning_rate": 1.9428931781171722e-07,
      "loss": 0.7143,
      "step": 31709
    },
    {
      "epoch": 0.9136748688987495,
      "grad_norm": 0.800317645072937,
      "learning_rate": 1.9416052916142624e-07,
      "loss": 0.6929,
      "step": 31710
    },
    {
      "epoch": 0.9137036823603988,
      "grad_norm": 0.7174084782600403,
      "learning_rate": 1.940317823651938e-07,
      "loss": 0.6652,
      "step": 31711
    },
    {
      "epoch": 0.9137324958220481,
      "grad_norm": 0.702360987663269,
      "learning_rate": 1.9390307742413904e-07,
      "loss": 0.6691,
      "step": 31712
    },
    {
      "epoch": 0.9137613092836974,
      "grad_norm": 0.714536190032959,
      "learning_rate": 1.9377441433938547e-07,
      "loss": 0.682,
      "step": 31713
    },
    {
      "epoch": 0.9137901227453467,
      "grad_norm": 0.7147461175918579,
      "learning_rate": 1.9364579311205222e-07,
      "loss": 0.6645,
      "step": 31714
    },
    {
      "epoch": 0.913818936206996,
      "grad_norm": 0.7263624668121338,
      "learning_rate": 1.935172137432595e-07,
      "loss": 0.693,
      "step": 31715
    },
    {
      "epoch": 0.9138477496686452,
      "grad_norm": 0.7286831140518188,
      "learning_rate": 1.933886762341264e-07,
      "loss": 0.6809,
      "step": 31716
    },
    {
      "epoch": 0.9138765631302945,
      "grad_norm": 0.7238035202026367,
      "learning_rate": 1.9326018058577423e-07,
      "loss": 0.6901,
      "step": 31717
    },
    {
      "epoch": 0.9139053765919437,
      "grad_norm": 0.7247554659843445,
      "learning_rate": 1.931317267993199e-07,
      "loss": 0.6659,
      "step": 31718
    },
    {
      "epoch": 0.913934190053593,
      "grad_norm": 0.7275213003158569,
      "learning_rate": 1.9300331487588307e-07,
      "loss": 0.6772,
      "step": 31719
    },
    {
      "epoch": 0.9139630035152423,
      "grad_norm": 0.7240639328956604,
      "learning_rate": 1.9287494481658287e-07,
      "loss": 0.6959,
      "step": 31720
    },
    {
      "epoch": 0.9139918169768916,
      "grad_norm": 0.7217124104499817,
      "learning_rate": 1.9274661662253558e-07,
      "loss": 0.6936,
      "step": 31721
    },
    {
      "epoch": 0.9140206304385409,
      "grad_norm": 0.7027405500411987,
      "learning_rate": 1.9261833029486087e-07,
      "loss": 0.6706,
      "step": 31722
    },
    {
      "epoch": 0.9140494439001902,
      "grad_norm": 0.7401775121688843,
      "learning_rate": 1.9249008583467398e-07,
      "loss": 0.6998,
      "step": 31723
    },
    {
      "epoch": 0.9140782573618395,
      "grad_norm": 0.747531533241272,
      "learning_rate": 1.923618832430929e-07,
      "loss": 0.6729,
      "step": 31724
    },
    {
      "epoch": 0.9141070708234887,
      "grad_norm": 0.7039724588394165,
      "learning_rate": 1.9223372252123395e-07,
      "loss": 0.6779,
      "step": 31725
    },
    {
      "epoch": 0.914135884285138,
      "grad_norm": 0.7352926135063171,
      "learning_rate": 1.9210560367021403e-07,
      "loss": 0.6823,
      "step": 31726
    },
    {
      "epoch": 0.9141646977467873,
      "grad_norm": 0.7432374954223633,
      "learning_rate": 1.9197752669114723e-07,
      "loss": 0.7003,
      "step": 31727
    },
    {
      "epoch": 0.9141935112084366,
      "grad_norm": 0.7242279648780823,
      "learning_rate": 1.9184949158515042e-07,
      "loss": 0.6629,
      "step": 31728
    },
    {
      "epoch": 0.9142223246700859,
      "grad_norm": 0.7371302247047424,
      "learning_rate": 1.9172149835333776e-07,
      "loss": 0.6888,
      "step": 31729
    },
    {
      "epoch": 0.9142511381317352,
      "grad_norm": 0.719394862651825,
      "learning_rate": 1.9159354699682496e-07,
      "loss": 0.7086,
      "step": 31730
    },
    {
      "epoch": 0.9142799515933844,
      "grad_norm": 0.7363806366920471,
      "learning_rate": 1.914656375167251e-07,
      "loss": 0.694,
      "step": 31731
    },
    {
      "epoch": 0.9143087650550337,
      "grad_norm": 0.7207635641098022,
      "learning_rate": 1.9133776991415332e-07,
      "loss": 0.6877,
      "step": 31732
    },
    {
      "epoch": 0.914337578516683,
      "grad_norm": 0.7118510603904724,
      "learning_rate": 1.912099441902221e-07,
      "loss": 0.6654,
      "step": 31733
    },
    {
      "epoch": 0.9143663919783322,
      "grad_norm": 0.7022647261619568,
      "learning_rate": 1.9108216034604498e-07,
      "loss": 0.6648,
      "step": 31734
    },
    {
      "epoch": 0.9143952054399815,
      "grad_norm": 0.7133031487464905,
      "learning_rate": 1.9095441838273608e-07,
      "loss": 0.6735,
      "step": 31735
    },
    {
      "epoch": 0.9144240189016308,
      "grad_norm": 0.720974862575531,
      "learning_rate": 1.908267183014062e-07,
      "loss": 0.68,
      "step": 31736
    },
    {
      "epoch": 0.9144528323632801,
      "grad_norm": 0.7158114910125732,
      "learning_rate": 1.9069906010316775e-07,
      "loss": 0.6624,
      "step": 31737
    },
    {
      "epoch": 0.9144816458249294,
      "grad_norm": 0.7143391370773315,
      "learning_rate": 1.9057144378913429e-07,
      "loss": 0.6808,
      "step": 31738
    },
    {
      "epoch": 0.9145104592865787,
      "grad_norm": 0.7288051843643188,
      "learning_rate": 1.904438693604149e-07,
      "loss": 0.6596,
      "step": 31739
    },
    {
      "epoch": 0.914539272748228,
      "grad_norm": 0.7129691243171692,
      "learning_rate": 1.9031633681812156e-07,
      "loss": 0.6677,
      "step": 31740
    },
    {
      "epoch": 0.9145680862098773,
      "grad_norm": 0.7324001789093018,
      "learning_rate": 1.9018884616336553e-07,
      "loss": 0.6839,
      "step": 31741
    },
    {
      "epoch": 0.9145968996715266,
      "grad_norm": 0.7305862903594971,
      "learning_rate": 1.9006139739725592e-07,
      "loss": 0.6733,
      "step": 31742
    },
    {
      "epoch": 0.9146257131331759,
      "grad_norm": 0.7063080072402954,
      "learning_rate": 1.8993399052090467e-07,
      "loss": 0.661,
      "step": 31743
    },
    {
      "epoch": 0.9146545265948252,
      "grad_norm": 0.7297947406768799,
      "learning_rate": 1.8980662553541862e-07,
      "loss": 0.6724,
      "step": 31744
    },
    {
      "epoch": 0.9146833400564743,
      "grad_norm": 0.7236226201057434,
      "learning_rate": 1.8967930244190969e-07,
      "loss": 0.6908,
      "step": 31745
    },
    {
      "epoch": 0.9147121535181236,
      "grad_norm": 0.7130443453788757,
      "learning_rate": 1.895520212414842e-07,
      "loss": 0.6712,
      "step": 31746
    },
    {
      "epoch": 0.9147409669797729,
      "grad_norm": 0.7165719270706177,
      "learning_rate": 1.8942478193525293e-07,
      "loss": 0.6926,
      "step": 31747
    },
    {
      "epoch": 0.9147697804414222,
      "grad_norm": 0.7294341921806335,
      "learning_rate": 1.8929758452432168e-07,
      "loss": 0.7093,
      "step": 31748
    },
    {
      "epoch": 0.9147985939030715,
      "grad_norm": 0.7131947875022888,
      "learning_rate": 1.8917042900980066e-07,
      "loss": 0.6833,
      "step": 31749
    },
    {
      "epoch": 0.9148274073647208,
      "grad_norm": 0.7371309995651245,
      "learning_rate": 1.8904331539279563e-07,
      "loss": 0.7137,
      "step": 31750
    },
    {
      "epoch": 0.9148562208263701,
      "grad_norm": 0.7020023465156555,
      "learning_rate": 1.8891624367441518e-07,
      "loss": 0.6677,
      "step": 31751
    },
    {
      "epoch": 0.9148850342880194,
      "grad_norm": 0.7049620747566223,
      "learning_rate": 1.88789213855764e-07,
      "loss": 0.6479,
      "step": 31752
    },
    {
      "epoch": 0.9149138477496687,
      "grad_norm": 0.7500728368759155,
      "learning_rate": 1.8866222593794946e-07,
      "loss": 0.6866,
      "step": 31753
    },
    {
      "epoch": 0.914942661211318,
      "grad_norm": 0.7033433318138123,
      "learning_rate": 1.8853527992207744e-07,
      "loss": 0.6632,
      "step": 31754
    },
    {
      "epoch": 0.9149714746729672,
      "grad_norm": 0.7117959856987,
      "learning_rate": 1.8840837580925365e-07,
      "loss": 0.6648,
      "step": 31755
    },
    {
      "epoch": 0.9150002881346165,
      "grad_norm": 0.7174088358879089,
      "learning_rate": 1.8828151360058222e-07,
      "loss": 0.6773,
      "step": 31756
    },
    {
      "epoch": 0.9150291015962658,
      "grad_norm": 0.7577311396598816,
      "learning_rate": 1.8815469329716952e-07,
      "loss": 0.6958,
      "step": 31757
    },
    {
      "epoch": 0.915057915057915,
      "grad_norm": 0.7519978880882263,
      "learning_rate": 1.8802791490011907e-07,
      "loss": 0.7208,
      "step": 31758
    },
    {
      "epoch": 0.9150867285195643,
      "grad_norm": 0.7039320468902588,
      "learning_rate": 1.8790117841053555e-07,
      "loss": 0.68,
      "step": 31759
    },
    {
      "epoch": 0.9151155419812136,
      "grad_norm": 0.7263427376747131,
      "learning_rate": 1.8777448382952203e-07,
      "loss": 0.6894,
      "step": 31760
    },
    {
      "epoch": 0.9151443554428629,
      "grad_norm": 0.7228322625160217,
      "learning_rate": 1.8764783115818198e-07,
      "loss": 0.6662,
      "step": 31761
    },
    {
      "epoch": 0.9151731689045122,
      "grad_norm": 0.7277122139930725,
      "learning_rate": 1.8752122039762011e-07,
      "loss": 0.6947,
      "step": 31762
    },
    {
      "epoch": 0.9152019823661615,
      "grad_norm": 0.6960169672966003,
      "learning_rate": 1.8739465154893666e-07,
      "loss": 0.6663,
      "step": 31763
    },
    {
      "epoch": 0.9152307958278108,
      "grad_norm": 0.7447896003723145,
      "learning_rate": 1.8726812461323518e-07,
      "loss": 0.6838,
      "step": 31764
    },
    {
      "epoch": 0.91525960928946,
      "grad_norm": 0.709328293800354,
      "learning_rate": 1.8714163959161758e-07,
      "loss": 0.6752,
      "step": 31765
    },
    {
      "epoch": 0.9152884227511093,
      "grad_norm": 0.7441851496696472,
      "learning_rate": 1.870151964851852e-07,
      "loss": 0.6651,
      "step": 31766
    },
    {
      "epoch": 0.9153172362127586,
      "grad_norm": 0.7168002128601074,
      "learning_rate": 1.8688879529503878e-07,
      "loss": 0.6656,
      "step": 31767
    },
    {
      "epoch": 0.9153460496744079,
      "grad_norm": 0.719778299331665,
      "learning_rate": 1.8676243602228028e-07,
      "loss": 0.6871,
      "step": 31768
    },
    {
      "epoch": 0.9153748631360572,
      "grad_norm": 0.7489109635353088,
      "learning_rate": 1.8663611866800878e-07,
      "loss": 0.6948,
      "step": 31769
    },
    {
      "epoch": 0.9154036765977065,
      "grad_norm": 0.7073710560798645,
      "learning_rate": 1.8650984323332566e-07,
      "loss": 0.6937,
      "step": 31770
    },
    {
      "epoch": 0.9154324900593558,
      "grad_norm": 0.7392420768737793,
      "learning_rate": 1.8638360971932944e-07,
      "loss": 0.6756,
      "step": 31771
    },
    {
      "epoch": 0.915461303521005,
      "grad_norm": 0.7289239168167114,
      "learning_rate": 1.8625741812712038e-07,
      "loss": 0.6918,
      "step": 31772
    },
    {
      "epoch": 0.9154901169826543,
      "grad_norm": 0.744040310382843,
      "learning_rate": 1.8613126845779704e-07,
      "loss": 0.6845,
      "step": 31773
    },
    {
      "epoch": 0.9155189304443035,
      "grad_norm": 0.7096034288406372,
      "learning_rate": 1.8600516071245855e-07,
      "loss": 0.6991,
      "step": 31774
    },
    {
      "epoch": 0.9155477439059528,
      "grad_norm": 0.7270529866218567,
      "learning_rate": 1.858790948922029e-07,
      "loss": 0.6731,
      "step": 31775
    },
    {
      "epoch": 0.9155765573676021,
      "grad_norm": 0.7331799864768982,
      "learning_rate": 1.857530709981281e-07,
      "loss": 0.6968,
      "step": 31776
    },
    {
      "epoch": 0.9156053708292514,
      "grad_norm": 0.7112353444099426,
      "learning_rate": 1.856270890313311e-07,
      "loss": 0.6872,
      "step": 31777
    },
    {
      "epoch": 0.9156341842909007,
      "grad_norm": 0.7131239771842957,
      "learning_rate": 1.8550114899290983e-07,
      "loss": 0.6696,
      "step": 31778
    },
    {
      "epoch": 0.91566299775255,
      "grad_norm": 0.720681369304657,
      "learning_rate": 1.8537525088396124e-07,
      "loss": 0.6645,
      "step": 31779
    },
    {
      "epoch": 0.9156918112141993,
      "grad_norm": 0.734103798866272,
      "learning_rate": 1.8524939470558056e-07,
      "loss": 0.6917,
      "step": 31780
    },
    {
      "epoch": 0.9157206246758486,
      "grad_norm": 0.7402558922767639,
      "learning_rate": 1.8512358045886525e-07,
      "loss": 0.6986,
      "step": 31781
    },
    {
      "epoch": 0.9157494381374979,
      "grad_norm": 0.7283292412757874,
      "learning_rate": 1.8499780814490996e-07,
      "loss": 0.6804,
      "step": 31782
    },
    {
      "epoch": 0.9157782515991472,
      "grad_norm": 0.7282847166061401,
      "learning_rate": 1.848720777648111e-07,
      "loss": 0.6798,
      "step": 31783
    },
    {
      "epoch": 0.9158070650607965,
      "grad_norm": 0.7224398255348206,
      "learning_rate": 1.8474638931966215e-07,
      "loss": 0.6668,
      "step": 31784
    },
    {
      "epoch": 0.9158358785224456,
      "grad_norm": 0.7285637259483337,
      "learning_rate": 1.8462074281055954e-07,
      "loss": 0.7015,
      "step": 31785
    },
    {
      "epoch": 0.9158646919840949,
      "grad_norm": 0.7215398550033569,
      "learning_rate": 1.8449513823859622e-07,
      "loss": 0.6884,
      "step": 31786
    },
    {
      "epoch": 0.9158935054457442,
      "grad_norm": 0.7357692718505859,
      "learning_rate": 1.843695756048669e-07,
      "loss": 0.7052,
      "step": 31787
    },
    {
      "epoch": 0.9159223189073935,
      "grad_norm": 0.7150059342384338,
      "learning_rate": 1.8424405491046404e-07,
      "loss": 0.6801,
      "step": 31788
    },
    {
      "epoch": 0.9159511323690428,
      "grad_norm": 0.7386103868484497,
      "learning_rate": 1.8411857615648233e-07,
      "loss": 0.6675,
      "step": 31789
    },
    {
      "epoch": 0.9159799458306921,
      "grad_norm": 0.725233793258667,
      "learning_rate": 1.839931393440131e-07,
      "loss": 0.6815,
      "step": 31790
    },
    {
      "epoch": 0.9160087592923414,
      "grad_norm": 0.742306649684906,
      "learning_rate": 1.838677444741499e-07,
      "loss": 0.6749,
      "step": 31791
    },
    {
      "epoch": 0.9160375727539907,
      "grad_norm": 0.7186639904975891,
      "learning_rate": 1.8374239154798357e-07,
      "loss": 0.6883,
      "step": 31792
    },
    {
      "epoch": 0.91606638621564,
      "grad_norm": 0.7187979221343994,
      "learning_rate": 1.836170805666071e-07,
      "loss": 0.6897,
      "step": 31793
    },
    {
      "epoch": 0.9160951996772893,
      "grad_norm": 0.7191603779792786,
      "learning_rate": 1.8349181153111074e-07,
      "loss": 0.6677,
      "step": 31794
    },
    {
      "epoch": 0.9161240131389385,
      "grad_norm": 0.7459250688552856,
      "learning_rate": 1.833665844425858e-07,
      "loss": 0.7075,
      "step": 31795
    },
    {
      "epoch": 0.9161528266005878,
      "grad_norm": 0.7128005027770996,
      "learning_rate": 1.832413993021237e-07,
      "loss": 0.6689,
      "step": 31796
    },
    {
      "epoch": 0.9161816400622371,
      "grad_norm": 0.7148571610450745,
      "learning_rate": 1.8311625611081408e-07,
      "loss": 0.6938,
      "step": 31797
    },
    {
      "epoch": 0.9162104535238864,
      "grad_norm": 0.7315261960029602,
      "learning_rate": 1.8299115486974606e-07,
      "loss": 0.6784,
      "step": 31798
    },
    {
      "epoch": 0.9162392669855356,
      "grad_norm": 0.7106136083602905,
      "learning_rate": 1.828660955800099e-07,
      "loss": 0.6532,
      "step": 31799
    },
    {
      "epoch": 0.9162680804471849,
      "grad_norm": 0.7376415729522705,
      "learning_rate": 1.8274107824269527e-07,
      "loss": 0.6872,
      "step": 31800
    },
    {
      "epoch": 0.9162968939088342,
      "grad_norm": 0.7213013768196106,
      "learning_rate": 1.8261610285888963e-07,
      "loss": 0.6876,
      "step": 31801
    },
    {
      "epoch": 0.9163257073704835,
      "grad_norm": 0.7101256847381592,
      "learning_rate": 1.8249116942968326e-07,
      "loss": 0.6767,
      "step": 31802
    },
    {
      "epoch": 0.9163545208321328,
      "grad_norm": 0.7206240296363831,
      "learning_rate": 1.823662779561619e-07,
      "loss": 0.6538,
      "step": 31803
    },
    {
      "epoch": 0.916383334293782,
      "grad_norm": 0.719513475894928,
      "learning_rate": 1.8224142843941527e-07,
      "loss": 0.6857,
      "step": 31804
    },
    {
      "epoch": 0.9164121477554313,
      "grad_norm": 0.7317131757736206,
      "learning_rate": 1.8211662088052917e-07,
      "loss": 0.6834,
      "step": 31805
    },
    {
      "epoch": 0.9164409612170806,
      "grad_norm": 0.7090208530426025,
      "learning_rate": 1.819918552805916e-07,
      "loss": 0.6737,
      "step": 31806
    },
    {
      "epoch": 0.9164697746787299,
      "grad_norm": 0.7028773427009583,
      "learning_rate": 1.8186713164068836e-07,
      "loss": 0.6487,
      "step": 31807
    },
    {
      "epoch": 0.9164985881403792,
      "grad_norm": 0.7201736569404602,
      "learning_rate": 1.817424499619058e-07,
      "loss": 0.6982,
      "step": 31808
    },
    {
      "epoch": 0.9165274016020285,
      "grad_norm": 0.7515879273414612,
      "learning_rate": 1.816178102453303e-07,
      "loss": 0.7003,
      "step": 31809
    },
    {
      "epoch": 0.9165562150636778,
      "grad_norm": 0.7259570360183716,
      "learning_rate": 1.8149321249204766e-07,
      "loss": 0.6696,
      "step": 31810
    },
    {
      "epoch": 0.9165850285253271,
      "grad_norm": 0.7929618954658508,
      "learning_rate": 1.813686567031414e-07,
      "loss": 0.7099,
      "step": 31811
    },
    {
      "epoch": 0.9166138419869763,
      "grad_norm": 0.7205314636230469,
      "learning_rate": 1.8124414287969794e-07,
      "loss": 0.6894,
      "step": 31812
    },
    {
      "epoch": 0.9166426554486256,
      "grad_norm": 0.712317168712616,
      "learning_rate": 1.8111967102280082e-07,
      "loss": 0.6682,
      "step": 31813
    },
    {
      "epoch": 0.9166714689102748,
      "grad_norm": 0.7113694548606873,
      "learning_rate": 1.8099524113353473e-07,
      "loss": 0.6856,
      "step": 31814
    },
    {
      "epoch": 0.9167002823719241,
      "grad_norm": 0.7668746709823608,
      "learning_rate": 1.8087085321298215e-07,
      "loss": 0.7095,
      "step": 31815
    },
    {
      "epoch": 0.9167290958335734,
      "grad_norm": 0.7102553248405457,
      "learning_rate": 1.8074650726222664e-07,
      "loss": 0.6766,
      "step": 31816
    },
    {
      "epoch": 0.9167579092952227,
      "grad_norm": 0.7059326171875,
      "learning_rate": 1.806222032823529e-07,
      "loss": 0.6863,
      "step": 31817
    },
    {
      "epoch": 0.916786722756872,
      "grad_norm": 0.7058245539665222,
      "learning_rate": 1.804979412744412e-07,
      "loss": 0.6893,
      "step": 31818
    },
    {
      "epoch": 0.9168155362185213,
      "grad_norm": 0.7203943133354187,
      "learning_rate": 1.8037372123957453e-07,
      "loss": 0.6736,
      "step": 31819
    },
    {
      "epoch": 0.9168443496801706,
      "grad_norm": 0.7070893049240112,
      "learning_rate": 1.8024954317883536e-07,
      "loss": 0.6853,
      "step": 31820
    },
    {
      "epoch": 0.9168731631418199,
      "grad_norm": 0.729640781879425,
      "learning_rate": 1.8012540709330452e-07,
      "loss": 0.7106,
      "step": 31821
    },
    {
      "epoch": 0.9169019766034692,
      "grad_norm": 0.7290536761283875,
      "learning_rate": 1.800013129840633e-07,
      "loss": 0.6855,
      "step": 31822
    },
    {
      "epoch": 0.9169307900651185,
      "grad_norm": 0.7124696373939514,
      "learning_rate": 1.798772608521926e-07,
      "loss": 0.675,
      "step": 31823
    },
    {
      "epoch": 0.9169596035267678,
      "grad_norm": 0.7164097428321838,
      "learning_rate": 1.7975325069877258e-07,
      "loss": 0.6828,
      "step": 31824
    },
    {
      "epoch": 0.916988416988417,
      "grad_norm": 0.7561217546463013,
      "learning_rate": 1.7962928252488354e-07,
      "loss": 0.7153,
      "step": 31825
    },
    {
      "epoch": 0.9170172304500662,
      "grad_norm": 0.7232717871665955,
      "learning_rate": 1.7950535633160404e-07,
      "loss": 0.6654,
      "step": 31826
    },
    {
      "epoch": 0.9170460439117155,
      "grad_norm": 0.7530947923660278,
      "learning_rate": 1.7938147212001488e-07,
      "loss": 0.6525,
      "step": 31827
    },
    {
      "epoch": 0.9170748573733648,
      "grad_norm": 0.7075875401496887,
      "learning_rate": 1.792576298911941e-07,
      "loss": 0.6668,
      "step": 31828
    },
    {
      "epoch": 0.9171036708350141,
      "grad_norm": 0.7159314751625061,
      "learning_rate": 1.791338296462203e-07,
      "loss": 0.66,
      "step": 31829
    },
    {
      "epoch": 0.9171324842966634,
      "grad_norm": 0.7264305949211121,
      "learning_rate": 1.7901007138617143e-07,
      "loss": 0.6543,
      "step": 31830
    },
    {
      "epoch": 0.9171612977583127,
      "grad_norm": 0.7368336915969849,
      "learning_rate": 1.788863551121256e-07,
      "loss": 0.6916,
      "step": 31831
    },
    {
      "epoch": 0.917190111219962,
      "grad_norm": 0.70856112241745,
      "learning_rate": 1.7876268082516023e-07,
      "loss": 0.6655,
      "step": 31832
    },
    {
      "epoch": 0.9172189246816113,
      "grad_norm": 0.7149574160575867,
      "learning_rate": 1.786390485263534e-07,
      "loss": 0.6689,
      "step": 31833
    },
    {
      "epoch": 0.9172477381432605,
      "grad_norm": 0.7258847951889038,
      "learning_rate": 1.7851545821677974e-07,
      "loss": 0.6653,
      "step": 31834
    },
    {
      "epoch": 0.9172765516049098,
      "grad_norm": 0.7193038463592529,
      "learning_rate": 1.7839190989751732e-07,
      "loss": 0.686,
      "step": 31835
    },
    {
      "epoch": 0.9173053650665591,
      "grad_norm": 0.7291536927223206,
      "learning_rate": 1.7826840356964138e-07,
      "loss": 0.6805,
      "step": 31836
    },
    {
      "epoch": 0.9173341785282084,
      "grad_norm": 0.7411293387413025,
      "learning_rate": 1.7814493923422716e-07,
      "loss": 0.6896,
      "step": 31837
    },
    {
      "epoch": 0.9173629919898577,
      "grad_norm": 0.7039452791213989,
      "learning_rate": 1.780215168923516e-07,
      "loss": 0.6637,
      "step": 31838
    },
    {
      "epoch": 0.9173918054515069,
      "grad_norm": 0.7454544901847839,
      "learning_rate": 1.7789813654508768e-07,
      "loss": 0.669,
      "step": 31839
    },
    {
      "epoch": 0.9174206189131562,
      "grad_norm": 0.721138596534729,
      "learning_rate": 1.7777479819351074e-07,
      "loss": 0.7078,
      "step": 31840
    },
    {
      "epoch": 0.9174494323748055,
      "grad_norm": 0.7212180495262146,
      "learning_rate": 1.7765150183869483e-07,
      "loss": 0.6658,
      "step": 31841
    },
    {
      "epoch": 0.9174782458364548,
      "grad_norm": 0.7165151834487915,
      "learning_rate": 1.7752824748171414e-07,
      "loss": 0.6598,
      "step": 31842
    },
    {
      "epoch": 0.917507059298104,
      "grad_norm": 0.7146291136741638,
      "learning_rate": 1.7740503512364004e-07,
      "loss": 0.6711,
      "step": 31843
    },
    {
      "epoch": 0.9175358727597533,
      "grad_norm": 0.7009353637695312,
      "learning_rate": 1.7728186476554943e-07,
      "loss": 0.6742,
      "step": 31844
    },
    {
      "epoch": 0.9175646862214026,
      "grad_norm": 0.7368918061256409,
      "learning_rate": 1.7715873640851144e-07,
      "loss": 0.6627,
      "step": 31845
    },
    {
      "epoch": 0.9175934996830519,
      "grad_norm": 0.7230787873268127,
      "learning_rate": 1.770356500536008e-07,
      "loss": 0.6844,
      "step": 31846
    },
    {
      "epoch": 0.9176223131447012,
      "grad_norm": 0.7119201421737671,
      "learning_rate": 1.7691260570188772e-07,
      "loss": 0.6586,
      "step": 31847
    },
    {
      "epoch": 0.9176511266063505,
      "grad_norm": 0.7361321449279785,
      "learning_rate": 1.7678960335444472e-07,
      "loss": 0.6933,
      "step": 31848
    },
    {
      "epoch": 0.9176799400679998,
      "grad_norm": 0.7259811758995056,
      "learning_rate": 1.766666430123426e-07,
      "loss": 0.6902,
      "step": 31849
    },
    {
      "epoch": 0.9177087535296491,
      "grad_norm": 0.7112424969673157,
      "learning_rate": 1.7654372467665325e-07,
      "loss": 0.6609,
      "step": 31850
    },
    {
      "epoch": 0.9177375669912984,
      "grad_norm": 0.7467532753944397,
      "learning_rate": 1.764208483484453e-07,
      "loss": 0.6846,
      "step": 31851
    },
    {
      "epoch": 0.9177663804529477,
      "grad_norm": 0.7151806950569153,
      "learning_rate": 1.7629801402879066e-07,
      "loss": 0.6621,
      "step": 31852
    },
    {
      "epoch": 0.9177951939145969,
      "grad_norm": 0.7000396251678467,
      "learning_rate": 1.7617522171875845e-07,
      "loss": 0.651,
      "step": 31853
    },
    {
      "epoch": 0.9178240073762461,
      "grad_norm": 0.7279137969017029,
      "learning_rate": 1.760524714194173e-07,
      "loss": 0.6772,
      "step": 31854
    },
    {
      "epoch": 0.9178528208378954,
      "grad_norm": 0.6967889666557312,
      "learning_rate": 1.7592976313183684e-07,
      "loss": 0.677,
      "step": 31855
    },
    {
      "epoch": 0.9178816342995447,
      "grad_norm": 0.7134008407592773,
      "learning_rate": 1.7580709685708687e-07,
      "loss": 0.6541,
      "step": 31856
    },
    {
      "epoch": 0.917910447761194,
      "grad_norm": 0.7233237028121948,
      "learning_rate": 1.7568447259623366e-07,
      "loss": 0.69,
      "step": 31857
    },
    {
      "epoch": 0.9179392612228433,
      "grad_norm": 0.7124274969100952,
      "learning_rate": 1.7556189035034642e-07,
      "loss": 0.6937,
      "step": 31858
    },
    {
      "epoch": 0.9179680746844926,
      "grad_norm": 0.732937216758728,
      "learning_rate": 1.7543935012049318e-07,
      "loss": 0.6775,
      "step": 31859
    },
    {
      "epoch": 0.9179968881461419,
      "grad_norm": 0.7119130492210388,
      "learning_rate": 1.7531685190773972e-07,
      "loss": 0.6844,
      "step": 31860
    },
    {
      "epoch": 0.9180257016077912,
      "grad_norm": 0.6936070322990417,
      "learning_rate": 1.751943957131541e-07,
      "loss": 0.6915,
      "step": 31861
    },
    {
      "epoch": 0.9180545150694405,
      "grad_norm": 0.7170293927192688,
      "learning_rate": 1.7507198153780158e-07,
      "loss": 0.6896,
      "step": 31862
    },
    {
      "epoch": 0.9180833285310898,
      "grad_norm": 0.7408912181854248,
      "learning_rate": 1.7494960938274962e-07,
      "loss": 0.6831,
      "step": 31863
    },
    {
      "epoch": 0.918112141992739,
      "grad_norm": 0.7401558756828308,
      "learning_rate": 1.748272792490624e-07,
      "loss": 0.6812,
      "step": 31864
    },
    {
      "epoch": 0.9181409554543883,
      "grad_norm": 0.7016807198524475,
      "learning_rate": 1.7470499113780736e-07,
      "loss": 0.6636,
      "step": 31865
    },
    {
      "epoch": 0.9181697689160375,
      "grad_norm": 0.7193365693092346,
      "learning_rate": 1.7458274505004702e-07,
      "loss": 0.678,
      "step": 31866
    },
    {
      "epoch": 0.9181985823776868,
      "grad_norm": 0.7136415243148804,
      "learning_rate": 1.7446054098684771e-07,
      "loss": 0.655,
      "step": 31867
    },
    {
      "epoch": 0.9182273958393361,
      "grad_norm": 0.7264600396156311,
      "learning_rate": 1.7433837894927364e-07,
      "loss": 0.6908,
      "step": 31868
    },
    {
      "epoch": 0.9182562093009854,
      "grad_norm": 0.7438633441925049,
      "learning_rate": 1.7421625893838835e-07,
      "loss": 0.7206,
      "step": 31869
    },
    {
      "epoch": 0.9182850227626347,
      "grad_norm": 0.7545730471611023,
      "learning_rate": 1.7409418095525542e-07,
      "loss": 0.6736,
      "step": 31870
    },
    {
      "epoch": 0.918313836224284,
      "grad_norm": 0.7081506252288818,
      "learning_rate": 1.7397214500093794e-07,
      "loss": 0.6906,
      "step": 31871
    },
    {
      "epoch": 0.9183426496859333,
      "grad_norm": 0.7148998379707336,
      "learning_rate": 1.7385015107649894e-07,
      "loss": 0.6603,
      "step": 31872
    },
    {
      "epoch": 0.9183714631475826,
      "grad_norm": 0.7229564189910889,
      "learning_rate": 1.7372819918300087e-07,
      "loss": 0.6649,
      "step": 31873
    },
    {
      "epoch": 0.9184002766092318,
      "grad_norm": 0.7043613195419312,
      "learning_rate": 1.7360628932150514e-07,
      "loss": 0.6592,
      "step": 31874
    },
    {
      "epoch": 0.9184290900708811,
      "grad_norm": 0.7297636866569519,
      "learning_rate": 1.734844214930742e-07,
      "loss": 0.6908,
      "step": 31875
    },
    {
      "epoch": 0.9184579035325304,
      "grad_norm": 0.7178249359130859,
      "learning_rate": 1.7336259569876947e-07,
      "loss": 0.6691,
      "step": 31876
    },
    {
      "epoch": 0.9184867169941797,
      "grad_norm": 0.7589846253395081,
      "learning_rate": 1.7324081193965114e-07,
      "loss": 0.6989,
      "step": 31877
    },
    {
      "epoch": 0.918515530455829,
      "grad_norm": 0.7233493328094482,
      "learning_rate": 1.7311907021678121e-07,
      "loss": 0.6643,
      "step": 31878
    },
    {
      "epoch": 0.9185443439174783,
      "grad_norm": 0.7157090902328491,
      "learning_rate": 1.7299737053121768e-07,
      "loss": 0.6772,
      "step": 31879
    },
    {
      "epoch": 0.9185731573791275,
      "grad_norm": 0.735048770904541,
      "learning_rate": 1.7287571288402304e-07,
      "loss": 0.6807,
      "step": 31880
    },
    {
      "epoch": 0.9186019708407768,
      "grad_norm": 0.7168009281158447,
      "learning_rate": 1.7275409727625481e-07,
      "loss": 0.6708,
      "step": 31881
    },
    {
      "epoch": 0.9186307843024261,
      "grad_norm": 0.720000147819519,
      "learning_rate": 1.7263252370897377e-07,
      "loss": 0.6755,
      "step": 31882
    },
    {
      "epoch": 0.9186595977640754,
      "grad_norm": 0.7277870774269104,
      "learning_rate": 1.7251099218323742e-07,
      "loss": 0.6992,
      "step": 31883
    },
    {
      "epoch": 0.9186884112257246,
      "grad_norm": 0.7089616060256958,
      "learning_rate": 1.723895027001049e-07,
      "loss": 0.6714,
      "step": 31884
    },
    {
      "epoch": 0.9187172246873739,
      "grad_norm": 0.7441685199737549,
      "learning_rate": 1.7226805526063318e-07,
      "loss": 0.6888,
      "step": 31885
    },
    {
      "epoch": 0.9187460381490232,
      "grad_norm": 0.716917097568512,
      "learning_rate": 1.7214664986588193e-07,
      "loss": 0.6915,
      "step": 31886
    },
    {
      "epoch": 0.9187748516106725,
      "grad_norm": 0.7230464816093445,
      "learning_rate": 1.7202528651690586e-07,
      "loss": 0.7052,
      "step": 31887
    },
    {
      "epoch": 0.9188036650723218,
      "grad_norm": 0.7313682436943054,
      "learning_rate": 1.719039652147647e-07,
      "loss": 0.6711,
      "step": 31888
    },
    {
      "epoch": 0.9188324785339711,
      "grad_norm": 0.7185753583908081,
      "learning_rate": 1.7178268596051262e-07,
      "loss": 0.69,
      "step": 31889
    },
    {
      "epoch": 0.9188612919956204,
      "grad_norm": 0.7220107913017273,
      "learning_rate": 1.7166144875520762e-07,
      "loss": 0.6743,
      "step": 31890
    },
    {
      "epoch": 0.9188901054572697,
      "grad_norm": 0.7201496958732605,
      "learning_rate": 1.715402535999039e-07,
      "loss": 0.7009,
      "step": 31891
    },
    {
      "epoch": 0.918918918918919,
      "grad_norm": 0.723504364490509,
      "learning_rate": 1.714191004956589e-07,
      "loss": 0.6624,
      "step": 31892
    },
    {
      "epoch": 0.9189477323805681,
      "grad_norm": 0.7252012491226196,
      "learning_rate": 1.712979894435257e-07,
      "loss": 0.6913,
      "step": 31893
    },
    {
      "epoch": 0.9189765458422174,
      "grad_norm": 0.7073095440864563,
      "learning_rate": 1.7117692044456124e-07,
      "loss": 0.7065,
      "step": 31894
    },
    {
      "epoch": 0.9190053593038667,
      "grad_norm": 0.7394603490829468,
      "learning_rate": 1.71055893499818e-07,
      "loss": 0.6762,
      "step": 31895
    },
    {
      "epoch": 0.919034172765516,
      "grad_norm": 0.7311581969261169,
      "learning_rate": 1.7093490861035066e-07,
      "loss": 0.6742,
      "step": 31896
    },
    {
      "epoch": 0.9190629862271653,
      "grad_norm": 0.737846314907074,
      "learning_rate": 1.7081396577721344e-07,
      "loss": 0.6903,
      "step": 31897
    },
    {
      "epoch": 0.9190917996888146,
      "grad_norm": 0.7299284338951111,
      "learning_rate": 1.7069306500145878e-07,
      "loss": 0.6785,
      "step": 31898
    },
    {
      "epoch": 0.9191206131504639,
      "grad_norm": 0.7215402722358704,
      "learning_rate": 1.7057220628414028e-07,
      "loss": 0.6644,
      "step": 31899
    },
    {
      "epoch": 0.9191494266121132,
      "grad_norm": 0.7344465255737305,
      "learning_rate": 1.704513896263099e-07,
      "loss": 0.6826,
      "step": 31900
    },
    {
      "epoch": 0.9191782400737625,
      "grad_norm": 0.7176703214645386,
      "learning_rate": 1.703306150290207e-07,
      "loss": 0.6793,
      "step": 31901
    },
    {
      "epoch": 0.9192070535354118,
      "grad_norm": 0.7196036577224731,
      "learning_rate": 1.7020988249332349e-07,
      "loss": 0.6708,
      "step": 31902
    },
    {
      "epoch": 0.9192358669970611,
      "grad_norm": 0.7102698087692261,
      "learning_rate": 1.7008919202027018e-07,
      "loss": 0.6585,
      "step": 31903
    },
    {
      "epoch": 0.9192646804587103,
      "grad_norm": 0.7289522886276245,
      "learning_rate": 1.6996854361091165e-07,
      "loss": 0.6698,
      "step": 31904
    },
    {
      "epoch": 0.9192934939203596,
      "grad_norm": 0.7501540184020996,
      "learning_rate": 1.6984793726629923e-07,
      "loss": 0.7093,
      "step": 31905
    },
    {
      "epoch": 0.9193223073820089,
      "grad_norm": 0.7289034724235535,
      "learning_rate": 1.6972737298748265e-07,
      "loss": 0.6825,
      "step": 31906
    },
    {
      "epoch": 0.9193511208436581,
      "grad_norm": 0.7138673663139343,
      "learning_rate": 1.6960685077551274e-07,
      "loss": 0.6935,
      "step": 31907
    },
    {
      "epoch": 0.9193799343053074,
      "grad_norm": 0.6873920559883118,
      "learning_rate": 1.6948637063143758e-07,
      "loss": 0.6834,
      "step": 31908
    },
    {
      "epoch": 0.9194087477669567,
      "grad_norm": 0.7230611443519592,
      "learning_rate": 1.693659325563085e-07,
      "loss": 0.6738,
      "step": 31909
    },
    {
      "epoch": 0.919437561228606,
      "grad_norm": 0.7167240977287292,
      "learning_rate": 1.6924553655117248e-07,
      "loss": 0.6951,
      "step": 31910
    },
    {
      "epoch": 0.9194663746902553,
      "grad_norm": 0.7341529726982117,
      "learning_rate": 1.6912518261707922e-07,
      "loss": 0.6475,
      "step": 31911
    },
    {
      "epoch": 0.9194951881519046,
      "grad_norm": 0.7313483953475952,
      "learning_rate": 1.6900487075507622e-07,
      "loss": 0.693,
      "step": 31912
    },
    {
      "epoch": 0.9195240016135539,
      "grad_norm": 0.7297524809837341,
      "learning_rate": 1.6888460096621151e-07,
      "loss": 0.6853,
      "step": 31913
    },
    {
      "epoch": 0.9195528150752031,
      "grad_norm": 0.7170259952545166,
      "learning_rate": 1.6876437325153262e-07,
      "loss": 0.6548,
      "step": 31914
    },
    {
      "epoch": 0.9195816285368524,
      "grad_norm": 0.6916748881340027,
      "learning_rate": 1.6864418761208644e-07,
      "loss": 0.6757,
      "step": 31915
    },
    {
      "epoch": 0.9196104419985017,
      "grad_norm": 0.745351254940033,
      "learning_rate": 1.6852404404891997e-07,
      "loss": 0.6784,
      "step": 31916
    },
    {
      "epoch": 0.919639255460151,
      "grad_norm": 0.7425205707550049,
      "learning_rate": 1.6840394256307958e-07,
      "loss": 0.7014,
      "step": 31917
    },
    {
      "epoch": 0.9196680689218003,
      "grad_norm": 0.7359927296638489,
      "learning_rate": 1.6828388315561107e-07,
      "loss": 0.6992,
      "step": 31918
    },
    {
      "epoch": 0.9196968823834496,
      "grad_norm": 0.7395313382148743,
      "learning_rate": 1.6816386582755972e-07,
      "loss": 0.6755,
      "step": 31919
    },
    {
      "epoch": 0.9197256958450988,
      "grad_norm": 0.7099166512489319,
      "learning_rate": 1.680438905799714e-07,
      "loss": 0.6537,
      "step": 31920
    },
    {
      "epoch": 0.9197545093067481,
      "grad_norm": 0.7245562076568604,
      "learning_rate": 1.6792395741388967e-07,
      "loss": 0.672,
      "step": 31921
    },
    {
      "epoch": 0.9197833227683974,
      "grad_norm": 0.7207152843475342,
      "learning_rate": 1.6780406633036094e-07,
      "loss": 0.6732,
      "step": 31922
    },
    {
      "epoch": 0.9198121362300467,
      "grad_norm": 0.7286526560783386,
      "learning_rate": 1.6768421733042827e-07,
      "loss": 0.7017,
      "step": 31923
    },
    {
      "epoch": 0.9198409496916959,
      "grad_norm": 0.7353089451789856,
      "learning_rate": 1.6756441041513528e-07,
      "loss": 0.695,
      "step": 31924
    },
    {
      "epoch": 0.9198697631533452,
      "grad_norm": 0.7492583394050598,
      "learning_rate": 1.6744464558552554e-07,
      "loss": 0.7106,
      "step": 31925
    },
    {
      "epoch": 0.9198985766149945,
      "grad_norm": 0.7357643842697144,
      "learning_rate": 1.6732492284264268e-07,
      "loss": 0.6841,
      "step": 31926
    },
    {
      "epoch": 0.9199273900766438,
      "grad_norm": 0.7279897928237915,
      "learning_rate": 1.6720524218752753e-07,
      "loss": 0.6838,
      "step": 31927
    },
    {
      "epoch": 0.9199562035382931,
      "grad_norm": 0.7313507199287415,
      "learning_rate": 1.6708560362122482e-07,
      "loss": 0.6686,
      "step": 31928
    },
    {
      "epoch": 0.9199850169999424,
      "grad_norm": 0.7289720773696899,
      "learning_rate": 1.6696600714477483e-07,
      "loss": 0.6943,
      "step": 31929
    },
    {
      "epoch": 0.9200138304615917,
      "grad_norm": 0.6921334266662598,
      "learning_rate": 1.6684645275922007e-07,
      "loss": 0.6738,
      "step": 31930
    },
    {
      "epoch": 0.920042643923241,
      "grad_norm": 0.721332848072052,
      "learning_rate": 1.6672694046560133e-07,
      "loss": 0.655,
      "step": 31931
    },
    {
      "epoch": 0.9200714573848903,
      "grad_norm": 0.7140746712684631,
      "learning_rate": 1.666074702649595e-07,
      "loss": 0.6747,
      "step": 31932
    },
    {
      "epoch": 0.9201002708465396,
      "grad_norm": 0.7216583490371704,
      "learning_rate": 1.6648804215833425e-07,
      "loss": 0.6968,
      "step": 31933
    },
    {
      "epoch": 0.9201290843081887,
      "grad_norm": 0.7289310693740845,
      "learning_rate": 1.66368656146767e-07,
      "loss": 0.6974,
      "step": 31934
    },
    {
      "epoch": 0.920157897769838,
      "grad_norm": 0.7297729253768921,
      "learning_rate": 1.6624931223129692e-07,
      "loss": 0.6961,
      "step": 31935
    },
    {
      "epoch": 0.9201867112314873,
      "grad_norm": 0.7224428057670593,
      "learning_rate": 1.6613001041296316e-07,
      "loss": 0.684,
      "step": 31936
    },
    {
      "epoch": 0.9202155246931366,
      "grad_norm": 0.7084766626358032,
      "learning_rate": 1.6601075069280493e-07,
      "loss": 0.6595,
      "step": 31937
    },
    {
      "epoch": 0.9202443381547859,
      "grad_norm": 0.7353036999702454,
      "learning_rate": 1.6589153307186078e-07,
      "loss": 0.6571,
      "step": 31938
    },
    {
      "epoch": 0.9202731516164352,
      "grad_norm": 0.7162424921989441,
      "learning_rate": 1.6577235755116882e-07,
      "loss": 0.6659,
      "step": 31939
    },
    {
      "epoch": 0.9203019650780845,
      "grad_norm": 0.7155792713165283,
      "learning_rate": 1.6565322413176765e-07,
      "loss": 0.6853,
      "step": 31940
    },
    {
      "epoch": 0.9203307785397338,
      "grad_norm": 0.7368266582489014,
      "learning_rate": 1.6553413281469422e-07,
      "loss": 0.6784,
      "step": 31941
    },
    {
      "epoch": 0.9203595920013831,
      "grad_norm": 0.7348142266273499,
      "learning_rate": 1.6541508360098545e-07,
      "loss": 0.6947,
      "step": 31942
    },
    {
      "epoch": 0.9203884054630324,
      "grad_norm": 0.7395752668380737,
      "learning_rate": 1.6529607649167945e-07,
      "loss": 0.6805,
      "step": 31943
    },
    {
      "epoch": 0.9204172189246816,
      "grad_norm": 0.7244359254837036,
      "learning_rate": 1.6517711148781035e-07,
      "loss": 0.6741,
      "step": 31944
    },
    {
      "epoch": 0.9204460323863309,
      "grad_norm": 0.7191879749298096,
      "learning_rate": 1.650581885904168e-07,
      "loss": 0.6809,
      "step": 31945
    },
    {
      "epoch": 0.9204748458479802,
      "grad_norm": 0.7375504374504089,
      "learning_rate": 1.6493930780053237e-07,
      "loss": 0.6708,
      "step": 31946
    },
    {
      "epoch": 0.9205036593096294,
      "grad_norm": 0.7148388028144836,
      "learning_rate": 1.6482046911919402e-07,
      "loss": 0.6898,
      "step": 31947
    },
    {
      "epoch": 0.9205324727712787,
      "grad_norm": 0.727471649646759,
      "learning_rate": 1.647016725474354e-07,
      "loss": 0.665,
      "step": 31948
    },
    {
      "epoch": 0.920561286232928,
      "grad_norm": 0.7316439151763916,
      "learning_rate": 1.6458291808629178e-07,
      "loss": 0.6798,
      "step": 31949
    },
    {
      "epoch": 0.9205900996945773,
      "grad_norm": 0.7119190692901611,
      "learning_rate": 1.6446420573679734e-07,
      "loss": 0.7103,
      "step": 31950
    },
    {
      "epoch": 0.9206189131562266,
      "grad_norm": 0.7176489233970642,
      "learning_rate": 1.6434553549998567e-07,
      "loss": 0.6868,
      "step": 31951
    },
    {
      "epoch": 0.9206477266178759,
      "grad_norm": 0.7171012163162231,
      "learning_rate": 1.642269073768904e-07,
      "loss": 0.6812,
      "step": 31952
    },
    {
      "epoch": 0.9206765400795252,
      "grad_norm": 0.7169389724731445,
      "learning_rate": 1.6410832136854515e-07,
      "loss": 0.6742,
      "step": 31953
    },
    {
      "epoch": 0.9207053535411744,
      "grad_norm": 0.7258777022361755,
      "learning_rate": 1.6398977747598245e-07,
      "loss": 0.6882,
      "step": 31954
    },
    {
      "epoch": 0.9207341670028237,
      "grad_norm": 0.7162753343582153,
      "learning_rate": 1.6387127570023421e-07,
      "loss": 0.6778,
      "step": 31955
    },
    {
      "epoch": 0.920762980464473,
      "grad_norm": 0.7220702171325684,
      "learning_rate": 1.6375281604233295e-07,
      "loss": 0.6743,
      "step": 31956
    },
    {
      "epoch": 0.9207917939261223,
      "grad_norm": 0.7209869623184204,
      "learning_rate": 1.6363439850331008e-07,
      "loss": 0.6889,
      "step": 31957
    },
    {
      "epoch": 0.9208206073877716,
      "grad_norm": 0.7191598415374756,
      "learning_rate": 1.63516023084197e-07,
      "loss": 0.6834,
      "step": 31958
    },
    {
      "epoch": 0.9208494208494209,
      "grad_norm": 0.7343031167984009,
      "learning_rate": 1.6339768978602455e-07,
      "loss": 0.6771,
      "step": 31959
    },
    {
      "epoch": 0.9208782343110702,
      "grad_norm": 0.7498605847358704,
      "learning_rate": 1.632793986098241e-07,
      "loss": 0.6682,
      "step": 31960
    },
    {
      "epoch": 0.9209070477727194,
      "grad_norm": 0.7197059988975525,
      "learning_rate": 1.631611495566243e-07,
      "loss": 0.6913,
      "step": 31961
    },
    {
      "epoch": 0.9209358612343687,
      "grad_norm": 0.7183993458747864,
      "learning_rate": 1.6304294262745656e-07,
      "loss": 0.6815,
      "step": 31962
    },
    {
      "epoch": 0.920964674696018,
      "grad_norm": 0.7164540886878967,
      "learning_rate": 1.6292477782334837e-07,
      "loss": 0.6858,
      "step": 31963
    },
    {
      "epoch": 0.9209934881576672,
      "grad_norm": 0.7147185206413269,
      "learning_rate": 1.628066551453311e-07,
      "loss": 0.6761,
      "step": 31964
    },
    {
      "epoch": 0.9210223016193165,
      "grad_norm": 0.6928073167800903,
      "learning_rate": 1.6268857459443176e-07,
      "loss": 0.6536,
      "step": 31965
    },
    {
      "epoch": 0.9210511150809658,
      "grad_norm": 0.7033431529998779,
      "learning_rate": 1.6257053617168063e-07,
      "loss": 0.6557,
      "step": 31966
    },
    {
      "epoch": 0.9210799285426151,
      "grad_norm": 0.7301777601242065,
      "learning_rate": 1.6245253987810294e-07,
      "loss": 0.6988,
      "step": 31967
    },
    {
      "epoch": 0.9211087420042644,
      "grad_norm": 0.7123041749000549,
      "learning_rate": 1.6233458571472906e-07,
      "loss": 0.6711,
      "step": 31968
    },
    {
      "epoch": 0.9211375554659137,
      "grad_norm": 0.7320629954338074,
      "learning_rate": 1.622166736825842e-07,
      "loss": 0.6961,
      "step": 31969
    },
    {
      "epoch": 0.921166368927563,
      "grad_norm": 0.7217870354652405,
      "learning_rate": 1.6209880378269704e-07,
      "loss": 0.6934,
      "step": 31970
    },
    {
      "epoch": 0.9211951823892123,
      "grad_norm": 0.7372502088546753,
      "learning_rate": 1.619809760160923e-07,
      "loss": 0.6727,
      "step": 31971
    },
    {
      "epoch": 0.9212239958508616,
      "grad_norm": 0.7471171617507935,
      "learning_rate": 1.6186319038379695e-07,
      "loss": 0.6907,
      "step": 31972
    },
    {
      "epoch": 0.9212528093125109,
      "grad_norm": 0.7100220322608948,
      "learning_rate": 1.617454468868379e-07,
      "loss": 0.6631,
      "step": 31973
    },
    {
      "epoch": 0.92128162277416,
      "grad_norm": 0.7237789034843445,
      "learning_rate": 1.6162774552623827e-07,
      "loss": 0.6596,
      "step": 31974
    },
    {
      "epoch": 0.9213104362358093,
      "grad_norm": 0.7329702377319336,
      "learning_rate": 1.61510086303025e-07,
      "loss": 0.6914,
      "step": 31975
    },
    {
      "epoch": 0.9213392496974586,
      "grad_norm": 0.7182474136352539,
      "learning_rate": 1.613924692182217e-07,
      "loss": 0.6577,
      "step": 31976
    },
    {
      "epoch": 0.9213680631591079,
      "grad_norm": 0.7299704551696777,
      "learning_rate": 1.612748942728537e-07,
      "loss": 0.6721,
      "step": 31977
    },
    {
      "epoch": 0.9213968766207572,
      "grad_norm": 0.725718080997467,
      "learning_rate": 1.6115736146794402e-07,
      "loss": 0.6582,
      "step": 31978
    },
    {
      "epoch": 0.9214256900824065,
      "grad_norm": 0.7516101002693176,
      "learning_rate": 1.6103987080451745e-07,
      "loss": 0.6522,
      "step": 31979
    },
    {
      "epoch": 0.9214545035440558,
      "grad_norm": 0.7201041579246521,
      "learning_rate": 1.6092242228359532e-07,
      "loss": 0.6971,
      "step": 31980
    },
    {
      "epoch": 0.9214833170057051,
      "grad_norm": 0.7019896507263184,
      "learning_rate": 1.6080501590620245e-07,
      "loss": 0.6595,
      "step": 31981
    },
    {
      "epoch": 0.9215121304673544,
      "grad_norm": 0.7453610301017761,
      "learning_rate": 1.606876516733602e-07,
      "loss": 0.6959,
      "step": 31982
    },
    {
      "epoch": 0.9215409439290037,
      "grad_norm": 0.7196817398071289,
      "learning_rate": 1.605703295860911e-07,
      "loss": 0.6734,
      "step": 31983
    },
    {
      "epoch": 0.9215697573906529,
      "grad_norm": 0.7392651438713074,
      "learning_rate": 1.6045304964541653e-07,
      "loss": 0.6854,
      "step": 31984
    },
    {
      "epoch": 0.9215985708523022,
      "grad_norm": 0.7236155867576599,
      "learning_rate": 1.6033581185235847e-07,
      "loss": 0.6603,
      "step": 31985
    },
    {
      "epoch": 0.9216273843139515,
      "grad_norm": 0.722159206867218,
      "learning_rate": 1.6021861620793666e-07,
      "loss": 0.6815,
      "step": 31986
    },
    {
      "epoch": 0.9216561977756008,
      "grad_norm": 0.7008645534515381,
      "learning_rate": 1.6010146271317307e-07,
      "loss": 0.6619,
      "step": 31987
    },
    {
      "epoch": 0.92168501123725,
      "grad_norm": 0.7027761340141296,
      "learning_rate": 1.5998435136908798e-07,
      "loss": 0.6542,
      "step": 31988
    },
    {
      "epoch": 0.9217138246988993,
      "grad_norm": 0.7252262234687805,
      "learning_rate": 1.598672821767011e-07,
      "loss": 0.7121,
      "step": 31989
    },
    {
      "epoch": 0.9217426381605486,
      "grad_norm": 0.7023928761482239,
      "learning_rate": 1.5975025513703112e-07,
      "loss": 0.6833,
      "step": 31990
    },
    {
      "epoch": 0.9217714516221979,
      "grad_norm": 0.7212831974029541,
      "learning_rate": 1.5963327025109888e-07,
      "loss": 0.6856,
      "step": 31991
    },
    {
      "epoch": 0.9218002650838472,
      "grad_norm": 0.7173635363578796,
      "learning_rate": 1.595163275199213e-07,
      "loss": 0.6713,
      "step": 31992
    },
    {
      "epoch": 0.9218290785454964,
      "grad_norm": 0.6947014927864075,
      "learning_rate": 1.5939942694451815e-07,
      "loss": 0.6595,
      "step": 31993
    },
    {
      "epoch": 0.9218578920071457,
      "grad_norm": 0.7335599064826965,
      "learning_rate": 1.5928256852590752e-07,
      "loss": 0.6742,
      "step": 31994
    },
    {
      "epoch": 0.921886705468795,
      "grad_norm": 0.749376654624939,
      "learning_rate": 1.5916575226510634e-07,
      "loss": 0.6464,
      "step": 31995
    },
    {
      "epoch": 0.9219155189304443,
      "grad_norm": 0.7288596034049988,
      "learning_rate": 1.590489781631327e-07,
      "loss": 0.6803,
      "step": 31996
    },
    {
      "epoch": 0.9219443323920936,
      "grad_norm": 0.7356389760971069,
      "learning_rate": 1.58932246221003e-07,
      "loss": 0.6977,
      "step": 31997
    },
    {
      "epoch": 0.9219731458537429,
      "grad_norm": 0.7289606928825378,
      "learning_rate": 1.588155564397348e-07,
      "loss": 0.6781,
      "step": 31998
    },
    {
      "epoch": 0.9220019593153922,
      "grad_norm": 0.706649899482727,
      "learning_rate": 1.5869890882034222e-07,
      "loss": 0.6525,
      "step": 31999
    },
    {
      "epoch": 0.9220307727770415,
      "grad_norm": 0.7274283170700073,
      "learning_rate": 1.5858230336384395e-07,
      "loss": 0.6852,
      "step": 32000
    },
    {
      "epoch": 0.9220595862386907,
      "grad_norm": 0.7231233716011047,
      "learning_rate": 1.584657400712536e-07,
      "loss": 0.6768,
      "step": 32001
    },
    {
      "epoch": 0.92208839970034,
      "grad_norm": 0.6980718970298767,
      "learning_rate": 1.58349218943587e-07,
      "loss": 0.6695,
      "step": 32002
    },
    {
      "epoch": 0.9221172131619892,
      "grad_norm": 0.7383769750595093,
      "learning_rate": 1.5823273998185896e-07,
      "loss": 0.7037,
      "step": 32003
    },
    {
      "epoch": 0.9221460266236385,
      "grad_norm": 0.6987137794494629,
      "learning_rate": 1.5811630318708417e-07,
      "loss": 0.687,
      "step": 32004
    },
    {
      "epoch": 0.9221748400852878,
      "grad_norm": 0.7196934819221497,
      "learning_rate": 1.5799990856027513e-07,
      "loss": 0.7077,
      "step": 32005
    },
    {
      "epoch": 0.9222036535469371,
      "grad_norm": 0.7186816930770874,
      "learning_rate": 1.5788355610244778e-07,
      "loss": 0.6745,
      "step": 32006
    },
    {
      "epoch": 0.9222324670085864,
      "grad_norm": 0.7379976511001587,
      "learning_rate": 1.57767245814614e-07,
      "loss": 0.6762,
      "step": 32007
    },
    {
      "epoch": 0.9222612804702357,
      "grad_norm": 0.7302510142326355,
      "learning_rate": 1.5765097769778692e-07,
      "loss": 0.6883,
      "step": 32008
    },
    {
      "epoch": 0.922290093931885,
      "grad_norm": 0.7110874056816101,
      "learning_rate": 1.5753475175297905e-07,
      "loss": 0.652,
      "step": 32009
    },
    {
      "epoch": 0.9223189073935343,
      "grad_norm": 0.7060189843177795,
      "learning_rate": 1.5741856798120292e-07,
      "loss": 0.6729,
      "step": 32010
    },
    {
      "epoch": 0.9223477208551836,
      "grad_norm": 0.7460392117500305,
      "learning_rate": 1.5730242638346992e-07,
      "loss": 0.7167,
      "step": 32011
    },
    {
      "epoch": 0.9223765343168329,
      "grad_norm": 0.6981826424598694,
      "learning_rate": 1.571863269607926e-07,
      "loss": 0.6613,
      "step": 32012
    },
    {
      "epoch": 0.9224053477784822,
      "grad_norm": 0.7413924932479858,
      "learning_rate": 1.570702697141807e-07,
      "loss": 0.6694,
      "step": 32013
    },
    {
      "epoch": 0.9224341612401313,
      "grad_norm": 0.7335978150367737,
      "learning_rate": 1.5695425464464563e-07,
      "loss": 0.6911,
      "step": 32014
    },
    {
      "epoch": 0.9224629747017806,
      "grad_norm": 0.7503931522369385,
      "learning_rate": 1.5683828175319827e-07,
      "loss": 0.7015,
      "step": 32015
    },
    {
      "epoch": 0.9224917881634299,
      "grad_norm": 0.7382579445838928,
      "learning_rate": 1.5672235104084776e-07,
      "loss": 0.6719,
      "step": 32016
    },
    {
      "epoch": 0.9225206016250792,
      "grad_norm": 0.7435264587402344,
      "learning_rate": 1.5660646250860445e-07,
      "loss": 0.6859,
      "step": 32017
    },
    {
      "epoch": 0.9225494150867285,
      "grad_norm": 0.7441328763961792,
      "learning_rate": 1.564906161574764e-07,
      "loss": 0.6834,
      "step": 32018
    },
    {
      "epoch": 0.9225782285483778,
      "grad_norm": 0.7232502102851868,
      "learning_rate": 1.5637481198847448e-07,
      "loss": 0.6824,
      "step": 32019
    },
    {
      "epoch": 0.9226070420100271,
      "grad_norm": 0.6953688263893127,
      "learning_rate": 1.562590500026051e-07,
      "loss": 0.6767,
      "step": 32020
    },
    {
      "epoch": 0.9226358554716764,
      "grad_norm": 0.7180366516113281,
      "learning_rate": 1.5614333020087802e-07,
      "loss": 0.6707,
      "step": 32021
    },
    {
      "epoch": 0.9226646689333257,
      "grad_norm": 0.7394012808799744,
      "learning_rate": 1.560276525843002e-07,
      "loss": 0.6772,
      "step": 32022
    },
    {
      "epoch": 0.922693482394975,
      "grad_norm": 0.7307381629943848,
      "learning_rate": 1.559120171538786e-07,
      "loss": 0.6885,
      "step": 32023
    },
    {
      "epoch": 0.9227222958566242,
      "grad_norm": 0.7238186001777649,
      "learning_rate": 1.5579642391062188e-07,
      "loss": 0.6842,
      "step": 32024
    },
    {
      "epoch": 0.9227511093182735,
      "grad_norm": 0.7289809584617615,
      "learning_rate": 1.556808728555359e-07,
      "loss": 0.6879,
      "step": 32025
    },
    {
      "epoch": 0.9227799227799228,
      "grad_norm": 0.7573821544647217,
      "learning_rate": 1.5556536398962652e-07,
      "loss": 0.6966,
      "step": 32026
    },
    {
      "epoch": 0.9228087362415721,
      "grad_norm": 0.720032274723053,
      "learning_rate": 1.554498973139007e-07,
      "loss": 0.661,
      "step": 32027
    },
    {
      "epoch": 0.9228375497032213,
      "grad_norm": 0.7232115268707275,
      "learning_rate": 1.553344728293632e-07,
      "loss": 0.6871,
      "step": 32028
    },
    {
      "epoch": 0.9228663631648706,
      "grad_norm": 0.723075807094574,
      "learning_rate": 1.5521909053701989e-07,
      "loss": 0.6788,
      "step": 32029
    },
    {
      "epoch": 0.9228951766265199,
      "grad_norm": 0.7503491044044495,
      "learning_rate": 1.5510375043787496e-07,
      "loss": 0.6916,
      "step": 32030
    },
    {
      "epoch": 0.9229239900881692,
      "grad_norm": 0.7396253943443298,
      "learning_rate": 1.549884525329326e-07,
      "loss": 0.6897,
      "step": 32031
    },
    {
      "epoch": 0.9229528035498185,
      "grad_norm": 0.7083670496940613,
      "learning_rate": 1.5487319682319867e-07,
      "loss": 0.6686,
      "step": 32032
    },
    {
      "epoch": 0.9229816170114677,
      "grad_norm": 0.7072106003761292,
      "learning_rate": 1.5475798330967517e-07,
      "loss": 0.6615,
      "step": 32033
    },
    {
      "epoch": 0.923010430473117,
      "grad_norm": 0.7261601686477661,
      "learning_rate": 1.5464281199336683e-07,
      "loss": 0.6825,
      "step": 32034
    },
    {
      "epoch": 0.9230392439347663,
      "grad_norm": 0.7057973742485046,
      "learning_rate": 1.5452768287527565e-07,
      "loss": 0.6761,
      "step": 32035
    },
    {
      "epoch": 0.9230680573964156,
      "grad_norm": 0.7315642237663269,
      "learning_rate": 1.5441259595640413e-07,
      "loss": 0.6961,
      "step": 32036
    },
    {
      "epoch": 0.9230968708580649,
      "grad_norm": 0.7301429510116577,
      "learning_rate": 1.5429755123775535e-07,
      "loss": 0.6671,
      "step": 32037
    },
    {
      "epoch": 0.9231256843197142,
      "grad_norm": 0.723421037197113,
      "learning_rate": 1.5418254872033135e-07,
      "loss": 0.6746,
      "step": 32038
    },
    {
      "epoch": 0.9231544977813635,
      "grad_norm": 0.7152710556983948,
      "learning_rate": 1.540675884051329e-07,
      "loss": 0.6646,
      "step": 32039
    },
    {
      "epoch": 0.9231833112430128,
      "grad_norm": 0.7005769610404968,
      "learning_rate": 1.5395267029316207e-07,
      "loss": 0.6473,
      "step": 32040
    },
    {
      "epoch": 0.923212124704662,
      "grad_norm": 0.6894041299819946,
      "learning_rate": 1.5383779438541913e-07,
      "loss": 0.6325,
      "step": 32041
    },
    {
      "epoch": 0.9232409381663113,
      "grad_norm": 0.731963574886322,
      "learning_rate": 1.5372296068290493e-07,
      "loss": 0.6905,
      "step": 32042
    },
    {
      "epoch": 0.9232697516279605,
      "grad_norm": 0.7351294755935669,
      "learning_rate": 1.5360816918661813e-07,
      "loss": 0.6908,
      "step": 32043
    },
    {
      "epoch": 0.9232985650896098,
      "grad_norm": 0.7144903540611267,
      "learning_rate": 1.5349341989756073e-07,
      "loss": 0.6801,
      "step": 32044
    },
    {
      "epoch": 0.9233273785512591,
      "grad_norm": 0.6984494924545288,
      "learning_rate": 1.533787128167302e-07,
      "loss": 0.6767,
      "step": 32045
    },
    {
      "epoch": 0.9233561920129084,
      "grad_norm": 0.7369824051856995,
      "learning_rate": 1.532640479451264e-07,
      "loss": 0.6708,
      "step": 32046
    },
    {
      "epoch": 0.9233850054745577,
      "grad_norm": 0.7090761065483093,
      "learning_rate": 1.5314942528374733e-07,
      "loss": 0.6374,
      "step": 32047
    },
    {
      "epoch": 0.923413818936207,
      "grad_norm": 0.7329209446907043,
      "learning_rate": 1.530348448335922e-07,
      "loss": 0.6908,
      "step": 32048
    },
    {
      "epoch": 0.9234426323978563,
      "grad_norm": 0.6995730996131897,
      "learning_rate": 1.5292030659565805e-07,
      "loss": 0.6519,
      "step": 32049
    },
    {
      "epoch": 0.9234714458595056,
      "grad_norm": 0.7299736738204956,
      "learning_rate": 1.5280581057094346e-07,
      "loss": 0.676,
      "step": 32050
    },
    {
      "epoch": 0.9235002593211549,
      "grad_norm": 0.7298166751861572,
      "learning_rate": 1.5269135676044434e-07,
      "loss": 0.6686,
      "step": 32051
    },
    {
      "epoch": 0.9235290727828042,
      "grad_norm": 0.7079244256019592,
      "learning_rate": 1.5257694516515765e-07,
      "loss": 0.6866,
      "step": 32052
    },
    {
      "epoch": 0.9235578862444535,
      "grad_norm": 0.7248063683509827,
      "learning_rate": 1.5246257578608093e-07,
      "loss": 0.6854,
      "step": 32053
    },
    {
      "epoch": 0.9235866997061027,
      "grad_norm": 0.6999165415763855,
      "learning_rate": 1.523482486242084e-07,
      "loss": 0.6597,
      "step": 32054
    },
    {
      "epoch": 0.9236155131677519,
      "grad_norm": 0.7073180079460144,
      "learning_rate": 1.5223396368053757e-07,
      "loss": 0.6848,
      "step": 32055
    },
    {
      "epoch": 0.9236443266294012,
      "grad_norm": 0.7080627679824829,
      "learning_rate": 1.5211972095606264e-07,
      "loss": 0.6923,
      "step": 32056
    },
    {
      "epoch": 0.9236731400910505,
      "grad_norm": 0.722517192363739,
      "learning_rate": 1.5200552045177897e-07,
      "loss": 0.7001,
      "step": 32057
    },
    {
      "epoch": 0.9237019535526998,
      "grad_norm": 0.70663982629776,
      "learning_rate": 1.5189136216868072e-07,
      "loss": 0.6655,
      "step": 32058
    },
    {
      "epoch": 0.9237307670143491,
      "grad_norm": 0.7257100343704224,
      "learning_rate": 1.5177724610776211e-07,
      "loss": 0.6685,
      "step": 32059
    },
    {
      "epoch": 0.9237595804759984,
      "grad_norm": 0.7142754197120667,
      "learning_rate": 1.5166317227001792e-07,
      "loss": 0.6851,
      "step": 32060
    },
    {
      "epoch": 0.9237883939376477,
      "grad_norm": 0.7167515158653259,
      "learning_rate": 1.5154914065644067e-07,
      "loss": 0.6912,
      "step": 32061
    },
    {
      "epoch": 0.923817207399297,
      "grad_norm": 0.717605471611023,
      "learning_rate": 1.5143515126802343e-07,
      "loss": 0.6846,
      "step": 32062
    },
    {
      "epoch": 0.9238460208609462,
      "grad_norm": 0.7146702408790588,
      "learning_rate": 1.513212041057599e-07,
      "loss": 0.6756,
      "step": 32063
    },
    {
      "epoch": 0.9238748343225955,
      "grad_norm": 0.7225658297538757,
      "learning_rate": 1.5120729917064147e-07,
      "loss": 0.6695,
      "step": 32064
    },
    {
      "epoch": 0.9239036477842448,
      "grad_norm": 0.7196248769760132,
      "learning_rate": 1.510934364636607e-07,
      "loss": 0.7028,
      "step": 32065
    },
    {
      "epoch": 0.9239324612458941,
      "grad_norm": 0.7350866198539734,
      "learning_rate": 1.5097961598580847e-07,
      "loss": 0.7035,
      "step": 32066
    },
    {
      "epoch": 0.9239612747075434,
      "grad_norm": 0.7050418257713318,
      "learning_rate": 1.5086583773807673e-07,
      "loss": 0.6773,
      "step": 32067
    },
    {
      "epoch": 0.9239900881691926,
      "grad_norm": 0.7296857237815857,
      "learning_rate": 1.5075210172145583e-07,
      "loss": 0.6811,
      "step": 32068
    },
    {
      "epoch": 0.9240189016308419,
      "grad_norm": 0.7371379733085632,
      "learning_rate": 1.5063840793693718e-07,
      "loss": 0.6854,
      "step": 32069
    },
    {
      "epoch": 0.9240477150924912,
      "grad_norm": 0.7302629351615906,
      "learning_rate": 1.5052475638551e-07,
      "loss": 0.6935,
      "step": 32070
    },
    {
      "epoch": 0.9240765285541405,
      "grad_norm": 0.7205797433853149,
      "learning_rate": 1.5041114706816462e-07,
      "loss": 0.6768,
      "step": 32071
    },
    {
      "epoch": 0.9241053420157898,
      "grad_norm": 0.7131739854812622,
      "learning_rate": 1.5029757998589023e-07,
      "loss": 0.6995,
      "step": 32072
    },
    {
      "epoch": 0.924134155477439,
      "grad_norm": 0.7158488035202026,
      "learning_rate": 1.501840551396766e-07,
      "loss": 0.6514,
      "step": 32073
    },
    {
      "epoch": 0.9241629689390883,
      "grad_norm": 0.7367962598800659,
      "learning_rate": 1.5007057253051127e-07,
      "loss": 0.6787,
      "step": 32074
    },
    {
      "epoch": 0.9241917824007376,
      "grad_norm": 0.7068071365356445,
      "learning_rate": 1.499571321593829e-07,
      "loss": 0.6524,
      "step": 32075
    },
    {
      "epoch": 0.9242205958623869,
      "grad_norm": 0.7176833748817444,
      "learning_rate": 1.4984373402728014e-07,
      "loss": 0.6871,
      "step": 32076
    },
    {
      "epoch": 0.9242494093240362,
      "grad_norm": 0.723149299621582,
      "learning_rate": 1.4973037813518998e-07,
      "loss": 0.6687,
      "step": 32077
    },
    {
      "epoch": 0.9242782227856855,
      "grad_norm": 0.7122412919998169,
      "learning_rate": 1.4961706448409995e-07,
      "loss": 0.6513,
      "step": 32078
    },
    {
      "epoch": 0.9243070362473348,
      "grad_norm": 0.751042366027832,
      "learning_rate": 1.4950379307499596e-07,
      "loss": 0.683,
      "step": 32079
    },
    {
      "epoch": 0.9243358497089841,
      "grad_norm": 0.7084032893180847,
      "learning_rate": 1.4939056390886608e-07,
      "loss": 0.6653,
      "step": 32080
    },
    {
      "epoch": 0.9243646631706334,
      "grad_norm": 0.7064367532730103,
      "learning_rate": 1.4927737698669453e-07,
      "loss": 0.6745,
      "step": 32081
    },
    {
      "epoch": 0.9243934766322826,
      "grad_norm": 0.7418463230133057,
      "learning_rate": 1.4916423230946885e-07,
      "loss": 0.7032,
      "step": 32082
    },
    {
      "epoch": 0.9244222900939318,
      "grad_norm": 0.7276022434234619,
      "learning_rate": 1.4905112987817326e-07,
      "loss": 0.6808,
      "step": 32083
    },
    {
      "epoch": 0.9244511035555811,
      "grad_norm": 0.7409821152687073,
      "learning_rate": 1.4893806969379364e-07,
      "loss": 0.6784,
      "step": 32084
    },
    {
      "epoch": 0.9244799170172304,
      "grad_norm": 0.7303400039672852,
      "learning_rate": 1.4882505175731367e-07,
      "loss": 0.6906,
      "step": 32085
    },
    {
      "epoch": 0.9245087304788797,
      "grad_norm": 0.7246436476707458,
      "learning_rate": 1.4871207606971804e-07,
      "loss": 0.665,
      "step": 32086
    },
    {
      "epoch": 0.924537543940529,
      "grad_norm": 0.7314152717590332,
      "learning_rate": 1.4859914263199103e-07,
      "loss": 0.6939,
      "step": 32087
    },
    {
      "epoch": 0.9245663574021783,
      "grad_norm": 0.728797972202301,
      "learning_rate": 1.484862514451163e-07,
      "loss": 0.6842,
      "step": 32088
    },
    {
      "epoch": 0.9245951708638276,
      "grad_norm": 0.7346632480621338,
      "learning_rate": 1.483734025100758e-07,
      "loss": 0.691,
      "step": 32089
    },
    {
      "epoch": 0.9246239843254769,
      "grad_norm": 0.7470504641532898,
      "learning_rate": 1.4826059582785323e-07,
      "loss": 0.6898,
      "step": 32090
    },
    {
      "epoch": 0.9246527977871262,
      "grad_norm": 0.7313433289527893,
      "learning_rate": 1.4814783139943112e-07,
      "loss": 0.6827,
      "step": 32091
    },
    {
      "epoch": 0.9246816112487755,
      "grad_norm": 0.7111681699752808,
      "learning_rate": 1.4803510922579089e-07,
      "loss": 0.6526,
      "step": 32092
    },
    {
      "epoch": 0.9247104247104247,
      "grad_norm": 0.7311742305755615,
      "learning_rate": 1.4792242930791568e-07,
      "loss": 0.6899,
      "step": 32093
    },
    {
      "epoch": 0.924739238172074,
      "grad_norm": 0.7532281279563904,
      "learning_rate": 1.4780979164678465e-07,
      "loss": 0.6881,
      "step": 32094
    },
    {
      "epoch": 0.9247680516337232,
      "grad_norm": 0.7368628978729248,
      "learning_rate": 1.476971962433804e-07,
      "loss": 0.6964,
      "step": 32095
    },
    {
      "epoch": 0.9247968650953725,
      "grad_norm": 0.7179852724075317,
      "learning_rate": 1.475846430986827e-07,
      "loss": 0.694,
      "step": 32096
    },
    {
      "epoch": 0.9248256785570218,
      "grad_norm": 0.7360732555389404,
      "learning_rate": 1.4747213221367297e-07,
      "loss": 0.681,
      "step": 32097
    },
    {
      "epoch": 0.9248544920186711,
      "grad_norm": 0.7295575737953186,
      "learning_rate": 1.473596635893293e-07,
      "loss": 0.6945,
      "step": 32098
    },
    {
      "epoch": 0.9248833054803204,
      "grad_norm": 0.7194663882255554,
      "learning_rate": 1.4724723722663315e-07,
      "loss": 0.6678,
      "step": 32099
    },
    {
      "epoch": 0.9249121189419697,
      "grad_norm": 0.7220659255981445,
      "learning_rate": 1.4713485312656152e-07,
      "loss": 0.6855,
      "step": 32100
    },
    {
      "epoch": 0.924940932403619,
      "grad_norm": 0.7200272679328918,
      "learning_rate": 1.4702251129009526e-07,
      "loss": 0.6949,
      "step": 32101
    },
    {
      "epoch": 0.9249697458652683,
      "grad_norm": 0.7292635440826416,
      "learning_rate": 1.4691021171821084e-07,
      "loss": 0.7012,
      "step": 32102
    },
    {
      "epoch": 0.9249985593269175,
      "grad_norm": 0.7396231889724731,
      "learning_rate": 1.4679795441188749e-07,
      "loss": 0.6966,
      "step": 32103
    },
    {
      "epoch": 0.9250273727885668,
      "grad_norm": 0.7538405656814575,
      "learning_rate": 1.4668573937210218e-07,
      "loss": 0.7154,
      "step": 32104
    },
    {
      "epoch": 0.9250561862502161,
      "grad_norm": 0.7224234342575073,
      "learning_rate": 1.4657356659983357e-07,
      "loss": 0.7056,
      "step": 32105
    },
    {
      "epoch": 0.9250849997118654,
      "grad_norm": 0.7054940462112427,
      "learning_rate": 1.464614360960559e-07,
      "loss": 0.6854,
      "step": 32106
    },
    {
      "epoch": 0.9251138131735147,
      "grad_norm": 0.7312548160552979,
      "learning_rate": 1.4634934786174892e-07,
      "loss": 0.7013,
      "step": 32107
    },
    {
      "epoch": 0.925142626635164,
      "grad_norm": 0.7285480499267578,
      "learning_rate": 1.462373018978863e-07,
      "loss": 0.7007,
      "step": 32108
    },
    {
      "epoch": 0.9251714400968132,
      "grad_norm": 0.7218990921974182,
      "learning_rate": 1.4612529820544508e-07,
      "loss": 0.6902,
      "step": 32109
    },
    {
      "epoch": 0.9252002535584625,
      "grad_norm": 0.7333155274391174,
      "learning_rate": 1.4601333678540053e-07,
      "loss": 0.6763,
      "step": 32110
    },
    {
      "epoch": 0.9252290670201118,
      "grad_norm": 0.7243610620498657,
      "learning_rate": 1.45901417638728e-07,
      "loss": 0.7043,
      "step": 32111
    },
    {
      "epoch": 0.925257880481761,
      "grad_norm": 0.7202507853507996,
      "learning_rate": 1.457895407664006e-07,
      "loss": 0.6599,
      "step": 32112
    },
    {
      "epoch": 0.9252866939434103,
      "grad_norm": 0.7040198445320129,
      "learning_rate": 1.4567770616939482e-07,
      "loss": 0.6638,
      "step": 32113
    },
    {
      "epoch": 0.9253155074050596,
      "grad_norm": 0.7393275499343872,
      "learning_rate": 1.4556591384868368e-07,
      "loss": 0.6914,
      "step": 32114
    },
    {
      "epoch": 0.9253443208667089,
      "grad_norm": 0.7311107516288757,
      "learning_rate": 1.4545416380524035e-07,
      "loss": 0.6857,
      "step": 32115
    },
    {
      "epoch": 0.9253731343283582,
      "grad_norm": 0.7181360125541687,
      "learning_rate": 1.4534245604003905e-07,
      "loss": 0.6832,
      "step": 32116
    },
    {
      "epoch": 0.9254019477900075,
      "grad_norm": 0.725062370300293,
      "learning_rate": 1.452307905540512e-07,
      "loss": 0.6802,
      "step": 32117
    },
    {
      "epoch": 0.9254307612516568,
      "grad_norm": 0.7288260459899902,
      "learning_rate": 1.451191673482505e-07,
      "loss": 0.6743,
      "step": 32118
    },
    {
      "epoch": 0.9254595747133061,
      "grad_norm": 0.7241653800010681,
      "learning_rate": 1.450075864236089e-07,
      "loss": 0.6776,
      "step": 32119
    },
    {
      "epoch": 0.9254883881749554,
      "grad_norm": 0.7266867160797119,
      "learning_rate": 1.4489604778109846e-07,
      "loss": 0.6531,
      "step": 32120
    },
    {
      "epoch": 0.9255172016366047,
      "grad_norm": 0.7181692719459534,
      "learning_rate": 1.4478455142168944e-07,
      "loss": 0.6805,
      "step": 32121
    },
    {
      "epoch": 0.9255460150982538,
      "grad_norm": 0.7465704679489136,
      "learning_rate": 1.4467309734635394e-07,
      "loss": 0.6903,
      "step": 32122
    },
    {
      "epoch": 0.9255748285599031,
      "grad_norm": 0.7057093977928162,
      "learning_rate": 1.4456168555606166e-07,
      "loss": 0.6604,
      "step": 32123
    },
    {
      "epoch": 0.9256036420215524,
      "grad_norm": 0.7364311814308167,
      "learning_rate": 1.4445031605178406e-07,
      "loss": 0.6799,
      "step": 32124
    },
    {
      "epoch": 0.9256324554832017,
      "grad_norm": 0.7015278339385986,
      "learning_rate": 1.443389888344904e-07,
      "loss": 0.6702,
      "step": 32125
    },
    {
      "epoch": 0.925661268944851,
      "grad_norm": 0.7196528911590576,
      "learning_rate": 1.4422770390515039e-07,
      "loss": 0.7004,
      "step": 32126
    },
    {
      "epoch": 0.9256900824065003,
      "grad_norm": 0.7283373475074768,
      "learning_rate": 1.4411646126473223e-07,
      "loss": 0.6727,
      "step": 32127
    },
    {
      "epoch": 0.9257188958681496,
      "grad_norm": 0.7078765034675598,
      "learning_rate": 1.440052609142062e-07,
      "loss": 0.6748,
      "step": 32128
    },
    {
      "epoch": 0.9257477093297989,
      "grad_norm": 0.7099056839942932,
      "learning_rate": 1.4389410285454042e-07,
      "loss": 0.6677,
      "step": 32129
    },
    {
      "epoch": 0.9257765227914482,
      "grad_norm": 0.7369797825813293,
      "learning_rate": 1.437829870867019e-07,
      "loss": 0.7053,
      "step": 32130
    },
    {
      "epoch": 0.9258053362530975,
      "grad_norm": 0.7581735849380493,
      "learning_rate": 1.4367191361165932e-07,
      "loss": 0.7129,
      "step": 32131
    },
    {
      "epoch": 0.9258341497147468,
      "grad_norm": 0.7191327214241028,
      "learning_rate": 1.4356088243038024e-07,
      "loss": 0.6641,
      "step": 32132
    },
    {
      "epoch": 0.925862963176396,
      "grad_norm": 0.7357496619224548,
      "learning_rate": 1.4344989354383056e-07,
      "loss": 0.6763,
      "step": 32133
    },
    {
      "epoch": 0.9258917766380453,
      "grad_norm": 0.718329906463623,
      "learning_rate": 1.433389469529778e-07,
      "loss": 0.653,
      "step": 32134
    },
    {
      "epoch": 0.9259205900996946,
      "grad_norm": 0.7218799591064453,
      "learning_rate": 1.4322804265878844e-07,
      "loss": 0.6689,
      "step": 32135
    },
    {
      "epoch": 0.9259494035613438,
      "grad_norm": 0.7134959697723389,
      "learning_rate": 1.4311718066222723e-07,
      "loss": 0.6664,
      "step": 32136
    },
    {
      "epoch": 0.9259782170229931,
      "grad_norm": 0.7145582437515259,
      "learning_rate": 1.4300636096426123e-07,
      "loss": 0.6948,
      "step": 32137
    },
    {
      "epoch": 0.9260070304846424,
      "grad_norm": 0.7052345275878906,
      "learning_rate": 1.4289558356585354e-07,
      "loss": 0.6728,
      "step": 32138
    },
    {
      "epoch": 0.9260358439462917,
      "grad_norm": 0.722538411617279,
      "learning_rate": 1.4278484846797058e-07,
      "loss": 0.6705,
      "step": 32139
    },
    {
      "epoch": 0.926064657407941,
      "grad_norm": 0.7206475734710693,
      "learning_rate": 1.4267415567157606e-07,
      "loss": 0.661,
      "step": 32140
    },
    {
      "epoch": 0.9260934708695903,
      "grad_norm": 0.7167229652404785,
      "learning_rate": 1.4256350517763417e-07,
      "loss": 0.689,
      "step": 32141
    },
    {
      "epoch": 0.9261222843312396,
      "grad_norm": 0.7292056679725647,
      "learning_rate": 1.424528969871075e-07,
      "loss": 0.687,
      "step": 32142
    },
    {
      "epoch": 0.9261510977928888,
      "grad_norm": 0.7361906170845032,
      "learning_rate": 1.423423311009614e-07,
      "loss": 0.6832,
      "step": 32143
    },
    {
      "epoch": 0.9261799112545381,
      "grad_norm": 0.7052541375160217,
      "learning_rate": 1.422318075201573e-07,
      "loss": 0.6727,
      "step": 32144
    },
    {
      "epoch": 0.9262087247161874,
      "grad_norm": 0.7313915491104126,
      "learning_rate": 1.4212132624565832e-07,
      "loss": 0.6622,
      "step": 32145
    },
    {
      "epoch": 0.9262375381778367,
      "grad_norm": 0.7367327809333801,
      "learning_rate": 1.4201088727842648e-07,
      "loss": 0.68,
      "step": 32146
    },
    {
      "epoch": 0.926266351639486,
      "grad_norm": 0.745039165019989,
      "learning_rate": 1.4190049061942378e-07,
      "loss": 0.6907,
      "step": 32147
    },
    {
      "epoch": 0.9262951651011353,
      "grad_norm": 0.7080026865005493,
      "learning_rate": 1.4179013626961113e-07,
      "loss": 0.659,
      "step": 32148
    },
    {
      "epoch": 0.9263239785627845,
      "grad_norm": 0.7214803695678711,
      "learning_rate": 1.4167982422994997e-07,
      "loss": 0.6776,
      "step": 32149
    },
    {
      "epoch": 0.9263527920244338,
      "grad_norm": 0.7360962629318237,
      "learning_rate": 1.415695545014012e-07,
      "loss": 0.6839,
      "step": 32150
    },
    {
      "epoch": 0.9263816054860831,
      "grad_norm": 0.7254083752632141,
      "learning_rate": 1.414593270849246e-07,
      "loss": 0.6579,
      "step": 32151
    },
    {
      "epoch": 0.9264104189477324,
      "grad_norm": 0.7188646793365479,
      "learning_rate": 1.4134914198148165e-07,
      "loss": 0.6817,
      "step": 32152
    },
    {
      "epoch": 0.9264392324093816,
      "grad_norm": 0.7667464017868042,
      "learning_rate": 1.412389991920293e-07,
      "loss": 0.7013,
      "step": 32153
    },
    {
      "epoch": 0.9264680458710309,
      "grad_norm": 0.7167931199073792,
      "learning_rate": 1.411288987175291e-07,
      "loss": 0.6839,
      "step": 32154
    },
    {
      "epoch": 0.9264968593326802,
      "grad_norm": 0.715951681137085,
      "learning_rate": 1.4101884055893854e-07,
      "loss": 0.6656,
      "step": 32155
    },
    {
      "epoch": 0.9265256727943295,
      "grad_norm": 0.7316824197769165,
      "learning_rate": 1.4090882471721745e-07,
      "loss": 0.672,
      "step": 32156
    },
    {
      "epoch": 0.9265544862559788,
      "grad_norm": 0.7263475656509399,
      "learning_rate": 1.4079885119332225e-07,
      "loss": 0.6799,
      "step": 32157
    },
    {
      "epoch": 0.9265832997176281,
      "grad_norm": 0.7313313484191895,
      "learning_rate": 1.4068891998821277e-07,
      "loss": 0.6978,
      "step": 32158
    },
    {
      "epoch": 0.9266121131792774,
      "grad_norm": 0.7086189985275269,
      "learning_rate": 1.4057903110284432e-07,
      "loss": 0.6932,
      "step": 32159
    },
    {
      "epoch": 0.9266409266409267,
      "grad_norm": 0.7282369136810303,
      "learning_rate": 1.4046918453817558e-07,
      "loss": 0.6928,
      "step": 32160
    },
    {
      "epoch": 0.926669740102576,
      "grad_norm": 0.7317196130752563,
      "learning_rate": 1.4035938029516194e-07,
      "loss": 0.7068,
      "step": 32161
    },
    {
      "epoch": 0.9266985535642253,
      "grad_norm": 0.748898983001709,
      "learning_rate": 1.4024961837476093e-07,
      "loss": 0.6818,
      "step": 32162
    },
    {
      "epoch": 0.9267273670258744,
      "grad_norm": 0.7070079445838928,
      "learning_rate": 1.4013989877792677e-07,
      "loss": 0.6824,
      "step": 32163
    },
    {
      "epoch": 0.9267561804875237,
      "grad_norm": 0.7186530828475952,
      "learning_rate": 1.4003022150561707e-07,
      "loss": 0.7049,
      "step": 32164
    },
    {
      "epoch": 0.926784993949173,
      "grad_norm": 0.7379255294799805,
      "learning_rate": 1.3992058655878494e-07,
      "loss": 0.6717,
      "step": 32165
    },
    {
      "epoch": 0.9268138074108223,
      "grad_norm": 0.7099356651306152,
      "learning_rate": 1.398109939383868e-07,
      "loss": 0.6735,
      "step": 32166
    },
    {
      "epoch": 0.9268426208724716,
      "grad_norm": 0.7112451195716858,
      "learning_rate": 1.3970144364537585e-07,
      "loss": 0.6788,
      "step": 32167
    },
    {
      "epoch": 0.9268714343341209,
      "grad_norm": 0.7120565176010132,
      "learning_rate": 1.3959193568070794e-07,
      "loss": 0.6785,
      "step": 32168
    },
    {
      "epoch": 0.9269002477957702,
      "grad_norm": 0.7264418005943298,
      "learning_rate": 1.3948247004533454e-07,
      "loss": 0.6829,
      "step": 32169
    },
    {
      "epoch": 0.9269290612574195,
      "grad_norm": 0.7080576419830322,
      "learning_rate": 1.3937304674021102e-07,
      "loss": 0.6517,
      "step": 32170
    },
    {
      "epoch": 0.9269578747190688,
      "grad_norm": 0.742623507976532,
      "learning_rate": 1.3926366576628881e-07,
      "loss": 0.7106,
      "step": 32171
    },
    {
      "epoch": 0.926986688180718,
      "grad_norm": 0.6984434723854065,
      "learning_rate": 1.3915432712452104e-07,
      "loss": 0.655,
      "step": 32172
    },
    {
      "epoch": 0.9270155016423673,
      "grad_norm": 0.7324645519256592,
      "learning_rate": 1.390450308158603e-07,
      "loss": 0.6847,
      "step": 32173
    },
    {
      "epoch": 0.9270443151040166,
      "grad_norm": 0.7261337637901306,
      "learning_rate": 1.3893577684125804e-07,
      "loss": 0.6786,
      "step": 32174
    },
    {
      "epoch": 0.9270731285656659,
      "grad_norm": 0.7481206655502319,
      "learning_rate": 1.3882656520166628e-07,
      "loss": 0.704,
      "step": 32175
    },
    {
      "epoch": 0.9271019420273151,
      "grad_norm": 0.70637446641922,
      "learning_rate": 1.387173958980348e-07,
      "loss": 0.7109,
      "step": 32176
    },
    {
      "epoch": 0.9271307554889644,
      "grad_norm": 0.7078364491462708,
      "learning_rate": 1.3860826893131673e-07,
      "loss": 0.674,
      "step": 32177
    },
    {
      "epoch": 0.9271595689506137,
      "grad_norm": 0.7102158069610596,
      "learning_rate": 1.3849918430245911e-07,
      "loss": 0.6744,
      "step": 32178
    },
    {
      "epoch": 0.927188382412263,
      "grad_norm": 0.7331808805465698,
      "learning_rate": 1.3839014201241506e-07,
      "loss": 0.6895,
      "step": 32179
    },
    {
      "epoch": 0.9272171958739123,
      "grad_norm": 0.7010431885719299,
      "learning_rate": 1.3828114206213273e-07,
      "loss": 0.6557,
      "step": 32180
    },
    {
      "epoch": 0.9272460093355616,
      "grad_norm": 0.7325404286384583,
      "learning_rate": 1.3817218445256243e-07,
      "loss": 0.6727,
      "step": 32181
    },
    {
      "epoch": 0.9272748227972109,
      "grad_norm": 0.7334927916526794,
      "learning_rate": 1.3806326918465173e-07,
      "loss": 0.6718,
      "step": 32182
    },
    {
      "epoch": 0.9273036362588601,
      "grad_norm": 0.7308361530303955,
      "learning_rate": 1.3795439625934992e-07,
      "loss": 0.6812,
      "step": 32183
    },
    {
      "epoch": 0.9273324497205094,
      "grad_norm": 0.7151080369949341,
      "learning_rate": 1.378455656776051e-07,
      "loss": 0.6855,
      "step": 32184
    },
    {
      "epoch": 0.9273612631821587,
      "grad_norm": 0.7084888815879822,
      "learning_rate": 1.377367774403654e-07,
      "loss": 0.6588,
      "step": 32185
    },
    {
      "epoch": 0.927390076643808,
      "grad_norm": 0.7406769394874573,
      "learning_rate": 1.376280315485773e-07,
      "loss": 0.6658,
      "step": 32186
    },
    {
      "epoch": 0.9274188901054573,
      "grad_norm": 0.7060643434524536,
      "learning_rate": 1.3751932800318834e-07,
      "loss": 0.6576,
      "step": 32187
    },
    {
      "epoch": 0.9274477035671066,
      "grad_norm": 0.7195641994476318,
      "learning_rate": 1.374106668051456e-07,
      "loss": 0.6762,
      "step": 32188
    },
    {
      "epoch": 0.9274765170287559,
      "grad_norm": 0.726001501083374,
      "learning_rate": 1.3730204795539494e-07,
      "loss": 0.679,
      "step": 32189
    },
    {
      "epoch": 0.9275053304904051,
      "grad_norm": 0.7509918212890625,
      "learning_rate": 1.371934714548828e-07,
      "loss": 0.7033,
      "step": 32190
    },
    {
      "epoch": 0.9275341439520544,
      "grad_norm": 0.7347385883331299,
      "learning_rate": 1.3708493730455464e-07,
      "loss": 0.6774,
      "step": 32191
    },
    {
      "epoch": 0.9275629574137036,
      "grad_norm": 0.7028802037239075,
      "learning_rate": 1.3697644550535515e-07,
      "loss": 0.6698,
      "step": 32192
    },
    {
      "epoch": 0.9275917708753529,
      "grad_norm": 0.6961618065834045,
      "learning_rate": 1.368679960582292e-07,
      "loss": 0.6707,
      "step": 32193
    },
    {
      "epoch": 0.9276205843370022,
      "grad_norm": 0.7156829237937927,
      "learning_rate": 1.3675958896412268e-07,
      "loss": 0.6481,
      "step": 32194
    },
    {
      "epoch": 0.9276493977986515,
      "grad_norm": 0.722297728061676,
      "learning_rate": 1.3665122422397813e-07,
      "loss": 0.6997,
      "step": 32195
    },
    {
      "epoch": 0.9276782112603008,
      "grad_norm": 0.7079833745956421,
      "learning_rate": 1.3654290183873985e-07,
      "loss": 0.6801,
      "step": 32196
    },
    {
      "epoch": 0.9277070247219501,
      "grad_norm": 0.7042922377586365,
      "learning_rate": 1.3643462180935096e-07,
      "loss": 0.6716,
      "step": 32197
    },
    {
      "epoch": 0.9277358381835994,
      "grad_norm": 0.7406887412071228,
      "learning_rate": 1.363263841367557e-07,
      "loss": 0.6704,
      "step": 32198
    },
    {
      "epoch": 0.9277646516452487,
      "grad_norm": 0.7171307802200317,
      "learning_rate": 1.3621818882189497e-07,
      "loss": 0.6848,
      "step": 32199
    },
    {
      "epoch": 0.927793465106898,
      "grad_norm": 0.7159425020217896,
      "learning_rate": 1.3611003586571192e-07,
      "loss": 0.6852,
      "step": 32200
    },
    {
      "epoch": 0.9278222785685473,
      "grad_norm": 0.7281004190444946,
      "learning_rate": 1.3600192526914802e-07,
      "loss": 0.6811,
      "step": 32201
    },
    {
      "epoch": 0.9278510920301966,
      "grad_norm": 0.7165586948394775,
      "learning_rate": 1.3589385703314527e-07,
      "loss": 0.6795,
      "step": 32202
    },
    {
      "epoch": 0.9278799054918457,
      "grad_norm": 0.7285547256469727,
      "learning_rate": 1.3578583115864519e-07,
      "loss": 0.672,
      "step": 32203
    },
    {
      "epoch": 0.927908718953495,
      "grad_norm": 0.7308185696601868,
      "learning_rate": 1.3567784764658809e-07,
      "loss": 0.7023,
      "step": 32204
    },
    {
      "epoch": 0.9279375324151443,
      "grad_norm": 0.727670431137085,
      "learning_rate": 1.355699064979138e-07,
      "loss": 0.685,
      "step": 32205
    },
    {
      "epoch": 0.9279663458767936,
      "grad_norm": 0.7217049598693848,
      "learning_rate": 1.354620077135632e-07,
      "loss": 0.6455,
      "step": 32206
    },
    {
      "epoch": 0.9279951593384429,
      "grad_norm": 0.7577047944068909,
      "learning_rate": 1.353541512944756e-07,
      "loss": 0.6983,
      "step": 32207
    },
    {
      "epoch": 0.9280239728000922,
      "grad_norm": 0.7119659185409546,
      "learning_rate": 1.3524633724159132e-07,
      "loss": 0.6921,
      "step": 32208
    },
    {
      "epoch": 0.9280527862617415,
      "grad_norm": 0.7009725570678711,
      "learning_rate": 1.3513856555584736e-07,
      "loss": 0.661,
      "step": 32209
    },
    {
      "epoch": 0.9280815997233908,
      "grad_norm": 0.7522858381271362,
      "learning_rate": 1.3503083623818413e-07,
      "loss": 0.6984,
      "step": 32210
    },
    {
      "epoch": 0.9281104131850401,
      "grad_norm": 0.7126774191856384,
      "learning_rate": 1.349231492895392e-07,
      "loss": 0.6971,
      "step": 32211
    },
    {
      "epoch": 0.9281392266466894,
      "grad_norm": 0.7236599922180176,
      "learning_rate": 1.348155047108496e-07,
      "loss": 0.6871,
      "step": 32212
    },
    {
      "epoch": 0.9281680401083386,
      "grad_norm": 0.7139468193054199,
      "learning_rate": 1.3470790250305454e-07,
      "loss": 0.6694,
      "step": 32213
    },
    {
      "epoch": 0.9281968535699879,
      "grad_norm": 0.7378641963005066,
      "learning_rate": 1.3460034266708887e-07,
      "loss": 0.6872,
      "step": 32214
    },
    {
      "epoch": 0.9282256670316372,
      "grad_norm": 0.7205795049667358,
      "learning_rate": 1.344928252038913e-07,
      "loss": 0.6886,
      "step": 32215
    },
    {
      "epoch": 0.9282544804932865,
      "grad_norm": 0.7307384610176086,
      "learning_rate": 1.3438535011439714e-07,
      "loss": 0.6586,
      "step": 32216
    },
    {
      "epoch": 0.9282832939549357,
      "grad_norm": 0.7170960903167725,
      "learning_rate": 1.3427791739954343e-07,
      "loss": 0.6805,
      "step": 32217
    },
    {
      "epoch": 0.928312107416585,
      "grad_norm": 0.71770840883255,
      "learning_rate": 1.3417052706026501e-07,
      "loss": 0.6984,
      "step": 32218
    },
    {
      "epoch": 0.9283409208782343,
      "grad_norm": 0.7442643642425537,
      "learning_rate": 1.340631790974972e-07,
      "loss": 0.6796,
      "step": 32219
    },
    {
      "epoch": 0.9283697343398836,
      "grad_norm": 0.7270926833152771,
      "learning_rate": 1.3395587351217433e-07,
      "loss": 0.6955,
      "step": 32220
    },
    {
      "epoch": 0.9283985478015329,
      "grad_norm": 0.7045441269874573,
      "learning_rate": 1.3384861030523278e-07,
      "loss": 0.6632,
      "step": 32221
    },
    {
      "epoch": 0.9284273612631821,
      "grad_norm": 0.724770188331604,
      "learning_rate": 1.3374138947760407e-07,
      "loss": 0.6878,
      "step": 32222
    },
    {
      "epoch": 0.9284561747248314,
      "grad_norm": 0.706847608089447,
      "learning_rate": 1.3363421103022467e-07,
      "loss": 0.6741,
      "step": 32223
    },
    {
      "epoch": 0.9284849881864807,
      "grad_norm": 0.7415145039558411,
      "learning_rate": 1.3352707496402551e-07,
      "loss": 0.6787,
      "step": 32224
    },
    {
      "epoch": 0.92851380164813,
      "grad_norm": 0.7201076149940491,
      "learning_rate": 1.3341998127994137e-07,
      "loss": 0.6779,
      "step": 32225
    },
    {
      "epoch": 0.9285426151097793,
      "grad_norm": 0.7153992056846619,
      "learning_rate": 1.3331292997890378e-07,
      "loss": 0.671,
      "step": 32226
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 0.7101707458496094,
      "learning_rate": 1.3320592106184694e-07,
      "loss": 0.6997,
      "step": 32227
    },
    {
      "epoch": 0.9286002420330779,
      "grad_norm": 0.7303780913352966,
      "learning_rate": 1.3309895452970067e-07,
      "loss": 0.6669,
      "step": 32228
    },
    {
      "epoch": 0.9286290554947272,
      "grad_norm": 0.7171802520751953,
      "learning_rate": 1.3299203038339759e-07,
      "loss": 0.6952,
      "step": 32229
    },
    {
      "epoch": 0.9286578689563764,
      "grad_norm": 0.7140617370605469,
      "learning_rate": 1.328851486238686e-07,
      "loss": 0.6557,
      "step": 32230
    },
    {
      "epoch": 0.9286866824180257,
      "grad_norm": 0.722307026386261,
      "learning_rate": 1.327783092520446e-07,
      "loss": 0.6872,
      "step": 32231
    },
    {
      "epoch": 0.928715495879675,
      "grad_norm": 0.7215783596038818,
      "learning_rate": 1.32671512268856e-07,
      "loss": 0.6605,
      "step": 32232
    },
    {
      "epoch": 0.9287443093413242,
      "grad_norm": 0.7098361253738403,
      "learning_rate": 1.3256475767523313e-07,
      "loss": 0.6711,
      "step": 32233
    },
    {
      "epoch": 0.9287731228029735,
      "grad_norm": 0.7193241119384766,
      "learning_rate": 1.3245804547210584e-07,
      "loss": 0.6939,
      "step": 32234
    },
    {
      "epoch": 0.9288019362646228,
      "grad_norm": 0.7305434942245483,
      "learning_rate": 1.323513756604028e-07,
      "loss": 0.6801,
      "step": 32235
    },
    {
      "epoch": 0.9288307497262721,
      "grad_norm": 0.7243821620941162,
      "learning_rate": 1.3224474824105381e-07,
      "loss": 0.6958,
      "step": 32236
    },
    {
      "epoch": 0.9288595631879214,
      "grad_norm": 0.7340030670166016,
      "learning_rate": 1.3213816321498707e-07,
      "loss": 0.7009,
      "step": 32237
    },
    {
      "epoch": 0.9288883766495707,
      "grad_norm": 0.7181099057197571,
      "learning_rate": 1.3203162058313012e-07,
      "loss": 0.6832,
      "step": 32238
    },
    {
      "epoch": 0.92891719011122,
      "grad_norm": 0.7223519086837769,
      "learning_rate": 1.319251203464117e-07,
      "loss": 0.6893,
      "step": 32239
    },
    {
      "epoch": 0.9289460035728693,
      "grad_norm": 0.7281912565231323,
      "learning_rate": 1.3181866250575992e-07,
      "loss": 0.7059,
      "step": 32240
    },
    {
      "epoch": 0.9289748170345186,
      "grad_norm": 0.6997873187065125,
      "learning_rate": 1.3171224706210073e-07,
      "loss": 0.6697,
      "step": 32241
    },
    {
      "epoch": 0.9290036304961679,
      "grad_norm": 0.7251102924346924,
      "learning_rate": 1.316058740163617e-07,
      "loss": 0.6783,
      "step": 32242
    },
    {
      "epoch": 0.9290324439578171,
      "grad_norm": 0.6983134150505066,
      "learning_rate": 1.314995433694688e-07,
      "loss": 0.6656,
      "step": 32243
    },
    {
      "epoch": 0.9290612574194663,
      "grad_norm": 0.721164882183075,
      "learning_rate": 1.3139325512234792e-07,
      "loss": 0.6696,
      "step": 32244
    },
    {
      "epoch": 0.9290900708811156,
      "grad_norm": 0.7253257036209106,
      "learning_rate": 1.3128700927592496e-07,
      "loss": 0.6883,
      "step": 32245
    },
    {
      "epoch": 0.9291188843427649,
      "grad_norm": 0.702004075050354,
      "learning_rate": 1.311808058311259e-07,
      "loss": 0.66,
      "step": 32246
    },
    {
      "epoch": 0.9291476978044142,
      "grad_norm": 0.7107614874839783,
      "learning_rate": 1.3107464478887387e-07,
      "loss": 0.6996,
      "step": 32247
    },
    {
      "epoch": 0.9291765112660635,
      "grad_norm": 0.7335656881332397,
      "learning_rate": 1.309685261500948e-07,
      "loss": 0.6792,
      "step": 32248
    },
    {
      "epoch": 0.9292053247277128,
      "grad_norm": 0.7038015127182007,
      "learning_rate": 1.3086244991571351e-07,
      "loss": 0.6506,
      "step": 32249
    },
    {
      "epoch": 0.9292341381893621,
      "grad_norm": 0.7607025504112244,
      "learning_rate": 1.30756416086652e-07,
      "loss": 0.7129,
      "step": 32250
    },
    {
      "epoch": 0.9292629516510114,
      "grad_norm": 0.7098591923713684,
      "learning_rate": 1.306504246638346e-07,
      "loss": 0.6589,
      "step": 32251
    },
    {
      "epoch": 0.9292917651126607,
      "grad_norm": 0.7298831939697266,
      "learning_rate": 1.3054447564818496e-07,
      "loss": 0.6976,
      "step": 32252
    },
    {
      "epoch": 0.9293205785743099,
      "grad_norm": 0.7334532141685486,
      "learning_rate": 1.3043856904062512e-07,
      "loss": 0.6749,
      "step": 32253
    },
    {
      "epoch": 0.9293493920359592,
      "grad_norm": 0.7144278287887573,
      "learning_rate": 1.3033270484207716e-07,
      "loss": 0.6674,
      "step": 32254
    },
    {
      "epoch": 0.9293782054976085,
      "grad_norm": 0.7050421237945557,
      "learning_rate": 1.302268830534642e-07,
      "loss": 0.6961,
      "step": 32255
    },
    {
      "epoch": 0.9294070189592578,
      "grad_norm": 0.7301818132400513,
      "learning_rate": 1.3012110367570662e-07,
      "loss": 0.6741,
      "step": 32256
    },
    {
      "epoch": 0.929435832420907,
      "grad_norm": 0.7122398614883423,
      "learning_rate": 1.3001536670972647e-07,
      "loss": 0.68,
      "step": 32257
    },
    {
      "epoch": 0.9294646458825563,
      "grad_norm": 0.7230674028396606,
      "learning_rate": 1.2990967215644413e-07,
      "loss": 0.6971,
      "step": 32258
    },
    {
      "epoch": 0.9294934593442056,
      "grad_norm": 0.7011846899986267,
      "learning_rate": 1.298040200167805e-07,
      "loss": 0.6773,
      "step": 32259
    },
    {
      "epoch": 0.9295222728058549,
      "grad_norm": 0.7469242811203003,
      "learning_rate": 1.2969841029165486e-07,
      "loss": 0.6965,
      "step": 32260
    },
    {
      "epoch": 0.9295510862675042,
      "grad_norm": 0.7404156923294067,
      "learning_rate": 1.295928429819876e-07,
      "loss": 0.6949,
      "step": 32261
    },
    {
      "epoch": 0.9295798997291534,
      "grad_norm": 0.7454007863998413,
      "learning_rate": 1.2948731808869797e-07,
      "loss": 0.6745,
      "step": 32262
    },
    {
      "epoch": 0.9296087131908027,
      "grad_norm": 0.7088204622268677,
      "learning_rate": 1.2938183561270522e-07,
      "loss": 0.6815,
      "step": 32263
    },
    {
      "epoch": 0.929637526652452,
      "grad_norm": 0.7105650901794434,
      "learning_rate": 1.292763955549281e-07,
      "loss": 0.6689,
      "step": 32264
    },
    {
      "epoch": 0.9296663401141013,
      "grad_norm": 0.7154534459114075,
      "learning_rate": 1.2917099791628475e-07,
      "loss": 0.6756,
      "step": 32265
    },
    {
      "epoch": 0.9296951535757506,
      "grad_norm": 0.71306312084198,
      "learning_rate": 1.2906564269769217e-07,
      "loss": 0.6686,
      "step": 32266
    },
    {
      "epoch": 0.9297239670373999,
      "grad_norm": 0.7461914420127869,
      "learning_rate": 1.289603299000697e-07,
      "loss": 0.6882,
      "step": 32267
    },
    {
      "epoch": 0.9297527804990492,
      "grad_norm": 0.711908757686615,
      "learning_rate": 1.2885505952433264e-07,
      "loss": 0.6645,
      "step": 32268
    },
    {
      "epoch": 0.9297815939606985,
      "grad_norm": 0.721996009349823,
      "learning_rate": 1.287498315713992e-07,
      "loss": 0.6864,
      "step": 32269
    },
    {
      "epoch": 0.9298104074223478,
      "grad_norm": 0.7176632881164551,
      "learning_rate": 1.286446460421853e-07,
      "loss": 0.6875,
      "step": 32270
    },
    {
      "epoch": 0.929839220883997,
      "grad_norm": 0.7051892280578613,
      "learning_rate": 1.2853950293760632e-07,
      "loss": 0.6942,
      "step": 32271
    },
    {
      "epoch": 0.9298680343456462,
      "grad_norm": 0.6999430656433105,
      "learning_rate": 1.2843440225857928e-07,
      "loss": 0.6636,
      "step": 32272
    },
    {
      "epoch": 0.9298968478072955,
      "grad_norm": 0.7243826985359192,
      "learning_rate": 1.2832934400601847e-07,
      "loss": 0.6783,
      "step": 32273
    },
    {
      "epoch": 0.9299256612689448,
      "grad_norm": 0.7159113883972168,
      "learning_rate": 1.282243281808393e-07,
      "loss": 0.6869,
      "step": 32274
    },
    {
      "epoch": 0.9299544747305941,
      "grad_norm": 0.7280781269073486,
      "learning_rate": 1.28119354783956e-07,
      "loss": 0.6962,
      "step": 32275
    },
    {
      "epoch": 0.9299832881922434,
      "grad_norm": 0.741755485534668,
      "learning_rate": 1.2801442381628337e-07,
      "loss": 0.7079,
      "step": 32276
    },
    {
      "epoch": 0.9300121016538927,
      "grad_norm": 0.7144396901130676,
      "learning_rate": 1.279095352787346e-07,
      "loss": 0.6763,
      "step": 32277
    },
    {
      "epoch": 0.930040915115542,
      "grad_norm": 0.7237197756767273,
      "learning_rate": 1.27804689172224e-07,
      "loss": 0.67,
      "step": 32278
    },
    {
      "epoch": 0.9300697285771913,
      "grad_norm": 0.7458667755126953,
      "learning_rate": 1.2769988549766353e-07,
      "loss": 0.7039,
      "step": 32279
    },
    {
      "epoch": 0.9300985420388406,
      "grad_norm": 0.7291690707206726,
      "learning_rate": 1.2759512425596698e-07,
      "loss": 0.667,
      "step": 32280
    },
    {
      "epoch": 0.9301273555004899,
      "grad_norm": 0.7284132242202759,
      "learning_rate": 1.274904054480458e-07,
      "loss": 0.674,
      "step": 32281
    },
    {
      "epoch": 0.9301561689621392,
      "grad_norm": 0.7307469248771667,
      "learning_rate": 1.2738572907481318e-07,
      "loss": 0.6872,
      "step": 32282
    },
    {
      "epoch": 0.9301849824237884,
      "grad_norm": 0.7382678985595703,
      "learning_rate": 1.2728109513717946e-07,
      "loss": 0.6915,
      "step": 32283
    },
    {
      "epoch": 0.9302137958854376,
      "grad_norm": 0.7067002058029175,
      "learning_rate": 1.271765036360567e-07,
      "loss": 0.6839,
      "step": 32284
    },
    {
      "epoch": 0.9302426093470869,
      "grad_norm": 0.738020122051239,
      "learning_rate": 1.2707195457235534e-07,
      "loss": 0.6806,
      "step": 32285
    },
    {
      "epoch": 0.9302714228087362,
      "grad_norm": 0.7091238498687744,
      "learning_rate": 1.2696744794698568e-07,
      "loss": 0.6579,
      "step": 32286
    },
    {
      "epoch": 0.9303002362703855,
      "grad_norm": 0.7340977191925049,
      "learning_rate": 1.268629837608587e-07,
      "loss": 0.7019,
      "step": 32287
    },
    {
      "epoch": 0.9303290497320348,
      "grad_norm": 0.7193780541419983,
      "learning_rate": 1.2675856201488422e-07,
      "loss": 0.6915,
      "step": 32288
    },
    {
      "epoch": 0.9303578631936841,
      "grad_norm": 0.7390285730361938,
      "learning_rate": 1.26654182709971e-07,
      "loss": 0.6778,
      "step": 32289
    },
    {
      "epoch": 0.9303866766553334,
      "grad_norm": 0.7217958569526672,
      "learning_rate": 1.2654984584702768e-07,
      "loss": 0.6937,
      "step": 32290
    },
    {
      "epoch": 0.9304154901169827,
      "grad_norm": 0.706248939037323,
      "learning_rate": 1.264455514269647e-07,
      "loss": 0.6887,
      "step": 32291
    },
    {
      "epoch": 0.930444303578632,
      "grad_norm": 0.7175605893135071,
      "learning_rate": 1.2634129945068797e-07,
      "loss": 0.6933,
      "step": 32292
    },
    {
      "epoch": 0.9304731170402812,
      "grad_norm": 0.7092130184173584,
      "learning_rate": 1.2623708991910788e-07,
      "loss": 0.6563,
      "step": 32293
    },
    {
      "epoch": 0.9305019305019305,
      "grad_norm": 0.709089994430542,
      "learning_rate": 1.2613292283312984e-07,
      "loss": 0.6891,
      "step": 32294
    },
    {
      "epoch": 0.9305307439635798,
      "grad_norm": 0.7158801555633545,
      "learning_rate": 1.2602879819366255e-07,
      "loss": 0.681,
      "step": 32295
    },
    {
      "epoch": 0.9305595574252291,
      "grad_norm": 0.7345147728919983,
      "learning_rate": 1.259247160016114e-07,
      "loss": 0.6864,
      "step": 32296
    },
    {
      "epoch": 0.9305883708868784,
      "grad_norm": 0.7088497877120972,
      "learning_rate": 1.2582067625788452e-07,
      "loss": 0.6708,
      "step": 32297
    },
    {
      "epoch": 0.9306171843485276,
      "grad_norm": 0.7048462629318237,
      "learning_rate": 1.2571667896338625e-07,
      "loss": 0.6688,
      "step": 32298
    },
    {
      "epoch": 0.9306459978101769,
      "grad_norm": 0.718242347240448,
      "learning_rate": 1.2561272411902415e-07,
      "loss": 0.6798,
      "step": 32299
    },
    {
      "epoch": 0.9306748112718262,
      "grad_norm": 0.7914443612098694,
      "learning_rate": 1.2550881172570195e-07,
      "loss": 0.689,
      "step": 32300
    },
    {
      "epoch": 0.9307036247334755,
      "grad_norm": 0.7186912298202515,
      "learning_rate": 1.2540494178432562e-07,
      "loss": 0.6662,
      "step": 32301
    },
    {
      "epoch": 0.9307324381951247,
      "grad_norm": 0.7473531365394592,
      "learning_rate": 1.253011142957994e-07,
      "loss": 0.6927,
      "step": 32302
    },
    {
      "epoch": 0.930761251656774,
      "grad_norm": 0.7154701948165894,
      "learning_rate": 1.2519732926102756e-07,
      "loss": 0.6567,
      "step": 32303
    },
    {
      "epoch": 0.9307900651184233,
      "grad_norm": 0.7362405061721802,
      "learning_rate": 1.2509358668091388e-07,
      "loss": 0.6858,
      "step": 32304
    },
    {
      "epoch": 0.9308188785800726,
      "grad_norm": 0.7083755731582642,
      "learning_rate": 1.2498988655636257e-07,
      "loss": 0.7072,
      "step": 32305
    },
    {
      "epoch": 0.9308476920417219,
      "grad_norm": 0.7135040163993835,
      "learning_rate": 1.2488622888827517e-07,
      "loss": 0.6769,
      "step": 32306
    },
    {
      "epoch": 0.9308765055033712,
      "grad_norm": 0.7086467742919922,
      "learning_rate": 1.247826136775554e-07,
      "loss": 0.6783,
      "step": 32307
    },
    {
      "epoch": 0.9309053189650205,
      "grad_norm": 0.7282932996749878,
      "learning_rate": 1.246790409251064e-07,
      "loss": 0.6917,
      "step": 32308
    },
    {
      "epoch": 0.9309341324266698,
      "grad_norm": 0.7262173295021057,
      "learning_rate": 1.2457551063182915e-07,
      "loss": 0.6792,
      "step": 32309
    },
    {
      "epoch": 0.9309629458883191,
      "grad_norm": 0.7383721470832825,
      "learning_rate": 1.2447202279862513e-07,
      "loss": 0.6723,
      "step": 32310
    },
    {
      "epoch": 0.9309917593499683,
      "grad_norm": 0.7374760508537292,
      "learning_rate": 1.2436857742639585e-07,
      "loss": 0.689,
      "step": 32311
    },
    {
      "epoch": 0.9310205728116175,
      "grad_norm": 0.7188015580177307,
      "learning_rate": 1.2426517451604337e-07,
      "loss": 0.6676,
      "step": 32312
    },
    {
      "epoch": 0.9310493862732668,
      "grad_norm": 0.7153781056404114,
      "learning_rate": 1.2416181406846695e-07,
      "loss": 0.6703,
      "step": 32313
    },
    {
      "epoch": 0.9310781997349161,
      "grad_norm": 0.7417482733726501,
      "learning_rate": 1.2405849608456699e-07,
      "loss": 0.6975,
      "step": 32314
    },
    {
      "epoch": 0.9311070131965654,
      "grad_norm": 0.6957493424415588,
      "learning_rate": 1.2395522056524333e-07,
      "loss": 0.6607,
      "step": 32315
    },
    {
      "epoch": 0.9311358266582147,
      "grad_norm": 0.7274899482727051,
      "learning_rate": 1.2385198751139582e-07,
      "loss": 0.6738,
      "step": 32316
    },
    {
      "epoch": 0.931164640119864,
      "grad_norm": 0.7143210768699646,
      "learning_rate": 1.2374879692392205e-07,
      "loss": 0.6578,
      "step": 32317
    },
    {
      "epoch": 0.9311934535815133,
      "grad_norm": 0.7312827110290527,
      "learning_rate": 1.2364564880372298e-07,
      "loss": 0.6653,
      "step": 32318
    },
    {
      "epoch": 0.9312222670431626,
      "grad_norm": 0.7266501188278198,
      "learning_rate": 1.2354254315169512e-07,
      "loss": 0.6715,
      "step": 32319
    },
    {
      "epoch": 0.9312510805048119,
      "grad_norm": 0.7109244465827942,
      "learning_rate": 1.2343947996873773e-07,
      "loss": 0.6906,
      "step": 32320
    },
    {
      "epoch": 0.9312798939664612,
      "grad_norm": 0.7432295680046082,
      "learning_rate": 1.2333645925574679e-07,
      "loss": 0.6962,
      "step": 32321
    },
    {
      "epoch": 0.9313087074281104,
      "grad_norm": 0.72198486328125,
      "learning_rate": 1.2323348101362042e-07,
      "loss": 0.6837,
      "step": 32322
    },
    {
      "epoch": 0.9313375208897597,
      "grad_norm": 0.7354377508163452,
      "learning_rate": 1.2313054524325575e-07,
      "loss": 0.6768,
      "step": 32323
    },
    {
      "epoch": 0.931366334351409,
      "grad_norm": 0.7315900325775146,
      "learning_rate": 1.2302765194554922e-07,
      "loss": 0.6759,
      "step": 32324
    },
    {
      "epoch": 0.9313951478130582,
      "grad_norm": 0.7259235978126526,
      "learning_rate": 1.2292480112139628e-07,
      "loss": 0.6821,
      "step": 32325
    },
    {
      "epoch": 0.9314239612747075,
      "grad_norm": 0.7087510228157043,
      "learning_rate": 1.2282199277169337e-07,
      "loss": 0.7104,
      "step": 32326
    },
    {
      "epoch": 0.9314527747363568,
      "grad_norm": 0.7105893492698669,
      "learning_rate": 1.2271922689733485e-07,
      "loss": 0.6843,
      "step": 32327
    },
    {
      "epoch": 0.9314815881980061,
      "grad_norm": 0.7170511484146118,
      "learning_rate": 1.2261650349921662e-07,
      "loss": 0.6792,
      "step": 32328
    },
    {
      "epoch": 0.9315104016596554,
      "grad_norm": 0.7437854409217834,
      "learning_rate": 1.2251382257823352e-07,
      "loss": 0.6855,
      "step": 32329
    },
    {
      "epoch": 0.9315392151213047,
      "grad_norm": 0.7207963466644287,
      "learning_rate": 1.2241118413527932e-07,
      "loss": 0.6881,
      "step": 32330
    },
    {
      "epoch": 0.931568028582954,
      "grad_norm": 0.7035330533981323,
      "learning_rate": 1.223085881712477e-07,
      "loss": 0.671,
      "step": 32331
    },
    {
      "epoch": 0.9315968420446032,
      "grad_norm": 0.7256706953048706,
      "learning_rate": 1.2220603468703186e-07,
      "loss": 0.6824,
      "step": 32332
    },
    {
      "epoch": 0.9316256555062525,
      "grad_norm": 0.7357416749000549,
      "learning_rate": 1.2210352368352661e-07,
      "loss": 0.6909,
      "step": 32333
    },
    {
      "epoch": 0.9316544689679018,
      "grad_norm": 0.7279637455940247,
      "learning_rate": 1.2200105516162185e-07,
      "loss": 0.657,
      "step": 32334
    },
    {
      "epoch": 0.9316832824295511,
      "grad_norm": 0.7171444296836853,
      "learning_rate": 1.2189862912221296e-07,
      "loss": 0.6845,
      "step": 32335
    },
    {
      "epoch": 0.9317120958912004,
      "grad_norm": 0.7264423370361328,
      "learning_rate": 1.217962455661903e-07,
      "loss": 0.6802,
      "step": 32336
    },
    {
      "epoch": 0.9317409093528497,
      "grad_norm": 0.7043594121932983,
      "learning_rate": 1.216939044944465e-07,
      "loss": 0.6662,
      "step": 32337
    },
    {
      "epoch": 0.9317697228144989,
      "grad_norm": 0.7228793501853943,
      "learning_rate": 1.2159160590787145e-07,
      "loss": 0.6877,
      "step": 32338
    },
    {
      "epoch": 0.9317985362761482,
      "grad_norm": 0.7552815675735474,
      "learning_rate": 1.2148934980735772e-07,
      "loss": 0.6926,
      "step": 32339
    },
    {
      "epoch": 0.9318273497377975,
      "grad_norm": 0.7294560074806213,
      "learning_rate": 1.2138713619379406e-07,
      "loss": 0.684,
      "step": 32340
    },
    {
      "epoch": 0.9318561631994468,
      "grad_norm": 0.8022223114967346,
      "learning_rate": 1.2128496506807251e-07,
      "loss": 0.69,
      "step": 32341
    },
    {
      "epoch": 0.931884976661096,
      "grad_norm": 0.7384412288665771,
      "learning_rate": 1.2118283643108132e-07,
      "loss": 0.68,
      "step": 32342
    },
    {
      "epoch": 0.9319137901227453,
      "grad_norm": 0.7257267832756042,
      "learning_rate": 1.2108075028371135e-07,
      "loss": 0.6732,
      "step": 32343
    },
    {
      "epoch": 0.9319426035843946,
      "grad_norm": 0.7355895638465881,
      "learning_rate": 1.2097870662685029e-07,
      "loss": 0.6967,
      "step": 32344
    },
    {
      "epoch": 0.9319714170460439,
      "grad_norm": 0.7317768335342407,
      "learning_rate": 1.2087670546138742e-07,
      "loss": 0.6828,
      "step": 32345
    },
    {
      "epoch": 0.9320002305076932,
      "grad_norm": 0.7179344296455383,
      "learning_rate": 1.2077474678821088e-07,
      "loss": 0.6839,
      "step": 32346
    },
    {
      "epoch": 0.9320290439693425,
      "grad_norm": 0.719768762588501,
      "learning_rate": 1.2067283060820944e-07,
      "loss": 0.6872,
      "step": 32347
    },
    {
      "epoch": 0.9320578574309918,
      "grad_norm": 0.7268955111503601,
      "learning_rate": 1.2057095692226962e-07,
      "loss": 0.6928,
      "step": 32348
    },
    {
      "epoch": 0.9320866708926411,
      "grad_norm": 0.6920902132987976,
      "learning_rate": 1.2046912573127956e-07,
      "loss": 0.6578,
      "step": 32349
    },
    {
      "epoch": 0.9321154843542904,
      "grad_norm": 0.743140697479248,
      "learning_rate": 1.203673370361258e-07,
      "loss": 0.7058,
      "step": 32350
    },
    {
      "epoch": 0.9321442978159397,
      "grad_norm": 0.7115433812141418,
      "learning_rate": 1.202655908376943e-07,
      "loss": 0.6777,
      "step": 32351
    },
    {
      "epoch": 0.9321731112775888,
      "grad_norm": 0.7307305335998535,
      "learning_rate": 1.2016388713687155e-07,
      "loss": 0.7001,
      "step": 32352
    },
    {
      "epoch": 0.9322019247392381,
      "grad_norm": 0.7155624628067017,
      "learning_rate": 1.2006222593454352e-07,
      "loss": 0.6697,
      "step": 32353
    },
    {
      "epoch": 0.9322307382008874,
      "grad_norm": 0.7741981744766235,
      "learning_rate": 1.199606072315951e-07,
      "loss": 0.6884,
      "step": 32354
    },
    {
      "epoch": 0.9322595516625367,
      "grad_norm": 0.7570966482162476,
      "learning_rate": 1.198590310289116e-07,
      "loss": 0.6846,
      "step": 32355
    },
    {
      "epoch": 0.932288365124186,
      "grad_norm": 0.7250014543533325,
      "learning_rate": 1.1975749732737795e-07,
      "loss": 0.6771,
      "step": 32356
    },
    {
      "epoch": 0.9323171785858353,
      "grad_norm": 0.6972936391830444,
      "learning_rate": 1.1965600612787732e-07,
      "loss": 0.6768,
      "step": 32357
    },
    {
      "epoch": 0.9323459920474846,
      "grad_norm": 0.7270781397819519,
      "learning_rate": 1.1955455743129506e-07,
      "loss": 0.6609,
      "step": 32358
    },
    {
      "epoch": 0.9323748055091339,
      "grad_norm": 0.7278904318809509,
      "learning_rate": 1.194531512385133e-07,
      "loss": 0.6771,
      "step": 32359
    },
    {
      "epoch": 0.9324036189707832,
      "grad_norm": 0.7413440346717834,
      "learning_rate": 1.1935178755041632e-07,
      "loss": 0.6871,
      "step": 32360
    },
    {
      "epoch": 0.9324324324324325,
      "grad_norm": 0.7306925058364868,
      "learning_rate": 1.1925046636788616e-07,
      "loss": 0.6965,
      "step": 32361
    },
    {
      "epoch": 0.9324612458940817,
      "grad_norm": 0.7236878871917725,
      "learning_rate": 1.1914918769180605e-07,
      "loss": 0.6682,
      "step": 32362
    },
    {
      "epoch": 0.932490059355731,
      "grad_norm": 0.7281646132469177,
      "learning_rate": 1.1904795152305637e-07,
      "loss": 0.6923,
      "step": 32363
    },
    {
      "epoch": 0.9325188728173803,
      "grad_norm": 0.7050867676734924,
      "learning_rate": 1.1894675786252086e-07,
      "loss": 0.661,
      "step": 32364
    },
    {
      "epoch": 0.9325476862790295,
      "grad_norm": 0.7086774110794067,
      "learning_rate": 1.1884560671107936e-07,
      "loss": 0.6754,
      "step": 32365
    },
    {
      "epoch": 0.9325764997406788,
      "grad_norm": 0.7327210307121277,
      "learning_rate": 1.1874449806961286e-07,
      "loss": 0.6799,
      "step": 32366
    },
    {
      "epoch": 0.9326053132023281,
      "grad_norm": 0.6963238716125488,
      "learning_rate": 1.186434319390034e-07,
      "loss": 0.6451,
      "step": 32367
    },
    {
      "epoch": 0.9326341266639774,
      "grad_norm": 0.7189613580703735,
      "learning_rate": 1.185424083201292e-07,
      "loss": 0.6983,
      "step": 32368
    },
    {
      "epoch": 0.9326629401256267,
      "grad_norm": 0.7186572551727295,
      "learning_rate": 1.184414272138712e-07,
      "loss": 0.6956,
      "step": 32369
    },
    {
      "epoch": 0.932691753587276,
      "grad_norm": 0.7148078680038452,
      "learning_rate": 1.1834048862110815e-07,
      "loss": 0.6638,
      "step": 32370
    },
    {
      "epoch": 0.9327205670489253,
      "grad_norm": 0.7227131724357605,
      "learning_rate": 1.1823959254272043e-07,
      "loss": 0.689,
      "step": 32371
    },
    {
      "epoch": 0.9327493805105745,
      "grad_norm": 0.7334926128387451,
      "learning_rate": 1.1813873897958517e-07,
      "loss": 0.6938,
      "step": 32372
    },
    {
      "epoch": 0.9327781939722238,
      "grad_norm": 0.7195969820022583,
      "learning_rate": 1.1803792793258218e-07,
      "loss": 0.7009,
      "step": 32373
    },
    {
      "epoch": 0.9328070074338731,
      "grad_norm": 0.7284544110298157,
      "learning_rate": 1.1793715940258798e-07,
      "loss": 0.6884,
      "step": 32374
    },
    {
      "epoch": 0.9328358208955224,
      "grad_norm": 0.7190991044044495,
      "learning_rate": 1.1783643339048135e-07,
      "loss": 0.6676,
      "step": 32375
    },
    {
      "epoch": 0.9328646343571717,
      "grad_norm": 0.7410521507263184,
      "learning_rate": 1.1773574989713821e-07,
      "loss": 0.6672,
      "step": 32376
    },
    {
      "epoch": 0.932893447818821,
      "grad_norm": 0.7248696088790894,
      "learning_rate": 1.1763510892343732e-07,
      "loss": 0.6718,
      "step": 32377
    },
    {
      "epoch": 0.9329222612804703,
      "grad_norm": 0.7182756066322327,
      "learning_rate": 1.17534510470253e-07,
      "loss": 0.6864,
      "step": 32378
    },
    {
      "epoch": 0.9329510747421195,
      "grad_norm": 0.7116816639900208,
      "learning_rate": 1.1743395453846341e-07,
      "loss": 0.6696,
      "step": 32379
    },
    {
      "epoch": 0.9329798882037688,
      "grad_norm": 0.7443142533302307,
      "learning_rate": 1.173334411289423e-07,
      "loss": 0.6995,
      "step": 32380
    },
    {
      "epoch": 0.933008701665418,
      "grad_norm": 0.7212308049201965,
      "learning_rate": 1.1723297024256674e-07,
      "loss": 0.7003,
      "step": 32381
    },
    {
      "epoch": 0.9330375151270673,
      "grad_norm": 0.7261930704116821,
      "learning_rate": 1.1713254188021051e-07,
      "loss": 0.6494,
      "step": 32382
    },
    {
      "epoch": 0.9330663285887166,
      "grad_norm": 0.7293357253074646,
      "learning_rate": 1.1703215604274898e-07,
      "loss": 0.6829,
      "step": 32383
    },
    {
      "epoch": 0.9330951420503659,
      "grad_norm": 0.7217013835906982,
      "learning_rate": 1.1693181273105647e-07,
      "loss": 0.6799,
      "step": 32384
    },
    {
      "epoch": 0.9331239555120152,
      "grad_norm": 0.7188169360160828,
      "learning_rate": 1.1683151194600618e-07,
      "loss": 0.6892,
      "step": 32385
    },
    {
      "epoch": 0.9331527689736645,
      "grad_norm": 0.7056083679199219,
      "learning_rate": 1.1673125368847238e-07,
      "loss": 0.6884,
      "step": 32386
    },
    {
      "epoch": 0.9331815824353138,
      "grad_norm": 0.7218092083930969,
      "learning_rate": 1.1663103795932717e-07,
      "loss": 0.6764,
      "step": 32387
    },
    {
      "epoch": 0.9332103958969631,
      "grad_norm": 0.7250014543533325,
      "learning_rate": 1.1653086475944487e-07,
      "loss": 0.6518,
      "step": 32388
    },
    {
      "epoch": 0.9332392093586124,
      "grad_norm": 0.7504866123199463,
      "learning_rate": 1.1643073408969696e-07,
      "loss": 0.6929,
      "step": 32389
    },
    {
      "epoch": 0.9332680228202617,
      "grad_norm": 0.7327980995178223,
      "learning_rate": 1.1633064595095556e-07,
      "loss": 0.6797,
      "step": 32390
    },
    {
      "epoch": 0.933296836281911,
      "grad_norm": 0.7359107732772827,
      "learning_rate": 1.1623060034409217e-07,
      "loss": 0.705,
      "step": 32391
    },
    {
      "epoch": 0.9333256497435601,
      "grad_norm": 0.7225313782691956,
      "learning_rate": 1.1613059726997834e-07,
      "loss": 0.6936,
      "step": 32392
    },
    {
      "epoch": 0.9333544632052094,
      "grad_norm": 0.7033274173736572,
      "learning_rate": 1.1603063672948445e-07,
      "loss": 0.6695,
      "step": 32393
    },
    {
      "epoch": 0.9333832766668587,
      "grad_norm": 0.7244781851768494,
      "learning_rate": 1.1593071872348205e-07,
      "loss": 0.7044,
      "step": 32394
    },
    {
      "epoch": 0.933412090128508,
      "grad_norm": 0.7139260768890381,
      "learning_rate": 1.1583084325284045e-07,
      "loss": 0.6501,
      "step": 32395
    },
    {
      "epoch": 0.9334409035901573,
      "grad_norm": 0.7339513897895813,
      "learning_rate": 1.1573101031843059e-07,
      "loss": 0.689,
      "step": 32396
    },
    {
      "epoch": 0.9334697170518066,
      "grad_norm": 0.7082197666168213,
      "learning_rate": 1.1563121992112014e-07,
      "loss": 0.6927,
      "step": 32397
    },
    {
      "epoch": 0.9334985305134559,
      "grad_norm": 0.7709944844245911,
      "learning_rate": 1.1553147206178006e-07,
      "loss": 0.6766,
      "step": 32398
    },
    {
      "epoch": 0.9335273439751052,
      "grad_norm": 0.7163041234016418,
      "learning_rate": 1.1543176674127743e-07,
      "loss": 0.662,
      "step": 32399
    },
    {
      "epoch": 0.9335561574367545,
      "grad_norm": 0.715567409992218,
      "learning_rate": 1.1533210396048211e-07,
      "loss": 0.6795,
      "step": 32400
    },
    {
      "epoch": 0.9335849708984038,
      "grad_norm": 0.7366184592247009,
      "learning_rate": 1.1523248372026064e-07,
      "loss": 0.697,
      "step": 32401
    },
    {
      "epoch": 0.933613784360053,
      "grad_norm": 0.7253105044364929,
      "learning_rate": 1.1513290602148175e-07,
      "loss": 0.6838,
      "step": 32402
    },
    {
      "epoch": 0.9336425978217023,
      "grad_norm": 0.7099575400352478,
      "learning_rate": 1.1503337086501199e-07,
      "loss": 0.6654,
      "step": 32403
    },
    {
      "epoch": 0.9336714112833516,
      "grad_norm": 0.714669942855835,
      "learning_rate": 1.1493387825171787e-07,
      "loss": 0.6852,
      "step": 32404
    },
    {
      "epoch": 0.9337002247450009,
      "grad_norm": 0.7081010341644287,
      "learning_rate": 1.1483442818246649e-07,
      "loss": 0.6818,
      "step": 32405
    },
    {
      "epoch": 0.9337290382066501,
      "grad_norm": 0.7480250000953674,
      "learning_rate": 1.1473502065812491e-07,
      "loss": 0.6922,
      "step": 32406
    },
    {
      "epoch": 0.9337578516682994,
      "grad_norm": 0.7057364583015442,
      "learning_rate": 1.1463565567955692e-07,
      "loss": 0.6954,
      "step": 32407
    },
    {
      "epoch": 0.9337866651299487,
      "grad_norm": 0.7058331966400146,
      "learning_rate": 1.1453633324762847e-07,
      "loss": 0.6636,
      "step": 32408
    },
    {
      "epoch": 0.933815478591598,
      "grad_norm": 0.7369085550308228,
      "learning_rate": 1.144370533632061e-07,
      "loss": 0.6768,
      "step": 32409
    },
    {
      "epoch": 0.9338442920532473,
      "grad_norm": 0.71738600730896,
      "learning_rate": 1.1433781602715189e-07,
      "loss": 0.6843,
      "step": 32410
    },
    {
      "epoch": 0.9338731055148966,
      "grad_norm": 0.7224649786949158,
      "learning_rate": 1.1423862124033236e-07,
      "loss": 0.6683,
      "step": 32411
    },
    {
      "epoch": 0.9339019189765458,
      "grad_norm": 0.7109125852584839,
      "learning_rate": 1.1413946900361017e-07,
      "loss": 0.6796,
      "step": 32412
    },
    {
      "epoch": 0.9339307324381951,
      "grad_norm": 0.7306683659553528,
      "learning_rate": 1.1404035931784963e-07,
      "loss": 0.7012,
      "step": 32413
    },
    {
      "epoch": 0.9339595458998444,
      "grad_norm": 0.7458248138427734,
      "learning_rate": 1.1394129218391226e-07,
      "loss": 0.6825,
      "step": 32414
    },
    {
      "epoch": 0.9339883593614937,
      "grad_norm": 0.7135549187660217,
      "learning_rate": 1.1384226760266292e-07,
      "loss": 0.6532,
      "step": 32415
    },
    {
      "epoch": 0.934017172823143,
      "grad_norm": 0.7319150567054749,
      "learning_rate": 1.1374328557496262e-07,
      "loss": 0.6856,
      "step": 32416
    },
    {
      "epoch": 0.9340459862847923,
      "grad_norm": 0.7140320539474487,
      "learning_rate": 1.1364434610167396e-07,
      "loss": 0.69,
      "step": 32417
    },
    {
      "epoch": 0.9340747997464416,
      "grad_norm": 0.757723867893219,
      "learning_rate": 1.1354544918365795e-07,
      "loss": 0.6957,
      "step": 32418
    },
    {
      "epoch": 0.9341036132080908,
      "grad_norm": 0.7308884263038635,
      "learning_rate": 1.1344659482177722e-07,
      "loss": 0.7105,
      "step": 32419
    },
    {
      "epoch": 0.9341324266697401,
      "grad_norm": 0.7108673453330994,
      "learning_rate": 1.1334778301689165e-07,
      "loss": 0.6697,
      "step": 32420
    },
    {
      "epoch": 0.9341612401313893,
      "grad_norm": 0.7244933247566223,
      "learning_rate": 1.1324901376986219e-07,
      "loss": 0.6797,
      "step": 32421
    },
    {
      "epoch": 0.9341900535930386,
      "grad_norm": 0.7384378910064697,
      "learning_rate": 1.1315028708154818e-07,
      "loss": 0.6895,
      "step": 32422
    },
    {
      "epoch": 0.9342188670546879,
      "grad_norm": 0.7237544655799866,
      "learning_rate": 1.1305160295281115e-07,
      "loss": 0.6658,
      "step": 32423
    },
    {
      "epoch": 0.9342476805163372,
      "grad_norm": 0.722978949546814,
      "learning_rate": 1.1295296138450874e-07,
      "loss": 0.673,
      "step": 32424
    },
    {
      "epoch": 0.9342764939779865,
      "grad_norm": 0.7168613076210022,
      "learning_rate": 1.128543623775008e-07,
      "loss": 0.6612,
      "step": 32425
    },
    {
      "epoch": 0.9343053074396358,
      "grad_norm": 0.7215561866760254,
      "learning_rate": 1.1275580593264613e-07,
      "loss": 0.6733,
      "step": 32426
    },
    {
      "epoch": 0.9343341209012851,
      "grad_norm": 0.7074502110481262,
      "learning_rate": 1.126572920508029e-07,
      "loss": 0.6627,
      "step": 32427
    },
    {
      "epoch": 0.9343629343629344,
      "grad_norm": 0.7200512290000916,
      "learning_rate": 1.1255882073282931e-07,
      "loss": 0.6791,
      "step": 32428
    },
    {
      "epoch": 0.9343917478245837,
      "grad_norm": 0.726701557636261,
      "learning_rate": 1.1246039197958248e-07,
      "loss": 0.6822,
      "step": 32429
    },
    {
      "epoch": 0.934420561286233,
      "grad_norm": 0.7411659359931946,
      "learning_rate": 1.1236200579192002e-07,
      "loss": 0.6616,
      "step": 32430
    },
    {
      "epoch": 0.9344493747478823,
      "grad_norm": 0.7154849767684937,
      "learning_rate": 1.1226366217069795e-07,
      "loss": 0.6826,
      "step": 32431
    },
    {
      "epoch": 0.9344781882095315,
      "grad_norm": 0.7262533903121948,
      "learning_rate": 1.1216536111677445e-07,
      "loss": 0.6999,
      "step": 32432
    },
    {
      "epoch": 0.9345070016711807,
      "grad_norm": 0.7124021053314209,
      "learning_rate": 1.120671026310044e-07,
      "loss": 0.6678,
      "step": 32433
    },
    {
      "epoch": 0.93453581513283,
      "grad_norm": 0.7057192921638489,
      "learning_rate": 1.1196888671424378e-07,
      "loss": 0.6635,
      "step": 32434
    },
    {
      "epoch": 0.9345646285944793,
      "grad_norm": 0.7139440178871155,
      "learning_rate": 1.1187071336734744e-07,
      "loss": 0.6911,
      "step": 32435
    },
    {
      "epoch": 0.9345934420561286,
      "grad_norm": 0.7378599643707275,
      "learning_rate": 1.1177258259117141e-07,
      "loss": 0.6766,
      "step": 32436
    },
    {
      "epoch": 0.9346222555177779,
      "grad_norm": 0.7380250096321106,
      "learning_rate": 1.1167449438656942e-07,
      "loss": 0.6778,
      "step": 32437
    },
    {
      "epoch": 0.9346510689794272,
      "grad_norm": 0.7265350818634033,
      "learning_rate": 1.1157644875439633e-07,
      "loss": 0.692,
      "step": 32438
    },
    {
      "epoch": 0.9346798824410765,
      "grad_norm": 0.7205135822296143,
      "learning_rate": 1.1147844569550537e-07,
      "loss": 0.6804,
      "step": 32439
    },
    {
      "epoch": 0.9347086959027258,
      "grad_norm": 0.7242056131362915,
      "learning_rate": 1.1138048521075084e-07,
      "loss": 0.6832,
      "step": 32440
    },
    {
      "epoch": 0.934737509364375,
      "grad_norm": 0.7185145616531372,
      "learning_rate": 1.112825673009843e-07,
      "loss": 0.6588,
      "step": 32441
    },
    {
      "epoch": 0.9347663228260243,
      "grad_norm": 0.6894364953041077,
      "learning_rate": 1.111846919670606e-07,
      "loss": 0.6908,
      "step": 32442
    },
    {
      "epoch": 0.9347951362876736,
      "grad_norm": 0.7238948345184326,
      "learning_rate": 1.1108685920983132e-07,
      "loss": 0.6643,
      "step": 32443
    },
    {
      "epoch": 0.9348239497493229,
      "grad_norm": 0.7142054438591003,
      "learning_rate": 1.109890690301485e-07,
      "loss": 0.6724,
      "step": 32444
    },
    {
      "epoch": 0.9348527632109722,
      "grad_norm": 0.7183628082275391,
      "learning_rate": 1.1089132142886317e-07,
      "loss": 0.6821,
      "step": 32445
    },
    {
      "epoch": 0.9348815766726214,
      "grad_norm": 0.7285300493240356,
      "learning_rate": 1.1079361640682684e-07,
      "loss": 0.7003,
      "step": 32446
    },
    {
      "epoch": 0.9349103901342707,
      "grad_norm": 0.7232831716537476,
      "learning_rate": 1.1069595396489163e-07,
      "loss": 0.6872,
      "step": 32447
    },
    {
      "epoch": 0.93493920359592,
      "grad_norm": 0.7512466311454773,
      "learning_rate": 1.105983341039063e-07,
      "loss": 0.6459,
      "step": 32448
    },
    {
      "epoch": 0.9349680170575693,
      "grad_norm": 0.7345504760742188,
      "learning_rate": 1.1050075682472238e-07,
      "loss": 0.6768,
      "step": 32449
    },
    {
      "epoch": 0.9349968305192186,
      "grad_norm": 0.7440471053123474,
      "learning_rate": 1.1040322212818921e-07,
      "loss": 0.7023,
      "step": 32450
    },
    {
      "epoch": 0.9350256439808678,
      "grad_norm": 0.7190771102905273,
      "learning_rate": 1.103057300151561e-07,
      "loss": 0.68,
      "step": 32451
    },
    {
      "epoch": 0.9350544574425171,
      "grad_norm": 0.693228542804718,
      "learning_rate": 1.1020828048647181e-07,
      "loss": 0.6311,
      "step": 32452
    },
    {
      "epoch": 0.9350832709041664,
      "grad_norm": 0.7237754464149475,
      "learning_rate": 1.1011087354298567e-07,
      "loss": 0.661,
      "step": 32453
    },
    {
      "epoch": 0.9351120843658157,
      "grad_norm": 0.7383900880813599,
      "learning_rate": 1.1001350918554532e-07,
      "loss": 0.7069,
      "step": 32454
    },
    {
      "epoch": 0.935140897827465,
      "grad_norm": 0.7051537036895752,
      "learning_rate": 1.0991618741499954e-07,
      "loss": 0.6731,
      "step": 32455
    },
    {
      "epoch": 0.9351697112891143,
      "grad_norm": 0.7514893412590027,
      "learning_rate": 1.0981890823219543e-07,
      "loss": 0.6901,
      "step": 32456
    },
    {
      "epoch": 0.9351985247507636,
      "grad_norm": 0.734704315662384,
      "learning_rate": 1.0972167163798064e-07,
      "loss": 0.6953,
      "step": 32457
    },
    {
      "epoch": 0.9352273382124129,
      "grad_norm": 0.7187172174453735,
      "learning_rate": 1.096244776332006e-07,
      "loss": 0.6822,
      "step": 32458
    },
    {
      "epoch": 0.9352561516740622,
      "grad_norm": 0.7276287078857422,
      "learning_rate": 1.0952732621870354e-07,
      "loss": 0.6639,
      "step": 32459
    },
    {
      "epoch": 0.9352849651357114,
      "grad_norm": 0.7408393621444702,
      "learning_rate": 1.0943021739533433e-07,
      "loss": 0.6888,
      "step": 32460
    },
    {
      "epoch": 0.9353137785973606,
      "grad_norm": 0.719184935092926,
      "learning_rate": 1.0933315116394006e-07,
      "loss": 0.6959,
      "step": 32461
    },
    {
      "epoch": 0.9353425920590099,
      "grad_norm": 0.7108325958251953,
      "learning_rate": 1.0923612752536395e-07,
      "loss": 0.6894,
      "step": 32462
    },
    {
      "epoch": 0.9353714055206592,
      "grad_norm": 0.7215291261672974,
      "learning_rate": 1.0913914648045254e-07,
      "loss": 0.6607,
      "step": 32463
    },
    {
      "epoch": 0.9354002189823085,
      "grad_norm": 0.7379334568977356,
      "learning_rate": 1.0904220803005072e-07,
      "loss": 0.683,
      "step": 32464
    },
    {
      "epoch": 0.9354290324439578,
      "grad_norm": 0.7258737087249756,
      "learning_rate": 1.089453121750017e-07,
      "loss": 0.6875,
      "step": 32465
    },
    {
      "epoch": 0.9354578459056071,
      "grad_norm": 0.72529137134552,
      "learning_rate": 1.0884845891614926e-07,
      "loss": 0.685,
      "step": 32466
    },
    {
      "epoch": 0.9354866593672564,
      "grad_norm": 0.7592012882232666,
      "learning_rate": 1.0875164825433771e-07,
      "loss": 0.6913,
      "step": 32467
    },
    {
      "epoch": 0.9355154728289057,
      "grad_norm": 0.7412685751914978,
      "learning_rate": 1.0865488019041026e-07,
      "loss": 0.6746,
      "step": 32468
    },
    {
      "epoch": 0.935544286290555,
      "grad_norm": 0.7355983257293701,
      "learning_rate": 1.0855815472520903e-07,
      "loss": 0.6909,
      "step": 32469
    },
    {
      "epoch": 0.9355730997522043,
      "grad_norm": 0.7048522233963013,
      "learning_rate": 1.0846147185957667e-07,
      "loss": 0.6741,
      "step": 32470
    },
    {
      "epoch": 0.9356019132138536,
      "grad_norm": 0.7470917701721191,
      "learning_rate": 1.0836483159435473e-07,
      "loss": 0.7121,
      "step": 32471
    },
    {
      "epoch": 0.9356307266755028,
      "grad_norm": 0.7098630666732788,
      "learning_rate": 1.0826823393038588e-07,
      "loss": 0.6623,
      "step": 32472
    },
    {
      "epoch": 0.935659540137152,
      "grad_norm": 0.7164126634597778,
      "learning_rate": 1.0817167886851055e-07,
      "loss": 0.688,
      "step": 32473
    },
    {
      "epoch": 0.9356883535988013,
      "grad_norm": 0.7069657444953918,
      "learning_rate": 1.0807516640956972e-07,
      "loss": 0.6618,
      "step": 32474
    },
    {
      "epoch": 0.9357171670604506,
      "grad_norm": 0.7300854325294495,
      "learning_rate": 1.0797869655440441e-07,
      "loss": 0.684,
      "step": 32475
    },
    {
      "epoch": 0.9357459805220999,
      "grad_norm": 0.7156822681427002,
      "learning_rate": 1.0788226930385393e-07,
      "loss": 0.677,
      "step": 32476
    },
    {
      "epoch": 0.9357747939837492,
      "grad_norm": 0.7138631939888,
      "learning_rate": 1.0778588465875817e-07,
      "loss": 0.6703,
      "step": 32477
    },
    {
      "epoch": 0.9358036074453985,
      "grad_norm": 0.7126980423927307,
      "learning_rate": 1.0768954261995812e-07,
      "loss": 0.6518,
      "step": 32478
    },
    {
      "epoch": 0.9358324209070478,
      "grad_norm": 0.722484290599823,
      "learning_rate": 1.075932431882909e-07,
      "loss": 0.6888,
      "step": 32479
    },
    {
      "epoch": 0.9358612343686971,
      "grad_norm": 0.7247666120529175,
      "learning_rate": 1.0749698636459582e-07,
      "loss": 0.6856,
      "step": 32480
    },
    {
      "epoch": 0.9358900478303463,
      "grad_norm": 0.7579116821289062,
      "learning_rate": 1.0740077214971167e-07,
      "loss": 0.6824,
      "step": 32481
    },
    {
      "epoch": 0.9359188612919956,
      "grad_norm": 0.6989560127258301,
      "learning_rate": 1.0730460054447611e-07,
      "loss": 0.6674,
      "step": 32482
    },
    {
      "epoch": 0.9359476747536449,
      "grad_norm": 0.7521011233329773,
      "learning_rate": 1.0720847154972568e-07,
      "loss": 0.6813,
      "step": 32483
    },
    {
      "epoch": 0.9359764882152942,
      "grad_norm": 0.7132394313812256,
      "learning_rate": 1.0711238516629918e-07,
      "loss": 0.667,
      "step": 32484
    },
    {
      "epoch": 0.9360053016769435,
      "grad_norm": 0.7340499758720398,
      "learning_rate": 1.0701634139503259e-07,
      "loss": 0.691,
      "step": 32485
    },
    {
      "epoch": 0.9360341151385928,
      "grad_norm": 0.7140748500823975,
      "learning_rate": 1.0692034023676246e-07,
      "loss": 0.6797,
      "step": 32486
    },
    {
      "epoch": 0.936062928600242,
      "grad_norm": 0.7091861367225647,
      "learning_rate": 1.0682438169232534e-07,
      "loss": 0.6481,
      "step": 32487
    },
    {
      "epoch": 0.9360917420618913,
      "grad_norm": 0.719754159450531,
      "learning_rate": 1.0672846576255558e-07,
      "loss": 0.6635,
      "step": 32488
    },
    {
      "epoch": 0.9361205555235406,
      "grad_norm": 0.7037558555603027,
      "learning_rate": 1.0663259244828971e-07,
      "loss": 0.675,
      "step": 32489
    },
    {
      "epoch": 0.9361493689851899,
      "grad_norm": 0.7162353992462158,
      "learning_rate": 1.065367617503621e-07,
      "loss": 0.6897,
      "step": 32490
    },
    {
      "epoch": 0.9361781824468391,
      "grad_norm": 0.7196863293647766,
      "learning_rate": 1.064409736696087e-07,
      "loss": 0.6688,
      "step": 32491
    },
    {
      "epoch": 0.9362069959084884,
      "grad_norm": 0.7202909588813782,
      "learning_rate": 1.0634522820686166e-07,
      "loss": 0.6862,
      "step": 32492
    },
    {
      "epoch": 0.9362358093701377,
      "grad_norm": 0.7292294502258301,
      "learning_rate": 1.0624952536295641e-07,
      "loss": 0.6713,
      "step": 32493
    },
    {
      "epoch": 0.936264622831787,
      "grad_norm": 0.7237583994865417,
      "learning_rate": 1.0615386513872506e-07,
      "loss": 0.6551,
      "step": 32494
    },
    {
      "epoch": 0.9362934362934363,
      "grad_norm": 0.709737241268158,
      "learning_rate": 1.0605824753500249e-07,
      "loss": 0.6875,
      "step": 32495
    },
    {
      "epoch": 0.9363222497550856,
      "grad_norm": 0.7250134944915771,
      "learning_rate": 1.0596267255261972e-07,
      "loss": 0.6827,
      "step": 32496
    },
    {
      "epoch": 0.9363510632167349,
      "grad_norm": 0.7195413708686829,
      "learning_rate": 1.058671401924105e-07,
      "loss": 0.6764,
      "step": 32497
    },
    {
      "epoch": 0.9363798766783842,
      "grad_norm": 0.749483585357666,
      "learning_rate": 1.0577165045520532e-07,
      "loss": 0.6697,
      "step": 32498
    },
    {
      "epoch": 0.9364086901400335,
      "grad_norm": 0.7620181441307068,
      "learning_rate": 1.0567620334183737e-07,
      "loss": 0.6772,
      "step": 32499
    },
    {
      "epoch": 0.9364375036016827,
      "grad_norm": 0.7139313220977783,
      "learning_rate": 1.0558079885313654e-07,
      "loss": 0.6869,
      "step": 32500
    },
    {
      "epoch": 0.9364663170633319,
      "grad_norm": 0.7101473212242126,
      "learning_rate": 1.0548543698993441e-07,
      "loss": 0.686,
      "step": 32501
    },
    {
      "epoch": 0.9364951305249812,
      "grad_norm": 0.7125884890556335,
      "learning_rate": 1.053901177530614e-07,
      "loss": 0.673,
      "step": 32502
    },
    {
      "epoch": 0.9365239439866305,
      "grad_norm": 0.7281217575073242,
      "learning_rate": 1.0529484114334797e-07,
      "loss": 0.6998,
      "step": 32503
    },
    {
      "epoch": 0.9365527574482798,
      "grad_norm": 0.7184041738510132,
      "learning_rate": 1.0519960716162347e-07,
      "loss": 0.6812,
      "step": 32504
    },
    {
      "epoch": 0.9365815709099291,
      "grad_norm": 0.7286884784698486,
      "learning_rate": 1.0510441580871667e-07,
      "loss": 0.6865,
      "step": 32505
    },
    {
      "epoch": 0.9366103843715784,
      "grad_norm": 0.7308574318885803,
      "learning_rate": 1.0500926708545856e-07,
      "loss": 0.6704,
      "step": 32506
    },
    {
      "epoch": 0.9366391978332277,
      "grad_norm": 0.7404448390007019,
      "learning_rate": 1.049141609926757e-07,
      "loss": 0.6684,
      "step": 32507
    },
    {
      "epoch": 0.936668011294877,
      "grad_norm": 0.7183594107627869,
      "learning_rate": 1.0481909753119746e-07,
      "loss": 0.6477,
      "step": 32508
    },
    {
      "epoch": 0.9366968247565263,
      "grad_norm": 0.7375451922416687,
      "learning_rate": 1.0472407670185148e-07,
      "loss": 0.7113,
      "step": 32509
    },
    {
      "epoch": 0.9367256382181756,
      "grad_norm": 0.7143509387969971,
      "learning_rate": 1.0462909850546543e-07,
      "loss": 0.6742,
      "step": 32510
    },
    {
      "epoch": 0.9367544516798249,
      "grad_norm": 0.739930272102356,
      "learning_rate": 1.0453416294286589e-07,
      "loss": 0.6787,
      "step": 32511
    },
    {
      "epoch": 0.9367832651414741,
      "grad_norm": 0.7100959420204163,
      "learning_rate": 1.0443927001488053e-07,
      "loss": 0.6702,
      "step": 32512
    },
    {
      "epoch": 0.9368120786031234,
      "grad_norm": 0.7097219228744507,
      "learning_rate": 1.0434441972233478e-07,
      "loss": 0.663,
      "step": 32513
    },
    {
      "epoch": 0.9368408920647726,
      "grad_norm": 0.7161419987678528,
      "learning_rate": 1.0424961206605633e-07,
      "loss": 0.692,
      "step": 32514
    },
    {
      "epoch": 0.9368697055264219,
      "grad_norm": 0.7243920564651489,
      "learning_rate": 1.0415484704686896e-07,
      "loss": 0.664,
      "step": 32515
    },
    {
      "epoch": 0.9368985189880712,
      "grad_norm": 0.7402157187461853,
      "learning_rate": 1.0406012466559922e-07,
      "loss": 0.6852,
      "step": 32516
    },
    {
      "epoch": 0.9369273324497205,
      "grad_norm": 0.7087120413780212,
      "learning_rate": 1.0396544492307148e-07,
      "loss": 0.686,
      "step": 32517
    },
    {
      "epoch": 0.9369561459113698,
      "grad_norm": 0.7309972047805786,
      "learning_rate": 1.038708078201106e-07,
      "loss": 0.6812,
      "step": 32518
    },
    {
      "epoch": 0.9369849593730191,
      "grad_norm": 0.711872935295105,
      "learning_rate": 1.037762133575404e-07,
      "loss": 0.6758,
      "step": 32519
    },
    {
      "epoch": 0.9370137728346684,
      "grad_norm": 0.7286151647567749,
      "learning_rate": 1.036816615361852e-07,
      "loss": 0.6839,
      "step": 32520
    },
    {
      "epoch": 0.9370425862963176,
      "grad_norm": 0.7398839592933655,
      "learning_rate": 1.0358715235686767e-07,
      "loss": 0.6872,
      "step": 32521
    },
    {
      "epoch": 0.9370713997579669,
      "grad_norm": 0.7182687520980835,
      "learning_rate": 1.0349268582041161e-07,
      "loss": 0.6826,
      "step": 32522
    },
    {
      "epoch": 0.9371002132196162,
      "grad_norm": 0.6947850584983826,
      "learning_rate": 1.0339826192764025e-07,
      "loss": 0.6576,
      "step": 32523
    },
    {
      "epoch": 0.9371290266812655,
      "grad_norm": 0.7424213886260986,
      "learning_rate": 1.0330388067937458e-07,
      "loss": 0.6967,
      "step": 32524
    },
    {
      "epoch": 0.9371578401429148,
      "grad_norm": 0.7066146731376648,
      "learning_rate": 1.032095420764373e-07,
      "loss": 0.6813,
      "step": 32525
    },
    {
      "epoch": 0.9371866536045641,
      "grad_norm": 0.7176642417907715,
      "learning_rate": 1.0311524611964995e-07,
      "loss": 0.6767,
      "step": 32526
    },
    {
      "epoch": 0.9372154670662133,
      "grad_norm": 0.72044837474823,
      "learning_rate": 1.0302099280983358e-07,
      "loss": 0.6709,
      "step": 32527
    },
    {
      "epoch": 0.9372442805278626,
      "grad_norm": 0.7331618666648865,
      "learning_rate": 1.0292678214780915e-07,
      "loss": 0.6881,
      "step": 32528
    },
    {
      "epoch": 0.9372730939895119,
      "grad_norm": 0.7241266369819641,
      "learning_rate": 1.028326141343977e-07,
      "loss": 0.6787,
      "step": 32529
    },
    {
      "epoch": 0.9373019074511612,
      "grad_norm": 0.7040556073188782,
      "learning_rate": 1.0273848877041804e-07,
      "loss": 0.6501,
      "step": 32530
    },
    {
      "epoch": 0.9373307209128104,
      "grad_norm": 0.7291517853736877,
      "learning_rate": 1.0264440605669168e-07,
      "loss": 0.6926,
      "step": 32531
    },
    {
      "epoch": 0.9373595343744597,
      "grad_norm": 0.7156556844711304,
      "learning_rate": 1.0255036599403578e-07,
      "loss": 0.6774,
      "step": 32532
    },
    {
      "epoch": 0.937388347836109,
      "grad_norm": 0.7177504301071167,
      "learning_rate": 1.024563685832719e-07,
      "loss": 0.6595,
      "step": 32533
    },
    {
      "epoch": 0.9374171612977583,
      "grad_norm": 0.7191804647445679,
      "learning_rate": 1.0236241382521605e-07,
      "loss": 0.6956,
      "step": 32534
    },
    {
      "epoch": 0.9374459747594076,
      "grad_norm": 0.7268571257591248,
      "learning_rate": 1.022685017206887e-07,
      "loss": 0.6702,
      "step": 32535
    },
    {
      "epoch": 0.9374747882210569,
      "grad_norm": 0.6984951496124268,
      "learning_rate": 1.0217463227050584e-07,
      "loss": 0.691,
      "step": 32536
    },
    {
      "epoch": 0.9375036016827062,
      "grad_norm": 0.7267343401908875,
      "learning_rate": 1.0208080547548627e-07,
      "loss": 0.6746,
      "step": 32537
    },
    {
      "epoch": 0.9375324151443555,
      "grad_norm": 0.7182497978210449,
      "learning_rate": 1.0198702133644655e-07,
      "loss": 0.6656,
      "step": 32538
    },
    {
      "epoch": 0.9375612286060048,
      "grad_norm": 0.7025377750396729,
      "learning_rate": 1.0189327985420438e-07,
      "loss": 0.6582,
      "step": 32539
    },
    {
      "epoch": 0.9375900420676541,
      "grad_norm": 0.7318335175514221,
      "learning_rate": 1.0179958102957466e-07,
      "loss": 0.6947,
      "step": 32540
    },
    {
      "epoch": 0.9376188555293032,
      "grad_norm": 0.7250258922576904,
      "learning_rate": 1.0170592486337449e-07,
      "loss": 0.6885,
      "step": 32541
    },
    {
      "epoch": 0.9376476689909525,
      "grad_norm": 0.7260282039642334,
      "learning_rate": 1.0161231135641935e-07,
      "loss": 0.6847,
      "step": 32542
    },
    {
      "epoch": 0.9376764824526018,
      "grad_norm": 0.7199572324752808,
      "learning_rate": 1.0151874050952415e-07,
      "loss": 0.6725,
      "step": 32543
    },
    {
      "epoch": 0.9377052959142511,
      "grad_norm": 0.7090157866477966,
      "learning_rate": 1.0142521232350433e-07,
      "loss": 0.6842,
      "step": 32544
    },
    {
      "epoch": 0.9377341093759004,
      "grad_norm": 0.7322820425033569,
      "learning_rate": 1.0133172679917369e-07,
      "loss": 0.7079,
      "step": 32545
    },
    {
      "epoch": 0.9377629228375497,
      "grad_norm": 0.7249948382377625,
      "learning_rate": 1.0123828393734714e-07,
      "loss": 0.6669,
      "step": 32546
    },
    {
      "epoch": 0.937791736299199,
      "grad_norm": 0.7221196293830872,
      "learning_rate": 1.011448837388379e-07,
      "loss": 0.7066,
      "step": 32547
    },
    {
      "epoch": 0.9378205497608483,
      "grad_norm": 0.7164840698242188,
      "learning_rate": 1.0105152620446034e-07,
      "loss": 0.6886,
      "step": 32548
    },
    {
      "epoch": 0.9378493632224976,
      "grad_norm": 0.7243072986602783,
      "learning_rate": 1.0095821133502548e-07,
      "loss": 0.659,
      "step": 32549
    },
    {
      "epoch": 0.9378781766841469,
      "grad_norm": 0.7230599522590637,
      "learning_rate": 1.0086493913134876e-07,
      "loss": 0.691,
      "step": 32550
    },
    {
      "epoch": 0.9379069901457961,
      "grad_norm": 0.7076433300971985,
      "learning_rate": 1.007717095942401e-07,
      "loss": 0.6673,
      "step": 32551
    },
    {
      "epoch": 0.9379358036074454,
      "grad_norm": 0.7190713286399841,
      "learning_rate": 1.0067852272451328e-07,
      "loss": 0.7003,
      "step": 32552
    },
    {
      "epoch": 0.9379646170690947,
      "grad_norm": 0.7306585907936096,
      "learning_rate": 1.0058537852297822e-07,
      "loss": 0.6987,
      "step": 32553
    },
    {
      "epoch": 0.9379934305307439,
      "grad_norm": 0.7179204225540161,
      "learning_rate": 1.0049227699044761e-07,
      "loss": 0.6744,
      "step": 32554
    },
    {
      "epoch": 0.9380222439923932,
      "grad_norm": 0.7008014917373657,
      "learning_rate": 1.0039921812773079e-07,
      "loss": 0.6935,
      "step": 32555
    },
    {
      "epoch": 0.9380510574540425,
      "grad_norm": 0.7185585498809814,
      "learning_rate": 1.0030620193563989e-07,
      "loss": 0.6898,
      "step": 32556
    },
    {
      "epoch": 0.9380798709156918,
      "grad_norm": 0.7234086394309998,
      "learning_rate": 1.0021322841498315e-07,
      "loss": 0.6918,
      "step": 32557
    },
    {
      "epoch": 0.9381086843773411,
      "grad_norm": 0.7136228084564209,
      "learning_rate": 1.0012029756657215e-07,
      "loss": 0.6804,
      "step": 32558
    },
    {
      "epoch": 0.9381374978389904,
      "grad_norm": 0.7270633578300476,
      "learning_rate": 1.0002740939121458e-07,
      "loss": 0.6675,
      "step": 32559
    },
    {
      "epoch": 0.9381663113006397,
      "grad_norm": 0.7326726913452148,
      "learning_rate": 9.993456388971979e-08,
      "loss": 0.6778,
      "step": 32560
    },
    {
      "epoch": 0.938195124762289,
      "grad_norm": 0.7095592617988586,
      "learning_rate": 9.984176106289711e-08,
      "loss": 0.688,
      "step": 32561
    },
    {
      "epoch": 0.9382239382239382,
      "grad_norm": 0.723297119140625,
      "learning_rate": 9.974900091155425e-08,
      "loss": 0.7179,
      "step": 32562
    },
    {
      "epoch": 0.9382527516855875,
      "grad_norm": 0.7263253927230835,
      "learning_rate": 9.965628343649947e-08,
      "loss": 0.6826,
      "step": 32563
    },
    {
      "epoch": 0.9382815651472368,
      "grad_norm": 0.7407798171043396,
      "learning_rate": 9.956360863853986e-08,
      "loss": 0.6823,
      "step": 32564
    },
    {
      "epoch": 0.9383103786088861,
      "grad_norm": 0.7434491515159607,
      "learning_rate": 9.947097651848203e-08,
      "loss": 0.6962,
      "step": 32565
    },
    {
      "epoch": 0.9383391920705354,
      "grad_norm": 0.7010563015937805,
      "learning_rate": 9.937838707713366e-08,
      "loss": 0.6868,
      "step": 32566
    },
    {
      "epoch": 0.9383680055321847,
      "grad_norm": 0.7348949909210205,
      "learning_rate": 9.928584031530076e-08,
      "loss": 0.6747,
      "step": 32567
    },
    {
      "epoch": 0.9383968189938339,
      "grad_norm": 0.7291305661201477,
      "learning_rate": 9.919333623378879e-08,
      "loss": 0.6919,
      "step": 32568
    },
    {
      "epoch": 0.9384256324554832,
      "grad_norm": 0.7049048542976379,
      "learning_rate": 9.910087483340436e-08,
      "loss": 0.689,
      "step": 32569
    },
    {
      "epoch": 0.9384544459171325,
      "grad_norm": 0.7600265741348267,
      "learning_rate": 9.90084561149518e-08,
      "loss": 0.6803,
      "step": 32570
    },
    {
      "epoch": 0.9384832593787817,
      "grad_norm": 0.7156667709350586,
      "learning_rate": 9.891608007923714e-08,
      "loss": 0.6887,
      "step": 32571
    },
    {
      "epoch": 0.938512072840431,
      "grad_norm": 0.7571211457252502,
      "learning_rate": 9.882374672706362e-08,
      "loss": 0.6773,
      "step": 32572
    },
    {
      "epoch": 0.9385408863020803,
      "grad_norm": 0.7157448530197144,
      "learning_rate": 9.873145605923562e-08,
      "loss": 0.6754,
      "step": 32573
    },
    {
      "epoch": 0.9385696997637296,
      "grad_norm": 0.7084189653396606,
      "learning_rate": 9.863920807655747e-08,
      "loss": 0.6775,
      "step": 32574
    },
    {
      "epoch": 0.9385985132253789,
      "grad_norm": 0.7175658345222473,
      "learning_rate": 9.854700277983242e-08,
      "loss": 0.6673,
      "step": 32575
    },
    {
      "epoch": 0.9386273266870282,
      "grad_norm": 0.7133278846740723,
      "learning_rate": 9.845484016986262e-08,
      "loss": 0.6341,
      "step": 32576
    },
    {
      "epoch": 0.9386561401486775,
      "grad_norm": 0.7171510457992554,
      "learning_rate": 9.836272024745241e-08,
      "loss": 0.663,
      "step": 32577
    },
    {
      "epoch": 0.9386849536103268,
      "grad_norm": 0.7377920150756836,
      "learning_rate": 9.827064301340228e-08,
      "loss": 0.6544,
      "step": 32578
    },
    {
      "epoch": 0.9387137670719761,
      "grad_norm": 0.7336946129798889,
      "learning_rate": 9.817860846851546e-08,
      "loss": 0.6776,
      "step": 32579
    },
    {
      "epoch": 0.9387425805336254,
      "grad_norm": 0.7296140193939209,
      "learning_rate": 9.808661661359242e-08,
      "loss": 0.6666,
      "step": 32580
    },
    {
      "epoch": 0.9387713939952745,
      "grad_norm": 0.7179545760154724,
      "learning_rate": 9.799466744943475e-08,
      "loss": 0.6964,
      "step": 32581
    },
    {
      "epoch": 0.9388002074569238,
      "grad_norm": 0.7501964569091797,
      "learning_rate": 9.790276097684403e-08,
      "loss": 0.6892,
      "step": 32582
    },
    {
      "epoch": 0.9388290209185731,
      "grad_norm": 0.7286567687988281,
      "learning_rate": 9.781089719661963e-08,
      "loss": 0.6845,
      "step": 32583
    },
    {
      "epoch": 0.9388578343802224,
      "grad_norm": 0.7219386696815491,
      "learning_rate": 9.771907610956199e-08,
      "loss": 0.6852,
      "step": 32584
    },
    {
      "epoch": 0.9388866478418717,
      "grad_norm": 0.7155758738517761,
      "learning_rate": 9.76272977164705e-08,
      "loss": 0.6787,
      "step": 32585
    },
    {
      "epoch": 0.938915461303521,
      "grad_norm": 0.7104179859161377,
      "learning_rate": 9.753556201814451e-08,
      "loss": 0.6598,
      "step": 32586
    },
    {
      "epoch": 0.9389442747651703,
      "grad_norm": 0.7409857511520386,
      "learning_rate": 9.744386901538339e-08,
      "loss": 0.6871,
      "step": 32587
    },
    {
      "epoch": 0.9389730882268196,
      "grad_norm": 0.7132978439331055,
      "learning_rate": 9.735221870898536e-08,
      "loss": 0.652,
      "step": 32588
    },
    {
      "epoch": 0.9390019016884689,
      "grad_norm": 0.7171538472175598,
      "learning_rate": 9.726061109974871e-08,
      "loss": 0.7043,
      "step": 32589
    },
    {
      "epoch": 0.9390307151501182,
      "grad_norm": 0.7210583686828613,
      "learning_rate": 9.716904618847111e-08,
      "loss": 0.6689,
      "step": 32590
    },
    {
      "epoch": 0.9390595286117674,
      "grad_norm": 0.7328103184700012,
      "learning_rate": 9.707752397594972e-08,
      "loss": 0.6824,
      "step": 32591
    },
    {
      "epoch": 0.9390883420734167,
      "grad_norm": 0.7370361685752869,
      "learning_rate": 9.698604446298276e-08,
      "loss": 0.6639,
      "step": 32592
    },
    {
      "epoch": 0.939117155535066,
      "grad_norm": 0.7150962948799133,
      "learning_rate": 9.689460765036574e-08,
      "loss": 0.6748,
      "step": 32593
    },
    {
      "epoch": 0.9391459689967153,
      "grad_norm": 0.7481749057769775,
      "learning_rate": 9.680321353889577e-08,
      "loss": 0.678,
      "step": 32594
    },
    {
      "epoch": 0.9391747824583645,
      "grad_norm": 0.7185974717140198,
      "learning_rate": 9.671186212936778e-08,
      "loss": 0.6927,
      "step": 32595
    },
    {
      "epoch": 0.9392035959200138,
      "grad_norm": 0.7237516045570374,
      "learning_rate": 9.662055342257837e-08,
      "loss": 0.7063,
      "step": 32596
    },
    {
      "epoch": 0.9392324093816631,
      "grad_norm": 0.739845871925354,
      "learning_rate": 9.652928741932133e-08,
      "loss": 0.6883,
      "step": 32597
    },
    {
      "epoch": 0.9392612228433124,
      "grad_norm": 0.7109028100967407,
      "learning_rate": 9.643806412039381e-08,
      "loss": 0.6568,
      "step": 32598
    },
    {
      "epoch": 0.9392900363049617,
      "grad_norm": 0.7269094586372375,
      "learning_rate": 9.634688352658849e-08,
      "loss": 0.6867,
      "step": 32599
    },
    {
      "epoch": 0.939318849766611,
      "grad_norm": 0.7073675394058228,
      "learning_rate": 9.625574563870032e-08,
      "loss": 0.6946,
      "step": 32600
    },
    {
      "epoch": 0.9393476632282602,
      "grad_norm": 0.7214757800102234,
      "learning_rate": 9.616465045752199e-08,
      "loss": 0.6708,
      "step": 32601
    },
    {
      "epoch": 0.9393764766899095,
      "grad_norm": 0.7303779721260071,
      "learning_rate": 9.607359798384785e-08,
      "loss": 0.6826,
      "step": 32602
    },
    {
      "epoch": 0.9394052901515588,
      "grad_norm": 0.731158971786499,
      "learning_rate": 9.598258821847062e-08,
      "loss": 0.6707,
      "step": 32603
    },
    {
      "epoch": 0.9394341036132081,
      "grad_norm": 0.716208815574646,
      "learning_rate": 9.589162116218243e-08,
      "loss": 0.6599,
      "step": 32604
    },
    {
      "epoch": 0.9394629170748574,
      "grad_norm": 0.7284276485443115,
      "learning_rate": 9.580069681577653e-08,
      "loss": 0.6784,
      "step": 32605
    },
    {
      "epoch": 0.9394917305365067,
      "grad_norm": 0.7158249616622925,
      "learning_rate": 9.570981518004397e-08,
      "loss": 0.6803,
      "step": 32606
    },
    {
      "epoch": 0.939520543998156,
      "grad_norm": 0.7313183546066284,
      "learning_rate": 9.561897625577633e-08,
      "loss": 0.6755,
      "step": 32607
    },
    {
      "epoch": 0.9395493574598052,
      "grad_norm": 0.7095748782157898,
      "learning_rate": 9.552818004376407e-08,
      "loss": 0.7089,
      "step": 32608
    },
    {
      "epoch": 0.9395781709214545,
      "grad_norm": 0.7313030362129211,
      "learning_rate": 9.543742654479992e-08,
      "loss": 0.6948,
      "step": 32609
    },
    {
      "epoch": 0.9396069843831037,
      "grad_norm": 0.7313181161880493,
      "learning_rate": 9.534671575967214e-08,
      "loss": 0.7029,
      "step": 32610
    },
    {
      "epoch": 0.939635797844753,
      "grad_norm": 0.7279845476150513,
      "learning_rate": 9.525604768917229e-08,
      "loss": 0.7111,
      "step": 32611
    },
    {
      "epoch": 0.9396646113064023,
      "grad_norm": 0.7254066467285156,
      "learning_rate": 9.516542233408921e-08,
      "loss": 0.6809,
      "step": 32612
    },
    {
      "epoch": 0.9396934247680516,
      "grad_norm": 0.7294055223464966,
      "learning_rate": 9.507483969521225e-08,
      "loss": 0.6661,
      "step": 32613
    },
    {
      "epoch": 0.9397222382297009,
      "grad_norm": 0.7071437835693359,
      "learning_rate": 9.498429977332968e-08,
      "loss": 0.6608,
      "step": 32614
    },
    {
      "epoch": 0.9397510516913502,
      "grad_norm": 0.715399980545044,
      "learning_rate": 9.489380256923142e-08,
      "loss": 0.6377,
      "step": 32615
    },
    {
      "epoch": 0.9397798651529995,
      "grad_norm": 0.7321871519088745,
      "learning_rate": 9.480334808370406e-08,
      "loss": 0.7004,
      "step": 32616
    },
    {
      "epoch": 0.9398086786146488,
      "grad_norm": 0.7166742086410522,
      "learning_rate": 9.471293631753697e-08,
      "loss": 0.6642,
      "step": 32617
    },
    {
      "epoch": 0.9398374920762981,
      "grad_norm": 0.7093321084976196,
      "learning_rate": 9.46225672715162e-08,
      "loss": 0.6596,
      "step": 32618
    },
    {
      "epoch": 0.9398663055379474,
      "grad_norm": 0.7206981778144836,
      "learning_rate": 9.453224094642943e-08,
      "loss": 0.659,
      "step": 32619
    },
    {
      "epoch": 0.9398951189995967,
      "grad_norm": 0.7196734547615051,
      "learning_rate": 9.444195734306271e-08,
      "loss": 0.6881,
      "step": 32620
    },
    {
      "epoch": 0.939923932461246,
      "grad_norm": 0.7064023613929749,
      "learning_rate": 9.435171646220376e-08,
      "loss": 0.6689,
      "step": 32621
    },
    {
      "epoch": 0.9399527459228951,
      "grad_norm": 0.7356323599815369,
      "learning_rate": 9.426151830463636e-08,
      "loss": 0.6777,
      "step": 32622
    },
    {
      "epoch": 0.9399815593845444,
      "grad_norm": 0.7313312888145447,
      "learning_rate": 9.417136287114826e-08,
      "loss": 0.6708,
      "step": 32623
    },
    {
      "epoch": 0.9400103728461937,
      "grad_norm": 0.7374277114868164,
      "learning_rate": 9.408125016252267e-08,
      "loss": 0.6867,
      "step": 32624
    },
    {
      "epoch": 0.940039186307843,
      "grad_norm": 0.8774962425231934,
      "learning_rate": 9.399118017954567e-08,
      "loss": 0.7265,
      "step": 32625
    },
    {
      "epoch": 0.9400679997694923,
      "grad_norm": 0.7170376777648926,
      "learning_rate": 9.390115292300161e-08,
      "loss": 0.6792,
      "step": 32626
    },
    {
      "epoch": 0.9400968132311416,
      "grad_norm": 0.711037278175354,
      "learning_rate": 9.381116839367376e-08,
      "loss": 0.6815,
      "step": 32627
    },
    {
      "epoch": 0.9401256266927909,
      "grad_norm": 0.7585611343383789,
      "learning_rate": 9.372122659234651e-08,
      "loss": 0.6736,
      "step": 32628
    },
    {
      "epoch": 0.9401544401544402,
      "grad_norm": 0.6995580792427063,
      "learning_rate": 9.363132751980253e-08,
      "loss": 0.6648,
      "step": 32629
    },
    {
      "epoch": 0.9401832536160895,
      "grad_norm": 0.7248316407203674,
      "learning_rate": 9.354147117682566e-08,
      "loss": 0.7109,
      "step": 32630
    },
    {
      "epoch": 0.9402120670777387,
      "grad_norm": 0.7175160050392151,
      "learning_rate": 9.34516575641975e-08,
      "loss": 0.6881,
      "step": 32631
    },
    {
      "epoch": 0.940240880539388,
      "grad_norm": 0.7238844633102417,
      "learning_rate": 9.336188668270019e-08,
      "loss": 0.6883,
      "step": 32632
    },
    {
      "epoch": 0.9402696940010373,
      "grad_norm": 0.7177978754043579,
      "learning_rate": 9.327215853311644e-08,
      "loss": 0.6839,
      "step": 32633
    },
    {
      "epoch": 0.9402985074626866,
      "grad_norm": 0.7467384338378906,
      "learning_rate": 9.318247311622785e-08,
      "loss": 0.7204,
      "step": 32634
    },
    {
      "epoch": 0.9403273209243358,
      "grad_norm": 0.7243945002555847,
      "learning_rate": 9.30928304328138e-08,
      "loss": 0.6947,
      "step": 32635
    },
    {
      "epoch": 0.9403561343859851,
      "grad_norm": 0.7254441976547241,
      "learning_rate": 9.300323048365701e-08,
      "loss": 0.6625,
      "step": 32636
    },
    {
      "epoch": 0.9403849478476344,
      "grad_norm": 0.7199514508247375,
      "learning_rate": 9.291367326953626e-08,
      "loss": 0.6693,
      "step": 32637
    },
    {
      "epoch": 0.9404137613092837,
      "grad_norm": 0.7197757363319397,
      "learning_rate": 9.282415879123318e-08,
      "loss": 0.6983,
      "step": 32638
    },
    {
      "epoch": 0.940442574770933,
      "grad_norm": 0.727324903011322,
      "learning_rate": 9.273468704952492e-08,
      "loss": 0.693,
      "step": 32639
    },
    {
      "epoch": 0.9404713882325823,
      "grad_norm": 0.7230806946754456,
      "learning_rate": 9.264525804519253e-08,
      "loss": 0.6734,
      "step": 32640
    },
    {
      "epoch": 0.9405002016942315,
      "grad_norm": 0.7265523076057434,
      "learning_rate": 9.255587177901482e-08,
      "loss": 0.6826,
      "step": 32641
    },
    {
      "epoch": 0.9405290151558808,
      "grad_norm": 0.715547502040863,
      "learning_rate": 9.246652825176949e-08,
      "loss": 0.6699,
      "step": 32642
    },
    {
      "epoch": 0.9405578286175301,
      "grad_norm": 0.6990086436271667,
      "learning_rate": 9.237722746423538e-08,
      "loss": 0.6889,
      "step": 32643
    },
    {
      "epoch": 0.9405866420791794,
      "grad_norm": 0.7336360216140747,
      "learning_rate": 9.228796941718909e-08,
      "loss": 0.7004,
      "step": 32644
    },
    {
      "epoch": 0.9406154555408287,
      "grad_norm": 0.7377986907958984,
      "learning_rate": 9.219875411140888e-08,
      "loss": 0.6911,
      "step": 32645
    },
    {
      "epoch": 0.940644269002478,
      "grad_norm": 0.7148873209953308,
      "learning_rate": 9.210958154767136e-08,
      "loss": 0.6812,
      "step": 32646
    },
    {
      "epoch": 0.9406730824641273,
      "grad_norm": 0.7270076870918274,
      "learning_rate": 9.202045172675367e-08,
      "loss": 0.6742,
      "step": 32647
    },
    {
      "epoch": 0.9407018959257766,
      "grad_norm": 0.7368793487548828,
      "learning_rate": 9.193136464943186e-08,
      "loss": 0.6797,
      "step": 32648
    },
    {
      "epoch": 0.9407307093874258,
      "grad_norm": 0.730563759803772,
      "learning_rate": 9.184232031648144e-08,
      "loss": 0.7043,
      "step": 32649
    },
    {
      "epoch": 0.940759522849075,
      "grad_norm": 0.7196224331855774,
      "learning_rate": 9.175331872867731e-08,
      "loss": 0.6842,
      "step": 32650
    },
    {
      "epoch": 0.9407883363107243,
      "grad_norm": 0.7181499004364014,
      "learning_rate": 9.166435988679667e-08,
      "loss": 0.6781,
      "step": 32651
    },
    {
      "epoch": 0.9408171497723736,
      "grad_norm": 0.7223621606826782,
      "learning_rate": 9.157544379161165e-08,
      "loss": 0.6668,
      "step": 32652
    },
    {
      "epoch": 0.9408459632340229,
      "grad_norm": 0.7555662393569946,
      "learning_rate": 9.148657044389886e-08,
      "loss": 0.6983,
      "step": 32653
    },
    {
      "epoch": 0.9408747766956722,
      "grad_norm": 0.7184902429580688,
      "learning_rate": 9.139773984443045e-08,
      "loss": 0.6634,
      "step": 32654
    },
    {
      "epoch": 0.9409035901573215,
      "grad_norm": 0.7362259030342102,
      "learning_rate": 9.130895199398194e-08,
      "loss": 0.7091,
      "step": 32655
    },
    {
      "epoch": 0.9409324036189708,
      "grad_norm": 0.7532487511634827,
      "learning_rate": 9.122020689332434e-08,
      "loss": 0.7118,
      "step": 32656
    },
    {
      "epoch": 0.9409612170806201,
      "grad_norm": 0.7510181665420532,
      "learning_rate": 9.11315045432326e-08,
      "loss": 0.7,
      "step": 32657
    },
    {
      "epoch": 0.9409900305422694,
      "grad_norm": 0.7404136657714844,
      "learning_rate": 9.10428449444778e-08,
      "loss": 0.6795,
      "step": 32658
    },
    {
      "epoch": 0.9410188440039187,
      "grad_norm": 0.7347903847694397,
      "learning_rate": 9.095422809783317e-08,
      "loss": 0.6944,
      "step": 32659
    },
    {
      "epoch": 0.941047657465568,
      "grad_norm": 0.7398279309272766,
      "learning_rate": 9.086565400406977e-08,
      "loss": 0.674,
      "step": 32660
    },
    {
      "epoch": 0.9410764709272172,
      "grad_norm": 0.745964527130127,
      "learning_rate": 9.077712266395922e-08,
      "loss": 0.6837,
      "step": 32661
    },
    {
      "epoch": 0.9411052843888664,
      "grad_norm": 0.7294275164604187,
      "learning_rate": 9.068863407827255e-08,
      "loss": 0.6928,
      "step": 32662
    },
    {
      "epoch": 0.9411340978505157,
      "grad_norm": 0.7187124490737915,
      "learning_rate": 9.060018824778027e-08,
      "loss": 0.6835,
      "step": 32663
    },
    {
      "epoch": 0.941162911312165,
      "grad_norm": 0.7231147885322571,
      "learning_rate": 9.051178517325287e-08,
      "loss": 0.6712,
      "step": 32664
    },
    {
      "epoch": 0.9411917247738143,
      "grad_norm": 0.7477354407310486,
      "learning_rate": 9.042342485545974e-08,
      "loss": 0.7006,
      "step": 32665
    },
    {
      "epoch": 0.9412205382354636,
      "grad_norm": 0.7305424809455872,
      "learning_rate": 9.033510729517137e-08,
      "loss": 0.7129,
      "step": 32666
    },
    {
      "epoch": 0.9412493516971129,
      "grad_norm": 0.7338594794273376,
      "learning_rate": 9.024683249315603e-08,
      "loss": 0.6841,
      "step": 32667
    },
    {
      "epoch": 0.9412781651587622,
      "grad_norm": 0.7312831878662109,
      "learning_rate": 9.015860045018254e-08,
      "loss": 0.6883,
      "step": 32668
    },
    {
      "epoch": 0.9413069786204115,
      "grad_norm": 0.7195521593093872,
      "learning_rate": 9.007041116701975e-08,
      "loss": 0.6895,
      "step": 32669
    },
    {
      "epoch": 0.9413357920820608,
      "grad_norm": 0.7265240550041199,
      "learning_rate": 8.998226464443594e-08,
      "loss": 0.6883,
      "step": 32670
    },
    {
      "epoch": 0.94136460554371,
      "grad_norm": 0.7241864204406738,
      "learning_rate": 8.989416088319825e-08,
      "loss": 0.7138,
      "step": 32671
    },
    {
      "epoch": 0.9413934190053593,
      "grad_norm": 3.6082499027252197,
      "learning_rate": 8.98060998840744e-08,
      "loss": 0.6842,
      "step": 32672
    },
    {
      "epoch": 0.9414222324670086,
      "grad_norm": 0.7529907822608948,
      "learning_rate": 8.971808164783046e-08,
      "loss": 0.7361,
      "step": 32673
    },
    {
      "epoch": 0.9414510459286579,
      "grad_norm": 0.7106491327285767,
      "learning_rate": 8.96301061752336e-08,
      "loss": 0.68,
      "step": 32674
    },
    {
      "epoch": 0.9414798593903072,
      "grad_norm": 0.7355031371116638,
      "learning_rate": 8.954217346704985e-08,
      "loss": 0.6834,
      "step": 32675
    },
    {
      "epoch": 0.9415086728519564,
      "grad_norm": 0.7235737442970276,
      "learning_rate": 8.945428352404583e-08,
      "loss": 0.6798,
      "step": 32676
    },
    {
      "epoch": 0.9415374863136057,
      "grad_norm": 0.71512770652771,
      "learning_rate": 8.936643634698538e-08,
      "loss": 0.6769,
      "step": 32677
    },
    {
      "epoch": 0.941566299775255,
      "grad_norm": 0.7254632115364075,
      "learning_rate": 8.927863193663455e-08,
      "loss": 0.6814,
      "step": 32678
    },
    {
      "epoch": 0.9415951132369043,
      "grad_norm": 0.7079590559005737,
      "learning_rate": 8.919087029375883e-08,
      "loss": 0.6941,
      "step": 32679
    },
    {
      "epoch": 0.9416239266985535,
      "grad_norm": 0.7345137596130371,
      "learning_rate": 8.91031514191204e-08,
      "loss": 0.6761,
      "step": 32680
    },
    {
      "epoch": 0.9416527401602028,
      "grad_norm": 0.7337262034416199,
      "learning_rate": 8.901547531348476e-08,
      "loss": 0.6669,
      "step": 32681
    },
    {
      "epoch": 0.9416815536218521,
      "grad_norm": 0.7199419140815735,
      "learning_rate": 8.892784197761572e-08,
      "loss": 0.6645,
      "step": 32682
    },
    {
      "epoch": 0.9417103670835014,
      "grad_norm": 0.7336764931678772,
      "learning_rate": 8.884025141227548e-08,
      "loss": 0.6846,
      "step": 32683
    },
    {
      "epoch": 0.9417391805451507,
      "grad_norm": 0.693731427192688,
      "learning_rate": 8.87527036182273e-08,
      "loss": 0.6444,
      "step": 32684
    },
    {
      "epoch": 0.9417679940068,
      "grad_norm": 0.7099249362945557,
      "learning_rate": 8.86651985962339e-08,
      "loss": 0.6818,
      "step": 32685
    },
    {
      "epoch": 0.9417968074684493,
      "grad_norm": 0.7304353713989258,
      "learning_rate": 8.857773634705635e-08,
      "loss": 0.6698,
      "step": 32686
    },
    {
      "epoch": 0.9418256209300986,
      "grad_norm": 0.7246975302696228,
      "learning_rate": 8.849031687145792e-08,
      "loss": 0.6769,
      "step": 32687
    },
    {
      "epoch": 0.9418544343917479,
      "grad_norm": 0.7110033631324768,
      "learning_rate": 8.840294017019857e-08,
      "loss": 0.6751,
      "step": 32688
    },
    {
      "epoch": 0.941883247853397,
      "grad_norm": 0.7501311302185059,
      "learning_rate": 8.831560624404046e-08,
      "loss": 0.6957,
      "step": 32689
    },
    {
      "epoch": 0.9419120613150463,
      "grad_norm": 0.7280110120773315,
      "learning_rate": 8.822831509374297e-08,
      "loss": 0.6944,
      "step": 32690
    },
    {
      "epoch": 0.9419408747766956,
      "grad_norm": 0.7291553020477295,
      "learning_rate": 8.814106672006716e-08,
      "loss": 0.6789,
      "step": 32691
    },
    {
      "epoch": 0.9419696882383449,
      "grad_norm": 0.7414475679397583,
      "learning_rate": 8.80538611237719e-08,
      "loss": 0.6955,
      "step": 32692
    },
    {
      "epoch": 0.9419985016999942,
      "grad_norm": 0.7424420118331909,
      "learning_rate": 8.796669830561821e-08,
      "loss": 0.6865,
      "step": 32693
    },
    {
      "epoch": 0.9420273151616435,
      "grad_norm": 0.7139185070991516,
      "learning_rate": 8.787957826636384e-08,
      "loss": 0.6901,
      "step": 32694
    },
    {
      "epoch": 0.9420561286232928,
      "grad_norm": 0.6986365914344788,
      "learning_rate": 8.779250100676872e-08,
      "loss": 0.6713,
      "step": 32695
    },
    {
      "epoch": 0.9420849420849421,
      "grad_norm": 0.7259053587913513,
      "learning_rate": 8.770546652758949e-08,
      "loss": 0.6809,
      "step": 32696
    },
    {
      "epoch": 0.9421137555465914,
      "grad_norm": 0.7333653569221497,
      "learning_rate": 8.761847482958608e-08,
      "loss": 0.695,
      "step": 32697
    },
    {
      "epoch": 0.9421425690082407,
      "grad_norm": 0.7761310935020447,
      "learning_rate": 8.753152591351454e-08,
      "loss": 0.7122,
      "step": 32698
    },
    {
      "epoch": 0.94217138246989,
      "grad_norm": 0.7142261862754822,
      "learning_rate": 8.74446197801332e-08,
      "loss": 0.6695,
      "step": 32699
    },
    {
      "epoch": 0.9422001959315393,
      "grad_norm": 0.7378647327423096,
      "learning_rate": 8.735775643019806e-08,
      "loss": 0.6875,
      "step": 32700
    },
    {
      "epoch": 0.9422290093931885,
      "grad_norm": 0.7152911424636841,
      "learning_rate": 8.727093586446633e-08,
      "loss": 0.6945,
      "step": 32701
    },
    {
      "epoch": 0.9422578228548378,
      "grad_norm": 0.7345799207687378,
      "learning_rate": 8.718415808369351e-08,
      "loss": 0.6755,
      "step": 32702
    },
    {
      "epoch": 0.942286636316487,
      "grad_norm": 0.7250019311904907,
      "learning_rate": 8.709742308863567e-08,
      "loss": 0.6874,
      "step": 32703
    },
    {
      "epoch": 0.9423154497781363,
      "grad_norm": 0.6950902938842773,
      "learning_rate": 8.70107308800483e-08,
      "loss": 0.6541,
      "step": 32704
    },
    {
      "epoch": 0.9423442632397856,
      "grad_norm": 0.7205755710601807,
      "learning_rate": 8.692408145868636e-08,
      "loss": 0.6502,
      "step": 32705
    },
    {
      "epoch": 0.9423730767014349,
      "grad_norm": 0.7206456661224365,
      "learning_rate": 8.683747482530425e-08,
      "loss": 0.6587,
      "step": 32706
    },
    {
      "epoch": 0.9424018901630842,
      "grad_norm": 0.7225560545921326,
      "learning_rate": 8.675091098065635e-08,
      "loss": 0.7132,
      "step": 32707
    },
    {
      "epoch": 0.9424307036247335,
      "grad_norm": 0.7222808003425598,
      "learning_rate": 8.666438992549709e-08,
      "loss": 0.675,
      "step": 32708
    },
    {
      "epoch": 0.9424595170863828,
      "grad_norm": 0.7228133678436279,
      "learning_rate": 8.65779116605786e-08,
      "loss": 0.6881,
      "step": 32709
    },
    {
      "epoch": 0.942488330548032,
      "grad_norm": 0.733566164970398,
      "learning_rate": 8.649147618665533e-08,
      "loss": 0.6734,
      "step": 32710
    },
    {
      "epoch": 0.9425171440096813,
      "grad_norm": 0.7219114303588867,
      "learning_rate": 8.64050835044794e-08,
      "loss": 0.7079,
      "step": 32711
    },
    {
      "epoch": 0.9425459574713306,
      "grad_norm": 0.7148144245147705,
      "learning_rate": 8.631873361480414e-08,
      "loss": 0.6717,
      "step": 32712
    },
    {
      "epoch": 0.9425747709329799,
      "grad_norm": 0.7328910231590271,
      "learning_rate": 8.623242651838005e-08,
      "loss": 0.6892,
      "step": 32713
    },
    {
      "epoch": 0.9426035843946292,
      "grad_norm": 0.7283966541290283,
      "learning_rate": 8.614616221595984e-08,
      "loss": 0.6814,
      "step": 32714
    },
    {
      "epoch": 0.9426323978562785,
      "grad_norm": 0.7253053784370422,
      "learning_rate": 8.605994070829404e-08,
      "loss": 0.6686,
      "step": 32715
    },
    {
      "epoch": 0.9426612113179277,
      "grad_norm": 0.7329664826393127,
      "learning_rate": 8.597376199613427e-08,
      "loss": 0.6807,
      "step": 32716
    },
    {
      "epoch": 0.942690024779577,
      "grad_norm": 0.7141796350479126,
      "learning_rate": 8.588762608023104e-08,
      "loss": 0.7007,
      "step": 32717
    },
    {
      "epoch": 0.9427188382412263,
      "grad_norm": 0.9046240448951721,
      "learning_rate": 8.580153296133431e-08,
      "loss": 0.6529,
      "step": 32718
    },
    {
      "epoch": 0.9427476517028756,
      "grad_norm": 0.7247472405433655,
      "learning_rate": 8.571548264019347e-08,
      "loss": 0.6818,
      "step": 32719
    },
    {
      "epoch": 0.9427764651645248,
      "grad_norm": 0.7362185716629028,
      "learning_rate": 8.562947511755848e-08,
      "loss": 0.6867,
      "step": 32720
    },
    {
      "epoch": 0.9428052786261741,
      "grad_norm": 0.7383981347084045,
      "learning_rate": 8.554351039417819e-08,
      "loss": 0.6889,
      "step": 32721
    },
    {
      "epoch": 0.9428340920878234,
      "grad_norm": 0.7028604745864868,
      "learning_rate": 8.545758847080144e-08,
      "loss": 0.6625,
      "step": 32722
    },
    {
      "epoch": 0.9428629055494727,
      "grad_norm": 0.732133686542511,
      "learning_rate": 8.537170934817596e-08,
      "loss": 0.6888,
      "step": 32723
    },
    {
      "epoch": 0.942891719011122,
      "grad_norm": 0.7196460962295532,
      "learning_rate": 8.528587302705005e-08,
      "loss": 0.6676,
      "step": 32724
    },
    {
      "epoch": 0.9429205324727713,
      "grad_norm": 0.7263713479042053,
      "learning_rate": 8.5200079508172e-08,
      "loss": 0.7104,
      "step": 32725
    },
    {
      "epoch": 0.9429493459344206,
      "grad_norm": 0.7409619092941284,
      "learning_rate": 8.511432879228732e-08,
      "loss": 0.6793,
      "step": 32726
    },
    {
      "epoch": 0.9429781593960699,
      "grad_norm": 0.7643332481384277,
      "learning_rate": 8.502862088014485e-08,
      "loss": 0.6779,
      "step": 32727
    },
    {
      "epoch": 0.9430069728577192,
      "grad_norm": 0.7138436436653137,
      "learning_rate": 8.494295577248845e-08,
      "loss": 0.6773,
      "step": 32728
    },
    {
      "epoch": 0.9430357863193685,
      "grad_norm": 0.7142900228500366,
      "learning_rate": 8.485733347006698e-08,
      "loss": 0.6672,
      "step": 32729
    },
    {
      "epoch": 0.9430645997810176,
      "grad_norm": 0.6967339515686035,
      "learning_rate": 8.477175397362369e-08,
      "loss": 0.6641,
      "step": 32730
    },
    {
      "epoch": 0.9430934132426669,
      "grad_norm": 0.723973274230957,
      "learning_rate": 8.46862172839058e-08,
      "loss": 0.6841,
      "step": 32731
    },
    {
      "epoch": 0.9431222267043162,
      "grad_norm": 0.701234757900238,
      "learning_rate": 8.460072340165715e-08,
      "loss": 0.6681,
      "step": 32732
    },
    {
      "epoch": 0.9431510401659655,
      "grad_norm": 0.7124210596084595,
      "learning_rate": 8.451527232762325e-08,
      "loss": 0.6926,
      "step": 32733
    },
    {
      "epoch": 0.9431798536276148,
      "grad_norm": 0.7061848640441895,
      "learning_rate": 8.442986406254682e-08,
      "loss": 0.6522,
      "step": 32734
    },
    {
      "epoch": 0.9432086670892641,
      "grad_norm": 0.719559371471405,
      "learning_rate": 8.434449860717287e-08,
      "loss": 0.6819,
      "step": 32735
    },
    {
      "epoch": 0.9432374805509134,
      "grad_norm": 0.7478271722793579,
      "learning_rate": 8.425917596224409e-08,
      "loss": 0.6784,
      "step": 32736
    },
    {
      "epoch": 0.9432662940125627,
      "grad_norm": 0.6985231041908264,
      "learning_rate": 8.417389612850435e-08,
      "loss": 0.6657,
      "step": 32737
    },
    {
      "epoch": 0.943295107474212,
      "grad_norm": 0.7147248387336731,
      "learning_rate": 8.408865910669583e-08,
      "loss": 0.6948,
      "step": 32738
    },
    {
      "epoch": 0.9433239209358613,
      "grad_norm": 0.7461055517196655,
      "learning_rate": 8.400346489756073e-08,
      "loss": 0.6772,
      "step": 32739
    },
    {
      "epoch": 0.9433527343975106,
      "grad_norm": 0.7157127857208252,
      "learning_rate": 8.391831350184121e-08,
      "loss": 0.6887,
      "step": 32740
    },
    {
      "epoch": 0.9433815478591598,
      "grad_norm": 0.7289564609527588,
      "learning_rate": 8.383320492027947e-08,
      "loss": 0.6793,
      "step": 32741
    },
    {
      "epoch": 0.9434103613208091,
      "grad_norm": 0.7194316983222961,
      "learning_rate": 8.374813915361546e-08,
      "loss": 0.6809,
      "step": 32742
    },
    {
      "epoch": 0.9434391747824583,
      "grad_norm": 0.7265852689743042,
      "learning_rate": 8.366311620259082e-08,
      "loss": 0.6638,
      "step": 32743
    },
    {
      "epoch": 0.9434679882441076,
      "grad_norm": 0.7132118344306946,
      "learning_rate": 8.357813606794663e-08,
      "loss": 0.6682,
      "step": 32744
    },
    {
      "epoch": 0.9434968017057569,
      "grad_norm": 0.7513648867607117,
      "learning_rate": 8.349319875042173e-08,
      "loss": 0.6977,
      "step": 32745
    },
    {
      "epoch": 0.9435256151674062,
      "grad_norm": 0.721165657043457,
      "learning_rate": 8.340830425075663e-08,
      "loss": 0.6836,
      "step": 32746
    },
    {
      "epoch": 0.9435544286290555,
      "grad_norm": 0.732294499874115,
      "learning_rate": 8.332345256968966e-08,
      "loss": 0.6675,
      "step": 32747
    },
    {
      "epoch": 0.9435832420907048,
      "grad_norm": 0.7152107954025269,
      "learning_rate": 8.32386437079613e-08,
      "loss": 0.671,
      "step": 32748
    },
    {
      "epoch": 0.9436120555523541,
      "grad_norm": 0.73384690284729,
      "learning_rate": 8.315387766630933e-08,
      "loss": 0.6597,
      "step": 32749
    },
    {
      "epoch": 0.9436408690140033,
      "grad_norm": 0.7442483901977539,
      "learning_rate": 8.306915444547203e-08,
      "loss": 0.6929,
      "step": 32750
    },
    {
      "epoch": 0.9436696824756526,
      "grad_norm": 0.7444148659706116,
      "learning_rate": 8.298447404618659e-08,
      "loss": 0.6852,
      "step": 32751
    },
    {
      "epoch": 0.9436984959373019,
      "grad_norm": 0.7083203792572021,
      "learning_rate": 8.289983646919186e-08,
      "loss": 0.6873,
      "step": 32752
    },
    {
      "epoch": 0.9437273093989512,
      "grad_norm": 0.722141444683075,
      "learning_rate": 8.281524171522393e-08,
      "loss": 0.6845,
      "step": 32753
    },
    {
      "epoch": 0.9437561228606005,
      "grad_norm": 0.7319706678390503,
      "learning_rate": 8.273068978501997e-08,
      "loss": 0.6895,
      "step": 32754
    },
    {
      "epoch": 0.9437849363222498,
      "grad_norm": 0.7179974913597107,
      "learning_rate": 8.264618067931607e-08,
      "loss": 0.6549,
      "step": 32755
    },
    {
      "epoch": 0.9438137497838991,
      "grad_norm": 0.7217414379119873,
      "learning_rate": 8.256171439884885e-08,
      "loss": 0.7005,
      "step": 32756
    },
    {
      "epoch": 0.9438425632455483,
      "grad_norm": 0.7213577628135681,
      "learning_rate": 8.247729094435331e-08,
      "loss": 0.6729,
      "step": 32757
    },
    {
      "epoch": 0.9438713767071976,
      "grad_norm": 0.7301923036575317,
      "learning_rate": 8.239291031656494e-08,
      "loss": 0.7024,
      "step": 32758
    },
    {
      "epoch": 0.9439001901688469,
      "grad_norm": 0.760154664516449,
      "learning_rate": 8.230857251621815e-08,
      "loss": 0.6848,
      "step": 32759
    },
    {
      "epoch": 0.9439290036304961,
      "grad_norm": 0.718386709690094,
      "learning_rate": 8.222427754404738e-08,
      "loss": 0.6812,
      "step": 32760
    },
    {
      "epoch": 0.9439578170921454,
      "grad_norm": 0.7345638275146484,
      "learning_rate": 8.214002540078814e-08,
      "loss": 0.6756,
      "step": 32761
    },
    {
      "epoch": 0.9439866305537947,
      "grad_norm": 0.7269642949104309,
      "learning_rate": 8.205581608717262e-08,
      "loss": 0.6955,
      "step": 32762
    },
    {
      "epoch": 0.944015444015444,
      "grad_norm": 0.714468240737915,
      "learning_rate": 8.197164960393522e-08,
      "loss": 0.6598,
      "step": 32763
    },
    {
      "epoch": 0.9440442574770933,
      "grad_norm": 0.7254258394241333,
      "learning_rate": 8.18875259518076e-08,
      "loss": 0.6743,
      "step": 32764
    },
    {
      "epoch": 0.9440730709387426,
      "grad_norm": 0.7232327461242676,
      "learning_rate": 8.180344513152416e-08,
      "loss": 0.6778,
      "step": 32765
    },
    {
      "epoch": 0.9441018844003919,
      "grad_norm": 0.7342329621315002,
      "learning_rate": 8.171940714381599e-08,
      "loss": 0.6693,
      "step": 32766
    },
    {
      "epoch": 0.9441306978620412,
      "grad_norm": 0.7288941144943237,
      "learning_rate": 8.163541198941582e-08,
      "loss": 0.6789,
      "step": 32767
    },
    {
      "epoch": 0.9441595113236905,
      "grad_norm": 0.7311551570892334,
      "learning_rate": 8.15514596690542e-08,
      "loss": 0.6682,
      "step": 32768
    },
    {
      "epoch": 0.9441883247853398,
      "grad_norm": 0.7240151762962341,
      "learning_rate": 8.146755018346331e-08,
      "loss": 0.69,
      "step": 32769
    },
    {
      "epoch": 0.9442171382469889,
      "grad_norm": 0.7147803902626038,
      "learning_rate": 8.138368353337255e-08,
      "loss": 0.6831,
      "step": 32770
    },
    {
      "epoch": 0.9442459517086382,
      "grad_norm": 0.719789445400238,
      "learning_rate": 8.129985971951359e-08,
      "loss": 0.7086,
      "step": 32771
    },
    {
      "epoch": 0.9442747651702875,
      "grad_norm": 0.7201776504516602,
      "learning_rate": 8.121607874261527e-08,
      "loss": 0.6599,
      "step": 32772
    },
    {
      "epoch": 0.9443035786319368,
      "grad_norm": 0.7012708783149719,
      "learning_rate": 8.113234060340869e-08,
      "loss": 0.6781,
      "step": 32773
    },
    {
      "epoch": 0.9443323920935861,
      "grad_norm": 0.7206727862358093,
      "learning_rate": 8.104864530262213e-08,
      "loss": 0.6529,
      "step": 32774
    },
    {
      "epoch": 0.9443612055552354,
      "grad_norm": 0.7133126258850098,
      "learning_rate": 8.096499284098447e-08,
      "loss": 0.6757,
      "step": 32775
    },
    {
      "epoch": 0.9443900190168847,
      "grad_norm": 0.7230690121650696,
      "learning_rate": 8.088138321922456e-08,
      "loss": 0.6735,
      "step": 32776
    },
    {
      "epoch": 0.944418832478534,
      "grad_norm": 0.7312441468238831,
      "learning_rate": 8.079781643807072e-08,
      "loss": 0.7019,
      "step": 32777
    },
    {
      "epoch": 0.9444476459401833,
      "grad_norm": 0.7223502993583679,
      "learning_rate": 8.071429249825013e-08,
      "loss": 0.6667,
      "step": 32778
    },
    {
      "epoch": 0.9444764594018326,
      "grad_norm": 0.7321391701698303,
      "learning_rate": 8.063081140049112e-08,
      "loss": 0.671,
      "step": 32779
    },
    {
      "epoch": 0.9445052728634818,
      "grad_norm": 0.7443061470985413,
      "learning_rate": 8.054737314551975e-08,
      "loss": 0.6625,
      "step": 32780
    },
    {
      "epoch": 0.9445340863251311,
      "grad_norm": 0.7226035594940186,
      "learning_rate": 8.046397773406268e-08,
      "loss": 0.6535,
      "step": 32781
    },
    {
      "epoch": 0.9445628997867804,
      "grad_norm": 0.7103449106216431,
      "learning_rate": 8.038062516684764e-08,
      "loss": 0.7,
      "step": 32782
    },
    {
      "epoch": 0.9445917132484297,
      "grad_norm": 0.7178061604499817,
      "learning_rate": 8.029731544459851e-08,
      "loss": 0.675,
      "step": 32783
    },
    {
      "epoch": 0.9446205267100789,
      "grad_norm": 0.7489979267120361,
      "learning_rate": 8.021404856804304e-08,
      "loss": 0.6987,
      "step": 32784
    },
    {
      "epoch": 0.9446493401717282,
      "grad_norm": 0.7256326079368591,
      "learning_rate": 8.0130824537904e-08,
      "loss": 0.6826,
      "step": 32785
    },
    {
      "epoch": 0.9446781536333775,
      "grad_norm": 0.7155330777168274,
      "learning_rate": 8.004764335490856e-08,
      "loss": 0.6798,
      "step": 32786
    },
    {
      "epoch": 0.9447069670950268,
      "grad_norm": 0.7076534032821655,
      "learning_rate": 7.996450501977893e-08,
      "loss": 0.6527,
      "step": 32787
    },
    {
      "epoch": 0.9447357805566761,
      "grad_norm": 0.7404805421829224,
      "learning_rate": 7.988140953324065e-08,
      "loss": 0.7,
      "step": 32788
    },
    {
      "epoch": 0.9447645940183254,
      "grad_norm": 0.7264813184738159,
      "learning_rate": 7.9798356896017e-08,
      "loss": 0.7077,
      "step": 32789
    },
    {
      "epoch": 0.9447934074799746,
      "grad_norm": 0.7085091471672058,
      "learning_rate": 7.971534710883133e-08,
      "loss": 0.6694,
      "step": 32790
    },
    {
      "epoch": 0.9448222209416239,
      "grad_norm": 0.704397976398468,
      "learning_rate": 7.963238017240638e-08,
      "loss": 0.6816,
      "step": 32791
    },
    {
      "epoch": 0.9448510344032732,
      "grad_norm": 0.7157861590385437,
      "learning_rate": 7.954945608746545e-08,
      "loss": 0.6822,
      "step": 32792
    },
    {
      "epoch": 0.9448798478649225,
      "grad_norm": 0.7160266041755676,
      "learning_rate": 7.946657485472908e-08,
      "loss": 0.6771,
      "step": 32793
    },
    {
      "epoch": 0.9449086613265718,
      "grad_norm": 0.7415517568588257,
      "learning_rate": 7.938373647492114e-08,
      "loss": 0.6846,
      "step": 32794
    },
    {
      "epoch": 0.9449374747882211,
      "grad_norm": 0.7314640283584595,
      "learning_rate": 7.930094094876106e-08,
      "loss": 0.6846,
      "step": 32795
    },
    {
      "epoch": 0.9449662882498704,
      "grad_norm": 0.7067468762397766,
      "learning_rate": 7.921818827697159e-08,
      "loss": 0.7007,
      "step": 32796
    },
    {
      "epoch": 0.9449951017115196,
      "grad_norm": 0.7327786684036255,
      "learning_rate": 7.91354784602727e-08,
      "loss": 0.6903,
      "step": 32797
    },
    {
      "epoch": 0.9450239151731689,
      "grad_norm": 0.7508102059364319,
      "learning_rate": 7.905281149938437e-08,
      "loss": 0.6733,
      "step": 32798
    },
    {
      "epoch": 0.9450527286348182,
      "grad_norm": 0.7021569609642029,
      "learning_rate": 7.897018739502716e-08,
      "loss": 0.6667,
      "step": 32799
    },
    {
      "epoch": 0.9450815420964674,
      "grad_norm": 0.7277011871337891,
      "learning_rate": 7.88876061479199e-08,
      "loss": 0.6898,
      "step": 32800
    },
    {
      "epoch": 0.9451103555581167,
      "grad_norm": 0.6977012753486633,
      "learning_rate": 7.880506775878261e-08,
      "loss": 0.6511,
      "step": 32801
    },
    {
      "epoch": 0.945139169019766,
      "grad_norm": 0.7170641422271729,
      "learning_rate": 7.872257222833357e-08,
      "loss": 0.6628,
      "step": 32802
    },
    {
      "epoch": 0.9451679824814153,
      "grad_norm": 0.7275039553642273,
      "learning_rate": 7.864011955729167e-08,
      "loss": 0.7053,
      "step": 32803
    },
    {
      "epoch": 0.9451967959430646,
      "grad_norm": 0.7161362767219543,
      "learning_rate": 7.855770974637467e-08,
      "loss": 0.6701,
      "step": 32804
    },
    {
      "epoch": 0.9452256094047139,
      "grad_norm": 0.7298725843429565,
      "learning_rate": 7.847534279630086e-08,
      "loss": 0.7,
      "step": 32805
    },
    {
      "epoch": 0.9452544228663632,
      "grad_norm": 0.7255651354789734,
      "learning_rate": 7.839301870778637e-08,
      "loss": 0.6964,
      "step": 32806
    },
    {
      "epoch": 0.9452832363280125,
      "grad_norm": 0.714293360710144,
      "learning_rate": 7.83107374815495e-08,
      "loss": 0.6756,
      "step": 32807
    },
    {
      "epoch": 0.9453120497896618,
      "grad_norm": 0.7189375162124634,
      "learning_rate": 7.822849911830577e-08,
      "loss": 0.6763,
      "step": 32808
    },
    {
      "epoch": 0.9453408632513111,
      "grad_norm": 0.7401279807090759,
      "learning_rate": 7.81463036187724e-08,
      "loss": 0.6576,
      "step": 32809
    },
    {
      "epoch": 0.9453696767129603,
      "grad_norm": 0.7067420482635498,
      "learning_rate": 7.806415098366438e-08,
      "loss": 0.6912,
      "step": 32810
    },
    {
      "epoch": 0.9453984901746095,
      "grad_norm": 0.7088650465011597,
      "learning_rate": 7.798204121369723e-08,
      "loss": 0.6625,
      "step": 32811
    },
    {
      "epoch": 0.9454273036362588,
      "grad_norm": 0.7294434905052185,
      "learning_rate": 7.789997430958595e-08,
      "loss": 0.6907,
      "step": 32812
    },
    {
      "epoch": 0.9454561170979081,
      "grad_norm": 0.6941937804222107,
      "learning_rate": 7.781795027204663e-08,
      "loss": 0.6595,
      "step": 32813
    },
    {
      "epoch": 0.9454849305595574,
      "grad_norm": 0.7089138627052307,
      "learning_rate": 7.773596910179149e-08,
      "loss": 0.6806,
      "step": 32814
    },
    {
      "epoch": 0.9455137440212067,
      "grad_norm": 0.7501052618026733,
      "learning_rate": 7.76540307995366e-08,
      "loss": 0.6985,
      "step": 32815
    },
    {
      "epoch": 0.945542557482856,
      "grad_norm": 0.7334443926811218,
      "learning_rate": 7.757213536599418e-08,
      "loss": 0.6998,
      "step": 32816
    },
    {
      "epoch": 0.9455713709445053,
      "grad_norm": 0.7123561501502991,
      "learning_rate": 7.749028280187753e-08,
      "loss": 0.6813,
      "step": 32817
    },
    {
      "epoch": 0.9456001844061546,
      "grad_norm": 0.7245783805847168,
      "learning_rate": 7.74084731079e-08,
      "loss": 0.6764,
      "step": 32818
    },
    {
      "epoch": 0.9456289978678039,
      "grad_norm": 0.7165985703468323,
      "learning_rate": 7.73267062847738e-08,
      "loss": 0.661,
      "step": 32819
    },
    {
      "epoch": 0.9456578113294531,
      "grad_norm": 0.7371382713317871,
      "learning_rate": 7.724498233321165e-08,
      "loss": 0.7096,
      "step": 32820
    },
    {
      "epoch": 0.9456866247911024,
      "grad_norm": 0.7052280902862549,
      "learning_rate": 7.716330125392413e-08,
      "loss": 0.6782,
      "step": 32821
    },
    {
      "epoch": 0.9457154382527517,
      "grad_norm": 0.7233642935752869,
      "learning_rate": 7.708166304762343e-08,
      "loss": 0.6983,
      "step": 32822
    },
    {
      "epoch": 0.945744251714401,
      "grad_norm": 0.7061492204666138,
      "learning_rate": 7.700006771502012e-08,
      "loss": 0.6477,
      "step": 32823
    },
    {
      "epoch": 0.9457730651760502,
      "grad_norm": 0.7057890295982361,
      "learning_rate": 7.691851525682525e-08,
      "loss": 0.6373,
      "step": 32824
    },
    {
      "epoch": 0.9458018786376995,
      "grad_norm": 0.7155765295028687,
      "learning_rate": 7.683700567374886e-08,
      "loss": 0.6685,
      "step": 32825
    },
    {
      "epoch": 0.9458306920993488,
      "grad_norm": 0.7008055448532104,
      "learning_rate": 7.67555389665009e-08,
      "loss": 0.6449,
      "step": 32826
    },
    {
      "epoch": 0.9458595055609981,
      "grad_norm": 0.7398987412452698,
      "learning_rate": 7.667411513579026e-08,
      "loss": 0.6908,
      "step": 32827
    },
    {
      "epoch": 0.9458883190226474,
      "grad_norm": 0.7159913182258606,
      "learning_rate": 7.659273418232693e-08,
      "loss": 0.6702,
      "step": 32828
    },
    {
      "epoch": 0.9459171324842967,
      "grad_norm": 0.7252392172813416,
      "learning_rate": 7.651139610681924e-08,
      "loss": 0.6649,
      "step": 32829
    },
    {
      "epoch": 0.9459459459459459,
      "grad_norm": 0.7293680310249329,
      "learning_rate": 7.643010090997604e-08,
      "loss": 0.6629,
      "step": 32830
    },
    {
      "epoch": 0.9459747594075952,
      "grad_norm": 0.7281396389007568,
      "learning_rate": 7.634884859250402e-08,
      "loss": 0.6781,
      "step": 32831
    },
    {
      "epoch": 0.9460035728692445,
      "grad_norm": 0.721231997013092,
      "learning_rate": 7.62676391551126e-08,
      "loss": 0.6754,
      "step": 32832
    },
    {
      "epoch": 0.9460323863308938,
      "grad_norm": 0.7276604771614075,
      "learning_rate": 7.618647259850731e-08,
      "loss": 0.6697,
      "step": 32833
    },
    {
      "epoch": 0.9460611997925431,
      "grad_norm": 0.7036352157592773,
      "learning_rate": 7.61053489233965e-08,
      "loss": 0.6604,
      "step": 32834
    },
    {
      "epoch": 0.9460900132541924,
      "grad_norm": 0.7192065119743347,
      "learning_rate": 7.602426813048513e-08,
      "loss": 0.6576,
      "step": 32835
    },
    {
      "epoch": 0.9461188267158417,
      "grad_norm": 0.70449298620224,
      "learning_rate": 7.5943230220481e-08,
      "loss": 0.6921,
      "step": 32836
    },
    {
      "epoch": 0.946147640177491,
      "grad_norm": 0.7445019483566284,
      "learning_rate": 7.586223519408853e-08,
      "loss": 0.6902,
      "step": 32837
    },
    {
      "epoch": 0.9461764536391402,
      "grad_norm": 0.7446838021278381,
      "learning_rate": 7.578128305201382e-08,
      "loss": 0.7039,
      "step": 32838
    },
    {
      "epoch": 0.9462052671007894,
      "grad_norm": 0.7308443188667297,
      "learning_rate": 7.570037379496187e-08,
      "loss": 0.6803,
      "step": 32839
    },
    {
      "epoch": 0.9462340805624387,
      "grad_norm": 0.6902689337730408,
      "learning_rate": 7.56195074236371e-08,
      "loss": 0.6566,
      "step": 32840
    },
    {
      "epoch": 0.946262894024088,
      "grad_norm": 0.730011522769928,
      "learning_rate": 7.553868393874397e-08,
      "loss": 0.6816,
      "step": 32841
    },
    {
      "epoch": 0.9462917074857373,
      "grad_norm": 0.7204684019088745,
      "learning_rate": 7.54579033409858e-08,
      "loss": 0.6918,
      "step": 32842
    },
    {
      "epoch": 0.9463205209473866,
      "grad_norm": 0.720387876033783,
      "learning_rate": 7.537716563106701e-08,
      "loss": 0.6754,
      "step": 32843
    },
    {
      "epoch": 0.9463493344090359,
      "grad_norm": 0.7121241092681885,
      "learning_rate": 7.529647080969039e-08,
      "loss": 0.6791,
      "step": 32844
    },
    {
      "epoch": 0.9463781478706852,
      "grad_norm": 0.7230473756790161,
      "learning_rate": 7.521581887755868e-08,
      "loss": 0.6876,
      "step": 32845
    },
    {
      "epoch": 0.9464069613323345,
      "grad_norm": 0.7156376838684082,
      "learning_rate": 7.513520983537359e-08,
      "loss": 0.6767,
      "step": 32846
    },
    {
      "epoch": 0.9464357747939838,
      "grad_norm": 0.7380304336547852,
      "learning_rate": 7.505464368383841e-08,
      "loss": 0.6767,
      "step": 32847
    },
    {
      "epoch": 0.9464645882556331,
      "grad_norm": 0.7456737160682678,
      "learning_rate": 7.497412042365316e-08,
      "loss": 0.6957,
      "step": 32848
    },
    {
      "epoch": 0.9464934017172824,
      "grad_norm": 0.721515953540802,
      "learning_rate": 7.489364005552113e-08,
      "loss": 0.687,
      "step": 32849
    },
    {
      "epoch": 0.9465222151789316,
      "grad_norm": 0.7301719188690186,
      "learning_rate": 7.481320258014124e-08,
      "loss": 0.68,
      "step": 32850
    },
    {
      "epoch": 0.9465510286405808,
      "grad_norm": 0.7309753894805908,
      "learning_rate": 7.473280799821624e-08,
      "loss": 0.6985,
      "step": 32851
    },
    {
      "epoch": 0.9465798421022301,
      "grad_norm": 0.7295039892196655,
      "learning_rate": 7.465245631044393e-08,
      "loss": 0.6863,
      "step": 32852
    },
    {
      "epoch": 0.9466086555638794,
      "grad_norm": 0.7263275384902954,
      "learning_rate": 7.457214751752595e-08,
      "loss": 0.6752,
      "step": 32853
    },
    {
      "epoch": 0.9466374690255287,
      "grad_norm": 0.7375873923301697,
      "learning_rate": 7.449188162016008e-08,
      "loss": 0.678,
      "step": 32854
    },
    {
      "epoch": 0.946666282487178,
      "grad_norm": 0.7213946580886841,
      "learning_rate": 7.441165861904686e-08,
      "loss": 0.701,
      "step": 32855
    },
    {
      "epoch": 0.9466950959488273,
      "grad_norm": 0.7035577893257141,
      "learning_rate": 7.43314785148841e-08,
      "loss": 0.665,
      "step": 32856
    },
    {
      "epoch": 0.9467239094104766,
      "grad_norm": 0.6952441930770874,
      "learning_rate": 7.425134130837009e-08,
      "loss": 0.6698,
      "step": 32857
    },
    {
      "epoch": 0.9467527228721259,
      "grad_norm": 0.7211264967918396,
      "learning_rate": 7.417124700020373e-08,
      "loss": 0.6803,
      "step": 32858
    },
    {
      "epoch": 0.9467815363337752,
      "grad_norm": 0.7306157946586609,
      "learning_rate": 7.409119559108113e-08,
      "loss": 0.6868,
      "step": 32859
    },
    {
      "epoch": 0.9468103497954244,
      "grad_norm": 0.7245948910713196,
      "learning_rate": 7.401118708170008e-08,
      "loss": 0.6773,
      "step": 32860
    },
    {
      "epoch": 0.9468391632570737,
      "grad_norm": 0.7399305701255798,
      "learning_rate": 7.393122147275722e-08,
      "loss": 0.6937,
      "step": 32861
    },
    {
      "epoch": 0.946867976718723,
      "grad_norm": 0.7258886694908142,
      "learning_rate": 7.385129876494979e-08,
      "loss": 0.7038,
      "step": 32862
    },
    {
      "epoch": 0.9468967901803723,
      "grad_norm": 0.7340220212936401,
      "learning_rate": 7.377141895897277e-08,
      "loss": 0.6829,
      "step": 32863
    },
    {
      "epoch": 0.9469256036420216,
      "grad_norm": 0.7064802050590515,
      "learning_rate": 7.36915820555223e-08,
      "loss": 0.6652,
      "step": 32864
    },
    {
      "epoch": 0.9469544171036708,
      "grad_norm": 0.7223984003067017,
      "learning_rate": 7.36117880552939e-08,
      "loss": 0.6686,
      "step": 32865
    },
    {
      "epoch": 0.9469832305653201,
      "grad_norm": 0.7349599003791809,
      "learning_rate": 7.353203695898203e-08,
      "loss": 0.6972,
      "step": 32866
    },
    {
      "epoch": 0.9470120440269694,
      "grad_norm": 0.7785478234291077,
      "learning_rate": 7.34523287672817e-08,
      "loss": 0.6902,
      "step": 32867
    },
    {
      "epoch": 0.9470408574886187,
      "grad_norm": 0.7382227182388306,
      "learning_rate": 7.337266348088678e-08,
      "loss": 0.6655,
      "step": 32868
    },
    {
      "epoch": 0.947069670950268,
      "grad_norm": 0.7226371765136719,
      "learning_rate": 7.329304110049062e-08,
      "loss": 0.6682,
      "step": 32869
    },
    {
      "epoch": 0.9470984844119172,
      "grad_norm": 0.7078553438186646,
      "learning_rate": 7.321346162678767e-08,
      "loss": 0.6775,
      "step": 32870
    },
    {
      "epoch": 0.9471272978735665,
      "grad_norm": 0.726775050163269,
      "learning_rate": 7.313392506047012e-08,
      "loss": 0.6652,
      "step": 32871
    },
    {
      "epoch": 0.9471561113352158,
      "grad_norm": 0.7297424077987671,
      "learning_rate": 7.305443140223134e-08,
      "loss": 0.6438,
      "step": 32872
    },
    {
      "epoch": 0.9471849247968651,
      "grad_norm": 0.725948691368103,
      "learning_rate": 7.297498065276299e-08,
      "loss": 0.6675,
      "step": 32873
    },
    {
      "epoch": 0.9472137382585144,
      "grad_norm": 0.7270875573158264,
      "learning_rate": 7.289557281275783e-08,
      "loss": 0.6845,
      "step": 32874
    },
    {
      "epoch": 0.9472425517201637,
      "grad_norm": 0.7486974000930786,
      "learning_rate": 7.281620788290644e-08,
      "loss": 0.6942,
      "step": 32875
    },
    {
      "epoch": 0.947271365181813,
      "grad_norm": 0.7357778549194336,
      "learning_rate": 7.273688586390104e-08,
      "loss": 0.7093,
      "step": 32876
    },
    {
      "epoch": 0.9473001786434623,
      "grad_norm": 0.7426135540008545,
      "learning_rate": 7.265760675643163e-08,
      "loss": 0.6923,
      "step": 32877
    },
    {
      "epoch": 0.9473289921051115,
      "grad_norm": 0.7139047980308533,
      "learning_rate": 7.257837056118877e-08,
      "loss": 0.6825,
      "step": 32878
    },
    {
      "epoch": 0.9473578055667607,
      "grad_norm": 0.7293376326560974,
      "learning_rate": 7.249917727886302e-08,
      "loss": 0.6812,
      "step": 32879
    },
    {
      "epoch": 0.94738661902841,
      "grad_norm": 0.7044110894203186,
      "learning_rate": 7.242002691014327e-08,
      "loss": 0.6696,
      "step": 32880
    },
    {
      "epoch": 0.9474154324900593,
      "grad_norm": 0.6893067955970764,
      "learning_rate": 7.234091945572008e-08,
      "loss": 0.6618,
      "step": 32881
    },
    {
      "epoch": 0.9474442459517086,
      "grad_norm": 0.7279964089393616,
      "learning_rate": 7.226185491628069e-08,
      "loss": 0.6889,
      "step": 32882
    },
    {
      "epoch": 0.9474730594133579,
      "grad_norm": 0.7237349152565002,
      "learning_rate": 7.218283329251508e-08,
      "loss": 0.6788,
      "step": 32883
    },
    {
      "epoch": 0.9475018728750072,
      "grad_norm": 0.7335673570632935,
      "learning_rate": 7.210385458511049e-08,
      "loss": 0.6739,
      "step": 32884
    },
    {
      "epoch": 0.9475306863366565,
      "grad_norm": 0.719901978969574,
      "learning_rate": 7.202491879475581e-08,
      "loss": 0.6687,
      "step": 32885
    },
    {
      "epoch": 0.9475594997983058,
      "grad_norm": 0.7169724106788635,
      "learning_rate": 7.194602592213773e-08,
      "loss": 0.6769,
      "step": 32886
    },
    {
      "epoch": 0.9475883132599551,
      "grad_norm": 0.7160339951515198,
      "learning_rate": 7.186717596794402e-08,
      "loss": 0.6662,
      "step": 32887
    },
    {
      "epoch": 0.9476171267216044,
      "grad_norm": 0.728252649307251,
      "learning_rate": 7.178836893286023e-08,
      "loss": 0.66,
      "step": 32888
    },
    {
      "epoch": 0.9476459401832537,
      "grad_norm": 0.7273474335670471,
      "learning_rate": 7.170960481757361e-08,
      "loss": 0.6847,
      "step": 32889
    },
    {
      "epoch": 0.947674753644903,
      "grad_norm": 0.7208877205848694,
      "learning_rate": 7.163088362276971e-08,
      "loss": 0.6835,
      "step": 32890
    },
    {
      "epoch": 0.9477035671065522,
      "grad_norm": 0.7266562581062317,
      "learning_rate": 7.155220534913465e-08,
      "loss": 0.6716,
      "step": 32891
    },
    {
      "epoch": 0.9477323805682014,
      "grad_norm": 0.7236741781234741,
      "learning_rate": 7.147356999735233e-08,
      "loss": 0.6626,
      "step": 32892
    },
    {
      "epoch": 0.9477611940298507,
      "grad_norm": 0.7051814794540405,
      "learning_rate": 7.139497756810943e-08,
      "loss": 0.6653,
      "step": 32893
    },
    {
      "epoch": 0.9477900074915,
      "grad_norm": 0.7189432978630066,
      "learning_rate": 7.131642806208927e-08,
      "loss": 0.6821,
      "step": 32894
    },
    {
      "epoch": 0.9478188209531493,
      "grad_norm": 0.7600775957107544,
      "learning_rate": 7.123792147997577e-08,
      "loss": 0.6947,
      "step": 32895
    },
    {
      "epoch": 0.9478476344147986,
      "grad_norm": 0.7039099335670471,
      "learning_rate": 7.115945782245282e-08,
      "loss": 0.6689,
      "step": 32896
    },
    {
      "epoch": 0.9478764478764479,
      "grad_norm": 0.6987795233726501,
      "learning_rate": 7.108103709020487e-08,
      "loss": 0.6837,
      "step": 32897
    },
    {
      "epoch": 0.9479052613380972,
      "grad_norm": 0.7017131447792053,
      "learning_rate": 7.100265928391303e-08,
      "loss": 0.6664,
      "step": 32898
    },
    {
      "epoch": 0.9479340747997465,
      "grad_norm": 0.7194840908050537,
      "learning_rate": 7.092432440426122e-08,
      "loss": 0.6609,
      "step": 32899
    },
    {
      "epoch": 0.9479628882613957,
      "grad_norm": 0.7775349020957947,
      "learning_rate": 7.084603245193167e-08,
      "loss": 0.6909,
      "step": 32900
    },
    {
      "epoch": 0.947991701723045,
      "grad_norm": 0.7411405444145203,
      "learning_rate": 7.076778342760493e-08,
      "loss": 0.6754,
      "step": 32901
    },
    {
      "epoch": 0.9480205151846943,
      "grad_norm": 0.7202645540237427,
      "learning_rate": 7.068957733196436e-08,
      "loss": 0.6783,
      "step": 32902
    },
    {
      "epoch": 0.9480493286463436,
      "grad_norm": 0.7099848389625549,
      "learning_rate": 7.06114141656894e-08,
      "loss": 0.6619,
      "step": 32903
    },
    {
      "epoch": 0.9480781421079929,
      "grad_norm": 0.7251721024513245,
      "learning_rate": 7.053329392946174e-08,
      "loss": 0.712,
      "step": 32904
    },
    {
      "epoch": 0.9481069555696421,
      "grad_norm": 0.7180392146110535,
      "learning_rate": 7.045521662396138e-08,
      "loss": 0.6785,
      "step": 32905
    },
    {
      "epoch": 0.9481357690312914,
      "grad_norm": 0.7198660373687744,
      "learning_rate": 7.037718224986833e-08,
      "loss": 0.6584,
      "step": 32906
    },
    {
      "epoch": 0.9481645824929407,
      "grad_norm": 0.7126910090446472,
      "learning_rate": 7.029919080786151e-08,
      "loss": 0.6943,
      "step": 32907
    },
    {
      "epoch": 0.94819339595459,
      "grad_norm": 0.727972686290741,
      "learning_rate": 7.022124229862204e-08,
      "loss": 0.6801,
      "step": 32908
    },
    {
      "epoch": 0.9482222094162392,
      "grad_norm": 0.7471716403961182,
      "learning_rate": 7.014333672282659e-08,
      "loss": 0.6848,
      "step": 32909
    },
    {
      "epoch": 0.9482510228778885,
      "grad_norm": 0.7009857296943665,
      "learning_rate": 7.006547408115572e-08,
      "loss": 0.6677,
      "step": 32910
    },
    {
      "epoch": 0.9482798363395378,
      "grad_norm": 0.6975641250610352,
      "learning_rate": 6.998765437428556e-08,
      "loss": 0.6502,
      "step": 32911
    },
    {
      "epoch": 0.9483086498011871,
      "grad_norm": 0.7241896390914917,
      "learning_rate": 6.990987760289502e-08,
      "loss": 0.7024,
      "step": 32912
    },
    {
      "epoch": 0.9483374632628364,
      "grad_norm": 0.7029203176498413,
      "learning_rate": 6.983214376766079e-08,
      "loss": 0.6581,
      "step": 32913
    },
    {
      "epoch": 0.9483662767244857,
      "grad_norm": 0.7247812151908875,
      "learning_rate": 6.975445286926064e-08,
      "loss": 0.6472,
      "step": 32914
    },
    {
      "epoch": 0.948395090186135,
      "grad_norm": 0.72111576795578,
      "learning_rate": 6.96768049083707e-08,
      "loss": 0.6635,
      "step": 32915
    },
    {
      "epoch": 0.9484239036477843,
      "grad_norm": 0.7493362426757812,
      "learning_rate": 6.95991998856671e-08,
      "loss": 0.6716,
      "step": 32916
    },
    {
      "epoch": 0.9484527171094336,
      "grad_norm": 0.7710627317428589,
      "learning_rate": 6.952163780182596e-08,
      "loss": 0.6841,
      "step": 32917
    },
    {
      "epoch": 0.9484815305710829,
      "grad_norm": 0.7438782453536987,
      "learning_rate": 6.944411865752288e-08,
      "loss": 0.6726,
      "step": 32918
    },
    {
      "epoch": 0.948510344032732,
      "grad_norm": 0.7942159175872803,
      "learning_rate": 6.936664245343227e-08,
      "loss": 0.6453,
      "step": 32919
    },
    {
      "epoch": 0.9485391574943813,
      "grad_norm": 0.733117938041687,
      "learning_rate": 6.928920919022974e-08,
      "loss": 0.6744,
      "step": 32920
    },
    {
      "epoch": 0.9485679709560306,
      "grad_norm": 0.7108358144760132,
      "learning_rate": 6.921181886858918e-08,
      "loss": 0.6722,
      "step": 32921
    },
    {
      "epoch": 0.9485967844176799,
      "grad_norm": 1.602455496788025,
      "learning_rate": 6.913447148918506e-08,
      "loss": 0.6799,
      "step": 32922
    },
    {
      "epoch": 0.9486255978793292,
      "grad_norm": 0.7311663031578064,
      "learning_rate": 6.905716705269017e-08,
      "loss": 0.6478,
      "step": 32923
    },
    {
      "epoch": 0.9486544113409785,
      "grad_norm": 0.7202883958816528,
      "learning_rate": 6.89799055597784e-08,
      "loss": 0.66,
      "step": 32924
    },
    {
      "epoch": 0.9486832248026278,
      "grad_norm": 0.7253447771072388,
      "learning_rate": 6.890268701112257e-08,
      "loss": 0.7069,
      "step": 32925
    },
    {
      "epoch": 0.9487120382642771,
      "grad_norm": 0.7334181070327759,
      "learning_rate": 6.88255114073949e-08,
      "loss": 0.7129,
      "step": 32926
    },
    {
      "epoch": 0.9487408517259264,
      "grad_norm": 0.7350688576698303,
      "learning_rate": 6.874837874926765e-08,
      "loss": 0.6976,
      "step": 32927
    },
    {
      "epoch": 0.9487696651875757,
      "grad_norm": 0.7390447854995728,
      "learning_rate": 6.867128903741249e-08,
      "loss": 0.6844,
      "step": 32928
    },
    {
      "epoch": 0.948798478649225,
      "grad_norm": 0.7226873636245728,
      "learning_rate": 6.859424227250112e-08,
      "loss": 0.7005,
      "step": 32929
    },
    {
      "epoch": 0.9488272921108742,
      "grad_norm": 0.7098739147186279,
      "learning_rate": 6.851723845520409e-08,
      "loss": 0.6525,
      "step": 32930
    },
    {
      "epoch": 0.9488561055725235,
      "grad_norm": 0.7196952700614929,
      "learning_rate": 6.844027758619253e-08,
      "loss": 0.6732,
      "step": 32931
    },
    {
      "epoch": 0.9488849190341727,
      "grad_norm": 0.7297313213348389,
      "learning_rate": 6.836335966613594e-08,
      "loss": 0.6636,
      "step": 32932
    },
    {
      "epoch": 0.948913732495822,
      "grad_norm": 1.7034271955490112,
      "learning_rate": 6.828648469570488e-08,
      "loss": 0.6947,
      "step": 32933
    },
    {
      "epoch": 0.9489425459574713,
      "grad_norm": 0.7291533350944519,
      "learning_rate": 6.820965267556822e-08,
      "loss": 0.6726,
      "step": 32934
    },
    {
      "epoch": 0.9489713594191206,
      "grad_norm": 0.707239031791687,
      "learning_rate": 6.813286360639659e-08,
      "loss": 0.67,
      "step": 32935
    },
    {
      "epoch": 0.9490001728807699,
      "grad_norm": 0.7337790131568909,
      "learning_rate": 6.805611748885666e-08,
      "loss": 0.6629,
      "step": 32936
    },
    {
      "epoch": 0.9490289863424192,
      "grad_norm": 0.7225649952888489,
      "learning_rate": 6.797941432361788e-08,
      "loss": 0.7086,
      "step": 32937
    },
    {
      "epoch": 0.9490577998040685,
      "grad_norm": 0.7279468774795532,
      "learning_rate": 6.790275411134861e-08,
      "loss": 0.677,
      "step": 32938
    },
    {
      "epoch": 0.9490866132657177,
      "grad_norm": 0.7212069630622864,
      "learning_rate": 6.782613685271555e-08,
      "loss": 0.672,
      "step": 32939
    },
    {
      "epoch": 0.949115426727367,
      "grad_norm": 0.7162386775016785,
      "learning_rate": 6.774956254838704e-08,
      "loss": 0.6937,
      "step": 32940
    },
    {
      "epoch": 0.9491442401890163,
      "grad_norm": 0.7398764491081238,
      "learning_rate": 6.767303119902924e-08,
      "loss": 0.7193,
      "step": 32941
    },
    {
      "epoch": 0.9491730536506656,
      "grad_norm": 0.7322613000869751,
      "learning_rate": 6.75965428053088e-08,
      "loss": 0.6852,
      "step": 32942
    },
    {
      "epoch": 0.9492018671123149,
      "grad_norm": 0.7090369462966919,
      "learning_rate": 6.752009736789134e-08,
      "loss": 0.6897,
      "step": 32943
    },
    {
      "epoch": 0.9492306805739642,
      "grad_norm": 0.7181916832923889,
      "learning_rate": 6.744369488744406e-08,
      "loss": 0.6577,
      "step": 32944
    },
    {
      "epoch": 0.9492594940356135,
      "grad_norm": 0.7208198308944702,
      "learning_rate": 6.736733536463091e-08,
      "loss": 0.6719,
      "step": 32945
    },
    {
      "epoch": 0.9492883074972627,
      "grad_norm": 0.7097761034965515,
      "learning_rate": 6.729101880011746e-08,
      "loss": 0.6672,
      "step": 32946
    },
    {
      "epoch": 0.949317120958912,
      "grad_norm": 0.7406944632530212,
      "learning_rate": 6.721474519456816e-08,
      "loss": 0.6895,
      "step": 32947
    },
    {
      "epoch": 0.9493459344205613,
      "grad_norm": 0.7145915031433105,
      "learning_rate": 6.713851454864806e-08,
      "loss": 0.6696,
      "step": 32948
    },
    {
      "epoch": 0.9493747478822105,
      "grad_norm": 0.7060731053352356,
      "learning_rate": 6.706232686301994e-08,
      "loss": 0.6616,
      "step": 32949
    },
    {
      "epoch": 0.9494035613438598,
      "grad_norm": 0.7204951643943787,
      "learning_rate": 6.698618213834829e-08,
      "loss": 0.6962,
      "step": 32950
    },
    {
      "epoch": 0.9494323748055091,
      "grad_norm": 0.7157513499259949,
      "learning_rate": 6.691008037529535e-08,
      "loss": 0.6961,
      "step": 32951
    },
    {
      "epoch": 0.9494611882671584,
      "grad_norm": 0.7289892435073853,
      "learning_rate": 6.683402157452446e-08,
      "loss": 0.6832,
      "step": 32952
    },
    {
      "epoch": 0.9494900017288077,
      "grad_norm": 0.7175893783569336,
      "learning_rate": 6.675800573669788e-08,
      "loss": 0.6455,
      "step": 32953
    },
    {
      "epoch": 0.949518815190457,
      "grad_norm": 0.7180024981498718,
      "learning_rate": 6.668203286247732e-08,
      "loss": 0.686,
      "step": 32954
    },
    {
      "epoch": 0.9495476286521063,
      "grad_norm": 0.7209937572479248,
      "learning_rate": 6.6606102952525e-08,
      "loss": 0.6835,
      "step": 32955
    },
    {
      "epoch": 0.9495764421137556,
      "grad_norm": 0.7085236310958862,
      "learning_rate": 6.65302160075021e-08,
      "loss": 0.6561,
      "step": 32956
    },
    {
      "epoch": 0.9496052555754049,
      "grad_norm": 0.7154252529144287,
      "learning_rate": 6.645437202806914e-08,
      "loss": 0.6714,
      "step": 32957
    },
    {
      "epoch": 0.9496340690370542,
      "grad_norm": 0.7337460517883301,
      "learning_rate": 6.637857101488676e-08,
      "loss": 0.6896,
      "step": 32958
    },
    {
      "epoch": 0.9496628824987033,
      "grad_norm": 0.729314923286438,
      "learning_rate": 6.630281296861552e-08,
      "loss": 0.6773,
      "step": 32959
    },
    {
      "epoch": 0.9496916959603526,
      "grad_norm": 0.7168089151382446,
      "learning_rate": 6.622709788991433e-08,
      "loss": 0.6801,
      "step": 32960
    },
    {
      "epoch": 0.9497205094220019,
      "grad_norm": 0.7311936020851135,
      "learning_rate": 6.615142577944378e-08,
      "loss": 0.685,
      "step": 32961
    },
    {
      "epoch": 0.9497493228836512,
      "grad_norm": 0.7298953533172607,
      "learning_rate": 6.607579663786223e-08,
      "loss": 0.6924,
      "step": 32962
    },
    {
      "epoch": 0.9497781363453005,
      "grad_norm": 0.7359702587127686,
      "learning_rate": 6.600021046582805e-08,
      "loss": 0.6746,
      "step": 32963
    },
    {
      "epoch": 0.9498069498069498,
      "grad_norm": 0.7119123339653015,
      "learning_rate": 6.592466726400016e-08,
      "loss": 0.6708,
      "step": 32964
    },
    {
      "epoch": 0.9498357632685991,
      "grad_norm": 0.724909245967865,
      "learning_rate": 6.584916703303579e-08,
      "loss": 0.6897,
      "step": 32965
    },
    {
      "epoch": 0.9498645767302484,
      "grad_norm": 0.7113542556762695,
      "learning_rate": 6.577370977359332e-08,
      "loss": 0.6527,
      "step": 32966
    },
    {
      "epoch": 0.9498933901918977,
      "grad_norm": 0.7134795188903809,
      "learning_rate": 6.569829548632889e-08,
      "loss": 0.6706,
      "step": 32967
    },
    {
      "epoch": 0.949922203653547,
      "grad_norm": 0.7183378338813782,
      "learning_rate": 6.562292417189975e-08,
      "loss": 0.6657,
      "step": 32968
    },
    {
      "epoch": 0.9499510171151963,
      "grad_norm": 0.7190958261489868,
      "learning_rate": 6.55475958309626e-08,
      "loss": 0.6596,
      "step": 32969
    },
    {
      "epoch": 0.9499798305768455,
      "grad_norm": 0.7020693421363831,
      "learning_rate": 6.547231046417357e-08,
      "loss": 0.6435,
      "step": 32970
    },
    {
      "epoch": 0.9500086440384948,
      "grad_norm": 0.7049104571342468,
      "learning_rate": 6.53970680721877e-08,
      "loss": 0.6703,
      "step": 32971
    },
    {
      "epoch": 0.9500374575001441,
      "grad_norm": 0.7464046478271484,
      "learning_rate": 6.532186865566059e-08,
      "loss": 0.6814,
      "step": 32972
    },
    {
      "epoch": 0.9500662709617933,
      "grad_norm": 0.7279570698738098,
      "learning_rate": 6.524671221524726e-08,
      "loss": 0.6742,
      "step": 32973
    },
    {
      "epoch": 0.9500950844234426,
      "grad_norm": 0.7339280843734741,
      "learning_rate": 6.517159875160162e-08,
      "loss": 0.6705,
      "step": 32974
    },
    {
      "epoch": 0.9501238978850919,
      "grad_norm": 1.5682127475738525,
      "learning_rate": 6.509652826537872e-08,
      "loss": 0.6823,
      "step": 32975
    },
    {
      "epoch": 0.9501527113467412,
      "grad_norm": 0.7238848209381104,
      "learning_rate": 6.50215007572319e-08,
      "loss": 0.6579,
      "step": 32976
    },
    {
      "epoch": 0.9501815248083905,
      "grad_norm": 0.7683212757110596,
      "learning_rate": 6.494651622781456e-08,
      "loss": 0.703,
      "step": 32977
    },
    {
      "epoch": 0.9502103382700398,
      "grad_norm": 0.7389832735061646,
      "learning_rate": 6.48715746777806e-08,
      "loss": 0.6824,
      "step": 32978
    },
    {
      "epoch": 0.950239151731689,
      "grad_norm": 0.7261906266212463,
      "learning_rate": 6.479667610778063e-08,
      "loss": 0.7056,
      "step": 32979
    },
    {
      "epoch": 0.9502679651933383,
      "grad_norm": 0.7192731499671936,
      "learning_rate": 6.472182051846909e-08,
      "loss": 0.6844,
      "step": 32980
    },
    {
      "epoch": 0.9502967786549876,
      "grad_norm": 0.7213901281356812,
      "learning_rate": 6.464700791049661e-08,
      "loss": 0.6794,
      "step": 32981
    },
    {
      "epoch": 0.9503255921166369,
      "grad_norm": 0.6979405283927917,
      "learning_rate": 6.457223828451542e-08,
      "loss": 0.6663,
      "step": 32982
    },
    {
      "epoch": 0.9503544055782862,
      "grad_norm": 0.7381592988967896,
      "learning_rate": 6.449751164117612e-08,
      "loss": 0.7055,
      "step": 32983
    },
    {
      "epoch": 0.9503832190399355,
      "grad_norm": 0.7296696305274963,
      "learning_rate": 6.442282798112987e-08,
      "loss": 0.6607,
      "step": 32984
    },
    {
      "epoch": 0.9504120325015848,
      "grad_norm": 0.7444167733192444,
      "learning_rate": 6.434818730502724e-08,
      "loss": 0.6893,
      "step": 32985
    },
    {
      "epoch": 0.950440845963234,
      "grad_norm": 0.7044515609741211,
      "learning_rate": 6.42735896135177e-08,
      "loss": 0.6687,
      "step": 32986
    },
    {
      "epoch": 0.9504696594248833,
      "grad_norm": 0.7386211156845093,
      "learning_rate": 6.419903490725188e-08,
      "loss": 0.698,
      "step": 32987
    },
    {
      "epoch": 0.9504984728865326,
      "grad_norm": 0.7248790264129639,
      "learning_rate": 6.412452318687812e-08,
      "loss": 0.688,
      "step": 32988
    },
    {
      "epoch": 0.9505272863481818,
      "grad_norm": 0.7016776204109192,
      "learning_rate": 6.405005445304536e-08,
      "loss": 0.6737,
      "step": 32989
    },
    {
      "epoch": 0.9505560998098311,
      "grad_norm": 0.7251853942871094,
      "learning_rate": 6.397562870640306e-08,
      "loss": 0.6696,
      "step": 32990
    },
    {
      "epoch": 0.9505849132714804,
      "grad_norm": 0.7242265343666077,
      "learning_rate": 6.390124594759794e-08,
      "loss": 0.688,
      "step": 32991
    },
    {
      "epoch": 0.9506137267331297,
      "grad_norm": 0.6999290585517883,
      "learning_rate": 6.382690617727949e-08,
      "loss": 0.6524,
      "step": 32992
    },
    {
      "epoch": 0.950642540194779,
      "grad_norm": 0.7222519516944885,
      "learning_rate": 6.375260939609385e-08,
      "loss": 0.684,
      "step": 32993
    },
    {
      "epoch": 0.9506713536564283,
      "grad_norm": 0.735093891620636,
      "learning_rate": 6.367835560468938e-08,
      "loss": 0.7171,
      "step": 32994
    },
    {
      "epoch": 0.9507001671180776,
      "grad_norm": 0.7235794067382812,
      "learning_rate": 6.360414480371113e-08,
      "loss": 0.6746,
      "step": 32995
    },
    {
      "epoch": 0.9507289805797269,
      "grad_norm": 0.7340496778488159,
      "learning_rate": 6.35299769938058e-08,
      "loss": 0.6838,
      "step": 32996
    },
    {
      "epoch": 0.9507577940413762,
      "grad_norm": 0.7588040828704834,
      "learning_rate": 6.345585217562066e-08,
      "loss": 0.6733,
      "step": 32997
    },
    {
      "epoch": 0.9507866075030255,
      "grad_norm": 0.6986546516418457,
      "learning_rate": 6.338177034979964e-08,
      "loss": 0.6547,
      "step": 32998
    },
    {
      "epoch": 0.9508154209646748,
      "grad_norm": 0.7199591398239136,
      "learning_rate": 6.330773151698944e-08,
      "loss": 0.6911,
      "step": 32999
    },
    {
      "epoch": 0.9508442344263239,
      "grad_norm": 0.7189459800720215,
      "learning_rate": 6.323373567783342e-08,
      "loss": 0.6714,
      "step": 33000
    },
    {
      "epoch": 0.9508730478879732,
      "grad_norm": 0.7392497658729553,
      "learning_rate": 6.315978283297663e-08,
      "loss": 0.6948,
      "step": 33001
    },
    {
      "epoch": 0.9509018613496225,
      "grad_norm": 0.7373888492584229,
      "learning_rate": 6.308587298306301e-08,
      "loss": 0.6958,
      "step": 33002
    },
    {
      "epoch": 0.9509306748112718,
      "grad_norm": 0.702984094619751,
      "learning_rate": 6.301200612873592e-08,
      "loss": 0.6695,
      "step": 33003
    },
    {
      "epoch": 0.9509594882729211,
      "grad_norm": 0.7220532894134521,
      "learning_rate": 6.293818227063985e-08,
      "loss": 0.6898,
      "step": 33004
    },
    {
      "epoch": 0.9509883017345704,
      "grad_norm": 0.7316688299179077,
      "learning_rate": 6.286440140941652e-08,
      "loss": 0.676,
      "step": 33005
    },
    {
      "epoch": 0.9510171151962197,
      "grad_norm": 0.7438511848449707,
      "learning_rate": 6.279066354570928e-08,
      "loss": 0.6718,
      "step": 33006
    },
    {
      "epoch": 0.951045928657869,
      "grad_norm": 0.7191818356513977,
      "learning_rate": 6.271696868015987e-08,
      "loss": 0.6622,
      "step": 33007
    },
    {
      "epoch": 0.9510747421195183,
      "grad_norm": 0.7276408672332764,
      "learning_rate": 6.264331681340996e-08,
      "loss": 0.6887,
      "step": 33008
    },
    {
      "epoch": 0.9511035555811675,
      "grad_norm": 0.7075391411781311,
      "learning_rate": 6.256970794610129e-08,
      "loss": 0.6871,
      "step": 33009
    },
    {
      "epoch": 0.9511323690428168,
      "grad_norm": 0.7195175290107727,
      "learning_rate": 6.2496142078875e-08,
      "loss": 0.6541,
      "step": 33010
    },
    {
      "epoch": 0.9511611825044661,
      "grad_norm": 0.7376233339309692,
      "learning_rate": 6.24226192123717e-08,
      "loss": 0.7026,
      "step": 33011
    },
    {
      "epoch": 0.9511899959661154,
      "grad_norm": 0.7280398607254028,
      "learning_rate": 6.234913934723142e-08,
      "loss": 0.7039,
      "step": 33012
    },
    {
      "epoch": 0.9512188094277646,
      "grad_norm": 0.7108452916145325,
      "learning_rate": 6.22757024840942e-08,
      "loss": 0.689,
      "step": 33013
    },
    {
      "epoch": 0.9512476228894139,
      "grad_norm": 0.7294231653213501,
      "learning_rate": 6.220230862359955e-08,
      "loss": 0.6908,
      "step": 33014
    },
    {
      "epoch": 0.9512764363510632,
      "grad_norm": 0.7219840288162231,
      "learning_rate": 6.212895776638694e-08,
      "loss": 0.6696,
      "step": 33015
    },
    {
      "epoch": 0.9513052498127125,
      "grad_norm": 0.7156194448471069,
      "learning_rate": 6.205564991309532e-08,
      "loss": 0.6653,
      "step": 33016
    },
    {
      "epoch": 0.9513340632743618,
      "grad_norm": 0.7233002185821533,
      "learning_rate": 6.19823850643625e-08,
      "loss": 0.6647,
      "step": 33017
    },
    {
      "epoch": 0.951362876736011,
      "grad_norm": 0.7387973666191101,
      "learning_rate": 6.19091632208274e-08,
      "loss": 0.6756,
      "step": 33018
    },
    {
      "epoch": 0.9513916901976603,
      "grad_norm": 0.7153002619743347,
      "learning_rate": 6.183598438312677e-08,
      "loss": 0.6823,
      "step": 33019
    },
    {
      "epoch": 0.9514205036593096,
      "grad_norm": 0.7402149438858032,
      "learning_rate": 6.17628485518984e-08,
      "loss": 0.6878,
      "step": 33020
    },
    {
      "epoch": 0.9514493171209589,
      "grad_norm": 0.7237066030502319,
      "learning_rate": 6.168975572777902e-08,
      "loss": 0.6758,
      "step": 33021
    },
    {
      "epoch": 0.9514781305826082,
      "grad_norm": 0.7263818979263306,
      "learning_rate": 6.161670591140589e-08,
      "loss": 0.675,
      "step": 33022
    },
    {
      "epoch": 0.9515069440442575,
      "grad_norm": 0.7018431425094604,
      "learning_rate": 6.154369910341407e-08,
      "loss": 0.6686,
      "step": 33023
    },
    {
      "epoch": 0.9515357575059068,
      "grad_norm": 0.7226274013519287,
      "learning_rate": 6.147073530444081e-08,
      "loss": 0.6788,
      "step": 33024
    },
    {
      "epoch": 0.9515645709675561,
      "grad_norm": 0.7326035499572754,
      "learning_rate": 6.139781451512006e-08,
      "loss": 0.6816,
      "step": 33025
    },
    {
      "epoch": 0.9515933844292054,
      "grad_norm": 0.7444204092025757,
      "learning_rate": 6.132493673608797e-08,
      "loss": 0.6901,
      "step": 33026
    },
    {
      "epoch": 0.9516221978908546,
      "grad_norm": 0.7258336544036865,
      "learning_rate": 6.125210196797793e-08,
      "loss": 0.6943,
      "step": 33027
    },
    {
      "epoch": 0.9516510113525039,
      "grad_norm": 0.7268320322036743,
      "learning_rate": 6.117931021142609e-08,
      "loss": 0.6747,
      "step": 33028
    },
    {
      "epoch": 0.9516798248141531,
      "grad_norm": 0.7353468537330627,
      "learning_rate": 6.110656146706473e-08,
      "loss": 0.684,
      "step": 33029
    },
    {
      "epoch": 0.9517086382758024,
      "grad_norm": 0.7247423529624939,
      "learning_rate": 6.103385573552889e-08,
      "loss": 0.6836,
      "step": 33030
    },
    {
      "epoch": 0.9517374517374517,
      "grad_norm": 0.7111450433731079,
      "learning_rate": 6.096119301745029e-08,
      "loss": 0.6822,
      "step": 33031
    },
    {
      "epoch": 0.951766265199101,
      "grad_norm": 0.7182079553604126,
      "learning_rate": 6.088857331346287e-08,
      "loss": 0.6538,
      "step": 33032
    },
    {
      "epoch": 0.9517950786607503,
      "grad_norm": 0.7149727940559387,
      "learning_rate": 6.081599662419834e-08,
      "loss": 0.6561,
      "step": 33033
    },
    {
      "epoch": 0.9518238921223996,
      "grad_norm": 0.7154989838600159,
      "learning_rate": 6.074346295028954e-08,
      "loss": 0.6839,
      "step": 33034
    },
    {
      "epoch": 0.9518527055840489,
      "grad_norm": 0.7272263765335083,
      "learning_rate": 6.067097229236763e-08,
      "loss": 0.6863,
      "step": 33035
    },
    {
      "epoch": 0.9518815190456982,
      "grad_norm": 0.7078881859779358,
      "learning_rate": 6.059852465106375e-08,
      "loss": 0.6629,
      "step": 33036
    },
    {
      "epoch": 0.9519103325073475,
      "grad_norm": 0.7020608186721802,
      "learning_rate": 6.052612002700964e-08,
      "loss": 0.6549,
      "step": 33037
    },
    {
      "epoch": 0.9519391459689968,
      "grad_norm": 0.7160305976867676,
      "learning_rate": 6.045375842083478e-08,
      "loss": 0.6708,
      "step": 33038
    },
    {
      "epoch": 0.951967959430646,
      "grad_norm": 0.7410402894020081,
      "learning_rate": 6.038143983317035e-08,
      "loss": 0.6786,
      "step": 33039
    },
    {
      "epoch": 0.9519967728922952,
      "grad_norm": 0.7443833351135254,
      "learning_rate": 6.030916426464528e-08,
      "loss": 0.7036,
      "step": 33040
    },
    {
      "epoch": 0.9520255863539445,
      "grad_norm": 0.7371975779533386,
      "learning_rate": 6.023693171589018e-08,
      "loss": 0.6852,
      "step": 33041
    },
    {
      "epoch": 0.9520543998155938,
      "grad_norm": 0.729499876499176,
      "learning_rate": 6.016474218753287e-08,
      "loss": 0.6629,
      "step": 33042
    },
    {
      "epoch": 0.9520832132772431,
      "grad_norm": 0.7116360068321228,
      "learning_rate": 6.009259568020287e-08,
      "loss": 0.6795,
      "step": 33043
    },
    {
      "epoch": 0.9521120267388924,
      "grad_norm": 0.7381777167320251,
      "learning_rate": 6.002049219452854e-08,
      "loss": 0.7052,
      "step": 33044
    },
    {
      "epoch": 0.9521408402005417,
      "grad_norm": 0.742514431476593,
      "learning_rate": 5.994843173113718e-08,
      "loss": 0.6944,
      "step": 33045
    },
    {
      "epoch": 0.952169653662191,
      "grad_norm": 0.721581220626831,
      "learning_rate": 5.987641429065661e-08,
      "loss": 0.6682,
      "step": 33046
    },
    {
      "epoch": 0.9521984671238403,
      "grad_norm": 0.719346821308136,
      "learning_rate": 5.980443987371465e-08,
      "loss": 0.671,
      "step": 33047
    },
    {
      "epoch": 0.9522272805854896,
      "grad_norm": 0.7168618440628052,
      "learning_rate": 5.973250848093748e-08,
      "loss": 0.6695,
      "step": 33048
    },
    {
      "epoch": 0.9522560940471388,
      "grad_norm": 0.7036823034286499,
      "learning_rate": 5.966062011295182e-08,
      "loss": 0.6764,
      "step": 33049
    },
    {
      "epoch": 0.9522849075087881,
      "grad_norm": 0.7235015034675598,
      "learning_rate": 5.9588774770383274e-08,
      "loss": 0.6634,
      "step": 33050
    },
    {
      "epoch": 0.9523137209704374,
      "grad_norm": 0.7112457752227783,
      "learning_rate": 5.951697245385746e-08,
      "loss": 0.6834,
      "step": 33051
    },
    {
      "epoch": 0.9523425344320867,
      "grad_norm": 0.7337682843208313,
      "learning_rate": 5.944521316400109e-08,
      "loss": 0.6854,
      "step": 33052
    },
    {
      "epoch": 0.952371347893736,
      "grad_norm": 0.7422839999198914,
      "learning_rate": 5.937349690143757e-08,
      "loss": 0.7086,
      "step": 33053
    },
    {
      "epoch": 0.9524001613553852,
      "grad_norm": 0.7051260471343994,
      "learning_rate": 5.930182366679249e-08,
      "loss": 0.6861,
      "step": 33054
    },
    {
      "epoch": 0.9524289748170345,
      "grad_norm": 0.7304602861404419,
      "learning_rate": 5.9230193460689256e-08,
      "loss": 0.6966,
      "step": 33055
    },
    {
      "epoch": 0.9524577882786838,
      "grad_norm": 0.7225950360298157,
      "learning_rate": 5.9158606283752365e-08,
      "loss": 0.7058,
      "step": 33056
    },
    {
      "epoch": 0.9524866017403331,
      "grad_norm": 0.7344672679901123,
      "learning_rate": 5.908706213660465e-08,
      "loss": 0.6821,
      "step": 33057
    },
    {
      "epoch": 0.9525154152019824,
      "grad_norm": 0.7130724191665649,
      "learning_rate": 5.901556101987005e-08,
      "loss": 0.6576,
      "step": 33058
    },
    {
      "epoch": 0.9525442286636316,
      "grad_norm": 0.7126675844192505,
      "learning_rate": 5.89441029341703e-08,
      "loss": 0.6775,
      "step": 33059
    },
    {
      "epoch": 0.9525730421252809,
      "grad_norm": 0.7272191643714905,
      "learning_rate": 5.8872687880128235e-08,
      "loss": 0.6915,
      "step": 33060
    },
    {
      "epoch": 0.9526018555869302,
      "grad_norm": 0.7253807187080383,
      "learning_rate": 5.880131585836557e-08,
      "loss": 0.6789,
      "step": 33061
    },
    {
      "epoch": 0.9526306690485795,
      "grad_norm": 0.7208541631698608,
      "learning_rate": 5.87299868695046e-08,
      "loss": 0.6723,
      "step": 33062
    },
    {
      "epoch": 0.9526594825102288,
      "grad_norm": 0.8147711753845215,
      "learning_rate": 5.865870091416537e-08,
      "loss": 0.6671,
      "step": 33063
    },
    {
      "epoch": 0.9526882959718781,
      "grad_norm": 0.7140071392059326,
      "learning_rate": 5.858745799296961e-08,
      "loss": 0.6704,
      "step": 33064
    },
    {
      "epoch": 0.9527171094335274,
      "grad_norm": 0.7206241488456726,
      "learning_rate": 5.8516258106537384e-08,
      "loss": 0.6745,
      "step": 33065
    },
    {
      "epoch": 0.9527459228951767,
      "grad_norm": 0.7083787322044373,
      "learning_rate": 5.84451012554893e-08,
      "loss": 0.6879,
      "step": 33066
    },
    {
      "epoch": 0.9527747363568259,
      "grad_norm": 0.7221071124076843,
      "learning_rate": 5.8373987440443755e-08,
      "loss": 0.6962,
      "step": 33067
    },
    {
      "epoch": 0.9528035498184751,
      "grad_norm": 0.708467960357666,
      "learning_rate": 5.830291666202192e-08,
      "loss": 0.6916,
      "step": 33068
    },
    {
      "epoch": 0.9528323632801244,
      "grad_norm": 0.7267075777053833,
      "learning_rate": 5.823188892084108e-08,
      "loss": 0.6797,
      "step": 33069
    },
    {
      "epoch": 0.9528611767417737,
      "grad_norm": 0.7179749608039856,
      "learning_rate": 5.81609042175213e-08,
      "loss": 0.6935,
      "step": 33070
    },
    {
      "epoch": 0.952889990203423,
      "grad_norm": 0.7087607979774475,
      "learning_rate": 5.808996255267929e-08,
      "loss": 0.6708,
      "step": 33071
    },
    {
      "epoch": 0.9529188036650723,
      "grad_norm": 0.729672372341156,
      "learning_rate": 5.801906392693346e-08,
      "loss": 0.7129,
      "step": 33072
    },
    {
      "epoch": 0.9529476171267216,
      "grad_norm": 0.727185845375061,
      "learning_rate": 5.7948208340902204e-08,
      "loss": 0.6767,
      "step": 33073
    },
    {
      "epoch": 0.9529764305883709,
      "grad_norm": 0.7541782855987549,
      "learning_rate": 5.787739579520113e-08,
      "loss": 0.6697,
      "step": 33074
    },
    {
      "epoch": 0.9530052440500202,
      "grad_norm": 0.7419357895851135,
      "learning_rate": 5.780662629044809e-08,
      "loss": 0.6943,
      "step": 33075
    },
    {
      "epoch": 0.9530340575116695,
      "grad_norm": 0.7275808453559875,
      "learning_rate": 5.773589982725869e-08,
      "loss": 0.6721,
      "step": 33076
    },
    {
      "epoch": 0.9530628709733188,
      "grad_norm": 0.7268728017807007,
      "learning_rate": 5.7665216406249115e-08,
      "loss": 0.6834,
      "step": 33077
    },
    {
      "epoch": 0.9530916844349681,
      "grad_norm": 0.7266389727592468,
      "learning_rate": 5.7594576028034975e-08,
      "loss": 0.6671,
      "step": 33078
    },
    {
      "epoch": 0.9531204978966173,
      "grad_norm": 0.7282724380493164,
      "learning_rate": 5.7523978693231893e-08,
      "loss": 0.6813,
      "step": 33079
    },
    {
      "epoch": 0.9531493113582665,
      "grad_norm": 0.7214998006820679,
      "learning_rate": 5.745342440245383e-08,
      "loss": 0.6818,
      "step": 33080
    },
    {
      "epoch": 0.9531781248199158,
      "grad_norm": 0.7314707636833191,
      "learning_rate": 5.7382913156315835e-08,
      "loss": 0.6664,
      "step": 33081
    },
    {
      "epoch": 0.9532069382815651,
      "grad_norm": 0.7328141331672668,
      "learning_rate": 5.731244495543187e-08,
      "loss": 0.6707,
      "step": 33082
    },
    {
      "epoch": 0.9532357517432144,
      "grad_norm": 0.7066517472267151,
      "learning_rate": 5.724201980041588e-08,
      "loss": 0.6622,
      "step": 33083
    },
    {
      "epoch": 0.9532645652048637,
      "grad_norm": 0.7228797078132629,
      "learning_rate": 5.717163769188072e-08,
      "loss": 0.6817,
      "step": 33084
    },
    {
      "epoch": 0.953293378666513,
      "grad_norm": 0.7166235446929932,
      "learning_rate": 5.7101298630439785e-08,
      "loss": 0.6568,
      "step": 33085
    },
    {
      "epoch": 0.9533221921281623,
      "grad_norm": 0.7294449210166931,
      "learning_rate": 5.70310026167048e-08,
      "loss": 0.6904,
      "step": 33086
    },
    {
      "epoch": 0.9533510055898116,
      "grad_norm": 0.7236230373382568,
      "learning_rate": 5.696074965128917e-08,
      "loss": 0.6837,
      "step": 33087
    },
    {
      "epoch": 0.9533798190514609,
      "grad_norm": 0.7084173560142517,
      "learning_rate": 5.689053973480407e-08,
      "loss": 0.6597,
      "step": 33088
    },
    {
      "epoch": 0.9534086325131101,
      "grad_norm": 0.7121330499649048,
      "learning_rate": 5.682037286786124e-08,
      "loss": 0.6697,
      "step": 33089
    },
    {
      "epoch": 0.9534374459747594,
      "grad_norm": 0.7408294081687927,
      "learning_rate": 5.6750249051071294e-08,
      "loss": 0.706,
      "step": 33090
    },
    {
      "epoch": 0.9534662594364087,
      "grad_norm": 0.7190052270889282,
      "learning_rate": 5.668016828504541e-08,
      "loss": 0.6826,
      "step": 33091
    },
    {
      "epoch": 0.953495072898058,
      "grad_norm": 0.719730794429779,
      "learning_rate": 5.6610130570393665e-08,
      "loss": 0.6905,
      "step": 33092
    },
    {
      "epoch": 0.9535238863597073,
      "grad_norm": 0.7301696538925171,
      "learning_rate": 5.6540135907726116e-08,
      "loss": 0.6882,
      "step": 33093
    },
    {
      "epoch": 0.9535526998213565,
      "grad_norm": 0.7281901836395264,
      "learning_rate": 5.6470184297652276e-08,
      "loss": 0.6894,
      "step": 33094
    },
    {
      "epoch": 0.9535815132830058,
      "grad_norm": 0.7263402938842773,
      "learning_rate": 5.640027574078111e-08,
      "loss": 0.6558,
      "step": 33095
    },
    {
      "epoch": 0.9536103267446551,
      "grad_norm": 0.7019894123077393,
      "learning_rate": 5.6330410237722676e-08,
      "loss": 0.6861,
      "step": 33096
    },
    {
      "epoch": 0.9536391402063044,
      "grad_norm": 0.7175272107124329,
      "learning_rate": 5.626058778908372e-08,
      "loss": 0.6651,
      "step": 33097
    },
    {
      "epoch": 0.9536679536679536,
      "grad_norm": 0.7129192352294922,
      "learning_rate": 5.619080839547375e-08,
      "loss": 0.6929,
      "step": 33098
    },
    {
      "epoch": 0.9536967671296029,
      "grad_norm": 0.706368088722229,
      "learning_rate": 5.6121072057498946e-08,
      "loss": 0.6401,
      "step": 33099
    },
    {
      "epoch": 0.9537255805912522,
      "grad_norm": 0.7406952381134033,
      "learning_rate": 5.605137877576827e-08,
      "loss": 0.6794,
      "step": 33100
    },
    {
      "epoch": 0.9537543940529015,
      "grad_norm": 0.7271577715873718,
      "learning_rate": 5.598172855088735e-08,
      "loss": 0.6885,
      "step": 33101
    },
    {
      "epoch": 0.9537832075145508,
      "grad_norm": 0.716860294342041,
      "learning_rate": 5.5912121383464025e-08,
      "loss": 0.6916,
      "step": 33102
    },
    {
      "epoch": 0.9538120209762001,
      "grad_norm": 0.7170162200927734,
      "learning_rate": 5.584255727410337e-08,
      "loss": 0.6874,
      "step": 33103
    },
    {
      "epoch": 0.9538408344378494,
      "grad_norm": 0.7307026386260986,
      "learning_rate": 5.5773036223412124e-08,
      "loss": 0.6732,
      "step": 33104
    },
    {
      "epoch": 0.9538696478994987,
      "grad_norm": 0.7570686936378479,
      "learning_rate": 5.5703558231994805e-08,
      "loss": 0.6831,
      "step": 33105
    },
    {
      "epoch": 0.953898461361148,
      "grad_norm": 0.7106505036354065,
      "learning_rate": 5.5634123300457585e-08,
      "loss": 0.6759,
      "step": 33106
    },
    {
      "epoch": 0.9539272748227972,
      "grad_norm": 0.7177777290344238,
      "learning_rate": 5.556473142940389e-08,
      "loss": 0.68,
      "step": 33107
    },
    {
      "epoch": 0.9539560882844464,
      "grad_norm": 0.7324459552764893,
      "learning_rate": 5.549538261943932e-08,
      "loss": 0.6926,
      "step": 33108
    },
    {
      "epoch": 0.9539849017460957,
      "grad_norm": 0.7088369131088257,
      "learning_rate": 5.542607687116674e-08,
      "loss": 0.6624,
      "step": 33109
    },
    {
      "epoch": 0.954013715207745,
      "grad_norm": 0.7130725383758545,
      "learning_rate": 5.5356814185190675e-08,
      "loss": 0.6814,
      "step": 33110
    },
    {
      "epoch": 0.9540425286693943,
      "grad_norm": 0.7230494618415833,
      "learning_rate": 5.528759456211397e-08,
      "loss": 0.6949,
      "step": 33111
    },
    {
      "epoch": 0.9540713421310436,
      "grad_norm": 0.7432096004486084,
      "learning_rate": 5.5218418002539464e-08,
      "loss": 0.7107,
      "step": 33112
    },
    {
      "epoch": 0.9541001555926929,
      "grad_norm": 0.7139397263526917,
      "learning_rate": 5.514928450706891e-08,
      "loss": 0.6914,
      "step": 33113
    },
    {
      "epoch": 0.9541289690543422,
      "grad_norm": 0.7463012933731079,
      "learning_rate": 5.5080194076305714e-08,
      "loss": 0.6794,
      "step": 33114
    },
    {
      "epoch": 0.9541577825159915,
      "grad_norm": 0.7276775240898132,
      "learning_rate": 5.501114671085106e-08,
      "loss": 0.6839,
      "step": 33115
    },
    {
      "epoch": 0.9541865959776408,
      "grad_norm": 0.7008837461471558,
      "learning_rate": 5.4942142411305576e-08,
      "loss": 0.6509,
      "step": 33116
    },
    {
      "epoch": 0.9542154094392901,
      "grad_norm": 0.7396460771560669,
      "learning_rate": 5.487318117827101e-08,
      "loss": 0.6943,
      "step": 33117
    },
    {
      "epoch": 0.9542442229009394,
      "grad_norm": 0.7827730774879456,
      "learning_rate": 5.4804263012347424e-08,
      "loss": 0.6699,
      "step": 33118
    },
    {
      "epoch": 0.9542730363625886,
      "grad_norm": 0.7265175580978394,
      "learning_rate": 5.473538791413602e-08,
      "loss": 0.6922,
      "step": 33119
    },
    {
      "epoch": 0.9543018498242379,
      "grad_norm": 0.7193102836608887,
      "learning_rate": 5.466655588423575e-08,
      "loss": 0.6942,
      "step": 33120
    },
    {
      "epoch": 0.9543306632858871,
      "grad_norm": 0.7185333967208862,
      "learning_rate": 5.459776692324614e-08,
      "loss": 0.6713,
      "step": 33121
    },
    {
      "epoch": 0.9543594767475364,
      "grad_norm": 0.7486719489097595,
      "learning_rate": 5.452902103176616e-08,
      "loss": 0.6725,
      "step": 33122
    },
    {
      "epoch": 0.9543882902091857,
      "grad_norm": 0.7277662754058838,
      "learning_rate": 5.4460318210394767e-08,
      "loss": 0.6774,
      "step": 33123
    },
    {
      "epoch": 0.954417103670835,
      "grad_norm": 0.7069145441055298,
      "learning_rate": 5.4391658459729824e-08,
      "loss": 0.6823,
      "step": 33124
    },
    {
      "epoch": 0.9544459171324843,
      "grad_norm": 0.729612410068512,
      "learning_rate": 5.4323041780370845e-08,
      "loss": 0.673,
      "step": 33125
    },
    {
      "epoch": 0.9544747305941336,
      "grad_norm": 0.7177229523658752,
      "learning_rate": 5.4254468172913466e-08,
      "loss": 0.692,
      "step": 33126
    },
    {
      "epoch": 0.9545035440557829,
      "grad_norm": 0.7587096691131592,
      "learning_rate": 5.41859376379561e-08,
      "loss": 0.6851,
      "step": 33127
    },
    {
      "epoch": 0.9545323575174322,
      "grad_norm": 0.7388092875480652,
      "learning_rate": 5.411745017609493e-08,
      "loss": 0.6906,
      "step": 33128
    },
    {
      "epoch": 0.9545611709790814,
      "grad_norm": 0.7347590327262878,
      "learning_rate": 5.404900578792727e-08,
      "loss": 0.6674,
      "step": 33129
    },
    {
      "epoch": 0.9545899844407307,
      "grad_norm": 0.722995936870575,
      "learning_rate": 5.398060447404818e-08,
      "loss": 0.6714,
      "step": 33130
    },
    {
      "epoch": 0.95461879790238,
      "grad_norm": 0.7205703854560852,
      "learning_rate": 5.3912246235053865e-08,
      "loss": 0.6654,
      "step": 33131
    },
    {
      "epoch": 0.9546476113640293,
      "grad_norm": 0.7435362935066223,
      "learning_rate": 5.3843931071539956e-08,
      "loss": 0.7033,
      "step": 33132
    },
    {
      "epoch": 0.9546764248256786,
      "grad_norm": 0.7002341747283936,
      "learning_rate": 5.377565898410042e-08,
      "loss": 0.6624,
      "step": 33133
    },
    {
      "epoch": 0.9547052382873278,
      "grad_norm": 0.758496105670929,
      "learning_rate": 5.3707429973330894e-08,
      "loss": 0.6721,
      "step": 33134
    },
    {
      "epoch": 0.9547340517489771,
      "grad_norm": 0.7313849329948425,
      "learning_rate": 5.363924403982479e-08,
      "loss": 0.6928,
      "step": 33135
    },
    {
      "epoch": 0.9547628652106264,
      "grad_norm": 0.7023991346359253,
      "learning_rate": 5.3571101184176633e-08,
      "loss": 0.6605,
      "step": 33136
    },
    {
      "epoch": 0.9547916786722757,
      "grad_norm": 0.7278786301612854,
      "learning_rate": 5.3503001406979283e-08,
      "loss": 0.6832,
      "step": 33137
    },
    {
      "epoch": 0.954820492133925,
      "grad_norm": 0.7333886027336121,
      "learning_rate": 5.343494470882671e-08,
      "loss": 0.6889,
      "step": 33138
    },
    {
      "epoch": 0.9548493055955742,
      "grad_norm": 0.7314276099205017,
      "learning_rate": 5.336693109031066e-08,
      "loss": 0.6902,
      "step": 33139
    },
    {
      "epoch": 0.9548781190572235,
      "grad_norm": 0.7105659246444702,
      "learning_rate": 5.329896055202399e-08,
      "loss": 0.6614,
      "step": 33140
    },
    {
      "epoch": 0.9549069325188728,
      "grad_norm": 0.7218896746635437,
      "learning_rate": 5.3231033094557904e-08,
      "loss": 0.6866,
      "step": 33141
    },
    {
      "epoch": 0.9549357459805221,
      "grad_norm": 0.724628210067749,
      "learning_rate": 5.316314871850525e-08,
      "loss": 0.6894,
      "step": 33142
    },
    {
      "epoch": 0.9549645594421714,
      "grad_norm": 0.7315995693206787,
      "learning_rate": 5.309530742445612e-08,
      "loss": 0.6948,
      "step": 33143
    },
    {
      "epoch": 0.9549933729038207,
      "grad_norm": 0.7222864031791687,
      "learning_rate": 5.302750921300226e-08,
      "loss": 0.6915,
      "step": 33144
    },
    {
      "epoch": 0.95502218636547,
      "grad_norm": 0.716228723526001,
      "learning_rate": 5.295975408473319e-08,
      "loss": 0.6658,
      "step": 33145
    },
    {
      "epoch": 0.9550509998271193,
      "grad_norm": 0.7275366187095642,
      "learning_rate": 5.2892042040239566e-08,
      "loss": 0.6695,
      "step": 33146
    },
    {
      "epoch": 0.9550798132887686,
      "grad_norm": 0.7216625213623047,
      "learning_rate": 5.28243730801109e-08,
      "loss": 0.6802,
      "step": 33147
    },
    {
      "epoch": 0.9551086267504177,
      "grad_norm": 0.7304631471633911,
      "learning_rate": 5.275674720493673e-08,
      "loss": 0.6847,
      "step": 33148
    },
    {
      "epoch": 0.955137440212067,
      "grad_norm": 0.7172819375991821,
      "learning_rate": 5.268916441530547e-08,
      "loss": 0.6758,
      "step": 33149
    },
    {
      "epoch": 0.9551662536737163,
      "grad_norm": 0.740520179271698,
      "learning_rate": 5.262162471180665e-08,
      "loss": 0.6792,
      "step": 33150
    },
    {
      "epoch": 0.9551950671353656,
      "grad_norm": 0.7110050320625305,
      "learning_rate": 5.255412809502758e-08,
      "loss": 0.6706,
      "step": 33151
    },
    {
      "epoch": 0.9552238805970149,
      "grad_norm": 0.7256067991256714,
      "learning_rate": 5.248667456555612e-08,
      "loss": 0.6886,
      "step": 33152
    },
    {
      "epoch": 0.9552526940586642,
      "grad_norm": 0.7322537302970886,
      "learning_rate": 5.241926412398068e-08,
      "loss": 0.6849,
      "step": 33153
    },
    {
      "epoch": 0.9552815075203135,
      "grad_norm": 0.7403976917266846,
      "learning_rate": 5.235189677088692e-08,
      "loss": 0.702,
      "step": 33154
    },
    {
      "epoch": 0.9553103209819628,
      "grad_norm": 0.7256758213043213,
      "learning_rate": 5.228457250686325e-08,
      "loss": 0.6891,
      "step": 33155
    },
    {
      "epoch": 0.9553391344436121,
      "grad_norm": 0.7209693789482117,
      "learning_rate": 5.22172913324942e-08,
      "loss": 0.6911,
      "step": 33156
    },
    {
      "epoch": 0.9553679479052614,
      "grad_norm": 0.7145487666130066,
      "learning_rate": 5.215005324836708e-08,
      "loss": 0.6807,
      "step": 33157
    },
    {
      "epoch": 0.9553967613669107,
      "grad_norm": 0.7460010051727295,
      "learning_rate": 5.208285825506643e-08,
      "loss": 0.6668,
      "step": 33158
    },
    {
      "epoch": 0.9554255748285599,
      "grad_norm": 0.729986846446991,
      "learning_rate": 5.201570635317843e-08,
      "loss": 0.6883,
      "step": 33159
    },
    {
      "epoch": 0.9554543882902092,
      "grad_norm": 0.730023980140686,
      "learning_rate": 5.194859754328707e-08,
      "loss": 0.6774,
      "step": 33160
    },
    {
      "epoch": 0.9554832017518584,
      "grad_norm": 0.7045608758926392,
      "learning_rate": 5.188153182597799e-08,
      "loss": 0.6668,
      "step": 33161
    },
    {
      "epoch": 0.9555120152135077,
      "grad_norm": 0.7421419024467468,
      "learning_rate": 5.181450920183351e-08,
      "loss": 0.665,
      "step": 33162
    },
    {
      "epoch": 0.955540828675157,
      "grad_norm": 0.7167927622795105,
      "learning_rate": 5.17475296714387e-08,
      "loss": 0.6927,
      "step": 33163
    },
    {
      "epoch": 0.9555696421368063,
      "grad_norm": 0.7282055020332336,
      "learning_rate": 5.1680593235376444e-08,
      "loss": 0.676,
      "step": 33164
    },
    {
      "epoch": 0.9555984555984556,
      "grad_norm": 0.7222775816917419,
      "learning_rate": 5.1613699894230154e-08,
      "loss": 0.6604,
      "step": 33165
    },
    {
      "epoch": 0.9556272690601049,
      "grad_norm": 0.7222873568534851,
      "learning_rate": 5.1546849648581035e-08,
      "loss": 0.6592,
      "step": 33166
    },
    {
      "epoch": 0.9556560825217542,
      "grad_norm": 0.7382300496101379,
      "learning_rate": 5.148004249901306e-08,
      "loss": 0.6966,
      "step": 33167
    },
    {
      "epoch": 0.9556848959834034,
      "grad_norm": 0.7202795743942261,
      "learning_rate": 5.141327844610688e-08,
      "loss": 0.6699,
      "step": 33168
    },
    {
      "epoch": 0.9557137094450527,
      "grad_norm": 0.7151008248329163,
      "learning_rate": 5.134655749044426e-08,
      "loss": 0.6736,
      "step": 33169
    },
    {
      "epoch": 0.955742522906702,
      "grad_norm": 0.7077386975288391,
      "learning_rate": 5.127987963260583e-08,
      "loss": 0.7067,
      "step": 33170
    },
    {
      "epoch": 0.9557713363683513,
      "grad_norm": 0.7190730571746826,
      "learning_rate": 5.1213244873173365e-08,
      "loss": 0.6514,
      "step": 33171
    },
    {
      "epoch": 0.9558001498300006,
      "grad_norm": 0.7199223041534424,
      "learning_rate": 5.114665321272638e-08,
      "loss": 0.7067,
      "step": 33172
    },
    {
      "epoch": 0.9558289632916499,
      "grad_norm": 0.7202188372612,
      "learning_rate": 5.108010465184499e-08,
      "loss": 0.6826,
      "step": 33173
    },
    {
      "epoch": 0.9558577767532992,
      "grad_norm": 0.7168198823928833,
      "learning_rate": 5.101359919110871e-08,
      "loss": 0.6643,
      "step": 33174
    },
    {
      "epoch": 0.9558865902149484,
      "grad_norm": 0.7148205637931824,
      "learning_rate": 5.094713683109709e-08,
      "loss": 0.6594,
      "step": 33175
    },
    {
      "epoch": 0.9559154036765977,
      "grad_norm": 0.7315611243247986,
      "learning_rate": 5.088071757238855e-08,
      "loss": 0.6847,
      "step": 33176
    },
    {
      "epoch": 0.955944217138247,
      "grad_norm": 0.7193521857261658,
      "learning_rate": 5.081434141556152e-08,
      "loss": 0.6651,
      "step": 33177
    },
    {
      "epoch": 0.9559730305998962,
      "grad_norm": 0.7431074976921082,
      "learning_rate": 5.0748008361194425e-08,
      "loss": 0.6814,
      "step": 33178
    },
    {
      "epoch": 0.9560018440615455,
      "grad_norm": 0.7103232145309448,
      "learning_rate": 5.068171840986402e-08,
      "loss": 0.659,
      "step": 33179
    },
    {
      "epoch": 0.9560306575231948,
      "grad_norm": 0.7072264552116394,
      "learning_rate": 5.06154715621493e-08,
      "loss": 0.6946,
      "step": 33180
    },
    {
      "epoch": 0.9560594709848441,
      "grad_norm": 0.7262304425239563,
      "learning_rate": 5.0549267818625905e-08,
      "loss": 0.683,
      "step": 33181
    },
    {
      "epoch": 0.9560882844464934,
      "grad_norm": 0.7241138219833374,
      "learning_rate": 5.04831071798706e-08,
      "loss": 0.6852,
      "step": 33182
    },
    {
      "epoch": 0.9561170979081427,
      "grad_norm": 0.7225103974342346,
      "learning_rate": 5.041698964645958e-08,
      "loss": 0.6713,
      "step": 33183
    },
    {
      "epoch": 0.956145911369792,
      "grad_norm": 0.7456777095794678,
      "learning_rate": 5.035091521896906e-08,
      "loss": 0.6701,
      "step": 33184
    },
    {
      "epoch": 0.9561747248314413,
      "grad_norm": 0.889451265335083,
      "learning_rate": 5.0284883897974146e-08,
      "loss": 0.6586,
      "step": 33185
    },
    {
      "epoch": 0.9562035382930906,
      "grad_norm": 0.7409698367118835,
      "learning_rate": 5.0218895684049916e-08,
      "loss": 0.6771,
      "step": 33186
    },
    {
      "epoch": 0.9562323517547399,
      "grad_norm": 0.7171009182929993,
      "learning_rate": 5.015295057777147e-08,
      "loss": 0.6663,
      "step": 33187
    },
    {
      "epoch": 0.956261165216389,
      "grad_norm": 0.7319186329841614,
      "learning_rate": 5.00870485797128e-08,
      "loss": 0.685,
      "step": 33188
    },
    {
      "epoch": 0.9562899786780383,
      "grad_norm": 0.7066362500190735,
      "learning_rate": 5.002118969044734e-08,
      "loss": 0.673,
      "step": 33189
    },
    {
      "epoch": 0.9563187921396876,
      "grad_norm": 0.7338864803314209,
      "learning_rate": 4.995537391054961e-08,
      "loss": 0.6885,
      "step": 33190
    },
    {
      "epoch": 0.9563476056013369,
      "grad_norm": 0.7489532232284546,
      "learning_rate": 4.988960124059195e-08,
      "loss": 0.6847,
      "step": 33191
    },
    {
      "epoch": 0.9563764190629862,
      "grad_norm": 0.7316240668296814,
      "learning_rate": 4.982387168114777e-08,
      "loss": 0.6795,
      "step": 33192
    },
    {
      "epoch": 0.9564052325246355,
      "grad_norm": 0.7507811784744263,
      "learning_rate": 4.9758185232789414e-08,
      "loss": 0.6931,
      "step": 33193
    },
    {
      "epoch": 0.9564340459862848,
      "grad_norm": 0.7171459794044495,
      "learning_rate": 4.969254189608863e-08,
      "loss": 0.6636,
      "step": 33194
    },
    {
      "epoch": 0.9564628594479341,
      "grad_norm": 0.7066651582717896,
      "learning_rate": 4.962694167161719e-08,
      "loss": 0.6619,
      "step": 33195
    },
    {
      "epoch": 0.9564916729095834,
      "grad_norm": 0.734440267086029,
      "learning_rate": 4.956138455994686e-08,
      "loss": 0.6925,
      "step": 33196
    },
    {
      "epoch": 0.9565204863712327,
      "grad_norm": 0.7212350964546204,
      "learning_rate": 4.9495870561648284e-08,
      "loss": 0.6707,
      "step": 33197
    },
    {
      "epoch": 0.956549299832882,
      "grad_norm": 0.7256790995597839,
      "learning_rate": 4.943039967729157e-08,
      "loss": 0.6996,
      "step": 33198
    },
    {
      "epoch": 0.9565781132945312,
      "grad_norm": 0.729870617389679,
      "learning_rate": 4.9364971907447936e-08,
      "loss": 0.6722,
      "step": 33199
    },
    {
      "epoch": 0.9566069267561805,
      "grad_norm": 0.735731303691864,
      "learning_rate": 4.92995872526858e-08,
      "loss": 0.6701,
      "step": 33200
    },
    {
      "epoch": 0.9566357402178298,
      "grad_norm": 0.7295433878898621,
      "learning_rate": 4.9234245713575824e-08,
      "loss": 0.6665,
      "step": 33201
    },
    {
      "epoch": 0.956664553679479,
      "grad_norm": 0.7248483300209045,
      "learning_rate": 4.916894729068644e-08,
      "loss": 0.6782,
      "step": 33202
    },
    {
      "epoch": 0.9566933671411283,
      "grad_norm": 0.7485484480857849,
      "learning_rate": 4.91036919845872e-08,
      "loss": 0.6894,
      "step": 33203
    },
    {
      "epoch": 0.9567221806027776,
      "grad_norm": 0.7327790856361389,
      "learning_rate": 4.903847979584486e-08,
      "loss": 0.6776,
      "step": 33204
    },
    {
      "epoch": 0.9567509940644269,
      "grad_norm": 0.7250493764877319,
      "learning_rate": 4.897331072502842e-08,
      "loss": 0.6834,
      "step": 33205
    },
    {
      "epoch": 0.9567798075260762,
      "grad_norm": 0.7002475261688232,
      "learning_rate": 4.89081847727052e-08,
      "loss": 0.6444,
      "step": 33206
    },
    {
      "epoch": 0.9568086209877255,
      "grad_norm": 0.7202341556549072,
      "learning_rate": 4.8843101939442526e-08,
      "loss": 0.6938,
      "step": 33207
    },
    {
      "epoch": 0.9568374344493747,
      "grad_norm": 0.7049057483673096,
      "learning_rate": 4.8778062225807167e-08,
      "loss": 0.663,
      "step": 33208
    },
    {
      "epoch": 0.956866247911024,
      "grad_norm": 0.7246479392051697,
      "learning_rate": 4.871306563236533e-08,
      "loss": 0.672,
      "step": 33209
    },
    {
      "epoch": 0.9568950613726733,
      "grad_norm": 0.7033329606056213,
      "learning_rate": 4.8648112159683236e-08,
      "loss": 0.6672,
      "step": 33210
    },
    {
      "epoch": 0.9569238748343226,
      "grad_norm": 0.724232017993927,
      "learning_rate": 4.858320180832654e-08,
      "loss": 0.7094,
      "step": 33211
    },
    {
      "epoch": 0.9569526882959719,
      "grad_norm": 0.7152987122535706,
      "learning_rate": 4.851833457886035e-08,
      "loss": 0.6572,
      "step": 33212
    },
    {
      "epoch": 0.9569815017576212,
      "grad_norm": 0.7246092557907104,
      "learning_rate": 4.845351047185032e-08,
      "loss": 0.6865,
      "step": 33213
    },
    {
      "epoch": 0.9570103152192705,
      "grad_norm": 0.7121405601501465,
      "learning_rate": 4.8388729487860445e-08,
      "loss": 0.6986,
      "step": 33214
    },
    {
      "epoch": 0.9570391286809197,
      "grad_norm": 0.7245972752571106,
      "learning_rate": 4.832399162745416e-08,
      "loss": 0.6741,
      "step": 33215
    },
    {
      "epoch": 0.957067942142569,
      "grad_norm": 0.7168328762054443,
      "learning_rate": 4.825929689119713e-08,
      "loss": 0.6849,
      "step": 33216
    },
    {
      "epoch": 0.9570967556042183,
      "grad_norm": 0.7072744369506836,
      "learning_rate": 4.8194645279650566e-08,
      "loss": 0.6656,
      "step": 33217
    },
    {
      "epoch": 0.9571255690658675,
      "grad_norm": 0.7291720509529114,
      "learning_rate": 4.813003679337958e-08,
      "loss": 0.6575,
      "step": 33218
    },
    {
      "epoch": 0.9571543825275168,
      "grad_norm": 0.7133007049560547,
      "learning_rate": 4.8065471432945375e-08,
      "loss": 0.6894,
      "step": 33219
    },
    {
      "epoch": 0.9571831959891661,
      "grad_norm": 0.7289758324623108,
      "learning_rate": 4.800094919891085e-08,
      "loss": 0.6816,
      "step": 33220
    },
    {
      "epoch": 0.9572120094508154,
      "grad_norm": 0.7411667108535767,
      "learning_rate": 4.793647009183777e-08,
      "loss": 0.7002,
      "step": 33221
    },
    {
      "epoch": 0.9572408229124647,
      "grad_norm": 0.7276767492294312,
      "learning_rate": 4.787203411228847e-08,
      "loss": 0.6683,
      "step": 33222
    },
    {
      "epoch": 0.957269636374114,
      "grad_norm": 0.7515687942504883,
      "learning_rate": 4.780764126082249e-08,
      "loss": 0.6811,
      "step": 33223
    },
    {
      "epoch": 0.9572984498357633,
      "grad_norm": 0.7160674333572388,
      "learning_rate": 4.774329153800272e-08,
      "loss": 0.678,
      "step": 33224
    },
    {
      "epoch": 0.9573272632974126,
      "grad_norm": 0.7150905728340149,
      "learning_rate": 4.7678984944387607e-08,
      "loss": 0.6707,
      "step": 33225
    },
    {
      "epoch": 0.9573560767590619,
      "grad_norm": 0.7255259156227112,
      "learning_rate": 4.761472148053836e-08,
      "loss": 0.6671,
      "step": 33226
    },
    {
      "epoch": 0.9573848902207112,
      "grad_norm": 0.7214296460151672,
      "learning_rate": 4.7550501147013986e-08,
      "loss": 0.6811,
      "step": 33227
    },
    {
      "epoch": 0.9574137036823605,
      "grad_norm": 0.7179398536682129,
      "learning_rate": 4.748632394437458e-08,
      "loss": 0.6608,
      "step": 33228
    },
    {
      "epoch": 0.9574425171440096,
      "grad_norm": 0.727565348148346,
      "learning_rate": 4.742218987317804e-08,
      "loss": 0.6901,
      "step": 33229
    },
    {
      "epoch": 0.9574713306056589,
      "grad_norm": 0.7297376990318298,
      "learning_rate": 4.735809893398335e-08,
      "loss": 0.6774,
      "step": 33230
    },
    {
      "epoch": 0.9575001440673082,
      "grad_norm": 0.7447986006736755,
      "learning_rate": 4.7294051127348975e-08,
      "loss": 0.6785,
      "step": 33231
    },
    {
      "epoch": 0.9575289575289575,
      "grad_norm": 0.7179223299026489,
      "learning_rate": 4.723004645383278e-08,
      "loss": 0.6992,
      "step": 33232
    },
    {
      "epoch": 0.9575577709906068,
      "grad_norm": 0.7203055024147034,
      "learning_rate": 4.716608491399211e-08,
      "loss": 0.694,
      "step": 33233
    },
    {
      "epoch": 0.9575865844522561,
      "grad_norm": 0.7292795777320862,
      "learning_rate": 4.7102166508383173e-08,
      "loss": 0.6817,
      "step": 33234
    },
    {
      "epoch": 0.9576153979139054,
      "grad_norm": 0.7231529951095581,
      "learning_rate": 4.7038291237563874e-08,
      "loss": 0.662,
      "step": 33235
    },
    {
      "epoch": 0.9576442113755547,
      "grad_norm": 0.7182670831680298,
      "learning_rate": 4.6974459102089306e-08,
      "loss": 0.6681,
      "step": 33236
    },
    {
      "epoch": 0.957673024837204,
      "grad_norm": 0.7213160395622253,
      "learning_rate": 4.691067010251682e-08,
      "loss": 0.6708,
      "step": 33237
    },
    {
      "epoch": 0.9577018382988532,
      "grad_norm": 0.7308720946311951,
      "learning_rate": 4.684692423940041e-08,
      "loss": 0.6868,
      "step": 33238
    },
    {
      "epoch": 0.9577306517605025,
      "grad_norm": 0.7230201959609985,
      "learning_rate": 4.6783221513296285e-08,
      "loss": 0.6666,
      "step": 33239
    },
    {
      "epoch": 0.9577594652221518,
      "grad_norm": 0.7014429569244385,
      "learning_rate": 4.671956192475901e-08,
      "loss": 0.6817,
      "step": 33240
    },
    {
      "epoch": 0.9577882786838011,
      "grad_norm": 0.7105153203010559,
      "learning_rate": 4.66559454743426e-08,
      "loss": 0.6954,
      "step": 33241
    },
    {
      "epoch": 0.9578170921454503,
      "grad_norm": 0.7334997057914734,
      "learning_rate": 4.659237216260104e-08,
      "loss": 0.6828,
      "step": 33242
    },
    {
      "epoch": 0.9578459056070996,
      "grad_norm": 0.7492963075637817,
      "learning_rate": 4.652884199008889e-08,
      "loss": 0.6958,
      "step": 33243
    },
    {
      "epoch": 0.9578747190687489,
      "grad_norm": 0.7380502820014954,
      "learning_rate": 4.6465354957358486e-08,
      "loss": 0.6703,
      "step": 33244
    },
    {
      "epoch": 0.9579035325303982,
      "grad_norm": 0.7006835341453552,
      "learning_rate": 4.6401911064963836e-08,
      "loss": 0.6726,
      "step": 33245
    },
    {
      "epoch": 0.9579323459920475,
      "grad_norm": 0.7440256476402283,
      "learning_rate": 4.6338510313456154e-08,
      "loss": 0.7042,
      "step": 33246
    },
    {
      "epoch": 0.9579611594536968,
      "grad_norm": 0.69636470079422,
      "learning_rate": 4.627515270338834e-08,
      "loss": 0.659,
      "step": 33247
    },
    {
      "epoch": 0.957989972915346,
      "grad_norm": 0.7393501400947571,
      "learning_rate": 4.621183823531217e-08,
      "loss": 0.6885,
      "step": 33248
    },
    {
      "epoch": 0.9580187863769953,
      "grad_norm": 0.7139809131622314,
      "learning_rate": 4.614856690977887e-08,
      "loss": 0.6625,
      "step": 33249
    },
    {
      "epoch": 0.9580475998386446,
      "grad_norm": 0.7221104502677917,
      "learning_rate": 4.608533872733911e-08,
      "loss": 0.7086,
      "step": 33250
    },
    {
      "epoch": 0.9580764133002939,
      "grad_norm": 0.7499048113822937,
      "learning_rate": 4.602215368854412e-08,
      "loss": 0.7234,
      "step": 33251
    },
    {
      "epoch": 0.9581052267619432,
      "grad_norm": 0.7375096082687378,
      "learning_rate": 4.595901179394402e-08,
      "loss": 0.6553,
      "step": 33252
    },
    {
      "epoch": 0.9581340402235925,
      "grad_norm": 0.7264094948768616,
      "learning_rate": 4.589591304408891e-08,
      "loss": 0.6774,
      "step": 33253
    },
    {
      "epoch": 0.9581628536852418,
      "grad_norm": 0.7151859998703003,
      "learning_rate": 4.583285743952781e-08,
      "loss": 0.6839,
      "step": 33254
    },
    {
      "epoch": 0.9581916671468911,
      "grad_norm": 0.7173026204109192,
      "learning_rate": 4.5769844980810276e-08,
      "loss": 0.6792,
      "step": 33255
    },
    {
      "epoch": 0.9582204806085403,
      "grad_norm": 0.7375285625457764,
      "learning_rate": 4.570687566848531e-08,
      "loss": 0.6825,
      "step": 33256
    },
    {
      "epoch": 0.9582492940701896,
      "grad_norm": 0.734733521938324,
      "learning_rate": 4.5643949503100806e-08,
      "loss": 0.6778,
      "step": 33257
    },
    {
      "epoch": 0.9582781075318388,
      "grad_norm": 0.7371625900268555,
      "learning_rate": 4.558106648520466e-08,
      "loss": 0.6597,
      "step": 33258
    },
    {
      "epoch": 0.9583069209934881,
      "grad_norm": 0.7314299941062927,
      "learning_rate": 4.5518226615344777e-08,
      "loss": 0.6807,
      "step": 33259
    },
    {
      "epoch": 0.9583357344551374,
      "grad_norm": 0.7223531007766724,
      "learning_rate": 4.545542989406904e-08,
      "loss": 0.6867,
      "step": 33260
    },
    {
      "epoch": 0.9583645479167867,
      "grad_norm": 0.7282570600509644,
      "learning_rate": 4.539267632192368e-08,
      "loss": 0.6782,
      "step": 33261
    },
    {
      "epoch": 0.958393361378436,
      "grad_norm": 0.7248566150665283,
      "learning_rate": 4.532996589945493e-08,
      "loss": 0.6891,
      "step": 33262
    },
    {
      "epoch": 0.9584221748400853,
      "grad_norm": 0.7208688855171204,
      "learning_rate": 4.526729862720958e-08,
      "loss": 0.6611,
      "step": 33263
    },
    {
      "epoch": 0.9584509883017346,
      "grad_norm": 0.7015169858932495,
      "learning_rate": 4.520467450573329e-08,
      "loss": 0.6471,
      "step": 33264
    },
    {
      "epoch": 0.9584798017633839,
      "grad_norm": 0.71895831823349,
      "learning_rate": 4.514209353557064e-08,
      "loss": 0.7039,
      "step": 33265
    },
    {
      "epoch": 0.9585086152250332,
      "grad_norm": 0.7323446273803711,
      "learning_rate": 4.5079555717267855e-08,
      "loss": 0.6853,
      "step": 33266
    },
    {
      "epoch": 0.9585374286866825,
      "grad_norm": 0.7243286371231079,
      "learning_rate": 4.501706105136838e-08,
      "loss": 0.6721,
      "step": 33267
    },
    {
      "epoch": 0.9585662421483317,
      "grad_norm": 0.7089462280273438,
      "learning_rate": 4.49546095384179e-08,
      "loss": 0.6658,
      "step": 33268
    },
    {
      "epoch": 0.9585950556099809,
      "grad_norm": 0.7374446392059326,
      "learning_rate": 4.489220117895932e-08,
      "loss": 0.6948,
      "step": 33269
    },
    {
      "epoch": 0.9586238690716302,
      "grad_norm": 0.7467707395553589,
      "learning_rate": 4.4829835973536625e-08,
      "loss": 0.6962,
      "step": 33270
    },
    {
      "epoch": 0.9586526825332795,
      "grad_norm": 0.7320290803909302,
      "learning_rate": 4.476751392269219e-08,
      "loss": 0.7057,
      "step": 33271
    },
    {
      "epoch": 0.9586814959949288,
      "grad_norm": 0.7239165306091309,
      "learning_rate": 4.470523502696944e-08,
      "loss": 0.6966,
      "step": 33272
    },
    {
      "epoch": 0.9587103094565781,
      "grad_norm": 0.7355203628540039,
      "learning_rate": 4.464299928691074e-08,
      "loss": 0.7035,
      "step": 33273
    },
    {
      "epoch": 0.9587391229182274,
      "grad_norm": 0.7378343939781189,
      "learning_rate": 4.4580806703057865e-08,
      "loss": 0.6903,
      "step": 33274
    },
    {
      "epoch": 0.9587679363798767,
      "grad_norm": 0.73824143409729,
      "learning_rate": 4.451865727595262e-08,
      "loss": 0.6977,
      "step": 33275
    },
    {
      "epoch": 0.958796749841526,
      "grad_norm": 0.7606808543205261,
      "learning_rate": 4.4456551006136216e-08,
      "loss": 0.7032,
      "step": 33276
    },
    {
      "epoch": 0.9588255633031753,
      "grad_norm": 0.6929210424423218,
      "learning_rate": 4.439448789414991e-08,
      "loss": 0.6441,
      "step": 33277
    },
    {
      "epoch": 0.9588543767648245,
      "grad_norm": 0.7225393056869507,
      "learning_rate": 4.4332467940532695e-08,
      "loss": 0.682,
      "step": 33278
    },
    {
      "epoch": 0.9588831902264738,
      "grad_norm": 0.7331193685531616,
      "learning_rate": 4.427049114582693e-08,
      "loss": 0.6905,
      "step": 33279
    },
    {
      "epoch": 0.9589120036881231,
      "grad_norm": 0.7332990765571594,
      "learning_rate": 4.420855751057051e-08,
      "loss": 0.6734,
      "step": 33280
    },
    {
      "epoch": 0.9589408171497724,
      "grad_norm": 0.738334059715271,
      "learning_rate": 4.414666703530412e-08,
      "loss": 0.7129,
      "step": 33281
    },
    {
      "epoch": 0.9589696306114217,
      "grad_norm": 0.7120084166526794,
      "learning_rate": 4.4084819720566216e-08,
      "loss": 0.6745,
      "step": 33282
    },
    {
      "epoch": 0.9589984440730709,
      "grad_norm": 0.743503749370575,
      "learning_rate": 4.4023015566895254e-08,
      "loss": 0.6788,
      "step": 33283
    },
    {
      "epoch": 0.9590272575347202,
      "grad_norm": 0.7096819877624512,
      "learning_rate": 4.396125457482969e-08,
      "loss": 0.6672,
      "step": 33284
    },
    {
      "epoch": 0.9590560709963695,
      "grad_norm": 0.7109824419021606,
      "learning_rate": 4.3899536744907436e-08,
      "loss": 0.6879,
      "step": 33285
    },
    {
      "epoch": 0.9590848844580188,
      "grad_norm": 0.7178646922111511,
      "learning_rate": 4.383786207766583e-08,
      "loss": 0.6742,
      "step": 33286
    },
    {
      "epoch": 0.959113697919668,
      "grad_norm": 0.7271789908409119,
      "learning_rate": 4.377623057364222e-08,
      "loss": 0.6907,
      "step": 33287
    },
    {
      "epoch": 0.9591425113813173,
      "grad_norm": 0.7230642437934875,
      "learning_rate": 4.37146422333734e-08,
      "loss": 0.6738,
      "step": 33288
    },
    {
      "epoch": 0.9591713248429666,
      "grad_norm": 0.7221435308456421,
      "learning_rate": 4.3653097057395067e-08,
      "loss": 0.6776,
      "step": 33289
    },
    {
      "epoch": 0.9592001383046159,
      "grad_norm": 0.740128219127655,
      "learning_rate": 4.359159504624399e-08,
      "loss": 0.7006,
      "step": 33290
    },
    {
      "epoch": 0.9592289517662652,
      "grad_norm": 0.7112074494361877,
      "learning_rate": 4.353013620045532e-08,
      "loss": 0.6723,
      "step": 33291
    },
    {
      "epoch": 0.9592577652279145,
      "grad_norm": 0.7234619855880737,
      "learning_rate": 4.3468720520564725e-08,
      "loss": 0.6861,
      "step": 33292
    },
    {
      "epoch": 0.9592865786895638,
      "grad_norm": 0.6999456286430359,
      "learning_rate": 4.3407348007106776e-08,
      "loss": 0.6806,
      "step": 33293
    },
    {
      "epoch": 0.9593153921512131,
      "grad_norm": 0.7202149629592896,
      "learning_rate": 4.334601866061605e-08,
      "loss": 0.6944,
      "step": 33294
    },
    {
      "epoch": 0.9593442056128624,
      "grad_norm": 0.7063403725624084,
      "learning_rate": 4.328473248162657e-08,
      "loss": 0.6512,
      "step": 33295
    },
    {
      "epoch": 0.9593730190745116,
      "grad_norm": 0.7157831192016602,
      "learning_rate": 4.322348947067234e-08,
      "loss": 0.6592,
      "step": 33296
    },
    {
      "epoch": 0.9594018325361608,
      "grad_norm": 0.7002303004264832,
      "learning_rate": 4.3162289628286837e-08,
      "loss": 0.6759,
      "step": 33297
    },
    {
      "epoch": 0.9594306459978101,
      "grad_norm": 0.7138728499412537,
      "learning_rate": 4.31011329550024e-08,
      "loss": 0.6958,
      "step": 33298
    },
    {
      "epoch": 0.9594594594594594,
      "grad_norm": 0.7138529419898987,
      "learning_rate": 4.3040019451351945e-08,
      "loss": 0.6692,
      "step": 33299
    },
    {
      "epoch": 0.9594882729211087,
      "grad_norm": 0.7240011692047119,
      "learning_rate": 4.297894911786837e-08,
      "loss": 0.6565,
      "step": 33300
    },
    {
      "epoch": 0.959517086382758,
      "grad_norm": 0.7239608764648438,
      "learning_rate": 4.291792195508182e-08,
      "loss": 0.6536,
      "step": 33301
    },
    {
      "epoch": 0.9595458998444073,
      "grad_norm": 0.7388504147529602,
      "learning_rate": 4.285693796352575e-08,
      "loss": 0.703,
      "step": 33302
    },
    {
      "epoch": 0.9595747133060566,
      "grad_norm": 0.7034814357757568,
      "learning_rate": 4.2795997143730285e-08,
      "loss": 0.6709,
      "step": 33303
    },
    {
      "epoch": 0.9596035267677059,
      "grad_norm": 0.7310378551483154,
      "learning_rate": 4.273509949622612e-08,
      "loss": 0.6829,
      "step": 33304
    },
    {
      "epoch": 0.9596323402293552,
      "grad_norm": 0.7274940609931946,
      "learning_rate": 4.2674245021543936e-08,
      "loss": 0.6736,
      "step": 33305
    },
    {
      "epoch": 0.9596611536910045,
      "grad_norm": 0.7267318367958069,
      "learning_rate": 4.2613433720213316e-08,
      "loss": 0.6644,
      "step": 33306
    },
    {
      "epoch": 0.9596899671526538,
      "grad_norm": 0.7259475588798523,
      "learning_rate": 4.255266559276438e-08,
      "loss": 0.6875,
      "step": 33307
    },
    {
      "epoch": 0.959718780614303,
      "grad_norm": 0.733466625213623,
      "learning_rate": 4.249194063972617e-08,
      "loss": 0.7009,
      "step": 33308
    },
    {
      "epoch": 0.9597475940759523,
      "grad_norm": 0.7014329433441162,
      "learning_rate": 4.2431258861627134e-08,
      "loss": 0.6451,
      "step": 33309
    },
    {
      "epoch": 0.9597764075376015,
      "grad_norm": 0.7256506681442261,
      "learning_rate": 4.23706202589963e-08,
      "loss": 0.707,
      "step": 33310
    },
    {
      "epoch": 0.9598052209992508,
      "grad_norm": 0.701535701751709,
      "learning_rate": 4.2310024832361576e-08,
      "loss": 0.6851,
      "step": 33311
    },
    {
      "epoch": 0.9598340344609001,
      "grad_norm": 0.7272307276725769,
      "learning_rate": 4.224947258225087e-08,
      "loss": 0.6857,
      "step": 33312
    },
    {
      "epoch": 0.9598628479225494,
      "grad_norm": 0.735345184803009,
      "learning_rate": 4.2188963509191e-08,
      "loss": 0.6832,
      "step": 33313
    },
    {
      "epoch": 0.9598916613841987,
      "grad_norm": 0.7414671778678894,
      "learning_rate": 4.2128497613708743e-08,
      "loss": 0.6782,
      "step": 33314
    },
    {
      "epoch": 0.959920474845848,
      "grad_norm": 0.7365714311599731,
      "learning_rate": 4.2068074896332025e-08,
      "loss": 0.6838,
      "step": 33315
    },
    {
      "epoch": 0.9599492883074973,
      "grad_norm": 0.705113410949707,
      "learning_rate": 4.200769535758598e-08,
      "loss": 0.6847,
      "step": 33316
    },
    {
      "epoch": 0.9599781017691466,
      "grad_norm": 0.7283338308334351,
      "learning_rate": 4.1947358997996843e-08,
      "loss": 0.6782,
      "step": 33317
    },
    {
      "epoch": 0.9600069152307958,
      "grad_norm": 0.7473440170288086,
      "learning_rate": 4.1887065818089766e-08,
      "loss": 0.6717,
      "step": 33318
    },
    {
      "epoch": 0.9600357286924451,
      "grad_norm": 0.689620316028595,
      "learning_rate": 4.1826815818390434e-08,
      "loss": 0.6782,
      "step": 33319
    },
    {
      "epoch": 0.9600645421540944,
      "grad_norm": 0.7279664874076843,
      "learning_rate": 4.1766608999422863e-08,
      "loss": 0.6931,
      "step": 33320
    },
    {
      "epoch": 0.9600933556157437,
      "grad_norm": 0.7326950430870056,
      "learning_rate": 4.1706445361711645e-08,
      "loss": 0.6768,
      "step": 33321
    },
    {
      "epoch": 0.960122169077393,
      "grad_norm": 0.6988683342933655,
      "learning_rate": 4.16463249057808e-08,
      "loss": 0.6593,
      "step": 33322
    },
    {
      "epoch": 0.9601509825390422,
      "grad_norm": 0.7361125946044922,
      "learning_rate": 4.1586247632154355e-08,
      "loss": 0.676,
      "step": 33323
    },
    {
      "epoch": 0.9601797960006915,
      "grad_norm": 0.7230987548828125,
      "learning_rate": 4.152621354135467e-08,
      "loss": 0.6753,
      "step": 33324
    },
    {
      "epoch": 0.9602086094623408,
      "grad_norm": 0.7415199279785156,
      "learning_rate": 4.1466222633905204e-08,
      "loss": 0.6564,
      "step": 33325
    },
    {
      "epoch": 0.9602374229239901,
      "grad_norm": 0.7147079706192017,
      "learning_rate": 4.1406274910327784e-08,
      "loss": 0.6853,
      "step": 33326
    },
    {
      "epoch": 0.9602662363856393,
      "grad_norm": 0.7264163494110107,
      "learning_rate": 4.134637037114586e-08,
      "loss": 0.7114,
      "step": 33327
    },
    {
      "epoch": 0.9602950498472886,
      "grad_norm": 0.733492910861969,
      "learning_rate": 4.128650901687903e-08,
      "loss": 0.6686,
      "step": 33328
    },
    {
      "epoch": 0.9603238633089379,
      "grad_norm": 0.7305555939674377,
      "learning_rate": 4.122669084805075e-08,
      "loss": 0.6864,
      "step": 33329
    },
    {
      "epoch": 0.9603526767705872,
      "grad_norm": 0.7159742712974548,
      "learning_rate": 4.1166915865180625e-08,
      "loss": 0.6888,
      "step": 33330
    },
    {
      "epoch": 0.9603814902322365,
      "grad_norm": 0.7201635241508484,
      "learning_rate": 4.1107184068789886e-08,
      "loss": 0.6916,
      "step": 33331
    },
    {
      "epoch": 0.9604103036938858,
      "grad_norm": 0.7301716804504395,
      "learning_rate": 4.1047495459398125e-08,
      "loss": 0.6607,
      "step": 33332
    },
    {
      "epoch": 0.9604391171555351,
      "grad_norm": 0.714677631855011,
      "learning_rate": 4.0987850037526036e-08,
      "loss": 0.6655,
      "step": 33333
    },
    {
      "epoch": 0.9604679306171844,
      "grad_norm": 0.7286683917045593,
      "learning_rate": 4.092824780369209e-08,
      "loss": 0.6856,
      "step": 33334
    },
    {
      "epoch": 0.9604967440788337,
      "grad_norm": 0.7262861728668213,
      "learning_rate": 4.086868875841588e-08,
      "loss": 0.6766,
      "step": 33335
    },
    {
      "epoch": 0.960525557540483,
      "grad_norm": 0.7215275168418884,
      "learning_rate": 4.080917290221642e-08,
      "loss": 0.6919,
      "step": 33336
    },
    {
      "epoch": 0.9605543710021321,
      "grad_norm": 0.7188915610313416,
      "learning_rate": 4.0749700235611645e-08,
      "loss": 0.6792,
      "step": 33337
    },
    {
      "epoch": 0.9605831844637814,
      "grad_norm": 0.7312301397323608,
      "learning_rate": 4.0690270759119466e-08,
      "loss": 0.6964,
      "step": 33338
    },
    {
      "epoch": 0.9606119979254307,
      "grad_norm": 0.7166825532913208,
      "learning_rate": 4.063088447325725e-08,
      "loss": 0.6768,
      "step": 33339
    },
    {
      "epoch": 0.96064081138708,
      "grad_norm": 0.7028646469116211,
      "learning_rate": 4.057154137854291e-08,
      "loss": 0.6679,
      "step": 33340
    },
    {
      "epoch": 0.9606696248487293,
      "grad_norm": 0.7302495241165161,
      "learning_rate": 4.0512241475492155e-08,
      "loss": 0.6734,
      "step": 33341
    },
    {
      "epoch": 0.9606984383103786,
      "grad_norm": 0.7318368554115295,
      "learning_rate": 4.0452984764622895e-08,
      "loss": 0.6728,
      "step": 33342
    },
    {
      "epoch": 0.9607272517720279,
      "grad_norm": 0.7236559391021729,
      "learning_rate": 4.039377124644972e-08,
      "loss": 0.6674,
      "step": 33343
    },
    {
      "epoch": 0.9607560652336772,
      "grad_norm": 0.7187796831130981,
      "learning_rate": 4.033460092149e-08,
      "loss": 0.6727,
      "step": 33344
    },
    {
      "epoch": 0.9607848786953265,
      "grad_norm": 0.6987348794937134,
      "learning_rate": 4.02754737902572e-08,
      "loss": 0.6801,
      "step": 33345
    },
    {
      "epoch": 0.9608136921569758,
      "grad_norm": 0.7122927904129028,
      "learning_rate": 4.021638985326759e-08,
      "loss": 0.6777,
      "step": 33346
    },
    {
      "epoch": 0.960842505618625,
      "grad_norm": 0.7116343379020691,
      "learning_rate": 4.015734911103464e-08,
      "loss": 0.6918,
      "step": 33347
    },
    {
      "epoch": 0.9608713190802743,
      "grad_norm": 0.7150068879127502,
      "learning_rate": 4.009835156407349e-08,
      "loss": 0.6615,
      "step": 33348
    },
    {
      "epoch": 0.9609001325419236,
      "grad_norm": 0.7273152470588684,
      "learning_rate": 4.0039397212897625e-08,
      "loss": 0.6814,
      "step": 33349
    },
    {
      "epoch": 0.9609289460035728,
      "grad_norm": 0.7196675539016724,
      "learning_rate": 3.998048605802052e-08,
      "loss": 0.6705,
      "step": 33350
    },
    {
      "epoch": 0.9609577594652221,
      "grad_norm": 0.7297621369361877,
      "learning_rate": 3.99216180999551e-08,
      "loss": 0.6811,
      "step": 33351
    },
    {
      "epoch": 0.9609865729268714,
      "grad_norm": 0.727055013179779,
      "learning_rate": 3.986279333921372e-08,
      "loss": 0.7048,
      "step": 33352
    },
    {
      "epoch": 0.9610153863885207,
      "grad_norm": 0.7202114462852478,
      "learning_rate": 3.980401177630988e-08,
      "loss": 0.6688,
      "step": 33353
    },
    {
      "epoch": 0.96104419985017,
      "grad_norm": 0.7396617531776428,
      "learning_rate": 3.974527341175427e-08,
      "loss": 0.6659,
      "step": 33354
    },
    {
      "epoch": 0.9610730133118193,
      "grad_norm": 0.7188145518302917,
      "learning_rate": 3.968657824605926e-08,
      "loss": 0.643,
      "step": 33355
    },
    {
      "epoch": 0.9611018267734686,
      "grad_norm": 0.7382585406303406,
      "learning_rate": 3.9627926279735e-08,
      "loss": 0.7058,
      "step": 33356
    },
    {
      "epoch": 0.9611306402351179,
      "grad_norm": 0.7148838043212891,
      "learning_rate": 3.956931751329385e-08,
      "loss": 0.6631,
      "step": 33357
    },
    {
      "epoch": 0.9611594536967671,
      "grad_norm": 0.7688509821891785,
      "learning_rate": 3.9510751947244296e-08,
      "loss": 0.6903,
      "step": 33358
    },
    {
      "epoch": 0.9611882671584164,
      "grad_norm": 0.7168763279914856,
      "learning_rate": 3.945222958209816e-08,
      "loss": 0.6562,
      "step": 33359
    },
    {
      "epoch": 0.9612170806200657,
      "grad_norm": 0.7149962186813354,
      "learning_rate": 3.939375041836391e-08,
      "loss": 0.6955,
      "step": 33360
    },
    {
      "epoch": 0.961245894081715,
      "grad_norm": 0.7154156565666199,
      "learning_rate": 3.93353144565517e-08,
      "loss": 0.6738,
      "step": 33361
    },
    {
      "epoch": 0.9612747075433643,
      "grad_norm": 0.7145744562149048,
      "learning_rate": 3.927692169716946e-08,
      "loss": 0.6465,
      "step": 33362
    },
    {
      "epoch": 0.9613035210050136,
      "grad_norm": 0.7117775082588196,
      "learning_rate": 3.921857214072677e-08,
      "loss": 0.6508,
      "step": 33363
    },
    {
      "epoch": 0.9613323344666628,
      "grad_norm": 0.7139551639556885,
      "learning_rate": 3.916026578773158e-08,
      "loss": 0.6935,
      "step": 33364
    },
    {
      "epoch": 0.9613611479283121,
      "grad_norm": 0.7338752746582031,
      "learning_rate": 3.910200263869124e-08,
      "loss": 0.6767,
      "step": 33365
    },
    {
      "epoch": 0.9613899613899614,
      "grad_norm": 0.7535126805305481,
      "learning_rate": 3.904378269411313e-08,
      "loss": 0.6892,
      "step": 33366
    },
    {
      "epoch": 0.9614187748516106,
      "grad_norm": 0.7517881393432617,
      "learning_rate": 3.898560595450518e-08,
      "loss": 0.6942,
      "step": 33367
    },
    {
      "epoch": 0.9614475883132599,
      "grad_norm": 0.7028617262840271,
      "learning_rate": 3.892747242037254e-08,
      "loss": 0.6782,
      "step": 33368
    },
    {
      "epoch": 0.9614764017749092,
      "grad_norm": 0.7248727679252625,
      "learning_rate": 3.8869382092222576e-08,
      "loss": 0.677,
      "step": 33369
    },
    {
      "epoch": 0.9615052152365585,
      "grad_norm": 0.7192000150680542,
      "learning_rate": 3.8811334970561555e-08,
      "loss": 0.6896,
      "step": 33370
    },
    {
      "epoch": 0.9615340286982078,
      "grad_norm": 0.710727870464325,
      "learning_rate": 3.8753331055894076e-08,
      "loss": 0.6888,
      "step": 33371
    },
    {
      "epoch": 0.9615628421598571,
      "grad_norm": 0.7189645767211914,
      "learning_rate": 3.8695370348725836e-08,
      "loss": 0.6991,
      "step": 33372
    },
    {
      "epoch": 0.9615916556215064,
      "grad_norm": 0.7546056509017944,
      "learning_rate": 3.8637452849560886e-08,
      "loss": 0.6695,
      "step": 33373
    },
    {
      "epoch": 0.9616204690831557,
      "grad_norm": 0.7351495027542114,
      "learning_rate": 3.857957855890438e-08,
      "loss": 0.6918,
      "step": 33374
    },
    {
      "epoch": 0.961649282544805,
      "grad_norm": 0.738084614276886,
      "learning_rate": 3.852174747726034e-08,
      "loss": 0.6827,
      "step": 33375
    },
    {
      "epoch": 0.9616780960064543,
      "grad_norm": 0.7187762260437012,
      "learning_rate": 3.846395960513227e-08,
      "loss": 0.6838,
      "step": 33376
    },
    {
      "epoch": 0.9617069094681034,
      "grad_norm": 0.7130988836288452,
      "learning_rate": 3.8406214943023104e-08,
      "loss": 0.6675,
      "step": 33377
    },
    {
      "epoch": 0.9617357229297527,
      "grad_norm": 0.7101413607597351,
      "learning_rate": 3.8348513491436315e-08,
      "loss": 0.6671,
      "step": 33378
    },
    {
      "epoch": 0.961764536391402,
      "grad_norm": 0.7310440540313721,
      "learning_rate": 3.829085525087373e-08,
      "loss": 0.6877,
      "step": 33379
    },
    {
      "epoch": 0.9617933498530513,
      "grad_norm": 0.7313530445098877,
      "learning_rate": 3.823324022183827e-08,
      "loss": 0.6855,
      "step": 33380
    },
    {
      "epoch": 0.9618221633147006,
      "grad_norm": 0.7375286221504211,
      "learning_rate": 3.817566840483122e-08,
      "loss": 0.682,
      "step": 33381
    },
    {
      "epoch": 0.9618509767763499,
      "grad_norm": 0.7064698338508606,
      "learning_rate": 3.811813980035439e-08,
      "loss": 0.6573,
      "step": 33382
    },
    {
      "epoch": 0.9618797902379992,
      "grad_norm": 0.7272899746894836,
      "learning_rate": 3.8060654408907936e-08,
      "loss": 0.6889,
      "step": 33383
    },
    {
      "epoch": 0.9619086036996485,
      "grad_norm": 0.7283053398132324,
      "learning_rate": 3.800321223099368e-08,
      "loss": 0.6767,
      "step": 33384
    },
    {
      "epoch": 0.9619374171612978,
      "grad_norm": 0.731003999710083,
      "learning_rate": 3.7945813267110666e-08,
      "loss": 0.6586,
      "step": 33385
    },
    {
      "epoch": 0.9619662306229471,
      "grad_norm": 0.7144953012466431,
      "learning_rate": 3.78884575177596e-08,
      "loss": 0.6632,
      "step": 33386
    },
    {
      "epoch": 0.9619950440845964,
      "grad_norm": 0.7256089448928833,
      "learning_rate": 3.783114498343954e-08,
      "loss": 0.6597,
      "step": 33387
    },
    {
      "epoch": 0.9620238575462456,
      "grad_norm": 0.7395055890083313,
      "learning_rate": 3.777387566465063e-08,
      "loss": 0.6888,
      "step": 33388
    },
    {
      "epoch": 0.9620526710078949,
      "grad_norm": 0.715872585773468,
      "learning_rate": 3.771664956189025e-08,
      "loss": 0.6983,
      "step": 33389
    },
    {
      "epoch": 0.9620814844695442,
      "grad_norm": 0.7269458174705505,
      "learning_rate": 3.765946667565745e-08,
      "loss": 0.7033,
      "step": 33390
    },
    {
      "epoch": 0.9621102979311934,
      "grad_norm": 0.7211047410964966,
      "learning_rate": 3.7602327006450166e-08,
      "loss": 0.6796,
      "step": 33391
    },
    {
      "epoch": 0.9621391113928427,
      "grad_norm": 0.7431938052177429,
      "learning_rate": 3.7545230554765774e-08,
      "loss": 0.6862,
      "step": 33392
    },
    {
      "epoch": 0.962167924854492,
      "grad_norm": 0.7265397906303406,
      "learning_rate": 3.7488177321102214e-08,
      "loss": 0.6932,
      "step": 33393
    },
    {
      "epoch": 0.9621967383161413,
      "grad_norm": 0.7260283827781677,
      "learning_rate": 3.743116730595575e-08,
      "loss": 0.6706,
      "step": 33394
    },
    {
      "epoch": 0.9622255517777906,
      "grad_norm": 0.741573691368103,
      "learning_rate": 3.7374200509823214e-08,
      "loss": 0.695,
      "step": 33395
    },
    {
      "epoch": 0.9622543652394399,
      "grad_norm": 0.7287968397140503,
      "learning_rate": 3.731727693320031e-08,
      "loss": 0.6983,
      "step": 33396
    },
    {
      "epoch": 0.9622831787010891,
      "grad_norm": 0.7421970367431641,
      "learning_rate": 3.726039657658331e-08,
      "loss": 0.6723,
      "step": 33397
    },
    {
      "epoch": 0.9623119921627384,
      "grad_norm": 0.7268663644790649,
      "learning_rate": 3.7203559440467385e-08,
      "loss": 0.658,
      "step": 33398
    },
    {
      "epoch": 0.9623408056243877,
      "grad_norm": 0.7234047651290894,
      "learning_rate": 3.714676552534713e-08,
      "loss": 0.6774,
      "step": 33399
    },
    {
      "epoch": 0.962369619086037,
      "grad_norm": 0.7143720984458923,
      "learning_rate": 3.70900148317177e-08,
      "loss": 0.6791,
      "step": 33400
    },
    {
      "epoch": 0.9623984325476863,
      "grad_norm": 0.7211921215057373,
      "learning_rate": 3.703330736007316e-08,
      "loss": 0.6716,
      "step": 33401
    },
    {
      "epoch": 0.9624272460093356,
      "grad_norm": 0.7129075527191162,
      "learning_rate": 3.697664311090754e-08,
      "loss": 0.6837,
      "step": 33402
    },
    {
      "epoch": 0.9624560594709849,
      "grad_norm": 0.754879355430603,
      "learning_rate": 3.692002208471435e-08,
      "loss": 0.6932,
      "step": 33403
    },
    {
      "epoch": 0.9624848729326341,
      "grad_norm": 0.7215328216552734,
      "learning_rate": 3.6863444281985963e-08,
      "loss": 0.6698,
      "step": 33404
    },
    {
      "epoch": 0.9625136863942834,
      "grad_norm": 0.7185789346694946,
      "learning_rate": 3.680690970321588e-08,
      "loss": 0.684,
      "step": 33405
    },
    {
      "epoch": 0.9625424998559327,
      "grad_norm": 0.7215775847434998,
      "learning_rate": 3.6750418348895924e-08,
      "loss": 0.6657,
      "step": 33406
    },
    {
      "epoch": 0.962571313317582,
      "grad_norm": 0.7297530174255371,
      "learning_rate": 3.669397021951848e-08,
      "loss": 0.6792,
      "step": 33407
    },
    {
      "epoch": 0.9626001267792312,
      "grad_norm": 0.7467246055603027,
      "learning_rate": 3.663756531557538e-08,
      "loss": 0.6769,
      "step": 33408
    },
    {
      "epoch": 0.9626289402408805,
      "grad_norm": 0.7156221270561218,
      "learning_rate": 3.658120363755735e-08,
      "loss": 0.6752,
      "step": 33409
    },
    {
      "epoch": 0.9626577537025298,
      "grad_norm": 0.715631902217865,
      "learning_rate": 3.652488518595565e-08,
      "loss": 0.6862,
      "step": 33410
    },
    {
      "epoch": 0.9626865671641791,
      "grad_norm": 0.7367952466011047,
      "learning_rate": 3.646860996125989e-08,
      "loss": 0.7002,
      "step": 33411
    },
    {
      "epoch": 0.9627153806258284,
      "grad_norm": 0.7556073069572449,
      "learning_rate": 3.641237796396191e-08,
      "loss": 0.6899,
      "step": 33412
    },
    {
      "epoch": 0.9627441940874777,
      "grad_norm": 0.7295610308647156,
      "learning_rate": 3.635618919454964e-08,
      "loss": 0.6812,
      "step": 33413
    },
    {
      "epoch": 0.962773007549127,
      "grad_norm": 0.7234138250350952,
      "learning_rate": 3.6300043653513253e-08,
      "loss": 0.677,
      "step": 33414
    },
    {
      "epoch": 0.9628018210107763,
      "grad_norm": 0.7078484296798706,
      "learning_rate": 3.6243941341341236e-08,
      "loss": 0.6881,
      "step": 33415
    },
    {
      "epoch": 0.9628306344724256,
      "grad_norm": 0.7129883170127869,
      "learning_rate": 3.618788225852321e-08,
      "loss": 0.6693,
      "step": 33416
    },
    {
      "epoch": 0.9628594479340749,
      "grad_norm": 0.7023024559020996,
      "learning_rate": 3.6131866405546556e-08,
      "loss": 0.6828,
      "step": 33417
    },
    {
      "epoch": 0.962888261395724,
      "grad_norm": 0.7350528836250305,
      "learning_rate": 3.607589378289922e-08,
      "loss": 0.6706,
      "step": 33418
    },
    {
      "epoch": 0.9629170748573733,
      "grad_norm": 0.7010917663574219,
      "learning_rate": 3.601996439106858e-08,
      "loss": 0.6535,
      "step": 33419
    },
    {
      "epoch": 0.9629458883190226,
      "grad_norm": 0.7272967100143433,
      "learning_rate": 3.596407823054204e-08,
      "loss": 0.6795,
      "step": 33420
    },
    {
      "epoch": 0.9629747017806719,
      "grad_norm": 0.7255515456199646,
      "learning_rate": 3.590823530180587e-08,
      "loss": 0.6766,
      "step": 33421
    },
    {
      "epoch": 0.9630035152423212,
      "grad_norm": 0.7039816975593567,
      "learning_rate": 3.5852435605346904e-08,
      "loss": 0.6622,
      "step": 33422
    },
    {
      "epoch": 0.9630323287039705,
      "grad_norm": 0.7171633839607239,
      "learning_rate": 3.579667914165086e-08,
      "loss": 0.6646,
      "step": 33423
    },
    {
      "epoch": 0.9630611421656198,
      "grad_norm": 0.7072150111198425,
      "learning_rate": 3.574096591120346e-08,
      "loss": 0.6972,
      "step": 33424
    },
    {
      "epoch": 0.9630899556272691,
      "grad_norm": 0.6905180215835571,
      "learning_rate": 3.568529591448988e-08,
      "loss": 0.6621,
      "step": 33425
    },
    {
      "epoch": 0.9631187690889184,
      "grad_norm": 0.7137725949287415,
      "learning_rate": 3.562966915199473e-08,
      "loss": 0.6922,
      "step": 33426
    },
    {
      "epoch": 0.9631475825505676,
      "grad_norm": 0.7130422592163086,
      "learning_rate": 3.5574085624202615e-08,
      "loss": 0.6791,
      "step": 33427
    },
    {
      "epoch": 0.9631763960122169,
      "grad_norm": 0.7259190678596497,
      "learning_rate": 3.5518545331597596e-08,
      "loss": 0.6824,
      "step": 33428
    },
    {
      "epoch": 0.9632052094738662,
      "grad_norm": 0.8276847004890442,
      "learning_rate": 3.5463048274663734e-08,
      "loss": 0.6666,
      "step": 33429
    },
    {
      "epoch": 0.9632340229355155,
      "grad_norm": 0.7312349677085876,
      "learning_rate": 3.5407594453883423e-08,
      "loss": 0.6827,
      "step": 33430
    },
    {
      "epoch": 0.9632628363971647,
      "grad_norm": 0.718543529510498,
      "learning_rate": 3.535218386974071e-08,
      "loss": 0.6822,
      "step": 33431
    },
    {
      "epoch": 0.963291649858814,
      "grad_norm": 0.715846061706543,
      "learning_rate": 3.529681652271688e-08,
      "loss": 0.6638,
      "step": 33432
    },
    {
      "epoch": 0.9633204633204633,
      "grad_norm": 0.7283624410629272,
      "learning_rate": 3.524149241329544e-08,
      "loss": 0.6987,
      "step": 33433
    },
    {
      "epoch": 0.9633492767821126,
      "grad_norm": 0.7315848469734192,
      "learning_rate": 3.518621154195767e-08,
      "loss": 0.6775,
      "step": 33434
    },
    {
      "epoch": 0.9633780902437619,
      "grad_norm": 0.7148677706718445,
      "learning_rate": 3.513097390918485e-08,
      "loss": 0.6521,
      "step": 33435
    },
    {
      "epoch": 0.9634069037054112,
      "grad_norm": 0.722964882850647,
      "learning_rate": 3.507577951545826e-08,
      "loss": 0.6608,
      "step": 33436
    },
    {
      "epoch": 0.9634357171670604,
      "grad_norm": 0.7159444689750671,
      "learning_rate": 3.502062836125863e-08,
      "loss": 0.6818,
      "step": 33437
    },
    {
      "epoch": 0.9634645306287097,
      "grad_norm": 0.7280224561691284,
      "learning_rate": 3.4965520447066136e-08,
      "loss": 0.6987,
      "step": 33438
    },
    {
      "epoch": 0.963493344090359,
      "grad_norm": 0.765659749507904,
      "learning_rate": 3.491045577336094e-08,
      "loss": 0.6923,
      "step": 33439
    },
    {
      "epoch": 0.9635221575520083,
      "grad_norm": 0.7178431749343872,
      "learning_rate": 3.4855434340622107e-08,
      "loss": 0.6682,
      "step": 33440
    },
    {
      "epoch": 0.9635509710136576,
      "grad_norm": 0.703353226184845,
      "learning_rate": 3.4800456149329255e-08,
      "loss": 0.6576,
      "step": 33441
    },
    {
      "epoch": 0.9635797844753069,
      "grad_norm": 0.7210720777511597,
      "learning_rate": 3.474552119996088e-08,
      "loss": 0.6462,
      "step": 33442
    },
    {
      "epoch": 0.9636085979369562,
      "grad_norm": 0.7122898697853088,
      "learning_rate": 3.469062949299551e-08,
      "loss": 0.6768,
      "step": 33443
    },
    {
      "epoch": 0.9636374113986055,
      "grad_norm": 0.7302985787391663,
      "learning_rate": 3.4635781028911074e-08,
      "loss": 0.695,
      "step": 33444
    },
    {
      "epoch": 0.9636662248602547,
      "grad_norm": 0.7076923251152039,
      "learning_rate": 3.458097580818609e-08,
      "loss": 0.672,
      "step": 33445
    },
    {
      "epoch": 0.963695038321904,
      "grad_norm": 0.7294363975524902,
      "learning_rate": 3.452621383129684e-08,
      "loss": 0.6968,
      "step": 33446
    },
    {
      "epoch": 0.9637238517835532,
      "grad_norm": 0.7166760563850403,
      "learning_rate": 3.447149509872072e-08,
      "loss": 0.6871,
      "step": 33447
    },
    {
      "epoch": 0.9637526652452025,
      "grad_norm": 0.7477336525917053,
      "learning_rate": 3.4416819610934016e-08,
      "loss": 0.6962,
      "step": 33448
    },
    {
      "epoch": 0.9637814787068518,
      "grad_norm": 0.7010495662689209,
      "learning_rate": 3.4362187368413014e-08,
      "loss": 0.6549,
      "step": 33449
    },
    {
      "epoch": 0.9638102921685011,
      "grad_norm": 0.7405457496643066,
      "learning_rate": 3.4307598371633445e-08,
      "loss": 0.6904,
      "step": 33450
    },
    {
      "epoch": 0.9638391056301504,
      "grad_norm": 0.7155120968818665,
      "learning_rate": 3.425305262107104e-08,
      "loss": 0.6805,
      "step": 33451
    },
    {
      "epoch": 0.9638679190917997,
      "grad_norm": 0.700106143951416,
      "learning_rate": 3.419855011720097e-08,
      "loss": 0.6637,
      "step": 33452
    },
    {
      "epoch": 0.963896732553449,
      "grad_norm": 0.7057281136512756,
      "learning_rate": 3.414409086049675e-08,
      "loss": 0.6462,
      "step": 33453
    },
    {
      "epoch": 0.9639255460150983,
      "grad_norm": 0.7137443423271179,
      "learning_rate": 3.408967485143411e-08,
      "loss": 0.6939,
      "step": 33454
    },
    {
      "epoch": 0.9639543594767476,
      "grad_norm": 0.7022228240966797,
      "learning_rate": 3.4035302090486e-08,
      "loss": 0.6607,
      "step": 33455
    },
    {
      "epoch": 0.9639831729383969,
      "grad_norm": 0.7351078987121582,
      "learning_rate": 3.398097257812649e-08,
      "loss": 0.6907,
      "step": 33456
    },
    {
      "epoch": 0.9640119864000462,
      "grad_norm": 0.7256629467010498,
      "learning_rate": 3.392668631482798e-08,
      "loss": 0.6871,
      "step": 33457
    },
    {
      "epoch": 0.9640407998616953,
      "grad_norm": 0.7643991708755493,
      "learning_rate": 3.387244330106454e-08,
      "loss": 0.6965,
      "step": 33458
    },
    {
      "epoch": 0.9640696133233446,
      "grad_norm": 0.7373877763748169,
      "learning_rate": 3.381824353730689e-08,
      "loss": 0.6787,
      "step": 33459
    },
    {
      "epoch": 0.9640984267849939,
      "grad_norm": 0.7223416566848755,
      "learning_rate": 3.3764087024029115e-08,
      "loss": 0.6735,
      "step": 33460
    },
    {
      "epoch": 0.9641272402466432,
      "grad_norm": 0.7295119762420654,
      "learning_rate": 3.370997376170082e-08,
      "loss": 0.7124,
      "step": 33461
    },
    {
      "epoch": 0.9641560537082925,
      "grad_norm": 0.7207087874412537,
      "learning_rate": 3.365590375079442e-08,
      "loss": 0.6808,
      "step": 33462
    },
    {
      "epoch": 0.9641848671699418,
      "grad_norm": 0.723496675491333,
      "learning_rate": 3.360187699178063e-08,
      "loss": 0.675,
      "step": 33463
    },
    {
      "epoch": 0.9642136806315911,
      "grad_norm": 0.7343319654464722,
      "learning_rate": 3.354789348513021e-08,
      "loss": 0.6762,
      "step": 33464
    },
    {
      "epoch": 0.9642424940932404,
      "grad_norm": 0.7330917119979858,
      "learning_rate": 3.349395323131277e-08,
      "loss": 0.6855,
      "step": 33465
    },
    {
      "epoch": 0.9642713075548897,
      "grad_norm": 0.7448094487190247,
      "learning_rate": 3.3440056230797934e-08,
      "loss": 0.6932,
      "step": 33466
    },
    {
      "epoch": 0.964300121016539,
      "grad_norm": 0.7072652578353882,
      "learning_rate": 3.338620248405589e-08,
      "loss": 0.6633,
      "step": 33467
    },
    {
      "epoch": 0.9643289344781882,
      "grad_norm": 0.7365376353263855,
      "learning_rate": 3.333239199155514e-08,
      "loss": 0.6692,
      "step": 33468
    },
    {
      "epoch": 0.9643577479398375,
      "grad_norm": 0.7208340167999268,
      "learning_rate": 3.32786247537642e-08,
      "loss": 0.6832,
      "step": 33469
    },
    {
      "epoch": 0.9643865614014868,
      "grad_norm": 0.7072609663009644,
      "learning_rate": 3.322490077115159e-08,
      "loss": 0.6575,
      "step": 33470
    },
    {
      "epoch": 0.9644153748631361,
      "grad_norm": 0.7094427943229675,
      "learning_rate": 3.317122004418527e-08,
      "loss": 0.678,
      "step": 33471
    },
    {
      "epoch": 0.9644441883247853,
      "grad_norm": 0.7665740847587585,
      "learning_rate": 3.311758257333264e-08,
      "loss": 0.7186,
      "step": 33472
    },
    {
      "epoch": 0.9644730017864346,
      "grad_norm": 0.7239341139793396,
      "learning_rate": 3.3063988359061105e-08,
      "loss": 0.6822,
      "step": 33473
    },
    {
      "epoch": 0.9645018152480839,
      "grad_norm": 0.7272130846977234,
      "learning_rate": 3.301043740183641e-08,
      "loss": 0.6727,
      "step": 33474
    },
    {
      "epoch": 0.9645306287097332,
      "grad_norm": 0.709176242351532,
      "learning_rate": 3.295692970212594e-08,
      "loss": 0.6767,
      "step": 33475
    },
    {
      "epoch": 0.9645594421713825,
      "grad_norm": 0.724764883518219,
      "learning_rate": 3.290346526039545e-08,
      "loss": 0.6913,
      "step": 33476
    },
    {
      "epoch": 0.9645882556330317,
      "grad_norm": 0.706534206867218,
      "learning_rate": 3.285004407711068e-08,
      "loss": 0.6662,
      "step": 33477
    },
    {
      "epoch": 0.964617069094681,
      "grad_norm": 0.7067723274230957,
      "learning_rate": 3.2796666152736245e-08,
      "loss": 0.6667,
      "step": 33478
    },
    {
      "epoch": 0.9646458825563303,
      "grad_norm": 0.7491518259048462,
      "learning_rate": 3.2743331487737897e-08,
      "loss": 0.6822,
      "step": 33479
    },
    {
      "epoch": 0.9646746960179796,
      "grad_norm": 0.7231647968292236,
      "learning_rate": 3.26900400825797e-08,
      "loss": 0.6743,
      "step": 33480
    },
    {
      "epoch": 0.9647035094796289,
      "grad_norm": 0.7389708161354065,
      "learning_rate": 3.2636791937725174e-08,
      "loss": 0.6951,
      "step": 33481
    },
    {
      "epoch": 0.9647323229412782,
      "grad_norm": 0.7589222192764282,
      "learning_rate": 3.258358705363896e-08,
      "loss": 0.701,
      "step": 33482
    },
    {
      "epoch": 0.9647611364029275,
      "grad_norm": 0.716552734375,
      "learning_rate": 3.253042543078455e-08,
      "loss": 0.6748,
      "step": 33483
    },
    {
      "epoch": 0.9647899498645768,
      "grad_norm": 0.7272996306419373,
      "learning_rate": 3.247730706962382e-08,
      "loss": 0.7064,
      "step": 33484
    },
    {
      "epoch": 0.964818763326226,
      "grad_norm": 0.7330442667007446,
      "learning_rate": 3.2424231970620835e-08,
      "loss": 0.6928,
      "step": 33485
    },
    {
      "epoch": 0.9648475767878752,
      "grad_norm": 0.7132602334022522,
      "learning_rate": 3.2371200134236336e-08,
      "loss": 0.6973,
      "step": 33486
    },
    {
      "epoch": 0.9648763902495245,
      "grad_norm": 0.7159009575843811,
      "learning_rate": 3.2318211560932736e-08,
      "loss": 0.6681,
      "step": 33487
    },
    {
      "epoch": 0.9649052037111738,
      "grad_norm": 0.7101324200630188,
      "learning_rate": 3.226526625117243e-08,
      "loss": 0.6748,
      "step": 33488
    },
    {
      "epoch": 0.9649340171728231,
      "grad_norm": 0.7201633453369141,
      "learning_rate": 3.221236420541507e-08,
      "loss": 0.6665,
      "step": 33489
    },
    {
      "epoch": 0.9649628306344724,
      "grad_norm": 0.7247509360313416,
      "learning_rate": 3.2159505424122495e-08,
      "loss": 0.6835,
      "step": 33490
    },
    {
      "epoch": 0.9649916440961217,
      "grad_norm": 0.7230544686317444,
      "learning_rate": 3.210668990775434e-08,
      "loss": 0.6691,
      "step": 33491
    },
    {
      "epoch": 0.965020457557771,
      "grad_norm": 0.7319961190223694,
      "learning_rate": 3.205391765677079e-08,
      "loss": 0.6833,
      "step": 33492
    },
    {
      "epoch": 0.9650492710194203,
      "grad_norm": 0.7433045506477356,
      "learning_rate": 3.200118867163149e-08,
      "loss": 0.7065,
      "step": 33493
    },
    {
      "epoch": 0.9650780844810696,
      "grad_norm": 0.7263614535331726,
      "learning_rate": 3.19485029527955e-08,
      "loss": 0.6729,
      "step": 33494
    },
    {
      "epoch": 0.9651068979427189,
      "grad_norm": 0.7256661653518677,
      "learning_rate": 3.189586050072191e-08,
      "loss": 0.6768,
      "step": 33495
    },
    {
      "epoch": 0.9651357114043682,
      "grad_norm": 0.7255401611328125,
      "learning_rate": 3.1843261315869786e-08,
      "loss": 0.6844,
      "step": 33496
    },
    {
      "epoch": 0.9651645248660174,
      "grad_norm": 0.6931940913200378,
      "learning_rate": 3.179070539869544e-08,
      "loss": 0.6524,
      "step": 33497
    },
    {
      "epoch": 0.9651933383276667,
      "grad_norm": 0.7291271090507507,
      "learning_rate": 3.17381927496585e-08,
      "loss": 0.6693,
      "step": 33498
    },
    {
      "epoch": 0.9652221517893159,
      "grad_norm": 0.7190123200416565,
      "learning_rate": 3.1685723369215274e-08,
      "loss": 0.6784,
      "step": 33499
    },
    {
      "epoch": 0.9652509652509652,
      "grad_norm": 0.7204697728157043,
      "learning_rate": 3.16332972578226e-08,
      "loss": 0.6632,
      "step": 33500
    },
    {
      "epoch": 0.9652797787126145,
      "grad_norm": 0.7024602890014648,
      "learning_rate": 3.158091441593736e-08,
      "loss": 0.6727,
      "step": 33501
    },
    {
      "epoch": 0.9653085921742638,
      "grad_norm": 0.7316771745681763,
      "learning_rate": 3.152857484401639e-08,
      "loss": 0.6956,
      "step": 33502
    },
    {
      "epoch": 0.9653374056359131,
      "grad_norm": 0.7333370447158813,
      "learning_rate": 3.147627854251434e-08,
      "loss": 0.6836,
      "step": 33503
    },
    {
      "epoch": 0.9653662190975624,
      "grad_norm": 0.7279486060142517,
      "learning_rate": 3.1424025511886947e-08,
      "loss": 0.6889,
      "step": 33504
    },
    {
      "epoch": 0.9653950325592117,
      "grad_norm": 0.7366810441017151,
      "learning_rate": 3.137181575258996e-08,
      "loss": 0.6913,
      "step": 33505
    },
    {
      "epoch": 0.965423846020861,
      "grad_norm": 0.7166081070899963,
      "learning_rate": 3.131964926507747e-08,
      "loss": 0.6712,
      "step": 33506
    },
    {
      "epoch": 0.9654526594825102,
      "grad_norm": 0.7253903746604919,
      "learning_rate": 3.1267526049804095e-08,
      "loss": 0.6875,
      "step": 33507
    },
    {
      "epoch": 0.9654814729441595,
      "grad_norm": 0.7267438769340515,
      "learning_rate": 3.121544610722394e-08,
      "loss": 0.697,
      "step": 33508
    },
    {
      "epoch": 0.9655102864058088,
      "grad_norm": 0.7131616473197937,
      "learning_rate": 3.1163409437789396e-08,
      "loss": 0.643,
      "step": 33509
    },
    {
      "epoch": 0.9655390998674581,
      "grad_norm": 0.7161325216293335,
      "learning_rate": 3.111141604195511e-08,
      "loss": 0.6578,
      "step": 33510
    },
    {
      "epoch": 0.9655679133291074,
      "grad_norm": 0.7091350555419922,
      "learning_rate": 3.10594659201735e-08,
      "loss": 0.6867,
      "step": 33511
    },
    {
      "epoch": 0.9655967267907566,
      "grad_norm": 0.7296358346939087,
      "learning_rate": 3.1007559072896433e-08,
      "loss": 0.692,
      "step": 33512
    },
    {
      "epoch": 0.9656255402524059,
      "grad_norm": 0.7136300802230835,
      "learning_rate": 3.0955695500576866e-08,
      "loss": 0.6782,
      "step": 33513
    },
    {
      "epoch": 0.9656543537140552,
      "grad_norm": 0.7138778567314148,
      "learning_rate": 3.090387520366556e-08,
      "loss": 0.6793,
      "step": 33514
    },
    {
      "epoch": 0.9656831671757045,
      "grad_norm": 0.7139182686805725,
      "learning_rate": 3.085209818261436e-08,
      "loss": 0.6745,
      "step": 33515
    },
    {
      "epoch": 0.9657119806373538,
      "grad_norm": 0.7185634970664978,
      "learning_rate": 3.080036443787404e-08,
      "loss": 0.673,
      "step": 33516
    },
    {
      "epoch": 0.965740794099003,
      "grad_norm": 0.7086306214332581,
      "learning_rate": 3.074867396989478e-08,
      "loss": 0.6568,
      "step": 33517
    },
    {
      "epoch": 0.9657696075606523,
      "grad_norm": 0.7489545345306396,
      "learning_rate": 3.069702677912734e-08,
      "loss": 0.6939,
      "step": 33518
    },
    {
      "epoch": 0.9657984210223016,
      "grad_norm": 0.7229114174842834,
      "learning_rate": 3.06454228660219e-08,
      "loss": 0.6745,
      "step": 33519
    },
    {
      "epoch": 0.9658272344839509,
      "grad_norm": 0.714368999004364,
      "learning_rate": 3.059386223102645e-08,
      "loss": 0.6482,
      "step": 33520
    },
    {
      "epoch": 0.9658560479456002,
      "grad_norm": 0.7145334482192993,
      "learning_rate": 3.0542344874591734e-08,
      "loss": 0.6695,
      "step": 33521
    },
    {
      "epoch": 0.9658848614072495,
      "grad_norm": 0.7066526412963867,
      "learning_rate": 3.049087079716462e-08,
      "loss": 0.6936,
      "step": 33522
    },
    {
      "epoch": 0.9659136748688988,
      "grad_norm": 0.7209342122077942,
      "learning_rate": 3.04394399991953e-08,
      "loss": 0.6936,
      "step": 33523
    },
    {
      "epoch": 0.9659424883305481,
      "grad_norm": 0.7262620329856873,
      "learning_rate": 3.0388052481130084e-08,
      "loss": 0.6714,
      "step": 33524
    },
    {
      "epoch": 0.9659713017921974,
      "grad_norm": 0.7392483949661255,
      "learning_rate": 3.033670824341695e-08,
      "loss": 0.6734,
      "step": 33525
    },
    {
      "epoch": 0.9660001152538465,
      "grad_norm": 0.7135957479476929,
      "learning_rate": 3.028540728650386e-08,
      "loss": 0.6849,
      "step": 33526
    },
    {
      "epoch": 0.9660289287154958,
      "grad_norm": 0.7335455417633057,
      "learning_rate": 3.0234149610836595e-08,
      "loss": 0.6827,
      "step": 33527
    },
    {
      "epoch": 0.9660577421771451,
      "grad_norm": 0.7114657759666443,
      "learning_rate": 3.0182935216861996e-08,
      "loss": 0.6849,
      "step": 33528
    },
    {
      "epoch": 0.9660865556387944,
      "grad_norm": 0.7285649180412292,
      "learning_rate": 3.013176410502583e-08,
      "loss": 0.6804,
      "step": 33529
    },
    {
      "epoch": 0.9661153691004437,
      "grad_norm": 0.7291226387023926,
      "learning_rate": 3.00806362757744e-08,
      "loss": 0.6665,
      "step": 33530
    },
    {
      "epoch": 0.966144182562093,
      "grad_norm": 0.7137896418571472,
      "learning_rate": 3.0029551729552353e-08,
      "loss": 0.6696,
      "step": 33531
    },
    {
      "epoch": 0.9661729960237423,
      "grad_norm": 0.7112722992897034,
      "learning_rate": 2.997851046680433e-08,
      "loss": 0.6922,
      "step": 33532
    },
    {
      "epoch": 0.9662018094853916,
      "grad_norm": 0.7207498550415039,
      "learning_rate": 2.992751248797554e-08,
      "loss": 0.6797,
      "step": 33533
    },
    {
      "epoch": 0.9662306229470409,
      "grad_norm": 0.7039243578910828,
      "learning_rate": 2.987655779351062e-08,
      "loss": 0.6794,
      "step": 33534
    },
    {
      "epoch": 0.9662594364086902,
      "grad_norm": 0.7108692526817322,
      "learning_rate": 2.9825646383851434e-08,
      "loss": 0.6596,
      "step": 33535
    },
    {
      "epoch": 0.9662882498703395,
      "grad_norm": 0.7207919955253601,
      "learning_rate": 2.9774778259443193e-08,
      "loss": 0.6895,
      "step": 33536
    },
    {
      "epoch": 0.9663170633319887,
      "grad_norm": 0.716136634349823,
      "learning_rate": 2.9723953420728313e-08,
      "loss": 0.662,
      "step": 33537
    },
    {
      "epoch": 0.966345876793638,
      "grad_norm": 0.7085817456245422,
      "learning_rate": 2.967317186814922e-08,
      "loss": 0.6923,
      "step": 33538
    },
    {
      "epoch": 0.9663746902552872,
      "grad_norm": 0.7245648503303528,
      "learning_rate": 2.9622433602147782e-08,
      "loss": 0.6693,
      "step": 33539
    },
    {
      "epoch": 0.9664035037169365,
      "grad_norm": 0.7297000885009766,
      "learning_rate": 2.9571738623166978e-08,
      "loss": 0.6598,
      "step": 33540
    },
    {
      "epoch": 0.9664323171785858,
      "grad_norm": 0.7150769829750061,
      "learning_rate": 2.9521086931647013e-08,
      "loss": 0.6688,
      "step": 33541
    },
    {
      "epoch": 0.9664611306402351,
      "grad_norm": 0.7107855081558228,
      "learning_rate": 2.947047852803031e-08,
      "loss": 0.6894,
      "step": 33542
    },
    {
      "epoch": 0.9664899441018844,
      "grad_norm": 0.724632740020752,
      "learning_rate": 2.9419913412757074e-08,
      "loss": 0.6853,
      "step": 33543
    },
    {
      "epoch": 0.9665187575635337,
      "grad_norm": 0.73424232006073,
      "learning_rate": 2.9369391586267504e-08,
      "loss": 0.6769,
      "step": 33544
    },
    {
      "epoch": 0.966547571025183,
      "grad_norm": 0.7230920195579529,
      "learning_rate": 2.9318913049001807e-08,
      "loss": 0.6877,
      "step": 33545
    },
    {
      "epoch": 0.9665763844868323,
      "grad_norm": 0.7170534729957581,
      "learning_rate": 2.9268477801399075e-08,
      "loss": 0.6587,
      "step": 33546
    },
    {
      "epoch": 0.9666051979484815,
      "grad_norm": 0.7208608984947205,
      "learning_rate": 2.9218085843899512e-08,
      "loss": 0.6896,
      "step": 33547
    },
    {
      "epoch": 0.9666340114101308,
      "grad_norm": 0.727950930595398,
      "learning_rate": 2.9167737176941102e-08,
      "loss": 0.6664,
      "step": 33548
    },
    {
      "epoch": 0.9666628248717801,
      "grad_norm": 0.7192612290382385,
      "learning_rate": 2.9117431800962935e-08,
      "loss": 0.6966,
      "step": 33549
    },
    {
      "epoch": 0.9666916383334294,
      "grad_norm": 0.7165942192077637,
      "learning_rate": 2.9067169716403e-08,
      "loss": 0.685,
      "step": 33550
    },
    {
      "epoch": 0.9667204517950787,
      "grad_norm": 0.7236785292625427,
      "learning_rate": 2.9016950923698718e-08,
      "loss": 0.6853,
      "step": 33551
    },
    {
      "epoch": 0.966749265256728,
      "grad_norm": 0.7089643478393555,
      "learning_rate": 2.8966775423287518e-08,
      "loss": 0.6813,
      "step": 33552
    },
    {
      "epoch": 0.9667780787183772,
      "grad_norm": 0.7112736105918884,
      "learning_rate": 2.8916643215606277e-08,
      "loss": 0.6636,
      "step": 33553
    },
    {
      "epoch": 0.9668068921800265,
      "grad_norm": 0.7127335667610168,
      "learning_rate": 2.8866554301091863e-08,
      "loss": 0.6842,
      "step": 33554
    },
    {
      "epoch": 0.9668357056416758,
      "grad_norm": 0.7072522044181824,
      "learning_rate": 2.8816508680181156e-08,
      "loss": 0.6532,
      "step": 33555
    },
    {
      "epoch": 0.966864519103325,
      "grad_norm": 0.7025697231292725,
      "learning_rate": 2.8766506353308798e-08,
      "loss": 0.6583,
      "step": 33556
    },
    {
      "epoch": 0.9668933325649743,
      "grad_norm": 0.719175398349762,
      "learning_rate": 2.8716547320910558e-08,
      "loss": 0.6918,
      "step": 33557
    },
    {
      "epoch": 0.9669221460266236,
      "grad_norm": 0.7585785388946533,
      "learning_rate": 2.8666631583422198e-08,
      "loss": 0.6585,
      "step": 33558
    },
    {
      "epoch": 0.9669509594882729,
      "grad_norm": 0.7178446054458618,
      "learning_rate": 2.8616759141277816e-08,
      "loss": 0.6582,
      "step": 33559
    },
    {
      "epoch": 0.9669797729499222,
      "grad_norm": 0.7219861149787903,
      "learning_rate": 2.8566929994911507e-08,
      "loss": 0.663,
      "step": 33560
    },
    {
      "epoch": 0.9670085864115715,
      "grad_norm": 0.7250496745109558,
      "learning_rate": 2.8517144144757924e-08,
      "loss": 0.6938,
      "step": 33561
    },
    {
      "epoch": 0.9670373998732208,
      "grad_norm": 0.7404246926307678,
      "learning_rate": 2.846740159125061e-08,
      "loss": 0.6991,
      "step": 33562
    },
    {
      "epoch": 0.9670662133348701,
      "grad_norm": 0.7390040159225464,
      "learning_rate": 2.8417702334821994e-08,
      "loss": 0.663,
      "step": 33563
    },
    {
      "epoch": 0.9670950267965194,
      "grad_norm": 0.7241301536560059,
      "learning_rate": 2.8368046375905623e-08,
      "loss": 0.6756,
      "step": 33564
    },
    {
      "epoch": 0.9671238402581687,
      "grad_norm": 0.7047439813613892,
      "learning_rate": 2.8318433714933926e-08,
      "loss": 0.6478,
      "step": 33565
    },
    {
      "epoch": 0.9671526537198178,
      "grad_norm": 0.7056583762168884,
      "learning_rate": 2.8268864352338776e-08,
      "loss": 0.6698,
      "step": 33566
    },
    {
      "epoch": 0.9671814671814671,
      "grad_norm": 0.7135624885559082,
      "learning_rate": 2.8219338288551502e-08,
      "loss": 0.6687,
      "step": 33567
    },
    {
      "epoch": 0.9672102806431164,
      "grad_norm": 0.7233796119689941,
      "learning_rate": 2.816985552400453e-08,
      "loss": 0.6978,
      "step": 33568
    },
    {
      "epoch": 0.9672390941047657,
      "grad_norm": 0.7232252955436707,
      "learning_rate": 2.812041605912752e-08,
      "loss": 0.6816,
      "step": 33569
    },
    {
      "epoch": 0.967267907566415,
      "grad_norm": 0.7388404607772827,
      "learning_rate": 2.807101989435179e-08,
      "loss": 0.7061,
      "step": 33570
    },
    {
      "epoch": 0.9672967210280643,
      "grad_norm": 0.7151744961738586,
      "learning_rate": 2.802166703010756e-08,
      "loss": 0.6612,
      "step": 33571
    },
    {
      "epoch": 0.9673255344897136,
      "grad_norm": 0.7255681753158569,
      "learning_rate": 2.7972357466824473e-08,
      "loss": 0.6968,
      "step": 33572
    },
    {
      "epoch": 0.9673543479513629,
      "grad_norm": 0.7331733703613281,
      "learning_rate": 2.792309120493164e-08,
      "loss": 0.7077,
      "step": 33573
    },
    {
      "epoch": 0.9673831614130122,
      "grad_norm": 0.7257696986198425,
      "learning_rate": 2.7873868244858716e-08,
      "loss": 0.6653,
      "step": 33574
    },
    {
      "epoch": 0.9674119748746615,
      "grad_norm": 0.7115204334259033,
      "learning_rate": 2.782468858703369e-08,
      "loss": 0.6627,
      "step": 33575
    },
    {
      "epoch": 0.9674407883363108,
      "grad_norm": 0.7276075482368469,
      "learning_rate": 2.777555223188566e-08,
      "loss": 0.6964,
      "step": 33576
    },
    {
      "epoch": 0.96746960179796,
      "grad_norm": 0.7522808313369751,
      "learning_rate": 2.7726459179841513e-08,
      "loss": 0.6576,
      "step": 33577
    },
    {
      "epoch": 0.9674984152596093,
      "grad_norm": 0.7090964913368225,
      "learning_rate": 2.7677409431330348e-08,
      "loss": 0.6841,
      "step": 33578
    },
    {
      "epoch": 0.9675272287212586,
      "grad_norm": 0.7006648778915405,
      "learning_rate": 2.7628402986777935e-08,
      "loss": 0.6572,
      "step": 33579
    },
    {
      "epoch": 0.9675560421829078,
      "grad_norm": 0.720775842666626,
      "learning_rate": 2.757943984661171e-08,
      "loss": 0.6761,
      "step": 33580
    },
    {
      "epoch": 0.9675848556445571,
      "grad_norm": 0.7179751992225647,
      "learning_rate": 2.7530520011258e-08,
      "loss": 0.6744,
      "step": 33581
    },
    {
      "epoch": 0.9676136691062064,
      "grad_norm": 0.7290374040603638,
      "learning_rate": 2.7481643481142573e-08,
      "loss": 0.6812,
      "step": 33582
    },
    {
      "epoch": 0.9676424825678557,
      "grad_norm": 0.7051748633384705,
      "learning_rate": 2.7432810256691757e-08,
      "loss": 0.6652,
      "step": 33583
    },
    {
      "epoch": 0.967671296029505,
      "grad_norm": 0.7078762650489807,
      "learning_rate": 2.738402033833021e-08,
      "loss": 0.6987,
      "step": 33584
    },
    {
      "epoch": 0.9677001094911543,
      "grad_norm": 0.7392968535423279,
      "learning_rate": 2.7335273726483146e-08,
      "loss": 0.69,
      "step": 33585
    },
    {
      "epoch": 0.9677289229528035,
      "grad_norm": 0.7079721093177795,
      "learning_rate": 2.7286570421574677e-08,
      "loss": 0.6787,
      "step": 33586
    },
    {
      "epoch": 0.9677577364144528,
      "grad_norm": 0.7150607705116272,
      "learning_rate": 2.7237910424029458e-08,
      "loss": 0.6896,
      "step": 33587
    },
    {
      "epoch": 0.9677865498761021,
      "grad_norm": 0.8886491656303406,
      "learning_rate": 2.7189293734271038e-08,
      "loss": 0.6619,
      "step": 33588
    },
    {
      "epoch": 0.9678153633377514,
      "grad_norm": 0.728836178779602,
      "learning_rate": 2.714072035272297e-08,
      "loss": 0.6974,
      "step": 33589
    },
    {
      "epoch": 0.9678441767994007,
      "grad_norm": 0.714057981967926,
      "learning_rate": 2.709219027980825e-08,
      "loss": 0.6857,
      "step": 33590
    },
    {
      "epoch": 0.96787299026105,
      "grad_norm": 0.7566927671432495,
      "learning_rate": 2.7043703515949316e-08,
      "loss": 0.67,
      "step": 33591
    },
    {
      "epoch": 0.9679018037226993,
      "grad_norm": 0.7379560470581055,
      "learning_rate": 2.6995260061568606e-08,
      "loss": 0.6711,
      "step": 33592
    },
    {
      "epoch": 0.9679306171843485,
      "grad_norm": 0.7277030944824219,
      "learning_rate": 2.694685991708801e-08,
      "loss": 0.6677,
      "step": 33593
    },
    {
      "epoch": 0.9679594306459978,
      "grad_norm": 0.7300306558609009,
      "learning_rate": 2.6898503082929404e-08,
      "loss": 0.6563,
      "step": 33594
    },
    {
      "epoch": 0.9679882441076471,
      "grad_norm": 0.7131714224815369,
      "learning_rate": 2.685018955951302e-08,
      "loss": 0.6858,
      "step": 33595
    },
    {
      "epoch": 0.9680170575692963,
      "grad_norm": 0.701349675655365,
      "learning_rate": 2.6801919347260173e-08,
      "loss": 0.6599,
      "step": 33596
    },
    {
      "epoch": 0.9680458710309456,
      "grad_norm": 0.7236615419387817,
      "learning_rate": 2.675369244659165e-08,
      "loss": 0.6699,
      "step": 33597
    },
    {
      "epoch": 0.9680746844925949,
      "grad_norm": 0.7131595611572266,
      "learning_rate": 2.6705508857926555e-08,
      "loss": 0.6777,
      "step": 33598
    },
    {
      "epoch": 0.9681034979542442,
      "grad_norm": 0.7239644527435303,
      "learning_rate": 2.6657368581685662e-08,
      "loss": 0.672,
      "step": 33599
    },
    {
      "epoch": 0.9681323114158935,
      "grad_norm": 0.710882306098938,
      "learning_rate": 2.660927161828697e-08,
      "loss": 0.6612,
      "step": 33600
    },
    {
      "epoch": 0.9681611248775428,
      "grad_norm": 0.7085238099098206,
      "learning_rate": 2.6561217968150144e-08,
      "loss": 0.6887,
      "step": 33601
    },
    {
      "epoch": 0.9681899383391921,
      "grad_norm": 0.7294787168502808,
      "learning_rate": 2.6513207631693184e-08,
      "loss": 0.6601,
      "step": 33602
    },
    {
      "epoch": 0.9682187518008414,
      "grad_norm": 0.6887037754058838,
      "learning_rate": 2.6465240609335196e-08,
      "loss": 0.6503,
      "step": 33603
    },
    {
      "epoch": 0.9682475652624907,
      "grad_norm": 0.7196118831634521,
      "learning_rate": 2.641731690149252e-08,
      "loss": 0.6818,
      "step": 33604
    },
    {
      "epoch": 0.96827637872414,
      "grad_norm": 0.7396547198295593,
      "learning_rate": 2.63694365085837e-08,
      "loss": 0.6875,
      "step": 33605
    },
    {
      "epoch": 0.9683051921857893,
      "grad_norm": 0.7262020111083984,
      "learning_rate": 2.6321599431025636e-08,
      "loss": 0.6982,
      "step": 33606
    },
    {
      "epoch": 0.9683340056474384,
      "grad_norm": 0.725700855255127,
      "learning_rate": 2.6273805669233543e-08,
      "loss": 0.6941,
      "step": 33607
    },
    {
      "epoch": 0.9683628191090877,
      "grad_norm": 0.7231022119522095,
      "learning_rate": 2.6226055223625424e-08,
      "loss": 0.662,
      "step": 33608
    },
    {
      "epoch": 0.968391632570737,
      "grad_norm": 0.7274203300476074,
      "learning_rate": 2.61783480946165e-08,
      "loss": 0.6775,
      "step": 33609
    },
    {
      "epoch": 0.9684204460323863,
      "grad_norm": 0.7400140166282654,
      "learning_rate": 2.6130684282622e-08,
      "loss": 0.6817,
      "step": 33610
    },
    {
      "epoch": 0.9684492594940356,
      "grad_norm": 0.7189903855323792,
      "learning_rate": 2.608306378805714e-08,
      "loss": 0.6616,
      "step": 33611
    },
    {
      "epoch": 0.9684780729556849,
      "grad_norm": 0.7187613844871521,
      "learning_rate": 2.603548661133659e-08,
      "loss": 0.6862,
      "step": 33612
    },
    {
      "epoch": 0.9685068864173342,
      "grad_norm": 0.7270729541778564,
      "learning_rate": 2.5987952752874467e-08,
      "loss": 0.6624,
      "step": 33613
    },
    {
      "epoch": 0.9685356998789835,
      "grad_norm": 0.7765740752220154,
      "learning_rate": 2.5940462213085992e-08,
      "loss": 0.6836,
      "step": 33614
    },
    {
      "epoch": 0.9685645133406328,
      "grad_norm": 0.7305033206939697,
      "learning_rate": 2.589301499238306e-08,
      "loss": 0.6712,
      "step": 33615
    },
    {
      "epoch": 0.968593326802282,
      "grad_norm": 0.7145012021064758,
      "learning_rate": 2.5845611091180334e-08,
      "loss": 0.6803,
      "step": 33616
    },
    {
      "epoch": 0.9686221402639313,
      "grad_norm": 0.7354626059532166,
      "learning_rate": 2.579825050988971e-08,
      "loss": 0.6756,
      "step": 33617
    },
    {
      "epoch": 0.9686509537255806,
      "grad_norm": 0.7084357738494873,
      "learning_rate": 2.575093324892364e-08,
      "loss": 0.6703,
      "step": 33618
    },
    {
      "epoch": 0.9686797671872299,
      "grad_norm": 0.7419735789299011,
      "learning_rate": 2.570365930869456e-08,
      "loss": 0.6945,
      "step": 33619
    },
    {
      "epoch": 0.9687085806488791,
      "grad_norm": 0.7395293116569519,
      "learning_rate": 2.565642868961493e-08,
      "loss": 0.6802,
      "step": 33620
    },
    {
      "epoch": 0.9687373941105284,
      "grad_norm": 0.7076479196548462,
      "learning_rate": 2.560924139209442e-08,
      "loss": 0.6678,
      "step": 33621
    },
    {
      "epoch": 0.9687662075721777,
      "grad_norm": 0.720928430557251,
      "learning_rate": 2.5562097416544918e-08,
      "loss": 0.6635,
      "step": 33622
    },
    {
      "epoch": 0.968795021033827,
      "grad_norm": 0.7206999063491821,
      "learning_rate": 2.551499676337721e-08,
      "loss": 0.6933,
      "step": 33623
    },
    {
      "epoch": 0.9688238344954763,
      "grad_norm": 0.7175103425979614,
      "learning_rate": 2.5467939433000967e-08,
      "loss": 0.6904,
      "step": 33624
    },
    {
      "epoch": 0.9688526479571256,
      "grad_norm": 0.7037082314491272,
      "learning_rate": 2.5420925425826414e-08,
      "loss": 0.6468,
      "step": 33625
    },
    {
      "epoch": 0.9688814614187748,
      "grad_norm": 0.7252397537231445,
      "learning_rate": 2.5373954742263228e-08,
      "loss": 0.6634,
      "step": 33626
    },
    {
      "epoch": 0.9689102748804241,
      "grad_norm": 0.7157010436058044,
      "learning_rate": 2.5327027382719417e-08,
      "loss": 0.662,
      "step": 33627
    },
    {
      "epoch": 0.9689390883420734,
      "grad_norm": 0.7315675020217896,
      "learning_rate": 2.5280143347604648e-08,
      "loss": 0.6882,
      "step": 33628
    },
    {
      "epoch": 0.9689679018037227,
      "grad_norm": 0.7086174488067627,
      "learning_rate": 2.5233302637326928e-08,
      "loss": 0.675,
      "step": 33629
    },
    {
      "epoch": 0.968996715265372,
      "grad_norm": 0.7173004746437073,
      "learning_rate": 2.518650525229427e-08,
      "loss": 0.681,
      "step": 33630
    },
    {
      "epoch": 0.9690255287270213,
      "grad_norm": 0.7376139760017395,
      "learning_rate": 2.5139751192914118e-08,
      "loss": 0.6874,
      "step": 33631
    },
    {
      "epoch": 0.9690543421886706,
      "grad_norm": 0.7174700498580933,
      "learning_rate": 2.5093040459593375e-08,
      "loss": 0.6542,
      "step": 33632
    },
    {
      "epoch": 0.9690831556503199,
      "grad_norm": 0.7215967774391174,
      "learning_rate": 2.5046373052740048e-08,
      "loss": 0.678,
      "step": 33633
    },
    {
      "epoch": 0.9691119691119691,
      "grad_norm": 0.710097074508667,
      "learning_rate": 2.499974897275881e-08,
      "loss": 0.6805,
      "step": 33634
    },
    {
      "epoch": 0.9691407825736184,
      "grad_norm": 0.7227267026901245,
      "learning_rate": 2.495316822005711e-08,
      "loss": 0.6642,
      "step": 33635
    },
    {
      "epoch": 0.9691695960352676,
      "grad_norm": 0.7315069437026978,
      "learning_rate": 2.4906630795039634e-08,
      "loss": 0.7105,
      "step": 33636
    },
    {
      "epoch": 0.9691984094969169,
      "grad_norm": 0.7293723225593567,
      "learning_rate": 2.486013669811216e-08,
      "loss": 0.703,
      "step": 33637
    },
    {
      "epoch": 0.9692272229585662,
      "grad_norm": 0.7261854410171509,
      "learning_rate": 2.4813685929679367e-08,
      "loss": 0.6606,
      "step": 33638
    },
    {
      "epoch": 0.9692560364202155,
      "grad_norm": 0.7361488938331604,
      "learning_rate": 2.4767278490146486e-08,
      "loss": 0.6802,
      "step": 33639
    },
    {
      "epoch": 0.9692848498818648,
      "grad_norm": 0.7247524857521057,
      "learning_rate": 2.4720914379917083e-08,
      "loss": 0.6892,
      "step": 33640
    },
    {
      "epoch": 0.9693136633435141,
      "grad_norm": 0.7183609008789062,
      "learning_rate": 2.467459359939528e-08,
      "loss": 0.6699,
      "step": 33641
    },
    {
      "epoch": 0.9693424768051634,
      "grad_norm": 0.7448099255561829,
      "learning_rate": 2.4628316148984088e-08,
      "loss": 0.6955,
      "step": 33642
    },
    {
      "epoch": 0.9693712902668127,
      "grad_norm": 0.7157201170921326,
      "learning_rate": 2.4582082029086517e-08,
      "loss": 0.691,
      "step": 33643
    },
    {
      "epoch": 0.969400103728462,
      "grad_norm": 0.7169587016105652,
      "learning_rate": 2.453589124010558e-08,
      "loss": 0.6691,
      "step": 33644
    },
    {
      "epoch": 0.9694289171901113,
      "grad_norm": 0.7168101668357849,
      "learning_rate": 2.4489743782443175e-08,
      "loss": 0.6851,
      "step": 33645
    },
    {
      "epoch": 0.9694577306517606,
      "grad_norm": 0.7140206694602966,
      "learning_rate": 2.444363965650176e-08,
      "loss": 0.6658,
      "step": 33646
    },
    {
      "epoch": 0.9694865441134097,
      "grad_norm": 0.7210639715194702,
      "learning_rate": 2.439757886268268e-08,
      "loss": 0.6343,
      "step": 33647
    },
    {
      "epoch": 0.969515357575059,
      "grad_norm": 0.7054141163825989,
      "learning_rate": 2.4351561401386726e-08,
      "loss": 0.6733,
      "step": 33648
    },
    {
      "epoch": 0.9695441710367083,
      "grad_norm": 0.7279097437858582,
      "learning_rate": 2.430558727301413e-08,
      "loss": 0.6722,
      "step": 33649
    },
    {
      "epoch": 0.9695729844983576,
      "grad_norm": 0.7405238747596741,
      "learning_rate": 2.42596564779668e-08,
      "loss": 0.6956,
      "step": 33650
    },
    {
      "epoch": 0.9696017979600069,
      "grad_norm": 0.7222398519515991,
      "learning_rate": 2.4213769016643298e-08,
      "loss": 0.6894,
      "step": 33651
    },
    {
      "epoch": 0.9696306114216562,
      "grad_norm": 0.7107517719268799,
      "learning_rate": 2.4167924889444417e-08,
      "loss": 0.6816,
      "step": 33652
    },
    {
      "epoch": 0.9696594248833055,
      "grad_norm": 0.7109967470169067,
      "learning_rate": 2.412212409676873e-08,
      "loss": 0.6662,
      "step": 33653
    },
    {
      "epoch": 0.9696882383449548,
      "grad_norm": 0.7158209681510925,
      "learning_rate": 2.4076366639015914e-08,
      "loss": 0.6664,
      "step": 33654
    },
    {
      "epoch": 0.9697170518066041,
      "grad_norm": 0.7316116690635681,
      "learning_rate": 2.4030652516582876e-08,
      "loss": 0.6746,
      "step": 33655
    },
    {
      "epoch": 0.9697458652682533,
      "grad_norm": 0.7093631029129028,
      "learning_rate": 2.3984981729869295e-08,
      "loss": 0.667,
      "step": 33656
    },
    {
      "epoch": 0.9697746787299026,
      "grad_norm": 0.7192418575286865,
      "learning_rate": 2.3939354279272077e-08,
      "loss": 0.6827,
      "step": 33657
    },
    {
      "epoch": 0.9698034921915519,
      "grad_norm": 0.723856508731842,
      "learning_rate": 2.3893770165189234e-08,
      "loss": 0.6589,
      "step": 33658
    },
    {
      "epoch": 0.9698323056532012,
      "grad_norm": 0.7376343607902527,
      "learning_rate": 2.3848229388016565e-08,
      "loss": 0.6843,
      "step": 33659
    },
    {
      "epoch": 0.9698611191148505,
      "grad_norm": 0.7313419580459595,
      "learning_rate": 2.380273194815208e-08,
      "loss": 0.6801,
      "step": 33660
    },
    {
      "epoch": 0.9698899325764997,
      "grad_norm": 0.7222840189933777,
      "learning_rate": 2.3757277845991022e-08,
      "loss": 0.6682,
      "step": 33661
    },
    {
      "epoch": 0.969918746038149,
      "grad_norm": 0.7368611097335815,
      "learning_rate": 2.3711867081930296e-08,
      "loss": 0.6835,
      "step": 33662
    },
    {
      "epoch": 0.9699475594997983,
      "grad_norm": 0.7466867566108704,
      "learning_rate": 2.366649965636403e-08,
      "loss": 0.6648,
      "step": 33663
    },
    {
      "epoch": 0.9699763729614476,
      "grad_norm": 0.7182645201683044,
      "learning_rate": 2.3621175569688572e-08,
      "loss": 0.6884,
      "step": 33664
    },
    {
      "epoch": 0.9700051864230969,
      "grad_norm": 0.7179938554763794,
      "learning_rate": 2.357589482229805e-08,
      "loss": 0.6739,
      "step": 33665
    },
    {
      "epoch": 0.9700339998847461,
      "grad_norm": 0.7155442237854004,
      "learning_rate": 2.35306574145866e-08,
      "loss": 0.6745,
      "step": 33666
    },
    {
      "epoch": 0.9700628133463954,
      "grad_norm": 0.7280820608139038,
      "learning_rate": 2.34854633469489e-08,
      "loss": 0.6802,
      "step": 33667
    },
    {
      "epoch": 0.9700916268080447,
      "grad_norm": 0.7308440804481506,
      "learning_rate": 2.3440312619777416e-08,
      "loss": 0.6843,
      "step": 33668
    },
    {
      "epoch": 0.970120440269694,
      "grad_norm": 0.7376726865768433,
      "learning_rate": 2.3395205233466832e-08,
      "loss": 0.6868,
      "step": 33669
    },
    {
      "epoch": 0.9701492537313433,
      "grad_norm": 0.7453028559684753,
      "learning_rate": 2.3350141188409058e-08,
      "loss": 0.6736,
      "step": 33670
    },
    {
      "epoch": 0.9701780671929926,
      "grad_norm": 0.7100004553794861,
      "learning_rate": 2.3305120484996558e-08,
      "loss": 0.6823,
      "step": 33671
    },
    {
      "epoch": 0.9702068806546419,
      "grad_norm": 0.7284939289093018,
      "learning_rate": 2.3260143123621236e-08,
      "loss": 0.6965,
      "step": 33672
    },
    {
      "epoch": 0.9702356941162912,
      "grad_norm": 0.736574649810791,
      "learning_rate": 2.321520910467556e-08,
      "loss": 0.6896,
      "step": 33673
    },
    {
      "epoch": 0.9702645075779404,
      "grad_norm": 0.7126044034957886,
      "learning_rate": 2.317031842855033e-08,
      "loss": 0.6713,
      "step": 33674
    },
    {
      "epoch": 0.9702933210395897,
      "grad_norm": 0.7158164978027344,
      "learning_rate": 2.3125471095636897e-08,
      "loss": 0.6842,
      "step": 33675
    },
    {
      "epoch": 0.9703221345012389,
      "grad_norm": 0.7024522423744202,
      "learning_rate": 2.308066710632495e-08,
      "loss": 0.6673,
      "step": 33676
    },
    {
      "epoch": 0.9703509479628882,
      "grad_norm": 0.7329788208007812,
      "learning_rate": 2.3035906461005843e-08,
      "loss": 0.6923,
      "step": 33677
    },
    {
      "epoch": 0.9703797614245375,
      "grad_norm": 0.7267927527427673,
      "learning_rate": 2.2991189160068704e-08,
      "loss": 0.6734,
      "step": 33678
    },
    {
      "epoch": 0.9704085748861868,
      "grad_norm": 0.7193641662597656,
      "learning_rate": 2.294651520390323e-08,
      "loss": 0.6531,
      "step": 33679
    },
    {
      "epoch": 0.9704373883478361,
      "grad_norm": 0.7129819393157959,
      "learning_rate": 2.290188459289855e-08,
      "loss": 0.6839,
      "step": 33680
    },
    {
      "epoch": 0.9704662018094854,
      "grad_norm": 0.7151196599006653,
      "learning_rate": 2.2857297327442684e-08,
      "loss": 0.6835,
      "step": 33681
    },
    {
      "epoch": 0.9704950152711347,
      "grad_norm": 0.713374674320221,
      "learning_rate": 2.281275340792477e-08,
      "loss": 0.663,
      "step": 33682
    },
    {
      "epoch": 0.970523828732784,
      "grad_norm": 0.7114974856376648,
      "learning_rate": 2.2768252834732275e-08,
      "loss": 0.6734,
      "step": 33683
    },
    {
      "epoch": 0.9705526421944333,
      "grad_norm": 0.739587128162384,
      "learning_rate": 2.2723795608252662e-08,
      "loss": 0.7013,
      "step": 33684
    },
    {
      "epoch": 0.9705814556560826,
      "grad_norm": 0.7445164322853088,
      "learning_rate": 2.2679381728873406e-08,
      "loss": 0.684,
      "step": 33685
    },
    {
      "epoch": 0.9706102691177318,
      "grad_norm": 0.7288552522659302,
      "learning_rate": 2.2635011196981417e-08,
      "loss": 0.6648,
      "step": 33686
    },
    {
      "epoch": 0.9706390825793811,
      "grad_norm": 0.7163743376731873,
      "learning_rate": 2.2590684012963047e-08,
      "loss": 0.6571,
      "step": 33687
    },
    {
      "epoch": 0.9706678960410303,
      "grad_norm": 0.7342977523803711,
      "learning_rate": 2.2546400177204107e-08,
      "loss": 0.6841,
      "step": 33688
    },
    {
      "epoch": 0.9706967095026796,
      "grad_norm": 0.714555025100708,
      "learning_rate": 2.2502159690090398e-08,
      "loss": 0.6834,
      "step": 33689
    },
    {
      "epoch": 0.9707255229643289,
      "grad_norm": 0.7156234979629517,
      "learning_rate": 2.2457962552007162e-08,
      "loss": 0.6815,
      "step": 33690
    },
    {
      "epoch": 0.9707543364259782,
      "grad_norm": 0.7496045827865601,
      "learning_rate": 2.2413808763339094e-08,
      "loss": 0.6935,
      "step": 33691
    },
    {
      "epoch": 0.9707831498876275,
      "grad_norm": 0.7236150503158569,
      "learning_rate": 2.2369698324471446e-08,
      "loss": 0.6718,
      "step": 33692
    },
    {
      "epoch": 0.9708119633492768,
      "grad_norm": 0.7156614065170288,
      "learning_rate": 2.2325631235787793e-08,
      "loss": 0.6945,
      "step": 33693
    },
    {
      "epoch": 0.9708407768109261,
      "grad_norm": 0.7163314819335938,
      "learning_rate": 2.2281607497672274e-08,
      "loss": 0.675,
      "step": 33694
    },
    {
      "epoch": 0.9708695902725754,
      "grad_norm": 0.7263767719268799,
      "learning_rate": 2.2237627110507367e-08,
      "loss": 0.6495,
      "step": 33695
    },
    {
      "epoch": 0.9708984037342246,
      "grad_norm": 0.7246874570846558,
      "learning_rate": 2.2193690074677198e-08,
      "loss": 0.6693,
      "step": 33696
    },
    {
      "epoch": 0.9709272171958739,
      "grad_norm": 0.7278626561164856,
      "learning_rate": 2.214979639056425e-08,
      "loss": 0.6777,
      "step": 33697
    },
    {
      "epoch": 0.9709560306575232,
      "grad_norm": 0.7255572080612183,
      "learning_rate": 2.2105946058549878e-08,
      "loss": 0.6686,
      "step": 33698
    },
    {
      "epoch": 0.9709848441191725,
      "grad_norm": 0.7123012542724609,
      "learning_rate": 2.206213907901711e-08,
      "loss": 0.6899,
      "step": 33699
    },
    {
      "epoch": 0.9710136575808218,
      "grad_norm": 0.7275829911231995,
      "learning_rate": 2.2018375452347307e-08,
      "loss": 0.6837,
      "step": 33700
    },
    {
      "epoch": 0.971042471042471,
      "grad_norm": 0.7306066751480103,
      "learning_rate": 2.1974655178920724e-08,
      "loss": 0.6852,
      "step": 33701
    },
    {
      "epoch": 0.9710712845041203,
      "grad_norm": 0.7258821725845337,
      "learning_rate": 2.1930978259118717e-08,
      "loss": 0.6668,
      "step": 33702
    },
    {
      "epoch": 0.9711000979657696,
      "grad_norm": 0.7399158477783203,
      "learning_rate": 2.18873446933221e-08,
      "loss": 0.6676,
      "step": 33703
    },
    {
      "epoch": 0.9711289114274189,
      "grad_norm": 0.7073233723640442,
      "learning_rate": 2.1843754481910006e-08,
      "loss": 0.6486,
      "step": 33704
    },
    {
      "epoch": 0.9711577248890682,
      "grad_norm": 0.7335332036018372,
      "learning_rate": 2.180020762526269e-08,
      "loss": 0.6889,
      "step": 33705
    },
    {
      "epoch": 0.9711865383507174,
      "grad_norm": 0.7370092868804932,
      "learning_rate": 2.1756704123758742e-08,
      "loss": 0.6775,
      "step": 33706
    },
    {
      "epoch": 0.9712153518123667,
      "grad_norm": 0.7231642007827759,
      "learning_rate": 2.1713243977778407e-08,
      "loss": 0.6833,
      "step": 33707
    },
    {
      "epoch": 0.971244165274016,
      "grad_norm": 0.7563712000846863,
      "learning_rate": 2.1669827187698057e-08,
      "loss": 0.6909,
      "step": 33708
    },
    {
      "epoch": 0.9712729787356653,
      "grad_norm": 0.7450374960899353,
      "learning_rate": 2.1626453753897936e-08,
      "loss": 0.7091,
      "step": 33709
    },
    {
      "epoch": 0.9713017921973146,
      "grad_norm": 0.7104480266571045,
      "learning_rate": 2.1583123676754415e-08,
      "loss": 0.6661,
      "step": 33710
    },
    {
      "epoch": 0.9713306056589639,
      "grad_norm": 0.721672773361206,
      "learning_rate": 2.1539836956645523e-08,
      "loss": 0.6607,
      "step": 33711
    },
    {
      "epoch": 0.9713594191206132,
      "grad_norm": 0.7137921452522278,
      "learning_rate": 2.1496593593947622e-08,
      "loss": 0.6884,
      "step": 33712
    },
    {
      "epoch": 0.9713882325822625,
      "grad_norm": 0.7323920726776123,
      "learning_rate": 2.1453393589038196e-08,
      "loss": 0.6933,
      "step": 33713
    },
    {
      "epoch": 0.9714170460439118,
      "grad_norm": 0.7307349443435669,
      "learning_rate": 2.141023694229305e-08,
      "loss": 0.6878,
      "step": 33714
    },
    {
      "epoch": 0.971445859505561,
      "grad_norm": 0.718382716178894,
      "learning_rate": 2.136712365408744e-08,
      "loss": 0.6817,
      "step": 33715
    },
    {
      "epoch": 0.9714746729672102,
      "grad_norm": 0.7272741198539734,
      "learning_rate": 2.1324053724797732e-08,
      "loss": 0.6866,
      "step": 33716
    },
    {
      "epoch": 0.9715034864288595,
      "grad_norm": 0.7666662931442261,
      "learning_rate": 2.128102715479863e-08,
      "loss": 0.6742,
      "step": 33717
    },
    {
      "epoch": 0.9715322998905088,
      "grad_norm": 0.7120427489280701,
      "learning_rate": 2.1238043944464826e-08,
      "loss": 0.6442,
      "step": 33718
    },
    {
      "epoch": 0.9715611133521581,
      "grad_norm": 0.7212215065956116,
      "learning_rate": 2.1195104094171028e-08,
      "loss": 0.6747,
      "step": 33719
    },
    {
      "epoch": 0.9715899268138074,
      "grad_norm": 0.7403568625450134,
      "learning_rate": 2.1152207604290265e-08,
      "loss": 0.6984,
      "step": 33720
    },
    {
      "epoch": 0.9716187402754567,
      "grad_norm": 0.7101213335990906,
      "learning_rate": 2.110935447519724e-08,
      "loss": 0.6772,
      "step": 33721
    },
    {
      "epoch": 0.971647553737106,
      "grad_norm": 0.7454568147659302,
      "learning_rate": 2.1066544707264437e-08,
      "loss": 0.6841,
      "step": 33722
    },
    {
      "epoch": 0.9716763671987553,
      "grad_norm": 0.7350903749465942,
      "learning_rate": 2.1023778300864884e-08,
      "loss": 0.6737,
      "step": 33723
    },
    {
      "epoch": 0.9717051806604046,
      "grad_norm": 0.7314348220825195,
      "learning_rate": 2.0981055256371064e-08,
      "loss": 0.6968,
      "step": 33724
    },
    {
      "epoch": 0.9717339941220539,
      "grad_norm": 0.7186235189437866,
      "learning_rate": 2.09383755741549e-08,
      "loss": 0.6961,
      "step": 33725
    },
    {
      "epoch": 0.9717628075837031,
      "grad_norm": 0.7313367128372192,
      "learning_rate": 2.0895739254588875e-08,
      "loss": 0.7092,
      "step": 33726
    },
    {
      "epoch": 0.9717916210453524,
      "grad_norm": 0.7268651723861694,
      "learning_rate": 2.0853146298043248e-08,
      "loss": 0.6846,
      "step": 33727
    },
    {
      "epoch": 0.9718204345070016,
      "grad_norm": 0.7319124341011047,
      "learning_rate": 2.0810596704889384e-08,
      "loss": 0.6962,
      "step": 33728
    },
    {
      "epoch": 0.9718492479686509,
      "grad_norm": 0.7476844191551208,
      "learning_rate": 2.076809047549755e-08,
      "loss": 0.7023,
      "step": 33729
    },
    {
      "epoch": 0.9718780614303002,
      "grad_norm": 0.7261422276496887,
      "learning_rate": 2.0725627610239106e-08,
      "loss": 0.6915,
      "step": 33730
    },
    {
      "epoch": 0.9719068748919495,
      "grad_norm": 0.7105291485786438,
      "learning_rate": 2.06832081094821e-08,
      "loss": 0.6821,
      "step": 33731
    },
    {
      "epoch": 0.9719356883535988,
      "grad_norm": 0.7296945452690125,
      "learning_rate": 2.0640831973597343e-08,
      "loss": 0.6726,
      "step": 33732
    },
    {
      "epoch": 0.9719645018152481,
      "grad_norm": 0.7335495352745056,
      "learning_rate": 2.0598499202953426e-08,
      "loss": 0.6817,
      "step": 33733
    },
    {
      "epoch": 0.9719933152768974,
      "grad_norm": 0.729090690612793,
      "learning_rate": 2.0556209797918946e-08,
      "loss": 0.691,
      "step": 33734
    },
    {
      "epoch": 0.9720221287385467,
      "grad_norm": 0.7160530686378479,
      "learning_rate": 2.0513963758861942e-08,
      "loss": 0.6758,
      "step": 33735
    },
    {
      "epoch": 0.9720509422001959,
      "grad_norm": 0.7698896527290344,
      "learning_rate": 2.047176108615101e-08,
      "loss": 0.6873,
      "step": 33736
    },
    {
      "epoch": 0.9720797556618452,
      "grad_norm": 0.7234950065612793,
      "learning_rate": 2.0429601780153075e-08,
      "loss": 0.6974,
      "step": 33737
    },
    {
      "epoch": 0.9721085691234945,
      "grad_norm": 0.705776035785675,
      "learning_rate": 2.038748584123562e-08,
      "loss": 0.6595,
      "step": 33738
    },
    {
      "epoch": 0.9721373825851438,
      "grad_norm": 0.7255802154541016,
      "learning_rate": 2.0345413269765578e-08,
      "loss": 0.6583,
      "step": 33739
    },
    {
      "epoch": 0.9721661960467931,
      "grad_norm": 0.7171810865402222,
      "learning_rate": 2.0303384066109323e-08,
      "loss": 0.701,
      "step": 33740
    },
    {
      "epoch": 0.9721950095084424,
      "grad_norm": 0.7051898837089539,
      "learning_rate": 2.0261398230632113e-08,
      "loss": 0.6744,
      "step": 33741
    },
    {
      "epoch": 0.9722238229700916,
      "grad_norm": 0.7045614719390869,
      "learning_rate": 2.0219455763700324e-08,
      "loss": 0.671,
      "step": 33742
    },
    {
      "epoch": 0.9722526364317409,
      "grad_norm": 0.7223173975944519,
      "learning_rate": 2.0177556665679777e-08,
      "loss": 0.6923,
      "step": 33743
    },
    {
      "epoch": 0.9722814498933902,
      "grad_norm": 0.7248306274414062,
      "learning_rate": 2.0135700936934066e-08,
      "loss": 0.6817,
      "step": 33744
    },
    {
      "epoch": 0.9723102633550395,
      "grad_norm": 0.7255518436431885,
      "learning_rate": 2.0093888577828458e-08,
      "loss": 0.6671,
      "step": 33745
    },
    {
      "epoch": 0.9723390768166887,
      "grad_norm": 0.7154199481010437,
      "learning_rate": 2.0052119588727105e-08,
      "loss": 0.6943,
      "step": 33746
    },
    {
      "epoch": 0.972367890278338,
      "grad_norm": 0.6974311470985413,
      "learning_rate": 2.00103939699936e-08,
      "loss": 0.6929,
      "step": 33747
    },
    {
      "epoch": 0.9723967037399873,
      "grad_norm": 0.7227771282196045,
      "learning_rate": 1.996871172199155e-08,
      "loss": 0.6896,
      "step": 33748
    },
    {
      "epoch": 0.9724255172016366,
      "grad_norm": 0.742169201374054,
      "learning_rate": 1.9927072845083438e-08,
      "loss": 0.6834,
      "step": 33749
    },
    {
      "epoch": 0.9724543306632859,
      "grad_norm": 0.7201616764068604,
      "learning_rate": 1.9885477339632864e-08,
      "loss": 0.685,
      "step": 33750
    },
    {
      "epoch": 0.9724831441249352,
      "grad_norm": 0.7252942323684692,
      "learning_rate": 1.9843925206001203e-08,
      "loss": 0.6692,
      "step": 33751
    },
    {
      "epoch": 0.9725119575865845,
      "grad_norm": 0.7382825613021851,
      "learning_rate": 1.9802416444550387e-08,
      "loss": 0.696,
      "step": 33752
    },
    {
      "epoch": 0.9725407710482338,
      "grad_norm": 0.7170879244804382,
      "learning_rate": 1.9760951055642354e-08,
      "loss": 0.6809,
      "step": 33753
    },
    {
      "epoch": 0.9725695845098831,
      "grad_norm": 0.7255776524543762,
      "learning_rate": 1.971952903963792e-08,
      "loss": 0.707,
      "step": 33754
    },
    {
      "epoch": 0.9725983979715322,
      "grad_norm": 0.7160409688949585,
      "learning_rate": 1.9678150396898464e-08,
      "loss": 0.6773,
      "step": 33755
    },
    {
      "epoch": 0.9726272114331815,
      "grad_norm": 0.7113226652145386,
      "learning_rate": 1.9636815127782593e-08,
      "loss": 0.6767,
      "step": 33756
    },
    {
      "epoch": 0.9726560248948308,
      "grad_norm": 0.7412599325180054,
      "learning_rate": 1.9595523232652792e-08,
      "loss": 0.6707,
      "step": 33757
    },
    {
      "epoch": 0.9726848383564801,
      "grad_norm": 0.7229412198066711,
      "learning_rate": 1.955427471186655e-08,
      "loss": 0.6777,
      "step": 33758
    },
    {
      "epoch": 0.9727136518181294,
      "grad_norm": 0.7214735150337219,
      "learning_rate": 1.9513069565784693e-08,
      "loss": 0.6746,
      "step": 33759
    },
    {
      "epoch": 0.9727424652797787,
      "grad_norm": 0.740103006362915,
      "learning_rate": 1.9471907794764712e-08,
      "loss": 0.7054,
      "step": 33760
    },
    {
      "epoch": 0.972771278741428,
      "grad_norm": 0.7250121831893921,
      "learning_rate": 1.9430789399166317e-08,
      "loss": 0.6652,
      "step": 33761
    },
    {
      "epoch": 0.9728000922030773,
      "grad_norm": 0.7065234184265137,
      "learning_rate": 1.9389714379346447e-08,
      "loss": 0.6704,
      "step": 33762
    },
    {
      "epoch": 0.9728289056647266,
      "grad_norm": 0.7323307394981384,
      "learning_rate": 1.9348682735663147e-08,
      "loss": 0.716,
      "step": 33763
    },
    {
      "epoch": 0.9728577191263759,
      "grad_norm": 0.7375823259353638,
      "learning_rate": 1.930769446847447e-08,
      "loss": 0.7018,
      "step": 33764
    },
    {
      "epoch": 0.9728865325880252,
      "grad_norm": 0.8730301260948181,
      "learning_rate": 1.926674957813679e-08,
      "loss": 0.689,
      "step": 33765
    },
    {
      "epoch": 0.9729153460496744,
      "grad_norm": 0.7239072918891907,
      "learning_rate": 1.9225848065007046e-08,
      "loss": 0.6959,
      "step": 33766
    },
    {
      "epoch": 0.9729441595113237,
      "grad_norm": 0.7305679321289062,
      "learning_rate": 1.9184989929441066e-08,
      "loss": 0.6552,
      "step": 33767
    },
    {
      "epoch": 0.972972972972973,
      "grad_norm": 0.719848096370697,
      "learning_rate": 1.9144175171794675e-08,
      "loss": 0.6433,
      "step": 33768
    },
    {
      "epoch": 0.9730017864346222,
      "grad_norm": 0.7100110054016113,
      "learning_rate": 1.9103403792423703e-08,
      "loss": 0.6789,
      "step": 33769
    },
    {
      "epoch": 0.9730305998962715,
      "grad_norm": 0.7308397889137268,
      "learning_rate": 1.906267579168286e-08,
      "loss": 0.6687,
      "step": 33770
    },
    {
      "epoch": 0.9730594133579208,
      "grad_norm": 0.7162421345710754,
      "learning_rate": 1.9021991169926867e-08,
      "loss": 0.6669,
      "step": 33771
    },
    {
      "epoch": 0.9730882268195701,
      "grad_norm": 0.7076525092124939,
      "learning_rate": 1.898134992751044e-08,
      "loss": 0.6836,
      "step": 33772
    },
    {
      "epoch": 0.9731170402812194,
      "grad_norm": 0.7303273677825928,
      "learning_rate": 1.8940752064787182e-08,
      "loss": 0.6895,
      "step": 33773
    },
    {
      "epoch": 0.9731458537428687,
      "grad_norm": 0.7029915452003479,
      "learning_rate": 1.89001975821107e-08,
      "loss": 0.6807,
      "step": 33774
    },
    {
      "epoch": 0.973174667204518,
      "grad_norm": 0.7302334308624268,
      "learning_rate": 1.8859686479834603e-08,
      "loss": 0.6619,
      "step": 33775
    },
    {
      "epoch": 0.9732034806661672,
      "grad_norm": 0.7152731418609619,
      "learning_rate": 1.881921875831083e-08,
      "loss": 0.6771,
      "step": 33776
    },
    {
      "epoch": 0.9732322941278165,
      "grad_norm": 0.7070074677467346,
      "learning_rate": 1.8778794417892432e-08,
      "loss": 0.6404,
      "step": 33777
    },
    {
      "epoch": 0.9732611075894658,
      "grad_norm": 0.7385187149047852,
      "learning_rate": 1.873841345893135e-08,
      "loss": 0.6942,
      "step": 33778
    },
    {
      "epoch": 0.9732899210511151,
      "grad_norm": 0.728495717048645,
      "learning_rate": 1.8698075881778964e-08,
      "loss": 0.6832,
      "step": 33779
    },
    {
      "epoch": 0.9733187345127644,
      "grad_norm": 0.6989607214927673,
      "learning_rate": 1.8657781686787225e-08,
      "loss": 0.6715,
      "step": 33780
    },
    {
      "epoch": 0.9733475479744137,
      "grad_norm": 0.7335667610168457,
      "learning_rate": 1.86175308743064e-08,
      "loss": 0.6863,
      "step": 33781
    },
    {
      "epoch": 0.9733763614360629,
      "grad_norm": 0.721379280090332,
      "learning_rate": 1.857732344468788e-08,
      "loss": 0.6732,
      "step": 33782
    },
    {
      "epoch": 0.9734051748977122,
      "grad_norm": 0.7250123023986816,
      "learning_rate": 1.8537159398280825e-08,
      "loss": 0.6906,
      "step": 33783
    },
    {
      "epoch": 0.9734339883593615,
      "grad_norm": 0.7246792912483215,
      "learning_rate": 1.8497038735435512e-08,
      "loss": 0.6648,
      "step": 33784
    },
    {
      "epoch": 0.9734628018210107,
      "grad_norm": 0.7265081405639648,
      "learning_rate": 1.845696145650111e-08,
      "loss": 0.6696,
      "step": 33785
    },
    {
      "epoch": 0.97349161528266,
      "grad_norm": 0.707960844039917,
      "learning_rate": 1.8416927561827337e-08,
      "loss": 0.6679,
      "step": 33786
    },
    {
      "epoch": 0.9735204287443093,
      "grad_norm": 0.7259548902511597,
      "learning_rate": 1.8376937051762246e-08,
      "loss": 0.6805,
      "step": 33787
    },
    {
      "epoch": 0.9735492422059586,
      "grad_norm": 0.7200501561164856,
      "learning_rate": 1.8336989926653893e-08,
      "loss": 0.6855,
      "step": 33788
    },
    {
      "epoch": 0.9735780556676079,
      "grad_norm": 0.7150112986564636,
      "learning_rate": 1.8297086186850888e-08,
      "loss": 0.6708,
      "step": 33789
    },
    {
      "epoch": 0.9736068691292572,
      "grad_norm": 0.7193000912666321,
      "learning_rate": 1.8257225832700175e-08,
      "loss": 0.6804,
      "step": 33790
    },
    {
      "epoch": 0.9736356825909065,
      "grad_norm": 0.7190248966217041,
      "learning_rate": 1.821740886454926e-08,
      "loss": 0.6747,
      "step": 33791
    },
    {
      "epoch": 0.9736644960525558,
      "grad_norm": 0.7074387073516846,
      "learning_rate": 1.817763528274452e-08,
      "loss": 0.6407,
      "step": 33792
    },
    {
      "epoch": 0.9736933095142051,
      "grad_norm": 0.7176186442375183,
      "learning_rate": 1.8137905087632913e-08,
      "loss": 0.6728,
      "step": 33793
    },
    {
      "epoch": 0.9737221229758544,
      "grad_norm": 0.7333939671516418,
      "learning_rate": 1.8098218279559708e-08,
      "loss": 0.6547,
      "step": 33794
    },
    {
      "epoch": 0.9737509364375037,
      "grad_norm": 0.7116740942001343,
      "learning_rate": 1.8058574858871304e-08,
      "loss": 0.676,
      "step": 33795
    },
    {
      "epoch": 0.9737797498991528,
      "grad_norm": 0.7157355546951294,
      "learning_rate": 1.8018974825912418e-08,
      "loss": 0.6752,
      "step": 33796
    },
    {
      "epoch": 0.9738085633608021,
      "grad_norm": 0.736598789691925,
      "learning_rate": 1.7979418181027775e-08,
      "loss": 0.7062,
      "step": 33797
    },
    {
      "epoch": 0.9738373768224514,
      "grad_norm": 0.7226254343986511,
      "learning_rate": 1.7939904924562657e-08,
      "loss": 0.6758,
      "step": 33798
    },
    {
      "epoch": 0.9738661902841007,
      "grad_norm": 0.709784209728241,
      "learning_rate": 1.7900435056860678e-08,
      "loss": 0.6855,
      "step": 33799
    },
    {
      "epoch": 0.97389500374575,
      "grad_norm": 0.7509883046150208,
      "learning_rate": 1.786100857826545e-08,
      "loss": 0.6929,
      "step": 33800
    },
    {
      "epoch": 0.9739238172073993,
      "grad_norm": 0.7175323963165283,
      "learning_rate": 1.7821625489120587e-08,
      "loss": 0.6811,
      "step": 33801
    },
    {
      "epoch": 0.9739526306690486,
      "grad_norm": 0.7433726191520691,
      "learning_rate": 1.778228578976915e-08,
      "loss": 0.6752,
      "step": 33802
    },
    {
      "epoch": 0.9739814441306979,
      "grad_norm": 0.7337684035301208,
      "learning_rate": 1.774298948055364e-08,
      "loss": 0.6754,
      "step": 33803
    },
    {
      "epoch": 0.9740102575923472,
      "grad_norm": 0.7406660318374634,
      "learning_rate": 1.770373656181601e-08,
      "loss": 0.6727,
      "step": 33804
    },
    {
      "epoch": 0.9740390710539965,
      "grad_norm": 0.7123318910598755,
      "learning_rate": 1.7664527033898203e-08,
      "loss": 0.6874,
      "step": 33805
    },
    {
      "epoch": 0.9740678845156457,
      "grad_norm": 0.7367600798606873,
      "learning_rate": 1.7625360897142728e-08,
      "loss": 0.6918,
      "step": 33806
    },
    {
      "epoch": 0.974096697977295,
      "grad_norm": 0.7231929302215576,
      "learning_rate": 1.7586238151888756e-08,
      "loss": 0.6768,
      "step": 33807
    },
    {
      "epoch": 0.9741255114389443,
      "grad_norm": 0.6928572058677673,
      "learning_rate": 1.754715879847879e-08,
      "loss": 0.6493,
      "step": 33808
    },
    {
      "epoch": 0.9741543249005935,
      "grad_norm": 0.70512855052948,
      "learning_rate": 1.7508122837252006e-08,
      "loss": 0.6602,
      "step": 33809
    },
    {
      "epoch": 0.9741831383622428,
      "grad_norm": 0.7186609506607056,
      "learning_rate": 1.746913026854924e-08,
      "loss": 0.6615,
      "step": 33810
    },
    {
      "epoch": 0.9742119518238921,
      "grad_norm": 0.7353992462158203,
      "learning_rate": 1.743018109270911e-08,
      "loss": 0.6898,
      "step": 33811
    },
    {
      "epoch": 0.9742407652855414,
      "grad_norm": 0.712670624256134,
      "learning_rate": 1.7391275310071897e-08,
      "loss": 0.6723,
      "step": 33812
    },
    {
      "epoch": 0.9742695787471907,
      "grad_norm": 0.7381763458251953,
      "learning_rate": 1.735241292097567e-08,
      "loss": 0.6889,
      "step": 33813
    },
    {
      "epoch": 0.97429839220884,
      "grad_norm": 0.7174704074859619,
      "learning_rate": 1.7313593925759043e-08,
      "loss": 0.678,
      "step": 33814
    },
    {
      "epoch": 0.9743272056704892,
      "grad_norm": 0.7521979212760925,
      "learning_rate": 1.7274818324760077e-08,
      "loss": 0.7054,
      "step": 33815
    },
    {
      "epoch": 0.9743560191321385,
      "grad_norm": 0.7472376227378845,
      "learning_rate": 1.7236086118316286e-08,
      "loss": 0.6854,
      "step": 33816
    },
    {
      "epoch": 0.9743848325937878,
      "grad_norm": 0.7112374305725098,
      "learning_rate": 1.719739730676573e-08,
      "loss": 0.6686,
      "step": 33817
    },
    {
      "epoch": 0.9744136460554371,
      "grad_norm": 0.7251826524734497,
      "learning_rate": 1.7158751890444804e-08,
      "loss": 0.6852,
      "step": 33818
    },
    {
      "epoch": 0.9744424595170864,
      "grad_norm": 0.7282723188400269,
      "learning_rate": 1.7120149869689906e-08,
      "loss": 0.6715,
      "step": 33819
    },
    {
      "epoch": 0.9744712729787357,
      "grad_norm": 0.7046553492546082,
      "learning_rate": 1.7081591244837993e-08,
      "loss": 0.6495,
      "step": 33820
    },
    {
      "epoch": 0.974500086440385,
      "grad_norm": 0.704649806022644,
      "learning_rate": 1.704307601622379e-08,
      "loss": 0.6846,
      "step": 33821
    },
    {
      "epoch": 0.9745288999020343,
      "grad_norm": 0.7154420614242554,
      "learning_rate": 1.7004604184183147e-08,
      "loss": 0.6862,
      "step": 33822
    },
    {
      "epoch": 0.9745577133636835,
      "grad_norm": 0.7429221868515015,
      "learning_rate": 1.696617574905135e-08,
      "loss": 0.6952,
      "step": 33823
    },
    {
      "epoch": 0.9745865268253328,
      "grad_norm": 0.7737873196601868,
      "learning_rate": 1.692779071116313e-08,
      "loss": 0.6795,
      "step": 33824
    },
    {
      "epoch": 0.974615340286982,
      "grad_norm": 0.7115550637245178,
      "learning_rate": 1.688944907085266e-08,
      "loss": 0.6798,
      "step": 33825
    },
    {
      "epoch": 0.9746441537486313,
      "grad_norm": 0.7286223769187927,
      "learning_rate": 1.6851150828453565e-08,
      "loss": 0.6791,
      "step": 33826
    },
    {
      "epoch": 0.9746729672102806,
      "grad_norm": 0.737419605255127,
      "learning_rate": 1.6812895984300026e-08,
      "loss": 0.6876,
      "step": 33827
    },
    {
      "epoch": 0.9747017806719299,
      "grad_norm": 0.7124108076095581,
      "learning_rate": 1.6774684538723994e-08,
      "loss": 0.6859,
      "step": 33828
    },
    {
      "epoch": 0.9747305941335792,
      "grad_norm": 0.7177229523658752,
      "learning_rate": 1.6736516492060207e-08,
      "loss": 0.6744,
      "step": 33829
    },
    {
      "epoch": 0.9747594075952285,
      "grad_norm": 0.7157834768295288,
      "learning_rate": 1.669839184463895e-08,
      "loss": 0.6678,
      "step": 33830
    },
    {
      "epoch": 0.9747882210568778,
      "grad_norm": 0.6982089281082153,
      "learning_rate": 1.666031059679385e-08,
      "loss": 0.691,
      "step": 33831
    },
    {
      "epoch": 0.9748170345185271,
      "grad_norm": 0.7048653960227966,
      "learning_rate": 1.6622272748855194e-08,
      "loss": 0.6895,
      "step": 33832
    },
    {
      "epoch": 0.9748458479801764,
      "grad_norm": 0.7192515134811401,
      "learning_rate": 1.6584278301155498e-08,
      "loss": 0.6948,
      "step": 33833
    },
    {
      "epoch": 0.9748746614418257,
      "grad_norm": 0.7161264419555664,
      "learning_rate": 1.654632725402505e-08,
      "loss": 0.6937,
      "step": 33834
    },
    {
      "epoch": 0.974903474903475,
      "grad_norm": 0.7128887176513672,
      "learning_rate": 1.65084196077947e-08,
      "loss": 0.665,
      "step": 33835
    },
    {
      "epoch": 0.9749322883651241,
      "grad_norm": 0.7240585684776306,
      "learning_rate": 1.6470555362794183e-08,
      "loss": 0.6665,
      "step": 33836
    },
    {
      "epoch": 0.9749611018267734,
      "grad_norm": 0.7275329232215881,
      "learning_rate": 1.6432734519353232e-08,
      "loss": 0.6925,
      "step": 33837
    },
    {
      "epoch": 0.9749899152884227,
      "grad_norm": 0.7297858595848083,
      "learning_rate": 1.6394957077801586e-08,
      "loss": 0.6808,
      "step": 33838
    },
    {
      "epoch": 0.975018728750072,
      "grad_norm": 0.7217445969581604,
      "learning_rate": 1.635722303846843e-08,
      "loss": 0.6953,
      "step": 33839
    },
    {
      "epoch": 0.9750475422117213,
      "grad_norm": 0.7461692094802856,
      "learning_rate": 1.631953240168127e-08,
      "loss": 0.6731,
      "step": 33840
    },
    {
      "epoch": 0.9750763556733706,
      "grad_norm": 0.720429539680481,
      "learning_rate": 1.628188516776985e-08,
      "loss": 0.6891,
      "step": 33841
    },
    {
      "epoch": 0.9751051691350199,
      "grad_norm": 0.7163847088813782,
      "learning_rate": 1.6244281337060575e-08,
      "loss": 0.6614,
      "step": 33842
    },
    {
      "epoch": 0.9751339825966692,
      "grad_norm": 0.7205718159675598,
      "learning_rate": 1.6206720909882068e-08,
      "loss": 0.6686,
      "step": 33843
    },
    {
      "epoch": 0.9751627960583185,
      "grad_norm": 0.746456503868103,
      "learning_rate": 1.616920388656129e-08,
      "loss": 0.669,
      "step": 33844
    },
    {
      "epoch": 0.9751916095199678,
      "grad_norm": 0.8526375889778137,
      "learning_rate": 1.613173026742465e-08,
      "loss": 0.7253,
      "step": 33845
    },
    {
      "epoch": 0.975220422981617,
      "grad_norm": 0.7235451340675354,
      "learning_rate": 1.609430005279855e-08,
      "loss": 0.6608,
      "step": 33846
    },
    {
      "epoch": 0.9752492364432663,
      "grad_norm": 0.725287914276123,
      "learning_rate": 1.6056913243008844e-08,
      "loss": 0.6876,
      "step": 33847
    },
    {
      "epoch": 0.9752780499049156,
      "grad_norm": 0.7499200701713562,
      "learning_rate": 1.6019569838381378e-08,
      "loss": 0.6939,
      "step": 33848
    },
    {
      "epoch": 0.9753068633665649,
      "grad_norm": 0.7304551601409912,
      "learning_rate": 1.598226983924145e-08,
      "loss": 0.7043,
      "step": 33849
    },
    {
      "epoch": 0.9753356768282141,
      "grad_norm": 0.7321456074714661,
      "learning_rate": 1.5945013245913797e-08,
      "loss": 0.6737,
      "step": 33850
    },
    {
      "epoch": 0.9753644902898634,
      "grad_norm": 0.7282432317733765,
      "learning_rate": 1.5907800058722612e-08,
      "loss": 0.6826,
      "step": 33851
    },
    {
      "epoch": 0.9753933037515127,
      "grad_norm": 0.7516763806343079,
      "learning_rate": 1.5870630277992072e-08,
      "loss": 0.6764,
      "step": 33852
    },
    {
      "epoch": 0.975422117213162,
      "grad_norm": 0.7308400273323059,
      "learning_rate": 1.5833503904046366e-08,
      "loss": 0.6723,
      "step": 33853
    },
    {
      "epoch": 0.9754509306748113,
      "grad_norm": 0.7284247875213623,
      "learning_rate": 1.5796420937208567e-08,
      "loss": 0.7079,
      "step": 33854
    },
    {
      "epoch": 0.9754797441364605,
      "grad_norm": 0.7441567778587341,
      "learning_rate": 1.5759381377801197e-08,
      "loss": 0.7025,
      "step": 33855
    },
    {
      "epoch": 0.9755085575981098,
      "grad_norm": 0.7225160002708435,
      "learning_rate": 1.572238522614733e-08,
      "loss": 0.6713,
      "step": 33856
    },
    {
      "epoch": 0.9755373710597591,
      "grad_norm": 0.7454972267150879,
      "learning_rate": 1.568543248256893e-08,
      "loss": 0.6916,
      "step": 33857
    },
    {
      "epoch": 0.9755661845214084,
      "grad_norm": 0.7346783876419067,
      "learning_rate": 1.5648523147388517e-08,
      "loss": 0.674,
      "step": 33858
    },
    {
      "epoch": 0.9755949979830577,
      "grad_norm": 0.7089521884918213,
      "learning_rate": 1.561165722092639e-08,
      "loss": 0.6756,
      "step": 33859
    },
    {
      "epoch": 0.975623811444707,
      "grad_norm": 0.7211604118347168,
      "learning_rate": 1.557483470350396e-08,
      "loss": 0.6835,
      "step": 33860
    },
    {
      "epoch": 0.9756526249063563,
      "grad_norm": 0.7122009992599487,
      "learning_rate": 1.5538055595442636e-08,
      "loss": 0.6871,
      "step": 33861
    },
    {
      "epoch": 0.9756814383680056,
      "grad_norm": 0.7342798709869385,
      "learning_rate": 1.550131989706216e-08,
      "loss": 0.6762,
      "step": 33862
    },
    {
      "epoch": 0.9757102518296548,
      "grad_norm": 0.703176736831665,
      "learning_rate": 1.546462760868228e-08,
      "loss": 0.6866,
      "step": 33863
    },
    {
      "epoch": 0.975739065291304,
      "grad_norm": 0.764380156993866,
      "learning_rate": 1.5427978730622738e-08,
      "loss": 0.6853,
      "step": 33864
    },
    {
      "epoch": 0.9757678787529533,
      "grad_norm": 0.7572965025901794,
      "learning_rate": 1.5391373263202725e-08,
      "loss": 0.6869,
      "step": 33865
    },
    {
      "epoch": 0.9757966922146026,
      "grad_norm": 0.7223266959190369,
      "learning_rate": 1.5354811206741426e-08,
      "loss": 0.6773,
      "step": 33866
    },
    {
      "epoch": 0.9758255056762519,
      "grad_norm": 0.7103146910667419,
      "learning_rate": 1.531829256155637e-08,
      "loss": 0.6753,
      "step": 33867
    },
    {
      "epoch": 0.9758543191379012,
      "grad_norm": 0.6986621022224426,
      "learning_rate": 1.5281817327966742e-08,
      "loss": 0.6652,
      "step": 33868
    },
    {
      "epoch": 0.9758831325995505,
      "grad_norm": 0.7251657843589783,
      "learning_rate": 1.524538550628951e-08,
      "loss": 0.7096,
      "step": 33869
    },
    {
      "epoch": 0.9759119460611998,
      "grad_norm": 0.7249884605407715,
      "learning_rate": 1.520899709684165e-08,
      "loss": 0.7076,
      "step": 33870
    },
    {
      "epoch": 0.9759407595228491,
      "grad_norm": 0.7489314675331116,
      "learning_rate": 1.5172652099940677e-08,
      "loss": 0.6752,
      "step": 33871
    },
    {
      "epoch": 0.9759695729844984,
      "grad_norm": 0.7198585271835327,
      "learning_rate": 1.513635051590301e-08,
      "loss": 0.6859,
      "step": 33872
    },
    {
      "epoch": 0.9759983864461477,
      "grad_norm": 0.7235687971115112,
      "learning_rate": 1.5100092345044505e-08,
      "loss": 0.6644,
      "step": 33873
    },
    {
      "epoch": 0.976027199907797,
      "grad_norm": 0.6979045271873474,
      "learning_rate": 1.506387758768102e-08,
      "loss": 0.6406,
      "step": 33874
    },
    {
      "epoch": 0.9760560133694463,
      "grad_norm": 0.7210391759872437,
      "learning_rate": 1.5027706244128414e-08,
      "loss": 0.636,
      "step": 33875
    },
    {
      "epoch": 0.9760848268310955,
      "grad_norm": 0.7294511795043945,
      "learning_rate": 1.4991578314700885e-08,
      "loss": 0.6911,
      "step": 33876
    },
    {
      "epoch": 0.9761136402927447,
      "grad_norm": 0.7219696640968323,
      "learning_rate": 1.495549379971373e-08,
      "loss": 0.6798,
      "step": 33877
    },
    {
      "epoch": 0.976142453754394,
      "grad_norm": 0.697284996509552,
      "learning_rate": 1.491945269948114e-08,
      "loss": 0.6712,
      "step": 33878
    },
    {
      "epoch": 0.9761712672160433,
      "grad_norm": 0.7245137095451355,
      "learning_rate": 1.488345501431676e-08,
      "loss": 0.6876,
      "step": 33879
    },
    {
      "epoch": 0.9762000806776926,
      "grad_norm": 0.7237629294395447,
      "learning_rate": 1.4847500744534226e-08,
      "loss": 0.6761,
      "step": 33880
    },
    {
      "epoch": 0.9762288941393419,
      "grad_norm": 0.7257941961288452,
      "learning_rate": 1.4811589890446066e-08,
      "loss": 0.676,
      "step": 33881
    },
    {
      "epoch": 0.9762577076009912,
      "grad_norm": 0.7025827765464783,
      "learning_rate": 1.477572245236647e-08,
      "loss": 0.6619,
      "step": 33882
    },
    {
      "epoch": 0.9762865210626405,
      "grad_norm": 0.7141879200935364,
      "learning_rate": 1.4739898430606304e-08,
      "loss": 0.6818,
      "step": 33883
    },
    {
      "epoch": 0.9763153345242898,
      "grad_norm": 0.7204766273498535,
      "learning_rate": 1.4704117825478647e-08,
      "loss": 0.6762,
      "step": 33884
    },
    {
      "epoch": 0.976344147985939,
      "grad_norm": 0.7262381911277771,
      "learning_rate": 1.4668380637294366e-08,
      "loss": 0.7015,
      "step": 33885
    },
    {
      "epoch": 0.9763729614475883,
      "grad_norm": 0.6978582739830017,
      "learning_rate": 1.463268686636543e-08,
      "loss": 0.6568,
      "step": 33886
    },
    {
      "epoch": 0.9764017749092376,
      "grad_norm": 0.734586775302887,
      "learning_rate": 1.4597036513001595e-08,
      "loss": 0.6747,
      "step": 33887
    },
    {
      "epoch": 0.9764305883708869,
      "grad_norm": 0.7437642812728882,
      "learning_rate": 1.4561429577514275e-08,
      "loss": 0.6955,
      "step": 33888
    },
    {
      "epoch": 0.9764594018325362,
      "grad_norm": 0.7136200666427612,
      "learning_rate": 1.4525866060213222e-08,
      "loss": 0.6928,
      "step": 33889
    },
    {
      "epoch": 0.9764882152941854,
      "grad_norm": 0.722253680229187,
      "learning_rate": 1.4490345961408748e-08,
      "loss": 0.6674,
      "step": 33890
    },
    {
      "epoch": 0.9765170287558347,
      "grad_norm": 0.7253884077072144,
      "learning_rate": 1.4454869281408935e-08,
      "loss": 0.671,
      "step": 33891
    },
    {
      "epoch": 0.976545842217484,
      "grad_norm": 0.702911376953125,
      "learning_rate": 1.4419436020524091e-08,
      "loss": 0.676,
      "step": 33892
    },
    {
      "epoch": 0.9765746556791333,
      "grad_norm": 0.7350261211395264,
      "learning_rate": 1.438404617906175e-08,
      "loss": 0.6611,
      "step": 33893
    },
    {
      "epoch": 0.9766034691407826,
      "grad_norm": 0.6961864233016968,
      "learning_rate": 1.4348699757330553e-08,
      "loss": 0.6608,
      "step": 33894
    },
    {
      "epoch": 0.9766322826024318,
      "grad_norm": 0.7250254154205322,
      "learning_rate": 1.4313396755638586e-08,
      "loss": 0.6927,
      "step": 33895
    },
    {
      "epoch": 0.9766610960640811,
      "grad_norm": 0.7153885364532471,
      "learning_rate": 1.4278137174293383e-08,
      "loss": 0.6736,
      "step": 33896
    },
    {
      "epoch": 0.9766899095257304,
      "grad_norm": 0.7554492354393005,
      "learning_rate": 1.4242921013600808e-08,
      "loss": 0.695,
      "step": 33897
    },
    {
      "epoch": 0.9767187229873797,
      "grad_norm": 0.7087939977645874,
      "learning_rate": 1.4207748273868949e-08,
      "loss": 0.6703,
      "step": 33898
    },
    {
      "epoch": 0.976747536449029,
      "grad_norm": 0.7335997223854065,
      "learning_rate": 1.4172618955404226e-08,
      "loss": 0.6984,
      "step": 33899
    },
    {
      "epoch": 0.9767763499106783,
      "grad_norm": 0.7279027104377747,
      "learning_rate": 1.4137533058511399e-08,
      "loss": 0.6762,
      "step": 33900
    },
    {
      "epoch": 0.9768051633723276,
      "grad_norm": 0.7387030720710754,
      "learning_rate": 1.4102490583496331e-08,
      "loss": 0.6707,
      "step": 33901
    },
    {
      "epoch": 0.9768339768339769,
      "grad_norm": 0.7347776889801025,
      "learning_rate": 1.406749153066489e-08,
      "loss": 0.6937,
      "step": 33902
    },
    {
      "epoch": 0.9768627902956262,
      "grad_norm": 0.7350603342056274,
      "learning_rate": 1.4032535900321275e-08,
      "loss": 0.6743,
      "step": 33903
    },
    {
      "epoch": 0.9768916037572754,
      "grad_norm": 0.7241284847259521,
      "learning_rate": 1.3997623692770246e-08,
      "loss": 0.7099,
      "step": 33904
    },
    {
      "epoch": 0.9769204172189246,
      "grad_norm": 0.7137587070465088,
      "learning_rate": 1.3962754908316001e-08,
      "loss": 0.691,
      "step": 33905
    },
    {
      "epoch": 0.9769492306805739,
      "grad_norm": 0.7171162962913513,
      "learning_rate": 1.3927929547261632e-08,
      "loss": 0.675,
      "step": 33906
    },
    {
      "epoch": 0.9769780441422232,
      "grad_norm": 0.7212259769439697,
      "learning_rate": 1.3893147609910785e-08,
      "loss": 0.6661,
      "step": 33907
    },
    {
      "epoch": 0.9770068576038725,
      "grad_norm": 0.7238363027572632,
      "learning_rate": 1.3858409096566549e-08,
      "loss": 0.7005,
      "step": 33908
    },
    {
      "epoch": 0.9770356710655218,
      "grad_norm": 0.7081483602523804,
      "learning_rate": 1.3823714007530909e-08,
      "loss": 0.657,
      "step": 33909
    },
    {
      "epoch": 0.9770644845271711,
      "grad_norm": 0.7222554683685303,
      "learning_rate": 1.3789062343106951e-08,
      "loss": 0.6726,
      "step": 33910
    },
    {
      "epoch": 0.9770932979888204,
      "grad_norm": 0.7385851740837097,
      "learning_rate": 1.3754454103595548e-08,
      "loss": 0.6838,
      "step": 33911
    },
    {
      "epoch": 0.9771221114504697,
      "grad_norm": 0.7227588295936584,
      "learning_rate": 1.3719889289298127e-08,
      "loss": 0.6764,
      "step": 33912
    },
    {
      "epoch": 0.977150924912119,
      "grad_norm": 0.7145118713378906,
      "learning_rate": 1.368536790051611e-08,
      "loss": 0.6634,
      "step": 33913
    },
    {
      "epoch": 0.9771797383737683,
      "grad_norm": 0.7081930637359619,
      "learning_rate": 1.365088993755037e-08,
      "loss": 0.679,
      "step": 33914
    },
    {
      "epoch": 0.9772085518354175,
      "grad_norm": 0.7318728566169739,
      "learning_rate": 1.3616455400700667e-08,
      "loss": 0.6902,
      "step": 33915
    },
    {
      "epoch": 0.9772373652970668,
      "grad_norm": 0.7222868800163269,
      "learning_rate": 1.358206429026676e-08,
      "loss": 0.6813,
      "step": 33916
    },
    {
      "epoch": 0.977266178758716,
      "grad_norm": 0.741873562335968,
      "learning_rate": 1.3547716606548967e-08,
      "loss": 0.6855,
      "step": 33917
    },
    {
      "epoch": 0.9772949922203653,
      "grad_norm": 0.7306778430938721,
      "learning_rate": 1.3513412349845934e-08,
      "loss": 0.6653,
      "step": 33918
    },
    {
      "epoch": 0.9773238056820146,
      "grad_norm": 0.7394886016845703,
      "learning_rate": 1.3479151520455758e-08,
      "loss": 0.6754,
      "step": 33919
    },
    {
      "epoch": 0.9773526191436639,
      "grad_norm": 0.7055610418319702,
      "learning_rate": 1.34449341186782e-08,
      "loss": 0.6914,
      "step": 33920
    },
    {
      "epoch": 0.9773814326053132,
      "grad_norm": 0.7260553240776062,
      "learning_rate": 1.3410760144810241e-08,
      "loss": 0.6942,
      "step": 33921
    },
    {
      "epoch": 0.9774102460669625,
      "grad_norm": 0.7227469086647034,
      "learning_rate": 1.3376629599149982e-08,
      "loss": 0.681,
      "step": 33922
    },
    {
      "epoch": 0.9774390595286118,
      "grad_norm": 0.723449170589447,
      "learning_rate": 1.3342542481994403e-08,
      "loss": 0.6795,
      "step": 33923
    },
    {
      "epoch": 0.977467872990261,
      "grad_norm": 0.7291659116744995,
      "learning_rate": 1.330849879364049e-08,
      "loss": 0.6811,
      "step": 33924
    },
    {
      "epoch": 0.9774966864519103,
      "grad_norm": 0.7201782464981079,
      "learning_rate": 1.3274498534384117e-08,
      "loss": 0.7016,
      "step": 33925
    },
    {
      "epoch": 0.9775254999135596,
      "grad_norm": 0.7439303994178772,
      "learning_rate": 1.3240541704522824e-08,
      "loss": 0.717,
      "step": 33926
    },
    {
      "epoch": 0.9775543133752089,
      "grad_norm": 0.7148926258087158,
      "learning_rate": 1.3206628304350822e-08,
      "loss": 0.688,
      "step": 33927
    },
    {
      "epoch": 0.9775831268368582,
      "grad_norm": 0.7164456844329834,
      "learning_rate": 1.3172758334164537e-08,
      "loss": 0.6887,
      "step": 33928
    },
    {
      "epoch": 0.9776119402985075,
      "grad_norm": 0.7440046072006226,
      "learning_rate": 1.3138931794258181e-08,
      "loss": 0.6672,
      "step": 33929
    },
    {
      "epoch": 0.9776407537601568,
      "grad_norm": 0.7150552868843079,
      "learning_rate": 1.3105148684927072e-08,
      "loss": 0.6725,
      "step": 33930
    },
    {
      "epoch": 0.977669567221806,
      "grad_norm": 0.7588425874710083,
      "learning_rate": 1.3071409006464863e-08,
      "loss": 0.7064,
      "step": 33931
    },
    {
      "epoch": 0.9776983806834553,
      "grad_norm": 0.7217293977737427,
      "learning_rate": 1.3037712759165766e-08,
      "loss": 0.6836,
      "step": 33932
    },
    {
      "epoch": 0.9777271941451046,
      "grad_norm": 0.7411016225814819,
      "learning_rate": 1.3004059943322878e-08,
      "loss": 0.6979,
      "step": 33933
    },
    {
      "epoch": 0.9777560076067539,
      "grad_norm": 0.7111285328865051,
      "learning_rate": 1.2970450559229298e-08,
      "loss": 0.6737,
      "step": 33934
    },
    {
      "epoch": 0.9777848210684031,
      "grad_norm": 0.7183176279067993,
      "learning_rate": 1.2936884607178124e-08,
      "loss": 0.6838,
      "step": 33935
    },
    {
      "epoch": 0.9778136345300524,
      "grad_norm": 0.7109542489051819,
      "learning_rate": 1.2903362087461346e-08,
      "loss": 0.6871,
      "step": 33936
    },
    {
      "epoch": 0.9778424479917017,
      "grad_norm": 0.7425012588500977,
      "learning_rate": 1.2869883000370954e-08,
      "loss": 0.7193,
      "step": 33937
    },
    {
      "epoch": 0.977871261453351,
      "grad_norm": 0.7101740837097168,
      "learning_rate": 1.2836447346198932e-08,
      "loss": 0.669,
      "step": 33938
    },
    {
      "epoch": 0.9779000749150003,
      "grad_norm": 0.7239184975624084,
      "learning_rate": 1.2803055125235608e-08,
      "loss": 0.7223,
      "step": 33939
    },
    {
      "epoch": 0.9779288883766496,
      "grad_norm": 0.7223260998725891,
      "learning_rate": 1.2769706337772414e-08,
      "loss": 0.6811,
      "step": 33940
    },
    {
      "epoch": 0.9779577018382989,
      "grad_norm": 0.724592387676239,
      "learning_rate": 1.2736400984099672e-08,
      "loss": 0.676,
      "step": 33941
    },
    {
      "epoch": 0.9779865152999482,
      "grad_norm": 0.7302352786064148,
      "learning_rate": 1.2703139064507707e-08,
      "loss": 0.6793,
      "step": 33942
    },
    {
      "epoch": 0.9780153287615975,
      "grad_norm": 0.7684609293937683,
      "learning_rate": 1.2669920579285733e-08,
      "loss": 0.6806,
      "step": 33943
    },
    {
      "epoch": 0.9780441422232466,
      "grad_norm": 0.7265501618385315,
      "learning_rate": 1.2636745528723515e-08,
      "loss": 0.6807,
      "step": 33944
    },
    {
      "epoch": 0.9780729556848959,
      "grad_norm": 0.7354202270507812,
      "learning_rate": 1.2603613913109713e-08,
      "loss": 0.6808,
      "step": 33945
    },
    {
      "epoch": 0.9781017691465452,
      "grad_norm": 0.7284291982650757,
      "learning_rate": 1.257052573273243e-08,
      "loss": 0.6752,
      "step": 33946
    },
    {
      "epoch": 0.9781305826081945,
      "grad_norm": 0.7120233774185181,
      "learning_rate": 1.2537480987880325e-08,
      "loss": 0.6734,
      "step": 33947
    },
    {
      "epoch": 0.9781593960698438,
      "grad_norm": 0.7376874089241028,
      "learning_rate": 1.2504479678840941e-08,
      "loss": 0.6852,
      "step": 33948
    },
    {
      "epoch": 0.9781882095314931,
      "grad_norm": 0.7164308428764343,
      "learning_rate": 1.2471521805902388e-08,
      "loss": 0.6746,
      "step": 33949
    },
    {
      "epoch": 0.9782170229931424,
      "grad_norm": 0.7012913823127747,
      "learning_rate": 1.2438607369351097e-08,
      "loss": 0.6701,
      "step": 33950
    },
    {
      "epoch": 0.9782458364547917,
      "grad_norm": 0.7234089970588684,
      "learning_rate": 1.2405736369473509e-08,
      "loss": 0.6849,
      "step": 33951
    },
    {
      "epoch": 0.978274649916441,
      "grad_norm": 0.7030882239341736,
      "learning_rate": 1.2372908806556616e-08,
      "loss": 0.696,
      "step": 33952
    },
    {
      "epoch": 0.9783034633780903,
      "grad_norm": 0.7151676416397095,
      "learning_rate": 1.2340124680885745e-08,
      "loss": 0.6825,
      "step": 33953
    },
    {
      "epoch": 0.9783322768397396,
      "grad_norm": 0.7287659645080566,
      "learning_rate": 1.2307383992746224e-08,
      "loss": 0.6789,
      "step": 33954
    },
    {
      "epoch": 0.9783610903013888,
      "grad_norm": 0.7046820521354675,
      "learning_rate": 1.2274686742423936e-08,
      "loss": 0.6711,
      "step": 33955
    },
    {
      "epoch": 0.9783899037630381,
      "grad_norm": 0.7245670557022095,
      "learning_rate": 1.2242032930202542e-08,
      "loss": 0.6763,
      "step": 33956
    },
    {
      "epoch": 0.9784187172246874,
      "grad_norm": 0.7145373821258545,
      "learning_rate": 1.2209422556367923e-08,
      "loss": 0.6716,
      "step": 33957
    },
    {
      "epoch": 0.9784475306863366,
      "grad_norm": 0.7037274837493896,
      "learning_rate": 1.2176855621202632e-08,
      "loss": 0.6526,
      "step": 33958
    },
    {
      "epoch": 0.9784763441479859,
      "grad_norm": 0.7247899174690247,
      "learning_rate": 1.2144332124991442e-08,
      "loss": 0.6989,
      "step": 33959
    },
    {
      "epoch": 0.9785051576096352,
      "grad_norm": 0.7354211211204529,
      "learning_rate": 1.2111852068016905e-08,
      "loss": 0.6725,
      "step": 33960
    },
    {
      "epoch": 0.9785339710712845,
      "grad_norm": 0.700685977935791,
      "learning_rate": 1.2079415450562127e-08,
      "loss": 0.6524,
      "step": 33961
    },
    {
      "epoch": 0.9785627845329338,
      "grad_norm": 0.7293309569358826,
      "learning_rate": 1.2047022272909103e-08,
      "loss": 0.697,
      "step": 33962
    },
    {
      "epoch": 0.9785915979945831,
      "grad_norm": 0.7075725793838501,
      "learning_rate": 1.2014672535340943e-08,
      "loss": 0.6625,
      "step": 33963
    },
    {
      "epoch": 0.9786204114562324,
      "grad_norm": 3.176854133605957,
      "learning_rate": 1.1982366238138532e-08,
      "loss": 0.6852,
      "step": 33964
    },
    {
      "epoch": 0.9786492249178816,
      "grad_norm": 0.7084444165229797,
      "learning_rate": 1.1950103381583867e-08,
      "loss": 0.6632,
      "step": 33965
    },
    {
      "epoch": 0.9786780383795309,
      "grad_norm": 0.7240648865699768,
      "learning_rate": 1.1917883965957278e-08,
      "loss": 0.6857,
      "step": 33966
    },
    {
      "epoch": 0.9787068518411802,
      "grad_norm": 0.7090832591056824,
      "learning_rate": 1.1885707991539652e-08,
      "loss": 0.6813,
      "step": 33967
    },
    {
      "epoch": 0.9787356653028295,
      "grad_norm": 0.7163602113723755,
      "learning_rate": 1.1853575458611322e-08,
      "loss": 0.6942,
      "step": 33968
    },
    {
      "epoch": 0.9787644787644788,
      "grad_norm": 0.7242376804351807,
      "learning_rate": 1.1821486367452062e-08,
      "loss": 0.7029,
      "step": 33969
    },
    {
      "epoch": 0.9787932922261281,
      "grad_norm": 0.7293387651443481,
      "learning_rate": 1.1789440718341094e-08,
      "loss": 0.6874,
      "step": 33970
    },
    {
      "epoch": 0.9788221056877773,
      "grad_norm": 0.728902280330658,
      "learning_rate": 1.175743851155764e-08,
      "loss": 0.6826,
      "step": 33971
    },
    {
      "epoch": 0.9788509191494266,
      "grad_norm": 0.7181075811386108,
      "learning_rate": 1.1725479747380919e-08,
      "loss": 0.6526,
      "step": 33972
    },
    {
      "epoch": 0.9788797326110759,
      "grad_norm": 0.7079761028289795,
      "learning_rate": 1.1693564426088488e-08,
      "loss": 0.6605,
      "step": 33973
    },
    {
      "epoch": 0.9789085460727251,
      "grad_norm": 0.7052693367004395,
      "learning_rate": 1.1661692547958458e-08,
      "loss": 0.6473,
      "step": 33974
    },
    {
      "epoch": 0.9789373595343744,
      "grad_norm": 0.7192304730415344,
      "learning_rate": 1.162986411326894e-08,
      "loss": 0.6693,
      "step": 33975
    },
    {
      "epoch": 0.9789661729960237,
      "grad_norm": 0.7304330468177795,
      "learning_rate": 1.1598079122296934e-08,
      "loss": 0.665,
      "step": 33976
    },
    {
      "epoch": 0.978994986457673,
      "grad_norm": 0.7141872048377991,
      "learning_rate": 1.1566337575318887e-08,
      "loss": 0.6541,
      "step": 33977
    },
    {
      "epoch": 0.9790237999193223,
      "grad_norm": 0.7171163558959961,
      "learning_rate": 1.1534639472611242e-08,
      "loss": 0.6544,
      "step": 33978
    },
    {
      "epoch": 0.9790526133809716,
      "grad_norm": 0.7364949584007263,
      "learning_rate": 1.1502984814450446e-08,
      "loss": 0.6692,
      "step": 33979
    },
    {
      "epoch": 0.9790814268426209,
      "grad_norm": 0.7230353951454163,
      "learning_rate": 1.1471373601111836e-08,
      "loss": 0.6942,
      "step": 33980
    },
    {
      "epoch": 0.9791102403042702,
      "grad_norm": 1.1560159921646118,
      "learning_rate": 1.1439805832870743e-08,
      "loss": 0.6904,
      "step": 33981
    },
    {
      "epoch": 0.9791390537659195,
      "grad_norm": 0.7306797504425049,
      "learning_rate": 1.1408281510002505e-08,
      "loss": 0.6802,
      "step": 33982
    },
    {
      "epoch": 0.9791678672275688,
      "grad_norm": 0.7344627976417542,
      "learning_rate": 1.1376800632780793e-08,
      "loss": 0.6876,
      "step": 33983
    },
    {
      "epoch": 0.9791966806892181,
      "grad_norm": 0.7319625616073608,
      "learning_rate": 1.1345363201480941e-08,
      "loss": 0.6723,
      "step": 33984
    },
    {
      "epoch": 0.9792254941508672,
      "grad_norm": 0.7126555442810059,
      "learning_rate": 1.1313969216376063e-08,
      "loss": 0.6848,
      "step": 33985
    },
    {
      "epoch": 0.9792543076125165,
      "grad_norm": 0.7127446532249451,
      "learning_rate": 1.1282618677739832e-08,
      "loss": 0.6693,
      "step": 33986
    },
    {
      "epoch": 0.9792831210741658,
      "grad_norm": 0.7131998538970947,
      "learning_rate": 1.1251311585844803e-08,
      "loss": 0.6388,
      "step": 33987
    },
    {
      "epoch": 0.9793119345358151,
      "grad_norm": 0.717686116695404,
      "learning_rate": 1.1220047940964096e-08,
      "loss": 0.6787,
      "step": 33988
    },
    {
      "epoch": 0.9793407479974644,
      "grad_norm": 0.7262203097343445,
      "learning_rate": 1.1188827743369713e-08,
      "loss": 0.6876,
      "step": 33989
    },
    {
      "epoch": 0.9793695614591137,
      "grad_norm": 0.7304558157920837,
      "learning_rate": 1.1157650993333658e-08,
      "loss": 0.6641,
      "step": 33990
    },
    {
      "epoch": 0.979398374920763,
      "grad_norm": 0.7028822302818298,
      "learning_rate": 1.112651769112738e-08,
      "loss": 0.6738,
      "step": 33991
    },
    {
      "epoch": 0.9794271883824123,
      "grad_norm": 0.7363387942314148,
      "learning_rate": 1.1095427837021777e-08,
      "loss": 0.6969,
      "step": 33992
    },
    {
      "epoch": 0.9794560018440616,
      "grad_norm": 0.7063617706298828,
      "learning_rate": 1.1064381431288295e-08,
      "loss": 0.6929,
      "step": 33993
    },
    {
      "epoch": 0.9794848153057109,
      "grad_norm": 0.6916977167129517,
      "learning_rate": 1.1033378474197276e-08,
      "loss": 0.665,
      "step": 33994
    },
    {
      "epoch": 0.9795136287673601,
      "grad_norm": 0.7223072052001953,
      "learning_rate": 1.1002418966017947e-08,
      "loss": 0.6809,
      "step": 33995
    },
    {
      "epoch": 0.9795424422290094,
      "grad_norm": 0.7239649295806885,
      "learning_rate": 1.0971502907020093e-08,
      "loss": 0.6796,
      "step": 33996
    },
    {
      "epoch": 0.9795712556906587,
      "grad_norm": 0.721644401550293,
      "learning_rate": 1.0940630297474053e-08,
      "loss": 0.6858,
      "step": 33997
    },
    {
      "epoch": 0.9796000691523079,
      "grad_norm": 0.7252795696258545,
      "learning_rate": 1.0909801137647391e-08,
      "loss": 0.6613,
      "step": 33998
    },
    {
      "epoch": 0.9796288826139572,
      "grad_norm": 0.7326292991638184,
      "learning_rate": 1.0879015427809337e-08,
      "loss": 0.6804,
      "step": 33999
    },
    {
      "epoch": 0.9796576960756065,
      "grad_norm": 0.7032758593559265,
      "learning_rate": 1.0848273168227452e-08,
      "loss": 0.6495,
      "step": 34000
    },
    {
      "epoch": 0.9796865095372558,
      "grad_norm": 0.7107229828834534,
      "learning_rate": 1.0817574359169858e-08,
      "loss": 0.6874,
      "step": 34001
    },
    {
      "epoch": 0.9797153229989051,
      "grad_norm": 0.7359609603881836,
      "learning_rate": 1.0786919000903562e-08,
      "loss": 0.696,
      "step": 34002
    },
    {
      "epoch": 0.9797441364605544,
      "grad_norm": 0.7315839529037476,
      "learning_rate": 1.0756307093696128e-08,
      "loss": 0.7049,
      "step": 34003
    },
    {
      "epoch": 0.9797729499222037,
      "grad_norm": 0.7369439005851746,
      "learning_rate": 1.072573863781401e-08,
      "loss": 0.6544,
      "step": 34004
    },
    {
      "epoch": 0.9798017633838529,
      "grad_norm": 0.7247198820114136,
      "learning_rate": 1.0695213633523105e-08,
      "loss": 0.6849,
      "step": 34005
    },
    {
      "epoch": 0.9798305768455022,
      "grad_norm": 0.7237599492073059,
      "learning_rate": 1.0664732081089868e-08,
      "loss": 0.6789,
      "step": 34006
    },
    {
      "epoch": 0.9798593903071515,
      "grad_norm": 0.7153318524360657,
      "learning_rate": 1.0634293980778532e-08,
      "loss": 0.6886,
      "step": 34007
    },
    {
      "epoch": 0.9798882037688008,
      "grad_norm": 0.7149328589439392,
      "learning_rate": 1.0603899332854995e-08,
      "loss": 0.6954,
      "step": 34008
    },
    {
      "epoch": 0.9799170172304501,
      "grad_norm": 0.7054517865180969,
      "learning_rate": 1.05735481375846e-08,
      "loss": 0.6714,
      "step": 34009
    },
    {
      "epoch": 0.9799458306920994,
      "grad_norm": 0.714666485786438,
      "learning_rate": 1.054324039523047e-08,
      "loss": 0.6749,
      "step": 34010
    },
    {
      "epoch": 0.9799746441537487,
      "grad_norm": 0.7479141354560852,
      "learning_rate": 1.0512976106057393e-08,
      "loss": 0.6802,
      "step": 34011
    },
    {
      "epoch": 0.9800034576153979,
      "grad_norm": 0.7279176115989685,
      "learning_rate": 1.0482755270328492e-08,
      "loss": 0.6922,
      "step": 34012
    },
    {
      "epoch": 0.9800322710770472,
      "grad_norm": 0.7122841477394104,
      "learning_rate": 1.0452577888306893e-08,
      "loss": 0.6813,
      "step": 34013
    },
    {
      "epoch": 0.9800610845386964,
      "grad_norm": 0.7196161150932312,
      "learning_rate": 1.0422443960255712e-08,
      "loss": 0.6746,
      "step": 34014
    },
    {
      "epoch": 0.9800898980003457,
      "grad_norm": 0.7257657051086426,
      "learning_rate": 1.0392353486437523e-08,
      "loss": 0.6818,
      "step": 34015
    },
    {
      "epoch": 0.980118711461995,
      "grad_norm": 0.7243902683258057,
      "learning_rate": 1.0362306467113781e-08,
      "loss": 0.6654,
      "step": 34016
    },
    {
      "epoch": 0.9801475249236443,
      "grad_norm": 0.7127824425697327,
      "learning_rate": 1.0332302902547053e-08,
      "loss": 0.6958,
      "step": 34017
    },
    {
      "epoch": 0.9801763383852936,
      "grad_norm": 0.7493074536323547,
      "learning_rate": 1.0302342792997688e-08,
      "loss": 0.6803,
      "step": 34018
    },
    {
      "epoch": 0.9802051518469429,
      "grad_norm": 0.7149170637130737,
      "learning_rate": 1.0272426138727143e-08,
      "loss": 0.6779,
      "step": 34019
    },
    {
      "epoch": 0.9802339653085922,
      "grad_norm": 0.7212996482849121,
      "learning_rate": 1.0242552939995765e-08,
      "loss": 0.6618,
      "step": 34020
    },
    {
      "epoch": 0.9802627787702415,
      "grad_norm": 0.7154707312583923,
      "learning_rate": 1.02127231970639e-08,
      "loss": 0.6732,
      "step": 34021
    },
    {
      "epoch": 0.9802915922318908,
      "grad_norm": 0.7059791088104248,
      "learning_rate": 1.0182936910191344e-08,
      "loss": 0.6813,
      "step": 34022
    },
    {
      "epoch": 0.9803204056935401,
      "grad_norm": 0.713217556476593,
      "learning_rate": 1.015319407963733e-08,
      "loss": 0.6852,
      "step": 34023
    },
    {
      "epoch": 0.9803492191551894,
      "grad_norm": 0.716072142124176,
      "learning_rate": 1.0123494705661096e-08,
      "loss": 0.6741,
      "step": 34024
    },
    {
      "epoch": 0.9803780326168385,
      "grad_norm": 0.7145424485206604,
      "learning_rate": 1.0093838788521326e-08,
      "loss": 0.6838,
      "step": 34025
    },
    {
      "epoch": 0.9804068460784878,
      "grad_norm": 0.7488791942596436,
      "learning_rate": 1.0064226328476145e-08,
      "loss": 0.6929,
      "step": 34026
    },
    {
      "epoch": 0.9804356595401371,
      "grad_norm": 0.7315341830253601,
      "learning_rate": 1.0034657325783126e-08,
      "loss": 0.6851,
      "step": 34027
    },
    {
      "epoch": 0.9804644730017864,
      "grad_norm": 0.7109079360961914,
      "learning_rate": 1.000513178069984e-08,
      "loss": 0.6864,
      "step": 34028
    },
    {
      "epoch": 0.9804932864634357,
      "grad_norm": 0.6892611384391785,
      "learning_rate": 9.975649693484413e-09,
      "loss": 0.664,
      "step": 34029
    },
    {
      "epoch": 0.980522099925085,
      "grad_norm": 0.7255356311798096,
      "learning_rate": 9.946211064392197e-09,
      "loss": 0.6845,
      "step": 34030
    },
    {
      "epoch": 0.9805509133867343,
      "grad_norm": 0.7028690576553345,
      "learning_rate": 9.916815893680765e-09,
      "loss": 0.6698,
      "step": 34031
    },
    {
      "epoch": 0.9805797268483836,
      "grad_norm": 0.7232012748718262,
      "learning_rate": 9.887464181605465e-09,
      "loss": 0.6733,
      "step": 34032
    },
    {
      "epoch": 0.9806085403100329,
      "grad_norm": 0.7194300889968872,
      "learning_rate": 9.858155928422208e-09,
      "loss": 0.6736,
      "step": 34033
    },
    {
      "epoch": 0.9806373537716822,
      "grad_norm": 0.7956731915473938,
      "learning_rate": 9.828891134385787e-09,
      "loss": 0.69,
      "step": 34034
    },
    {
      "epoch": 0.9806661672333314,
      "grad_norm": 0.69757080078125,
      "learning_rate": 9.799669799751554e-09,
      "loss": 0.671,
      "step": 34035
    },
    {
      "epoch": 0.9806949806949807,
      "grad_norm": 0.7094818353652954,
      "learning_rate": 9.77049192477375e-09,
      "loss": 0.6721,
      "step": 34036
    },
    {
      "epoch": 0.98072379415663,
      "grad_norm": 0.7211748957633972,
      "learning_rate": 9.741357509706618e-09,
      "loss": 0.6696,
      "step": 34037
    },
    {
      "epoch": 0.9807526076182793,
      "grad_norm": 0.7210122346878052,
      "learning_rate": 9.712266554803285e-09,
      "loss": 0.679,
      "step": 34038
    },
    {
      "epoch": 0.9807814210799285,
      "grad_norm": 0.7380317449569702,
      "learning_rate": 9.68321906031855e-09,
      "loss": 0.6822,
      "step": 34039
    },
    {
      "epoch": 0.9808102345415778,
      "grad_norm": 0.7280035614967346,
      "learning_rate": 9.654215026503877e-09,
      "loss": 0.6565,
      "step": 34040
    },
    {
      "epoch": 0.9808390480032271,
      "grad_norm": 0.7267963290214539,
      "learning_rate": 9.625254453612954e-09,
      "loss": 0.6854,
      "step": 34041
    },
    {
      "epoch": 0.9808678614648764,
      "grad_norm": 0.7151725888252258,
      "learning_rate": 9.596337341897244e-09,
      "loss": 0.6958,
      "step": 34042
    },
    {
      "epoch": 0.9808966749265257,
      "grad_norm": 0.7279757261276245,
      "learning_rate": 9.56746369160877e-09,
      "loss": 0.6775,
      "step": 34043
    },
    {
      "epoch": 0.980925488388175,
      "grad_norm": 0.71843421459198,
      "learning_rate": 9.538633502998995e-09,
      "loss": 0.6813,
      "step": 34044
    },
    {
      "epoch": 0.9809543018498242,
      "grad_norm": 0.7273525595664978,
      "learning_rate": 9.509846776319387e-09,
      "loss": 0.705,
      "step": 34045
    },
    {
      "epoch": 0.9809831153114735,
      "grad_norm": 0.7148823142051697,
      "learning_rate": 9.4811035118203e-09,
      "loss": 0.6729,
      "step": 34046
    },
    {
      "epoch": 0.9810119287731228,
      "grad_norm": 0.7126555442810059,
      "learning_rate": 9.452403709751533e-09,
      "loss": 0.6497,
      "step": 34047
    },
    {
      "epoch": 0.9810407422347721,
      "grad_norm": 0.7185085415840149,
      "learning_rate": 9.423747370364555e-09,
      "loss": 0.6576,
      "step": 34048
    },
    {
      "epoch": 0.9810695556964214,
      "grad_norm": 0.7079927921295166,
      "learning_rate": 9.395134493907498e-09,
      "loss": 0.6624,
      "step": 34049
    },
    {
      "epoch": 0.9810983691580707,
      "grad_norm": 0.7161393165588379,
      "learning_rate": 9.366565080630164e-09,
      "loss": 0.6695,
      "step": 34050
    },
    {
      "epoch": 0.98112718261972,
      "grad_norm": 0.7254663109779358,
      "learning_rate": 9.338039130780686e-09,
      "loss": 0.6721,
      "step": 34051
    },
    {
      "epoch": 0.9811559960813692,
      "grad_norm": 0.7500337958335876,
      "learning_rate": 9.309556644608864e-09,
      "loss": 0.7106,
      "step": 34052
    },
    {
      "epoch": 0.9811848095430185,
      "grad_norm": 0.730827808380127,
      "learning_rate": 9.281117622361724e-09,
      "loss": 0.67,
      "step": 34053
    },
    {
      "epoch": 0.9812136230046677,
      "grad_norm": 0.7324656248092651,
      "learning_rate": 9.252722064286845e-09,
      "loss": 0.6588,
      "step": 34054
    },
    {
      "epoch": 0.981242436466317,
      "grad_norm": 0.7202053666114807,
      "learning_rate": 9.224369970631809e-09,
      "loss": 0.6828,
      "step": 34055
    },
    {
      "epoch": 0.9812712499279663,
      "grad_norm": 0.7192789912223816,
      "learning_rate": 9.196061341644191e-09,
      "loss": 0.6792,
      "step": 34056
    },
    {
      "epoch": 0.9813000633896156,
      "grad_norm": 0.7307013273239136,
      "learning_rate": 9.167796177569354e-09,
      "loss": 0.6644,
      "step": 34057
    },
    {
      "epoch": 0.9813288768512649,
      "grad_norm": 0.7193447947502136,
      "learning_rate": 9.139574478654323e-09,
      "loss": 0.685,
      "step": 34058
    },
    {
      "epoch": 0.9813576903129142,
      "grad_norm": 0.722980797290802,
      "learning_rate": 9.1113962451439e-09,
      "loss": 0.6712,
      "step": 34059
    },
    {
      "epoch": 0.9813865037745635,
      "grad_norm": 0.7230120897293091,
      "learning_rate": 9.083261477284555e-09,
      "loss": 0.6716,
      "step": 34060
    },
    {
      "epoch": 0.9814153172362128,
      "grad_norm": 0.7227864265441895,
      "learning_rate": 9.055170175321093e-09,
      "loss": 0.6701,
      "step": 34061
    },
    {
      "epoch": 0.9814441306978621,
      "grad_norm": 0.7262833118438721,
      "learning_rate": 9.027122339497763e-09,
      "loss": 0.6568,
      "step": 34062
    },
    {
      "epoch": 0.9814729441595114,
      "grad_norm": 0.7301275134086609,
      "learning_rate": 8.999117970059369e-09,
      "loss": 0.6728,
      "step": 34063
    },
    {
      "epoch": 0.9815017576211607,
      "grad_norm": 0.7200565934181213,
      "learning_rate": 8.971157067249048e-09,
      "loss": 0.6535,
      "step": 34064
    },
    {
      "epoch": 0.9815305710828099,
      "grad_norm": 0.7257611751556396,
      "learning_rate": 8.94323963131105e-09,
      "loss": 0.685,
      "step": 34065
    },
    {
      "epoch": 0.9815593845444591,
      "grad_norm": 0.7325758337974548,
      "learning_rate": 8.915365662488517e-09,
      "loss": 0.6742,
      "step": 34066
    },
    {
      "epoch": 0.9815881980061084,
      "grad_norm": 0.708106279373169,
      "learning_rate": 8.887535161023475e-09,
      "loss": 0.6973,
      "step": 34067
    },
    {
      "epoch": 0.9816170114677577,
      "grad_norm": 0.7270239591598511,
      "learning_rate": 8.859748127158507e-09,
      "loss": 0.676,
      "step": 34068
    },
    {
      "epoch": 0.981645824929407,
      "grad_norm": 0.720406174659729,
      "learning_rate": 8.832004561136198e-09,
      "loss": 0.6667,
      "step": 34069
    },
    {
      "epoch": 0.9816746383910563,
      "grad_norm": 0.6948850750923157,
      "learning_rate": 8.804304463197466e-09,
      "loss": 0.6641,
      "step": 34070
    },
    {
      "epoch": 0.9817034518527056,
      "grad_norm": 0.7067196369171143,
      "learning_rate": 8.77664783358434e-09,
      "loss": 0.655,
      "step": 34071
    },
    {
      "epoch": 0.9817322653143549,
      "grad_norm": 0.7107172608375549,
      "learning_rate": 8.749034672536627e-09,
      "loss": 0.6802,
      "step": 34072
    },
    {
      "epoch": 0.9817610787760042,
      "grad_norm": 0.7339120507240295,
      "learning_rate": 8.721464980295802e-09,
      "loss": 0.6833,
      "step": 34073
    },
    {
      "epoch": 0.9817898922376534,
      "grad_norm": 0.7115541100502014,
      "learning_rate": 8.693938757101673e-09,
      "loss": 0.6635,
      "step": 34074
    },
    {
      "epoch": 0.9818187056993027,
      "grad_norm": 0.7311985492706299,
      "learning_rate": 8.66645600319349e-09,
      "loss": 0.6888,
      "step": 34075
    },
    {
      "epoch": 0.981847519160952,
      "grad_norm": 0.7271917462348938,
      "learning_rate": 8.639016718811066e-09,
      "loss": 0.6702,
      "step": 34076
    },
    {
      "epoch": 0.9818763326226013,
      "grad_norm": 0.7134053111076355,
      "learning_rate": 8.611620904193097e-09,
      "loss": 0.6566,
      "step": 34077
    },
    {
      "epoch": 0.9819051460842506,
      "grad_norm": 0.7314469218254089,
      "learning_rate": 8.584268559578834e-09,
      "loss": 0.6868,
      "step": 34078
    },
    {
      "epoch": 0.9819339595458998,
      "grad_norm": 0.7263697981834412,
      "learning_rate": 8.556959685205869e-09,
      "loss": 0.657,
      "step": 34079
    },
    {
      "epoch": 0.9819627730075491,
      "grad_norm": 0.7168598771095276,
      "learning_rate": 8.529694281311784e-09,
      "loss": 0.6736,
      "step": 34080
    },
    {
      "epoch": 0.9819915864691984,
      "grad_norm": 0.7085413932800293,
      "learning_rate": 8.502472348134727e-09,
      "loss": 0.6527,
      "step": 34081
    },
    {
      "epoch": 0.9820203999308477,
      "grad_norm": 0.7030443549156189,
      "learning_rate": 8.475293885911173e-09,
      "loss": 0.6929,
      "step": 34082
    },
    {
      "epoch": 0.982049213392497,
      "grad_norm": 0.7300214171409607,
      "learning_rate": 8.448158894878711e-09,
      "loss": 0.6828,
      "step": 34083
    },
    {
      "epoch": 0.9820780268541462,
      "grad_norm": 0.7017415761947632,
      "learning_rate": 8.421067375272708e-09,
      "loss": 0.6559,
      "step": 34084
    },
    {
      "epoch": 0.9821068403157955,
      "grad_norm": 0.7264453768730164,
      "learning_rate": 8.394019327329083e-09,
      "loss": 0.6696,
      "step": 34085
    },
    {
      "epoch": 0.9821356537774448,
      "grad_norm": 0.7279742956161499,
      "learning_rate": 8.367014751284319e-09,
      "loss": 0.6916,
      "step": 34086
    },
    {
      "epoch": 0.9821644672390941,
      "grad_norm": 0.712223470211029,
      "learning_rate": 8.34005364737267e-09,
      "loss": 0.6581,
      "step": 34087
    },
    {
      "epoch": 0.9821932807007434,
      "grad_norm": 0.7070525288581848,
      "learning_rate": 8.313136015830058e-09,
      "loss": 0.6818,
      "step": 34088
    },
    {
      "epoch": 0.9822220941623927,
      "grad_norm": 0.7252669930458069,
      "learning_rate": 8.286261856889077e-09,
      "loss": 0.6814,
      "step": 34089
    },
    {
      "epoch": 0.982250907624042,
      "grad_norm": 0.8745826482772827,
      "learning_rate": 8.259431170785648e-09,
      "loss": 0.6532,
      "step": 34090
    },
    {
      "epoch": 0.9822797210856913,
      "grad_norm": 0.7268363833427429,
      "learning_rate": 8.232643957752361e-09,
      "loss": 0.6832,
      "step": 34091
    },
    {
      "epoch": 0.9823085345473406,
      "grad_norm": 0.7221406698226929,
      "learning_rate": 8.205900218022923e-09,
      "loss": 0.6734,
      "step": 34092
    },
    {
      "epoch": 0.9823373480089898,
      "grad_norm": 0.7134835720062256,
      "learning_rate": 8.179199951829919e-09,
      "loss": 0.6471,
      "step": 34093
    },
    {
      "epoch": 0.982366161470639,
      "grad_norm": 0.75246661901474,
      "learning_rate": 8.1525431594065e-09,
      "loss": 0.6937,
      "step": 34094
    },
    {
      "epoch": 0.9823949749322883,
      "grad_norm": 0.722145140171051,
      "learning_rate": 8.125929840984148e-09,
      "loss": 0.6954,
      "step": 34095
    },
    {
      "epoch": 0.9824237883939376,
      "grad_norm": 0.7339918613433838,
      "learning_rate": 8.099359996794898e-09,
      "loss": 0.6628,
      "step": 34096
    },
    {
      "epoch": 0.9824526018555869,
      "grad_norm": 0.7249178290367126,
      "learning_rate": 8.07283362707023e-09,
      "loss": 0.6707,
      "step": 34097
    },
    {
      "epoch": 0.9824814153172362,
      "grad_norm": 0.7103301882743835,
      "learning_rate": 8.046350732041075e-09,
      "loss": 0.6633,
      "step": 34098
    },
    {
      "epoch": 0.9825102287788855,
      "grad_norm": 0.7275580167770386,
      "learning_rate": 8.019911311937801e-09,
      "loss": 0.6748,
      "step": 34099
    },
    {
      "epoch": 0.9825390422405348,
      "grad_norm": 0.7279692888259888,
      "learning_rate": 7.993515366991334e-09,
      "loss": 0.6871,
      "step": 34100
    },
    {
      "epoch": 0.9825678557021841,
      "grad_norm": 0.7385299801826477,
      "learning_rate": 7.967162897430936e-09,
      "loss": 0.6646,
      "step": 34101
    },
    {
      "epoch": 0.9825966691638334,
      "grad_norm": 0.7356212139129639,
      "learning_rate": 7.940853903486424e-09,
      "loss": 0.6855,
      "step": 34102
    },
    {
      "epoch": 0.9826254826254827,
      "grad_norm": 0.7153186798095703,
      "learning_rate": 7.9145883853865e-09,
      "loss": 0.6693,
      "step": 34103
    },
    {
      "epoch": 0.982654296087132,
      "grad_norm": 0.7352673411369324,
      "learning_rate": 7.888366343360432e-09,
      "loss": 0.6734,
      "step": 34104
    },
    {
      "epoch": 0.9826831095487812,
      "grad_norm": 0.7329410314559937,
      "learning_rate": 7.86218777763692e-09,
      "loss": 0.6973,
      "step": 34105
    },
    {
      "epoch": 0.9827119230104304,
      "grad_norm": 0.7259053587913513,
      "learning_rate": 7.836052688443008e-09,
      "loss": 0.6828,
      "step": 34106
    },
    {
      "epoch": 0.9827407364720797,
      "grad_norm": 0.713359534740448,
      "learning_rate": 7.809961076006845e-09,
      "loss": 0.6931,
      "step": 34107
    },
    {
      "epoch": 0.982769549933729,
      "grad_norm": 0.7169216275215149,
      "learning_rate": 7.783912940555472e-09,
      "loss": 0.6763,
      "step": 34108
    },
    {
      "epoch": 0.9827983633953783,
      "grad_norm": 0.7299913167953491,
      "learning_rate": 7.757908282315929e-09,
      "loss": 0.676,
      "step": 34109
    },
    {
      "epoch": 0.9828271768570276,
      "grad_norm": 0.7120156288146973,
      "learning_rate": 7.731947101514702e-09,
      "loss": 0.668,
      "step": 34110
    },
    {
      "epoch": 0.9828559903186769,
      "grad_norm": 0.8676798939704895,
      "learning_rate": 7.706029398377168e-09,
      "loss": 0.6648,
      "step": 34111
    },
    {
      "epoch": 0.9828848037803262,
      "grad_norm": 0.73637455701828,
      "learning_rate": 7.680155173130366e-09,
      "loss": 0.6855,
      "step": 34112
    },
    {
      "epoch": 0.9829136172419755,
      "grad_norm": 0.725886881351471,
      "learning_rate": 7.65432442599856e-09,
      "loss": 0.6944,
      "step": 34113
    },
    {
      "epoch": 0.9829424307036247,
      "grad_norm": 0.7479415535926819,
      "learning_rate": 7.628537157207128e-09,
      "loss": 0.697,
      "step": 34114
    },
    {
      "epoch": 0.982971244165274,
      "grad_norm": 0.7295541763305664,
      "learning_rate": 7.602793366980888e-09,
      "loss": 0.6991,
      "step": 34115
    },
    {
      "epoch": 0.9830000576269233,
      "grad_norm": 0.7343343496322632,
      "learning_rate": 7.577093055542995e-09,
      "loss": 0.6714,
      "step": 34116
    },
    {
      "epoch": 0.9830288710885726,
      "grad_norm": 0.71192467212677,
      "learning_rate": 7.551436223118824e-09,
      "loss": 0.6668,
      "step": 34117
    },
    {
      "epoch": 0.9830576845502219,
      "grad_norm": 0.7409781217575073,
      "learning_rate": 7.525822869930422e-09,
      "loss": 0.6598,
      "step": 34118
    },
    {
      "epoch": 0.9830864980118712,
      "grad_norm": 0.7409491539001465,
      "learning_rate": 7.50025299620205e-09,
      "loss": 0.6861,
      "step": 34119
    },
    {
      "epoch": 0.9831153114735204,
      "grad_norm": 0.7320320010185242,
      "learning_rate": 7.474726602155202e-09,
      "loss": 0.6891,
      "step": 34120
    },
    {
      "epoch": 0.9831441249351697,
      "grad_norm": 0.7321122884750366,
      "learning_rate": 7.449243688013031e-09,
      "loss": 0.6661,
      "step": 34121
    },
    {
      "epoch": 0.983172938396819,
      "grad_norm": 0.779488205909729,
      "learning_rate": 7.42380425399758e-09,
      "loss": 0.6953,
      "step": 34122
    },
    {
      "epoch": 0.9832017518584683,
      "grad_norm": 0.724743664264679,
      "learning_rate": 7.398408300329784e-09,
      "loss": 0.695,
      "step": 34123
    },
    {
      "epoch": 0.9832305653201175,
      "grad_norm": 0.7416858673095703,
      "learning_rate": 7.3730558272305795e-09,
      "loss": 0.6893,
      "step": 34124
    },
    {
      "epoch": 0.9832593787817668,
      "grad_norm": 0.733474850654602,
      "learning_rate": 7.34774683492201e-09,
      "loss": 0.6924,
      "step": 34125
    },
    {
      "epoch": 0.9832881922434161,
      "grad_norm": 0.6856785416603088,
      "learning_rate": 7.3224813236233425e-09,
      "loss": 0.6574,
      "step": 34126
    },
    {
      "epoch": 0.9833170057050654,
      "grad_norm": 0.7095852494239807,
      "learning_rate": 7.297259293554959e-09,
      "loss": 0.6858,
      "step": 34127
    },
    {
      "epoch": 0.9833458191667147,
      "grad_norm": 0.7074068784713745,
      "learning_rate": 7.2720807449366825e-09,
      "loss": 0.6653,
      "step": 34128
    },
    {
      "epoch": 0.983374632628364,
      "grad_norm": 0.7148502469062805,
      "learning_rate": 7.246945677987782e-09,
      "loss": 0.6735,
      "step": 34129
    },
    {
      "epoch": 0.9834034460900133,
      "grad_norm": 0.7100455164909363,
      "learning_rate": 7.221854092926972e-09,
      "loss": 0.6725,
      "step": 34130
    },
    {
      "epoch": 0.9834322595516626,
      "grad_norm": 0.7194573283195496,
      "learning_rate": 7.1968059899729655e-09,
      "loss": 0.682,
      "step": 34131
    },
    {
      "epoch": 0.9834610730133119,
      "grad_norm": 0.7045111656188965,
      "learning_rate": 7.171801369343922e-09,
      "loss": 0.691,
      "step": 34132
    },
    {
      "epoch": 0.983489886474961,
      "grad_norm": 0.708720326423645,
      "learning_rate": 7.146840231256891e-09,
      "loss": 0.6918,
      "step": 34133
    },
    {
      "epoch": 0.9835186999366103,
      "grad_norm": 0.7157507538795471,
      "learning_rate": 7.1219225759305845e-09,
      "loss": 0.6959,
      "step": 34134
    },
    {
      "epoch": 0.9835475133982596,
      "grad_norm": 0.7440839409828186,
      "learning_rate": 7.097048403580941e-09,
      "loss": 0.7208,
      "step": 34135
    },
    {
      "epoch": 0.9835763268599089,
      "grad_norm": 0.7239024639129639,
      "learning_rate": 7.072217714424456e-09,
      "loss": 0.7066,
      "step": 34136
    },
    {
      "epoch": 0.9836051403215582,
      "grad_norm": 0.7120130658149719,
      "learning_rate": 7.047430508678732e-09,
      "loss": 0.6743,
      "step": 34137
    },
    {
      "epoch": 0.9836339537832075,
      "grad_norm": 0.6974732875823975,
      "learning_rate": 7.022686786558042e-09,
      "loss": 0.6601,
      "step": 34138
    },
    {
      "epoch": 0.9836627672448568,
      "grad_norm": 0.7535877823829651,
      "learning_rate": 6.9979865482788786e-09,
      "loss": 0.687,
      "step": 34139
    },
    {
      "epoch": 0.9836915807065061,
      "grad_norm": 0.725976824760437,
      "learning_rate": 6.973329794056072e-09,
      "loss": 0.6922,
      "step": 34140
    },
    {
      "epoch": 0.9837203941681554,
      "grad_norm": 0.7220659255981445,
      "learning_rate": 6.948716524104449e-09,
      "loss": 0.6668,
      "step": 34141
    },
    {
      "epoch": 0.9837492076298047,
      "grad_norm": 0.7069684863090515,
      "learning_rate": 6.924146738638282e-09,
      "loss": 0.6624,
      "step": 34142
    },
    {
      "epoch": 0.983778021091454,
      "grad_norm": 0.75835782289505,
      "learning_rate": 6.899620437871846e-09,
      "loss": 0.6903,
      "step": 34143
    },
    {
      "epoch": 0.9838068345531032,
      "grad_norm": 0.7036932110786438,
      "learning_rate": 6.875137622017747e-09,
      "loss": 0.6618,
      "step": 34144
    },
    {
      "epoch": 0.9838356480147525,
      "grad_norm": 0.7461510896682739,
      "learning_rate": 6.850698291290259e-09,
      "loss": 0.707,
      "step": 34145
    },
    {
      "epoch": 0.9838644614764018,
      "grad_norm": 0.7078622579574585,
      "learning_rate": 6.826302445901989e-09,
      "loss": 0.6508,
      "step": 34146
    },
    {
      "epoch": 0.983893274938051,
      "grad_norm": 0.7104282379150391,
      "learning_rate": 6.8019500860649904e-09,
      "loss": 0.6819,
      "step": 34147
    },
    {
      "epoch": 0.9839220883997003,
      "grad_norm": 0.7298935651779175,
      "learning_rate": 6.77764121199187e-09,
      "loss": 0.6856,
      "step": 34148
    },
    {
      "epoch": 0.9839509018613496,
      "grad_norm": 0.7248514294624329,
      "learning_rate": 6.753375823894126e-09,
      "loss": 0.6993,
      "step": 34149
    },
    {
      "epoch": 0.9839797153229989,
      "grad_norm": 0.7434964179992676,
      "learning_rate": 6.729153921982701e-09,
      "loss": 0.6923,
      "step": 34150
    },
    {
      "epoch": 0.9840085287846482,
      "grad_norm": 0.7487187385559082,
      "learning_rate": 6.704975506469092e-09,
      "loss": 0.6969,
      "step": 34151
    },
    {
      "epoch": 0.9840373422462975,
      "grad_norm": 0.7243791222572327,
      "learning_rate": 6.6808405775636855e-09,
      "loss": 0.6695,
      "step": 34152
    },
    {
      "epoch": 0.9840661557079468,
      "grad_norm": 0.7469386458396912,
      "learning_rate": 6.65674913547687e-09,
      "loss": 0.6656,
      "step": 34153
    },
    {
      "epoch": 0.984094969169596,
      "grad_norm": 0.7391794919967651,
      "learning_rate": 6.632701180418477e-09,
      "loss": 0.6856,
      "step": 34154
    },
    {
      "epoch": 0.9841237826312453,
      "grad_norm": 0.7401858568191528,
      "learning_rate": 6.608696712597229e-09,
      "loss": 0.6796,
      "step": 34155
    },
    {
      "epoch": 0.9841525960928946,
      "grad_norm": 0.7304913997650146,
      "learning_rate": 6.584735732222402e-09,
      "loss": 0.6901,
      "step": 34156
    },
    {
      "epoch": 0.9841814095545439,
      "grad_norm": 0.7257526516914368,
      "learning_rate": 6.560818239503275e-09,
      "loss": 0.6644,
      "step": 34157
    },
    {
      "epoch": 0.9842102230161932,
      "grad_norm": 0.7281669974327087,
      "learning_rate": 6.536944234648013e-09,
      "loss": 0.6812,
      "step": 34158
    },
    {
      "epoch": 0.9842390364778425,
      "grad_norm": 0.7276433110237122,
      "learning_rate": 6.513113717864228e-09,
      "loss": 0.6833,
      "step": 34159
    },
    {
      "epoch": 0.9842678499394917,
      "grad_norm": 0.7454199194908142,
      "learning_rate": 6.489326689359532e-09,
      "loss": 0.6641,
      "step": 34160
    },
    {
      "epoch": 0.984296663401141,
      "grad_norm": 0.7060880064964294,
      "learning_rate": 6.465583149340981e-09,
      "loss": 0.6631,
      "step": 34161
    },
    {
      "epoch": 0.9843254768627903,
      "grad_norm": 0.7133337259292603,
      "learning_rate": 6.441883098015633e-09,
      "loss": 0.6559,
      "step": 34162
    },
    {
      "epoch": 0.9843542903244396,
      "grad_norm": 0.7356348037719727,
      "learning_rate": 6.418226535589989e-09,
      "loss": 0.68,
      "step": 34163
    },
    {
      "epoch": 0.9843831037860888,
      "grad_norm": 0.7237750887870789,
      "learning_rate": 6.394613462269439e-09,
      "loss": 0.6676,
      "step": 34164
    },
    {
      "epoch": 0.9844119172477381,
      "grad_norm": 0.7264835238456726,
      "learning_rate": 6.37104387825993e-09,
      "loss": 0.68,
      "step": 34165
    },
    {
      "epoch": 0.9844407307093874,
      "grad_norm": 0.7282779812812805,
      "learning_rate": 6.3475177837674095e-09,
      "loss": 0.6858,
      "step": 34166
    },
    {
      "epoch": 0.9844695441710367,
      "grad_norm": 0.7258532643318176,
      "learning_rate": 6.324035178995603e-09,
      "loss": 0.6743,
      "step": 34167
    },
    {
      "epoch": 0.984498357632686,
      "grad_norm": 0.7261528372764587,
      "learning_rate": 6.3005960641498995e-09,
      "loss": 0.6998,
      "step": 34168
    },
    {
      "epoch": 0.9845271710943353,
      "grad_norm": 0.7211989164352417,
      "learning_rate": 6.277200439434028e-09,
      "loss": 0.6752,
      "step": 34169
    },
    {
      "epoch": 0.9845559845559846,
      "grad_norm": 0.734246015548706,
      "learning_rate": 6.253848305052268e-09,
      "loss": 0.6921,
      "step": 34170
    },
    {
      "epoch": 0.9845847980176339,
      "grad_norm": 0.7152836322784424,
      "learning_rate": 6.23053966120668e-09,
      "loss": 0.6923,
      "step": 34171
    },
    {
      "epoch": 0.9846136114792832,
      "grad_norm": 0.7284923195838928,
      "learning_rate": 6.207274508102101e-09,
      "loss": 0.6904,
      "step": 34172
    },
    {
      "epoch": 0.9846424249409323,
      "grad_norm": 0.7528612017631531,
      "learning_rate": 6.184052845939481e-09,
      "loss": 0.6823,
      "step": 34173
    },
    {
      "epoch": 0.9846712384025816,
      "grad_norm": 0.7109158039093018,
      "learning_rate": 6.16087467492199e-09,
      "loss": 0.7036,
      "step": 34174
    },
    {
      "epoch": 0.9847000518642309,
      "grad_norm": 0.7193949818611145,
      "learning_rate": 6.137739995251135e-09,
      "loss": 0.6869,
      "step": 34175
    },
    {
      "epoch": 0.9847288653258802,
      "grad_norm": 0.7270771861076355,
      "learning_rate": 6.114648807128975e-09,
      "loss": 0.6655,
      "step": 34176
    },
    {
      "epoch": 0.9847576787875295,
      "grad_norm": 0.7442782521247864,
      "learning_rate": 6.091601110755907e-09,
      "loss": 0.7101,
      "step": 34177
    },
    {
      "epoch": 0.9847864922491788,
      "grad_norm": 0.7121527791023254,
      "learning_rate": 6.06859690633288e-09,
      "loss": 0.6984,
      "step": 34178
    },
    {
      "epoch": 0.9848153057108281,
      "grad_norm": 0.7349875569343567,
      "learning_rate": 6.045636194060289e-09,
      "loss": 0.6953,
      "step": 34179
    },
    {
      "epoch": 0.9848441191724774,
      "grad_norm": 0.71563321352005,
      "learning_rate": 6.022718974137976e-09,
      "loss": 0.6862,
      "step": 34180
    },
    {
      "epoch": 0.9848729326341267,
      "grad_norm": 0.7330766916275024,
      "learning_rate": 5.999845246765779e-09,
      "loss": 0.6788,
      "step": 34181
    },
    {
      "epoch": 0.984901746095776,
      "grad_norm": 0.7521395683288574,
      "learning_rate": 5.9770150121429834e-09,
      "loss": 0.6584,
      "step": 34182
    },
    {
      "epoch": 0.9849305595574253,
      "grad_norm": 0.7206390500068665,
      "learning_rate": 5.954228270467765e-09,
      "loss": 0.6533,
      "step": 34183
    },
    {
      "epoch": 0.9849593730190745,
      "grad_norm": 0.703665018081665,
      "learning_rate": 5.931485021938854e-09,
      "loss": 0.6792,
      "step": 34184
    },
    {
      "epoch": 0.9849881864807238,
      "grad_norm": 0.7303602695465088,
      "learning_rate": 5.90878526675498e-09,
      "loss": 0.6866,
      "step": 34185
    },
    {
      "epoch": 0.9850169999423731,
      "grad_norm": 0.7143299579620361,
      "learning_rate": 5.886129005113206e-09,
      "loss": 0.6865,
      "step": 34186
    },
    {
      "epoch": 0.9850458134040223,
      "grad_norm": 0.7030778527259827,
      "learning_rate": 5.863516237210598e-09,
      "loss": 0.683,
      "step": 34187
    },
    {
      "epoch": 0.9850746268656716,
      "grad_norm": 0.7003796696662903,
      "learning_rate": 5.840946963244776e-09,
      "loss": 0.6542,
      "step": 34188
    },
    {
      "epoch": 0.9851034403273209,
      "grad_norm": 0.7318357825279236,
      "learning_rate": 5.818421183412248e-09,
      "loss": 0.6882,
      "step": 34189
    },
    {
      "epoch": 0.9851322537889702,
      "grad_norm": 0.7337246537208557,
      "learning_rate": 5.795938897908415e-09,
      "loss": 0.6672,
      "step": 34190
    },
    {
      "epoch": 0.9851610672506195,
      "grad_norm": 0.7248355746269226,
      "learning_rate": 5.773500106929785e-09,
      "loss": 0.6732,
      "step": 34191
    },
    {
      "epoch": 0.9851898807122688,
      "grad_norm": 0.731322169303894,
      "learning_rate": 5.751104810671759e-09,
      "loss": 0.6781,
      "step": 34192
    },
    {
      "epoch": 0.985218694173918,
      "grad_norm": 0.7236550450325012,
      "learning_rate": 5.728753009328625e-09,
      "loss": 0.6872,
      "step": 34193
    },
    {
      "epoch": 0.9852475076355673,
      "grad_norm": 0.7002314329147339,
      "learning_rate": 5.706444703096337e-09,
      "loss": 0.6427,
      "step": 34194
    },
    {
      "epoch": 0.9852763210972166,
      "grad_norm": 0.7249597311019897,
      "learning_rate": 5.684179892168074e-09,
      "loss": 0.6715,
      "step": 34195
    },
    {
      "epoch": 0.9853051345588659,
      "grad_norm": 0.7172713875770569,
      "learning_rate": 5.661958576738125e-09,
      "loss": 0.6659,
      "step": 34196
    },
    {
      "epoch": 0.9853339480205152,
      "grad_norm": 0.7138645052909851,
      "learning_rate": 5.6397807569996685e-09,
      "loss": 0.6837,
      "step": 34197
    },
    {
      "epoch": 0.9853627614821645,
      "grad_norm": 0.7159733176231384,
      "learning_rate": 5.61764643314644e-09,
      "loss": 0.6688,
      "step": 34198
    },
    {
      "epoch": 0.9853915749438138,
      "grad_norm": 0.7548254132270813,
      "learning_rate": 5.5955556053710615e-09,
      "loss": 0.6864,
      "step": 34199
    },
    {
      "epoch": 0.985420388405463,
      "grad_norm": 0.7257395386695862,
      "learning_rate": 5.573508273866157e-09,
      "loss": 0.6726,
      "step": 34200
    },
    {
      "epoch": 0.9854492018671123,
      "grad_norm": 0.7250145077705383,
      "learning_rate": 5.5515044388226855e-09,
      "loss": 0.6715,
      "step": 34201
    },
    {
      "epoch": 0.9854780153287616,
      "grad_norm": 0.7168300747871399,
      "learning_rate": 5.5295441004332706e-09,
      "loss": 0.666,
      "step": 34202
    },
    {
      "epoch": 0.9855068287904108,
      "grad_norm": 0.7174007892608643,
      "learning_rate": 5.50762725888887e-09,
      "loss": 0.6634,
      "step": 34203
    },
    {
      "epoch": 0.9855356422520601,
      "grad_norm": 0.7351424098014832,
      "learning_rate": 5.485753914380443e-09,
      "loss": 0.6769,
      "step": 34204
    },
    {
      "epoch": 0.9855644557137094,
      "grad_norm": 0.7250854969024658,
      "learning_rate": 5.4639240670983915e-09,
      "loss": 0.6526,
      "step": 34205
    },
    {
      "epoch": 0.9855932691753587,
      "grad_norm": 0.7142854928970337,
      "learning_rate": 5.44213771723312e-09,
      "loss": 0.6627,
      "step": 34206
    },
    {
      "epoch": 0.985622082637008,
      "grad_norm": 0.7296443581581116,
      "learning_rate": 5.420394864973921e-09,
      "loss": 0.6773,
      "step": 34207
    },
    {
      "epoch": 0.9856508960986573,
      "grad_norm": 0.7190730571746826,
      "learning_rate": 5.398695510510088e-09,
      "loss": 0.671,
      "step": 34208
    },
    {
      "epoch": 0.9856797095603066,
      "grad_norm": 0.7243430614471436,
      "learning_rate": 5.377039654030913e-09,
      "loss": 0.7092,
      "step": 34209
    },
    {
      "epoch": 0.9857085230219559,
      "grad_norm": 0.7333064675331116,
      "learning_rate": 5.355427295725135e-09,
      "loss": 0.6623,
      "step": 34210
    },
    {
      "epoch": 0.9857373364836052,
      "grad_norm": 0.7125558853149414,
      "learning_rate": 5.333858435780936e-09,
      "loss": 0.6727,
      "step": 34211
    },
    {
      "epoch": 0.9857661499452545,
      "grad_norm": 0.7368280291557312,
      "learning_rate": 5.31233307438539e-09,
      "loss": 0.6778,
      "step": 34212
    },
    {
      "epoch": 0.9857949634069038,
      "grad_norm": 0.7514309287071228,
      "learning_rate": 5.290851211727233e-09,
      "loss": 0.6807,
      "step": 34213
    },
    {
      "epoch": 0.9858237768685529,
      "grad_norm": 0.7106051445007324,
      "learning_rate": 5.269412847992428e-09,
      "loss": 0.6902,
      "step": 34214
    },
    {
      "epoch": 0.9858525903302022,
      "grad_norm": 0.7472414970397949,
      "learning_rate": 5.248017983368048e-09,
      "loss": 0.6816,
      "step": 34215
    },
    {
      "epoch": 0.9858814037918515,
      "grad_norm": 0.7150521874427795,
      "learning_rate": 5.22666661804061e-09,
      "loss": 0.6662,
      "step": 34216
    },
    {
      "epoch": 0.9859102172535008,
      "grad_norm": 0.7047604322433472,
      "learning_rate": 5.205358752196077e-09,
      "loss": 0.6447,
      "step": 34217
    },
    {
      "epoch": 0.9859390307151501,
      "grad_norm": 0.7297126650810242,
      "learning_rate": 5.184094386019856e-09,
      "loss": 0.6814,
      "step": 34218
    },
    {
      "epoch": 0.9859678441767994,
      "grad_norm": 0.704267680644989,
      "learning_rate": 5.162873519697353e-09,
      "loss": 0.677,
      "step": 34219
    },
    {
      "epoch": 0.9859966576384487,
      "grad_norm": 0.7093814611434937,
      "learning_rate": 5.1416961534128675e-09,
      "loss": 0.6664,
      "step": 34220
    },
    {
      "epoch": 0.986025471100098,
      "grad_norm": 0.8159517645835876,
      "learning_rate": 5.120562287351805e-09,
      "loss": 0.6779,
      "step": 34221
    },
    {
      "epoch": 0.9860542845617473,
      "grad_norm": 0.7218825221061707,
      "learning_rate": 5.099471921696797e-09,
      "loss": 0.6675,
      "step": 34222
    },
    {
      "epoch": 0.9860830980233966,
      "grad_norm": 0.7015059590339661,
      "learning_rate": 5.078425056632696e-09,
      "loss": 0.6897,
      "step": 34223
    },
    {
      "epoch": 0.9861119114850458,
      "grad_norm": 0.7189076542854309,
      "learning_rate": 5.05742169234269e-09,
      "loss": 0.6834,
      "step": 34224
    },
    {
      "epoch": 0.9861407249466951,
      "grad_norm": 0.7148973345756531,
      "learning_rate": 5.036461829008854e-09,
      "loss": 0.6704,
      "step": 34225
    },
    {
      "epoch": 0.9861695384083444,
      "grad_norm": 0.7083612680435181,
      "learning_rate": 5.015545466814931e-09,
      "loss": 0.6701,
      "step": 34226
    },
    {
      "epoch": 0.9861983518699936,
      "grad_norm": 0.7268385291099548,
      "learning_rate": 4.994672605941886e-09,
      "loss": 0.6729,
      "step": 34227
    },
    {
      "epoch": 0.9862271653316429,
      "grad_norm": 0.7306458950042725,
      "learning_rate": 4.973843246572352e-09,
      "loss": 0.677,
      "step": 34228
    },
    {
      "epoch": 0.9862559787932922,
      "grad_norm": 0.7154067754745483,
      "learning_rate": 4.953057388887295e-09,
      "loss": 0.667,
      "step": 34229
    },
    {
      "epoch": 0.9862847922549415,
      "grad_norm": 0.7199912667274475,
      "learning_rate": 4.932315033068236e-09,
      "loss": 0.6849,
      "step": 34230
    },
    {
      "epoch": 0.9863136057165908,
      "grad_norm": 0.7305731177330017,
      "learning_rate": 4.911616179295031e-09,
      "loss": 0.6766,
      "step": 34231
    },
    {
      "epoch": 0.9863424191782401,
      "grad_norm": 0.7339119911193848,
      "learning_rate": 4.890960827748648e-09,
      "loss": 0.6808,
      "step": 34232
    },
    {
      "epoch": 0.9863712326398894,
      "grad_norm": 0.7363855838775635,
      "learning_rate": 4.870348978608386e-09,
      "loss": 0.6885,
      "step": 34233
    },
    {
      "epoch": 0.9864000461015386,
      "grad_norm": 0.7390724420547485,
      "learning_rate": 4.849780632054657e-09,
      "loss": 0.6811,
      "step": 34234
    },
    {
      "epoch": 0.9864288595631879,
      "grad_norm": 0.7572529911994934,
      "learning_rate": 4.8292557882656525e-09,
      "loss": 0.6874,
      "step": 34235
    },
    {
      "epoch": 0.9864576730248372,
      "grad_norm": 0.7261450886726379,
      "learning_rate": 4.808774447420117e-09,
      "loss": 0.692,
      "step": 34236
    },
    {
      "epoch": 0.9864864864864865,
      "grad_norm": 0.7386103868484497,
      "learning_rate": 4.788336609697353e-09,
      "loss": 0.6658,
      "step": 34237
    },
    {
      "epoch": 0.9865152999481358,
      "grad_norm": 0.7380800843238831,
      "learning_rate": 4.767942275274995e-09,
      "loss": 0.6778,
      "step": 34238
    },
    {
      "epoch": 0.9865441134097851,
      "grad_norm": 0.7266929149627686,
      "learning_rate": 4.747591444329569e-09,
      "loss": 0.6958,
      "step": 34239
    },
    {
      "epoch": 0.9865729268714344,
      "grad_norm": 0.7349887490272522,
      "learning_rate": 4.727284117039821e-09,
      "loss": 0.6947,
      "step": 34240
    },
    {
      "epoch": 0.9866017403330836,
      "grad_norm": 0.7105131149291992,
      "learning_rate": 4.7070202935822765e-09,
      "loss": 0.6466,
      "step": 34241
    },
    {
      "epoch": 0.9866305537947329,
      "grad_norm": 0.7311797142028809,
      "learning_rate": 4.68679997413235e-09,
      "loss": 0.6921,
      "step": 34242
    },
    {
      "epoch": 0.9866593672563821,
      "grad_norm": 0.7099502086639404,
      "learning_rate": 4.6666231588676784e-09,
      "loss": 0.6737,
      "step": 34243
    },
    {
      "epoch": 0.9866881807180314,
      "grad_norm": 0.7224089503288269,
      "learning_rate": 4.646489847962565e-09,
      "loss": 0.6876,
      "step": 34244
    },
    {
      "epoch": 0.9867169941796807,
      "grad_norm": 0.7352033257484436,
      "learning_rate": 4.626400041593538e-09,
      "loss": 0.6862,
      "step": 34245
    },
    {
      "epoch": 0.98674580764133,
      "grad_norm": 0.753876805305481,
      "learning_rate": 4.606353739934344e-09,
      "loss": 0.7229,
      "step": 34246
    },
    {
      "epoch": 0.9867746211029793,
      "grad_norm": 0.7477352023124695,
      "learning_rate": 4.586350943160956e-09,
      "loss": 0.6899,
      "step": 34247
    },
    {
      "epoch": 0.9868034345646286,
      "grad_norm": 0.7221672534942627,
      "learning_rate": 4.566391651446567e-09,
      "loss": 0.6858,
      "step": 34248
    },
    {
      "epoch": 0.9868322480262779,
      "grad_norm": 0.7179524302482605,
      "learning_rate": 4.5464758649649274e-09,
      "loss": 0.6815,
      "step": 34249
    },
    {
      "epoch": 0.9868610614879272,
      "grad_norm": 0.746958315372467,
      "learning_rate": 4.526603583890343e-09,
      "loss": 0.6657,
      "step": 34250
    },
    {
      "epoch": 0.9868898749495765,
      "grad_norm": 0.7313018441200256,
      "learning_rate": 4.5067748083954535e-09,
      "loss": 0.6989,
      "step": 34251
    },
    {
      "epoch": 0.9869186884112258,
      "grad_norm": 0.7126665711402893,
      "learning_rate": 4.4869895386528975e-09,
      "loss": 0.673,
      "step": 34252
    },
    {
      "epoch": 0.986947501872875,
      "grad_norm": 0.7106180191040039,
      "learning_rate": 4.4672477748347595e-09,
      "loss": 0.6757,
      "step": 34253
    },
    {
      "epoch": 0.9869763153345242,
      "grad_norm": 0.7310202717781067,
      "learning_rate": 4.447549517113126e-09,
      "loss": 0.6846,
      "step": 34254
    },
    {
      "epoch": 0.9870051287961735,
      "grad_norm": 0.7242233753204346,
      "learning_rate": 4.427894765659524e-09,
      "loss": 0.6909,
      "step": 34255
    },
    {
      "epoch": 0.9870339422578228,
      "grad_norm": 0.7138737440109253,
      "learning_rate": 4.408283520645484e-09,
      "loss": 0.6814,
      "step": 34256
    },
    {
      "epoch": 0.9870627557194721,
      "grad_norm": 0.718786358833313,
      "learning_rate": 4.3887157822414265e-09,
      "loss": 0.6817,
      "step": 34257
    },
    {
      "epoch": 0.9870915691811214,
      "grad_norm": 0.7384078502655029,
      "learning_rate": 4.369191550617768e-09,
      "loss": 0.6923,
      "step": 34258
    },
    {
      "epoch": 0.9871203826427707,
      "grad_norm": 0.7331740260124207,
      "learning_rate": 4.349710825944931e-09,
      "loss": 0.7001,
      "step": 34259
    },
    {
      "epoch": 0.98714919610442,
      "grad_norm": 0.7330279350280762,
      "learning_rate": 4.330273608392221e-09,
      "loss": 0.705,
      "step": 34260
    },
    {
      "epoch": 0.9871780095660693,
      "grad_norm": 0.7254275679588318,
      "learning_rate": 4.31087989812895e-09,
      "loss": 0.6845,
      "step": 34261
    },
    {
      "epoch": 0.9872068230277186,
      "grad_norm": 0.7245600819587708,
      "learning_rate": 4.2915296953238705e-09,
      "loss": 0.6886,
      "step": 34262
    },
    {
      "epoch": 0.9872356364893679,
      "grad_norm": 0.7267093062400818,
      "learning_rate": 4.272223000145737e-09,
      "loss": 0.6874,
      "step": 34263
    },
    {
      "epoch": 0.9872644499510171,
      "grad_norm": 0.7234851717948914,
      "learning_rate": 4.252959812762747e-09,
      "loss": 0.6657,
      "step": 34264
    },
    {
      "epoch": 0.9872932634126664,
      "grad_norm": 0.715837299823761,
      "learning_rate": 4.233740133342545e-09,
      "loss": 0.6744,
      "step": 34265
    },
    {
      "epoch": 0.9873220768743157,
      "grad_norm": 0.7158999443054199,
      "learning_rate": 4.214563962052221e-09,
      "loss": 0.6947,
      "step": 34266
    },
    {
      "epoch": 0.987350890335965,
      "grad_norm": 0.7145724296569824,
      "learning_rate": 4.195431299059416e-09,
      "loss": 0.6623,
      "step": 34267
    },
    {
      "epoch": 0.9873797037976142,
      "grad_norm": 0.7110761404037476,
      "learning_rate": 4.176342144530665e-09,
      "loss": 0.6733,
      "step": 34268
    },
    {
      "epoch": 0.9874085172592635,
      "grad_norm": 0.7200140953063965,
      "learning_rate": 4.1572964986319465e-09,
      "loss": 0.6786,
      "step": 34269
    },
    {
      "epoch": 0.9874373307209128,
      "grad_norm": 0.7236116528511047,
      "learning_rate": 4.138294361528683e-09,
      "loss": 0.6712,
      "step": 34270
    },
    {
      "epoch": 0.9874661441825621,
      "grad_norm": 0.743343710899353,
      "learning_rate": 4.119335733386853e-09,
      "loss": 0.6834,
      "step": 34271
    },
    {
      "epoch": 0.9874949576442114,
      "grad_norm": 0.7286434173583984,
      "learning_rate": 4.1004206143718805e-09,
      "loss": 0.6886,
      "step": 34272
    },
    {
      "epoch": 0.9875237711058606,
      "grad_norm": 0.7249632477760315,
      "learning_rate": 4.0815490046480775e-09,
      "loss": 0.6785,
      "step": 34273
    },
    {
      "epoch": 0.9875525845675099,
      "grad_norm": 0.7197278738021851,
      "learning_rate": 4.062720904379758e-09,
      "loss": 0.6594,
      "step": 34274
    },
    {
      "epoch": 0.9875813980291592,
      "grad_norm": 0.7141082882881165,
      "learning_rate": 4.043936313731234e-09,
      "loss": 0.6709,
      "step": 34275
    },
    {
      "epoch": 0.9876102114908085,
      "grad_norm": 0.73548424243927,
      "learning_rate": 4.025195232865709e-09,
      "loss": 0.6943,
      "step": 34276
    },
    {
      "epoch": 0.9876390249524578,
      "grad_norm": 0.7349528074264526,
      "learning_rate": 4.00649766194694e-09,
      "loss": 0.6891,
      "step": 34277
    },
    {
      "epoch": 0.9876678384141071,
      "grad_norm": 0.7430084943771362,
      "learning_rate": 3.987843601137021e-09,
      "loss": 0.7108,
      "step": 34278
    },
    {
      "epoch": 0.9876966518757564,
      "grad_norm": 0.7121701240539551,
      "learning_rate": 3.969233050599153e-09,
      "loss": 0.6761,
      "step": 34279
    },
    {
      "epoch": 0.9877254653374057,
      "grad_norm": 0.7246484756469727,
      "learning_rate": 3.950666010494875e-09,
      "loss": 0.6763,
      "step": 34280
    },
    {
      "epoch": 0.9877542787990549,
      "grad_norm": 0.6993350386619568,
      "learning_rate": 3.932142480985723e-09,
      "loss": 0.6663,
      "step": 34281
    },
    {
      "epoch": 0.9877830922607042,
      "grad_norm": 0.8679801225662231,
      "learning_rate": 3.9136624622337914e-09,
      "loss": 0.6899,
      "step": 34282
    },
    {
      "epoch": 0.9878119057223534,
      "grad_norm": 0.7131001353263855,
      "learning_rate": 3.895225954399506e-09,
      "loss": 0.6932,
      "step": 34283
    },
    {
      "epoch": 0.9878407191840027,
      "grad_norm": 0.7022652626037598,
      "learning_rate": 3.87683295764385e-09,
      "loss": 0.6818,
      "step": 34284
    },
    {
      "epoch": 0.987869532645652,
      "grad_norm": 0.7154098749160767,
      "learning_rate": 3.85848347212614e-09,
      "loss": 0.6811,
      "step": 34285
    },
    {
      "epoch": 0.9878983461073013,
      "grad_norm": 0.6933647990226746,
      "learning_rate": 3.840177498006803e-09,
      "loss": 0.6284,
      "step": 34286
    },
    {
      "epoch": 0.9879271595689506,
      "grad_norm": 0.7329391837120056,
      "learning_rate": 3.821915035445711e-09,
      "loss": 0.6896,
      "step": 34287
    },
    {
      "epoch": 0.9879559730305999,
      "grad_norm": 0.7163298726081848,
      "learning_rate": 3.803696084601072e-09,
      "loss": 0.6709,
      "step": 34288
    },
    {
      "epoch": 0.9879847864922492,
      "grad_norm": 0.7314774394035339,
      "learning_rate": 3.785520645631646e-09,
      "loss": 0.6673,
      "step": 34289
    },
    {
      "epoch": 0.9880135999538985,
      "grad_norm": 0.733492374420166,
      "learning_rate": 3.7673887186961965e-09,
      "loss": 0.7033,
      "step": 34290
    },
    {
      "epoch": 0.9880424134155478,
      "grad_norm": 0.7129850387573242,
      "learning_rate": 3.749300303952375e-09,
      "loss": 0.69,
      "step": 34291
    },
    {
      "epoch": 0.9880712268771971,
      "grad_norm": 0.7200227975845337,
      "learning_rate": 3.731255401557277e-09,
      "loss": 0.6851,
      "step": 34292
    },
    {
      "epoch": 0.9881000403388464,
      "grad_norm": 0.7457100749015808,
      "learning_rate": 3.7132540116691095e-09,
      "loss": 0.7071,
      "step": 34293
    },
    {
      "epoch": 0.9881288538004956,
      "grad_norm": 0.7097764015197754,
      "learning_rate": 3.6952961344438597e-09,
      "loss": 0.6751,
      "step": 34294
    },
    {
      "epoch": 0.9881576672621448,
      "grad_norm": 0.711920976638794,
      "learning_rate": 3.6773817700380687e-09,
      "loss": 0.6785,
      "step": 34295
    },
    {
      "epoch": 0.9881864807237941,
      "grad_norm": 0.7151310443878174,
      "learning_rate": 3.6595109186077226e-09,
      "loss": 0.6914,
      "step": 34296
    },
    {
      "epoch": 0.9882152941854434,
      "grad_norm": 0.7292062640190125,
      "learning_rate": 3.641683580308253e-09,
      "loss": 0.6992,
      "step": 34297
    },
    {
      "epoch": 0.9882441076470927,
      "grad_norm": 0.7229759097099304,
      "learning_rate": 3.6238997552956456e-09,
      "loss": 0.6742,
      "step": 34298
    },
    {
      "epoch": 0.988272921108742,
      "grad_norm": 0.7481561899185181,
      "learning_rate": 3.606159443724222e-09,
      "loss": 0.6796,
      "step": 34299
    },
    {
      "epoch": 0.9883017345703913,
      "grad_norm": 0.7267009019851685,
      "learning_rate": 3.588462645748858e-09,
      "loss": 0.6954,
      "step": 34300
    },
    {
      "epoch": 0.9883305480320406,
      "grad_norm": 0.7504401206970215,
      "learning_rate": 3.5708093615227647e-09,
      "loss": 0.6801,
      "step": 34301
    },
    {
      "epoch": 0.9883593614936899,
      "grad_norm": 1.5968221426010132,
      "learning_rate": 3.553199591200818e-09,
      "loss": 0.6787,
      "step": 34302
    },
    {
      "epoch": 0.9883881749553391,
      "grad_norm": 0.7427374720573425,
      "learning_rate": 3.5356333349356733e-09,
      "loss": 0.6946,
      "step": 34303
    },
    {
      "epoch": 0.9884169884169884,
      "grad_norm": 0.6959028840065002,
      "learning_rate": 3.5181105928805414e-09,
      "loss": 0.6729,
      "step": 34304
    },
    {
      "epoch": 0.9884458018786377,
      "grad_norm": 0.7091915607452393,
      "learning_rate": 3.5006313651880785e-09,
      "loss": 0.6593,
      "step": 34305
    },
    {
      "epoch": 0.988474615340287,
      "grad_norm": 0.7349546551704407,
      "learning_rate": 3.4831956520103847e-09,
      "loss": 0.6744,
      "step": 34306
    },
    {
      "epoch": 0.9885034288019363,
      "grad_norm": 0.7155954837799072,
      "learning_rate": 3.465803453499561e-09,
      "loss": 0.6798,
      "step": 34307
    },
    {
      "epoch": 0.9885322422635855,
      "grad_norm": 0.7027056217193604,
      "learning_rate": 3.4484547698065973e-09,
      "loss": 0.66,
      "step": 34308
    },
    {
      "epoch": 0.9885610557252348,
      "grad_norm": 0.7384201884269714,
      "learning_rate": 3.431149601083039e-09,
      "loss": 0.6648,
      "step": 34309
    },
    {
      "epoch": 0.9885898691868841,
      "grad_norm": 0.7098839282989502,
      "learning_rate": 3.413887947479322e-09,
      "loss": 0.6754,
      "step": 34310
    },
    {
      "epoch": 0.9886186826485334,
      "grad_norm": 0.7095677852630615,
      "learning_rate": 3.3966698091458804e-09,
      "loss": 0.6812,
      "step": 34311
    },
    {
      "epoch": 0.9886474961101827,
      "grad_norm": 0.7524491548538208,
      "learning_rate": 3.379495186232595e-09,
      "loss": 0.6968,
      "step": 34312
    },
    {
      "epoch": 0.988676309571832,
      "grad_norm": 0.746586799621582,
      "learning_rate": 3.362364078889346e-09,
      "loss": 0.6879,
      "step": 34313
    },
    {
      "epoch": 0.9887051230334812,
      "grad_norm": 0.741540253162384,
      "learning_rate": 3.3452764872649035e-09,
      "loss": 0.6978,
      "step": 34314
    },
    {
      "epoch": 0.9887339364951305,
      "grad_norm": 0.6956606507301331,
      "learning_rate": 3.3282324115085917e-09,
      "loss": 0.6489,
      "step": 34315
    },
    {
      "epoch": 0.9887627499567798,
      "grad_norm": 0.7351986169815063,
      "learning_rate": 3.311231851768626e-09,
      "loss": 0.6899,
      "step": 34316
    },
    {
      "epoch": 0.9887915634184291,
      "grad_norm": 0.7374492883682251,
      "learning_rate": 3.2942748081921105e-09,
      "loss": 0.6634,
      "step": 34317
    },
    {
      "epoch": 0.9888203768800784,
      "grad_norm": 0.6953107714653015,
      "learning_rate": 3.2773612809283707e-09,
      "loss": 0.6778,
      "step": 34318
    },
    {
      "epoch": 0.9888491903417277,
      "grad_norm": 0.7087225914001465,
      "learning_rate": 3.2604912701234003e-09,
      "loss": 0.659,
      "step": 34319
    },
    {
      "epoch": 0.988878003803377,
      "grad_norm": 0.7273982167243958,
      "learning_rate": 3.2436647759254147e-09,
      "loss": 0.6877,
      "step": 34320
    },
    {
      "epoch": 0.9889068172650263,
      "grad_norm": 0.7274070978164673,
      "learning_rate": 3.2268817984792976e-09,
      "loss": 0.6861,
      "step": 34321
    },
    {
      "epoch": 0.9889356307266755,
      "grad_norm": 0.7203271985054016,
      "learning_rate": 3.210142337932709e-09,
      "loss": 0.6925,
      "step": 34322
    },
    {
      "epoch": 0.9889644441883247,
      "grad_norm": 0.7074299454689026,
      "learning_rate": 3.1934463944305326e-09,
      "loss": 0.6718,
      "step": 34323
    },
    {
      "epoch": 0.988993257649974,
      "grad_norm": 0.7143237590789795,
      "learning_rate": 3.1767939681187632e-09,
      "loss": 0.667,
      "step": 34324
    },
    {
      "epoch": 0.9890220711116233,
      "grad_norm": 0.7112151384353638,
      "learning_rate": 3.1601850591417293e-09,
      "loss": 0.6685,
      "step": 34325
    },
    {
      "epoch": 0.9890508845732726,
      "grad_norm": 0.7180307507514954,
      "learning_rate": 3.1436196676448703e-09,
      "loss": 0.6511,
      "step": 34326
    },
    {
      "epoch": 0.9890796980349219,
      "grad_norm": 0.7228859066963196,
      "learning_rate": 3.1270977937714055e-09,
      "loss": 0.6744,
      "step": 34327
    },
    {
      "epoch": 0.9891085114965712,
      "grad_norm": 0.7825208902359009,
      "learning_rate": 3.110619437666218e-09,
      "loss": 0.6711,
      "step": 34328
    },
    {
      "epoch": 0.9891373249582205,
      "grad_norm": 0.7286614179611206,
      "learning_rate": 3.094184599472527e-09,
      "loss": 0.6591,
      "step": 34329
    },
    {
      "epoch": 0.9891661384198698,
      "grad_norm": 0.7454304695129395,
      "learning_rate": 3.077793279333552e-09,
      "loss": 0.67,
      "step": 34330
    },
    {
      "epoch": 0.9891949518815191,
      "grad_norm": 0.731380820274353,
      "learning_rate": 3.0614454773919555e-09,
      "loss": 0.6524,
      "step": 34331
    },
    {
      "epoch": 0.9892237653431684,
      "grad_norm": 0.6930665969848633,
      "learning_rate": 3.045141193789847e-09,
      "loss": 0.6603,
      "step": 34332
    },
    {
      "epoch": 0.9892525788048177,
      "grad_norm": 0.7301313877105713,
      "learning_rate": 3.028880428669334e-09,
      "loss": 0.6644,
      "step": 34333
    },
    {
      "epoch": 0.9892813922664669,
      "grad_norm": 0.7535630464553833,
      "learning_rate": 3.012663182171971e-09,
      "loss": 0.6995,
      "step": 34334
    },
    {
      "epoch": 0.9893102057281161,
      "grad_norm": 0.7380533218383789,
      "learning_rate": 2.996489454439866e-09,
      "loss": 0.675,
      "step": 34335
    },
    {
      "epoch": 0.9893390191897654,
      "grad_norm": 0.7228555679321289,
      "learning_rate": 2.980359245612352e-09,
      "loss": 0.651,
      "step": 34336
    },
    {
      "epoch": 0.9893678326514147,
      "grad_norm": 0.7103008031845093,
      "learning_rate": 2.964272555831538e-09,
      "loss": 0.6945,
      "step": 34337
    },
    {
      "epoch": 0.989396646113064,
      "grad_norm": 0.7140816450119019,
      "learning_rate": 2.948229385236201e-09,
      "loss": 0.6806,
      "step": 34338
    },
    {
      "epoch": 0.9894254595747133,
      "grad_norm": 0.7232003211975098,
      "learning_rate": 2.932229733966785e-09,
      "loss": 0.6927,
      "step": 34339
    },
    {
      "epoch": 0.9894542730363626,
      "grad_norm": 0.735901415348053,
      "learning_rate": 2.9162736021620675e-09,
      "loss": 0.7005,
      "step": 34340
    },
    {
      "epoch": 0.9894830864980119,
      "grad_norm": 0.7209230661392212,
      "learning_rate": 2.9003609899613816e-09,
      "loss": 0.6641,
      "step": 34341
    },
    {
      "epoch": 0.9895118999596612,
      "grad_norm": 0.7175900340080261,
      "learning_rate": 2.8844918975035052e-09,
      "loss": 0.6908,
      "step": 34342
    },
    {
      "epoch": 0.9895407134213104,
      "grad_norm": 0.7019049525260925,
      "learning_rate": 2.868666324926106e-09,
      "loss": 0.6561,
      "step": 34343
    },
    {
      "epoch": 0.9895695268829597,
      "grad_norm": 0.7075583934783936,
      "learning_rate": 2.8528842723679616e-09,
      "loss": 0.6765,
      "step": 34344
    },
    {
      "epoch": 0.989598340344609,
      "grad_norm": 0.7231972217559814,
      "learning_rate": 2.8371457399650747e-09,
      "loss": 0.6851,
      "step": 34345
    },
    {
      "epoch": 0.9896271538062583,
      "grad_norm": 0.7430007457733154,
      "learning_rate": 2.821450727855668e-09,
      "loss": 0.6826,
      "step": 34346
    },
    {
      "epoch": 0.9896559672679076,
      "grad_norm": 0.7154543399810791,
      "learning_rate": 2.8057992361762988e-09,
      "loss": 0.6668,
      "step": 34347
    },
    {
      "epoch": 0.9896847807295569,
      "grad_norm": 0.7282072305679321,
      "learning_rate": 2.790191265062969e-09,
      "loss": 0.681,
      "step": 34348
    },
    {
      "epoch": 0.9897135941912061,
      "grad_norm": 0.7531213760375977,
      "learning_rate": 2.774626814651682e-09,
      "loss": 0.707,
      "step": 34349
    },
    {
      "epoch": 0.9897424076528554,
      "grad_norm": 0.7132784128189087,
      "learning_rate": 2.759105885077884e-09,
      "loss": 0.6702,
      "step": 34350
    },
    {
      "epoch": 0.9897712211145047,
      "grad_norm": 0.6969124674797058,
      "learning_rate": 2.7436284764770227e-09,
      "loss": 0.6892,
      "step": 34351
    },
    {
      "epoch": 0.989800034576154,
      "grad_norm": 0.7201002836227417,
      "learning_rate": 2.728194588983435e-09,
      "loss": 0.6734,
      "step": 34352
    },
    {
      "epoch": 0.9898288480378032,
      "grad_norm": 0.71903395652771,
      "learning_rate": 2.7128042227320127e-09,
      "loss": 0.6965,
      "step": 34353
    },
    {
      "epoch": 0.9898576614994525,
      "grad_norm": 0.7079131603240967,
      "learning_rate": 2.6974573778565383e-09,
      "loss": 0.6718,
      "step": 34354
    },
    {
      "epoch": 0.9898864749611018,
      "grad_norm": 0.7412815690040588,
      "learning_rate": 2.6821540544907933e-09,
      "loss": 0.6964,
      "step": 34355
    },
    {
      "epoch": 0.9899152884227511,
      "grad_norm": 0.7346335053443909,
      "learning_rate": 2.666894252768004e-09,
      "loss": 0.6958,
      "step": 34356
    },
    {
      "epoch": 0.9899441018844004,
      "grad_norm": 0.7019926905632019,
      "learning_rate": 2.651677972820843e-09,
      "loss": 0.6541,
      "step": 34357
    },
    {
      "epoch": 0.9899729153460497,
      "grad_norm": 0.7188987731933594,
      "learning_rate": 2.6365052147825366e-09,
      "loss": 0.6875,
      "step": 34358
    },
    {
      "epoch": 0.990001728807699,
      "grad_norm": 0.7390496134757996,
      "learning_rate": 2.621375978784091e-09,
      "loss": 0.6573,
      "step": 34359
    },
    {
      "epoch": 0.9900305422693483,
      "grad_norm": 0.7131441235542297,
      "learning_rate": 2.606290264958733e-09,
      "loss": 0.692,
      "step": 34360
    },
    {
      "epoch": 0.9900593557309976,
      "grad_norm": 0.7214800715446472,
      "learning_rate": 2.591248073436359e-09,
      "loss": 0.6906,
      "step": 34361
    },
    {
      "epoch": 0.9900881691926467,
      "grad_norm": 0.7521846890449524,
      "learning_rate": 2.5762494043485297e-09,
      "loss": 0.6737,
      "step": 34362
    },
    {
      "epoch": 0.990116982654296,
      "grad_norm": 0.7294525504112244,
      "learning_rate": 2.5612942578262524e-09,
      "loss": 0.658,
      "step": 34363
    },
    {
      "epoch": 0.9901457961159453,
      "grad_norm": 0.7052424550056458,
      "learning_rate": 2.5463826339999774e-09,
      "loss": 0.6551,
      "step": 34364
    },
    {
      "epoch": 0.9901746095775946,
      "grad_norm": 0.7288758754730225,
      "learning_rate": 2.5315145329984915e-09,
      "loss": 0.6777,
      "step": 34365
    },
    {
      "epoch": 0.9902034230392439,
      "grad_norm": 0.7306530475616455,
      "learning_rate": 2.51668995495169e-09,
      "loss": 0.7128,
      "step": 34366
    },
    {
      "epoch": 0.9902322365008932,
      "grad_norm": 0.7121623754501343,
      "learning_rate": 2.5019088999894693e-09,
      "loss": 0.6832,
      "step": 34367
    },
    {
      "epoch": 0.9902610499625425,
      "grad_norm": 0.7549595236778259,
      "learning_rate": 2.4871713682395048e-09,
      "loss": 0.6851,
      "step": 34368
    },
    {
      "epoch": 0.9902898634241918,
      "grad_norm": 0.732154905796051,
      "learning_rate": 2.472477359830583e-09,
      "loss": 0.6801,
      "step": 34369
    },
    {
      "epoch": 0.9903186768858411,
      "grad_norm": 0.7263842225074768,
      "learning_rate": 2.457826874890934e-09,
      "loss": 0.6748,
      "step": 34370
    },
    {
      "epoch": 0.9903474903474904,
      "grad_norm": 0.7292749285697937,
      "learning_rate": 2.443219913548234e-09,
      "loss": 0.6899,
      "step": 34371
    },
    {
      "epoch": 0.9903763038091397,
      "grad_norm": 0.7110562920570374,
      "learning_rate": 2.4286564759284923e-09,
      "loss": 0.6512,
      "step": 34372
    },
    {
      "epoch": 0.990405117270789,
      "grad_norm": 0.7100409269332886,
      "learning_rate": 2.414136562159941e-09,
      "loss": 0.6601,
      "step": 34373
    },
    {
      "epoch": 0.9904339307324382,
      "grad_norm": 0.6980181932449341,
      "learning_rate": 2.3996601723685896e-09,
      "loss": 0.6771,
      "step": 34374
    },
    {
      "epoch": 0.9904627441940875,
      "grad_norm": 0.7159878611564636,
      "learning_rate": 2.3852273066804488e-09,
      "loss": 0.6839,
      "step": 34375
    },
    {
      "epoch": 0.9904915576557367,
      "grad_norm": 0.723852276802063,
      "learning_rate": 2.3708379652209734e-09,
      "loss": 0.6651,
      "step": 34376
    },
    {
      "epoch": 0.990520371117386,
      "grad_norm": 0.7186260223388672,
      "learning_rate": 2.3564921481156187e-09,
      "loss": 0.6663,
      "step": 34377
    },
    {
      "epoch": 0.9905491845790353,
      "grad_norm": 0.7341119647026062,
      "learning_rate": 2.3421898554892852e-09,
      "loss": 0.6793,
      "step": 34378
    },
    {
      "epoch": 0.9905779980406846,
      "grad_norm": 0.7173287272453308,
      "learning_rate": 2.3279310874668726e-09,
      "loss": 0.6878,
      "step": 34379
    },
    {
      "epoch": 0.9906068115023339,
      "grad_norm": 0.7122073769569397,
      "learning_rate": 2.3137158441721706e-09,
      "loss": 0.6861,
      "step": 34380
    },
    {
      "epoch": 0.9906356249639832,
      "grad_norm": 0.7206083536148071,
      "learning_rate": 2.2995441257295247e-09,
      "loss": 0.6633,
      "step": 34381
    },
    {
      "epoch": 0.9906644384256325,
      "grad_norm": 0.7259880304336548,
      "learning_rate": 2.2854159322616144e-09,
      "loss": 0.6633,
      "step": 34382
    },
    {
      "epoch": 0.9906932518872817,
      "grad_norm": 0.7144859433174133,
      "learning_rate": 2.271331263892229e-09,
      "loss": 0.6819,
      "step": 34383
    },
    {
      "epoch": 0.990722065348931,
      "grad_norm": 0.7331462502479553,
      "learning_rate": 2.257290120743494e-09,
      "loss": 0.6842,
      "step": 34384
    },
    {
      "epoch": 0.9907508788105803,
      "grad_norm": 0.752197802066803,
      "learning_rate": 2.243292502937533e-09,
      "loss": 0.6872,
      "step": 34385
    },
    {
      "epoch": 0.9907796922722296,
      "grad_norm": 0.692468523979187,
      "learning_rate": 2.229338410597026e-09,
      "loss": 0.654,
      "step": 34386
    },
    {
      "epoch": 0.9908085057338789,
      "grad_norm": 0.7164486050605774,
      "learning_rate": 2.2154278438429876e-09,
      "loss": 0.6767,
      "step": 34387
    },
    {
      "epoch": 0.9908373191955282,
      "grad_norm": 0.7422334551811218,
      "learning_rate": 2.201560802796432e-09,
      "loss": 0.6841,
      "step": 34388
    },
    {
      "epoch": 0.9908661326571774,
      "grad_norm": 0.7100372314453125,
      "learning_rate": 2.187737287578373e-09,
      "loss": 0.6549,
      "step": 34389
    },
    {
      "epoch": 0.9908949461188267,
      "grad_norm": 0.7086052298545837,
      "learning_rate": 2.1739572983087153e-09,
      "loss": 0.6572,
      "step": 34390
    },
    {
      "epoch": 0.990923759580476,
      "grad_norm": 0.7149611115455627,
      "learning_rate": 2.160220835108473e-09,
      "loss": 0.6747,
      "step": 34391
    },
    {
      "epoch": 0.9909525730421253,
      "grad_norm": 0.7141755819320679,
      "learning_rate": 2.14652789809644e-09,
      "loss": 0.6787,
      "step": 34392
    },
    {
      "epoch": 0.9909813865037745,
      "grad_norm": 0.7265552282333374,
      "learning_rate": 2.132878487391965e-09,
      "loss": 0.6782,
      "step": 34393
    },
    {
      "epoch": 0.9910101999654238,
      "grad_norm": 0.7102692127227783,
      "learning_rate": 2.1192726031143973e-09,
      "loss": 0.6635,
      "step": 34394
    },
    {
      "epoch": 0.9910390134270731,
      "grad_norm": 0.7308573722839355,
      "learning_rate": 2.1057102453819754e-09,
      "loss": 0.7055,
      "step": 34395
    },
    {
      "epoch": 0.9910678268887224,
      "grad_norm": 0.7305545210838318,
      "learning_rate": 2.0921914143123836e-09,
      "loss": 0.689,
      "step": 34396
    },
    {
      "epoch": 0.9910966403503717,
      "grad_norm": 0.7027189135551453,
      "learning_rate": 2.0787161100238595e-09,
      "loss": 0.6915,
      "step": 34397
    },
    {
      "epoch": 0.991125453812021,
      "grad_norm": 0.7281208038330078,
      "learning_rate": 2.0652843326340878e-09,
      "loss": 0.6823,
      "step": 34398
    },
    {
      "epoch": 0.9911542672736703,
      "grad_norm": 0.7243311405181885,
      "learning_rate": 2.051896082259086e-09,
      "loss": 0.6875,
      "step": 34399
    },
    {
      "epoch": 0.9911830807353196,
      "grad_norm": 0.8113660216331482,
      "learning_rate": 2.0385513590159835e-09,
      "loss": 0.6892,
      "step": 34400
    },
    {
      "epoch": 0.9912118941969689,
      "grad_norm": 0.7271645069122314,
      "learning_rate": 2.0252501630207975e-09,
      "loss": 0.6913,
      "step": 34401
    },
    {
      "epoch": 0.9912407076586182,
      "grad_norm": 0.7303289175033569,
      "learning_rate": 2.0119924943901026e-09,
      "loss": 0.6707,
      "step": 34402
    },
    {
      "epoch": 0.9912695211202673,
      "grad_norm": 0.7100399136543274,
      "learning_rate": 1.998778353238251e-09,
      "loss": 0.6729,
      "step": 34403
    },
    {
      "epoch": 0.9912983345819166,
      "grad_norm": 0.7288795113563538,
      "learning_rate": 1.9856077396812612e-09,
      "loss": 0.6797,
      "step": 34404
    },
    {
      "epoch": 0.9913271480435659,
      "grad_norm": 0.7265059947967529,
      "learning_rate": 1.972480653832931e-09,
      "loss": 0.6737,
      "step": 34405
    },
    {
      "epoch": 0.9913559615052152,
      "grad_norm": 0.7178830504417419,
      "learning_rate": 1.959397095808724e-09,
      "loss": 0.6991,
      "step": 34406
    },
    {
      "epoch": 0.9913847749668645,
      "grad_norm": 0.7023162245750427,
      "learning_rate": 1.946357065721327e-09,
      "loss": 0.6548,
      "step": 34407
    },
    {
      "epoch": 0.9914135884285138,
      "grad_norm": 0.7420100569725037,
      "learning_rate": 1.933360563685649e-09,
      "loss": 0.6935,
      "step": 34408
    },
    {
      "epoch": 0.9914424018901631,
      "grad_norm": 0.7095827460289001,
      "learning_rate": 1.9204075898138217e-09,
      "loss": 0.6597,
      "step": 34409
    },
    {
      "epoch": 0.9914712153518124,
      "grad_norm": 0.732720136642456,
      "learning_rate": 1.9074981442185337e-09,
      "loss": 0.6796,
      "step": 34410
    },
    {
      "epoch": 0.9915000288134617,
      "grad_norm": 0.7193937301635742,
      "learning_rate": 1.8946322270135818e-09,
      "loss": 0.6875,
      "step": 34411
    },
    {
      "epoch": 0.991528842275111,
      "grad_norm": 0.712355375289917,
      "learning_rate": 1.881809838309434e-09,
      "loss": 0.6732,
      "step": 34412
    },
    {
      "epoch": 0.9915576557367602,
      "grad_norm": 0.724773108959198,
      "learning_rate": 1.8690309782187777e-09,
      "loss": 0.6952,
      "step": 34413
    },
    {
      "epoch": 0.9915864691984095,
      "grad_norm": 0.723576545715332,
      "learning_rate": 1.85629564685208e-09,
      "loss": 0.6716,
      "step": 34414
    },
    {
      "epoch": 0.9916152826600588,
      "grad_norm": 0.7099675536155701,
      "learning_rate": 1.843603844320918e-09,
      "loss": 0.6703,
      "step": 34415
    },
    {
      "epoch": 0.991644096121708,
      "grad_norm": 0.7311878204345703,
      "learning_rate": 1.8309555707357595e-09,
      "loss": 0.677,
      "step": 34416
    },
    {
      "epoch": 0.9916729095833573,
      "grad_norm": 0.7277650833129883,
      "learning_rate": 1.8183508262070715e-09,
      "loss": 0.6639,
      "step": 34417
    },
    {
      "epoch": 0.9917017230450066,
      "grad_norm": 0.717572808265686,
      "learning_rate": 1.8057896108436557e-09,
      "loss": 0.6751,
      "step": 34418
    },
    {
      "epoch": 0.9917305365066559,
      "grad_norm": 0.7082982063293457,
      "learning_rate": 1.7932719247554241e-09,
      "loss": 0.6827,
      "step": 34419
    },
    {
      "epoch": 0.9917593499683052,
      "grad_norm": 0.710599422454834,
      "learning_rate": 1.7807977680517341e-09,
      "loss": 0.6897,
      "step": 34420
    },
    {
      "epoch": 0.9917881634299545,
      "grad_norm": 0.7126237154006958,
      "learning_rate": 1.768367140840832e-09,
      "loss": 0.6627,
      "step": 34421
    },
    {
      "epoch": 0.9918169768916038,
      "grad_norm": 0.7245080471038818,
      "learning_rate": 1.7559800432309647e-09,
      "loss": 0.6733,
      "step": 34422
    },
    {
      "epoch": 0.991845790353253,
      "grad_norm": 0.7213675379753113,
      "learning_rate": 1.7436364753298241e-09,
      "loss": 0.6818,
      "step": 34423
    },
    {
      "epoch": 0.9918746038149023,
      "grad_norm": 0.7419689297676086,
      "learning_rate": 1.7313364372456565e-09,
      "loss": 0.6609,
      "step": 34424
    },
    {
      "epoch": 0.9919034172765516,
      "grad_norm": 0.7109841704368591,
      "learning_rate": 1.7190799290850434e-09,
      "loss": 0.6727,
      "step": 34425
    },
    {
      "epoch": 0.9919322307382009,
      "grad_norm": 0.7155051827430725,
      "learning_rate": 1.7068669509545666e-09,
      "loss": 0.6943,
      "step": 34426
    },
    {
      "epoch": 0.9919610441998502,
      "grad_norm": 0.7321915030479431,
      "learning_rate": 1.694697502960807e-09,
      "loss": 0.6859,
      "step": 34427
    },
    {
      "epoch": 0.9919898576614995,
      "grad_norm": 0.7460507154464722,
      "learning_rate": 1.6825715852097912e-09,
      "loss": 0.6869,
      "step": 34428
    },
    {
      "epoch": 0.9920186711231488,
      "grad_norm": 0.7074359655380249,
      "learning_rate": 1.6704891978069904e-09,
      "loss": 0.6595,
      "step": 34429
    },
    {
      "epoch": 0.992047484584798,
      "grad_norm": 0.7350157499313354,
      "learning_rate": 1.6584503408578757e-09,
      "loss": 0.6877,
      "step": 34430
    },
    {
      "epoch": 0.9920762980464473,
      "grad_norm": 0.7077158093452454,
      "learning_rate": 1.646455014467363e-09,
      "loss": 0.6866,
      "step": 34431
    },
    {
      "epoch": 0.9921051115080965,
      "grad_norm": 0.7181913256645203,
      "learning_rate": 1.6345032187392584e-09,
      "loss": 0.6779,
      "step": 34432
    },
    {
      "epoch": 0.9921339249697458,
      "grad_norm": 0.7288509011268616,
      "learning_rate": 1.622594953778478e-09,
      "loss": 0.6954,
      "step": 34433
    },
    {
      "epoch": 0.9921627384313951,
      "grad_norm": 0.7088630795478821,
      "learning_rate": 1.6107302196882725e-09,
      "loss": 0.6761,
      "step": 34434
    },
    {
      "epoch": 0.9921915518930444,
      "grad_norm": 0.7168551087379456,
      "learning_rate": 1.5989090165718924e-09,
      "loss": 0.6986,
      "step": 34435
    },
    {
      "epoch": 0.9922203653546937,
      "grad_norm": 0.7114540338516235,
      "learning_rate": 1.5871313445325887e-09,
      "loss": 0.6598,
      "step": 34436
    },
    {
      "epoch": 0.992249178816343,
      "grad_norm": 0.7421863079071045,
      "learning_rate": 1.575397203672502e-09,
      "loss": 0.6792,
      "step": 34437
    },
    {
      "epoch": 0.9922779922779923,
      "grad_norm": 0.7194856405258179,
      "learning_rate": 1.5637065940943274e-09,
      "loss": 0.6684,
      "step": 34438
    },
    {
      "epoch": 0.9923068057396416,
      "grad_norm": 0.7272813320159912,
      "learning_rate": 1.552059515899651e-09,
      "loss": 0.6778,
      "step": 34439
    },
    {
      "epoch": 0.9923356192012909,
      "grad_norm": 0.7285915613174438,
      "learning_rate": 1.5404559691900577e-09,
      "loss": 0.672,
      "step": 34440
    },
    {
      "epoch": 0.9923644326629402,
      "grad_norm": 0.7370971441268921,
      "learning_rate": 1.5288959540660232e-09,
      "loss": 0.6934,
      "step": 34441
    },
    {
      "epoch": 0.9923932461245895,
      "grad_norm": 0.7555999159812927,
      "learning_rate": 1.5173794706291322e-09,
      "loss": 0.6927,
      "step": 34442
    },
    {
      "epoch": 0.9924220595862386,
      "grad_norm": 0.7685306668281555,
      "learning_rate": 1.5059065189787502e-09,
      "loss": 0.6509,
      "step": 34443
    },
    {
      "epoch": 0.9924508730478879,
      "grad_norm": 0.7581409811973572,
      "learning_rate": 1.4944770992159075e-09,
      "loss": 0.6966,
      "step": 34444
    },
    {
      "epoch": 0.9924796865095372,
      "grad_norm": 0.7501012682914734,
      "learning_rate": 1.4830912114388585e-09,
      "loss": 0.6561,
      "step": 34445
    },
    {
      "epoch": 0.9925084999711865,
      "grad_norm": 0.7320894002914429,
      "learning_rate": 1.4717488557475235e-09,
      "loss": 0.666,
      "step": 34446
    },
    {
      "epoch": 0.9925373134328358,
      "grad_norm": 0.7362608313560486,
      "learning_rate": 1.4604500322407123e-09,
      "loss": 0.6582,
      "step": 34447
    },
    {
      "epoch": 0.9925661268944851,
      "grad_norm": 0.73738032579422,
      "learning_rate": 1.4491947410166795e-09,
      "loss": 0.6763,
      "step": 34448
    },
    {
      "epoch": 0.9925949403561344,
      "grad_norm": 0.710390031337738,
      "learning_rate": 1.4379829821736802e-09,
      "loss": 0.6581,
      "step": 34449
    },
    {
      "epoch": 0.9926237538177837,
      "grad_norm": 0.7079100012779236,
      "learning_rate": 1.4268147558088585e-09,
      "loss": 0.6626,
      "step": 34450
    },
    {
      "epoch": 0.992652567279433,
      "grad_norm": 0.7187714576721191,
      "learning_rate": 1.4156900620193592e-09,
      "loss": 0.6696,
      "step": 34451
    },
    {
      "epoch": 0.9926813807410823,
      "grad_norm": 0.7191385626792908,
      "learning_rate": 1.4046089009028817e-09,
      "loss": 0.6901,
      "step": 34452
    },
    {
      "epoch": 0.9927101942027315,
      "grad_norm": 0.7292196154594421,
      "learning_rate": 1.3935712725554606e-09,
      "loss": 0.6732,
      "step": 34453
    },
    {
      "epoch": 0.9927390076643808,
      "grad_norm": 0.7115371823310852,
      "learning_rate": 1.38257717707313e-09,
      "loss": 0.6691,
      "step": 34454
    },
    {
      "epoch": 0.9927678211260301,
      "grad_norm": 0.7148070335388184,
      "learning_rate": 1.3716266145519242e-09,
      "loss": 0.6666,
      "step": 34455
    },
    {
      "epoch": 0.9927966345876794,
      "grad_norm": 0.7305977940559387,
      "learning_rate": 1.3607195850873223e-09,
      "loss": 0.6503,
      "step": 34456
    },
    {
      "epoch": 0.9928254480493286,
      "grad_norm": 0.718529224395752,
      "learning_rate": 1.3498560887736932e-09,
      "loss": 0.6815,
      "step": 34457
    },
    {
      "epoch": 0.9928542615109779,
      "grad_norm": 0.7016100287437439,
      "learning_rate": 1.3390361257059614e-09,
      "loss": 0.6704,
      "step": 34458
    },
    {
      "epoch": 0.9928830749726272,
      "grad_norm": 0.7136935591697693,
      "learning_rate": 1.3282596959784954e-09,
      "loss": 0.667,
      "step": 34459
    },
    {
      "epoch": 0.9929118884342765,
      "grad_norm": 0.7283472418785095,
      "learning_rate": 1.3175267996851094e-09,
      "loss": 0.6965,
      "step": 34460
    },
    {
      "epoch": 0.9929407018959258,
      "grad_norm": 0.7276084423065186,
      "learning_rate": 1.3068374369190618e-09,
      "loss": 0.6697,
      "step": 34461
    },
    {
      "epoch": 0.992969515357575,
      "grad_norm": 0.7216910719871521,
      "learning_rate": 1.2961916077736115e-09,
      "loss": 0.6614,
      "step": 34462
    },
    {
      "epoch": 0.9929983288192243,
      "grad_norm": 0.7313213348388672,
      "learning_rate": 1.285589312341462e-09,
      "loss": 0.6783,
      "step": 34463
    },
    {
      "epoch": 0.9930271422808736,
      "grad_norm": 0.7195072174072266,
      "learning_rate": 1.275030550714762e-09,
      "loss": 0.6909,
      "step": 34464
    },
    {
      "epoch": 0.9930559557425229,
      "grad_norm": 0.7314409017562866,
      "learning_rate": 1.26451532298566e-09,
      "loss": 0.6569,
      "step": 34465
    },
    {
      "epoch": 0.9930847692041722,
      "grad_norm": 0.701261579990387,
      "learning_rate": 1.2540436292463042e-09,
      "loss": 0.6699,
      "step": 34466
    },
    {
      "epoch": 0.9931135826658215,
      "grad_norm": 0.727707028388977,
      "learning_rate": 1.243615469586623e-09,
      "loss": 0.6827,
      "step": 34467
    },
    {
      "epoch": 0.9931423961274708,
      "grad_norm": 0.7211867570877075,
      "learning_rate": 1.233230844098765e-09,
      "loss": 0.6942,
      "step": 34468
    },
    {
      "epoch": 0.9931712095891201,
      "grad_norm": 0.7291938066482544,
      "learning_rate": 1.2228897528721028e-09,
      "loss": 0.7028,
      "step": 34469
    },
    {
      "epoch": 0.9932000230507693,
      "grad_norm": 0.7066771984100342,
      "learning_rate": 1.21259219599712e-09,
      "loss": 0.6806,
      "step": 34470
    },
    {
      "epoch": 0.9932288365124186,
      "grad_norm": 0.7428270578384399,
      "learning_rate": 1.202338173563189e-09,
      "loss": 0.6861,
      "step": 34471
    },
    {
      "epoch": 0.9932576499740678,
      "grad_norm": 0.7086887359619141,
      "learning_rate": 1.1921276856607933e-09,
      "loss": 0.6742,
      "step": 34472
    },
    {
      "epoch": 0.9932864634357171,
      "grad_norm": 0.7175560593605042,
      "learning_rate": 1.1819607323770853e-09,
      "loss": 0.6908,
      "step": 34473
    },
    {
      "epoch": 0.9933152768973664,
      "grad_norm": 0.709022045135498,
      "learning_rate": 1.1718373138019934e-09,
      "loss": 0.6819,
      "step": 34474
    },
    {
      "epoch": 0.9933440903590157,
      "grad_norm": 0.7390811443328857,
      "learning_rate": 1.1617574300232248e-09,
      "loss": 0.6718,
      "step": 34475
    },
    {
      "epoch": 0.993372903820665,
      "grad_norm": 0.7238076329231262,
      "learning_rate": 1.1517210811284874e-09,
      "loss": 0.6561,
      "step": 34476
    },
    {
      "epoch": 0.9934017172823143,
      "grad_norm": 0.706977367401123,
      "learning_rate": 1.1417282672054885e-09,
      "loss": 0.6652,
      "step": 34477
    },
    {
      "epoch": 0.9934305307439636,
      "grad_norm": 0.6988046765327454,
      "learning_rate": 1.1317789883402708e-09,
      "loss": 0.648,
      "step": 34478
    },
    {
      "epoch": 0.9934593442056129,
      "grad_norm": 0.7196896076202393,
      "learning_rate": 1.1218732446210967e-09,
      "loss": 0.6588,
      "step": 34479
    },
    {
      "epoch": 0.9934881576672622,
      "grad_norm": 0.7090527415275574,
      "learning_rate": 1.1120110361328984e-09,
      "loss": 0.6593,
      "step": 34480
    },
    {
      "epoch": 0.9935169711289115,
      "grad_norm": 0.7277007699012756,
      "learning_rate": 1.1021923629617182e-09,
      "loss": 0.6671,
      "step": 34481
    },
    {
      "epoch": 0.9935457845905608,
      "grad_norm": 0.7247980237007141,
      "learning_rate": 1.0924172251941533e-09,
      "loss": 0.6943,
      "step": 34482
    },
    {
      "epoch": 0.99357459805221,
      "grad_norm": 0.748701274394989,
      "learning_rate": 1.0826856229140258e-09,
      "loss": 0.6915,
      "step": 34483
    },
    {
      "epoch": 0.9936034115138592,
      "grad_norm": 0.7272992134094238,
      "learning_rate": 1.0729975562068229e-09,
      "loss": 0.6832,
      "step": 34484
    },
    {
      "epoch": 0.9936322249755085,
      "grad_norm": 0.747133195400238,
      "learning_rate": 1.0633530251563663e-09,
      "loss": 0.6554,
      "step": 34485
    },
    {
      "epoch": 0.9936610384371578,
      "grad_norm": 0.7228357195854187,
      "learning_rate": 1.0537520298470327e-09,
      "loss": 0.6926,
      "step": 34486
    },
    {
      "epoch": 0.9936898518988071,
      "grad_norm": 0.7274424433708191,
      "learning_rate": 1.0441945703626444e-09,
      "loss": 0.6887,
      "step": 34487
    },
    {
      "epoch": 0.9937186653604564,
      "grad_norm": 0.7272101640701294,
      "learning_rate": 1.0346806467864678e-09,
      "loss": 0.6912,
      "step": 34488
    },
    {
      "epoch": 0.9937474788221057,
      "grad_norm": 0.7019572854042053,
      "learning_rate": 1.0252102592006597e-09,
      "loss": 0.6615,
      "step": 34489
    },
    {
      "epoch": 0.993776292283755,
      "grad_norm": 0.717170000076294,
      "learning_rate": 1.0157834076879314e-09,
      "loss": 0.6884,
      "step": 34490
    },
    {
      "epoch": 0.9938051057454043,
      "grad_norm": 0.7336384057998657,
      "learning_rate": 1.0064000923309947e-09,
      "loss": 0.6639,
      "step": 34491
    },
    {
      "epoch": 0.9938339192070536,
      "grad_norm": 0.7209432721138,
      "learning_rate": 9.970603132108957e-10,
      "loss": 0.6875,
      "step": 34492
    },
    {
      "epoch": 0.9938627326687028,
      "grad_norm": 0.7404378056526184,
      "learning_rate": 9.877640704097913e-10,
      "loss": 0.6671,
      "step": 34493
    },
    {
      "epoch": 0.9938915461303521,
      "grad_norm": 0.7274737358093262,
      "learning_rate": 9.78511364007617e-10,
      "loss": 0.7001,
      "step": 34494
    },
    {
      "epoch": 0.9939203595920014,
      "grad_norm": 0.7356138825416565,
      "learning_rate": 9.693021940854198e-10,
      "loss": 0.674,
      "step": 34495
    },
    {
      "epoch": 0.9939491730536507,
      "grad_norm": 0.733368992805481,
      "learning_rate": 9.60136560723135e-10,
      "loss": 0.6797,
      "step": 34496
    },
    {
      "epoch": 0.9939779865152999,
      "grad_norm": 0.7330193519592285,
      "learning_rate": 9.510144640012542e-10,
      "loss": 0.6766,
      "step": 34497
    },
    {
      "epoch": 0.9940067999769492,
      "grad_norm": 0.6974572539329529,
      "learning_rate": 9.419359039986032e-10,
      "loss": 0.6757,
      "step": 34498
    },
    {
      "epoch": 0.9940356134385985,
      "grad_norm": 0.7279320955276489,
      "learning_rate": 9.329008807945628e-10,
      "loss": 0.6993,
      "step": 34499
    },
    {
      "epoch": 0.9940644269002478,
      "grad_norm": 0.7118632793426514,
      "learning_rate": 9.239093944679589e-10,
      "loss": 0.65,
      "step": 34500
    },
    {
      "epoch": 0.9940932403618971,
      "grad_norm": 0.7040707468986511,
      "learning_rate": 9.149614450965072e-10,
      "loss": 0.6842,
      "step": 34501
    },
    {
      "epoch": 0.9941220538235463,
      "grad_norm": 0.7043389678001404,
      "learning_rate": 9.060570327584784e-10,
      "loss": 0.6514,
      "step": 34502
    },
    {
      "epoch": 0.9941508672851956,
      "grad_norm": 0.7074833512306213,
      "learning_rate": 8.971961575315879e-10,
      "loss": 0.6699,
      "step": 34503
    },
    {
      "epoch": 0.9941796807468449,
      "grad_norm": 0.7134007811546326,
      "learning_rate": 8.883788194929965e-10,
      "loss": 0.6804,
      "step": 34504
    },
    {
      "epoch": 0.9942084942084942,
      "grad_norm": 0.7155733704566956,
      "learning_rate": 8.796050187193095e-10,
      "loss": 0.6745,
      "step": 34505
    },
    {
      "epoch": 0.9942373076701435,
      "grad_norm": 0.733695924282074,
      "learning_rate": 8.70874755286577e-10,
      "loss": 0.7126,
      "step": 34506
    },
    {
      "epoch": 0.9942661211317928,
      "grad_norm": 0.7086948752403259,
      "learning_rate": 8.621880292714046e-10,
      "loss": 0.6863,
      "step": 34507
    },
    {
      "epoch": 0.9942949345934421,
      "grad_norm": 0.7233319282531738,
      "learning_rate": 8.535448407492875e-10,
      "loss": 0.6598,
      "step": 34508
    },
    {
      "epoch": 0.9943237480550914,
      "grad_norm": 0.7187408804893494,
      "learning_rate": 8.449451897957206e-10,
      "loss": 0.6788,
      "step": 34509
    },
    {
      "epoch": 0.9943525615167407,
      "grad_norm": 0.7115600109100342,
      "learning_rate": 8.36389076485089e-10,
      "loss": 0.6783,
      "step": 34510
    },
    {
      "epoch": 0.9943813749783899,
      "grad_norm": 0.7164440751075745,
      "learning_rate": 8.278765008923328e-10,
      "loss": 0.6662,
      "step": 34511
    },
    {
      "epoch": 0.9944101884400391,
      "grad_norm": 0.7254422307014465,
      "learning_rate": 8.194074630912818e-10,
      "loss": 0.6856,
      "step": 34512
    },
    {
      "epoch": 0.9944390019016884,
      "grad_norm": 0.7409167289733887,
      "learning_rate": 8.109819631557659e-10,
      "loss": 0.6779,
      "step": 34513
    },
    {
      "epoch": 0.9944678153633377,
      "grad_norm": 0.7076181769371033,
      "learning_rate": 8.026000011596147e-10,
      "loss": 0.6829,
      "step": 34514
    },
    {
      "epoch": 0.994496628824987,
      "grad_norm": 0.7317229509353638,
      "learning_rate": 7.942615771749928e-10,
      "loss": 0.701,
      "step": 34515
    },
    {
      "epoch": 0.9945254422866363,
      "grad_norm": 0.7197121381759644,
      "learning_rate": 7.859666912751751e-10,
      "loss": 0.6658,
      "step": 34516
    },
    {
      "epoch": 0.9945542557482856,
      "grad_norm": 0.7421066761016846,
      "learning_rate": 7.777153435323259e-10,
      "loss": 0.7059,
      "step": 34517
    },
    {
      "epoch": 0.9945830692099349,
      "grad_norm": 0.7243902087211609,
      "learning_rate": 7.695075340180547e-10,
      "loss": 0.6513,
      "step": 34518
    },
    {
      "epoch": 0.9946118826715842,
      "grad_norm": 0.7155764698982239,
      "learning_rate": 7.613432628039708e-10,
      "loss": 0.6971,
      "step": 34519
    },
    {
      "epoch": 0.9946406961332335,
      "grad_norm": 0.7556325197219849,
      "learning_rate": 7.532225299611284e-10,
      "loss": 0.7017,
      "step": 34520
    },
    {
      "epoch": 0.9946695095948828,
      "grad_norm": 0.7159278988838196,
      "learning_rate": 7.451453355600269e-10,
      "loss": 0.6677,
      "step": 34521
    },
    {
      "epoch": 0.994698323056532,
      "grad_norm": 0.7228487730026245,
      "learning_rate": 7.371116796717203e-10,
      "loss": 0.6841,
      "step": 34522
    },
    {
      "epoch": 0.9947271365181813,
      "grad_norm": 0.7273836731910706,
      "learning_rate": 7.291215623655979e-10,
      "loss": 0.6869,
      "step": 34523
    },
    {
      "epoch": 0.9947559499798305,
      "grad_norm": 0.7235155701637268,
      "learning_rate": 7.211749837116033e-10,
      "loss": 0.6675,
      "step": 34524
    },
    {
      "epoch": 0.9947847634414798,
      "grad_norm": 0.7239829301834106,
      "learning_rate": 7.132719437785707e-10,
      "loss": 0.6986,
      "step": 34525
    },
    {
      "epoch": 0.9948135769031291,
      "grad_norm": 0.7332563400268555,
      "learning_rate": 7.054124426353337e-10,
      "loss": 0.6787,
      "step": 34526
    },
    {
      "epoch": 0.9948423903647784,
      "grad_norm": 0.7165006399154663,
      "learning_rate": 6.975964803507262e-10,
      "loss": 0.6388,
      "step": 34527
    },
    {
      "epoch": 0.9948712038264277,
      "grad_norm": 0.7474826574325562,
      "learning_rate": 6.898240569924719e-10,
      "loss": 0.701,
      "step": 34528
    },
    {
      "epoch": 0.994900017288077,
      "grad_norm": 0.7170189023017883,
      "learning_rate": 6.820951726282943e-10,
      "loss": 0.666,
      "step": 34529
    },
    {
      "epoch": 0.9949288307497263,
      "grad_norm": 0.7127454280853271,
      "learning_rate": 6.74409827325917e-10,
      "loss": 0.6722,
      "step": 34530
    },
    {
      "epoch": 0.9949576442113756,
      "grad_norm": 0.7148706912994385,
      "learning_rate": 6.667680211513982e-10,
      "loss": 0.6521,
      "step": 34531
    },
    {
      "epoch": 0.9949864576730248,
      "grad_norm": 0.7154721021652222,
      "learning_rate": 6.591697541719066e-10,
      "loss": 0.6871,
      "step": 34532
    },
    {
      "epoch": 0.9950152711346741,
      "grad_norm": 0.7089959383010864,
      "learning_rate": 6.516150264540555e-10,
      "loss": 0.6685,
      "step": 34533
    },
    {
      "epoch": 0.9950440845963234,
      "grad_norm": 0.724007248878479,
      "learning_rate": 6.441038380627929e-10,
      "loss": 0.6873,
      "step": 34534
    },
    {
      "epoch": 0.9950728980579727,
      "grad_norm": 0.7363383769989014,
      "learning_rate": 6.366361890636219e-10,
      "loss": 0.6718,
      "step": 34535
    },
    {
      "epoch": 0.995101711519622,
      "grad_norm": 0.7230014204978943,
      "learning_rate": 6.292120795220457e-10,
      "loss": 0.6772,
      "step": 34536
    },
    {
      "epoch": 0.9951305249812713,
      "grad_norm": 0.7114102244377136,
      "learning_rate": 6.218315095024574e-10,
      "loss": 0.6871,
      "step": 34537
    },
    {
      "epoch": 0.9951593384429205,
      "grad_norm": 0.711692750453949,
      "learning_rate": 6.144944790692498e-10,
      "loss": 0.6594,
      "step": 34538
    },
    {
      "epoch": 0.9951881519045698,
      "grad_norm": 0.7178075313568115,
      "learning_rate": 6.072009882862606e-10,
      "loss": 0.6768,
      "step": 34539
    },
    {
      "epoch": 0.9952169653662191,
      "grad_norm": 0.7278192639350891,
      "learning_rate": 5.999510372167727e-10,
      "loss": 0.6969,
      "step": 34540
    },
    {
      "epoch": 0.9952457788278684,
      "grad_norm": 0.7250493168830872,
      "learning_rate": 5.927446259246239e-10,
      "loss": 0.686,
      "step": 34541
    },
    {
      "epoch": 0.9952745922895176,
      "grad_norm": 0.7251787185668945,
      "learning_rate": 5.855817544714315e-10,
      "loss": 0.6828,
      "step": 34542
    },
    {
      "epoch": 0.9953034057511669,
      "grad_norm": 0.7354498505592346,
      "learning_rate": 5.784624229204783e-10,
      "loss": 0.6936,
      "step": 34543
    },
    {
      "epoch": 0.9953322192128162,
      "grad_norm": 0.7182382345199585,
      "learning_rate": 5.713866313339367e-10,
      "loss": 0.6785,
      "step": 34544
    },
    {
      "epoch": 0.9953610326744655,
      "grad_norm": 0.7547792196273804,
      "learning_rate": 5.643543797723139e-10,
      "loss": 0.6771,
      "step": 34545
    },
    {
      "epoch": 0.9953898461361148,
      "grad_norm": 0.8363136649131775,
      "learning_rate": 5.573656682977824e-10,
      "loss": 0.6678,
      "step": 34546
    },
    {
      "epoch": 0.9954186595977641,
      "grad_norm": 0.7044817805290222,
      "learning_rate": 5.504204969714044e-10,
      "loss": 0.6737,
      "step": 34547
    },
    {
      "epoch": 0.9954474730594134,
      "grad_norm": 0.7186312675476074,
      "learning_rate": 5.435188658531321e-10,
      "loss": 0.6881,
      "step": 34548
    },
    {
      "epoch": 0.9954762865210627,
      "grad_norm": 0.7482162714004517,
      "learning_rate": 5.366607750029174e-10,
      "loss": 0.6915,
      "step": 34549
    },
    {
      "epoch": 0.995505099982712,
      "grad_norm": 0.7169263958930969,
      "learning_rate": 5.298462244807124e-10,
      "loss": 0.6816,
      "step": 34550
    },
    {
      "epoch": 0.9955339134443612,
      "grad_norm": 0.7382341623306274,
      "learning_rate": 5.230752143459139e-10,
      "loss": 0.6929,
      "step": 34551
    },
    {
      "epoch": 0.9955627269060104,
      "grad_norm": 0.7355203628540039,
      "learning_rate": 5.16347744657919e-10,
      "loss": 0.7091,
      "step": 34552
    },
    {
      "epoch": 0.9955915403676597,
      "grad_norm": 0.7444164752960205,
      "learning_rate": 5.096638154744593e-10,
      "loss": 0.6871,
      "step": 34553
    },
    {
      "epoch": 0.995620353829309,
      "grad_norm": 0.7050867080688477,
      "learning_rate": 5.030234268543765e-10,
      "loss": 0.6426,
      "step": 34554
    },
    {
      "epoch": 0.9956491672909583,
      "grad_norm": 0.7215266823768616,
      "learning_rate": 4.964265788548472e-10,
      "loss": 0.7081,
      "step": 34555
    },
    {
      "epoch": 0.9956779807526076,
      "grad_norm": 0.7209382057189941,
      "learning_rate": 4.898732715341581e-10,
      "loss": 0.6602,
      "step": 34556
    },
    {
      "epoch": 0.9957067942142569,
      "grad_norm": 0.7348635196685791,
      "learning_rate": 4.833635049489305e-10,
      "loss": 0.6814,
      "step": 34557
    },
    {
      "epoch": 0.9957356076759062,
      "grad_norm": 0.7164553999900818,
      "learning_rate": 4.768972791563409e-10,
      "loss": 0.6647,
      "step": 34558
    },
    {
      "epoch": 0.9957644211375555,
      "grad_norm": 0.6933764815330505,
      "learning_rate": 4.704745942119005e-10,
      "loss": 0.6778,
      "step": 34559
    },
    {
      "epoch": 0.9957932345992048,
      "grad_norm": 0.706795334815979,
      "learning_rate": 4.6409545017167547e-10,
      "loss": 0.6524,
      "step": 34560
    },
    {
      "epoch": 0.9958220480608541,
      "grad_norm": 0.7273809909820557,
      "learning_rate": 4.5775984709173216e-10,
      "loss": 0.6793,
      "step": 34561
    },
    {
      "epoch": 0.9958508615225034,
      "grad_norm": 0.8202216625213623,
      "learning_rate": 4.514677850270266e-10,
      "loss": 0.6845,
      "step": 34562
    },
    {
      "epoch": 0.9958796749841526,
      "grad_norm": 0.7048245072364807,
      "learning_rate": 4.452192640319597e-10,
      "loss": 0.687,
      "step": 34563
    },
    {
      "epoch": 0.9959084884458019,
      "grad_norm": 0.7059239149093628,
      "learning_rate": 4.390142841620426e-10,
      "loss": 0.6826,
      "step": 34564
    },
    {
      "epoch": 0.9959373019074511,
      "grad_norm": 0.7125080227851868,
      "learning_rate": 4.328528454700109e-10,
      "loss": 0.668,
      "step": 34565
    },
    {
      "epoch": 0.9959661153691004,
      "grad_norm": 0.7130905985832214,
      "learning_rate": 4.267349480102656e-10,
      "loss": 0.6814,
      "step": 34566
    },
    {
      "epoch": 0.9959949288307497,
      "grad_norm": 0.7236467599868774,
      "learning_rate": 4.206605918355422e-10,
      "loss": 0.6709,
      "step": 34567
    },
    {
      "epoch": 0.996023742292399,
      "grad_norm": 0.7374907732009888,
      "learning_rate": 4.1462977699968656e-10,
      "loss": 0.6962,
      "step": 34568
    },
    {
      "epoch": 0.9960525557540483,
      "grad_norm": 0.7355034351348877,
      "learning_rate": 4.08642503554324e-10,
      "loss": 0.6989,
      "step": 34569
    },
    {
      "epoch": 0.9960813692156976,
      "grad_norm": 0.7202180027961731,
      "learning_rate": 4.0269877155219017e-10,
      "loss": 0.6566,
      "step": 34570
    },
    {
      "epoch": 0.9961101826773469,
      "grad_norm": 0.7286995649337769,
      "learning_rate": 3.9679858104435533e-10,
      "loss": 0.6875,
      "step": 34571
    },
    {
      "epoch": 0.9961389961389961,
      "grad_norm": 0.7194452881813049,
      "learning_rate": 3.909419320829999e-10,
      "loss": 0.6705,
      "step": 34572
    },
    {
      "epoch": 0.9961678096006454,
      "grad_norm": 0.7255688309669495,
      "learning_rate": 3.851288247186391e-10,
      "loss": 0.6745,
      "step": 34573
    },
    {
      "epoch": 0.9961966230622947,
      "grad_norm": 0.7116386890411377,
      "learning_rate": 3.7935925900178804e-10,
      "loss": 0.6668,
      "step": 34574
    },
    {
      "epoch": 0.996225436523944,
      "grad_norm": 0.7427639365196228,
      "learning_rate": 3.73633234983517e-10,
      "loss": 0.6846,
      "step": 34575
    },
    {
      "epoch": 0.9962542499855933,
      "grad_norm": 0.7334703207015991,
      "learning_rate": 3.679507527126758e-10,
      "loss": 0.6533,
      "step": 34576
    },
    {
      "epoch": 0.9962830634472426,
      "grad_norm": 0.7419028282165527,
      "learning_rate": 3.623118122392244e-10,
      "loss": 0.6784,
      "step": 34577
    },
    {
      "epoch": 0.9963118769088918,
      "grad_norm": 0.7181965112686157,
      "learning_rate": 3.567164136120127e-10,
      "loss": 0.6633,
      "step": 34578
    },
    {
      "epoch": 0.9963406903705411,
      "grad_norm": 0.7230302691459656,
      "learning_rate": 3.511645568804456e-10,
      "loss": 0.6784,
      "step": 34579
    },
    {
      "epoch": 0.9963695038321904,
      "grad_norm": 0.7111204266548157,
      "learning_rate": 3.4565624209226265e-10,
      "loss": 0.6809,
      "step": 34580
    },
    {
      "epoch": 0.9963983172938397,
      "grad_norm": 0.7401975989341736,
      "learning_rate": 3.401914692952035e-10,
      "loss": 0.6718,
      "step": 34581
    },
    {
      "epoch": 0.9964271307554889,
      "grad_norm": 0.7007699608802795,
      "learning_rate": 3.3477023853756283e-10,
      "loss": 0.6695,
      "step": 34582
    },
    {
      "epoch": 0.9964559442171382,
      "grad_norm": 0.7053753137588501,
      "learning_rate": 3.2939254986652513e-10,
      "loss": 0.6305,
      "step": 34583
    },
    {
      "epoch": 0.9964847576787875,
      "grad_norm": 0.7194659113883972,
      "learning_rate": 3.2405840332816463e-10,
      "loss": 0.668,
      "step": 34584
    },
    {
      "epoch": 0.9965135711404368,
      "grad_norm": 0.7275649309158325,
      "learning_rate": 3.187677989696658e-10,
      "loss": 0.7015,
      "step": 34585
    },
    {
      "epoch": 0.9965423846020861,
      "grad_norm": 0.7160215973854065,
      "learning_rate": 3.1352073683654784e-10,
      "loss": 0.7052,
      "step": 34586
    },
    {
      "epoch": 0.9965711980637354,
      "grad_norm": 0.7251039743423462,
      "learning_rate": 3.0831721697544004e-10,
      "loss": 0.6686,
      "step": 34587
    },
    {
      "epoch": 0.9966000115253847,
      "grad_norm": 0.7066527605056763,
      "learning_rate": 3.0315723943019625e-10,
      "loss": 0.6687,
      "step": 34588
    },
    {
      "epoch": 0.996628824987034,
      "grad_norm": 0.7049278020858765,
      "learning_rate": 2.9804080424744587e-10,
      "loss": 0.667,
      "step": 34589
    },
    {
      "epoch": 0.9966576384486833,
      "grad_norm": 0.7236176133155823,
      "learning_rate": 2.929679114704875e-10,
      "loss": 0.6709,
      "step": 34590
    },
    {
      "epoch": 0.9966864519103326,
      "grad_norm": 0.7298858761787415,
      "learning_rate": 2.879385611437302e-10,
      "loss": 0.6474,
      "step": 34591
    },
    {
      "epoch": 0.9967152653719817,
      "grad_norm": 0.7355515956878662,
      "learning_rate": 2.829527533110277e-10,
      "loss": 0.6947,
      "step": 34592
    },
    {
      "epoch": 0.996744078833631,
      "grad_norm": 0.7149889469146729,
      "learning_rate": 2.7801048801623376e-10,
      "loss": 0.6661,
      "step": 34593
    },
    {
      "epoch": 0.9967728922952803,
      "grad_norm": 0.7076429128646851,
      "learning_rate": 2.7311176530209205e-10,
      "loss": 0.7047,
      "step": 34594
    },
    {
      "epoch": 0.9968017057569296,
      "grad_norm": 0.7159367799758911,
      "learning_rate": 2.6825658521134614e-10,
      "loss": 0.6777,
      "step": 34595
    },
    {
      "epoch": 0.9968305192185789,
      "grad_norm": 0.7176808714866638,
      "learning_rate": 2.6344494778562933e-10,
      "loss": 0.7016,
      "step": 34596
    },
    {
      "epoch": 0.9968593326802282,
      "grad_norm": 0.7232877016067505,
      "learning_rate": 2.5867685306824043e-10,
      "loss": 0.6916,
      "step": 34597
    },
    {
      "epoch": 0.9968881461418775,
      "grad_norm": 0.726573646068573,
      "learning_rate": 2.5395230109914736e-10,
      "loss": 0.6643,
      "step": 34598
    },
    {
      "epoch": 0.9969169596035268,
      "grad_norm": 0.7399914860725403,
      "learning_rate": 2.4927129192053866e-10,
      "loss": 0.6724,
      "step": 34599
    },
    {
      "epoch": 0.9969457730651761,
      "grad_norm": 0.7338676452636719,
      "learning_rate": 2.446338255729375e-10,
      "loss": 0.6549,
      "step": 34600
    },
    {
      "epoch": 0.9969745865268254,
      "grad_norm": 0.7158873677253723,
      "learning_rate": 2.400399020963118e-10,
      "loss": 0.6615,
      "step": 34601
    },
    {
      "epoch": 0.9970033999884746,
      "grad_norm": 0.7213645577430725,
      "learning_rate": 2.3548952153118474e-10,
      "loss": 0.6722,
      "step": 34602
    },
    {
      "epoch": 0.9970322134501239,
      "grad_norm": 0.7290063500404358,
      "learning_rate": 2.3098268391696933e-10,
      "loss": 0.655,
      "step": 34603
    },
    {
      "epoch": 0.9970610269117732,
      "grad_norm": 0.72430020570755,
      "learning_rate": 2.2651938929307838e-10,
      "loss": 0.6534,
      "step": 34604
    },
    {
      "epoch": 0.9970898403734224,
      "grad_norm": 0.7168049216270447,
      "learning_rate": 2.2209963769836973e-10,
      "loss": 0.7012,
      "step": 34605
    },
    {
      "epoch": 0.9971186538350717,
      "grad_norm": 0.7338898181915283,
      "learning_rate": 2.1772342917059096e-10,
      "loss": 0.6957,
      "step": 34606
    },
    {
      "epoch": 0.997147467296721,
      "grad_norm": 0.7209183573722839,
      "learning_rate": 2.1339076374915502e-10,
      "loss": 0.6638,
      "step": 34607
    },
    {
      "epoch": 0.9971762807583703,
      "grad_norm": 0.7395858764648438,
      "learning_rate": 2.0910164147069923e-10,
      "loss": 0.6881,
      "step": 34608
    },
    {
      "epoch": 0.9972050942200196,
      "grad_norm": 0.727520763874054,
      "learning_rate": 2.048560623735263e-10,
      "loss": 0.6749,
      "step": 34609
    },
    {
      "epoch": 0.9972339076816689,
      "grad_norm": 0.7004662752151489,
      "learning_rate": 2.0065402649371845e-10,
      "loss": 0.658,
      "step": 34610
    },
    {
      "epoch": 0.9972627211433182,
      "grad_norm": 0.7382463216781616,
      "learning_rate": 1.9649553386846821e-10,
      "loss": 0.6937,
      "step": 34611
    },
    {
      "epoch": 0.9972915346049674,
      "grad_norm": 0.7267391085624695,
      "learning_rate": 1.9238058453330266e-10,
      "loss": 0.6811,
      "step": 34612
    },
    {
      "epoch": 0.9973203480666167,
      "grad_norm": 0.7123645544052124,
      "learning_rate": 1.883091785248592e-10,
      "loss": 0.6748,
      "step": 34613
    },
    {
      "epoch": 0.997349161528266,
      "grad_norm": 0.7124285101890564,
      "learning_rate": 1.842813158781098e-10,
      "loss": 0.6596,
      "step": 34614
    },
    {
      "epoch": 0.9973779749899153,
      "grad_norm": 0.7258379459381104,
      "learning_rate": 1.802969966285817e-10,
      "loss": 0.6868,
      "step": 34615
    },
    {
      "epoch": 0.9974067884515646,
      "grad_norm": 0.7065138816833496,
      "learning_rate": 1.763562208106917e-10,
      "loss": 0.691,
      "step": 34616
    },
    {
      "epoch": 0.9974356019132139,
      "grad_norm": 0.7289742231369019,
      "learning_rate": 1.7245898845830167e-10,
      "loss": 0.673,
      "step": 34617
    },
    {
      "epoch": 0.9974644153748632,
      "grad_norm": 0.7348135709762573,
      "learning_rate": 1.686052996063836e-10,
      "loss": 0.6952,
      "step": 34618
    },
    {
      "epoch": 0.9974932288365124,
      "grad_norm": 0.7127604484558105,
      "learning_rate": 1.64795154287134e-10,
      "loss": 0.6798,
      "step": 34619
    },
    {
      "epoch": 0.9975220422981617,
      "grad_norm": 0.7251308560371399,
      "learning_rate": 1.6102855253496973e-10,
      "loss": 0.6655,
      "step": 34620
    },
    {
      "epoch": 0.997550855759811,
      "grad_norm": 0.711446225643158,
      "learning_rate": 1.5730549438264242e-10,
      "loss": 0.6762,
      "step": 34621
    },
    {
      "epoch": 0.9975796692214602,
      "grad_norm": 0.7313432693481445,
      "learning_rate": 1.5362597986123828e-10,
      "loss": 0.6834,
      "step": 34622
    },
    {
      "epoch": 0.9976084826831095,
      "grad_norm": 0.7135794162750244,
      "learning_rate": 1.4999000900461912e-10,
      "loss": 0.6893,
      "step": 34623
    },
    {
      "epoch": 0.9976372961447588,
      "grad_norm": 0.7240836024284363,
      "learning_rate": 1.463975818433161e-10,
      "loss": 0.671,
      "step": 34624
    },
    {
      "epoch": 0.9976661096064081,
      "grad_norm": 0.7184834480285645,
      "learning_rate": 1.4284869840841543e-10,
      "loss": 0.6854,
      "step": 34625
    },
    {
      "epoch": 0.9976949230680574,
      "grad_norm": 0.7244728803634644,
      "learning_rate": 1.3934335873155848e-10,
      "loss": 0.6766,
      "step": 34626
    },
    {
      "epoch": 0.9977237365297067,
      "grad_norm": 0.7315801382064819,
      "learning_rate": 1.3588156284327637e-10,
      "loss": 0.6673,
      "step": 34627
    },
    {
      "epoch": 0.997752549991356,
      "grad_norm": 0.7115014791488647,
      "learning_rate": 1.3246331077299e-10,
      "loss": 0.6796,
      "step": 34628
    },
    {
      "epoch": 0.9977813634530053,
      "grad_norm": 0.7145321369171143,
      "learning_rate": 1.2908860255123057e-10,
      "loss": 0.69,
      "step": 34629
    },
    {
      "epoch": 0.9978101769146546,
      "grad_norm": 0.7192953824996948,
      "learning_rate": 1.257574382068638e-10,
      "loss": 0.6685,
      "step": 34630
    },
    {
      "epoch": 0.9978389903763039,
      "grad_norm": 0.7088042497634888,
      "learning_rate": 1.2246981776931065e-10,
      "loss": 0.6636,
      "step": 34631
    },
    {
      "epoch": 0.997867803837953,
      "grad_norm": 0.7314918041229248,
      "learning_rate": 1.1922574126688181e-10,
      "loss": 0.6709,
      "step": 34632
    },
    {
      "epoch": 0.9978966172996023,
      "grad_norm": 0.7227196097373962,
      "learning_rate": 1.1602520872788792e-10,
      "loss": 0.6776,
      "step": 34633
    },
    {
      "epoch": 0.9979254307612516,
      "grad_norm": 0.7202069759368896,
      "learning_rate": 1.1286822018008459e-10,
      "loss": 0.6803,
      "step": 34634
    },
    {
      "epoch": 0.9979542442229009,
      "grad_norm": 0.7667086720466614,
      "learning_rate": 1.097547756517825e-10,
      "loss": 0.722,
      "step": 34635
    },
    {
      "epoch": 0.9979830576845502,
      "grad_norm": 0.7190220952033997,
      "learning_rate": 1.066848751690719e-10,
      "loss": 0.6879,
      "step": 34636
    },
    {
      "epoch": 0.9980118711461995,
      "grad_norm": 0.7720209360122681,
      "learning_rate": 1.0365851875859812e-10,
      "loss": 0.6734,
      "step": 34637
    },
    {
      "epoch": 0.9980406846078488,
      "grad_norm": 0.7194583415985107,
      "learning_rate": 1.0067570644756164e-10,
      "loss": 0.6652,
      "step": 34638
    },
    {
      "epoch": 0.9980694980694981,
      "grad_norm": 0.7303584218025208,
      "learning_rate": 9.773643826205271e-11,
      "loss": 0.7004,
      "step": 34639
    },
    {
      "epoch": 0.9980983115311474,
      "grad_norm": 0.7158762812614441,
      "learning_rate": 9.484071422649621e-11,
      "loss": 0.6764,
      "step": 34640
    },
    {
      "epoch": 0.9981271249927967,
      "grad_norm": 0.7162401676177979,
      "learning_rate": 9.19885343669824e-11,
      "loss": 0.681,
      "step": 34641
    },
    {
      "epoch": 0.998155938454446,
      "grad_norm": 0.7092809677124023,
      "learning_rate": 8.917989870849131e-11,
      "loss": 0.6531,
      "step": 34642
    },
    {
      "epoch": 0.9981847519160952,
      "grad_norm": 0.7267215847969055,
      "learning_rate": 8.641480727489271e-11,
      "loss": 0.6765,
      "step": 34643
    },
    {
      "epoch": 0.9982135653777445,
      "grad_norm": 0.708720326423645,
      "learning_rate": 8.369326009061152e-11,
      "loss": 0.651,
      "step": 34644
    },
    {
      "epoch": 0.9982423788393938,
      "grad_norm": 0.7186529040336609,
      "learning_rate": 8.101525717951753e-11,
      "loss": 0.6664,
      "step": 34645
    },
    {
      "epoch": 0.998271192301043,
      "grad_norm": 0.810303807258606,
      "learning_rate": 7.838079856437031e-11,
      "loss": 0.6686,
      "step": 34646
    },
    {
      "epoch": 0.9983000057626923,
      "grad_norm": 0.819257378578186,
      "learning_rate": 7.578988426848455e-11,
      "loss": 0.6663,
      "step": 34647
    },
    {
      "epoch": 0.9983288192243416,
      "grad_norm": 0.7364847660064697,
      "learning_rate": 7.324251431406471e-11,
      "loss": 0.6974,
      "step": 34648
    },
    {
      "epoch": 0.9983576326859909,
      "grad_norm": 0.7085341215133667,
      "learning_rate": 7.073868872387035e-11,
      "loss": 0.6873,
      "step": 34649
    },
    {
      "epoch": 0.9983864461476402,
      "grad_norm": 0.7059379816055298,
      "learning_rate": 6.827840751955084e-11,
      "loss": 0.6704,
      "step": 34650
    },
    {
      "epoch": 0.9984152596092895,
      "grad_norm": 0.7279603481292725,
      "learning_rate": 6.58616707222004e-11,
      "loss": 0.6668,
      "step": 34651
    },
    {
      "epoch": 0.9984440730709387,
      "grad_norm": 0.7306922674179077,
      "learning_rate": 6.348847835346838e-11,
      "loss": 0.6971,
      "step": 34652
    },
    {
      "epoch": 0.998472886532588,
      "grad_norm": 0.7430827021598816,
      "learning_rate": 6.115883043333881e-11,
      "loss": 0.6818,
      "step": 34653
    },
    {
      "epoch": 0.9985016999942373,
      "grad_norm": 0.7318741679191589,
      "learning_rate": 5.88727269823508e-11,
      "loss": 0.6907,
      "step": 34654
    },
    {
      "epoch": 0.9985305134558866,
      "grad_norm": 0.7384433150291443,
      "learning_rate": 5.6630168021043484e-11,
      "loss": 0.6994,
      "step": 34655
    },
    {
      "epoch": 0.9985593269175359,
      "grad_norm": 0.7415874004364014,
      "learning_rate": 5.443115356773554e-11,
      "loss": 0.6679,
      "step": 34656
    },
    {
      "epoch": 0.9985881403791852,
      "grad_norm": 0.7389257550239563,
      "learning_rate": 5.227568364241098e-11,
      "loss": 0.6804,
      "step": 34657
    },
    {
      "epoch": 0.9986169538408345,
      "grad_norm": 0.7135925889015198,
      "learning_rate": 5.01637582639436e-11,
      "loss": 0.6877,
      "step": 34658
    },
    {
      "epoch": 0.9986457673024837,
      "grad_norm": 0.7132635116577148,
      "learning_rate": 4.809537745009696e-11,
      "loss": 0.6527,
      "step": 34659
    },
    {
      "epoch": 0.998674580764133,
      "grad_norm": 0.7262187004089355,
      "learning_rate": 4.607054121918975e-11,
      "loss": 0.6613,
      "step": 34660
    },
    {
      "epoch": 0.9987033942257822,
      "grad_norm": 0.6995380520820618,
      "learning_rate": 4.408924958898553e-11,
      "loss": 0.687,
      "step": 34661
    },
    {
      "epoch": 0.9987322076874315,
      "grad_norm": 0.7369899749755859,
      "learning_rate": 4.215150257669276e-11,
      "loss": 0.6802,
      "step": 34662
    },
    {
      "epoch": 0.9987610211490808,
      "grad_norm": 0.7305655479431152,
      "learning_rate": 4.025730019896479e-11,
      "loss": 0.6908,
      "step": 34663
    },
    {
      "epoch": 0.9987898346107301,
      "grad_norm": 0.7419024109840393,
      "learning_rate": 3.840664247245496e-11,
      "loss": 0.686,
      "step": 34664
    },
    {
      "epoch": 0.9988186480723794,
      "grad_norm": 0.71360844373703,
      "learning_rate": 3.65995294132615e-11,
      "loss": 0.6783,
      "step": 34665
    },
    {
      "epoch": 0.9988474615340287,
      "grad_norm": 0.7099508047103882,
      "learning_rate": 3.4835961037482656e-11,
      "loss": 0.6568,
      "step": 34666
    },
    {
      "epoch": 0.998876274995678,
      "grad_norm": 0.7176823019981384,
      "learning_rate": 3.311593735955132e-11,
      "loss": 0.6717,
      "step": 34667
    },
    {
      "epoch": 0.9989050884573273,
      "grad_norm": 0.7169075012207031,
      "learning_rate": 3.1439458395565725e-11,
      "loss": 0.6577,
      "step": 34668
    },
    {
      "epoch": 0.9989339019189766,
      "grad_norm": 0.7184366583824158,
      "learning_rate": 2.980652415940366e-11,
      "loss": 0.6653,
      "step": 34669
    },
    {
      "epoch": 0.9989627153806259,
      "grad_norm": 0.7195708751678467,
      "learning_rate": 2.8217134665498026e-11,
      "loss": 0.6732,
      "step": 34670
    },
    {
      "epoch": 0.9989915288422752,
      "grad_norm": 0.7145923376083374,
      "learning_rate": 2.6671289927726608e-11,
      "loss": 0.6647,
      "step": 34671
    },
    {
      "epoch": 0.9990203423039244,
      "grad_norm": 0.7103452086448669,
      "learning_rate": 2.5168989959967193e-11,
      "loss": 0.6689,
      "step": 34672
    },
    {
      "epoch": 0.9990491557655736,
      "grad_norm": 0.7203572392463684,
      "learning_rate": 2.371023477443224e-11,
      "loss": 0.692,
      "step": 34673
    },
    {
      "epoch": 0.9990779692272229,
      "grad_norm": 0.7227464318275452,
      "learning_rate": 2.2295024383889307e-11,
      "loss": 0.6707,
      "step": 34674
    },
    {
      "epoch": 0.9991067826888722,
      "grad_norm": 0.7134599685668945,
      "learning_rate": 2.0923358801105964e-11,
      "loss": 0.6716,
      "step": 34675
    },
    {
      "epoch": 0.9991355961505215,
      "grad_norm": 0.7139137983322144,
      "learning_rate": 1.9595238038294662e-11,
      "loss": 0.6849,
      "step": 34676
    },
    {
      "epoch": 0.9991644096121708,
      "grad_norm": 0.7177690267562866,
      "learning_rate": 1.831066210655763e-11,
      "loss": 0.6604,
      "step": 34677
    },
    {
      "epoch": 0.9991932230738201,
      "grad_norm": 0.7483103275299072,
      "learning_rate": 1.70696310169971e-11,
      "loss": 0.6749,
      "step": 34678
    },
    {
      "epoch": 0.9992220365354694,
      "grad_norm": 0.726240873336792,
      "learning_rate": 1.5872144780715305e-11,
      "loss": 0.6807,
      "step": 34679
    },
    {
      "epoch": 0.9992508499971187,
      "grad_norm": 0.7520367503166199,
      "learning_rate": 1.471820340825936e-11,
      "loss": 0.7166,
      "step": 34680
    },
    {
      "epoch": 0.999279663458768,
      "grad_norm": 0.7303938269615173,
      "learning_rate": 1.3607806909066157e-11,
      "loss": 0.6955,
      "step": 34681
    },
    {
      "epoch": 0.9993084769204172,
      "grad_norm": 0.7080749869346619,
      "learning_rate": 1.2540955293682822e-11,
      "loss": 0.6745,
      "step": 34682
    },
    {
      "epoch": 0.9993372903820665,
      "grad_norm": 0.7424649000167847,
      "learning_rate": 1.1517648570436024e-11,
      "loss": 0.6749,
      "step": 34683
    },
    {
      "epoch": 0.9993661038437158,
      "grad_norm": 0.7378020286560059,
      "learning_rate": 1.053788674876266e-11,
      "loss": 0.6908,
      "step": 34684
    },
    {
      "epoch": 0.9993949173053651,
      "grad_norm": 0.7242977619171143,
      "learning_rate": 9.601669837544514e-12,
      "loss": 0.6886,
      "step": 34685
    },
    {
      "epoch": 0.9994237307670143,
      "grad_norm": 0.7251747846603394,
      "learning_rate": 8.708997843998035e-12,
      "loss": 0.67,
      "step": 34686
    },
    {
      "epoch": 0.9994525442286636,
      "grad_norm": 0.7186242938041687,
      "learning_rate": 7.859870776449896e-12,
      "loss": 0.6612,
      "step": 34687
    },
    {
      "epoch": 0.9994813576903129,
      "grad_norm": 0.7119255065917969,
      "learning_rate": 7.054288642671659e-12,
      "loss": 0.6666,
      "step": 34688
    },
    {
      "epoch": 0.9995101711519622,
      "grad_norm": 0.7239856123924255,
      "learning_rate": 6.29225144932466e-12,
      "loss": 0.6758,
      "step": 34689
    },
    {
      "epoch": 0.9995389846136115,
      "grad_norm": 0.7260589003562927,
      "learning_rate": 5.573759202515128e-12,
      "loss": 0.7004,
      "step": 34690
    },
    {
      "epoch": 0.9995677980752607,
      "grad_norm": 0.7314122915267944,
      "learning_rate": 4.898811909459511e-12,
      "loss": 0.6831,
      "step": 34691
    },
    {
      "epoch": 0.99959661153691,
      "grad_norm": 0.7068117260932922,
      "learning_rate": 4.267409575153814e-12,
      "loss": 0.6762,
      "step": 34692
    },
    {
      "epoch": 0.9996254249985593,
      "grad_norm": 0.7228374481201172,
      "learning_rate": 3.679552205704262e-12,
      "loss": 0.665,
      "step": 34693
    },
    {
      "epoch": 0.9996542384602086,
      "grad_norm": 0.7389020919799805,
      "learning_rate": 3.13523980610686e-12,
      "loss": 0.6685,
      "step": 34694
    },
    {
      "epoch": 0.9996830519218579,
      "grad_norm": 0.7064288854598999,
      "learning_rate": 2.634472381357611e-12,
      "loss": 0.672,
      "step": 34695
    },
    {
      "epoch": 0.9997118653835072,
      "grad_norm": 0.7307515740394592,
      "learning_rate": 2.1772499347871846e-12,
      "loss": 0.6644,
      "step": 34696
    },
    {
      "epoch": 0.9997406788451565,
      "grad_norm": 0.7196272015571594,
      "learning_rate": 1.7635724719466952e-12,
      "loss": 0.6715,
      "step": 34697
    },
    {
      "epoch": 0.9997694923068058,
      "grad_norm": 0.7297257781028748,
      "learning_rate": 1.393439995056589e-12,
      "loss": 0.6861,
      "step": 34698
    },
    {
      "epoch": 0.9997983057684551,
      "grad_norm": 0.7199012637138367,
      "learning_rate": 1.0668525074475355e-12,
      "loss": 0.653,
      "step": 34699
    },
    {
      "epoch": 0.9998271192301043,
      "grad_norm": 0.7287864685058594,
      "learning_rate": 7.83810013005315e-13,
      "loss": 0.6814,
      "step": 34700
    },
    {
      "epoch": 0.9998559326917535,
      "grad_norm": 0.7408960461616516,
      "learning_rate": 5.44312513395262e-13,
      "loss": 0.7141,
      "step": 34701
    },
    {
      "epoch": 0.9998847461534028,
      "grad_norm": 0.7222028374671936,
      "learning_rate": 3.4836001083782266e-13,
      "loss": 0.6818,
      "step": 34702
    },
    {
      "epoch": 0.9999135596150521,
      "grad_norm": 0.7881169319152832,
      "learning_rate": 1.9595250699833145e-13,
      "loss": 0.6825,
      "step": 34703
    },
    {
      "epoch": 0.9999423730767014,
      "grad_norm": 0.7135899066925049,
      "learning_rate": 8.709000354212294e-14,
      "loss": 0.6517,
      "step": 34704
    },
    {
      "epoch": 0.9999711865383507,
      "grad_norm": 0.7539904117584229,
      "learning_rate": 2.1772501024308612e-14,
      "loss": 0.6668,
      "step": 34705
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7211441993713379,
      "learning_rate": 0.0,
      "loss": 0.6751,
      "step": 34706
    },
    {
      "epoch": 1.0,
      "step": 34706,
      "total_flos": 1.865158684212818e+20,
      "train_loss": 0.7671146870263954,
      "train_runtime": 88327.5037,
      "train_samples_per_second": 201.173,
      "train_steps_per_second": 0.393
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 34706,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 3000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.865158684212818e+20,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}